How Goal Specificity Shapes Motivation: A Reference Points Perspective

How Goal Specificity Shapes Motivation: A Reference Points Perspective Consumers often pursue goals that lack specific end states, such as goals to lose as much weight as possible or to pay off as much debt as possible. Yet despite considerable interest in the consequences of setting nonspecific (vs. specific) goals, how goal specificity affects motivation throughout goal pursuit is less well understood. The current research explores the role of reference points in shaping goal specificity’s effects. We propose that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit: for specific goals, the end state tends to be more salient, but for nonspecific goals, the initial state should be more salient. Five studies investigate how this difference in focal reference points shapes (1) the relationship between goal progress and motivation, (2) when (i.e., at what level of goal progress) goal specificity produces the greatest difference in motivation, and (3) the underlying process driving these effects. Our findings advance understanding of the relationship between goal specificity, goal progress, and motivation, and in doing so, underscore the critical role that reference points play in goal-directed behavior. In addition, the findings offer practical insight into how best to set important financial, health, and other consumer goals to enhance motivation. goals, reference points, motivation The relationship between goal progress and motivation is one of the most robust and well-known findings in the goal pursuit literature (Hull 1932; Kivetz, Urminsky, and Zheng 2006; Louro, Pieters, and Zeelenberg 2007; Nunes and Drèze 2011; Soman and Shi 2003). Often called the “goal gradient” or “goal looms larger” effect, accumulating progress toward a goal tends to make consumers more motivated to pursue it. Scholars have described this phenomenon as “the main insight from classic and modern research on motivation” (Koo and Fishbach 2012). A prominent explanation for this effect comes from the theory of goals as reference points (Heath, Larrick, and Wu 1999). This theory posits that the desired end state of a goal serves as a reference point during goal pursuit, producing a “value function” (Kahneman and Tversky 1979) that drives motivation as a function of distance to the goal end state. Because the value function is steeper closer to the reference point, as consumers accumulate goal progress (i.e., grow closer to the goal’s end state), each unit of marginal goal progress is perceived to have a greater impact on the overall goal, and this increases subsequent motivation. For example, a dieter with a goal to lose six pounds will be more motivated to lose the next pound when he has lost four pounds versus two pounds so far, because he is on a steeper part of the value function (i.e., closer to the goal end state) and therefore sees losing the next pound as more impactful. But what about goals that lack specific end states? While this “reference points” explanation assumes that goals are defined by a specific end state, many consumer goals are not. Rather than striving to lose six pounds, for example, dieters may simply try to lose as much weight as possible, and rather than aiming to pay off $500 of debt, consumers may simply try to pay off as much debt as possible. Such nonspecific “do your best” goals are both common and important. When we asked US adults (N = 149, 19 to 82 years, mean age 35.14 years, 60.8% male) to list a series of personal goals and note whether each was associated with a specific end state, half of the listed goals were nonspecific (i.e., 611 out of 1,188 goals lacked a specific performance objective). Participants also viewed these nonspecific goals as equally important (1 = Not important at all, 7 = Extremely important) as their specific goals (Mnonspecific = 5.73 vs. Mspecific = 5.78; t < 1). How does goal specificity shape motivation during goal pursuit? What effect might the absence of a specific end state have on the relationship between goal progress and motivation? What role might reference points play in goal specificity’s effects? The present research examines these questions. We propose that, lacking a specific end state, nonspecific goal pursuers will use the initial state (i.e., where goal pursuit began) as the reference point instead. Drawing on the value function’s features of diminishing sensitivity and loss aversion (Kahneman and Tversky 1979), we develop a series of hypotheses that describe how this difference in focal reference points shapes the relationship between goal progress and motivation. We first consider how accumulating goal progress affects motivation to pursue nonspecific (vs. specific) goals. Then, we examine when (i.e., at what level of goal progress) goal specificity produces the greatest difference in motivation. Finally, we explore the underlying mechanism driving these effects. The findings make three main contributions. First, this research furthers understanding of the relationship between goal progress and motivation. While a substantial body of work shows that accumulating goal progress can increase motivation (Hull 1932; Kivetz et al. 2006; Nunes and Drèze 2011; Soman and Shi 2003), our findings provide a more nuanced perspective: whether accumulating goal progress increases or decreases subsequent motivation critically depends on goal specificity (i.e., the presence of an end-state reference point). Second, this work furthers understanding of how goal specificity shapes motivation. Whereas goal specificity’s effects have previously been attributed to ambiguity in how performance is evaluated (Naylor and Ilgen 1984; Wright and Kacmar 1994), we introduce a theoretical framework that predicts how motivation to pursue nonspecific versus specific goals differs as a function of salient reference points. Third, this work generalizes the theory of goals as reference points beyond goals that have specific performance objectives. Whereas previous tests of the existing framework have exclusively considered goals that provide an end-state reference point (Bonezzi, Brendl, and De Angelis 2011; Heath et al. 1999; Koo and Fishbach 2012), we develop and test novel predictions for goals that do not (i.e., nonspecific goals). The findings underscore that goal specificity plays a key role in determining what reference points consumers adopt during goal pursuit. GOAL SPECIFICITY Goal specificity is a defining characteristic of consumer goals. Unlike specific goals, nonspecific goals have some degree of “ambiguity or diffuseness in the exact level of performance required” (Hollenbeck and Klein 1987; Naylor and Ilgen 1984; Wright and Kacmar 1994). Whereas specific goals define a desired end-state objective (e.g., lose six pounds, pay off $500 of debt), nonspecific goals do not (e.g., lose as much weight as possible, pay off as much debt as possible). Nonspecific goals can take different forms (e.g., range goals, Scott and Nowlis 2013), but the most common “do your best” type of nonspecific goal lacks an end state entirely (Locke and Latham 1990; Wright and Kacmar 1994). Setting nonspecific versus specific goals has a variety of consequences. Prior work finds that nonspecific (vs. specific) goals are perceived as less difficult and more attainable (Ülkümen and Cheema 2011), which encourages people to adopt them more readily (Locke and Latham 1990; Naylor and Ilgen 1984). Nonspecific (vs. specific) goals are also less likely to evoke feelings of failure, which reduces goal abandonment (Kirschenbaum, Humphrey, and Malett 1981; Soman and Cheema 2004). People also tend to feel less committed to nonspecific (vs. specific) goals (Hollenbeck and Klein 1987; Naylor and Ilgen 1984). This makes nonspecific goals more likely to be revised (Wright and Kacmar 1994), creates greater variability in performance outcomes (Klein, Whitener, and Ilgen 1990; Locke et al. 1989), and can lead to worse performance overall (Locke and Latham 1990; Locke et al. 1981). To explain these prior findings, researchers have argued that the absence of a specific end state introduces ambiguity into how performance is evaluated (Naylor and Ilgen 1984; Wright and Kacmar 1994). Because for nonspecific goals, the goal objective is less precisely defined, a broader range of outcomes can constitute success. For instance, whereas for a goal to lose six pounds, only that single outcome (losing six pounds) would achieve the goal, for a goal to lose as much weight as possible, multiple outcomes (e.g., losing four, six, or eight pounds) could potentially seem sufficient. While this reasoning helps explain the previously documented effects, it offers limited insight into how goal specificity shapes motivation during goal pursuit. A growing body of research reveals goal pursuit to be a dynamic process in which motivation changes as consumers accumulate goal progress (Amir and Ariely 2008; Etkin and Ratner 2012; Huang, Zhang, and Broniarczyk 2012; Kivetz et al. 2006; Koo and Fishbach 2008, 2012). For nonspecific (vs. specific) goals, how motivated will consumers be after accumulating different amounts of goal progress? If a dieter has a goal to lose as much weight as possible, for instance, how motivated would he be to lose more weight having lost two versus four (vs. six, etc.) pounds so far? And for a given level of goal progress (e.g., four pounds lost), how would motivation differ if, rather than lose as much weight as possible, the dieter’s goal was instead to lose six pounds exactly? To address these questions and provide deeper insight into goal specificity’s effects, the current research develops a theoretical framework that describes how the absence of a specific end state influences motivation during goal pursuit. Central to our theorizing is the notion of reference points. GOAL SPECIFICITY: A REFERENCE POINTS APPROACH We propose that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit, and that this difference in focal reference points has important implications for the relationship between goal progress and motivation. A “reference point” divides the space of outcomes into regions of gain and loss (Kahneman and Tversky 1979). Outcomes above the reference point are evaluated as gains and outcomes below the reference point are evaluated as losses. The valuation of these gains and losses varies systematically based on the slope of Prospect Theory’s value function, which is steeper closer to the reference point (i.e., diminishing sensitivity) and steeper on the loss side than on the gain side of the reference point (i.e., loss aversion, Kahneman and Tversky 1979). Differences between outcomes along a steeper part of the value function have a greater influence on subsequent decisions (Kahneman 1992; Larrick, Heath, and Wu 2009; Tversky and Kahneman 1991). While the notion of reference points has long been established, more recent research has attempted to understand where reference points originate (Abeler et al. 2011; Allen et al. 2016; Barberis 2013). One important source is consumers’ goals. The theory of goals as reference points (Heath et al. 1999) posits that the desired end state of a goal serves as the reference point during goal pursuit, and goal-related outcomes (i.e., levels of goal progress) are evaluated relative to that end state. For instance, if a dieter has a goal to lose six pounds, the dieter’s reference point will be the goal objective (six pounds lost) and he will evaluate his current goal progress (e.g., four pounds lost so far) relative to that desired end state. We propose that, absent a specific end state to serve as a reference point, nonspecific goal pursuers will use the initial state (i.e., where goal pursuit began) instead. Recent work finds that, in addition to the end state, the initial state of a specific goal can also serve as a reference point (Bonezzi et al. 2011; Carton et al. 2011; Koo and Fishbach 2008, 2012; Touré-Tillery and Fishbach 2012). For a goal to lose six pounds, for instance, either the end state (i.e., the six-pound goal objective) or the initial state (i.e., the dieter’s previous weight, or zero pounds lost) could serve as the reference point. While the end state is naturally more salient for specific goals (Heath et al. 1999; Kivetz et al. 2006), incidental factors that make the initial state more salient (e.g., goal progress feedback, Koo and Fishbach 2012; visual cues, Bonezzi et al. 2011) can encourage people to adopt it as the reference point instead. Because the absence of an end state should make the initial state more salient, we argue that nonspecific goal pursuers will spontaneously adopt the initial state as the focal reference point. CONSEQUENCES FOR MOTIVATION We propose that this difference in focal reference points plays a key role in how goal specificity shapes motivation. According to the theory of goals as reference points (Heath et al. 1999), the slope of the value function determines motivation by changing people’s subjective valuation of the impact of marginal goal progress (i.e., the “next step” of goal progress). Because the value function is nonlinear, the same objective increase in goal progress (e.g., losing one more pound) can be perceived as contributing more or less to the overall goal (e.g., lose six pounds). When one’s current goal progress falls on a steeper part of the value function, marginal goal progress seems more impactful. By determining the shape of the value function, salient reference points influence the subjective impact of marginal goal progress, and thus motivation. As previously discussed, diminishing sensitivity makes the value function steeper when one’s current state is closer to the reference point, and loss aversion makes the value function steeper when one’s current state is below the reference point (i.e., on the loss rather than the gain side of the value function). Consequently, because marginal goal progress seems more impactful when the value function is steeper, motivation is higher when consumers’ current goal progress puts them closer to their focal reference point or on the loss side of that reference point (Bonezzi et al. 2011; Heath et al. 1999; Koo and Fishbach 2012). We argue that goal specificity influences the shape of the value function, and thus changes how accumulating goal progress affects subsequent motivation. For specific goals, diminishing sensitivity should make the value function steeper closer to the (more salient) end state (Heath et al. 1999). Consequently, as consumers accumulate goal progress, they move closer to their focal reference point (and onto a steeper part of the value function), which makes marginal goal progress seem more impactful and increases subsequent motivation (i.e., the “goal gradient” effect, Kivetz et al. 2006). For nonspecific goals, however, diminishing sensitivity should make the value function steeper closer to the (more salient) initial state. Consequently, as consumers accumulate goal progress, they move further away from their focal reference point (and onto a shallower part of the value function). This should make marginal goal progress seem less impactful and therefore decrease subsequent motivation. For example, the dieter with a goal to lose as much weight as possible should see losing the next pound as having less of an impact on his overall weight loss goal, and thus be less motivated to lose more weight, after having lost four pounds (further from zero) versus two pounds (closer to zero) so far. For nonspecific goals, we thus predict a reverse goal gradient: accumulating goal progress will decrease subsequent motivation, driven by a decrease in the subjective impact of marginal goal progress. Our reasoning thus far describes a crossover interaction between goal specificity and goal progress (figure 1): for specific goals, motivation starts low (far from the focal end-state reference point) and increases with accumulated goal progress; for nonspecific goals, motivation starts high (near the focal initial-state reference point) and decreases with accumulated goal progress. This suggests that when goal progress is relatively high, nonspecific goals should be less motivating than specific goals, but when goal progress is relatively low, nonspecific goals should be more motivating than specific goals. FIGURE 1 View largeDownload slide PREDICTED EFFECTS OF GOAL SPECIFICITY AND GOAL PROGRESS ON SUBSEQUENT MOTIVATION NOTE.—Due to diminishing sensitivity, accumulating goal progress decreases (increases) motivation to pursue nonspecific (specific) goals. Due to loss aversion, the effect of goal specificity on motivation is greater at higher (vs. lower) levels of goal progress. FIGURE 1 View largeDownload slide PREDICTED EFFECTS OF GOAL SPECIFICITY AND GOAL PROGRESS ON SUBSEQUENT MOTIVATION NOTE.—Due to diminishing sensitivity, accumulating goal progress decreases (increases) motivation to pursue nonspecific (specific) goals. Due to loss aversion, the effect of goal specificity on motivation is greater at higher (vs. lower) levels of goal progress. Rather than a symmetrical crossover, however, we argue that loss aversion will produce an asymmetry in this interaction (figure 1). Whereas focusing on the end state locates current goal progress on the loss side of the value function (e.g., $250 below a savings goal of $500), focusing on the initial state locates current goal progress on the gain side (e.g., $250 above a starting point of $0). Because loss aversion makes losses steeper than gains, for a given level of goal progress, focusing on the end state (vs. initial state) as the reference point should put consumers on a steeper part of the value function. Together with the effect of diminishing sensitivity, this suggests that the value function should be at its steepest (shallowest) when current goal progress is both close to (far from) the focal reference point and on the loss (gain) side of that reference point. Consequently, with respect to goal specificity, the value function should be steepest for specific goals at high goal progress (loss side of the value function, close to the end-state reference point) and shallowest for nonspecific goals at high goal progress (gain side of the value function, far from the initial-state reference point); see figure 2. At low goal progress, the effects of diminishing sensitivity and loss aversion should act in opposition, with a net result of more moderate motivation for both specific and nonspecific goals (figure 2). FIGURE 2 View largeDownload slide POSITION RELATIVE TO FOCAL REFERENCE POINT NOTE.—Goal specificity interacts with accumulated goal progress to shape motivation via the slope of the value function. Motivation is greater when current goal progress is closer to (vs. further from) the focal reference point and when it is on the loss (vs. gain) side of the value function (i.e., below vs. above the reference point). FIGURE 2 View largeDownload slide POSITION RELATIVE TO FOCAL REFERENCE POINT NOTE.—Goal specificity interacts with accumulated goal progress to shape motivation via the slope of the value function. Motivation is greater when current goal progress is closer to (vs. further from) the focal reference point and when it is on the loss (vs. gain) side of the value function (i.e., below vs. above the reference point). We thus predict that goal specificity will produce a greater difference in the subjective impact of marginal goal progress (and thus motivation) at higher (vs. lower) levels of goal progress. In particular, when goal progress is high, nonspecific goals should decrease subjective impact (and motivation) relative to specific goals, but when goal progress is low, these effects should be attenuated.1 In summary, we predict: H1: For nonspecific goals (specific goals), accumulating goal progress decreases (increases) subsequent motivation. H2: When current goal progress is high, nonspecific goals reduce motivation relative to specific goals, but this effect is attenuated when current goal progress is low. H3: These effects are driven by the subjective impact of marginal goal progress. Five studies tested our hypotheses. Study 1 used an effortful lab task to examine how goal specificity shapes motivation. Studies 2 and 3 used realistic scenarios in important consumer goal domains (debt repayment in study 2; weight loss in study 3) to provide more controlled tests of our motivation predictions and examine the proposed underlying role of the subjective impact of marginal goal progress. Studies 4a and 4b further tested the proposed underlying process by directly manipulating the focal reference point. Together the findings show that how goal specificity shapes the dynamics of motivation depends on the different reference points that nonspecific (vs. specific) goals make salient. STUDY 1 Study 1 tests our first two hypotheses by examining effort on a goal-directed task: proofreading passages of text. We manipulated goal specificity and then measured motivation (i.e., persistence) at different points throughout the task. In the specific goal condition, we predicted that accumulating goal progress would increase subsequent motivation: after finding a greater number of errors, participants should work harder to find additional errors. In the nonspecific goal condition, however, we predicted that accumulating goal progress would instead decrease subsequent motivation: after finding a greater number of errors, participants should work less hard to find more errors. Further, due to loss aversion, we predicted that the nonspecific goal would be less motivating than the specific goal at the highest level of goal progress, but this effect would be reduced at lower progress levels. Design and Method Participants (N = 155) were recruited from a university behavioral lab in exchange for course credit. In this and subsequent lab studies, lab capacity and participant availability determined the sample size. Ten individuals (6%) reported technical problems completing the study (e.g., failure to load a page) and were excluded from the analyses, leaving a sample of 145 (average age = 24.67 years, 59% female). Participants were randomly assigned to one condition of a 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design. Participants read that they would be proofreading a series of short text passages and that there was one spelling error in each passage. In the specific goal condition, we told participants that their goal was to “find 10 errors in a row.” In the nonspecific goal condition, we told participants that their goal was to “find as many errors as possible in a row.” All participants read that if they failed to find the error in a given passage, their streak would end, and they would not be able to restart their streak or revisit the failed passage. After completing a practice passage, participants began the main proofreading task. Participants proceeded through the proofreading task as instructed, and were given a running count of how many errors they had found so far (which equaled the number of passages they had completed). After finding two (low progress condition), five (intermediate progress condition), or eight errors (high progress condition), we paused the task. We told participants that the remaining proofreading passages would be more difficult, and that if they failed to find the spelling error in one of the passages, they could quit the task (and end their streak). Then, participants returned to the main task, and we measured motivation. The next (target) text passage contained no spelling errors, meaning that in order to advance beyond this page, all participants eventually had to quit. We recorded how long participants persisted (i.e., how much effort they invested) in trying to find the error before quitting. Persistence time was log-transformed for analysis to correct for non-normality (Kolmogorov-Smirnov test statistic: .11, p < .01); raw means are reported for ease of interpretation. Results A 2 (goal specificity) × 3 (goal progress) ANOVA on motivation (i.e., persistence time) revealed a main effect of goal specificity (Mspecific = 146.53, Mnonspecific = 105.93, F(1, 139) = 7.24, p = .008), qualified by the predicted interaction (F(2, 139) = 6.19, p = .003; figure 3). There was no main effect of goal progress (F < 1). FIGURE 3 View largeDownload slide GOAL SPECIFICITY AFFECTS PROOFREADING EFFORT FIGURE 3 View largeDownload slide GOAL SPECIFICITY AFFECTS PROOFREADING EFFORT As expected, in the specific goal condition, accumulating goal progress increased subsequent motivation (linear contrast: F(1, 139) = 4.90, p = .028). Participants expended more effort on the target passage (i.e., worked harder to find the nonexistent error) after finding five errors (M = 140.56 sec) versus two errors (M = 117.35 sec), and after finding eight errors (M = 188.15 sec) versus five errors. However, supporting hypothesis 1, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased subsequent motivation (linear contrast: F(1, 139) = 7.78, p = .006). Participants expended less effort on the target passage after finding five errors (M = 105.71 sec) versus two errors (M = 136.08 sec), and after finding eight errors (M = 78.11 sec) versus five errors. See table 1 for pairwise contrasts. Table 1 Pairwise Contrasts between Goal Progress Conditions     Specific goal   Nonspecific goal   Study (DV)    Low vs. middle  Middle vs. high  Low vs. high  Low vs. middle  Middle vs. high  Low vs. high  1 (Log time)  F(1, 139)  2.17  .77  4.90**  1.69  1.72  7.78***  p  .143  .381  .028  .196  .192  .006  2 (Log WTP)  F(1, 301)  1.64  3.54*  8.83***  6.50**  .91  12.65***  p  .202  .061  .003  .011  .341  < .001  2 (Impact)  F(1, 301)  2.29  1.43  6.48**  7.88***  0.50  12.70***  p  .132  .232  .011  .005  .479  < .001      Specific goal   Nonspecific goal   Study (DV)    Low vs. middle  Middle vs. high  Low vs. high  Low vs. middle  Middle vs. high  Low vs. high  1 (Log time)  F(1, 139)  2.17  .77  4.90**  1.69  1.72  7.78***  p  .143  .381  .028  .196  .192  .006  2 (Log WTP)  F(1, 301)  1.64  3.54*  8.83***  6.50**  .91  12.65***  p  .202  .061  .003  .011  .341  < .001  2 (Impact)  F(1, 301)  2.29  1.43  6.48**  7.88***  0.50  12.70***  p  .132  .232  .011  .005  .479  < .001  NOTE.—Pairwise contrasts in each goal specificity condition of studies 1 and 2: low versus intermediate goal progress, intermediate versus high goal progress, and low versus high goal progress. As expected, the low versus high goal progress contrast emerged as significant in each case. * p < .10, **p < .05, ***p < .01. Further, supporting hypothesis 2, when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. After finding eight errors, participants in the nonspecific goal condition expended less effort on the target passage than did those in the specific goal condition (Mnonspecific = 78.11 sec, Mspecific = 188.15 sec; F(1, 139) = 15.21, p < .001). This effect was reduced, however, at intermediate progress (five errors, Mnonspecific = 105.71 sec, Mspecific = 140.56 sec; F(1, 139) = 3.00, p = .085) and directionally reversed at low goal progress (two errors, Mnonspecific = 136.08 sec, Mspecific = 117.35 sec; F(1, 139) = 1.06, p = .305). Discussion Study 1 supports our first two hypotheses with effortful behavior on a goal-directed task. Consistent with prior work (Heath et al. 1999; Kivetz et al. 2006; Nunes and Drèze 2011), when participants had a specific goal, accumulating goal progress increased subsequent motivation. The more progress participants had made in the proofreading task (i.e., the greater the number of errors they had found so far), the more effort they expended on the target passage. Importantly, as predicted (hypothesis 1), when participants had a nonspecific goal, the opposite occurred: accumulating goal progress decreased subsequent motivation. The more progress participants had made in the task, the less effort they expended on the target passage. Also as predicted (hypothesis 2), when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. This effect was attenuated (and directionally reversed), however, at the lower goal progress levels. Because high goal progress is where the effect of loss aversion reinforces that of diminishing sensitivity (figure 2), this is where goal specificity had the biggest effect on subsequent motivation. STUDY 2 Study 2 tests the proposed underlying process (hypothesis 3) in a common and important goal domain: debt repayment. We manipulated the specificity of a debt repayment goal, provided participants with goal progress feedback, and then, in addition to measuring motivation, asked participants to rate the impact of an incremental step of goal progress (saving an additional $25) on their overall debt repayment goal. In this and subsequent studies, we use realistic goal scenarios to provide more controlled tests of our predictions. Although study 1 held the (perceived) rate of goal progress constant, the overall amount of effort invested prior to reaching the target passage differed by condition. Consequently, in the nonspecific goal condition, depletion may have played a role in why accumulating goal progress decreased subsequent motivation. By manipulating goal progress independently of actual effort investment, scenario-based paradigms eliminate this potential confound, allowing us to more precisely test the proposed underlying mechanism. This approach is consistent with prior research on goals as reference points, which has extensively used scenario-based paradigms (Heath et al. 1999) and found them to show the same effects as real behavior (see Wu, Heath, and Larrick 2008 for a review). Design and Method Participants (N = 320) were recruited from Amazon Mechanical Turk in exchange for small payment. In this web-based study, a target rule of 50–60 participants per condition determined the sample size. Thirteen individuals (4%) initiated the study but failed to complete it, leaving a sample of 307 responses for analysis (average age = 35.04 years, 44.0% female). Due to the online format, we were unable to ask participants their reasons for exiting, but attrition did not differ across conditions. Participants were randomly assigned to one condition in a 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design. First, we manipulated goal specificity. We asked participants to imagine they had $10,000 in loans to pay off over time, and that they decided to pay down their loans faster by cutting back on spending. In the specific goal condition, participants read that their goal this month was to “pay off an extra $500.” A pretest conducted with a separate sample from the same population confirmed that this amount ($500) was comparable to what people would naturally plan to pay off in a month (see appendix A for details). In the nonspecific goal condition, participants read that their goal this month was to “pay off as much extra as you can.” Second, we provided goal progress feedback. Participants read that partway through the month, they were planning to go out to dinner with a friend, and that so far this month they had saved $50 (low progress condition), $250 (intermediate progress condition), or $450 (high progress condition) to put toward their loans. A pretest conducted with a separate sample from the same population confirmed that in both goal specificity conditions, perceived goal progress increased from $50 to $250 and from $250 to $500 (see appendix B for details). Third, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much of an impact would saving an additional $25 have on helping you reach your goal for the month?” (1 = No impact at all, 7 = Very large impact). Finally, we measured motivation. We reasoned that the more motivated participants were to put money toward their debt repayment goal, the less money they should be willing to spend on dinner with their friend. Accordingly, we asked them, “How much money would you be willing to spend on this dinner with your friend?” (open-ended in dollars), where larger values indicated lower motivation to conserve money. Willingness to pay was log-transformed for analysis to correct for non-normality (Kolmogorov-Smirnov test statistic: .23, p < .01); raw means are reported for ease of interpretation. Results Motivation A 2 (goal specificity) × 3 (goal progress) ANOVA on motivation (i.e., willingness to pay) revealed a marginal main effect of goal specificity (Mnonspecific = $30.83, Mspecific = $26.75; F(1, 301) = 3.25, p = .073) qualified by the predicted interaction (F(2, 301) = 10.76, p < .001; figure 4). There was no main effect of goal progress (F < 1). FIGURE 4 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO PAY OFF DEBT NOTE.—Willingness to pay for dinner corresponds to lower motivation to conserve money. FIGURE 4 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO PAY OFF DEBT NOTE.—Willingness to pay for dinner corresponds to lower motivation to conserve money. Consistent with study 1, in the specific goal condition, accumulating goal progress increased subsequent motivation (linear contrast: F(1, 301) = 8.83, p = .003). Participants were more motivated to conserve money (i.e., willing to spend less money on dinner) after putting $250 (M = $26.92) versus $50 (M = $31.90) toward their loans, and after putting $450 (M = $21.37) versus $250 toward their loans. However, supporting hypothesis 1, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased subsequent motivation (linear contrast: F(1, 301) = 12.65, p < .001). Participants were less motivated to conserve money (i.e., willing to spend more money on dinner) after putting $250 (M = $32.13) versus $50 (M = $23.59) toward their loans, and after putting $450 (M = $37.27) versus $250 toward their loans. See table 1 for pairwise contrasts. Further, consistent with study 1 and supporting hypothesis 2, when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. After paying off $450, participants in the nonspecific goal condition were less motivated to conserve money (i.e., willing to spend more on dinner) than those in the specific goal condition (Mnonspecific = $37.27, Mspecific = $21.37; F(1, 301) = 16.71, p < .001). This effect was reduced, however, at intermediate goal progress ($250; Mnonspecific = $32.13, Mspecific = $26.92; F(1, 301) = 2.04, p = .154), and it reversed (although the effect was smaller, consistent with our theory) at low goal progress ($50; Mnonspecific = $23.59, Mspecific = $31.90; F(1, 301) = 5.89, p = .016). Subjective Impact A 2 (goal specificity) × 3 (goal progress) ANOVA on subjective impact revealed only the predicted interaction (F(2, 314) = 11.54, p < .001). There was no main effect of goal specificity (F(1, 301) = 2.51, p = .114) or goal progress (F < 1). As expected, in the specific goal condition, accumulating goal progress increased the subjective impact of marginal goal progress (linear contrast: F(1, 301) = 6.48, p = .011). Saving an additional $25 was perceived to have a bigger impact on the overall debt repayment goal when participants had already put $250 (M = 4.16) versus $50 (M = 3.74) toward their loans, and when they had already put $450 (M = 4.49) versus $250 toward their loans. However, supporting our theory, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased the subjective impact of marginal goal progress (linear contrast: F(1, 301) = 12.70, p < .001). Saving an additional $25 was perceived to have less of an impact on the overall debt repayment goal when participants had already put $250 (M = 3.65) versus $50 (M = 4.49) toward their loans, and when they had already put $450 (M = 3.44) versus $250 toward their loans. See table 1 for pairwise contrasts. Further supporting our reasoning, when goal progress was high, the nonspecific (vs. specific) goal reduced the subjective impact of marginal goal progress. After paying off $450, participants in the nonspecific goal condition perceived saving an additional $25 as less impactful than those in the specific goal condition (Mnonspecific = 3.44, Mspecific = 4.49; F(1, 301) = 12.45, p < .001). This effect was reduced, however, at intermediate goal progress ($250, Mnonspecific = 3.65, Mspecific = 4.16; F(1, 301) = 3.17, p = .077), and it reversed (although the effect was smaller, consistent with our theory) at low goal progress ($50, Mnonspecific = 4.49, Mspecific = 3.73; F(1, 301) = 7.88, p = .005). Underlying Process To examine the proposed underlying role of the subjective impact of marginal goal progress, we ran a bias-corrected bootstrapping mediated moderation analysis with 5,000 samples (PROCESS model 7, Hayes 2013). Results supported hypothesis 3, revealing a significant index of mediated moderation (index: –.06, 95% CI [–.12 to –.01]). In the specific goal condition, accumulating goal progress increased motivation to conserve money (i.e., decreased willingness to spend money on dinner), driven by seeing marginal progress as more impactful (ab = –.02, 95% CI [–.06 to –.002]). In the nonspecific goal condition, however, accumulating goal progress decreased motivation to conserve money (i.e., increased willingness to spend money on dinner), driven by seeing marginal progress as less impactful (ab = .03, 95% CI [.008 to .07]). Discussion Study 2 supports our motivation hypotheses in an important goal domain (debt repayment) and demonstrates the underlying process. Consistent with study 1 and supporting hypothesis 1, when participants had a specific (nonspecific) goal to pay off debt, accumulating goal progress increased (decreased) subsequent motivation to conserve money (i.e., reduced the amount participants were willing to spend on dinner). Also consistent with study 1 and supporting hypothesis 2, when goal progress was high, the nonspecific debt repayment goal reduced motivation relative to the specific goal, but this effect was attenuated at the lower goal progress levels (and even reversed at the lowest level). Importantly, supporting hypothesis 3, these effects were driven by the perceived impact of marginal goal progress (i.e., of saving an additional $25) on the overall debt repayment goal. When participants had a specific (nonspecific) goal, accumulating goal progress made marginal goal progress seem more (less) impactful, and these judgments of subjective impact determined subsequent motivation. Lastly, that we found support for our hypotheses in this controlled realistic goal scenario casts doubt on the possibility that depletion or other differences between the previous study’s goal progress conditions might have explained the effects. STUDY 3 Study 3 underscores the findings of study 2 in a different important goal domain: weight loss. We manipulated the specificity of a weight loss goal, provided participants with goal progress feedback, and then measured motivation and the subjective impact of marginal goal progress (i.e., losing an additional pound) on their overall weight loss goal. To rule alternative explanations, we made two key adjustments to the study design. First, to ensure that the specific goal is not artificially increasing participants’ aspiration level, we set the specific goal level lower than the natural aspiration level. Doing so rules out the possibility that the asymmetry in the goal specificity × goal progress interaction, which we argue is driven by loss aversion, could be attributed to differences in goal difficulty (i.e., the nonspecific goal being less challenging than the specific goal) or perceived goal completion (i.e., nonspecific goal pursuers inferring goal completion at higher progress levels). Second, to rule out potential order effects, we measured subjective impact after motivation. Design and Method Participants (N = 229) were recruited from a university behavioral lab in exchange for course credit. Four individuals declined to participate after reading an eating-disorder trigger warning on the consent page, and 10 participants (4%) began the study but did not complete it, leaving a sample of 215 (average age = 25.21 years, 66.5% female). Participants were randomly assigned to one condition in a 2 (goal specificity: specific, nonspecific) × 2 (goal progress: low, high) between-subjects design. First, we manipulated goal specificity. We asked participants to imagine that they had a goal to lose weight over the next eight weeks. In the specific goal condition, participants reported their current body weight and read that their goal was to “lose 6 pounds over the next 8 weeks” from this starting weight. A pretest conducted with a separate sample from the same population indicated that this goal was just 63% of what people would naturally aim to lose in eight weeks (see appendix A for details). In the nonspecific goal condition, participants reported their current body weight and read that their goal was to “lose as much weight as you can over the next 8 weeks” from this starting weight. Second, we provided goal progress feedback. In the low (high) goal progress condition, participants read that, “near the beginning (end) of the 8 weeks, you’ve lost 1.5 pounds (4.5 pounds) so far. Based on your starting weight of [X] pounds, you would weigh [Y] pounds at this point.” The exact current weight was automatically calculated for each participant based on his or her reported starting weight. A pretest conducted with a separate sample confirmed that in both goal specificity conditions, perceived goal progress increased from 1.5 pounds to 4.5 pounds (see appendix B for details). Third, we measured motivation. Participants responded to two questions: “How motivated would you be to lose weight at this point?” (1 = Not motivated at all, 7 = Extremely motivated) and “How hard would you be willing to work to lose weight at this point?” (1 = Unwilling to work hard, 7 = Willing to work very hard). These items were highly correlated (r = .79) and combined. Finally, we measured the subjective impact of marginal progress. We asked participants, “At this point, how much of an impact would losing the next pound have on your weight loss goal?” (1 = No impact at all, 7 = Very large impact). Results Motivation A 2 (goal specificity) × 2 (goal progress) ANOVA on motivation revealed only the predicted interaction (F(1, 211) = 12.49, p = .001; figure 5). There was no main effect of goal specificity or goal progress (Fs < 1). FIGURE 5 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO LOSE WEIGHT FIGURE 5 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO LOSE WEIGHT Consistent with studies 1 and 2, in the specific goal condition, accumulating goal progress increased subsequent motivation (F(1, 211) = 5.90, p = .016). Specific goal pursuers were more motivated to lose additional weight after losing 4.5 pounds (M = 5.59) versus 1.5 pounds so far (M = 4.82). Supporting hypothesis 1, however, in the nonspecific goal condition, the opposite occurred (F(1, 211) = 6.64, p = .011). Participants were less motivated to lose additional weight after losing 4.5 pounds (M = 4.65) versus 1.5 pounds so far (M = 5.41). Further, consistent with the previous studies and supporting hypothesis 2, when goal progress was high, the nonspecific (vs. specific) goal reduced motivation (Mnonspecific = 4.65, Mspecific = 5.59; F(1, 211) = 9.82, p = .002). When goal progress was low, however, this effect (marginally) reversed (Mnonspecific = 5.41, Mspecific = 4.82; F(1, 211) = 3.56, p = .060). Subjective Impact A 2 (goal specificity) × 2 (goal progress) ANOVA on subjective impact revealed a significant main effect of goal specificity (Mnonspecific = 4.89, Mspecific = 5.37; F(1, 211) = 4.69, p = .03), qualified by the predicted interaction (F(2, 211) = 12.86, p < .001). There was no main effect of goal progress (F < 1). Consistent with study 2, in the specific goal condition, accumulating goal progress increased the subjective impact of marginal goal progress (F(1, 211) = 4.46, p = .036). Losing an additional pound was perceived to have a bigger impact when participants had already lost 4.5 pounds (M = 5.68) versus 1.5 pounds (M = 5.06). As expected, however, in the nonspecific goal condition, the opposite occurred (F(1, 211) = 8.93, p = .003). Losing an additional pound was perceived to have less of an impact when participants had lost 4.5 pounds (M = 4.52) versus 1.5 pounds (M = 5.35). Further, also as expected, when goal progress was high, the nonspecific (vs. specific) goal reduced the subjective impact of marginal goal progress (Mnonspecific = 4.52, Mspecific = 5.68; F(1, 211) = 17.16, p < .001). When goal progress was low, however, this effect directionally reversed (Mnonspecific = 5.35, Mspecific = 5.06; F(1, 211) = .973, p = .325). Underlying Process As in study 2, we ran a bias-corrected bootstrapping mediated moderation analysis to examine the underlying process. Results supported hypothesis 3, revealing a significant index of mediated moderation (index: .91, 95% CI [.41 to 1.51]). In the specific goal condition, accumulating goal progress increased motivation to lose weight, driven by seeing marginal goal progress as more impactful (ab = .39, 95% CI [.05 to .79]). In the nonspecific goal condition, however, accumulating goal progress decreased motivation to lose weight, driven by seeing marginal goal progress as less impactful (ab = –.52, 95% CI [–.92 to –.19]). Discussion Study 3 provides further support for our theory in a different goal domain: weight loss. Consistent with the previous studies and supporting hypothesis 1, when participants had a specific (nonspecific) goal to lose weight, accumulating goal progress increased (decreased) subsequent motivation. Consistent with the previous studies and supporting hypothesis 2, when goal progress was high, the nonspecific (vs. specific) goal reduced motivation, but this effect was eliminated (and directionally reversed) when goal progress was low. Finally, consistent with study 2 and supporting hypothesis 3, these motivation effects were driven by the perceived impact of marginal goal progress (i.e., losing an additional pound) on the overall weight loss goal. Notably, because the specific goal value was calibrated to be one-third lower than participants’ natural targets, these results cannot be explained by nonspecific goal pursuers being less ambitious or feeling that the higher progress level constituted goal completion. We have argued that goal specificity alters the relationship between goal progress and motivation because it makes different reference points salient: the end state for specific goals and the initial state for nonspecific goals. To test the role of reference points more directly, our next two studies (4a and 4b) manipulate the focal reference point for specific goal pursuers—the end state (as is naturally the case) or the initial state—and compare their judgments of subjective impact and motivation to nonspecific goal pursuers. If a difference in salient reference points underlies goal specificity’s effects, as we suggest, then encouraging specific goal pursuers to instead use the initial state as the reference point should make them appear like nonspecific goal pursuers: exhibiting a reverse goal gradient (study 4a) and reducing motivation relative to specific goal pursuers focused on the end state at high goal progress (studies 4a and 4b). Notably, if manipulating specific goal pursuers’ focal reference point attenuates goal specificity’s effects, as we expect, this would further rule out potential alternative explanations due to goal difficulty or goal completion (which rely on unrelated differences between nonspecific and specific goals). STUDY 4A Study 4a directly tests the proposed role of reference point focus in generating goal specificity’s effects. Following the paradigm of study 2, we manipulated the specificity of a debt repayment goal and the current level of goal progress. In addition, in the specific goal condition, we directed some participants to focus on the initial state as the reference point. We expected that, rather than motivation increasing with accumulated goal progress, motivation would decrease with accumulated goal progress (i.e., a reverse goal gradient) in this case. Further, as with nonspecific goal pursuers, we expected that specific goal pursuers focused on the initial state would be less motivated at high (vs. low) goal progress than those in the specific control condition. Design and Method Participants (N = 312) were recruited from a university behavioral lab in exchange for course credit. Four individuals (1%) reported technical problems and failed to complete the study, leaving a sample of 308 (average age = 22.48 years, 70.5% female). Participants were each randomly assigned to one condition in a 3 (reference point focus: specific control, specific initial-state focus, nonspecific) × 2 (goal progress: low, high) between-subjects design. Note that (here and in study 4b) there was no “nonspecific end-state focus” condition, because based on our conceptualization, the end-state reference point does not exist for nonspecific goals. First, we manipulated goal specificity. As in study 2, we asked participants to imagine they were paying off loans over time. In the two specific goal conditions, participants read that their goal this month was to “pay off an extra $500.” In the nonspecific goal condition, participants read that their goal this month was to “pay off as much extra as you can.” Second, we provided goal progress feedback. Participants read that partway through the month, they were planning to go out to dinner with a friend, and that so far this month they had put $50 (low progress condition) or $450 (high progress condition) toward their loans. Third, we manipulated the focal reference point. On the same page as the goal progress feedback, participants viewed a progress bar with a dotted line indicating their current progress level (see appendix C). Following a manipulation used in prior work (Bonezzi et al. 2011; Koo and Fishbach 2012), in the specific initial-state focus condition, we instructed participants to highlight the portion of the progress bar corresponding to their accumulated goal progress (i.e., the area between their current state and the initial state). This encouraged them to compare their current goal progress to the initial-state (rather than the end-state) reference point. Participants in the specific control and nonspecific conditions proceeded directly to the next part of the study. Fourth, we measured motivation. All participants received information about two potential restaurants for the dinner with their friend: Restaurant A, which was described as a restaurant the friend liked with an average cost of $35 per person for dinner, and Restaurant B, which was described as another restaurant the friend liked with an average cost of $20 per person for dinner. We reasoned that the more motivated participants were to put money toward their debt repayment goal, the more they should prefer Restaurant B (the less expensive option) to Restaurant A. Accordingly, we asked them, “Would you be more likely to choose Restaurant A or the less-expensive Restaurant B?” (1 = Definitely Restaurant A, 7 = Definitely Restaurant B). Finally, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much of an impact would saving an additional $15 have on helping you reach your goal for the month?” (1 = No impact at all, 7 = Very large impact). Results Motivation A 3 (reference point focus) × 2 (goal progress) ANOVA on motivation (i.e., preference for the inexpensive restaurant) revealed only the predicted interaction (F(2, 302) = 3.46, p = .033; figure 6). There was no main effect of reference point focus (F(2, 302) = 2.16, p = .118) or goal progress condition (F(1, 302) = 2.54, p = .112). FIGURE 6 View largeDownload slide REFERENCE POINT FOCUS AFFECTS MOTIVATION TO PAY OFF DEBT FIGURE 6 View largeDownload slide REFERENCE POINT FOCUS AFFECTS MOTIVATION TO PAY OFF DEBT Consistent with the prior studies, in the specific control condition, accumulating goal progress (i.e., putting $50 vs. $450 toward the loans) increased subsequent preference for the inexpensive restaurant (although the effect was only directional in this case; Mlow = 5.91, Mhigh = 6.23; F(1, 302) = 1.42, p = .235). In the nonspecific goal condition, however, accumulating goal progress decreased preference for the inexpensive restaurant (Mlow = 6.29, Mhigh = 5.68; F(1, 302) = 5.41, p = .021). Importantly, supporting our theory, in the specific initial-state focus condition, accumulating goal progress also decreased subsequent preference for the inexpensive restaurant (albeit marginally, F(1, 302) = 2.73, p = .099). When we encouraged specific goal pursuers to adopt the initial state as their reference point, as nonspecific goal pursuers do naturally, they were less motivated to conserve money after putting $450 (M = 5.49) versus $50 (M = 5.92) toward their loans (similar to those in the nonspecific goal condition). Also consistent with the prior studies, when goal progress was high, participants in the nonspecific goal condition were less motivated than those in the specific control condition (Mnonspecific = 5.68, Mspecific-control = 6.23; F(1, 302) = 4.17, p = .042). However, supporting our theory, this difference was eliminated when specific goal pursuers were encouraged to focus on the initial state: motivation was lower (M = 5.49) than in the specific control condition (F(1, 302) = 7.65, p = .006) and no different from the nonspecific goal condition (F < 1). When goal progress was low, motivation did not differ between the nonspecific goal (M = 6.29), specific control (M = 5.91), and specific initial-state focus conditions (M = 5.92). See table 2 for pairwise contrasts. Table 2 Pairwise Contrasts between Reference Point Focus Conditions Study, progress (DV)    Specific control vs. nonspecific  Specific control vs. specific initial-state focus  Nonspecific vs. specific initial-state focus  4a, high (preference)  F(1, 302)  4.17**  7.65***  .55  p  .042  .006  .458  4a, high (impact)  F(1, 302)  12.32***  15.11***  .13  p  .001  < .001  .715  4a, low (preference)  F(1, 302)  2.12  < .001  1.91  p  .146  .986  .168  4a, low (impact)  F(1, 302)  2.62  .50  .77  p  .107  .481  .381  4b, high (motivation)  F(1, 189)  10.89***  9.67***  .09  p  .001  .002  .771  4b, high (impact)  F(1, 189)  3.25*  5.42**  .21  p  .073  .021  .646  Study, progress (DV)    Specific control vs. nonspecific  Specific control vs. specific initial-state focus  Nonspecific vs. specific initial-state focus  4a, high (preference)  F(1, 302)  4.17**  7.65***  .55  p  .042  .006  .458  4a, high (impact)  F(1, 302)  12.32***  15.11***  .13  p  .001  < .001  .715  4a, low (preference)  F(1, 302)  2.12  < .001  1.91  p  .146  .986  .168  4a, low (impact)  F(1, 302)  2.62  .50  .77  p  .107  .481  .381  4b, high (motivation)  F(1, 189)  10.89***  9.67***  .09  p  .001  .002  .771  4b, high (impact)  F(1, 189)  3.25*  5.42**  .21  p  .073  .021  .646  NOTE.—Pairwise contrasts in each reference point focus condition of studies 4a and 4b: specific control versus nonspecific, specific control versus specific initial-state focus, nonspecific versus specific initial-state focus. As expected, in the high goal progress condition, the specific control contrasts emerged as significant in each case, whereas the nonspecific versus specific initial-state focus contrast did not. * p < .10, **p < .05, ***p < .01. Subjective Impact A 3 (reference point focus) × 2 (goal progress) ANOVA on subjective impact revealed a marginal main effect of reference point focus (F(2, 302) = 2.73, p = .07), qualified by the predicted interaction (F(2, 302) = 8.09, p < .001). There was no main effect of goal progress (F < 1). Consistent with studies 2 and 3, in the specific control condition, accumulating goal progress (i.e., putting $450 vs. $50 toward the loans) increased the subjective impact of marginal goal progress (Mlow = 4.36, Mhigh = 5.16; F(1, 302) = 8.33, p = .004). In the nonspecific goal condition, however, accumulating goal progress decreased the subjective impact of marginal goal progress (Mlow = 4.80, Mhigh = 4.17; F(1, 302) = 5.23, p = .020). Importantly, as expected, in the specific initial-state focus condition, accumulating goal progress also decreased the subjective impact of marginal goal progress (albeit marginally, F(1, 302) = 3.11, p = .080). When we encouraged specific goal pursuers to adopt the initial state as their reference point, as nonspecific goal pursuers do naturally, they perceived saving an additional $15 to have less of an impact on the overall goal after putting $450 (M = 4.07) versus $50 (M = 4.55) toward their loans (similar to those in the nonspecific goal condition). Also consistent with studies 2 and 3, when goal progress was high, participants in the nonspecific goal condition saw marginal goal progress as more impactful than did those in the specific control condition (Mnonspecific = 4.17, Mspecific-control = 5.16; F(1, 302) = 12.32, p = .001). This difference was eliminated, however, when specific goal pursuers focused on the initial state as the reference point: subjective impact was lower (M = 4.07) than in the specific control condition (F(1, 302) = 15.11, p < .001) and no different from the nonspecific condition (F < 1). When goal progress was low, subjective impact did not differ between the nonspecific goal (M = 4.80), specific control (M = 4.36), and specific initial-state focus conditions (M = 4.55). See table 2 for pairwise contrasts. Underlying Process As in the previous studies, we ran a bias-corrected bootstrapping mediated moderation analysis to examine the underlying process. Because we expected (and found) similar effects in the two conditions where the initial state was salient, these were combined for this analysis (effects are the same if each is separately compared to the specific control condition). Results supported our theory, revealing a significant index of mediated moderation (index: .44, 95% CI [.21 to .76]). In the specific control condition, where participants naturally focused on the end state as the reference point, accumulating goal progress increased motivation by making marginal goal progress seem more impactful (ab = .26, 95% CI [.10 to .48]). In the other two conditions, where participants focused on the initial state as the reference point, accumulating goal progress decreased motivation by making marginal progress seem less impactful (ab = –.18, 95% CI [–.37 to –.05]). Further, in the high goal progress condition, focusing on the initial state (nonspecific and specific initial-state focus conditions) was less motivating than focusing on the end state (specific control), because it made marginal goal progress seem less impactful (ab = .11, 95% CI [.06 to .19]). In the low goal progress condition, the indirect effect was not significant (ab = –.03, 95% CI [–.09 to .01]). Discussion Study 4a provides further insight into the underlying process by directly manipulating the focal reference point. When we encouraged specific goal pursuers to focus on the initial state as the reference point instead, their motivation (and judgments of subjective impact) no longer increased, but decreased with accumulated goal progress (like their nonspecific goal counterparts). Moreover, when current goal progress was high, specific goal pursuers focused on the initial state were less motivated than those in the specific control condition, like those in the nonspecific goal condition. Together these results provide direct evidence for the role of reference points (rather than other potential differences related to goal difficulty or goal completion) in shaping goal specificity’s effects. STUDY 4B Building on study 4a, study 4b further explored the role of reference point focus in determining how goal specificity shapes motivation. Following a similar paradigm to study 3, we manipulated whether specific goal pursuers focused on the initial-state (vs. end-state) reference point. If a difference in salient reference points underlies goal specificity’s effects, as our theory suggests, then encouraging specific goal pursuers to instead use the initial state as the reference point should attenuate the difference between nonspecific and specific goals. We tested this prediction at a high level of goal progress, where goal specificity produces the strongest divergence. Design and Method Participants (N = 192, average age = 22.95 years, 66.1% female) were recruited from a university behavioral lab in exchange for course credit. All recruited participants completed the study and all were included in the analyses. Participants were randomly assigned to a reference point focus condition: specific end-state focus, specific initial-state focus, or nonspecific. First, we manipulated goal specificity. Similar to study 3, in the two specific goal conditions, participants reported their current body weight and read that their goal was to “lose six pounds” from this starting weight. In the nonspecific goal condition, participants reported their current body weight and read that their goal was to “lose as much weight as you can” from this starting weight. Second, we provided high goal progress feedback. All participants read that a few weeks later, they weighed themselves again, and their current weight was five pounds less than their starting weight. The exact current weight value was automatically calculated for each participant based on his or her reported starting weight. Third, we manipulated the focal reference point. Similar to study 4a, participants viewed a progress bar (on a sheet of loose paper) with a dotted line indicating their current level of goal progress and an arrow either pointing left (toward the initial state) or right (toward the end state) (see appendix D). In the specific end-state focus condition, the arrow pointed to the right. In the specific initial-state focus and nonspecific conditions, the arrow pointed to the left. All participants were instructed to shade in the progress bar with a pencil, starting from the dotted line in the direction of the arrow. This encouraged them to compare their current goal progress to either the initial state or end state, depending on condition. Fourth, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much would losing an additional pound impact your weight loss goal?” (1 = No impact at all, 7 = Very large impact). Finally, we measured motivation. Participants answered the same two questions from study 4a, which we combined (r = .79). Results Motivation A one-way ANOVA on motivation revealed a significant effect (F(2, 189) = 6.87, p = .001). Consistent with study 4a and supporting our theory, at this high level of goal progress, participants in the nonspecific goal condition were less motivated than those in the specific end-state focus condition (Mnonspecific = 4.21, Mspecific end-state = 5.20; F(1, 189) = 10.89, p = .001). This difference was eliminated, however, when specific goal pursuers were encouraged to focus on the initial state instead (M = 4.29; vs. the specific end-state focus condition: F(1, 189) = 9.67, p = .002; vs. the nonspecific goal condition: F < 1). Subjective Impact A one-way ANOVA on the subjective impact of marginal goal progress also revealed a significant effect (F(2, 189) = 2.99, p = .053). Consistent with study 4a and supporting our theory, at this high level of goal progress, subjective impact was lower in the nonspecific versus the specific end-state focus condition (Mnonspecific = 3.97, Mspecific = 4.59; F(1, 189) = 3.25, p = .073), but this difference was eliminated in the specific initial-state focus condition (M = 3.81; the specific end-state focus condition: F(1, 189) = 5.42, p = .021; the nonspecific condition: F < 1). Underlying Process Similar to the previous studies, we ran a bias-corrected bootstrapping mediation analysis to examine the underlying process. Because we expected (and found) no difference between the two conditions where the initial state was salient, these were combined for this analysis (effects are the same if each is separately compared to the specific end-state focus condition). Results supported our reasoning: at this high level of goal progress, focusing on the initial state—regardless of whether the goal was nonspecific or specific—was less motivating than focusing on the end state, because it made marginal goal progress seem less impactful (ab = .13, 95% CI [.02 to .24]). Discussion Study 4b underscores the role of reference points in shaping goal specificity’s effects. When focused on the naturally more salient reference point (end state for specific goals and initial state for nonspecific goals), the nonspecific goal reduced subjective impact and motivation relative to the specific goal. When specific goal pursuers were directed to focus on the initial state as the reference point instead, however, this effect was attenuated. These findings support our theory that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit, and this difference in focal reference points underlies the documented effects of goal specificity on subsequent motivation. GENERAL DISCUSSION Nonspecific goals are both common and important in consumers’ lives. Yet despite considerable interest in the consequences of setting nonspecific (vs. specific) goals (Locke et al. 1989; Locke and Latham 1990; Naylor and Ilgen 1984; Soman and Cheema 2004; Ülkümen and Cheema 2011; Wright and Kacmar 1994), an understanding of how goal specificity shapes motivation during goal pursuit is more limited. To provide deeper insight into goal specificity’s effects, the current research developed a series of hypotheses that describe how goal specificity and goal progress jointly influence subsequent motivation. Our central proposition is that goal specificity alters what reference point consumers adopt during goal pursuit: for specific goals, the goal objective or specific end state serves as the focal reference point, but for nonspecific goals, which lack a specific end state, the initial state serves as the focal reference point. We argued that this difference in focal reference points has important consequences for (1) how accumulating goal progress shapes motivation to pursue nonspecific (vs. specific) goals, and (2) when (i.e., at what level of goal progress) nonspecific goals reduce (or increase) motivation relative to specific goals. Five studies supported our hypotheses. Across a variety of goal domains (task performance, debt repayment, weight loss), paradigms (lab tasks and realistic goal scenarios), and measures of motivation, consistent results emerged. First, for specific goals, accumulating goal progress increases subsequent motivation, but for nonspecific goals, accumulating goal progress decreases subsequent motivation (studies 1–4a, hypothesis 1). Second, nonspecific (vs. specific) goals are less motivating at higher levels of goal progress (studies 1–4b, hypothesis 2), but this difference is attenuated (and in some cases reversed) at lower levels of goal progress (studies 1–4a, hypothesis 2). Third, the subjective impact of marginal goal progress, which is determined by the shape of the value function, drives these effects (studies 2–4b; hypothesis 3). Our final two studies provided direct evidence for the role of reference points by manipulating whether specific goal pursuers focused on the initial state. When specific goal pursuers were encouraged to adopt the initial state as their reference point (as nonspecific goal pursuers do naturally), their motivation decreased with accumulated goal progress (study 4a) and they were no longer more motivated than nonspecific goal pursuers when current goal progress was high (studies 4a and 4b). These findings underscore that goal specificity’s effects rely on natural differences in reference point focus. Notably, our findings support the proposed roles of both diminishing sensitivity and loss aversion in determining how goal specificity shapes motivation. That the subjective impact of marginal goal progress (and motivation) decreased (increased) with accumulated goal progress for nonspecific (specific) goals underscores that proximity to one’s salient reference point influences motivation (diminishing sensitivity). Moreover, that goal specificity produced a greater effect on subjective impact and motivation at higher (vs. lower) goal progress levels underscores that whether one is below or above the salient reference point (and thus in losses or gains) influences motivation (loss aversion). Further support for the role of loss aversion comes from examining the intermediate level of goal progress. Based on our theory, when consumers’ current level of goal progress is equidistant from the initial-state and end-state reference points, nonspecific goals should tend to reduce motivation relative to specific goals. Because distance from the focal reference point is held constant, loss aversion, rather than diminishing sensitivity, should be the sole determinant of the subjective impact of marginal goal progress, and nonspecific (specific) goals should put people in losses (gains). A single-paper meta-analysis (McShane and Böckenholt 2017) on the intermediate progress level conditions of studies 1 and 2 supported this reasoning. When nonspecific and specific goal pursuers were equally far from their respective reference points, specific goal pursuers showed greater motivation (contrast = 0.23, SE = .11, p = .035).2 These results bolster empirical support for the proposed role of loss aversion in determining how goal specificity shapes motivation. Theoretical Contributions This research makes three main theoretical contributions. First, our findings inform the relationship between goal progress and motivation. A large body of research demonstrates that accumulating goal progress increases subsequent motivation (e.g., the “goal gradient” or “goal looms larger” effect; Hull 1932; Kivetz et al. 2006; Louro et al. 2007; Nunes and Drèze 2011; Soman and Shi 2003). More recently, a few articles have suggested that accumulating goal progress can both increase and decrease subsequent motivation, depending on whether the starting point (i.e., initial state) or ending point (i.e., end state) is salient (e.g., the “stuck in the middle” effect or “small area hypothesis”; Bonezzi et al. 2011; Carton et al. 2011; Koo and Fishbach 2012; Touré-Tillery and Fishbach 2012). Building on these findings, our research identifies goal specificity as a key determinant of the relationship between goal progress and motivation. By influencing what reference point consumers naturally adopt, goal specificity determines whether accumulating goal progress increases or decreases subsequent motivation. Second, this research advances understanding of how goal specificity shapes motivation. Goal specificity is known to influence many aspects of goal pursuit, including goal commitment and performance (Locke et al. 1989; Naylor and Ilgen 1984; Soman and Cheema 2004; Ülkümen and Cheema 2011; Wright and Kacmar 1994). Prior research has explained these effects by noting that nonspecific goals introduce ambiguity into how performance is evaluated (Locke and Latham 1990; Wright and Kacmar 1994). Yet, while this reasoning is consistent with the previously documented effects, it provides limited ability to predict how motivated consumers will be at specific points during goal pursuit (i.e., having accumulated different amounts of goal progress). The current research proposes that, beyond simply making performance evaluation more ambiguous, goal specificity fundamentally changes what reference point consumers adopt during goal pursuit, and that this difference in focal reference points determines how accumulating goal progress affects subsequent motivation. In addition, the current work informs previously documented advantages and disadvantages of specific goals relative to nonspecific goals. For example, prior work finds that setting specific (vs. nonspecific) goals tends to lead to better performance outcomes (Locke and Latham 1990; Locke et al. 1981). Consistent with this, we also find a performance advantage of specific goals relative to nonspecific goals, but show that this occurs primarily at higher levels of goal progress. Prior work also demonstrates that specific goals can exhibit a “starting problem,” such that goal pursuers are reluctant to take initial steps toward very distant goals (Heath et al. 1999), but nonspecific goals, which lack a specific end state, show this less. Consistent with this, we also find that specific (vs. nonspecific) goals can be disadvantageous at lower progress levels, because the value function is less steep. Third, this research generalizes the theory of goals as reference points beyond goals that have specific performance objectives. Since the seminal article introducing this framework (Heath et al. 1999), research has explored its consequences for decision making (Larrick et al. 2009; Medvec and Savitsky 1997) and behavior (Allen et al. 2016; Berger and Pope 2011; Bonezzi et al. 2011; Kivetz et al. 2006; Medvec, Madey, and Gilovich 1995; Pope and Simonsohn 2011), but focused less on further developing the theory. The current work contributes in two important ways: (1) by identifying goal specificity as a key determinant of what reference point consumers spontaneously rely on to evaluate their current goal progress, and (2) by specifying implications of both diminishing sensitivity and loss aversion for goal pursuers focused on a goal’s initial state (vs. end state) as the reference point. Notably, our work is also the first to empirically show that the subjective impact of marginal goal progress, determined by the shape of the value function, drives the effects of reference point focus on subsequent motivation. Prior work has speculated about this underlying process (Bonezzi et al. 2011; Heath et al. 1999; Kivetz et al. 2006; Koo and Fishbach 2012), but only measured its downstream effects on motivation and behavior. By eliciting explicit judgments of the subjective impact of marginal goal progress, the present studies provide direct support for its role in predicting motivation. Practical Implications The findings also have practical implications. For consumers, our work suggests that setting specific goals can lead to greater motivation (Locke et al. 1989; Locke and Latham 1990), but might not always. Early on in the pursuit of a specific goal, when reaching the goal objective is far off, focusing on the end state as the reference point may prove less motivating than focusing on the initial state. This may put specific goals at a disadvantage when consumers’ current level of goal progress is relatively low (e.g., after losing just a couple of pounds or repaying a small portion of a loan). Indeed, in our studies that examined lower levels of goal progress, the nonspecific goal was (at least directionally) more motivating than the specific goal in each case. Particularly for challenging specific goals, which have end states further from the initial state, consumers could thus benefit from deliberately adopting the initial state as the reference point early on or by breaking up their overall goal into smaller subgoals (Fishbach, Dhar, and Zhang 2006; Heath et al. 1999). For marketers, our findings suggest that effective strategies for motivating consumers to pursue specific goals may be ineffective or even harmful for nonspecific goals. Whereas marketers can encourage consumers to work toward specific goals (e.g., loyalty program rewards, product collections) by emphasizing earned progress or endowing unearned progress (Kivetz et al. 2006; Nunes and Drèze 2006; Zhang and Huang 2010), these strategies may backfire for nonspecific goals. For both individual nonspecific goals (e.g., reward programs) and group nonspecific goals (e.g., fundraising drives, petitions), perceiving greater goal progress may undermine motivation by making subsequent actions seem less impactful. Marketers may be able to overcome this demotivating effect by directly bolstering the subjective impact of marginal progress or by encouraging consumers to compare their current progress to external reference points (e.g., social comparisons, prior performance) rather than where they started. Future Research Directions This work suggests several interesting opportunities for future research. One is to investigate the effects of range goals (e.g., lose 10–15 pounds) on motivation. In past research, range goals have been treated as an intermediate level of goal specificity, falling between purely specific goals and nonspecific “do your best” goals (Locke et al. 1989; Naylor and Ilgen 1984; Scott and Nowlis 2013). However, from a reference points perspective, range goals differ from these goal types in that they contain two potential end-state reference points, as opposed to just one for specific goals and none for nonspecific goals. Future work could consider how consumers utilize the additional end-state reference point provided by range goals and explore its consequences for motivation. Another opportunity for further research is to more deeply explore other aspects of goal specificity. Implicit in the current theorizing is that goals provide a specific initial state that can be used as a reference point in the absence of an end state. Yet consumers can also pursue goals that have neither a specific initial state nor a specific end state, such as goals to “get in shape” or “get rich.” When pursuing such goals, consumers may draw potential reference points from alternative sources like round numbers (Pope and Simonsohn 2011) or resource constraints (March and Shapira 1992; Spiller 2011). Future research could further explore alternative sources of reference points during goal pursuit, as well as factors that affect how consumers direct their attention when multiple potential reference points are available. Finally, future work may wish to consider the role of goal difficulty more explicitly. To facilitate fair comparisons between nonspecific and specific goals, our studies used specific goals set at or below consumers’ natural aspiration level (see appendix A). How nonspecific goals compare to more challenging specific goals could also be interesting to test. Based on our theory, we expect that setting specific goals further from the starting point will exacerbate the effects of goal specificity: compared to the effects shown here, higher specific goals should more strongly enhance (reduce) motivation when goal progress is high (low). By exploring these and related phenomena, future research could build on our findings to further understanding of how goal specificity and salient reference points shape consumer motivation. DATA COLLECTION INFORMATION The first author collected pilot study data through Amazon Mechanical Turk in fall 2014. Both authors supervised data collection for studies 1, 4a, and 4b by the lab manager at the University of Pennsylvania’s Wharton Behavioral Lab in winter 2016, fall 2016, and spring 2015, respectively. The first author collected data for study 2 through Amazon Mechanical Turk in winter 2016. Both authors supervised data collection for study 3 by the lab manager at Duke University’s Fuqua Behavioral Lab in fall 2016. The first author conducted analyses for all studies. The authors thank Jonah Berger, Jim Bettman, Gavan Fitzsimons, and Rick Larrick for their helpful comments on prior versions of the manuscript. APPENDIX A GOAL CALIBRATION PRETESTS Calibration of Loan Payment Goal (Studies 2 and 4a) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 130, average age = 34.05 years, 40.8% female). They read the loan payment scenario from study 2, with any reference to the goals or goal progress omitted to avoid biasing responses. We asked participants how much debt they would aim to pay off in one month if they were setting a goal for themselves (open-ended in dollars). Results The average self-generated debt repayment goal was $565.77 (SD = 1015.72). This confirms that the specific goal assigned in studies 2 and 4a (pay off $500 of debt) is aligned with participants’ natural aspiration level and appropriately calibrated for the study. The average self-generated debt repayment goal was also greater than the high goal progress level ($450), indicating that participants in the nonspecific goal condition who received the high goal progress feedback were unlikely to infer that they had already achieved the goal. Calibration of Weight Loss Goal (Studies 3 and 4b) Method Pretest participants were recruited from a university behavioral lab in exchange for course credit (N = 27, average age = 22.93 years, 59.3% female). They read the weight loss scenario from study 3, with any reference to specific goals or goal progress omitted to avoid biasing responses. We asked participants how many pounds they would aim to lose in eight weeks if they were setting a goal for themselves (open-ended in pounds). Results The average self-generated weight loss goal was 9.58 pounds (SD = 7.07). This confirms that the specific goal assigned in studies 3 and 4b (lose six pounds) is below participants’ natural aspiration level and thus could not be artificially inflating their target. The average self-generated weight loss goal was also greater than the high goal progress level (4.5 pounds lost in study 3 and 5 pounds lost in study 4b), indicating that participants in the nonspecific goal condition who received the high progress feedback were unlikely to infer that they had already achieved the goal. APPENDIX B GOAL PROGRESS MANIPULATION PRETESTS Progress Pretest for Loan Payment Goals (Studies 2 and 4a) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 292, average age = 33.88 years, 37.3% female). Participants were randomly assigned to one condition in the same 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design used in study 2. To verify the effect of our progress manipulation, we measured goal progress perceptions using two measures: “At this point, how much progress would you feel you had made?” (1 = A little, 7 = A lot) and “At this point, how much money would you feel you had saved to put toward your loans for the month?” (1 = A little, 7 = A lot). These items were highly correlated (r = .89) and combined. Results A 2 (goal specificity) × 3 (goal progress) ANOVA on perceived goal progress revealed a main effect of goal specificity (F(1, 286) = 12.17, p < .001), such that, overall, participants in the specific goal condition perceived greater goal progress than did those in the nonspecific goal condition (Mnonspecific = 3.20 vs. Mspecific = 3.76). Importantly, this analysis also revealed the expected main effect of goal progress (F(2, 286) = 165.83, p < .001). Confirming that the manipulation worked as intended, in the specific goal condition, perceived goal progress significantly increased from the low to the intermediate goal progress condition (Mlow = 1.68 vs. Mintermediate = 3.88; F(1, 286) = 66.46, p < .001), and from the intermediate to the high goal progress condition (Mintermediate = 3.88 vs. Mhigh = 5.78; F(1, 286) = 48.66, p < .001); likewise, in the nonspecific goal condition, perceived goal progress significantly increased from the low to the intermediate goal progress condition (Mlow = 1.75 vs. Mintermediate = 3.30; F(1, 286) = 33.32, p < .001), and from the intermediate to the high goal progress condition (Mintermediate = 3.30 vs. Mhigh = 4.64; F(1, 286) = 23.98, p < .001). The 2 (goal specificity) × 3 (goal progress) ANOVA also revealed an interaction (F(2, 286) = 4.95, p = .008), simply reflecting a difference in the magnitude of the effect of the goal progress manipulation across goal specificity conditions. Most relevant to the present research, the pretest results demonstrate that the goal progress manipulation had the intended effect on goal progress perceptions in both goal specificity conditions. Progress Pretest for Weight Loss Goals (Study 3) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 243, average age = 34.00 years, 37.4% female). Participants were randomly assigned to one condition in the same 2 (goal specificity: specific, nonspecific) × 2 (goal progress: low, high) between-subjects design used in study 3. To verify the effect of our progress manipulation, we measured goal progress perceptions using two measures: “At this point, how much progress would you feel you had made?” (1 = A little, 7 = A lot) and “At this point, how much weight would you feel you had lost so far?” (1 = A little, 7 = A lot). These items were highly correlated (r = .84) and combined. Results A 2 (goal specificity) × 2 (goal progress) ANOVA on perceived goal progress revealed a main effect of goal specificity (F(1, 239) = 24.15, p < .001), such that, overall, participants in the specific goal condition perceived greater goal progress than did those in the nonspecific goal condition (Mnonspecific = 3.30 vs. Mspecific = 4.11). Importantly, this analysis also revealed the expected main effect of goal progress (F(1, 239) = 42.02, p < .001). Confirming that the manipulation worked as intended, the progress manipulation increased perceived progress in both the specific goal (Mlow = 3.46 vs. Mhigh = 4.99; F(1, 239) = 27.55, p < .001) and the nonspecific goal condition (Mlow = 2.65 vs. Mhigh = 3.79; F(1, 239) = 15.33, p < .001). There was no interaction between goal specificity and goal progress (F < 1). APPENDIX C REFERENCE POINT MANIPULATION STIMULI (STUDY 4A) Specific Control Condition: Specific Initial-State Focus Condition: NOTE.—Highlighted segments were filled in as participants selected them. Nonspecific Condition: APPENDIX D REFERENCE POINT MANIPULATION STIMULI (STUDY 4B) Specific End-State Focus Condition: Specific Initial-State Focus Condition: Nonspecific Condition: Footnotes 1 When current goal progress is low, the subjective impact of marginal goal progress (and motivation) will depend on the tension between diminishing sensitivity (which should favor nonspecific goals) and loss aversion (which should favor specific goals). If goal progress is sufficiently low (i.e., the distance from the focal reference point is sufficiently small) to outweigh the effect of being in gains (vs. losses), then nonspecific goals may in fact increase motivation relative to specific goals (figure 2). 2 This analysis uses the focal measures of motivation in each study (persistence in study 1 and WTP in study 2, both log-transformed). Due to the reversed coding in study 2 (i.e., lower WTP indicates higher motivation), cell means in that study were reflected around the grand mean. If we instead use the study 2 subjective impact measure, to avoid reverse coding, the focal effect is even stronger (contrast = 0.47, SE = .18, p = .011). REFERENCES Abeler Johannes, Falk Armin, Goette Lorenz, Huffman David ( 2011), “Reference Points and Effort Provision,” American Economic Review , 101 2, 470– 92. Google Scholar CrossRef Search ADS   Allen Eric J., Dechow Patricia M., Pope Devin G., Wu George ( 2016), “Reference-Dependent Preferences: Evidence from Marathon Runners,” Management Science , forthcoming. Amir On, Ariely Dan ( 2008), “Resting on Laurels: The Effects of Discrete Progress Markers as Subgoals on Task Performance and Preferences,” Journal of Experimental Psychology: Learning, Memory, and Cognition , 34 5, 1158– 71. Google Scholar CrossRef Search ADS PubMed  Barberis Nicholas C. ( 2013), “Thirty Years of Prospect Theory in Economics: A Review and Assessment,” Journal of Economic Perspectives,   27 1, 173– 95. Google Scholar CrossRef Search ADS   Berger Jonah, Pope Devin ( 2011), “Can Losing Lead to Winning?”  Management Science,   57 5, 817– 27. Google Scholar CrossRef Search ADS   Bonezzi Andrea, Brendl C. Miguel, De Angelis Matteo ( 2011), “Stuck in the Middle: The Psychophysics of Goal Pursuit,” Psychological Science,   22 5, 607– 12. Google Scholar CrossRef Search ADS   Carton Andrew, Larrick Richard P., Page L. ( 2011), “Back to the Grind: How Attention Affects Satisfaction during Goal Pursuit,” unpublished manuscript, Fuqua School of Business, Duke University, Durham, NC 27708. Etkin Jordan, Ratner Rebecca K. ( 2012), “The Dynamic Impact of Variety among Means on Motivation,” Journal of Consumer Research , 38 ( April), 1076– 92. Google Scholar CrossRef Search ADS   Fishbach Ayelet, Dhar Ravi, Zhang Ying ( 2006), “Subgoals as Substitutes or Complements: The Role of Goal Accessibility,” Journal of Personality and Social Psychology,   91 2, 232– 42. Google Scholar CrossRef Search ADS   Hayes Andrew F. ( 2013), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-based Approach , New York: Guilford Press. Heath Chip, Larrick Richard P., Wu George ( 1999), “Goals as Reference Points,”  Cognitive Psychology , 38 1, 79– 109. Google Scholar CrossRef Search ADS PubMed  Hollenbeck John R., Klein Howard J. ( 1987), “Goal Commitment and the Goal-Setting Process: Problems, Prospects, and Proposals for Future Research,” Journal of Applied Psychology , 74, 18– 23. Google Scholar CrossRef Search ADS   Huang Szu-chi, Zhang Ying, Broniarczyk Susan M. ( 2012), “So Near and Yet So Far: The Mental Representation of Goal Progress,”  Journal of Personality and Social Psychology,   103 2, 225– 41. Google Scholar CrossRef Search ADS   Hull Clark L. ( 1932), “The Goal-Gradient Hypothesis and Maze Learning,” Psychological Review , 39 1, 25– 43. Google Scholar CrossRef Search ADS   Kahneman Daniel ( 1992), “Reference Points, Anchors, Norms, and Mixed Feelings,” Organizational Behavior and Human Decision Processes , 51 2, 296– 312. Google Scholar CrossRef Search ADS   Kahneman Daniel, Tversky Amos ( 1979), “Prospect Theory: An Analysis of Decision under Risk,”  Econometrica: Journal of the Econometric Society , 47 2, 263– 91. Google Scholar CrossRef Search ADS   Kirschenbaum Daniel S., Humphrey Laura L., Malett Sheldon D. ( 1981), “Specificity of Planning in Adult Self-Control: An Applied Investigation,” Journal of Personality and Social Psychology , 40 5, 941– 50. Google Scholar CrossRef Search ADS   Kivetz Ran, Urminsky Oleg, Zheng Yuhuang ( 2006), “The Goal-Gradient Hypothesis Resurrected: Purchase Acceleration, Illusionary Goal Progress, and Customer Retention,”  Journal of Marketing Research,   43 1, 39– 58. Google Scholar CrossRef Search ADS   Klein Howard J., Whitener Ellen M., Ilgen Daniel R. ( 1990), “The Role of Goal Specificity in the Goal-Setting Process,” Motivation and Emotion , 14 3, 179– 93. Google Scholar CrossRef Search ADS   Koo Minjung, Fishbach Ayelet ( 2008), “Dynamics of Self-Regulation: How (Un)accomplished Goal Actions Affect Motivation,” Journal of Personality and Social Psychology , 94 2, 183– 95. Google Scholar CrossRef Search ADS PubMed  Koo Minjung, Fishbach Ayelet ( 2012), “The Small-Area Hypothesis: Effects of Progress Monitoring on Goal Adherence,” Journal of Consumer Research , 39 3, 493– 509. Google Scholar CrossRef Search ADS   Larrick Richard P., Heath Chip, Wu George ( 2009), “Goal-Induced Risk Taking in Negotiation and Decision Making,” Social Cognition , 27 3, 342– 64. Google Scholar CrossRef Search ADS   Locke Edwin A., Chah Dong-Ok, Harrison Scott, Lustgarten Nancy ( 1989), “Separating the Effects of Goal Specificity from Goal Level,” Organizational Behavior and Human Decision Processes , 43 2, 270– 87. Google Scholar CrossRef Search ADS   Locke Edwin A., Latham Gary P. ( 1990), A Theory of Goal Setting and Task Performance , Englewood Cliffs, NJ: Prentice-Hall. Locke Edwin A., Shaw Karyll N., Saari Lise M., Latham Gary P. ( 1981), “Goal Setting and Task Performance: 1960–1980,” Psychological Bulletin , 90 1, 125– 52. Google Scholar CrossRef Search ADS   Louro Maria J., Pieters Rik, Zeelenberg Marcel ( 2007), “Dynamics of Multiple-Goal Pursuit,”  Journal of Personality and Social Psychology,   93 2, 174– 93. Google Scholar CrossRef Search ADS   March James G., Shapira Zur ( 1992), “Variable Risk Preferences and the Focus of Attention,” Psychological Review , 99 1, 172– 83. Google Scholar CrossRef Search ADS   McShane Blakely B., Böckenholt Ulf ( 2017), “Single Paper Meta-Analysis: Benefits for Study Summary, Theory Testing, and Replicability,” Journal of Consumer Research , 43 6, 1048– 63. Medvec Victoria Husted, Madey Scott F., Gilovich Thomas ( 1995), “When Less Is More: Counterfactual Thinking and Satisfaction among Olympic Medalists,” Journal of Personality and Social Psychology , 69 4, 603– 10. Google Scholar CrossRef Search ADS PubMed  Medvec Victoria Husted, Savitsky Kenneth ( 1997), “When Doing Better Means Feeling Worse: The Effects of Categorical Cutoff Points on Counterfactual Thinking and Satisfaction,” Journal of Personality and Social Psychology , 72 6, 1284– 96. Google Scholar CrossRef Search ADS   Naylor James C., Ilgen Daniel R. ( 1984), “Goal Setting: A Theoretical Analysis of a Motivational Technology,” Research in Organizational Behavior , 6, 95– 140. Nunes Joseph C., Drèze Xavier ( 2006), “The Endowed Progress Effect: How Artificial Advancement Increases Effort,” Journal of Consumer Research , 32 4, 504– 12. Google Scholar CrossRef Search ADS   Pope Devin, Simonsohn Uri ( 2011), “Round Numbers as Goals: Evidence from Baseball, SAT Takers, and the Lab,” Psychological Science,   22 1, 71– 9. Google Scholar CrossRef Search ADS   Scott Maura L., Nowlis Stephen M. ( 2013), “The Effect of Goal Specificity on Consumer Goal Reengagement,”  Journal of Consumer Research,   40 3, 444– 59. Google Scholar CrossRef Search ADS   Soman Dilip, Cheema Amar ( 2004), “When Goals Are Counterproductive: The Effects of Violation of a Behavioral Goal on Subsequent Performance,” Journal of Consumer Research,   31 1, 52– 62. Google Scholar CrossRef Search ADS   Soman Dilip, Shi Mengze ( 2003), “Virtual Progress: The Effect of Path Characteristics on Perceptions of Progress and Choice” Management Science , 49 9, 1229– 50. Google Scholar CrossRef Search ADS   Spiller Stephen A. ( 2011), “Opportunity Cost Consideration,” Journal of Consumer Research,   38 4, 595– 610. Google Scholar CrossRef Search ADS   Touré-Tillery Maferima, Fishbach Ayelet ( 2012), “The End Justifies the Means, but Only in the Middle,” Journal of Experimental Psychology: General , 141 3, 570– 83. Google Scholar CrossRef Search ADS PubMed  Tversky Amos, Kahneman Daniel ( 1991), “Loss Aversion in Riskless Choice: A Reference-Dependent Model,” Quarterly Journal of Economics , 106 4, 1039– 61. Google Scholar CrossRef Search ADS   Ülkümen Gülden, Cheema Amar ( 2011), “Framing Goals to Influence Personal Savings: The Role of Specificity and Construal Level,”  Journal of Marketing Research,   48 6, 958– 69. Google Scholar CrossRef Search ADS   Wright Patrick M., Kacmar K. Michele ( 1994), “Goal Specificity as a Determinant of Goal Commitment and Goal Change,” Organizational Behavior and Human Decision Processes , 59 2, 242– 60. Google Scholar CrossRef Search ADS   Wu George, Heath Chip, Larrick Richard P. ( 2008), “A Prospect Theory Model of Goal Behavior,” working paper, Graduate School of Business, University of Chicago, Chicago, IL 60637. Zhang Ying, Huang Szu-Chi ( 2010), “How Endowed versus Earned Progress Affects Consumer Goal Commitment and Motivation,” Journal of Consumer Research , 37 4, 641– 54. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Consumer Research Oxford University Press

How Goal Specificity Shapes Motivation: A Reference Points Perspective

Loading next page...
 
/lp/ou_press/how-goal-specificity-shapes-motivation-a-reference-points-perspective-HRBgAg77gr
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
0093-5301
eISSN
1537-5277
D.O.I.
10.1093/jcr/ucx082
Publisher site
See Article on Publisher Site

Abstract

Consumers often pursue goals that lack specific end states, such as goals to lose as much weight as possible or to pay off as much debt as possible. Yet despite considerable interest in the consequences of setting nonspecific (vs. specific) goals, how goal specificity affects motivation throughout goal pursuit is less well understood. The current research explores the role of reference points in shaping goal specificity’s effects. We propose that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit: for specific goals, the end state tends to be more salient, but for nonspecific goals, the initial state should be more salient. Five studies investigate how this difference in focal reference points shapes (1) the relationship between goal progress and motivation, (2) when (i.e., at what level of goal progress) goal specificity produces the greatest difference in motivation, and (3) the underlying process driving these effects. Our findings advance understanding of the relationship between goal specificity, goal progress, and motivation, and in doing so, underscore the critical role that reference points play in goal-directed behavior. In addition, the findings offer practical insight into how best to set important financial, health, and other consumer goals to enhance motivation. goals, reference points, motivation The relationship between goal progress and motivation is one of the most robust and well-known findings in the goal pursuit literature (Hull 1932; Kivetz, Urminsky, and Zheng 2006; Louro, Pieters, and Zeelenberg 2007; Nunes and Drèze 2011; Soman and Shi 2003). Often called the “goal gradient” or “goal looms larger” effect, accumulating progress toward a goal tends to make consumers more motivated to pursue it. Scholars have described this phenomenon as “the main insight from classic and modern research on motivation” (Koo and Fishbach 2012). A prominent explanation for this effect comes from the theory of goals as reference points (Heath, Larrick, and Wu 1999). This theory posits that the desired end state of a goal serves as a reference point during goal pursuit, producing a “value function” (Kahneman and Tversky 1979) that drives motivation as a function of distance to the goal end state. Because the value function is steeper closer to the reference point, as consumers accumulate goal progress (i.e., grow closer to the goal’s end state), each unit of marginal goal progress is perceived to have a greater impact on the overall goal, and this increases subsequent motivation. For example, a dieter with a goal to lose six pounds will be more motivated to lose the next pound when he has lost four pounds versus two pounds so far, because he is on a steeper part of the value function (i.e., closer to the goal end state) and therefore sees losing the next pound as more impactful. But what about goals that lack specific end states? While this “reference points” explanation assumes that goals are defined by a specific end state, many consumer goals are not. Rather than striving to lose six pounds, for example, dieters may simply try to lose as much weight as possible, and rather than aiming to pay off $500 of debt, consumers may simply try to pay off as much debt as possible. Such nonspecific “do your best” goals are both common and important. When we asked US adults (N = 149, 19 to 82 years, mean age 35.14 years, 60.8% male) to list a series of personal goals and note whether each was associated with a specific end state, half of the listed goals were nonspecific (i.e., 611 out of 1,188 goals lacked a specific performance objective). Participants also viewed these nonspecific goals as equally important (1 = Not important at all, 7 = Extremely important) as their specific goals (Mnonspecific = 5.73 vs. Mspecific = 5.78; t < 1). How does goal specificity shape motivation during goal pursuit? What effect might the absence of a specific end state have on the relationship between goal progress and motivation? What role might reference points play in goal specificity’s effects? The present research examines these questions. We propose that, lacking a specific end state, nonspecific goal pursuers will use the initial state (i.e., where goal pursuit began) as the reference point instead. Drawing on the value function’s features of diminishing sensitivity and loss aversion (Kahneman and Tversky 1979), we develop a series of hypotheses that describe how this difference in focal reference points shapes the relationship between goal progress and motivation. We first consider how accumulating goal progress affects motivation to pursue nonspecific (vs. specific) goals. Then, we examine when (i.e., at what level of goal progress) goal specificity produces the greatest difference in motivation. Finally, we explore the underlying mechanism driving these effects. The findings make three main contributions. First, this research furthers understanding of the relationship between goal progress and motivation. While a substantial body of work shows that accumulating goal progress can increase motivation (Hull 1932; Kivetz et al. 2006; Nunes and Drèze 2011; Soman and Shi 2003), our findings provide a more nuanced perspective: whether accumulating goal progress increases or decreases subsequent motivation critically depends on goal specificity (i.e., the presence of an end-state reference point). Second, this work furthers understanding of how goal specificity shapes motivation. Whereas goal specificity’s effects have previously been attributed to ambiguity in how performance is evaluated (Naylor and Ilgen 1984; Wright and Kacmar 1994), we introduce a theoretical framework that predicts how motivation to pursue nonspecific versus specific goals differs as a function of salient reference points. Third, this work generalizes the theory of goals as reference points beyond goals that have specific performance objectives. Whereas previous tests of the existing framework have exclusively considered goals that provide an end-state reference point (Bonezzi, Brendl, and De Angelis 2011; Heath et al. 1999; Koo and Fishbach 2012), we develop and test novel predictions for goals that do not (i.e., nonspecific goals). The findings underscore that goal specificity plays a key role in determining what reference points consumers adopt during goal pursuit. GOAL SPECIFICITY Goal specificity is a defining characteristic of consumer goals. Unlike specific goals, nonspecific goals have some degree of “ambiguity or diffuseness in the exact level of performance required” (Hollenbeck and Klein 1987; Naylor and Ilgen 1984; Wright and Kacmar 1994). Whereas specific goals define a desired end-state objective (e.g., lose six pounds, pay off $500 of debt), nonspecific goals do not (e.g., lose as much weight as possible, pay off as much debt as possible). Nonspecific goals can take different forms (e.g., range goals, Scott and Nowlis 2013), but the most common “do your best” type of nonspecific goal lacks an end state entirely (Locke and Latham 1990; Wright and Kacmar 1994). Setting nonspecific versus specific goals has a variety of consequences. Prior work finds that nonspecific (vs. specific) goals are perceived as less difficult and more attainable (Ülkümen and Cheema 2011), which encourages people to adopt them more readily (Locke and Latham 1990; Naylor and Ilgen 1984). Nonspecific (vs. specific) goals are also less likely to evoke feelings of failure, which reduces goal abandonment (Kirschenbaum, Humphrey, and Malett 1981; Soman and Cheema 2004). People also tend to feel less committed to nonspecific (vs. specific) goals (Hollenbeck and Klein 1987; Naylor and Ilgen 1984). This makes nonspecific goals more likely to be revised (Wright and Kacmar 1994), creates greater variability in performance outcomes (Klein, Whitener, and Ilgen 1990; Locke et al. 1989), and can lead to worse performance overall (Locke and Latham 1990; Locke et al. 1981). To explain these prior findings, researchers have argued that the absence of a specific end state introduces ambiguity into how performance is evaluated (Naylor and Ilgen 1984; Wright and Kacmar 1994). Because for nonspecific goals, the goal objective is less precisely defined, a broader range of outcomes can constitute success. For instance, whereas for a goal to lose six pounds, only that single outcome (losing six pounds) would achieve the goal, for a goal to lose as much weight as possible, multiple outcomes (e.g., losing four, six, or eight pounds) could potentially seem sufficient. While this reasoning helps explain the previously documented effects, it offers limited insight into how goal specificity shapes motivation during goal pursuit. A growing body of research reveals goal pursuit to be a dynamic process in which motivation changes as consumers accumulate goal progress (Amir and Ariely 2008; Etkin and Ratner 2012; Huang, Zhang, and Broniarczyk 2012; Kivetz et al. 2006; Koo and Fishbach 2008, 2012). For nonspecific (vs. specific) goals, how motivated will consumers be after accumulating different amounts of goal progress? If a dieter has a goal to lose as much weight as possible, for instance, how motivated would he be to lose more weight having lost two versus four (vs. six, etc.) pounds so far? And for a given level of goal progress (e.g., four pounds lost), how would motivation differ if, rather than lose as much weight as possible, the dieter’s goal was instead to lose six pounds exactly? To address these questions and provide deeper insight into goal specificity’s effects, the current research develops a theoretical framework that describes how the absence of a specific end state influences motivation during goal pursuit. Central to our theorizing is the notion of reference points. GOAL SPECIFICITY: A REFERENCE POINTS APPROACH We propose that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit, and that this difference in focal reference points has important implications for the relationship between goal progress and motivation. A “reference point” divides the space of outcomes into regions of gain and loss (Kahneman and Tversky 1979). Outcomes above the reference point are evaluated as gains and outcomes below the reference point are evaluated as losses. The valuation of these gains and losses varies systematically based on the slope of Prospect Theory’s value function, which is steeper closer to the reference point (i.e., diminishing sensitivity) and steeper on the loss side than on the gain side of the reference point (i.e., loss aversion, Kahneman and Tversky 1979). Differences between outcomes along a steeper part of the value function have a greater influence on subsequent decisions (Kahneman 1992; Larrick, Heath, and Wu 2009; Tversky and Kahneman 1991). While the notion of reference points has long been established, more recent research has attempted to understand where reference points originate (Abeler et al. 2011; Allen et al. 2016; Barberis 2013). One important source is consumers’ goals. The theory of goals as reference points (Heath et al. 1999) posits that the desired end state of a goal serves as the reference point during goal pursuit, and goal-related outcomes (i.e., levels of goal progress) are evaluated relative to that end state. For instance, if a dieter has a goal to lose six pounds, the dieter’s reference point will be the goal objective (six pounds lost) and he will evaluate his current goal progress (e.g., four pounds lost so far) relative to that desired end state. We propose that, absent a specific end state to serve as a reference point, nonspecific goal pursuers will use the initial state (i.e., where goal pursuit began) instead. Recent work finds that, in addition to the end state, the initial state of a specific goal can also serve as a reference point (Bonezzi et al. 2011; Carton et al. 2011; Koo and Fishbach 2008, 2012; Touré-Tillery and Fishbach 2012). For a goal to lose six pounds, for instance, either the end state (i.e., the six-pound goal objective) or the initial state (i.e., the dieter’s previous weight, or zero pounds lost) could serve as the reference point. While the end state is naturally more salient for specific goals (Heath et al. 1999; Kivetz et al. 2006), incidental factors that make the initial state more salient (e.g., goal progress feedback, Koo and Fishbach 2012; visual cues, Bonezzi et al. 2011) can encourage people to adopt it as the reference point instead. Because the absence of an end state should make the initial state more salient, we argue that nonspecific goal pursuers will spontaneously adopt the initial state as the focal reference point. CONSEQUENCES FOR MOTIVATION We propose that this difference in focal reference points plays a key role in how goal specificity shapes motivation. According to the theory of goals as reference points (Heath et al. 1999), the slope of the value function determines motivation by changing people’s subjective valuation of the impact of marginal goal progress (i.e., the “next step” of goal progress). Because the value function is nonlinear, the same objective increase in goal progress (e.g., losing one more pound) can be perceived as contributing more or less to the overall goal (e.g., lose six pounds). When one’s current goal progress falls on a steeper part of the value function, marginal goal progress seems more impactful. By determining the shape of the value function, salient reference points influence the subjective impact of marginal goal progress, and thus motivation. As previously discussed, diminishing sensitivity makes the value function steeper when one’s current state is closer to the reference point, and loss aversion makes the value function steeper when one’s current state is below the reference point (i.e., on the loss rather than the gain side of the value function). Consequently, because marginal goal progress seems more impactful when the value function is steeper, motivation is higher when consumers’ current goal progress puts them closer to their focal reference point or on the loss side of that reference point (Bonezzi et al. 2011; Heath et al. 1999; Koo and Fishbach 2012). We argue that goal specificity influences the shape of the value function, and thus changes how accumulating goal progress affects subsequent motivation. For specific goals, diminishing sensitivity should make the value function steeper closer to the (more salient) end state (Heath et al. 1999). Consequently, as consumers accumulate goal progress, they move closer to their focal reference point (and onto a steeper part of the value function), which makes marginal goal progress seem more impactful and increases subsequent motivation (i.e., the “goal gradient” effect, Kivetz et al. 2006). For nonspecific goals, however, diminishing sensitivity should make the value function steeper closer to the (more salient) initial state. Consequently, as consumers accumulate goal progress, they move further away from their focal reference point (and onto a shallower part of the value function). This should make marginal goal progress seem less impactful and therefore decrease subsequent motivation. For example, the dieter with a goal to lose as much weight as possible should see losing the next pound as having less of an impact on his overall weight loss goal, and thus be less motivated to lose more weight, after having lost four pounds (further from zero) versus two pounds (closer to zero) so far. For nonspecific goals, we thus predict a reverse goal gradient: accumulating goal progress will decrease subsequent motivation, driven by a decrease in the subjective impact of marginal goal progress. Our reasoning thus far describes a crossover interaction between goal specificity and goal progress (figure 1): for specific goals, motivation starts low (far from the focal end-state reference point) and increases with accumulated goal progress; for nonspecific goals, motivation starts high (near the focal initial-state reference point) and decreases with accumulated goal progress. This suggests that when goal progress is relatively high, nonspecific goals should be less motivating than specific goals, but when goal progress is relatively low, nonspecific goals should be more motivating than specific goals. FIGURE 1 View largeDownload slide PREDICTED EFFECTS OF GOAL SPECIFICITY AND GOAL PROGRESS ON SUBSEQUENT MOTIVATION NOTE.—Due to diminishing sensitivity, accumulating goal progress decreases (increases) motivation to pursue nonspecific (specific) goals. Due to loss aversion, the effect of goal specificity on motivation is greater at higher (vs. lower) levels of goal progress. FIGURE 1 View largeDownload slide PREDICTED EFFECTS OF GOAL SPECIFICITY AND GOAL PROGRESS ON SUBSEQUENT MOTIVATION NOTE.—Due to diminishing sensitivity, accumulating goal progress decreases (increases) motivation to pursue nonspecific (specific) goals. Due to loss aversion, the effect of goal specificity on motivation is greater at higher (vs. lower) levels of goal progress. Rather than a symmetrical crossover, however, we argue that loss aversion will produce an asymmetry in this interaction (figure 1). Whereas focusing on the end state locates current goal progress on the loss side of the value function (e.g., $250 below a savings goal of $500), focusing on the initial state locates current goal progress on the gain side (e.g., $250 above a starting point of $0). Because loss aversion makes losses steeper than gains, for a given level of goal progress, focusing on the end state (vs. initial state) as the reference point should put consumers on a steeper part of the value function. Together with the effect of diminishing sensitivity, this suggests that the value function should be at its steepest (shallowest) when current goal progress is both close to (far from) the focal reference point and on the loss (gain) side of that reference point. Consequently, with respect to goal specificity, the value function should be steepest for specific goals at high goal progress (loss side of the value function, close to the end-state reference point) and shallowest for nonspecific goals at high goal progress (gain side of the value function, far from the initial-state reference point); see figure 2. At low goal progress, the effects of diminishing sensitivity and loss aversion should act in opposition, with a net result of more moderate motivation for both specific and nonspecific goals (figure 2). FIGURE 2 View largeDownload slide POSITION RELATIVE TO FOCAL REFERENCE POINT NOTE.—Goal specificity interacts with accumulated goal progress to shape motivation via the slope of the value function. Motivation is greater when current goal progress is closer to (vs. further from) the focal reference point and when it is on the loss (vs. gain) side of the value function (i.e., below vs. above the reference point). FIGURE 2 View largeDownload slide POSITION RELATIVE TO FOCAL REFERENCE POINT NOTE.—Goal specificity interacts with accumulated goal progress to shape motivation via the slope of the value function. Motivation is greater when current goal progress is closer to (vs. further from) the focal reference point and when it is on the loss (vs. gain) side of the value function (i.e., below vs. above the reference point). We thus predict that goal specificity will produce a greater difference in the subjective impact of marginal goal progress (and thus motivation) at higher (vs. lower) levels of goal progress. In particular, when goal progress is high, nonspecific goals should decrease subjective impact (and motivation) relative to specific goals, but when goal progress is low, these effects should be attenuated.1 In summary, we predict: H1: For nonspecific goals (specific goals), accumulating goal progress decreases (increases) subsequent motivation. H2: When current goal progress is high, nonspecific goals reduce motivation relative to specific goals, but this effect is attenuated when current goal progress is low. H3: These effects are driven by the subjective impact of marginal goal progress. Five studies tested our hypotheses. Study 1 used an effortful lab task to examine how goal specificity shapes motivation. Studies 2 and 3 used realistic scenarios in important consumer goal domains (debt repayment in study 2; weight loss in study 3) to provide more controlled tests of our motivation predictions and examine the proposed underlying role of the subjective impact of marginal goal progress. Studies 4a and 4b further tested the proposed underlying process by directly manipulating the focal reference point. Together the findings show that how goal specificity shapes the dynamics of motivation depends on the different reference points that nonspecific (vs. specific) goals make salient. STUDY 1 Study 1 tests our first two hypotheses by examining effort on a goal-directed task: proofreading passages of text. We manipulated goal specificity and then measured motivation (i.e., persistence) at different points throughout the task. In the specific goal condition, we predicted that accumulating goal progress would increase subsequent motivation: after finding a greater number of errors, participants should work harder to find additional errors. In the nonspecific goal condition, however, we predicted that accumulating goal progress would instead decrease subsequent motivation: after finding a greater number of errors, participants should work less hard to find more errors. Further, due to loss aversion, we predicted that the nonspecific goal would be less motivating than the specific goal at the highest level of goal progress, but this effect would be reduced at lower progress levels. Design and Method Participants (N = 155) were recruited from a university behavioral lab in exchange for course credit. In this and subsequent lab studies, lab capacity and participant availability determined the sample size. Ten individuals (6%) reported technical problems completing the study (e.g., failure to load a page) and were excluded from the analyses, leaving a sample of 145 (average age = 24.67 years, 59% female). Participants were randomly assigned to one condition of a 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design. Participants read that they would be proofreading a series of short text passages and that there was one spelling error in each passage. In the specific goal condition, we told participants that their goal was to “find 10 errors in a row.” In the nonspecific goal condition, we told participants that their goal was to “find as many errors as possible in a row.” All participants read that if they failed to find the error in a given passage, their streak would end, and they would not be able to restart their streak or revisit the failed passage. After completing a practice passage, participants began the main proofreading task. Participants proceeded through the proofreading task as instructed, and were given a running count of how many errors they had found so far (which equaled the number of passages they had completed). After finding two (low progress condition), five (intermediate progress condition), or eight errors (high progress condition), we paused the task. We told participants that the remaining proofreading passages would be more difficult, and that if they failed to find the spelling error in one of the passages, they could quit the task (and end their streak). Then, participants returned to the main task, and we measured motivation. The next (target) text passage contained no spelling errors, meaning that in order to advance beyond this page, all participants eventually had to quit. We recorded how long participants persisted (i.e., how much effort they invested) in trying to find the error before quitting. Persistence time was log-transformed for analysis to correct for non-normality (Kolmogorov-Smirnov test statistic: .11, p < .01); raw means are reported for ease of interpretation. Results A 2 (goal specificity) × 3 (goal progress) ANOVA on motivation (i.e., persistence time) revealed a main effect of goal specificity (Mspecific = 146.53, Mnonspecific = 105.93, F(1, 139) = 7.24, p = .008), qualified by the predicted interaction (F(2, 139) = 6.19, p = .003; figure 3). There was no main effect of goal progress (F < 1). FIGURE 3 View largeDownload slide GOAL SPECIFICITY AFFECTS PROOFREADING EFFORT FIGURE 3 View largeDownload slide GOAL SPECIFICITY AFFECTS PROOFREADING EFFORT As expected, in the specific goal condition, accumulating goal progress increased subsequent motivation (linear contrast: F(1, 139) = 4.90, p = .028). Participants expended more effort on the target passage (i.e., worked harder to find the nonexistent error) after finding five errors (M = 140.56 sec) versus two errors (M = 117.35 sec), and after finding eight errors (M = 188.15 sec) versus five errors. However, supporting hypothesis 1, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased subsequent motivation (linear contrast: F(1, 139) = 7.78, p = .006). Participants expended less effort on the target passage after finding five errors (M = 105.71 sec) versus two errors (M = 136.08 sec), and after finding eight errors (M = 78.11 sec) versus five errors. See table 1 for pairwise contrasts. Table 1 Pairwise Contrasts between Goal Progress Conditions     Specific goal   Nonspecific goal   Study (DV)    Low vs. middle  Middle vs. high  Low vs. high  Low vs. middle  Middle vs. high  Low vs. high  1 (Log time)  F(1, 139)  2.17  .77  4.90**  1.69  1.72  7.78***  p  .143  .381  .028  .196  .192  .006  2 (Log WTP)  F(1, 301)  1.64  3.54*  8.83***  6.50**  .91  12.65***  p  .202  .061  .003  .011  .341  < .001  2 (Impact)  F(1, 301)  2.29  1.43  6.48**  7.88***  0.50  12.70***  p  .132  .232  .011  .005  .479  < .001      Specific goal   Nonspecific goal   Study (DV)    Low vs. middle  Middle vs. high  Low vs. high  Low vs. middle  Middle vs. high  Low vs. high  1 (Log time)  F(1, 139)  2.17  .77  4.90**  1.69  1.72  7.78***  p  .143  .381  .028  .196  .192  .006  2 (Log WTP)  F(1, 301)  1.64  3.54*  8.83***  6.50**  .91  12.65***  p  .202  .061  .003  .011  .341  < .001  2 (Impact)  F(1, 301)  2.29  1.43  6.48**  7.88***  0.50  12.70***  p  .132  .232  .011  .005  .479  < .001  NOTE.—Pairwise contrasts in each goal specificity condition of studies 1 and 2: low versus intermediate goal progress, intermediate versus high goal progress, and low versus high goal progress. As expected, the low versus high goal progress contrast emerged as significant in each case. * p < .10, **p < .05, ***p < .01. Further, supporting hypothesis 2, when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. After finding eight errors, participants in the nonspecific goal condition expended less effort on the target passage than did those in the specific goal condition (Mnonspecific = 78.11 sec, Mspecific = 188.15 sec; F(1, 139) = 15.21, p < .001). This effect was reduced, however, at intermediate progress (five errors, Mnonspecific = 105.71 sec, Mspecific = 140.56 sec; F(1, 139) = 3.00, p = .085) and directionally reversed at low goal progress (two errors, Mnonspecific = 136.08 sec, Mspecific = 117.35 sec; F(1, 139) = 1.06, p = .305). Discussion Study 1 supports our first two hypotheses with effortful behavior on a goal-directed task. Consistent with prior work (Heath et al. 1999; Kivetz et al. 2006; Nunes and Drèze 2011), when participants had a specific goal, accumulating goal progress increased subsequent motivation. The more progress participants had made in the proofreading task (i.e., the greater the number of errors they had found so far), the more effort they expended on the target passage. Importantly, as predicted (hypothesis 1), when participants had a nonspecific goal, the opposite occurred: accumulating goal progress decreased subsequent motivation. The more progress participants had made in the task, the less effort they expended on the target passage. Also as predicted (hypothesis 2), when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. This effect was attenuated (and directionally reversed), however, at the lower goal progress levels. Because high goal progress is where the effect of loss aversion reinforces that of diminishing sensitivity (figure 2), this is where goal specificity had the biggest effect on subsequent motivation. STUDY 2 Study 2 tests the proposed underlying process (hypothesis 3) in a common and important goal domain: debt repayment. We manipulated the specificity of a debt repayment goal, provided participants with goal progress feedback, and then, in addition to measuring motivation, asked participants to rate the impact of an incremental step of goal progress (saving an additional $25) on their overall debt repayment goal. In this and subsequent studies, we use realistic goal scenarios to provide more controlled tests of our predictions. Although study 1 held the (perceived) rate of goal progress constant, the overall amount of effort invested prior to reaching the target passage differed by condition. Consequently, in the nonspecific goal condition, depletion may have played a role in why accumulating goal progress decreased subsequent motivation. By manipulating goal progress independently of actual effort investment, scenario-based paradigms eliminate this potential confound, allowing us to more precisely test the proposed underlying mechanism. This approach is consistent with prior research on goals as reference points, which has extensively used scenario-based paradigms (Heath et al. 1999) and found them to show the same effects as real behavior (see Wu, Heath, and Larrick 2008 for a review). Design and Method Participants (N = 320) were recruited from Amazon Mechanical Turk in exchange for small payment. In this web-based study, a target rule of 50–60 participants per condition determined the sample size. Thirteen individuals (4%) initiated the study but failed to complete it, leaving a sample of 307 responses for analysis (average age = 35.04 years, 44.0% female). Due to the online format, we were unable to ask participants their reasons for exiting, but attrition did not differ across conditions. Participants were randomly assigned to one condition in a 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design. First, we manipulated goal specificity. We asked participants to imagine they had $10,000 in loans to pay off over time, and that they decided to pay down their loans faster by cutting back on spending. In the specific goal condition, participants read that their goal this month was to “pay off an extra $500.” A pretest conducted with a separate sample from the same population confirmed that this amount ($500) was comparable to what people would naturally plan to pay off in a month (see appendix A for details). In the nonspecific goal condition, participants read that their goal this month was to “pay off as much extra as you can.” Second, we provided goal progress feedback. Participants read that partway through the month, they were planning to go out to dinner with a friend, and that so far this month they had saved $50 (low progress condition), $250 (intermediate progress condition), or $450 (high progress condition) to put toward their loans. A pretest conducted with a separate sample from the same population confirmed that in both goal specificity conditions, perceived goal progress increased from $50 to $250 and from $250 to $500 (see appendix B for details). Third, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much of an impact would saving an additional $25 have on helping you reach your goal for the month?” (1 = No impact at all, 7 = Very large impact). Finally, we measured motivation. We reasoned that the more motivated participants were to put money toward their debt repayment goal, the less money they should be willing to spend on dinner with their friend. Accordingly, we asked them, “How much money would you be willing to spend on this dinner with your friend?” (open-ended in dollars), where larger values indicated lower motivation to conserve money. Willingness to pay was log-transformed for analysis to correct for non-normality (Kolmogorov-Smirnov test statistic: .23, p < .01); raw means are reported for ease of interpretation. Results Motivation A 2 (goal specificity) × 3 (goal progress) ANOVA on motivation (i.e., willingness to pay) revealed a marginal main effect of goal specificity (Mnonspecific = $30.83, Mspecific = $26.75; F(1, 301) = 3.25, p = .073) qualified by the predicted interaction (F(2, 301) = 10.76, p < .001; figure 4). There was no main effect of goal progress (F < 1). FIGURE 4 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO PAY OFF DEBT NOTE.—Willingness to pay for dinner corresponds to lower motivation to conserve money. FIGURE 4 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO PAY OFF DEBT NOTE.—Willingness to pay for dinner corresponds to lower motivation to conserve money. Consistent with study 1, in the specific goal condition, accumulating goal progress increased subsequent motivation (linear contrast: F(1, 301) = 8.83, p = .003). Participants were more motivated to conserve money (i.e., willing to spend less money on dinner) after putting $250 (M = $26.92) versus $50 (M = $31.90) toward their loans, and after putting $450 (M = $21.37) versus $250 toward their loans. However, supporting hypothesis 1, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased subsequent motivation (linear contrast: F(1, 301) = 12.65, p < .001). Participants were less motivated to conserve money (i.e., willing to spend more money on dinner) after putting $250 (M = $32.13) versus $50 (M = $23.59) toward their loans, and after putting $450 (M = $37.27) versus $250 toward their loans. See table 1 for pairwise contrasts. Further, consistent with study 1 and supporting hypothesis 2, when goal progress was high, the nonspecific goal reduced motivation relative to the specific goal. After paying off $450, participants in the nonspecific goal condition were less motivated to conserve money (i.e., willing to spend more on dinner) than those in the specific goal condition (Mnonspecific = $37.27, Mspecific = $21.37; F(1, 301) = 16.71, p < .001). This effect was reduced, however, at intermediate goal progress ($250; Mnonspecific = $32.13, Mspecific = $26.92; F(1, 301) = 2.04, p = .154), and it reversed (although the effect was smaller, consistent with our theory) at low goal progress ($50; Mnonspecific = $23.59, Mspecific = $31.90; F(1, 301) = 5.89, p = .016). Subjective Impact A 2 (goal specificity) × 3 (goal progress) ANOVA on subjective impact revealed only the predicted interaction (F(2, 314) = 11.54, p < .001). There was no main effect of goal specificity (F(1, 301) = 2.51, p = .114) or goal progress (F < 1). As expected, in the specific goal condition, accumulating goal progress increased the subjective impact of marginal goal progress (linear contrast: F(1, 301) = 6.48, p = .011). Saving an additional $25 was perceived to have a bigger impact on the overall debt repayment goal when participants had already put $250 (M = 4.16) versus $50 (M = 3.74) toward their loans, and when they had already put $450 (M = 4.49) versus $250 toward their loans. However, supporting our theory, in the nonspecific goal condition, the opposite occurred: accumulating goal progress decreased the subjective impact of marginal goal progress (linear contrast: F(1, 301) = 12.70, p < .001). Saving an additional $25 was perceived to have less of an impact on the overall debt repayment goal when participants had already put $250 (M = 3.65) versus $50 (M = 4.49) toward their loans, and when they had already put $450 (M = 3.44) versus $250 toward their loans. See table 1 for pairwise contrasts. Further supporting our reasoning, when goal progress was high, the nonspecific (vs. specific) goal reduced the subjective impact of marginal goal progress. After paying off $450, participants in the nonspecific goal condition perceived saving an additional $25 as less impactful than those in the specific goal condition (Mnonspecific = 3.44, Mspecific = 4.49; F(1, 301) = 12.45, p < .001). This effect was reduced, however, at intermediate goal progress ($250, Mnonspecific = 3.65, Mspecific = 4.16; F(1, 301) = 3.17, p = .077), and it reversed (although the effect was smaller, consistent with our theory) at low goal progress ($50, Mnonspecific = 4.49, Mspecific = 3.73; F(1, 301) = 7.88, p = .005). Underlying Process To examine the proposed underlying role of the subjective impact of marginal goal progress, we ran a bias-corrected bootstrapping mediated moderation analysis with 5,000 samples (PROCESS model 7, Hayes 2013). Results supported hypothesis 3, revealing a significant index of mediated moderation (index: –.06, 95% CI [–.12 to –.01]). In the specific goal condition, accumulating goal progress increased motivation to conserve money (i.e., decreased willingness to spend money on dinner), driven by seeing marginal progress as more impactful (ab = –.02, 95% CI [–.06 to –.002]). In the nonspecific goal condition, however, accumulating goal progress decreased motivation to conserve money (i.e., increased willingness to spend money on dinner), driven by seeing marginal progress as less impactful (ab = .03, 95% CI [.008 to .07]). Discussion Study 2 supports our motivation hypotheses in an important goal domain (debt repayment) and demonstrates the underlying process. Consistent with study 1 and supporting hypothesis 1, when participants had a specific (nonspecific) goal to pay off debt, accumulating goal progress increased (decreased) subsequent motivation to conserve money (i.e., reduced the amount participants were willing to spend on dinner). Also consistent with study 1 and supporting hypothesis 2, when goal progress was high, the nonspecific debt repayment goal reduced motivation relative to the specific goal, but this effect was attenuated at the lower goal progress levels (and even reversed at the lowest level). Importantly, supporting hypothesis 3, these effects were driven by the perceived impact of marginal goal progress (i.e., of saving an additional $25) on the overall debt repayment goal. When participants had a specific (nonspecific) goal, accumulating goal progress made marginal goal progress seem more (less) impactful, and these judgments of subjective impact determined subsequent motivation. Lastly, that we found support for our hypotheses in this controlled realistic goal scenario casts doubt on the possibility that depletion or other differences between the previous study’s goal progress conditions might have explained the effects. STUDY 3 Study 3 underscores the findings of study 2 in a different important goal domain: weight loss. We manipulated the specificity of a weight loss goal, provided participants with goal progress feedback, and then measured motivation and the subjective impact of marginal goal progress (i.e., losing an additional pound) on their overall weight loss goal. To rule alternative explanations, we made two key adjustments to the study design. First, to ensure that the specific goal is not artificially increasing participants’ aspiration level, we set the specific goal level lower than the natural aspiration level. Doing so rules out the possibility that the asymmetry in the goal specificity × goal progress interaction, which we argue is driven by loss aversion, could be attributed to differences in goal difficulty (i.e., the nonspecific goal being less challenging than the specific goal) or perceived goal completion (i.e., nonspecific goal pursuers inferring goal completion at higher progress levels). Second, to rule out potential order effects, we measured subjective impact after motivation. Design and Method Participants (N = 229) were recruited from a university behavioral lab in exchange for course credit. Four individuals declined to participate after reading an eating-disorder trigger warning on the consent page, and 10 participants (4%) began the study but did not complete it, leaving a sample of 215 (average age = 25.21 years, 66.5% female). Participants were randomly assigned to one condition in a 2 (goal specificity: specific, nonspecific) × 2 (goal progress: low, high) between-subjects design. First, we manipulated goal specificity. We asked participants to imagine that they had a goal to lose weight over the next eight weeks. In the specific goal condition, participants reported their current body weight and read that their goal was to “lose 6 pounds over the next 8 weeks” from this starting weight. A pretest conducted with a separate sample from the same population indicated that this goal was just 63% of what people would naturally aim to lose in eight weeks (see appendix A for details). In the nonspecific goal condition, participants reported their current body weight and read that their goal was to “lose as much weight as you can over the next 8 weeks” from this starting weight. Second, we provided goal progress feedback. In the low (high) goal progress condition, participants read that, “near the beginning (end) of the 8 weeks, you’ve lost 1.5 pounds (4.5 pounds) so far. Based on your starting weight of [X] pounds, you would weigh [Y] pounds at this point.” The exact current weight was automatically calculated for each participant based on his or her reported starting weight. A pretest conducted with a separate sample confirmed that in both goal specificity conditions, perceived goal progress increased from 1.5 pounds to 4.5 pounds (see appendix B for details). Third, we measured motivation. Participants responded to two questions: “How motivated would you be to lose weight at this point?” (1 = Not motivated at all, 7 = Extremely motivated) and “How hard would you be willing to work to lose weight at this point?” (1 = Unwilling to work hard, 7 = Willing to work very hard). These items were highly correlated (r = .79) and combined. Finally, we measured the subjective impact of marginal progress. We asked participants, “At this point, how much of an impact would losing the next pound have on your weight loss goal?” (1 = No impact at all, 7 = Very large impact). Results Motivation A 2 (goal specificity) × 2 (goal progress) ANOVA on motivation revealed only the predicted interaction (F(1, 211) = 12.49, p = .001; figure 5). There was no main effect of goal specificity or goal progress (Fs < 1). FIGURE 5 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO LOSE WEIGHT FIGURE 5 View largeDownload slide GOAL SPECIFICITY AFFECTS MOTIVATION TO LOSE WEIGHT Consistent with studies 1 and 2, in the specific goal condition, accumulating goal progress increased subsequent motivation (F(1, 211) = 5.90, p = .016). Specific goal pursuers were more motivated to lose additional weight after losing 4.5 pounds (M = 5.59) versus 1.5 pounds so far (M = 4.82). Supporting hypothesis 1, however, in the nonspecific goal condition, the opposite occurred (F(1, 211) = 6.64, p = .011). Participants were less motivated to lose additional weight after losing 4.5 pounds (M = 4.65) versus 1.5 pounds so far (M = 5.41). Further, consistent with the previous studies and supporting hypothesis 2, when goal progress was high, the nonspecific (vs. specific) goal reduced motivation (Mnonspecific = 4.65, Mspecific = 5.59; F(1, 211) = 9.82, p = .002). When goal progress was low, however, this effect (marginally) reversed (Mnonspecific = 5.41, Mspecific = 4.82; F(1, 211) = 3.56, p = .060). Subjective Impact A 2 (goal specificity) × 2 (goal progress) ANOVA on subjective impact revealed a significant main effect of goal specificity (Mnonspecific = 4.89, Mspecific = 5.37; F(1, 211) = 4.69, p = .03), qualified by the predicted interaction (F(2, 211) = 12.86, p < .001). There was no main effect of goal progress (F < 1). Consistent with study 2, in the specific goal condition, accumulating goal progress increased the subjective impact of marginal goal progress (F(1, 211) = 4.46, p = .036). Losing an additional pound was perceived to have a bigger impact when participants had already lost 4.5 pounds (M = 5.68) versus 1.5 pounds (M = 5.06). As expected, however, in the nonspecific goal condition, the opposite occurred (F(1, 211) = 8.93, p = .003). Losing an additional pound was perceived to have less of an impact when participants had lost 4.5 pounds (M = 4.52) versus 1.5 pounds (M = 5.35). Further, also as expected, when goal progress was high, the nonspecific (vs. specific) goal reduced the subjective impact of marginal goal progress (Mnonspecific = 4.52, Mspecific = 5.68; F(1, 211) = 17.16, p < .001). When goal progress was low, however, this effect directionally reversed (Mnonspecific = 5.35, Mspecific = 5.06; F(1, 211) = .973, p = .325). Underlying Process As in study 2, we ran a bias-corrected bootstrapping mediated moderation analysis to examine the underlying process. Results supported hypothesis 3, revealing a significant index of mediated moderation (index: .91, 95% CI [.41 to 1.51]). In the specific goal condition, accumulating goal progress increased motivation to lose weight, driven by seeing marginal goal progress as more impactful (ab = .39, 95% CI [.05 to .79]). In the nonspecific goal condition, however, accumulating goal progress decreased motivation to lose weight, driven by seeing marginal goal progress as less impactful (ab = –.52, 95% CI [–.92 to –.19]). Discussion Study 3 provides further support for our theory in a different goal domain: weight loss. Consistent with the previous studies and supporting hypothesis 1, when participants had a specific (nonspecific) goal to lose weight, accumulating goal progress increased (decreased) subsequent motivation. Consistent with the previous studies and supporting hypothesis 2, when goal progress was high, the nonspecific (vs. specific) goal reduced motivation, but this effect was eliminated (and directionally reversed) when goal progress was low. Finally, consistent with study 2 and supporting hypothesis 3, these motivation effects were driven by the perceived impact of marginal goal progress (i.e., losing an additional pound) on the overall weight loss goal. Notably, because the specific goal value was calibrated to be one-third lower than participants’ natural targets, these results cannot be explained by nonspecific goal pursuers being less ambitious or feeling that the higher progress level constituted goal completion. We have argued that goal specificity alters the relationship between goal progress and motivation because it makes different reference points salient: the end state for specific goals and the initial state for nonspecific goals. To test the role of reference points more directly, our next two studies (4a and 4b) manipulate the focal reference point for specific goal pursuers—the end state (as is naturally the case) or the initial state—and compare their judgments of subjective impact and motivation to nonspecific goal pursuers. If a difference in salient reference points underlies goal specificity’s effects, as we suggest, then encouraging specific goal pursuers to instead use the initial state as the reference point should make them appear like nonspecific goal pursuers: exhibiting a reverse goal gradient (study 4a) and reducing motivation relative to specific goal pursuers focused on the end state at high goal progress (studies 4a and 4b). Notably, if manipulating specific goal pursuers’ focal reference point attenuates goal specificity’s effects, as we expect, this would further rule out potential alternative explanations due to goal difficulty or goal completion (which rely on unrelated differences between nonspecific and specific goals). STUDY 4A Study 4a directly tests the proposed role of reference point focus in generating goal specificity’s effects. Following the paradigm of study 2, we manipulated the specificity of a debt repayment goal and the current level of goal progress. In addition, in the specific goal condition, we directed some participants to focus on the initial state as the reference point. We expected that, rather than motivation increasing with accumulated goal progress, motivation would decrease with accumulated goal progress (i.e., a reverse goal gradient) in this case. Further, as with nonspecific goal pursuers, we expected that specific goal pursuers focused on the initial state would be less motivated at high (vs. low) goal progress than those in the specific control condition. Design and Method Participants (N = 312) were recruited from a university behavioral lab in exchange for course credit. Four individuals (1%) reported technical problems and failed to complete the study, leaving a sample of 308 (average age = 22.48 years, 70.5% female). Participants were each randomly assigned to one condition in a 3 (reference point focus: specific control, specific initial-state focus, nonspecific) × 2 (goal progress: low, high) between-subjects design. Note that (here and in study 4b) there was no “nonspecific end-state focus” condition, because based on our conceptualization, the end-state reference point does not exist for nonspecific goals. First, we manipulated goal specificity. As in study 2, we asked participants to imagine they were paying off loans over time. In the two specific goal conditions, participants read that their goal this month was to “pay off an extra $500.” In the nonspecific goal condition, participants read that their goal this month was to “pay off as much extra as you can.” Second, we provided goal progress feedback. Participants read that partway through the month, they were planning to go out to dinner with a friend, and that so far this month they had put $50 (low progress condition) or $450 (high progress condition) toward their loans. Third, we manipulated the focal reference point. On the same page as the goal progress feedback, participants viewed a progress bar with a dotted line indicating their current progress level (see appendix C). Following a manipulation used in prior work (Bonezzi et al. 2011; Koo and Fishbach 2012), in the specific initial-state focus condition, we instructed participants to highlight the portion of the progress bar corresponding to their accumulated goal progress (i.e., the area between their current state and the initial state). This encouraged them to compare their current goal progress to the initial-state (rather than the end-state) reference point. Participants in the specific control and nonspecific conditions proceeded directly to the next part of the study. Fourth, we measured motivation. All participants received information about two potential restaurants for the dinner with their friend: Restaurant A, which was described as a restaurant the friend liked with an average cost of $35 per person for dinner, and Restaurant B, which was described as another restaurant the friend liked with an average cost of $20 per person for dinner. We reasoned that the more motivated participants were to put money toward their debt repayment goal, the more they should prefer Restaurant B (the less expensive option) to Restaurant A. Accordingly, we asked them, “Would you be more likely to choose Restaurant A or the less-expensive Restaurant B?” (1 = Definitely Restaurant A, 7 = Definitely Restaurant B). Finally, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much of an impact would saving an additional $15 have on helping you reach your goal for the month?” (1 = No impact at all, 7 = Very large impact). Results Motivation A 3 (reference point focus) × 2 (goal progress) ANOVA on motivation (i.e., preference for the inexpensive restaurant) revealed only the predicted interaction (F(2, 302) = 3.46, p = .033; figure 6). There was no main effect of reference point focus (F(2, 302) = 2.16, p = .118) or goal progress condition (F(1, 302) = 2.54, p = .112). FIGURE 6 View largeDownload slide REFERENCE POINT FOCUS AFFECTS MOTIVATION TO PAY OFF DEBT FIGURE 6 View largeDownload slide REFERENCE POINT FOCUS AFFECTS MOTIVATION TO PAY OFF DEBT Consistent with the prior studies, in the specific control condition, accumulating goal progress (i.e., putting $50 vs. $450 toward the loans) increased subsequent preference for the inexpensive restaurant (although the effect was only directional in this case; Mlow = 5.91, Mhigh = 6.23; F(1, 302) = 1.42, p = .235). In the nonspecific goal condition, however, accumulating goal progress decreased preference for the inexpensive restaurant (Mlow = 6.29, Mhigh = 5.68; F(1, 302) = 5.41, p = .021). Importantly, supporting our theory, in the specific initial-state focus condition, accumulating goal progress also decreased subsequent preference for the inexpensive restaurant (albeit marginally, F(1, 302) = 2.73, p = .099). When we encouraged specific goal pursuers to adopt the initial state as their reference point, as nonspecific goal pursuers do naturally, they were less motivated to conserve money after putting $450 (M = 5.49) versus $50 (M = 5.92) toward their loans (similar to those in the nonspecific goal condition). Also consistent with the prior studies, when goal progress was high, participants in the nonspecific goal condition were less motivated than those in the specific control condition (Mnonspecific = 5.68, Mspecific-control = 6.23; F(1, 302) = 4.17, p = .042). However, supporting our theory, this difference was eliminated when specific goal pursuers were encouraged to focus on the initial state: motivation was lower (M = 5.49) than in the specific control condition (F(1, 302) = 7.65, p = .006) and no different from the nonspecific goal condition (F < 1). When goal progress was low, motivation did not differ between the nonspecific goal (M = 6.29), specific control (M = 5.91), and specific initial-state focus conditions (M = 5.92). See table 2 for pairwise contrasts. Table 2 Pairwise Contrasts between Reference Point Focus Conditions Study, progress (DV)    Specific control vs. nonspecific  Specific control vs. specific initial-state focus  Nonspecific vs. specific initial-state focus  4a, high (preference)  F(1, 302)  4.17**  7.65***  .55  p  .042  .006  .458  4a, high (impact)  F(1, 302)  12.32***  15.11***  .13  p  .001  < .001  .715  4a, low (preference)  F(1, 302)  2.12  < .001  1.91  p  .146  .986  .168  4a, low (impact)  F(1, 302)  2.62  .50  .77  p  .107  .481  .381  4b, high (motivation)  F(1, 189)  10.89***  9.67***  .09  p  .001  .002  .771  4b, high (impact)  F(1, 189)  3.25*  5.42**  .21  p  .073  .021  .646  Study, progress (DV)    Specific control vs. nonspecific  Specific control vs. specific initial-state focus  Nonspecific vs. specific initial-state focus  4a, high (preference)  F(1, 302)  4.17**  7.65***  .55  p  .042  .006  .458  4a, high (impact)  F(1, 302)  12.32***  15.11***  .13  p  .001  < .001  .715  4a, low (preference)  F(1, 302)  2.12  < .001  1.91  p  .146  .986  .168  4a, low (impact)  F(1, 302)  2.62  .50  .77  p  .107  .481  .381  4b, high (motivation)  F(1, 189)  10.89***  9.67***  .09  p  .001  .002  .771  4b, high (impact)  F(1, 189)  3.25*  5.42**  .21  p  .073  .021  .646  NOTE.—Pairwise contrasts in each reference point focus condition of studies 4a and 4b: specific control versus nonspecific, specific control versus specific initial-state focus, nonspecific versus specific initial-state focus. As expected, in the high goal progress condition, the specific control contrasts emerged as significant in each case, whereas the nonspecific versus specific initial-state focus contrast did not. * p < .10, **p < .05, ***p < .01. Subjective Impact A 3 (reference point focus) × 2 (goal progress) ANOVA on subjective impact revealed a marginal main effect of reference point focus (F(2, 302) = 2.73, p = .07), qualified by the predicted interaction (F(2, 302) = 8.09, p < .001). There was no main effect of goal progress (F < 1). Consistent with studies 2 and 3, in the specific control condition, accumulating goal progress (i.e., putting $450 vs. $50 toward the loans) increased the subjective impact of marginal goal progress (Mlow = 4.36, Mhigh = 5.16; F(1, 302) = 8.33, p = .004). In the nonspecific goal condition, however, accumulating goal progress decreased the subjective impact of marginal goal progress (Mlow = 4.80, Mhigh = 4.17; F(1, 302) = 5.23, p = .020). Importantly, as expected, in the specific initial-state focus condition, accumulating goal progress also decreased the subjective impact of marginal goal progress (albeit marginally, F(1, 302) = 3.11, p = .080). When we encouraged specific goal pursuers to adopt the initial state as their reference point, as nonspecific goal pursuers do naturally, they perceived saving an additional $15 to have less of an impact on the overall goal after putting $450 (M = 4.07) versus $50 (M = 4.55) toward their loans (similar to those in the nonspecific goal condition). Also consistent with studies 2 and 3, when goal progress was high, participants in the nonspecific goal condition saw marginal goal progress as more impactful than did those in the specific control condition (Mnonspecific = 4.17, Mspecific-control = 5.16; F(1, 302) = 12.32, p = .001). This difference was eliminated, however, when specific goal pursuers focused on the initial state as the reference point: subjective impact was lower (M = 4.07) than in the specific control condition (F(1, 302) = 15.11, p < .001) and no different from the nonspecific condition (F < 1). When goal progress was low, subjective impact did not differ between the nonspecific goal (M = 4.80), specific control (M = 4.36), and specific initial-state focus conditions (M = 4.55). See table 2 for pairwise contrasts. Underlying Process As in the previous studies, we ran a bias-corrected bootstrapping mediated moderation analysis to examine the underlying process. Because we expected (and found) similar effects in the two conditions where the initial state was salient, these were combined for this analysis (effects are the same if each is separately compared to the specific control condition). Results supported our theory, revealing a significant index of mediated moderation (index: .44, 95% CI [.21 to .76]). In the specific control condition, where participants naturally focused on the end state as the reference point, accumulating goal progress increased motivation by making marginal goal progress seem more impactful (ab = .26, 95% CI [.10 to .48]). In the other two conditions, where participants focused on the initial state as the reference point, accumulating goal progress decreased motivation by making marginal progress seem less impactful (ab = –.18, 95% CI [–.37 to –.05]). Further, in the high goal progress condition, focusing on the initial state (nonspecific and specific initial-state focus conditions) was less motivating than focusing on the end state (specific control), because it made marginal goal progress seem less impactful (ab = .11, 95% CI [.06 to .19]). In the low goal progress condition, the indirect effect was not significant (ab = –.03, 95% CI [–.09 to .01]). Discussion Study 4a provides further insight into the underlying process by directly manipulating the focal reference point. When we encouraged specific goal pursuers to focus on the initial state as the reference point instead, their motivation (and judgments of subjective impact) no longer increased, but decreased with accumulated goal progress (like their nonspecific goal counterparts). Moreover, when current goal progress was high, specific goal pursuers focused on the initial state were less motivated than those in the specific control condition, like those in the nonspecific goal condition. Together these results provide direct evidence for the role of reference points (rather than other potential differences related to goal difficulty or goal completion) in shaping goal specificity’s effects. STUDY 4B Building on study 4a, study 4b further explored the role of reference point focus in determining how goal specificity shapes motivation. Following a similar paradigm to study 3, we manipulated whether specific goal pursuers focused on the initial-state (vs. end-state) reference point. If a difference in salient reference points underlies goal specificity’s effects, as our theory suggests, then encouraging specific goal pursuers to instead use the initial state as the reference point should attenuate the difference between nonspecific and specific goals. We tested this prediction at a high level of goal progress, where goal specificity produces the strongest divergence. Design and Method Participants (N = 192, average age = 22.95 years, 66.1% female) were recruited from a university behavioral lab in exchange for course credit. All recruited participants completed the study and all were included in the analyses. Participants were randomly assigned to a reference point focus condition: specific end-state focus, specific initial-state focus, or nonspecific. First, we manipulated goal specificity. Similar to study 3, in the two specific goal conditions, participants reported their current body weight and read that their goal was to “lose six pounds” from this starting weight. In the nonspecific goal condition, participants reported their current body weight and read that their goal was to “lose as much weight as you can” from this starting weight. Second, we provided high goal progress feedback. All participants read that a few weeks later, they weighed themselves again, and their current weight was five pounds less than their starting weight. The exact current weight value was automatically calculated for each participant based on his or her reported starting weight. Third, we manipulated the focal reference point. Similar to study 4a, participants viewed a progress bar (on a sheet of loose paper) with a dotted line indicating their current level of goal progress and an arrow either pointing left (toward the initial state) or right (toward the end state) (see appendix D). In the specific end-state focus condition, the arrow pointed to the right. In the specific initial-state focus and nonspecific conditions, the arrow pointed to the left. All participants were instructed to shade in the progress bar with a pencil, starting from the dotted line in the direction of the arrow. This encouraged them to compare their current goal progress to either the initial state or end state, depending on condition. Fourth, we measured the subjective impact of marginal goal progress. We asked participants, “At this point, how much would losing an additional pound impact your weight loss goal?” (1 = No impact at all, 7 = Very large impact). Finally, we measured motivation. Participants answered the same two questions from study 4a, which we combined (r = .79). Results Motivation A one-way ANOVA on motivation revealed a significant effect (F(2, 189) = 6.87, p = .001). Consistent with study 4a and supporting our theory, at this high level of goal progress, participants in the nonspecific goal condition were less motivated than those in the specific end-state focus condition (Mnonspecific = 4.21, Mspecific end-state = 5.20; F(1, 189) = 10.89, p = .001). This difference was eliminated, however, when specific goal pursuers were encouraged to focus on the initial state instead (M = 4.29; vs. the specific end-state focus condition: F(1, 189) = 9.67, p = .002; vs. the nonspecific goal condition: F < 1). Subjective Impact A one-way ANOVA on the subjective impact of marginal goal progress also revealed a significant effect (F(2, 189) = 2.99, p = .053). Consistent with study 4a and supporting our theory, at this high level of goal progress, subjective impact was lower in the nonspecific versus the specific end-state focus condition (Mnonspecific = 3.97, Mspecific = 4.59; F(1, 189) = 3.25, p = .073), but this difference was eliminated in the specific initial-state focus condition (M = 3.81; the specific end-state focus condition: F(1, 189) = 5.42, p = .021; the nonspecific condition: F < 1). Underlying Process Similar to the previous studies, we ran a bias-corrected bootstrapping mediation analysis to examine the underlying process. Because we expected (and found) no difference between the two conditions where the initial state was salient, these were combined for this analysis (effects are the same if each is separately compared to the specific end-state focus condition). Results supported our reasoning: at this high level of goal progress, focusing on the initial state—regardless of whether the goal was nonspecific or specific—was less motivating than focusing on the end state, because it made marginal goal progress seem less impactful (ab = .13, 95% CI [.02 to .24]). Discussion Study 4b underscores the role of reference points in shaping goal specificity’s effects. When focused on the naturally more salient reference point (end state for specific goals and initial state for nonspecific goals), the nonspecific goal reduced subjective impact and motivation relative to the specific goal. When specific goal pursuers were directed to focus on the initial state as the reference point instead, however, this effect was attenuated. These findings support our theory that goal specificity alters what reference point consumers spontaneously adopt during goal pursuit, and this difference in focal reference points underlies the documented effects of goal specificity on subsequent motivation. GENERAL DISCUSSION Nonspecific goals are both common and important in consumers’ lives. Yet despite considerable interest in the consequences of setting nonspecific (vs. specific) goals (Locke et al. 1989; Locke and Latham 1990; Naylor and Ilgen 1984; Soman and Cheema 2004; Ülkümen and Cheema 2011; Wright and Kacmar 1994), an understanding of how goal specificity shapes motivation during goal pursuit is more limited. To provide deeper insight into goal specificity’s effects, the current research developed a series of hypotheses that describe how goal specificity and goal progress jointly influence subsequent motivation. Our central proposition is that goal specificity alters what reference point consumers adopt during goal pursuit: for specific goals, the goal objective or specific end state serves as the focal reference point, but for nonspecific goals, which lack a specific end state, the initial state serves as the focal reference point. We argued that this difference in focal reference points has important consequences for (1) how accumulating goal progress shapes motivation to pursue nonspecific (vs. specific) goals, and (2) when (i.e., at what level of goal progress) nonspecific goals reduce (or increase) motivation relative to specific goals. Five studies supported our hypotheses. Across a variety of goal domains (task performance, debt repayment, weight loss), paradigms (lab tasks and realistic goal scenarios), and measures of motivation, consistent results emerged. First, for specific goals, accumulating goal progress increases subsequent motivation, but for nonspecific goals, accumulating goal progress decreases subsequent motivation (studies 1–4a, hypothesis 1). Second, nonspecific (vs. specific) goals are less motivating at higher levels of goal progress (studies 1–4b, hypothesis 2), but this difference is attenuated (and in some cases reversed) at lower levels of goal progress (studies 1–4a, hypothesis 2). Third, the subjective impact of marginal goal progress, which is determined by the shape of the value function, drives these effects (studies 2–4b; hypothesis 3). Our final two studies provided direct evidence for the role of reference points by manipulating whether specific goal pursuers focused on the initial state. When specific goal pursuers were encouraged to adopt the initial state as their reference point (as nonspecific goal pursuers do naturally), their motivation decreased with accumulated goal progress (study 4a) and they were no longer more motivated than nonspecific goal pursuers when current goal progress was high (studies 4a and 4b). These findings underscore that goal specificity’s effects rely on natural differences in reference point focus. Notably, our findings support the proposed roles of both diminishing sensitivity and loss aversion in determining how goal specificity shapes motivation. That the subjective impact of marginal goal progress (and motivation) decreased (increased) with accumulated goal progress for nonspecific (specific) goals underscores that proximity to one’s salient reference point influences motivation (diminishing sensitivity). Moreover, that goal specificity produced a greater effect on subjective impact and motivation at higher (vs. lower) goal progress levels underscores that whether one is below or above the salient reference point (and thus in losses or gains) influences motivation (loss aversion). Further support for the role of loss aversion comes from examining the intermediate level of goal progress. Based on our theory, when consumers’ current level of goal progress is equidistant from the initial-state and end-state reference points, nonspecific goals should tend to reduce motivation relative to specific goals. Because distance from the focal reference point is held constant, loss aversion, rather than diminishing sensitivity, should be the sole determinant of the subjective impact of marginal goal progress, and nonspecific (specific) goals should put people in losses (gains). A single-paper meta-analysis (McShane and Böckenholt 2017) on the intermediate progress level conditions of studies 1 and 2 supported this reasoning. When nonspecific and specific goal pursuers were equally far from their respective reference points, specific goal pursuers showed greater motivation (contrast = 0.23, SE = .11, p = .035).2 These results bolster empirical support for the proposed role of loss aversion in determining how goal specificity shapes motivation. Theoretical Contributions This research makes three main theoretical contributions. First, our findings inform the relationship between goal progress and motivation. A large body of research demonstrates that accumulating goal progress increases subsequent motivation (e.g., the “goal gradient” or “goal looms larger” effect; Hull 1932; Kivetz et al. 2006; Louro et al. 2007; Nunes and Drèze 2011; Soman and Shi 2003). More recently, a few articles have suggested that accumulating goal progress can both increase and decrease subsequent motivation, depending on whether the starting point (i.e., initial state) or ending point (i.e., end state) is salient (e.g., the “stuck in the middle” effect or “small area hypothesis”; Bonezzi et al. 2011; Carton et al. 2011; Koo and Fishbach 2012; Touré-Tillery and Fishbach 2012). Building on these findings, our research identifies goal specificity as a key determinant of the relationship between goal progress and motivation. By influencing what reference point consumers naturally adopt, goal specificity determines whether accumulating goal progress increases or decreases subsequent motivation. Second, this research advances understanding of how goal specificity shapes motivation. Goal specificity is known to influence many aspects of goal pursuit, including goal commitment and performance (Locke et al. 1989; Naylor and Ilgen 1984; Soman and Cheema 2004; Ülkümen and Cheema 2011; Wright and Kacmar 1994). Prior research has explained these effects by noting that nonspecific goals introduce ambiguity into how performance is evaluated (Locke and Latham 1990; Wright and Kacmar 1994). Yet, while this reasoning is consistent with the previously documented effects, it provides limited ability to predict how motivated consumers will be at specific points during goal pursuit (i.e., having accumulated different amounts of goal progress). The current research proposes that, beyond simply making performance evaluation more ambiguous, goal specificity fundamentally changes what reference point consumers adopt during goal pursuit, and that this difference in focal reference points determines how accumulating goal progress affects subsequent motivation. In addition, the current work informs previously documented advantages and disadvantages of specific goals relative to nonspecific goals. For example, prior work finds that setting specific (vs. nonspecific) goals tends to lead to better performance outcomes (Locke and Latham 1990; Locke et al. 1981). Consistent with this, we also find a performance advantage of specific goals relative to nonspecific goals, but show that this occurs primarily at higher levels of goal progress. Prior work also demonstrates that specific goals can exhibit a “starting problem,” such that goal pursuers are reluctant to take initial steps toward very distant goals (Heath et al. 1999), but nonspecific goals, which lack a specific end state, show this less. Consistent with this, we also find that specific (vs. nonspecific) goals can be disadvantageous at lower progress levels, because the value function is less steep. Third, this research generalizes the theory of goals as reference points beyond goals that have specific performance objectives. Since the seminal article introducing this framework (Heath et al. 1999), research has explored its consequences for decision making (Larrick et al. 2009; Medvec and Savitsky 1997) and behavior (Allen et al. 2016; Berger and Pope 2011; Bonezzi et al. 2011; Kivetz et al. 2006; Medvec, Madey, and Gilovich 1995; Pope and Simonsohn 2011), but focused less on further developing the theory. The current work contributes in two important ways: (1) by identifying goal specificity as a key determinant of what reference point consumers spontaneously rely on to evaluate their current goal progress, and (2) by specifying implications of both diminishing sensitivity and loss aversion for goal pursuers focused on a goal’s initial state (vs. end state) as the reference point. Notably, our work is also the first to empirically show that the subjective impact of marginal goal progress, determined by the shape of the value function, drives the effects of reference point focus on subsequent motivation. Prior work has speculated about this underlying process (Bonezzi et al. 2011; Heath et al. 1999; Kivetz et al. 2006; Koo and Fishbach 2012), but only measured its downstream effects on motivation and behavior. By eliciting explicit judgments of the subjective impact of marginal goal progress, the present studies provide direct support for its role in predicting motivation. Practical Implications The findings also have practical implications. For consumers, our work suggests that setting specific goals can lead to greater motivation (Locke et al. 1989; Locke and Latham 1990), but might not always. Early on in the pursuit of a specific goal, when reaching the goal objective is far off, focusing on the end state as the reference point may prove less motivating than focusing on the initial state. This may put specific goals at a disadvantage when consumers’ current level of goal progress is relatively low (e.g., after losing just a couple of pounds or repaying a small portion of a loan). Indeed, in our studies that examined lower levels of goal progress, the nonspecific goal was (at least directionally) more motivating than the specific goal in each case. Particularly for challenging specific goals, which have end states further from the initial state, consumers could thus benefit from deliberately adopting the initial state as the reference point early on or by breaking up their overall goal into smaller subgoals (Fishbach, Dhar, and Zhang 2006; Heath et al. 1999). For marketers, our findings suggest that effective strategies for motivating consumers to pursue specific goals may be ineffective or even harmful for nonspecific goals. Whereas marketers can encourage consumers to work toward specific goals (e.g., loyalty program rewards, product collections) by emphasizing earned progress or endowing unearned progress (Kivetz et al. 2006; Nunes and Drèze 2006; Zhang and Huang 2010), these strategies may backfire for nonspecific goals. For both individual nonspecific goals (e.g., reward programs) and group nonspecific goals (e.g., fundraising drives, petitions), perceiving greater goal progress may undermine motivation by making subsequent actions seem less impactful. Marketers may be able to overcome this demotivating effect by directly bolstering the subjective impact of marginal progress or by encouraging consumers to compare their current progress to external reference points (e.g., social comparisons, prior performance) rather than where they started. Future Research Directions This work suggests several interesting opportunities for future research. One is to investigate the effects of range goals (e.g., lose 10–15 pounds) on motivation. In past research, range goals have been treated as an intermediate level of goal specificity, falling between purely specific goals and nonspecific “do your best” goals (Locke et al. 1989; Naylor and Ilgen 1984; Scott and Nowlis 2013). However, from a reference points perspective, range goals differ from these goal types in that they contain two potential end-state reference points, as opposed to just one for specific goals and none for nonspecific goals. Future work could consider how consumers utilize the additional end-state reference point provided by range goals and explore its consequences for motivation. Another opportunity for further research is to more deeply explore other aspects of goal specificity. Implicit in the current theorizing is that goals provide a specific initial state that can be used as a reference point in the absence of an end state. Yet consumers can also pursue goals that have neither a specific initial state nor a specific end state, such as goals to “get in shape” or “get rich.” When pursuing such goals, consumers may draw potential reference points from alternative sources like round numbers (Pope and Simonsohn 2011) or resource constraints (March and Shapira 1992; Spiller 2011). Future research could further explore alternative sources of reference points during goal pursuit, as well as factors that affect how consumers direct their attention when multiple potential reference points are available. Finally, future work may wish to consider the role of goal difficulty more explicitly. To facilitate fair comparisons between nonspecific and specific goals, our studies used specific goals set at or below consumers’ natural aspiration level (see appendix A). How nonspecific goals compare to more challenging specific goals could also be interesting to test. Based on our theory, we expect that setting specific goals further from the starting point will exacerbate the effects of goal specificity: compared to the effects shown here, higher specific goals should more strongly enhance (reduce) motivation when goal progress is high (low). By exploring these and related phenomena, future research could build on our findings to further understanding of how goal specificity and salient reference points shape consumer motivation. DATA COLLECTION INFORMATION The first author collected pilot study data through Amazon Mechanical Turk in fall 2014. Both authors supervised data collection for studies 1, 4a, and 4b by the lab manager at the University of Pennsylvania’s Wharton Behavioral Lab in winter 2016, fall 2016, and spring 2015, respectively. The first author collected data for study 2 through Amazon Mechanical Turk in winter 2016. Both authors supervised data collection for study 3 by the lab manager at Duke University’s Fuqua Behavioral Lab in fall 2016. The first author conducted analyses for all studies. The authors thank Jonah Berger, Jim Bettman, Gavan Fitzsimons, and Rick Larrick for their helpful comments on prior versions of the manuscript. APPENDIX A GOAL CALIBRATION PRETESTS Calibration of Loan Payment Goal (Studies 2 and 4a) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 130, average age = 34.05 years, 40.8% female). They read the loan payment scenario from study 2, with any reference to the goals or goal progress omitted to avoid biasing responses. We asked participants how much debt they would aim to pay off in one month if they were setting a goal for themselves (open-ended in dollars). Results The average self-generated debt repayment goal was $565.77 (SD = 1015.72). This confirms that the specific goal assigned in studies 2 and 4a (pay off $500 of debt) is aligned with participants’ natural aspiration level and appropriately calibrated for the study. The average self-generated debt repayment goal was also greater than the high goal progress level ($450), indicating that participants in the nonspecific goal condition who received the high goal progress feedback were unlikely to infer that they had already achieved the goal. Calibration of Weight Loss Goal (Studies 3 and 4b) Method Pretest participants were recruited from a university behavioral lab in exchange for course credit (N = 27, average age = 22.93 years, 59.3% female). They read the weight loss scenario from study 3, with any reference to specific goals or goal progress omitted to avoid biasing responses. We asked participants how many pounds they would aim to lose in eight weeks if they were setting a goal for themselves (open-ended in pounds). Results The average self-generated weight loss goal was 9.58 pounds (SD = 7.07). This confirms that the specific goal assigned in studies 3 and 4b (lose six pounds) is below participants’ natural aspiration level and thus could not be artificially inflating their target. The average self-generated weight loss goal was also greater than the high goal progress level (4.5 pounds lost in study 3 and 5 pounds lost in study 4b), indicating that participants in the nonspecific goal condition who received the high progress feedback were unlikely to infer that they had already achieved the goal. APPENDIX B GOAL PROGRESS MANIPULATION PRETESTS Progress Pretest for Loan Payment Goals (Studies 2 and 4a) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 292, average age = 33.88 years, 37.3% female). Participants were randomly assigned to one condition in the same 2 (goal specificity: specific, nonspecific) × 3 (goal progress: low, intermediate, high) between-subjects design used in study 2. To verify the effect of our progress manipulation, we measured goal progress perceptions using two measures: “At this point, how much progress would you feel you had made?” (1 = A little, 7 = A lot) and “At this point, how much money would you feel you had saved to put toward your loans for the month?” (1 = A little, 7 = A lot). These items were highly correlated (r = .89) and combined. Results A 2 (goal specificity) × 3 (goal progress) ANOVA on perceived goal progress revealed a main effect of goal specificity (F(1, 286) = 12.17, p < .001), such that, overall, participants in the specific goal condition perceived greater goal progress than did those in the nonspecific goal condition (Mnonspecific = 3.20 vs. Mspecific = 3.76). Importantly, this analysis also revealed the expected main effect of goal progress (F(2, 286) = 165.83, p < .001). Confirming that the manipulation worked as intended, in the specific goal condition, perceived goal progress significantly increased from the low to the intermediate goal progress condition (Mlow = 1.68 vs. Mintermediate = 3.88; F(1, 286) = 66.46, p < .001), and from the intermediate to the high goal progress condition (Mintermediate = 3.88 vs. Mhigh = 5.78; F(1, 286) = 48.66, p < .001); likewise, in the nonspecific goal condition, perceived goal progress significantly increased from the low to the intermediate goal progress condition (Mlow = 1.75 vs. Mintermediate = 3.30; F(1, 286) = 33.32, p < .001), and from the intermediate to the high goal progress condition (Mintermediate = 3.30 vs. Mhigh = 4.64; F(1, 286) = 23.98, p < .001). The 2 (goal specificity) × 3 (goal progress) ANOVA also revealed an interaction (F(2, 286) = 4.95, p = .008), simply reflecting a difference in the magnitude of the effect of the goal progress manipulation across goal specificity conditions. Most relevant to the present research, the pretest results demonstrate that the goal progress manipulation had the intended effect on goal progress perceptions in both goal specificity conditions. Progress Pretest for Weight Loss Goals (Study 3) Method Pretest participants were recruited from Amazon Mechanical Turk in exchange for small payment (N = 243, average age = 34.00 years, 37.4% female). Participants were randomly assigned to one condition in the same 2 (goal specificity: specific, nonspecific) × 2 (goal progress: low, high) between-subjects design used in study 3. To verify the effect of our progress manipulation, we measured goal progress perceptions using two measures: “At this point, how much progress would you feel you had made?” (1 = A little, 7 = A lot) and “At this point, how much weight would you feel you had lost so far?” (1 = A little, 7 = A lot). These items were highly correlated (r = .84) and combined. Results A 2 (goal specificity) × 2 (goal progress) ANOVA on perceived goal progress revealed a main effect of goal specificity (F(1, 239) = 24.15, p < .001), such that, overall, participants in the specific goal condition perceived greater goal progress than did those in the nonspecific goal condition (Mnonspecific = 3.30 vs. Mspecific = 4.11). Importantly, this analysis also revealed the expected main effect of goal progress (F(1, 239) = 42.02, p < .001). Confirming that the manipulation worked as intended, the progress manipulation increased perceived progress in both the specific goal (Mlow = 3.46 vs. Mhigh = 4.99; F(1, 239) = 27.55, p < .001) and the nonspecific goal condition (Mlow = 2.65 vs. Mhigh = 3.79; F(1, 239) = 15.33, p < .001). There was no interaction between goal specificity and goal progress (F < 1). APPENDIX C REFERENCE POINT MANIPULATION STIMULI (STUDY 4A) Specific Control Condition: Specific Initial-State Focus Condition: NOTE.—Highlighted segments were filled in as participants selected them. Nonspecific Condition: APPENDIX D REFERENCE POINT MANIPULATION STIMULI (STUDY 4B) Specific End-State Focus Condition: Specific Initial-State Focus Condition: Nonspecific Condition: Footnotes 1 When current goal progress is low, the subjective impact of marginal goal progress (and motivation) will depend on the tension between diminishing sensitivity (which should favor nonspecific goals) and loss aversion (which should favor specific goals). If goal progress is sufficiently low (i.e., the distance from the focal reference point is sufficiently small) to outweigh the effect of being in gains (vs. losses), then nonspecific goals may in fact increase motivation relative to specific goals (figure 2). 2 This analysis uses the focal measures of motivation in each study (persistence in study 1 and WTP in study 2, both log-transformed). Due to the reversed coding in study 2 (i.e., lower WTP indicates higher motivation), cell means in that study were reflected around the grand mean. If we instead use the study 2 subjective impact measure, to avoid reverse coding, the focal effect is even stronger (contrast = 0.47, SE = .18, p = .011). REFERENCES Abeler Johannes, Falk Armin, Goette Lorenz, Huffman David ( 2011), “Reference Points and Effort Provision,” American Economic Review , 101 2, 470– 92. Google Scholar CrossRef Search ADS   Allen Eric J., Dechow Patricia M., Pope Devin G., Wu George ( 2016), “Reference-Dependent Preferences: Evidence from Marathon Runners,” Management Science , forthcoming. Amir On, Ariely Dan ( 2008), “Resting on Laurels: The Effects of Discrete Progress Markers as Subgoals on Task Performance and Preferences,” Journal of Experimental Psychology: Learning, Memory, and Cognition , 34 5, 1158– 71. Google Scholar CrossRef Search ADS PubMed  Barberis Nicholas C. ( 2013), “Thirty Years of Prospect Theory in Economics: A Review and Assessment,” Journal of Economic Perspectives,   27 1, 173– 95. Google Scholar CrossRef Search ADS   Berger Jonah, Pope Devin ( 2011), “Can Losing Lead to Winning?”  Management Science,   57 5, 817– 27. Google Scholar CrossRef Search ADS   Bonezzi Andrea, Brendl C. Miguel, De Angelis Matteo ( 2011), “Stuck in the Middle: The Psychophysics of Goal Pursuit,” Psychological Science,   22 5, 607– 12. Google Scholar CrossRef Search ADS   Carton Andrew, Larrick Richard P., Page L. ( 2011), “Back to the Grind: How Attention Affects Satisfaction during Goal Pursuit,” unpublished manuscript, Fuqua School of Business, Duke University, Durham, NC 27708. Etkin Jordan, Ratner Rebecca K. ( 2012), “The Dynamic Impact of Variety among Means on Motivation,” Journal of Consumer Research , 38 ( April), 1076– 92. Google Scholar CrossRef Search ADS   Fishbach Ayelet, Dhar Ravi, Zhang Ying ( 2006), “Subgoals as Substitutes or Complements: The Role of Goal Accessibility,” Journal of Personality and Social Psychology,   91 2, 232– 42. Google Scholar CrossRef Search ADS   Hayes Andrew F. ( 2013), Introduction to Mediation, Moderation, and Conditional Process Analysis: A Regression-based Approach , New York: Guilford Press. Heath Chip, Larrick Richard P., Wu George ( 1999), “Goals as Reference Points,”  Cognitive Psychology , 38 1, 79– 109. Google Scholar CrossRef Search ADS PubMed  Hollenbeck John R., Klein Howard J. ( 1987), “Goal Commitment and the Goal-Setting Process: Problems, Prospects, and Proposals for Future Research,” Journal of Applied Psychology , 74, 18– 23. Google Scholar CrossRef Search ADS   Huang Szu-chi, Zhang Ying, Broniarczyk Susan M. ( 2012), “So Near and Yet So Far: The Mental Representation of Goal Progress,”  Journal of Personality and Social Psychology,   103 2, 225– 41. Google Scholar CrossRef Search ADS   Hull Clark L. ( 1932), “The Goal-Gradient Hypothesis and Maze Learning,” Psychological Review , 39 1, 25– 43. Google Scholar CrossRef Search ADS   Kahneman Daniel ( 1992), “Reference Points, Anchors, Norms, and Mixed Feelings,” Organizational Behavior and Human Decision Processes , 51 2, 296– 312. Google Scholar CrossRef Search ADS   Kahneman Daniel, Tversky Amos ( 1979), “Prospect Theory: An Analysis of Decision under Risk,”  Econometrica: Journal of the Econometric Society , 47 2, 263– 91. Google Scholar CrossRef Search ADS   Kirschenbaum Daniel S., Humphrey Laura L., Malett Sheldon D. ( 1981), “Specificity of Planning in Adult Self-Control: An Applied Investigation,” Journal of Personality and Social Psychology , 40 5, 941– 50. Google Scholar CrossRef Search ADS   Kivetz Ran, Urminsky Oleg, Zheng Yuhuang ( 2006), “The Goal-Gradient Hypothesis Resurrected: Purchase Acceleration, Illusionary Goal Progress, and Customer Retention,”  Journal of Marketing Research,   43 1, 39– 58. Google Scholar CrossRef Search ADS   Klein Howard J., Whitener Ellen M., Ilgen Daniel R. ( 1990), “The Role of Goal Specificity in the Goal-Setting Process,” Motivation and Emotion , 14 3, 179– 93. Google Scholar CrossRef Search ADS   Koo Minjung, Fishbach Ayelet ( 2008), “Dynamics of Self-Regulation: How (Un)accomplished Goal Actions Affect Motivation,” Journal of Personality and Social Psychology , 94 2, 183– 95. Google Scholar CrossRef Search ADS PubMed  Koo Minjung, Fishbach Ayelet ( 2012), “The Small-Area Hypothesis: Effects of Progress Monitoring on Goal Adherence,” Journal of Consumer Research , 39 3, 493– 509. Google Scholar CrossRef Search ADS   Larrick Richard P., Heath Chip, Wu George ( 2009), “Goal-Induced Risk Taking in Negotiation and Decision Making,” Social Cognition , 27 3, 342– 64. Google Scholar CrossRef Search ADS   Locke Edwin A., Chah Dong-Ok, Harrison Scott, Lustgarten Nancy ( 1989), “Separating the Effects of Goal Specificity from Goal Level,” Organizational Behavior and Human Decision Processes , 43 2, 270– 87. Google Scholar CrossRef Search ADS   Locke Edwin A., Latham Gary P. ( 1990), A Theory of Goal Setting and Task Performance , Englewood Cliffs, NJ: Prentice-Hall. Locke Edwin A., Shaw Karyll N., Saari Lise M., Latham Gary P. ( 1981), “Goal Setting and Task Performance: 1960–1980,” Psychological Bulletin , 90 1, 125– 52. Google Scholar CrossRef Search ADS   Louro Maria J., Pieters Rik, Zeelenberg Marcel ( 2007), “Dynamics of Multiple-Goal Pursuit,”  Journal of Personality and Social Psychology,   93 2, 174– 93. Google Scholar CrossRef Search ADS   March James G., Shapira Zur ( 1992), “Variable Risk Preferences and the Focus of Attention,” Psychological Review , 99 1, 172– 83. Google Scholar CrossRef Search ADS   McShane Blakely B., Böckenholt Ulf ( 2017), “Single Paper Meta-Analysis: Benefits for Study Summary, Theory Testing, and Replicability,” Journal of Consumer Research , 43 6, 1048– 63. Medvec Victoria Husted, Madey Scott F., Gilovich Thomas ( 1995), “When Less Is More: Counterfactual Thinking and Satisfaction among Olympic Medalists,” Journal of Personality and Social Psychology , 69 4, 603– 10. Google Scholar CrossRef Search ADS PubMed  Medvec Victoria Husted, Savitsky Kenneth ( 1997), “When Doing Better Means Feeling Worse: The Effects of Categorical Cutoff Points on Counterfactual Thinking and Satisfaction,” Journal of Personality and Social Psychology , 72 6, 1284– 96. Google Scholar CrossRef Search ADS   Naylor James C., Ilgen Daniel R. ( 1984), “Goal Setting: A Theoretical Analysis of a Motivational Technology,” Research in Organizational Behavior , 6, 95– 140. Nunes Joseph C., Drèze Xavier ( 2006), “The Endowed Progress Effect: How Artificial Advancement Increases Effort,” Journal of Consumer Research , 32 4, 504– 12. Google Scholar CrossRef Search ADS   Pope Devin, Simonsohn Uri ( 2011), “Round Numbers as Goals: Evidence from Baseball, SAT Takers, and the Lab,” Psychological Science,   22 1, 71– 9. Google Scholar CrossRef Search ADS   Scott Maura L., Nowlis Stephen M. ( 2013), “The Effect of Goal Specificity on Consumer Goal Reengagement,”  Journal of Consumer Research,   40 3, 444– 59. Google Scholar CrossRef Search ADS   Soman Dilip, Cheema Amar ( 2004), “When Goals Are Counterproductive: The Effects of Violation of a Behavioral Goal on Subsequent Performance,” Journal of Consumer Research,   31 1, 52– 62. Google Scholar CrossRef Search ADS   Soman Dilip, Shi Mengze ( 2003), “Virtual Progress: The Effect of Path Characteristics on Perceptions of Progress and Choice” Management Science , 49 9, 1229– 50. Google Scholar CrossRef Search ADS   Spiller Stephen A. ( 2011), “Opportunity Cost Consideration,” Journal of Consumer Research,   38 4, 595– 610. Google Scholar CrossRef Search ADS   Touré-Tillery Maferima, Fishbach Ayelet ( 2012), “The End Justifies the Means, but Only in the Middle,” Journal of Experimental Psychology: General , 141 3, 570– 83. Google Scholar CrossRef Search ADS PubMed  Tversky Amos, Kahneman Daniel ( 1991), “Loss Aversion in Riskless Choice: A Reference-Dependent Model,” Quarterly Journal of Economics , 106 4, 1039– 61. Google Scholar CrossRef Search ADS   Ülkümen Gülden, Cheema Amar ( 2011), “Framing Goals to Influence Personal Savings: The Role of Specificity and Construal Level,”  Journal of Marketing Research,   48 6, 958– 69. Google Scholar CrossRef Search ADS   Wright Patrick M., Kacmar K. Michele ( 1994), “Goal Specificity as a Determinant of Goal Commitment and Goal Change,” Organizational Behavior and Human Decision Processes , 59 2, 242– 60. Google Scholar CrossRef Search ADS   Wu George, Heath Chip, Larrick Richard P. ( 2008), “A Prospect Theory Model of Goal Behavior,” working paper, Graduate School of Business, University of Chicago, Chicago, IL 60637. Zhang Ying, Huang Szu-Chi ( 2010), “How Endowed versus Earned Progress Affects Consumer Goal Commitment and Motivation,” Journal of Consumer Research , 37 4, 641– 54. Google Scholar CrossRef Search ADS   © The Author 2017. Published by Oxford University Press on behalf of Journal of Consumer Research, Inc. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Journal of Consumer ResearchOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off