Dietary Motivation and Hedonic Hunger Predict Palatable Food Consumption: An Intensive Longitudinal Study of Adolescents

Dietary Motivation and Hedonic Hunger Predict Palatable Food Consumption: An Intensive... Abstract Background Understanding interactions between stable characteristics and fluctuating states underlying youth’s food choices may inform methods for promoting more healthful food intake. Purpose This study examined dietary motivation and hedonic hunger as interacting predictors of adolescents’ consumption of palatable foods. Methods Intensive longitudinal data were collected from 50 adolescents (aged 13–18) over 20 days. Participants completed a measure of dietary motivation at baseline and reported on hedonic hunger and palatable food consumption via a smartphone app at the end of each day. Results Results indicated that 66.7% of the variability in hedonic hunger was between-person (BP) and 33.3% was within-person (WP). BP hedonic hunger was positively associated with fatty food consumption (β = 0.28, p < .05), and WP hedonic hunger was positively associated with starchy food consumption (β = 0.38, p < .0001). Autonomous motivation was negatively associated with consumption of fast foods (β = −0.14, p < .05). Significant cross-level interactions were found: WP hedonic hunger and controlled motivation were positively associated with starchy food consumption, and WP hedonic hunger and autonomous motivation were negatively associated with fast food consumption. Conclusions Findings indicated that hedonic hunger has the potential to fluctuate, and conceptualization of the variable as both trait and state may be most appropriate. Adolescents with controlled dietary motivation may be vulnerable to the influence of hedonic hunger and prone to eating higher quantities of starchy foods. Adolescents with autonomous dietary motivation may be less vulnerable to hedonic hunger and less likely to consume fast food. Hedonic hunger, Dietary motivation, Ecological momentary assessment, Palatable food, Food consumption Current dietary guidelines highlight the need for regulating food consumption by avoiding intake that exceeds caloric need and moderating calories from added sugars, trans fats, and saturated fats [1]. These recent recommendations highlight the importance of all food and beverage choices and emphasize healthy eating across the life span [1]. Adolescence in particular is an important period to study dietary behavior, given the expectation of increased independence, including more freedom to make one’s own food choices [2]. Indeed, the result of this increased independence is often marked by poor dietary choices and low-nutrient, high-calorie food consumption [3]. This widespread poor diet is a main contributor to the continuing national epidemic of overweight and obesity [4], high blood pressure, elevated cholesterol levels, and insulin resistance, all of which lead to morbidity and mortality in adulthood [5]. Understanding the processes that influence adolescents’ food consumption will inform efforts to improve youth’s physical health and weight status. Dietary Motivation and Food Intake Self-determination theory (SDT) provides a useful framework for understanding the motives that may drive food consumption [6]. Within SDT, autonomous motivation is thought to guide behavior that is important or valuable to oneself, independent of judgement from others. Moreover, this type of motivation is characterized by taking responsibility and pride in the choices one makes. In contrast, controlled motivation is characterized by a desire to please others, fit in with social norms, gain respect, or avoid guilt or shame. A recent study in adults found autonomous motivation and goal setting to predict more healthful food choices, such as eating more fruits and vegetables, and controlled motivation to predict choices for less healthful foods, such as eating more sweet and savory palatable foods [7]. Autonomous regulation of eating behavior has also been associated with less consumption of palatable food [8]. One recent study in a youth sample found adolescents’ intrinsic motivation for healthy eating to be linked to more fruit and vegetable consumption [9]. It is relevant to note that although some interventions have successfully increased autonomous motivation for particular health behaviors, such changes seem to require focused and intentional effort. Dietary motivation appears to be stable even in light of a specific intervention [10] and should not change spontaneously in an individual. Hedonic Hunger and Food Intake Hedonic hunger refers to an appetitive drive to consume highly palatable foods for pleasure, which is in contrast to the physiological need for calories that characterizes homeostatic hunger [11]. Food items that are considered palatable are those that quickly replenish energy stores, may be calorie dense, and/or have an effect of reward or pleasure on the consumer. Therefore, high-carbohydrate and high-fat foods could be considered palatable, and the desire to eat each of these classes of palatable foods has been widely studied [12–15]. Our modern society has transitioned into a food environment where palatable food is widely available to adults and children, in terms of both convenience and economic cost [16], allowing many opportunities for hedonic appetite. This obesogenic environment is a complex ecological model with influences from the home, school, and community that promote overeating and inactivity [17]. Multiple interrelated factors (e.g., increased portion sizes and energy density, low cost of palatable foods) create surroundings that provide no restriction in the quantity of food available and, thus, contribute to increased food intake and subsequent weight gain [18]. The Power of Food Scale (PFS), the most used self-report measure of hedonic hunger, assesses individual differences in appetite for palatable foods when respondents are not food deprived. Recent studies have suggested that hedonic hunger may be accurately assessed regardless of energy status and have found hedonic hunger not to be affected by varying levels of homeostatic hunger [19, 20]. Findings of parallel fluctuations in weight and hedonic hunger following bariatric surgery imply a relationship between the construct and dietary behavior in adolescents [21]. In studies of adults, hedonic appetite has sometimes been associated with overeating, particularly in combination with weak inhibitory control [22]. A recent study found that hedonic hunger did not predict short-term effects of overconsumption of highly palatable foods, but did predict higher food consumption overall [23]. Other recent work has found support for measuring hedonic hunger in youth samples [24] and found hedonic appetite to be present in children as young as 9 years old [25]. Still, findings are mixed, as some studies show no relationship of hedonic hunger with excess food consumption or weight status [11]. Further investigation is needed to explore the relationships among hedonic appetite, intake of palatable foods, and implications for weight status in youth. The literature to date has not specified whether hedonic hunger is best measured as a between-person (BP; trait) or within-person (WP; state) variable, or whether both elements should be modeled within a given individual. While major shifts in body weight due to surgery or diet have indeed been associated with changes in hedonic hunger [21], these adjustments occur on a longer time scale (e.g., 18–24 months). Therefore, it remains to be known whether hedonic hunger is subject to day-to-day fluctuations within a person. The existing literature indicates that hedonic eating may be best measured contemporaneously with its occurrence [11]. However, findings from another study suggest that hedonic hunger can be considered a stable construct that should not vary significantly with daily variations in hunger or exposure to food in the immediate environment [20]. Ecological momentary assessment (EMA) is a methodology that assesses phenomena, usually with repeated observations, as they occur in an individual’s natural environment [26]. Such methods have shown to be particularly feasible and effective in adolescent samples [27]. In the current study examining hedonic hunger, EMA methods allowed for an empirical answer to whether hedonic hunger contains both BP and WP components, and subsequently, how to model its effect on food intake. Present Study This study aimed to determine whether the construct of hedonic hunger includes a state (WP) element that merits attention, in addition to its known trait (BP) function. We investigated hedonic hunger’s function as a BP and WP construct and examined the relationships among dietary motivation, hedonic hunger, and food intake. We sought to examine whether the potential effects of a fluctuating drive state, such as hedonic hunger, may moderate the effects of a more stable trait variable, such as dietary motivation, on palatable food consumption. The following hypotheses were developed a priori and were tested: (a) Hedonic hunger was expected to vary over time and be modeled best by including both BP and WP variabilities; (b) Hedonic hunger (both BP and WP) was expected to be positively related to consumption of palatable foods (i.e., sweet, starchy, fatty, and fast foods); (c) Autonomous motivation was expected to be negatively related to consumption of palatable foods; (d) Controlled motivation was expected to be positively related to consumption of palatable foods; (e) WP hedonic hunger was expected to moderate the respective relationships between dietary motivation and palatable food consumption, such that high hedonic hunger and high controlled dietary motivation were expected to predict the most consumption of these foods, and that low hedonic hunger and high autonomous motivation were expected to predict the least consumption of these foods. Methods Participants The participant sample consisted of 50 adolescents, aged 13–18 (M = 14.70, SD = 1.49), recruited in the community of a Midwestern city. Recruitment tactics included posting flyers in local businesses and areas of public recreation, reaching out to school principals for assistance in providing students and parents with information about the study, and distributing information at community events (e.g., farmers’ markets and sporting events). Eligibility criteria included the ability to read at grade level in English, the absence of significant visual impairments, and the absence of any physical conditions that would limit physical activity. For example, adolescents with developmental delays or physical disabilities were excluded, as we aimed to study health behavior processes in typically functioning adolescents for this particular study. The 50 adolescents enrolled represented 62.5% of total parent–adolescent dyads who contacted the research team with interest in the study. Procedure The procedures described below were part of a larger EMA study of physical activity, sleep, and diet in adolescents. All protocols and materials were approved by the Institutional Review Board prior to commencing recruitment or study procedures. Interested participants or parents who learned about the study through community recruitment contacted the research team by telephone to obtain information and complete a brief screening. Adolescents and their parents were told that the purpose of the study was to learn more about physical activity, sleep, and diet in adolescents throughout the day, using technology. It was communicated that the study was observational, meaning participants would not be asked to do anything differently (e.g., exercise more, eat more healthfully) than they usually do. The adolescent–parent dyad were scheduled to visit the local university for consenting and the baseline visit. Initial study visit At the initial study visit, the research staff reviewed the informed consent form with parents and the assent form with adolescents. Each participant then completed a demographic form, an activity calendar of any exercise they had scheduled, and their height and weight were recorded using measurement from a calibrated scale and stadiometer. The research staff directed adolescent participants to complete baseline questionnaires at a computer. The initial visit lasted approximately 1 hr in duration, and an exit visit was scheduled approximately 20 days later. Smartphone App The PETE smartphone app was developed by the research team for internal use in this study to measure time-varying (WP) constructs [27]. The app can be programmed to administer surveys at specific times throughout the day. When it is time to complete a survey, an alarm sounds to notify the participant and continues to sound until the first question has been answered. If a participant happens to have the phone off during a survey notification, the alarm would then sound once the phone was turned back on, and the participant could complete the survey at that time. EMA procedures After completing measures, participants completed training on use of the PETE smartphone app that administered survey questions over the 20-day course of the study. Adolescents and parents discussed daily schedules with the research staff to determine four times throughout each day (e.g., 8 a.m., 12 p.m., 3:30 p.m., and 8 p.m.) during which the adolescent could complete a brief 3–5 min survey. Only the final survey of the day was used in this study, as this was the only daily occasion when the PFS and food consumption measures were administered. The research staff programmed the agreed upon survey times into the smartphone app and showed the participants how the survey app would function. Participants were informed that the phone would produce an alarm at the time of each scheduled survey (i.e., just one alarm for each survey and no additional reminders). If participants were otherwise engaged at the time of the scheduled survey, they could answer the first survey question to stop the alarm and then return to answer the remainder of the survey as soon as possible, and before the next scheduled survey. Participants were also informed that they would be allowed to turn the phone off, if needed, so that an alarm would not sound (e.g., at a movie theater). Participants were provided the smartphone, locked with a passcode so that only the survey app would be accessible. Research staff explained that participants would receive maximum payment for the phone survey portion of the study ($25) if they complete all four surveys on at least 17 of the 20 days they are in the study (i.e., 85% of surveys). Objective measures of physical activity and sleep were also included in the EMA data collection procedures, but these data were not used for this study. Final study visit At the exit visit, participants returned equipment and completed questionnaires. Research staff downloaded participants’ answers to the phone surveys and determined how many questionnaires were completed over the study period. Participants were paid accordingly, based on their compliance with the EMA protocol. Measures Height and weight Each participant’s height and weight were measured at the initial study visit. Body mass index (BMI) percentile was calculated based on age and sex, as indicated by the Centers for Disease Control [28]. Demographics Participants completed a demographic questionnaire, with assistance from parents as needed. The questionnaire included items about gender, date of birth, age, race and ethnicity, and indicators of family socioeconomic status (SES). Autonomous and controlled motivation The Treatment Self-Regulation Questionnaire for diet (TSRQ-D) is a 15-item scale that assesses motivation for eating a healthy diet [29]. The measure begins with The reason I would eat a healthy diet is: and asks responders to rate responses pertaining to dietary motivation. Items are rated on a 7-point Likert scale with options ranging from 1 (not at all true) to 4 (somewhat true), to 7 (very true; i.e., three of the responses provide text anchors and four do not). Because participants may respond not at all true, they are not obligated to endorse any motive for healthy eating. The measure is scored with respective subscales for autonomous motivation and controlled motivation. Responses on each of the subscales are averaged to create separate mean scores. The TSRQ-D has been found to have internally consistent subscales (α values ≥ 0.73; 29). In this sample, the subscales for autonomous motivation and controlled motivation were found to be highly reliable (α = 0.94 and 0.85, respectively), and the TSRQ-D also had high test–retest reliability in this sample (r = 0.84). Previous assessments of construct validity have found the subscales of the TSRQ to correlate with respective health outcomes, with autonomous motivation correlated with perceived confidence in the ability to change one’s diet (r = .54, p < .01; 29). The TSRQ-D was completed as part of the initial study visit surveys. Hedonic hunger The PFS is a 15-item measure that assesses the construct of hedonic hunger [18, 30]. Items assess participants’ thoughts and feelings about eating outside of physiological need with particular attention to highly palatable foods. The modified daily measure is designed to help the respondent make a distinction between hedonic and homeostatic appetite with items such as “Today I found myself thinking about food even when I was not physically hungry.” Response options are on a 5-point Likert scale, ranging from 1 (don’t agree at all) to 5 (strongly agree). A mean of the total score was calculated, with higher scores indicating higher hedonic hunger (range 1–5). The PFS has been found to have good reliability (α = 0.91) and validity, correlating significantly with several measures of eating attitude and behavior [11, 18]. The PFS was highly reliable in this particular sample (α = 0.94). Recent data indicate that in a sample of children and adolescents, the PFS replicates the same three-factor structure (i.e., food available, present, and tasted) with one higher order total score, as it has shown in adults [18, 30, 31]. The PFS was completed once daily within the fourth survey administered through the PETE app on the smartphone. Food consumption The daily food consumption variables of interest were intake of servings of high sugar (sweet: chocolate, cookies, cake, or candy), high carbohydrate (starchy: cereal, sandwich bread, rolls), high fat (fatty: fried chicken, bacon, or sausage), and fast foods (fast food: French fries, chips, pizza, or hamburgers), reported at the end of each study day. The foods used as examples in each category were items with the highest factor loadings on each food construct from a well-validated measure called the Food Craving Inventory [15]. This approach does not have existing psychometric validity, but it was used to avoid participant burden by minimizing the number of daily diet questions. An example of one of these prompts is: How many servings of SWEETS (foods like chocolate, cookies, cake, or candy) have you eaten today? Participants answered one, two, three, four, or five or more to indicate their daily food consumption. The lower end of the scale was intentionally set at 1 (i.e., zero was not a response choice) in an attempt to minimize socially desirable responding that is typical in self-reported dietary assessments. Data Analysis While the data in this study were collected as part of a larger project, the analyses presented here were planned and hypotheses were conceptualized a priori. The current project collected intensive longitudinal data (ILD) using EMA, which was analyzed using multilevel modeling. Analyses were conducted using SAS PROC MIXED. In addition to ILD, baseline assessment was conducted to establish time invariant levels of the constructs of interest to examine as BP effects. Data screening Of the 1,000 expected EMA observations (one daily survey for 50 participants over 20 study days), 69.4% were fully completed by study participants (n = 694), which is lower, yet comparable to compliance rates in other EMA studies of adolescents [27]. The lower rate of compliance in this study was likely due to the rigor required in screening for invalid data, and possibly due to the timing of survey administration at the end of each study day. Specifically, 30.6% of the surveys were missing because participants never started them, and 1.2% of the total expected EMA observations were started but not completed. Data were also screened for uniform responding. Our recent latent profile analysis of EMA data has revealed that participants sometimes provide invalid response by repeatedly pressing the same response choice until the survey is completed [32]. Of the 1,000 total expected observations, 10.3% of cases demonstrated this invalid pattern of responding and were counted as missing in subsequent analysis. Lastly, 3.4% of the total expected EMA observations were excluded if the survey was completed on the morning following the prompt rather than at the end of the study day. The data screening process resulted in one participant not having sufficient valid data for analysis (i.e., their hedonic hunger rating was mistakenly completed in the morning, rather than the evening prior). Following data screening, 54.5% (n = 545) of expected cases were included in the analysis. To be clear, while the proportion of expected responses was 54.5%, the dataset that was entered into SAS PROC MIXED was complete with no missing values. Our dependent variables were the last questions in the EMA survey; therefore, the number of included data points is the same as if we had analyzed all of the expected responses because PROC MIXED excludes cases with missing dependent variables. Establishing WP variability Each dependent variable was entered into a multilevel model with persons at Level 2 and observations at Level 1, and no predictors. An intraclass correlation coefficient (ICC) was computed using the formula ICC = (Random intercept variance/Total variance) × 100 to obtain BP and WP variabilities. Partitioning the variance To test whether multilevel models were needed, we conducted a nested model comparison in PROC MIXED. First we fit an empty model for each dependent variable and then fit a nested model with the residual variance constrained to zero. In each case, the model fit was significantly worse when residual variances were set to zero. Therefore, we concluded that multilevel models were necessary for each dependent variable. Once it was established that multilevel models were necessary, covariates (independent variables) were partitioned into BP and WP components. BP components consisted of the mean of each participant’s responses over time. After grand-mean centering the variables, WP variables were computed by subtracting the person mean described in the previous step from each observation (person-mean centering) resulting in a value that was relative to one’s typical score on an indicator. Modeling time To evaluate how hedonic hunger performs over time (Hypothesis 1), a multilevel model with a fixed linear effect of time was fit, and alternate models were also tested (i.e., random linear, fixed quadratic, and random quadratic) to determine how best to represent time in the final model. Model fit was assessed using nested model comparisons using a −2LL difference test. The procedure above was repeated for all four types of food consumption. Preliminary analyses To determine whether engagement in exercise contributed to consumption of more servings of food over the study days, participants were coded as “Athletes” and “Nonathletes” based on the information they provided in the scheduled activity calendar. Participants who reported engaging in at least one scheduled exercise activity were coded as “Athletes” (54% of participants) and participants who reported having no particular physical activity scheduled were coded as “Nonathletes” (46% of participants), and four multilevel models were run with each type of food consumption as the dependent variable. The results were nonsignificant, and therefore, the variable was not included in subsequent models. Evaluating hypotheses To examine the effects of motivation and hedonic hunger, respectively, on food consumption (Hypotheses 2, 3, and 4), four multilevel models were fit with each type of palatable food consumption as a dependent variable. Equations for each of the four models are included in Fig. 1. Models were specified by adding substantive predictors to the model for time. Predictors for each model included BP hedonic hunger, WP hedonic hunger, BP autonomous motivation, and BP controlled motivation. To evaluate the proposed moderation effect (Hypothesis 5), interaction terms of WP hedonic hunger with each motivation construct were also tested as predictors of each type of food consumption, with the expectation that the interaction between high hedonic hunger and controlled motivation would predict the most consumption of each of the four types of palatable food, and that the interaction between low hedonic hunger and high autonomous motivation would predict the least. As it is critical to evaluate simple slopes, regions of significance, and confidence bands for interactions in multilevel models, specific computational tools for post-hoc probing in multilevel modeling were used to determine the nature of any significant interactions [33]. Fig. 1. View largeDownload slide The γ terms represent fixed effects while terms denoted by u represent random effects. The term i is used to denote a measurement occasion while the j term is used to denote the person-level cluster. Fig. 1. View largeDownload slide The γ terms represent fixed effects while terms denoted by u represent random effects. The term i is used to denote a measurement occasion while the j term is used to denote the person-level cluster. Results Descriptive Statistics The adolescent sample was 60% female and 70% Caucasian, and most individuals were from upper-middle class families (52%; Table 1). The average level of controlled motivation for the sample was 2.93 (SD = 1.26, Min: 1.0, Max: 5.5) while the average reported level of autonomous motivation was 4.76 (SD = 1.53, Min: 1.0, Max: 7.0). The average daily level of hedonic hunger for the sample was 1.71 (SD = 0.79, Min: 1.0, Max: 4.5). On average, participants completed their rating of daily hedonic hunger around 9:00 in the evening (M = 8:54 pm, SD = 84 min). Table 1 Demographic Characteristics of Adolescent Participants and Their Families Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 One participant did not report race/ethnicity. M mean; SD standard deviation; BMI body mass index. View Large Table 1 Demographic Characteristics of Adolescent Participants and Their Families Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 One participant did not report race/ethnicity. M mean; SD standard deviation; BMI body mass index. View Large Screening for Covariates Additional descriptive statistics and bivariate correlations were conducted for the analytic sample (Table 2). Approximate family income was significantly associated with motivation, such that participants reporting higher family income also reported higher levels of controlled and autonomous motivation. An independent samples t-test revealed that female participants reported significantly higher daily consumption of servings of sweet foods than male participants reported (female: M = 2.15, SD = 1.15; male: M = 1.95, SD = 0.90; p < .05). Additionally, male participants reported significantly higher levels of controlled motivation and higher BP hedonic hunger than female participants did. Thus, all models controlled for the variables of approximate family income and gender. Results indicated that associations with these covariates were nonsignificant in each of the four models (Table 3). Observations obtained from participants occurred from June through February of the following year. Results of a model testing the effect of month of year were nonsignificant; and it was not included in subsequent models. Table 2 Summary of Correlations for Income, Motivation, Hedonic Hunger, and Food Consumption Variables Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Means for measures 2–5 are derived from centered variables. Measures 2–3 indicate scores on subscales of TSRQ-D administered at baseline. Measures 4–5 are derived from daily reported scores on the PFS. Measures 6–9 indicate self-reported servings of respective foods, ranging from 1 to 5 or more. M mean; SD standard deviation; TSRQ-D Treatment Self-Regulation Questionnaire–Diet; PFS Power of Food Scale. *p < .05, ** p < .01. View Large Table 2 Summary of Correlations for Income, Motivation, Hedonic Hunger, and Food Consumption Variables Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Means for measures 2–5 are derived from centered variables. Measures 2–3 indicate scores on subscales of TSRQ-D administered at baseline. Measures 4–5 are derived from daily reported scores on the PFS. Measures 6–9 indicate self-reported servings of respective foods, ranging from 1 to 5 or more. M mean; SD standard deviation; TSRQ-D Treatment Self-Regulation Questionnaire–Diet; PFS Power of Food Scale. *p < .05, ** p < .01. View Large Table 3 Associations of Predictors and Covariates With Food Consumption-Dependent Variables Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Intercepts and slopes are reported from the final models for each dependent variable. Each column represents a separate model. Bold text denotes significant effects. BP between-person; WP within-person. View Large Table 3 Associations of Predictors and Covariates With Food Consumption-Dependent Variables Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Intercepts and slopes are reported from the final models for each dependent variable. Each column represents a separate model. Bold text denotes significant effects. BP between-person; WP within-person. View Large Variability, Effects of Time, and Associations With Dependent Variables Hedonic hunger The ICC for hedonic hunger indicated that 66.7% of the variability was BP and 33.3% of the variability was WP. A random quadratic effect of time was established for the hedonic hunger variable, indicating that hedonic hunger plateaued randomly for individuals across the study period. Sweet food consumption The ICC for the sweet dependent variable indicated that 21.4% of the variability was BP, and therefore, 78.7% was WP. The intercept for the empty model was 2.01, and a random quadratic effect of time was established for the dependent variable of sweet food consumption. No significant associations were found between the independent variables and sweet food consumption, and no significant interactions were found. Starchy food consumption The ICC for the starchy food dependent variable indicated that 34.5% of the variability was BP and 65.5% was WP. The intercept for this variable in the empty model was 2.26, and a random linear effect of time was established for the dependent variable of starchy food consumption. WP hedonic hunger was positively associated with consumption of starchy foods (β = .38, p < .0001), such that individuals experiencing higher hedonic hunger than they typically experienced reported consuming more servings of starchy foods. The interaction term of WP hedonic hunger and controlled motivation was also positively associated with starchy food consumption (β = .17, p = .02). Results of probing significant interactions to interpret the conditional effects indicated that, at higher levels of hedonic hunger, the slope relating controlled motivation to starchy food consumption was more strongly positive. At the conditional value of hedonic hunger 1 SD below the mean, the simple slope was 0.06 (p = .55, not significant). At the mean of hedonic hunger, the simple slope was 0.13 (p = .17, not significant). At the conditional value of hedonic hunger 1 SD above the mean, the simple slope was 0.20 (p = .04, significant), indicating that controlled motivation was a significant predictor of consumption of more servings of starchy foods, at higher levels of hedonic hunger. The region of significance for the moderator (hedonic hunger) ranged from −5.42 to 0.35, indicating that any given simple slope outside of this range was statistically significant (Fig. 2). Given that the centered WP hedonic hunger variable had a mean of −0.02 and a standard deviation of 0.40, this indicated that the effect of controlled motivation on starchy food consumption was significant only for relatively high observed values of hedonic hunger. Fig. 2. View largeDownload slide Mean plot illustrating the interaction of controlled motivation and hedonic hunger predicting starchy food consumption (left). The interaction was significant for “High hedonic hunger,” meaning that the top dotted line represents a significant positive relationship between controlled dietary motivation and starchy food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted parallel lines mark the regions of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −5.42 and higher than 0.35. Fig. 2. View largeDownload slide Mean plot illustrating the interaction of controlled motivation and hedonic hunger predicting starchy food consumption (left). The interaction was significant for “High hedonic hunger,” meaning that the top dotted line represents a significant positive relationship between controlled dietary motivation and starchy food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted parallel lines mark the regions of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −5.42 and higher than 0.35. Fatty food consumption The ICC for the fatty food dependent variable indicated that 27.5% of the variability was BP, and 72.5% was WP. The intercept for the empty model was 1.73, and a random linear effect of time was established for the dependent variable of fatty food consumption. BP hedonic hunger was positively associated with consumption of fatty foods (β = .28, p = .03), such that individuals who reported higher levels of hedonic hunger than others also reported consuming more servings of fatty foods. No significant interactions were found to be predictors of fatty food consumption. Fast food consumption The ICC for the fast food dependent variable indicated that 31.8% of the variability was BP and 68.3% was WP. The intercept for the empty model was 1.46, and none of the alternate models for time fit better than the empty model. Autonomous motivation was negatively associated with consumption of fast foods (β = −.14, p = .02). Additionally, the interaction term of WP hedonic hunger and autonomous motivation was negatively associated with fast food consumption (β = −.10, p < .05). Results of probing significant interactions to interpret the conditional effects indicated that, at higher levels of hedonic hunger, the slope relating autonomous motivation to fast food consumption was more strongly negative. At the conditional value of hedonic hunger one standard deviation below the mean, the simple slope was −0.10 (p = .11, not significant). At the mean of hedonic hunger, the simple slope was −0.14 (p = .02, significant). At the conditional value of hedonic hunger one standard deviation above the mean, the simple slope was −0.18 (p = .004, significant), indicating that autonomous motivation was a significant predictor of consumption of fewer servings of fast foods, at average or higher levels of hedonic hunger. The region of significance for the moderator (hedonic hunger) ranged from −0.22 to 420.79, indicating that any given simple slope outside of this range was statistically significant (Fig. 3). Fig. 3. View largeDownload slide Mean plot illustrating the interaction of autonomous motivation and hedonic hunger predicting fast food consumption (left). The interaction was significant for “Average hedonic hunger” and “High hedonic hunger,” meaning that the top two dotted lines represent significant negative relationships between autonomous dietary motivation and fast food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted line marks the region of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −0.22. As the upper bound is 420.79, this dotted line is not pictured in the figure. Fig. 3. View largeDownload slide Mean plot illustrating the interaction of autonomous motivation and hedonic hunger predicting fast food consumption (left). The interaction was significant for “Average hedonic hunger” and “High hedonic hunger,” meaning that the top two dotted lines represent significant negative relationships between autonomous dietary motivation and fast food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted line marks the region of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −0.22. As the upper bound is 420.79, this dotted line is not pictured in the figure. Discussion This study aimed to determine whether hedonic hunger includes a state function, in addition to its current conceptualization as a variable with a trait function, and to examine dietary motivation and hedonic hunger as predictors of adolescents’ consumption of specific types of palatable food. The first hypothesis, that hedonic hunger was expected to vary over time and be modeled best by including both BP and WP variabilities, was supported. Hedonic hunger did, in fact, demonstrate both state and trait properties (66.7%, BP, 33.3% WP variability). The BP variability helps to explain why lab-based protocols are able to detect the effect of hedonic hunger at the group level with a single observation [20, 22, 23]. The WP variability observed in this study answers calls to examine hedonic hunger as a temporally fluctuating variable and highlights the importance of including state conceptualizations of the construct in future research protocols [34, 35]. The evidence that one-third of the variability in hedonic hunger exists WP indicates that continuing to study the variable solely as a trait, and only with BP variability, would ignore a substantial portion of the phenomenon. This observation of both BP and WP variabilities may be considered the necessary first step in establishing the two-part process involved in the construct of hedonic hunger. Additionally, Hypothesis 2, that BP and WP hedonic hunger would be positively related to palatable food consumption, was partially supported. Results indicated that BP hedonic hunger predicted fatty food consumption and WP hedonic hunger predicted starchy food consumption. This indicates that adolescents’ consumption of palatable food may be differentially influenced by whether hedonic hunger is conceptualized as a state or trait variable. That is, adolescents who experience higher hedonic hunger than their peers may be more likely to consume fatty foods, which aligns with previous research associating hedonic hunger with higher unhealthy snack intake in adolescents [36]. On the other hand, the current findings suggest that any adolescent, regardless of how their hedonic hunger compares to their peers’, may be susceptible to consumption of starchy food in association with a spike in their own hedonic hunger. Some researchers have begun to assess the food environment through EMA protocols [35], but this is the first known study to examine WP fluctuations in hedonic hunger in association with food intake. Therefore, future investigations that conceptualize hedonic hunger as a trait variable subject to state fluctuations are needed to cover the range of influences exerted by the construct. The third hypothesis, that autonomous dietary motivation would be negatively related to palatable food consumption, was also partially supported, with results indicating that autonomous motivation was negatively related to consumption of fast foods. This fits with current literature on dietary motivation, which suggests that intrinsic motivation to consume a healthy diet is associated with healthier food choices and the ability to resist unhealthy foods [7, 36]. Regarding the effect of the interaction between autonomous motivation and WP hedonic hunger, the significant interaction predicting fast food consumption was not entirely consistent with Hypothesis 5, as the interaction was significant at mean and higher levels of hedonic hunger, rather than low levels. This suggests that adolescents with high intrinsic motivation to consume a healthful diet may be able to resist the influence of hedonic hunger, even when it is higher than usual, and still ultimately consume fewer servings of fast food. These findings suggest that at higher levels of hedonic hunger, autonomous motivation may be stronger. Results from a qualitative study align with this finding, in that adolescents expressed opinions about taking more autonomous responsibility for healthy food choices after having experienced incidents where fast food made them feel ill or negatively affected their functioning [37]. While unexpected, this finding provides valuable information that fluctuations in hedonic hunger may have important associations with the behavior of adolescents who have low autonomous motivation for a healthy diet. Findings were inconsistent with the fourth hypothesis, that controlled dietary motivation would be positively related to palatable food consumption. Controlled dietary motivation was not a significant predictor of any of the palatable food consumption. It is possible that controlled motivation may not always be directly related to choices that negatively influence health, but that the choices it drives do not foster feelings of satisfaction or of worth [38]. Thus, the effects associated with adolescents’ extrinsically driven dietary choices merit further attention. However, controlled motivation and WP hedonic hunger interacted to predict starchy food consumption, which was consistent with hypothesis 5 (i.e., WP hedonic hunger would moderate relationships between dietary motivation and palatable food consumption) and provided support for WP hedonic hunger as a moderator. Adolescents with high controlled dietary motivation who also experienced higher hedonic hunger than was typical for them reported consuming more servings of starchy foods. This suggests that adolescents with externally motivated reasons for consuming a healthful diet may be more vulnerable to the influence of hedonic hunger and may consume more servings of starchy foods. This finding helps identify the specific combination of controlled dietary motivation and high WP hedonic hunger as a factor that may be related to the consumption of palatable starchy foods. To summarize, the moderation effects of Hypothesis 5 were partially supported, as WP hedonic hunger moderated the respective relationships between controlled motivation and starchy food consumption, and the relationship between autonomous motivation and fast food consumption. Findings from this study align with those of one study that examined a similar combination of variables [36] with the added novelty of considering the effects of adolescents’ individual time-varying fluctuations in hedonic hunger as well as using the variables of interest to predict consumption of particular types of palatable foods. Results of this study confirm the importance of dietary motivation as indicated by prior studies [9, 29] and contribute evidence that hedonic hunger is also a significant time-varying factor that may account for choices in food consumption. The significant multilevel interactions confirm that unique relationships exist between trait dietary motivation and fluctuating hedonic hunger, and that the interactions of these variables on an individual level may hold value in understanding and addressing unhealthful dietary choices (i.e., eating palatable foods in excess). Clinical Implications Findings from this study indicate that adolescents may be capable of resisting the influence of hedonic hunger for fast food most of the time if they hold strong intrinsic motivation to eat a healthy diet. It is possible that having autonomous motivation for a healthy diet would protect an adolescent from engaging in a deliberate and planned unhealthy behavior (e.g., taking a drive to purchase fast food), even when he or she is experiencing high hedonic hunger. Alternatively, autonomous motivation for diet in adolescents may be related to values of the family household, as certain parents may model values of maintaining a healthful diet and therefore do not typically facilitate access to fast food. In contrast, the interaction effect may not be present for other food types because they may be readily available in an adolescent’s home or school, and more subject to impulsive consumption. While dietary motivation does not appear to fluctuate rapidly, there exists some evidence that novel clinical strategies may allow for shifts in motivation over extended periods of time. For example, as mindfulness has been shown to play a role in the development of autonomous regulation and motivation [38], clinical exercises to promote mindful eating strategies may be useful [39, 40]. Clinical efforts for health promotion and prevention could also encourage autonomous motivation through nutrition education to understand the impact of dietary choices on one’s own health and well-being as a means of decreasing fast food consumption. With regard to fatty and starchy foods, autonomous motivation does not appear to serve as a protective factor. Therefore, stimulus control efforts such as removing these foods from the home, storing them in infrequently accessed locations, and avoiding purchasing them at the store are likely to be helpful intervention strategies. Limitations This study used self-reported food consumption and is limited by the fact that our method of assessing palatable food intake was not previously validated. While the food consumption categories were derived from a well-validated measure, we cannot ensure that all adolescents interpreted the items similarly. Although survey items instructed adolescents in how to categorize food consumption, it is possible that the particular categories of sugary and starchy overlapped, and that the categories of fatty and fast food overlapped, potentially leading to variation among adolescents’ categorization of certain foods. Another limitation of the self-reported food consumption was the expectation that adolescents could estimate the amount constituting a serving of the foods indicated. It is possible that adolescents may have underestimated or overestimated serving sizes, resulting in some variation in the amount of food consumption reported. Overall, accurately assessing food intake is a consistent challenge in dietary research, as most individuals tend to underreport their intake, particularly for highly palatable energy-dense foods [41, 42]. Based on past research regarding such underreporting, it is possible that adolescents’ self-report of palatable food consumption in this study was lower than what they truly consumed [41, 42]. The limitations associated with the measurement of palatable food consumption could be improved by using a 24-hr dietary recall to overlap with EMA dietary assessment in future studies. Additionally, while we chose a highly reliable and well-validated measure of dietary motivation (i.e., TSRQ-D; 29), the fact that the particular measure has not been validated in an adolescent sample is a limitation of the present study. Regarding other measurements in the study, it is possible that recency of eating could have influenced daily ratings of hedonic hunger. Although participants were asked to consider their experience of hedonic hunger over the day, it is possible that their rating could be related to their current levels of hunger or satiety. In relation to this, our findings may be limited by the use of only one daily assessment. Additional assessments to evaluate fluctuations within a given day may enhance understanding of real-time fluctuations. Lastly, homogeneity of the sample limits generalizability of this study’s findings. Though recruitment efforts were made to recruit a diverse sample with respect to gender, race, and SES, the resulting sample was predominantly Caucasian and upper-middle class. In relation to this, though our exclusion of adolescents with substantial physical activity limitations was helpful in obtaining our targeted sample, we did not screen participants for particular conditions that may influence eating habits (e.g., diabetes). Future studies could be improved by accounting for such conditions in the research design and analysis. Conclusions and Future Direction This study enhances current knowledge about the function of hedonic hunger as a variable through evidence that it does vary over time and includes substantial BP and WP variabilities. Future studies may more closely examine fluctuations in hedonic hunger by testing whether it changes at different times of day, or in relation to an individual’s daily experiences and previous food consumption. Moreover, this study presents a novel investigation of the relationship between trait dietary motivation and time-varying hedonic hunger, which were found to predict palatable food consumption. While findings about the association between BMI and hedonic hunger have been mixed [31, 43], now that the initial relationships between state and trait hedonic hunger have been investigated, a larger study with appropriate stratification of BMI may include weight status as a predictor variable. Future research may continue to study the TSRQ-D as a measure of dietary motivation in youth samples. The present study considered dietary motivation as a stable construct, not expected to fluctuate day to day within an individual, and focused on examination of such fluctuations specifically in hedonic hunger. However, continued research may allow us to examine whether dietary motivation may also have a WP component or confirm whether it is primarily a BP variable that does not typically fluctuate. Our finding involving autonomous motivation in particular may further support past research in which the combination of inhibitory control and high hedonic hunger predicted greater intake of palatable food [22]. This line of research may gain clarity through inclusion of hedonic hunger, dietary motivation, and inhibitory control together in subsequent studies. Additionally, a variety of other factors may influence state-based shifts in hedonic hunger (e.g., exposure to food cues, recency of consumption), and these should also be included in future investigations. It is recommended that future studies continue to examine adolescents’ dietary behavior with more advanced measures of food consumption, such as three-day dietary recall [44], as well as examine these particular relationships in more diverse samples. Future research should also continue to examine hedonic hunger as both a BP and WP variable, and seek to determine whether the respective uses of the variable differentially predict various dietary choices. Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards Authors Carolina M. Bejarano and Christopher C. Cushing declare that they have no conflict of interest. All procedures, including the informed consent process, were conducted in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. Compliance with Ethical Standards Authors’ Contributions Carolina Bejarano assisted in the conceptualization of the project, led the writing of the document, and led reporting of the findings. Dr. Cushing obtained funding for the work, conceptualized the project, contributed substantial original writing, provided the statistical and methodological design, conducted the statistical analyses, collected the data, edited the work, and supervised Carolina Bejarano's activities. Ethical Approval The current study involved human participants, and all procedures were approved and monitored by the Institutional Review Board at the University of Kansas. No animal subjects were used. Informed Consent Informed consent was obtained from all guardians of all adolescents under the age of 18, and the minor child also provided assent to participate. When the participant was over the age of 18, they provided consent for themselves. Acknowledgements This research was supported in part by a Targeted Research Grant from the Society of Pediatric Psychology awarded to C.C.C. References 1. United States Department of Agriculture . Dietary guidelines for Americans ; 2015 . Available at http://health.gov/dietaryguidelines/2015/guidelines/. Accessibility verified February 12, 2016 . 2. Stok FM , De Ridder DT , Adriaanse MA , De Wit JB . Looking cool or attaining self-rule. Different motives for autonomy and their effects on unhealthy snack purchase . Appetite . 2010 ; 54 ( 3 ): 607 – 610 . Google Scholar CrossRef Search ADS PubMed 3. Reedy J , Krebs-Smith SM . Dietary sources of energy, solid fats, and added sugars among children and adolescents in the United States . J Am Diet Assoc . 2010 ; 110 ( 10 ): 1477 – 1484 . Google Scholar CrossRef Search ADS PubMed 4. Iannotti RJ , Wang J . Trends in physical activity, sedentary behavior, diet, and BMI among US adolescents, 2001–2009 . Pediatrics . 2013 ; 132 ( 4 ): 606 – 614 . Google Scholar CrossRef Search ADS PubMed 5. Günther AL , Schulze MB , Kroke A et al. Early diet and later cancer risk: Prospective associations of dietary patterns during critical periods of childhood with the GH-IGF Axis, insulin resistance and body fatness in younger adulthood . Nutr Cancer . 2015 ; 67 ( 6 ): 877 – 892 . Google Scholar CrossRef Search ADS PubMed 6. Deci EL , Ryan RM. Intrinsic Motivation and Self-determination in Human Behavior . New York, NY : Plenum ; 1985 . Google Scholar CrossRef Search ADS 7. Hartmann C , Dohle S , Siegrist M . A self-determination theory approach to adults’ healthy body weight motivation: A longitudinal study focussing on food choices and recreational physical activity . Psychol Health . 2015 ; 30 ( 8 ): 924 – 948 . Google Scholar CrossRef Search ADS PubMed 8. Leong SL , Madden C , Gray A , Horwath C . Self-determined, autonomous regulation of eating behavior is related to lower body mass index in a nationwide survey of middle-aged women . J Acad Nutr Diet . 2012 ; 112 ( 9 ): 1337 – 1346 . Google Scholar CrossRef Search ADS PubMed 9. Niermann CY , Kremers SP , Renner B , Woll A . Family health climate and adolescents’ physical activity and healthy eating: A cross-sectional study with mother-father-adolescent triads . PLoS One . 2015 ; 10 ( 11 ): e0143599 . Google Scholar CrossRef Search ADS PubMed 10. Rutten GM , Meis JJ , Hendriks MR , Hamers FJ , Veenhof C , Kremers SP . The contribution of lifestyle coaching of overweight patients in primary care to more autonomous motivation for physical activity and healthy dietary behaviour: Results of a longitudinal study . Int J Behav Nutr Phys Act . 2014 ; 11(1) : 86 . Google Scholar CrossRef Search ADS 11. Lowe MR , Butryn ML . Hedonic hunger: A new dimension of appetite ? Physiol Behav . 2007 ; 91 ( 4 ): 432 – 439 . Google Scholar CrossRef Search ADS PubMed 12. Corsica JA , Spring BJ . Carbohydrate craving: A double-blind, placebo-controlled test of the self-medication hypothesis . Eat Behav . 2008 ; 9 ( 4 ): 447 – 454 . Google Scholar CrossRef Search ADS PubMed 13. Erlanson‐Albertsson C . How palatable food disrupts appetite regulation . Basic Clin Pharmacol Toxicol . 2005 ; 97(2) : 61 – 73 . Google Scholar CrossRef Search ADS 14. Oliver G , Wardle J . Perceived effects of stress on food choice . Physiol Behav . 1999 ; 66 ( 3 ): 511 – 515 . Google Scholar CrossRef Search ADS PubMed 15. White MA , Whisenhunt BL , Williamson DA , Greenway FL , Netemeyer RG . Development and validation of the food-craving inventory . Obes Res . 2002 ; 10 ( 2 ): 107 – 114 . Google Scholar CrossRef Search ADS PubMed 16. Borradaile KE , Sherman S , Vander Veur SS et al. Snacking in children: The role of urban corner stores . Pediatrics . 2009 ; 124 ( 5 ): 1293 – 1298 . Google Scholar CrossRef Search ADS PubMed 17. Gorin AA , Crane MM . The obesogenic environment . In: Jelalian E , Steele R , eds. Handbook of Childhood and Adolescent Obesity . New York, NY: Springer US ; 2008 : 145 – 161 . Google Scholar CrossRef Search ADS 18. Lowe MR , Butryn ML , Didie ER et al. The power of food scale. A new measure of the psychological influence of the food environment . Appetite . 2009 ; 53 ( 1 ): 114 – 118 . Google Scholar CrossRef Search ADS PubMed 19. Witt AA , Lowe MR . Hedonic hunger and binge eating among women with eating disorders . Int J Eat Disord . 2014 ; 47 ( 3 ): 273 – 280 . Google Scholar CrossRef Search ADS PubMed 20. Witt AA , Raggio GA , Butryn ML , Lowe MR . Do hunger and exposure to food affect scores on a measure of hedonic hunger? An experimental study . Appetite . 2014 ; 74 : 1 – 5 . Google Scholar CrossRef Search ADS PubMed 21. Cushing CC , Benoit SC , Peugh JL , Reiter-Purtill J , Inge TH , Zeller MH . Longitudinal trends in hedonic hunger after Roux-en-Y gastric bypass in adolescents . Surg Obes Relat Dis . 2014 ; 10 ( 1 ): 125 – 130 . Google Scholar CrossRef Search ADS PubMed 22. Appelhans BM , Woolf K , Pagoto SL , Schneider KL , Whited MC , Liebman R . Inhibiting food reward: Delay discounting, food reward sensitivity, and palatable food intake in overweight and obese women . Obesity (Silver Spring) . 2011 ; 19 ( 11 ): 2175 – 2182 . Google Scholar CrossRef Search ADS PubMed 23. Ely AV , Howard J , Lowe MR . Delayed discounting and hedonic hunger in the prediction of lab-based eating behavior . Eat Behav . 2015 ; 19 : 72 – 75 . Google Scholar CrossRef Search ADS PubMed 24. Laurent JS . Psychometric properties for the children’s power of food scale in a diverse sample of pre-adolescent youth . Appl Nurs Res . 2015 ; 28 ( 2 ): 127 – 131 . Google Scholar CrossRef Search ADS PubMed 25. Laurent JS , Sibold J . Addictive-like eating, body mass index, and psychological correlates in a community sample of preadolescents . J Pediatr Health Care . 2016 ; 30 ( 3 ): 216 – 223 . Google Scholar CrossRef Search ADS PubMed 26. Stone AA , Shiffman S . Ecological momentary assessment (EMA) in behavioral medicine . Ann Behav Med . 1994 ; 16(3) : 199 – 202 . 27. Brannon EE , Cushing CC , Crick CJ , Mitchell TB . The promise of wearable sensors and ecological momentary assessment measures for dynamical systems modeling in adolescents: A feasibility and acceptability study . Transl Behav Med . 2016 ; 6 ( 4 ): 558 – 565 . doi: 10.1007/s13142-016-0442-4 Google Scholar CrossRef Search ADS PubMed 28. Centers for Disease Control and Prevention (CDC) . Growth chart training: A SAS program for the CDC growth charts [Computer software] . Available at http://www.cdc.gov/growth charts/computer_programs.htm. Accessibility verified July 12, 2016 . 29. Levesque CS , Williams GC , Elliot D , Pickering MA , Bodenhamer B , Finley PJ . Validating the theoretical structure of the treatment self-regulation questionnaire (TSRQ) across three different health behaviors . Health Educ Res . 2007 ; 22 ( 5 ): 691 – 702 . Google Scholar CrossRef Search ADS PubMed 30. Cappelleri JC , Bushmakin AG , Gerber RA et al. Evaluating the power of food scale in obese subjects and a general sample of individuals: Development and measurement properties . Int J Obes (Lond) . 2009 ; 33 ( 8 ): 913 – 922 . Google Scholar CrossRef Search ADS PubMed 31. Mitchell TB , Cushing CC , Amaro CM . Psychometric properties of the power of food scale in a community sample of preadolescents and adolescents . J Child Family Stud . 2016; 25(9):2733–2739 32. Cushing CC, Marker AM, Bejarano CM, Crick CJ, Huffhines LP. Latent variable mixture modeling of ecological momentary assessment data: Implications for screening and adolescent mood profiles. J Child Family Stud. 2017; 26(6):1565–1572. 33. Preacher KJ , Curran PJ , Bauer DJ . Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis . J Educ Behav Stat . 2006 ; 31(4) : 437 – 448 . Google Scholar CrossRef Search ADS 34. Boggiano MM , Wenger LE , Turan B et al. Real-time sampling of reasons for hedonic food consumption: Further validation of the Palatable Eating Motives Scale . Front Psychol . 2015 ; 6 : 744 . Google Scholar CrossRef Search ADS PubMed 35. Thomas JG , Doshi S , Crosby RD , Lowe MR . Ecological momentary assessment of obesogenic eating behavior: Combining person-specific and environmental predictors . Obesity (Silver Spring) . 2011 ; 19 ( 8 ): 1574 – 1579 . Google Scholar CrossRef Search ADS PubMed 36. Stok FM , De Vet E , Wardle J , Chu MT , De Wit J , De Ridder DT . Navigating the obesogenic environment: How psychological sensitivity to the food environment and self-regulatory competence are associated with adolescent unhealthy snacking . Eat Behav . 2015 ; 17 : 19 – 22 . Google Scholar CrossRef Search ADS PubMed 37. Bassett R , Chapman GE , Beagan BL . Autonomy and control: The co-construction of adolescent food choice . Appetite . 2008 ; 50 ( 2–3 ): 325 – 332 . Google Scholar CrossRef Search ADS PubMed 38. Deci EL , Ryan RM . Self-determination theory: A macrotheory of human motivation, development, and health . Can Psychol . 2008 ; 49(3) : 182 . Google Scholar CrossRef Search ADS 39. Dalen J , Brody JL , Staples JK , Sedillo D . A Conceptual framework for the expansion of behavioral interventions for youth obesity: A family-based mindful eating approach . Child Obes . 2015 ; 11 ( 5 ): 577 – 584 . Google Scholar CrossRef Search ADS PubMed 40. Forman EM , Butryn ML . A new look at the science of weight control: How acceptance and commitment strategies can address the challenge of self-regulation . Appetite . 2015 ; 84(1) : 171 – 180 . Google Scholar CrossRef Search ADS 41. Johansson G , Wikman A , Ahrén AM , Hallmans G , Johansson I . Underreporting of energy intake in repeated 24-hour recalls related to gender, age, weight status, day of interview, educational level, reported food intake, smoking habits and area of living . Public Health Nutr . 2001 ; 4 ( 4 ): 919 – 927 . Google Scholar CrossRef Search ADS PubMed 42. Macdiarmid J , Blundell J . Assessing dietary intake: Who, what and why of under-reporting . Nutr Res Rev . 1998 ; 11 ( 2 ): 231 – 253 . Google Scholar CrossRef Search ADS PubMed 43. Schultes B , Ernst B , Wilms B , Thurnheer M , Hallschmid M . Hedonic hunger is increased in severely obese patients and is reduced after gastric bypass surgery . Am J Clin Nutr . 2010 ; 92 ( 2 ): 277 – 283 . Google Scholar CrossRef Search ADS PubMed 44. Subar AF , Freedman LS , Tooze JA et al. Addressing current criticism regarding the value of self-report dietary data . J Nutr . 2015 ; 145 ( 12 ): 2639 – 2645 . Google Scholar CrossRef Search ADS PubMed © Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Annals of Behavioral Medicine Oxford University Press

Dietary Motivation and Hedonic Hunger Predict Palatable Food Consumption: An Intensive Longitudinal Study of Adolescents

Loading next page...
 
/lp/ou_press/dietary-motivation-and-hedonic-hunger-predict-palatable-food-WR2UtP0oL4
Copyright
© Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0883-6612
eISSN
1532-4796
D.O.I.
10.1093/abm/kax051
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Understanding interactions between stable characteristics and fluctuating states underlying youth’s food choices may inform methods for promoting more healthful food intake. Purpose This study examined dietary motivation and hedonic hunger as interacting predictors of adolescents’ consumption of palatable foods. Methods Intensive longitudinal data were collected from 50 adolescents (aged 13–18) over 20 days. Participants completed a measure of dietary motivation at baseline and reported on hedonic hunger and palatable food consumption via a smartphone app at the end of each day. Results Results indicated that 66.7% of the variability in hedonic hunger was between-person (BP) and 33.3% was within-person (WP). BP hedonic hunger was positively associated with fatty food consumption (β = 0.28, p < .05), and WP hedonic hunger was positively associated with starchy food consumption (β = 0.38, p < .0001). Autonomous motivation was negatively associated with consumption of fast foods (β = −0.14, p < .05). Significant cross-level interactions were found: WP hedonic hunger and controlled motivation were positively associated with starchy food consumption, and WP hedonic hunger and autonomous motivation were negatively associated with fast food consumption. Conclusions Findings indicated that hedonic hunger has the potential to fluctuate, and conceptualization of the variable as both trait and state may be most appropriate. Adolescents with controlled dietary motivation may be vulnerable to the influence of hedonic hunger and prone to eating higher quantities of starchy foods. Adolescents with autonomous dietary motivation may be less vulnerable to hedonic hunger and less likely to consume fast food. Hedonic hunger, Dietary motivation, Ecological momentary assessment, Palatable food, Food consumption Current dietary guidelines highlight the need for regulating food consumption by avoiding intake that exceeds caloric need and moderating calories from added sugars, trans fats, and saturated fats [1]. These recent recommendations highlight the importance of all food and beverage choices and emphasize healthy eating across the life span [1]. Adolescence in particular is an important period to study dietary behavior, given the expectation of increased independence, including more freedom to make one’s own food choices [2]. Indeed, the result of this increased independence is often marked by poor dietary choices and low-nutrient, high-calorie food consumption [3]. This widespread poor diet is a main contributor to the continuing national epidemic of overweight and obesity [4], high blood pressure, elevated cholesterol levels, and insulin resistance, all of which lead to morbidity and mortality in adulthood [5]. Understanding the processes that influence adolescents’ food consumption will inform efforts to improve youth’s physical health and weight status. Dietary Motivation and Food Intake Self-determination theory (SDT) provides a useful framework for understanding the motives that may drive food consumption [6]. Within SDT, autonomous motivation is thought to guide behavior that is important or valuable to oneself, independent of judgement from others. Moreover, this type of motivation is characterized by taking responsibility and pride in the choices one makes. In contrast, controlled motivation is characterized by a desire to please others, fit in with social norms, gain respect, or avoid guilt or shame. A recent study in adults found autonomous motivation and goal setting to predict more healthful food choices, such as eating more fruits and vegetables, and controlled motivation to predict choices for less healthful foods, such as eating more sweet and savory palatable foods [7]. Autonomous regulation of eating behavior has also been associated with less consumption of palatable food [8]. One recent study in a youth sample found adolescents’ intrinsic motivation for healthy eating to be linked to more fruit and vegetable consumption [9]. It is relevant to note that although some interventions have successfully increased autonomous motivation for particular health behaviors, such changes seem to require focused and intentional effort. Dietary motivation appears to be stable even in light of a specific intervention [10] and should not change spontaneously in an individual. Hedonic Hunger and Food Intake Hedonic hunger refers to an appetitive drive to consume highly palatable foods for pleasure, which is in contrast to the physiological need for calories that characterizes homeostatic hunger [11]. Food items that are considered palatable are those that quickly replenish energy stores, may be calorie dense, and/or have an effect of reward or pleasure on the consumer. Therefore, high-carbohydrate and high-fat foods could be considered palatable, and the desire to eat each of these classes of palatable foods has been widely studied [12–15]. Our modern society has transitioned into a food environment where palatable food is widely available to adults and children, in terms of both convenience and economic cost [16], allowing many opportunities for hedonic appetite. This obesogenic environment is a complex ecological model with influences from the home, school, and community that promote overeating and inactivity [17]. Multiple interrelated factors (e.g., increased portion sizes and energy density, low cost of palatable foods) create surroundings that provide no restriction in the quantity of food available and, thus, contribute to increased food intake and subsequent weight gain [18]. The Power of Food Scale (PFS), the most used self-report measure of hedonic hunger, assesses individual differences in appetite for palatable foods when respondents are not food deprived. Recent studies have suggested that hedonic hunger may be accurately assessed regardless of energy status and have found hedonic hunger not to be affected by varying levels of homeostatic hunger [19, 20]. Findings of parallel fluctuations in weight and hedonic hunger following bariatric surgery imply a relationship between the construct and dietary behavior in adolescents [21]. In studies of adults, hedonic appetite has sometimes been associated with overeating, particularly in combination with weak inhibitory control [22]. A recent study found that hedonic hunger did not predict short-term effects of overconsumption of highly palatable foods, but did predict higher food consumption overall [23]. Other recent work has found support for measuring hedonic hunger in youth samples [24] and found hedonic appetite to be present in children as young as 9 years old [25]. Still, findings are mixed, as some studies show no relationship of hedonic hunger with excess food consumption or weight status [11]. Further investigation is needed to explore the relationships among hedonic appetite, intake of palatable foods, and implications for weight status in youth. The literature to date has not specified whether hedonic hunger is best measured as a between-person (BP; trait) or within-person (WP; state) variable, or whether both elements should be modeled within a given individual. While major shifts in body weight due to surgery or diet have indeed been associated with changes in hedonic hunger [21], these adjustments occur on a longer time scale (e.g., 18–24 months). Therefore, it remains to be known whether hedonic hunger is subject to day-to-day fluctuations within a person. The existing literature indicates that hedonic eating may be best measured contemporaneously with its occurrence [11]. However, findings from another study suggest that hedonic hunger can be considered a stable construct that should not vary significantly with daily variations in hunger or exposure to food in the immediate environment [20]. Ecological momentary assessment (EMA) is a methodology that assesses phenomena, usually with repeated observations, as they occur in an individual’s natural environment [26]. Such methods have shown to be particularly feasible and effective in adolescent samples [27]. In the current study examining hedonic hunger, EMA methods allowed for an empirical answer to whether hedonic hunger contains both BP and WP components, and subsequently, how to model its effect on food intake. Present Study This study aimed to determine whether the construct of hedonic hunger includes a state (WP) element that merits attention, in addition to its known trait (BP) function. We investigated hedonic hunger’s function as a BP and WP construct and examined the relationships among dietary motivation, hedonic hunger, and food intake. We sought to examine whether the potential effects of a fluctuating drive state, such as hedonic hunger, may moderate the effects of a more stable trait variable, such as dietary motivation, on palatable food consumption. The following hypotheses were developed a priori and were tested: (a) Hedonic hunger was expected to vary over time and be modeled best by including both BP and WP variabilities; (b) Hedonic hunger (both BP and WP) was expected to be positively related to consumption of palatable foods (i.e., sweet, starchy, fatty, and fast foods); (c) Autonomous motivation was expected to be negatively related to consumption of palatable foods; (d) Controlled motivation was expected to be positively related to consumption of palatable foods; (e) WP hedonic hunger was expected to moderate the respective relationships between dietary motivation and palatable food consumption, such that high hedonic hunger and high controlled dietary motivation were expected to predict the most consumption of these foods, and that low hedonic hunger and high autonomous motivation were expected to predict the least consumption of these foods. Methods Participants The participant sample consisted of 50 adolescents, aged 13–18 (M = 14.70, SD = 1.49), recruited in the community of a Midwestern city. Recruitment tactics included posting flyers in local businesses and areas of public recreation, reaching out to school principals for assistance in providing students and parents with information about the study, and distributing information at community events (e.g., farmers’ markets and sporting events). Eligibility criteria included the ability to read at grade level in English, the absence of significant visual impairments, and the absence of any physical conditions that would limit physical activity. For example, adolescents with developmental delays or physical disabilities were excluded, as we aimed to study health behavior processes in typically functioning adolescents for this particular study. The 50 adolescents enrolled represented 62.5% of total parent–adolescent dyads who contacted the research team with interest in the study. Procedure The procedures described below were part of a larger EMA study of physical activity, sleep, and diet in adolescents. All protocols and materials were approved by the Institutional Review Board prior to commencing recruitment or study procedures. Interested participants or parents who learned about the study through community recruitment contacted the research team by telephone to obtain information and complete a brief screening. Adolescents and their parents were told that the purpose of the study was to learn more about physical activity, sleep, and diet in adolescents throughout the day, using technology. It was communicated that the study was observational, meaning participants would not be asked to do anything differently (e.g., exercise more, eat more healthfully) than they usually do. The adolescent–parent dyad were scheduled to visit the local university for consenting and the baseline visit. Initial study visit At the initial study visit, the research staff reviewed the informed consent form with parents and the assent form with adolescents. Each participant then completed a demographic form, an activity calendar of any exercise they had scheduled, and their height and weight were recorded using measurement from a calibrated scale and stadiometer. The research staff directed adolescent participants to complete baseline questionnaires at a computer. The initial visit lasted approximately 1 hr in duration, and an exit visit was scheduled approximately 20 days later. Smartphone App The PETE smartphone app was developed by the research team for internal use in this study to measure time-varying (WP) constructs [27]. The app can be programmed to administer surveys at specific times throughout the day. When it is time to complete a survey, an alarm sounds to notify the participant and continues to sound until the first question has been answered. If a participant happens to have the phone off during a survey notification, the alarm would then sound once the phone was turned back on, and the participant could complete the survey at that time. EMA procedures After completing measures, participants completed training on use of the PETE smartphone app that administered survey questions over the 20-day course of the study. Adolescents and parents discussed daily schedules with the research staff to determine four times throughout each day (e.g., 8 a.m., 12 p.m., 3:30 p.m., and 8 p.m.) during which the adolescent could complete a brief 3–5 min survey. Only the final survey of the day was used in this study, as this was the only daily occasion when the PFS and food consumption measures were administered. The research staff programmed the agreed upon survey times into the smartphone app and showed the participants how the survey app would function. Participants were informed that the phone would produce an alarm at the time of each scheduled survey (i.e., just one alarm for each survey and no additional reminders). If participants were otherwise engaged at the time of the scheduled survey, they could answer the first survey question to stop the alarm and then return to answer the remainder of the survey as soon as possible, and before the next scheduled survey. Participants were also informed that they would be allowed to turn the phone off, if needed, so that an alarm would not sound (e.g., at a movie theater). Participants were provided the smartphone, locked with a passcode so that only the survey app would be accessible. Research staff explained that participants would receive maximum payment for the phone survey portion of the study ($25) if they complete all four surveys on at least 17 of the 20 days they are in the study (i.e., 85% of surveys). Objective measures of physical activity and sleep were also included in the EMA data collection procedures, but these data were not used for this study. Final study visit At the exit visit, participants returned equipment and completed questionnaires. Research staff downloaded participants’ answers to the phone surveys and determined how many questionnaires were completed over the study period. Participants were paid accordingly, based on their compliance with the EMA protocol. Measures Height and weight Each participant’s height and weight were measured at the initial study visit. Body mass index (BMI) percentile was calculated based on age and sex, as indicated by the Centers for Disease Control [28]. Demographics Participants completed a demographic questionnaire, with assistance from parents as needed. The questionnaire included items about gender, date of birth, age, race and ethnicity, and indicators of family socioeconomic status (SES). Autonomous and controlled motivation The Treatment Self-Regulation Questionnaire for diet (TSRQ-D) is a 15-item scale that assesses motivation for eating a healthy diet [29]. The measure begins with The reason I would eat a healthy diet is: and asks responders to rate responses pertaining to dietary motivation. Items are rated on a 7-point Likert scale with options ranging from 1 (not at all true) to 4 (somewhat true), to 7 (very true; i.e., three of the responses provide text anchors and four do not). Because participants may respond not at all true, they are not obligated to endorse any motive for healthy eating. The measure is scored with respective subscales for autonomous motivation and controlled motivation. Responses on each of the subscales are averaged to create separate mean scores. The TSRQ-D has been found to have internally consistent subscales (α values ≥ 0.73; 29). In this sample, the subscales for autonomous motivation and controlled motivation were found to be highly reliable (α = 0.94 and 0.85, respectively), and the TSRQ-D also had high test–retest reliability in this sample (r = 0.84). Previous assessments of construct validity have found the subscales of the TSRQ to correlate with respective health outcomes, with autonomous motivation correlated with perceived confidence in the ability to change one’s diet (r = .54, p < .01; 29). The TSRQ-D was completed as part of the initial study visit surveys. Hedonic hunger The PFS is a 15-item measure that assesses the construct of hedonic hunger [18, 30]. Items assess participants’ thoughts and feelings about eating outside of physiological need with particular attention to highly palatable foods. The modified daily measure is designed to help the respondent make a distinction between hedonic and homeostatic appetite with items such as “Today I found myself thinking about food even when I was not physically hungry.” Response options are on a 5-point Likert scale, ranging from 1 (don’t agree at all) to 5 (strongly agree). A mean of the total score was calculated, with higher scores indicating higher hedonic hunger (range 1–5). The PFS has been found to have good reliability (α = 0.91) and validity, correlating significantly with several measures of eating attitude and behavior [11, 18]. The PFS was highly reliable in this particular sample (α = 0.94). Recent data indicate that in a sample of children and adolescents, the PFS replicates the same three-factor structure (i.e., food available, present, and tasted) with one higher order total score, as it has shown in adults [18, 30, 31]. The PFS was completed once daily within the fourth survey administered through the PETE app on the smartphone. Food consumption The daily food consumption variables of interest were intake of servings of high sugar (sweet: chocolate, cookies, cake, or candy), high carbohydrate (starchy: cereal, sandwich bread, rolls), high fat (fatty: fried chicken, bacon, or sausage), and fast foods (fast food: French fries, chips, pizza, or hamburgers), reported at the end of each study day. The foods used as examples in each category were items with the highest factor loadings on each food construct from a well-validated measure called the Food Craving Inventory [15]. This approach does not have existing psychometric validity, but it was used to avoid participant burden by minimizing the number of daily diet questions. An example of one of these prompts is: How many servings of SWEETS (foods like chocolate, cookies, cake, or candy) have you eaten today? Participants answered one, two, three, four, or five or more to indicate their daily food consumption. The lower end of the scale was intentionally set at 1 (i.e., zero was not a response choice) in an attempt to minimize socially desirable responding that is typical in self-reported dietary assessments. Data Analysis While the data in this study were collected as part of a larger project, the analyses presented here were planned and hypotheses were conceptualized a priori. The current project collected intensive longitudinal data (ILD) using EMA, which was analyzed using multilevel modeling. Analyses were conducted using SAS PROC MIXED. In addition to ILD, baseline assessment was conducted to establish time invariant levels of the constructs of interest to examine as BP effects. Data screening Of the 1,000 expected EMA observations (one daily survey for 50 participants over 20 study days), 69.4% were fully completed by study participants (n = 694), which is lower, yet comparable to compliance rates in other EMA studies of adolescents [27]. The lower rate of compliance in this study was likely due to the rigor required in screening for invalid data, and possibly due to the timing of survey administration at the end of each study day. Specifically, 30.6% of the surveys were missing because participants never started them, and 1.2% of the total expected EMA observations were started but not completed. Data were also screened for uniform responding. Our recent latent profile analysis of EMA data has revealed that participants sometimes provide invalid response by repeatedly pressing the same response choice until the survey is completed [32]. Of the 1,000 total expected observations, 10.3% of cases demonstrated this invalid pattern of responding and were counted as missing in subsequent analysis. Lastly, 3.4% of the total expected EMA observations were excluded if the survey was completed on the morning following the prompt rather than at the end of the study day. The data screening process resulted in one participant not having sufficient valid data for analysis (i.e., their hedonic hunger rating was mistakenly completed in the morning, rather than the evening prior). Following data screening, 54.5% (n = 545) of expected cases were included in the analysis. To be clear, while the proportion of expected responses was 54.5%, the dataset that was entered into SAS PROC MIXED was complete with no missing values. Our dependent variables were the last questions in the EMA survey; therefore, the number of included data points is the same as if we had analyzed all of the expected responses because PROC MIXED excludes cases with missing dependent variables. Establishing WP variability Each dependent variable was entered into a multilevel model with persons at Level 2 and observations at Level 1, and no predictors. An intraclass correlation coefficient (ICC) was computed using the formula ICC = (Random intercept variance/Total variance) × 100 to obtain BP and WP variabilities. Partitioning the variance To test whether multilevel models were needed, we conducted a nested model comparison in PROC MIXED. First we fit an empty model for each dependent variable and then fit a nested model with the residual variance constrained to zero. In each case, the model fit was significantly worse when residual variances were set to zero. Therefore, we concluded that multilevel models were necessary for each dependent variable. Once it was established that multilevel models were necessary, covariates (independent variables) were partitioned into BP and WP components. BP components consisted of the mean of each participant’s responses over time. After grand-mean centering the variables, WP variables were computed by subtracting the person mean described in the previous step from each observation (person-mean centering) resulting in a value that was relative to one’s typical score on an indicator. Modeling time To evaluate how hedonic hunger performs over time (Hypothesis 1), a multilevel model with a fixed linear effect of time was fit, and alternate models were also tested (i.e., random linear, fixed quadratic, and random quadratic) to determine how best to represent time in the final model. Model fit was assessed using nested model comparisons using a −2LL difference test. The procedure above was repeated for all four types of food consumption. Preliminary analyses To determine whether engagement in exercise contributed to consumption of more servings of food over the study days, participants were coded as “Athletes” and “Nonathletes” based on the information they provided in the scheduled activity calendar. Participants who reported engaging in at least one scheduled exercise activity were coded as “Athletes” (54% of participants) and participants who reported having no particular physical activity scheduled were coded as “Nonathletes” (46% of participants), and four multilevel models were run with each type of food consumption as the dependent variable. The results were nonsignificant, and therefore, the variable was not included in subsequent models. Evaluating hypotheses To examine the effects of motivation and hedonic hunger, respectively, on food consumption (Hypotheses 2, 3, and 4), four multilevel models were fit with each type of palatable food consumption as a dependent variable. Equations for each of the four models are included in Fig. 1. Models were specified by adding substantive predictors to the model for time. Predictors for each model included BP hedonic hunger, WP hedonic hunger, BP autonomous motivation, and BP controlled motivation. To evaluate the proposed moderation effect (Hypothesis 5), interaction terms of WP hedonic hunger with each motivation construct were also tested as predictors of each type of food consumption, with the expectation that the interaction between high hedonic hunger and controlled motivation would predict the most consumption of each of the four types of palatable food, and that the interaction between low hedonic hunger and high autonomous motivation would predict the least. As it is critical to evaluate simple slopes, regions of significance, and confidence bands for interactions in multilevel models, specific computational tools for post-hoc probing in multilevel modeling were used to determine the nature of any significant interactions [33]. Fig. 1. View largeDownload slide The γ terms represent fixed effects while terms denoted by u represent random effects. The term i is used to denote a measurement occasion while the j term is used to denote the person-level cluster. Fig. 1. View largeDownload slide The γ terms represent fixed effects while terms denoted by u represent random effects. The term i is used to denote a measurement occasion while the j term is used to denote the person-level cluster. Results Descriptive Statistics The adolescent sample was 60% female and 70% Caucasian, and most individuals were from upper-middle class families (52%; Table 1). The average level of controlled motivation for the sample was 2.93 (SD = 1.26, Min: 1.0, Max: 5.5) while the average reported level of autonomous motivation was 4.76 (SD = 1.53, Min: 1.0, Max: 7.0). The average daily level of hedonic hunger for the sample was 1.71 (SD = 0.79, Min: 1.0, Max: 4.5). On average, participants completed their rating of daily hedonic hunger around 9:00 in the evening (M = 8:54 pm, SD = 84 min). Table 1 Demographic Characteristics of Adolescent Participants and Their Families Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 One participant did not report race/ethnicity. M mean; SD standard deviation; BMI body mass index. View Large Table 1 Demographic Characteristics of Adolescent Participants and Their Families Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 Demographic variables n = 50 % Gender  Male 20 40  Female 30 60 Race/Ethnicity  Caucasian 35 70  African American 2 4  Hispanic/Latino 7 14  Asian 1 2  Other/Multiracial 4 8 Approximate family income  Less than $10,000 1 2  $10,000–$20,000 3 6  $21,000–$30,000 3 6  $31,000–$40,000 2 4  $41,000–$50,000 2 4  $51,000–$60,000 13 26  More than $60,000 26 52 Mother’s highest level of education  High school graduate 8 16  College graduate 25 50  Master’s degree 12 24  PhD/JD, MD 3 6  Other 2 4 Father’s highest level of education  High school graduate 15 30  College graduate 16 32  Master’s degree 4 8  PhD/JD, MD 5 10  Other 10 20 M SD Adolescent’s age at baseline (years) 14.70 1.49 BMI percentile 60.78 29.16 One participant did not report race/ethnicity. M mean; SD standard deviation; BMI body mass index. View Large Screening for Covariates Additional descriptive statistics and bivariate correlations were conducted for the analytic sample (Table 2). Approximate family income was significantly associated with motivation, such that participants reporting higher family income also reported higher levels of controlled and autonomous motivation. An independent samples t-test revealed that female participants reported significantly higher daily consumption of servings of sweet foods than male participants reported (female: M = 2.15, SD = 1.15; male: M = 1.95, SD = 0.90; p < .05). Additionally, male participants reported significantly higher levels of controlled motivation and higher BP hedonic hunger than female participants did. Thus, all models controlled for the variables of approximate family income and gender. Results indicated that associations with these covariates were nonsignificant in each of the four models (Table 3). Observations obtained from participants occurred from June through February of the following year. Results of a model testing the effect of month of year were nonsignificant; and it was not included in subsequent models. Table 2 Summary of Correlations for Income, Motivation, Hedonic Hunger, and Food Consumption Variables Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Means for measures 2–5 are derived from centered variables. Measures 2–3 indicate scores on subscales of TSRQ-D administered at baseline. Measures 4–5 are derived from daily reported scores on the PFS. Measures 6–9 indicate self-reported servings of respective foods, ranging from 1 to 5 or more. M mean; SD standard deviation; TSRQ-D Treatment Self-Regulation Questionnaire–Diet; PFS Power of Food Scale. *p < .05, ** p < .01. View Large Table 2 Summary of Correlations for Income, Motivation, Hedonic Hunger, and Food Consumption Variables Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Measures 1 2 3 4 5 6 7 8 9 M SD 1. Approximate family income – 0.23** 0.19** 0.01 0.00 −0.20** −0.09* −0.08 −0.18** 2. Autonomous motivation – 0.50** −0.04 0.01 −0.13** −0.13** 0.05 −0.19** 0.005 1.53 3. Controlled motivation – 0.02 0.00 −0.06 0.01 0.18** −0.02 −0.004 1.26 4. Hedonic hunger (between) – −0.00 0.02 0.21** −0.06 0.08 0.017 0.69 5. Hedonic hunger (within) – 0.04 0.05 0.15** 0.07 −0.020 0.40 6. Sweet – 0.32** 0.22** 0.15** 2.07 1.05 7. Fatty – 0.12** 0.43** 1.80 0.93 8. Starchy – 0.14** 2.27 1.05 9. Fast food – 1.47 0.84 Means for measures 2–5 are derived from centered variables. Measures 2–3 indicate scores on subscales of TSRQ-D administered at baseline. Measures 4–5 are derived from daily reported scores on the PFS. Measures 6–9 indicate self-reported servings of respective foods, ranging from 1 to 5 or more. M mean; SD standard deviation; TSRQ-D Treatment Self-Regulation Questionnaire–Diet; PFS Power of Food Scale. *p < .05, ** p < .01. View Large Table 3 Associations of Predictors and Covariates With Food Consumption-Dependent Variables Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Intercepts and slopes are reported from the final models for each dependent variable. Each column represents a separate model. Bold text denotes significant effects. BP between-person; WP within-person. View Large Table 3 Associations of Predictors and Covariates With Food Consumption-Dependent Variables Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Sweet Starchy Fatty Fast food β (SE) p β (SE) p β (SE) p β (SE) p Fixed effects  Intercept 1.98 (0.45) <.0001 2.52 (0.59) <.0001 2.13 (0.42) <.0001 2.08 (0.41) <.0001  Between-person   Gender 0.15 (0.17) .39 −0.02 (0.22) .93 −0.06 (0.16) .71 −0.14 (0.16) .39   Income −0.09 (0.05) .07 −0.05 (0.06) .42 −0.04 (0.04) .33 −0.06 (0.04) .16   BP hedonic hunger 0.15 (0.13) .25 −0.00 (0.17) .98 0.28 (0.12) .03 0.11 (0.12) .38   Autonomous motivation −0.05 (0.06) .46 0.01 (0.08) .94 −0.11 (0.06) .06 −0.14 (0.06) .02   Controlled motivation −0.00 (0.07) .98 0.13 (0.09) .17 0.03 (0.07) .61 0.07 (0.07) .30  Within-person   Time −0.01 (0.00) .01 0.01 (0.01) .48 −0.00 (0.01) .87 –   WP hedonic hunger 0.20 (0.11) .06 0.38 (0.10) <.0001 0.05 (0.09) .57 0.14 (0.08) .08  Cross-level interactions   WP hedonic hunger × Autonomous motivation 0.00 (0.07) .98 −0.09 (0.06) .18 −0.03 (0.06) .61 -0.10 (0.05) .05   WP hedonic hunger × Controlled motivation 0.07 (0.08) .41 0.17 (0.08) .02 −0.06 (0.07) .34 0.09 (0.06) .15  Random effects   (1,1) Intercept variance 0.23 (0.14) .04 0.52 (0.16) .00 0.37 (0.12) <.001 0.20 (0.06) <.0001   (2,2) Time slope variance 0.02 (0.01) .03 0.00 (0.00) .03 0.00 (0.00) .00 –   (2,1) Intercept–Time slope covariance −0.03 (0.03) .27 −0.01 (0.01) .12 −0.02 (0.01) .02 –   (3,3) Random quadratic time variance 0.00 (0.00) .04 – – –   (3,2) Random quadratic time slope covariance −0.00 (0.00) .07 – – –   (3,1) Intercept– Random quadratic time slope covariance 0.00 (0.00) .24 – – –   Residual 0.78 (0.06) <.0001 0.64 (0.04) <.0001 0.56 (0.04) <.0001 0.49 (0.03) <.0001 Intercepts and slopes are reported from the final models for each dependent variable. Each column represents a separate model. Bold text denotes significant effects. BP between-person; WP within-person. View Large Variability, Effects of Time, and Associations With Dependent Variables Hedonic hunger The ICC for hedonic hunger indicated that 66.7% of the variability was BP and 33.3% of the variability was WP. A random quadratic effect of time was established for the hedonic hunger variable, indicating that hedonic hunger plateaued randomly for individuals across the study period. Sweet food consumption The ICC for the sweet dependent variable indicated that 21.4% of the variability was BP, and therefore, 78.7% was WP. The intercept for the empty model was 2.01, and a random quadratic effect of time was established for the dependent variable of sweet food consumption. No significant associations were found between the independent variables and sweet food consumption, and no significant interactions were found. Starchy food consumption The ICC for the starchy food dependent variable indicated that 34.5% of the variability was BP and 65.5% was WP. The intercept for this variable in the empty model was 2.26, and a random linear effect of time was established for the dependent variable of starchy food consumption. WP hedonic hunger was positively associated with consumption of starchy foods (β = .38, p < .0001), such that individuals experiencing higher hedonic hunger than they typically experienced reported consuming more servings of starchy foods. The interaction term of WP hedonic hunger and controlled motivation was also positively associated with starchy food consumption (β = .17, p = .02). Results of probing significant interactions to interpret the conditional effects indicated that, at higher levels of hedonic hunger, the slope relating controlled motivation to starchy food consumption was more strongly positive. At the conditional value of hedonic hunger 1 SD below the mean, the simple slope was 0.06 (p = .55, not significant). At the mean of hedonic hunger, the simple slope was 0.13 (p = .17, not significant). At the conditional value of hedonic hunger 1 SD above the mean, the simple slope was 0.20 (p = .04, significant), indicating that controlled motivation was a significant predictor of consumption of more servings of starchy foods, at higher levels of hedonic hunger. The region of significance for the moderator (hedonic hunger) ranged from −5.42 to 0.35, indicating that any given simple slope outside of this range was statistically significant (Fig. 2). Given that the centered WP hedonic hunger variable had a mean of −0.02 and a standard deviation of 0.40, this indicated that the effect of controlled motivation on starchy food consumption was significant only for relatively high observed values of hedonic hunger. Fig. 2. View largeDownload slide Mean plot illustrating the interaction of controlled motivation and hedonic hunger predicting starchy food consumption (left). The interaction was significant for “High hedonic hunger,” meaning that the top dotted line represents a significant positive relationship between controlled dietary motivation and starchy food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted parallel lines mark the regions of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −5.42 and higher than 0.35. Fig. 2. View largeDownload slide Mean plot illustrating the interaction of controlled motivation and hedonic hunger predicting starchy food consumption (left). The interaction was significant for “High hedonic hunger,” meaning that the top dotted line represents a significant positive relationship between controlled dietary motivation and starchy food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted parallel lines mark the regions of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −5.42 and higher than 0.35. Fatty food consumption The ICC for the fatty food dependent variable indicated that 27.5% of the variability was BP, and 72.5% was WP. The intercept for the empty model was 1.73, and a random linear effect of time was established for the dependent variable of fatty food consumption. BP hedonic hunger was positively associated with consumption of fatty foods (β = .28, p = .03), such that individuals who reported higher levels of hedonic hunger than others also reported consuming more servings of fatty foods. No significant interactions were found to be predictors of fatty food consumption. Fast food consumption The ICC for the fast food dependent variable indicated that 31.8% of the variability was BP and 68.3% was WP. The intercept for the empty model was 1.46, and none of the alternate models for time fit better than the empty model. Autonomous motivation was negatively associated with consumption of fast foods (β = −.14, p = .02). Additionally, the interaction term of WP hedonic hunger and autonomous motivation was negatively associated with fast food consumption (β = −.10, p < .05). Results of probing significant interactions to interpret the conditional effects indicated that, at higher levels of hedonic hunger, the slope relating autonomous motivation to fast food consumption was more strongly negative. At the conditional value of hedonic hunger one standard deviation below the mean, the simple slope was −0.10 (p = .11, not significant). At the mean of hedonic hunger, the simple slope was −0.14 (p = .02, significant). At the conditional value of hedonic hunger one standard deviation above the mean, the simple slope was −0.18 (p = .004, significant), indicating that autonomous motivation was a significant predictor of consumption of fewer servings of fast foods, at average or higher levels of hedonic hunger. The region of significance for the moderator (hedonic hunger) ranged from −0.22 to 420.79, indicating that any given simple slope outside of this range was statistically significant (Fig. 3). Fig. 3. View largeDownload slide Mean plot illustrating the interaction of autonomous motivation and hedonic hunger predicting fast food consumption (left). The interaction was significant for “Average hedonic hunger” and “High hedonic hunger,” meaning that the top two dotted lines represent significant negative relationships between autonomous dietary motivation and fast food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted line marks the region of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −0.22. As the upper bound is 420.79, this dotted line is not pictured in the figure. Fig. 3. View largeDownload slide Mean plot illustrating the interaction of autonomous motivation and hedonic hunger predicting fast food consumption (left). The interaction was significant for “Average hedonic hunger” and “High hedonic hunger,” meaning that the top two dotted lines represent significant negative relationships between autonomous dietary motivation and fast food consumption. Plot illustrating confidence bands for observed sample values of hedonic hunger (right). The dotted line marks the region of significance, meaning that simple slopes are significant at values of WP hedonic hunger lower than −0.22. As the upper bound is 420.79, this dotted line is not pictured in the figure. Discussion This study aimed to determine whether hedonic hunger includes a state function, in addition to its current conceptualization as a variable with a trait function, and to examine dietary motivation and hedonic hunger as predictors of adolescents’ consumption of specific types of palatable food. The first hypothesis, that hedonic hunger was expected to vary over time and be modeled best by including both BP and WP variabilities, was supported. Hedonic hunger did, in fact, demonstrate both state and trait properties (66.7%, BP, 33.3% WP variability). The BP variability helps to explain why lab-based protocols are able to detect the effect of hedonic hunger at the group level with a single observation [20, 22, 23]. The WP variability observed in this study answers calls to examine hedonic hunger as a temporally fluctuating variable and highlights the importance of including state conceptualizations of the construct in future research protocols [34, 35]. The evidence that one-third of the variability in hedonic hunger exists WP indicates that continuing to study the variable solely as a trait, and only with BP variability, would ignore a substantial portion of the phenomenon. This observation of both BP and WP variabilities may be considered the necessary first step in establishing the two-part process involved in the construct of hedonic hunger. Additionally, Hypothesis 2, that BP and WP hedonic hunger would be positively related to palatable food consumption, was partially supported. Results indicated that BP hedonic hunger predicted fatty food consumption and WP hedonic hunger predicted starchy food consumption. This indicates that adolescents’ consumption of palatable food may be differentially influenced by whether hedonic hunger is conceptualized as a state or trait variable. That is, adolescents who experience higher hedonic hunger than their peers may be more likely to consume fatty foods, which aligns with previous research associating hedonic hunger with higher unhealthy snack intake in adolescents [36]. On the other hand, the current findings suggest that any adolescent, regardless of how their hedonic hunger compares to their peers’, may be susceptible to consumption of starchy food in association with a spike in their own hedonic hunger. Some researchers have begun to assess the food environment through EMA protocols [35], but this is the first known study to examine WP fluctuations in hedonic hunger in association with food intake. Therefore, future investigations that conceptualize hedonic hunger as a trait variable subject to state fluctuations are needed to cover the range of influences exerted by the construct. The third hypothesis, that autonomous dietary motivation would be negatively related to palatable food consumption, was also partially supported, with results indicating that autonomous motivation was negatively related to consumption of fast foods. This fits with current literature on dietary motivation, which suggests that intrinsic motivation to consume a healthy diet is associated with healthier food choices and the ability to resist unhealthy foods [7, 36]. Regarding the effect of the interaction between autonomous motivation and WP hedonic hunger, the significant interaction predicting fast food consumption was not entirely consistent with Hypothesis 5, as the interaction was significant at mean and higher levels of hedonic hunger, rather than low levels. This suggests that adolescents with high intrinsic motivation to consume a healthful diet may be able to resist the influence of hedonic hunger, even when it is higher than usual, and still ultimately consume fewer servings of fast food. These findings suggest that at higher levels of hedonic hunger, autonomous motivation may be stronger. Results from a qualitative study align with this finding, in that adolescents expressed opinions about taking more autonomous responsibility for healthy food choices after having experienced incidents where fast food made them feel ill or negatively affected their functioning [37]. While unexpected, this finding provides valuable information that fluctuations in hedonic hunger may have important associations with the behavior of adolescents who have low autonomous motivation for a healthy diet. Findings were inconsistent with the fourth hypothesis, that controlled dietary motivation would be positively related to palatable food consumption. Controlled dietary motivation was not a significant predictor of any of the palatable food consumption. It is possible that controlled motivation may not always be directly related to choices that negatively influence health, but that the choices it drives do not foster feelings of satisfaction or of worth [38]. Thus, the effects associated with adolescents’ extrinsically driven dietary choices merit further attention. However, controlled motivation and WP hedonic hunger interacted to predict starchy food consumption, which was consistent with hypothesis 5 (i.e., WP hedonic hunger would moderate relationships between dietary motivation and palatable food consumption) and provided support for WP hedonic hunger as a moderator. Adolescents with high controlled dietary motivation who also experienced higher hedonic hunger than was typical for them reported consuming more servings of starchy foods. This suggests that adolescents with externally motivated reasons for consuming a healthful diet may be more vulnerable to the influence of hedonic hunger and may consume more servings of starchy foods. This finding helps identify the specific combination of controlled dietary motivation and high WP hedonic hunger as a factor that may be related to the consumption of palatable starchy foods. To summarize, the moderation effects of Hypothesis 5 were partially supported, as WP hedonic hunger moderated the respective relationships between controlled motivation and starchy food consumption, and the relationship between autonomous motivation and fast food consumption. Findings from this study align with those of one study that examined a similar combination of variables [36] with the added novelty of considering the effects of adolescents’ individual time-varying fluctuations in hedonic hunger as well as using the variables of interest to predict consumption of particular types of palatable foods. Results of this study confirm the importance of dietary motivation as indicated by prior studies [9, 29] and contribute evidence that hedonic hunger is also a significant time-varying factor that may account for choices in food consumption. The significant multilevel interactions confirm that unique relationships exist between trait dietary motivation and fluctuating hedonic hunger, and that the interactions of these variables on an individual level may hold value in understanding and addressing unhealthful dietary choices (i.e., eating palatable foods in excess). Clinical Implications Findings from this study indicate that adolescents may be capable of resisting the influence of hedonic hunger for fast food most of the time if they hold strong intrinsic motivation to eat a healthy diet. It is possible that having autonomous motivation for a healthy diet would protect an adolescent from engaging in a deliberate and planned unhealthy behavior (e.g., taking a drive to purchase fast food), even when he or she is experiencing high hedonic hunger. Alternatively, autonomous motivation for diet in adolescents may be related to values of the family household, as certain parents may model values of maintaining a healthful diet and therefore do not typically facilitate access to fast food. In contrast, the interaction effect may not be present for other food types because they may be readily available in an adolescent’s home or school, and more subject to impulsive consumption. While dietary motivation does not appear to fluctuate rapidly, there exists some evidence that novel clinical strategies may allow for shifts in motivation over extended periods of time. For example, as mindfulness has been shown to play a role in the development of autonomous regulation and motivation [38], clinical exercises to promote mindful eating strategies may be useful [39, 40]. Clinical efforts for health promotion and prevention could also encourage autonomous motivation through nutrition education to understand the impact of dietary choices on one’s own health and well-being as a means of decreasing fast food consumption. With regard to fatty and starchy foods, autonomous motivation does not appear to serve as a protective factor. Therefore, stimulus control efforts such as removing these foods from the home, storing them in infrequently accessed locations, and avoiding purchasing them at the store are likely to be helpful intervention strategies. Limitations This study used self-reported food consumption and is limited by the fact that our method of assessing palatable food intake was not previously validated. While the food consumption categories were derived from a well-validated measure, we cannot ensure that all adolescents interpreted the items similarly. Although survey items instructed adolescents in how to categorize food consumption, it is possible that the particular categories of sugary and starchy overlapped, and that the categories of fatty and fast food overlapped, potentially leading to variation among adolescents’ categorization of certain foods. Another limitation of the self-reported food consumption was the expectation that adolescents could estimate the amount constituting a serving of the foods indicated. It is possible that adolescents may have underestimated or overestimated serving sizes, resulting in some variation in the amount of food consumption reported. Overall, accurately assessing food intake is a consistent challenge in dietary research, as most individuals tend to underreport their intake, particularly for highly palatable energy-dense foods [41, 42]. Based on past research regarding such underreporting, it is possible that adolescents’ self-report of palatable food consumption in this study was lower than what they truly consumed [41, 42]. The limitations associated with the measurement of palatable food consumption could be improved by using a 24-hr dietary recall to overlap with EMA dietary assessment in future studies. Additionally, while we chose a highly reliable and well-validated measure of dietary motivation (i.e., TSRQ-D; 29), the fact that the particular measure has not been validated in an adolescent sample is a limitation of the present study. Regarding other measurements in the study, it is possible that recency of eating could have influenced daily ratings of hedonic hunger. Although participants were asked to consider their experience of hedonic hunger over the day, it is possible that their rating could be related to their current levels of hunger or satiety. In relation to this, our findings may be limited by the use of only one daily assessment. Additional assessments to evaluate fluctuations within a given day may enhance understanding of real-time fluctuations. Lastly, homogeneity of the sample limits generalizability of this study’s findings. Though recruitment efforts were made to recruit a diverse sample with respect to gender, race, and SES, the resulting sample was predominantly Caucasian and upper-middle class. In relation to this, though our exclusion of adolescents with substantial physical activity limitations was helpful in obtaining our targeted sample, we did not screen participants for particular conditions that may influence eating habits (e.g., diabetes). Future studies could be improved by accounting for such conditions in the research design and analysis. Conclusions and Future Direction This study enhances current knowledge about the function of hedonic hunger as a variable through evidence that it does vary over time and includes substantial BP and WP variabilities. Future studies may more closely examine fluctuations in hedonic hunger by testing whether it changes at different times of day, or in relation to an individual’s daily experiences and previous food consumption. Moreover, this study presents a novel investigation of the relationship between trait dietary motivation and time-varying hedonic hunger, which were found to predict palatable food consumption. While findings about the association between BMI and hedonic hunger have been mixed [31, 43], now that the initial relationships between state and trait hedonic hunger have been investigated, a larger study with appropriate stratification of BMI may include weight status as a predictor variable. Future research may continue to study the TSRQ-D as a measure of dietary motivation in youth samples. The present study considered dietary motivation as a stable construct, not expected to fluctuate day to day within an individual, and focused on examination of such fluctuations specifically in hedonic hunger. However, continued research may allow us to examine whether dietary motivation may also have a WP component or confirm whether it is primarily a BP variable that does not typically fluctuate. Our finding involving autonomous motivation in particular may further support past research in which the combination of inhibitory control and high hedonic hunger predicted greater intake of palatable food [22]. This line of research may gain clarity through inclusion of hedonic hunger, dietary motivation, and inhibitory control together in subsequent studies. Additionally, a variety of other factors may influence state-based shifts in hedonic hunger (e.g., exposure to food cues, recency of consumption), and these should also be included in future investigations. It is recommended that future studies continue to examine adolescents’ dietary behavior with more advanced measures of food consumption, such as three-day dietary recall [44], as well as examine these particular relationships in more diverse samples. Future research should also continue to examine hedonic hunger as both a BP and WP variable, and seek to determine whether the respective uses of the variable differentially predict various dietary choices. Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards Authors Carolina M. Bejarano and Christopher C. Cushing declare that they have no conflict of interest. All procedures, including the informed consent process, were conducted in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. Compliance with Ethical Standards Authors’ Contributions Carolina Bejarano assisted in the conceptualization of the project, led the writing of the document, and led reporting of the findings. Dr. Cushing obtained funding for the work, conceptualized the project, contributed substantial original writing, provided the statistical and methodological design, conducted the statistical analyses, collected the data, edited the work, and supervised Carolina Bejarano's activities. Ethical Approval The current study involved human participants, and all procedures were approved and monitored by the Institutional Review Board at the University of Kansas. No animal subjects were used. Informed Consent Informed consent was obtained from all guardians of all adolescents under the age of 18, and the minor child also provided assent to participate. When the participant was over the age of 18, they provided consent for themselves. Acknowledgements This research was supported in part by a Targeted Research Grant from the Society of Pediatric Psychology awarded to C.C.C. References 1. United States Department of Agriculture . Dietary guidelines for Americans ; 2015 . Available at http://health.gov/dietaryguidelines/2015/guidelines/. Accessibility verified February 12, 2016 . 2. Stok FM , De Ridder DT , Adriaanse MA , De Wit JB . Looking cool or attaining self-rule. Different motives for autonomy and their effects on unhealthy snack purchase . Appetite . 2010 ; 54 ( 3 ): 607 – 610 . Google Scholar CrossRef Search ADS PubMed 3. Reedy J , Krebs-Smith SM . Dietary sources of energy, solid fats, and added sugars among children and adolescents in the United States . J Am Diet Assoc . 2010 ; 110 ( 10 ): 1477 – 1484 . Google Scholar CrossRef Search ADS PubMed 4. Iannotti RJ , Wang J . Trends in physical activity, sedentary behavior, diet, and BMI among US adolescents, 2001–2009 . Pediatrics . 2013 ; 132 ( 4 ): 606 – 614 . Google Scholar CrossRef Search ADS PubMed 5. Günther AL , Schulze MB , Kroke A et al. Early diet and later cancer risk: Prospective associations of dietary patterns during critical periods of childhood with the GH-IGF Axis, insulin resistance and body fatness in younger adulthood . Nutr Cancer . 2015 ; 67 ( 6 ): 877 – 892 . Google Scholar CrossRef Search ADS PubMed 6. Deci EL , Ryan RM. Intrinsic Motivation and Self-determination in Human Behavior . New York, NY : Plenum ; 1985 . Google Scholar CrossRef Search ADS 7. Hartmann C , Dohle S , Siegrist M . A self-determination theory approach to adults’ healthy body weight motivation: A longitudinal study focussing on food choices and recreational physical activity . Psychol Health . 2015 ; 30 ( 8 ): 924 – 948 . Google Scholar CrossRef Search ADS PubMed 8. Leong SL , Madden C , Gray A , Horwath C . Self-determined, autonomous regulation of eating behavior is related to lower body mass index in a nationwide survey of middle-aged women . J Acad Nutr Diet . 2012 ; 112 ( 9 ): 1337 – 1346 . Google Scholar CrossRef Search ADS PubMed 9. Niermann CY , Kremers SP , Renner B , Woll A . Family health climate and adolescents’ physical activity and healthy eating: A cross-sectional study with mother-father-adolescent triads . PLoS One . 2015 ; 10 ( 11 ): e0143599 . Google Scholar CrossRef Search ADS PubMed 10. Rutten GM , Meis JJ , Hendriks MR , Hamers FJ , Veenhof C , Kremers SP . The contribution of lifestyle coaching of overweight patients in primary care to more autonomous motivation for physical activity and healthy dietary behaviour: Results of a longitudinal study . Int J Behav Nutr Phys Act . 2014 ; 11(1) : 86 . Google Scholar CrossRef Search ADS 11. Lowe MR , Butryn ML . Hedonic hunger: A new dimension of appetite ? Physiol Behav . 2007 ; 91 ( 4 ): 432 – 439 . Google Scholar CrossRef Search ADS PubMed 12. Corsica JA , Spring BJ . Carbohydrate craving: A double-blind, placebo-controlled test of the self-medication hypothesis . Eat Behav . 2008 ; 9 ( 4 ): 447 – 454 . Google Scholar CrossRef Search ADS PubMed 13. Erlanson‐Albertsson C . How palatable food disrupts appetite regulation . Basic Clin Pharmacol Toxicol . 2005 ; 97(2) : 61 – 73 . Google Scholar CrossRef Search ADS 14. Oliver G , Wardle J . Perceived effects of stress on food choice . Physiol Behav . 1999 ; 66 ( 3 ): 511 – 515 . Google Scholar CrossRef Search ADS PubMed 15. White MA , Whisenhunt BL , Williamson DA , Greenway FL , Netemeyer RG . Development and validation of the food-craving inventory . Obes Res . 2002 ; 10 ( 2 ): 107 – 114 . Google Scholar CrossRef Search ADS PubMed 16. Borradaile KE , Sherman S , Vander Veur SS et al. Snacking in children: The role of urban corner stores . Pediatrics . 2009 ; 124 ( 5 ): 1293 – 1298 . Google Scholar CrossRef Search ADS PubMed 17. Gorin AA , Crane MM . The obesogenic environment . In: Jelalian E , Steele R , eds. Handbook of Childhood and Adolescent Obesity . New York, NY: Springer US ; 2008 : 145 – 161 . Google Scholar CrossRef Search ADS 18. Lowe MR , Butryn ML , Didie ER et al. The power of food scale. A new measure of the psychological influence of the food environment . Appetite . 2009 ; 53 ( 1 ): 114 – 118 . Google Scholar CrossRef Search ADS PubMed 19. Witt AA , Lowe MR . Hedonic hunger and binge eating among women with eating disorders . Int J Eat Disord . 2014 ; 47 ( 3 ): 273 – 280 . Google Scholar CrossRef Search ADS PubMed 20. Witt AA , Raggio GA , Butryn ML , Lowe MR . Do hunger and exposure to food affect scores on a measure of hedonic hunger? An experimental study . Appetite . 2014 ; 74 : 1 – 5 . Google Scholar CrossRef Search ADS PubMed 21. Cushing CC , Benoit SC , Peugh JL , Reiter-Purtill J , Inge TH , Zeller MH . Longitudinal trends in hedonic hunger after Roux-en-Y gastric bypass in adolescents . Surg Obes Relat Dis . 2014 ; 10 ( 1 ): 125 – 130 . Google Scholar CrossRef Search ADS PubMed 22. Appelhans BM , Woolf K , Pagoto SL , Schneider KL , Whited MC , Liebman R . Inhibiting food reward: Delay discounting, food reward sensitivity, and palatable food intake in overweight and obese women . Obesity (Silver Spring) . 2011 ; 19 ( 11 ): 2175 – 2182 . Google Scholar CrossRef Search ADS PubMed 23. Ely AV , Howard J , Lowe MR . Delayed discounting and hedonic hunger in the prediction of lab-based eating behavior . Eat Behav . 2015 ; 19 : 72 – 75 . Google Scholar CrossRef Search ADS PubMed 24. Laurent JS . Psychometric properties for the children’s power of food scale in a diverse sample of pre-adolescent youth . Appl Nurs Res . 2015 ; 28 ( 2 ): 127 – 131 . Google Scholar CrossRef Search ADS PubMed 25. Laurent JS , Sibold J . Addictive-like eating, body mass index, and psychological correlates in a community sample of preadolescents . J Pediatr Health Care . 2016 ; 30 ( 3 ): 216 – 223 . Google Scholar CrossRef Search ADS PubMed 26. Stone AA , Shiffman S . Ecological momentary assessment (EMA) in behavioral medicine . Ann Behav Med . 1994 ; 16(3) : 199 – 202 . 27. Brannon EE , Cushing CC , Crick CJ , Mitchell TB . The promise of wearable sensors and ecological momentary assessment measures for dynamical systems modeling in adolescents: A feasibility and acceptability study . Transl Behav Med . 2016 ; 6 ( 4 ): 558 – 565 . doi: 10.1007/s13142-016-0442-4 Google Scholar CrossRef Search ADS PubMed 28. Centers for Disease Control and Prevention (CDC) . Growth chart training: A SAS program for the CDC growth charts [Computer software] . Available at http://www.cdc.gov/growth charts/computer_programs.htm. Accessibility verified July 12, 2016 . 29. Levesque CS , Williams GC , Elliot D , Pickering MA , Bodenhamer B , Finley PJ . Validating the theoretical structure of the treatment self-regulation questionnaire (TSRQ) across three different health behaviors . Health Educ Res . 2007 ; 22 ( 5 ): 691 – 702 . Google Scholar CrossRef Search ADS PubMed 30. Cappelleri JC , Bushmakin AG , Gerber RA et al. Evaluating the power of food scale in obese subjects and a general sample of individuals: Development and measurement properties . Int J Obes (Lond) . 2009 ; 33 ( 8 ): 913 – 922 . Google Scholar CrossRef Search ADS PubMed 31. Mitchell TB , Cushing CC , Amaro CM . Psychometric properties of the power of food scale in a community sample of preadolescents and adolescents . J Child Family Stud . 2016; 25(9):2733–2739 32. Cushing CC, Marker AM, Bejarano CM, Crick CJ, Huffhines LP. Latent variable mixture modeling of ecological momentary assessment data: Implications for screening and adolescent mood profiles. J Child Family Stud. 2017; 26(6):1565–1572. 33. Preacher KJ , Curran PJ , Bauer DJ . Computational tools for probing interactions in multiple linear regression, multilevel modeling, and latent curve analysis . J Educ Behav Stat . 2006 ; 31(4) : 437 – 448 . Google Scholar CrossRef Search ADS 34. Boggiano MM , Wenger LE , Turan B et al. Real-time sampling of reasons for hedonic food consumption: Further validation of the Palatable Eating Motives Scale . Front Psychol . 2015 ; 6 : 744 . Google Scholar CrossRef Search ADS PubMed 35. Thomas JG , Doshi S , Crosby RD , Lowe MR . Ecological momentary assessment of obesogenic eating behavior: Combining person-specific and environmental predictors . Obesity (Silver Spring) . 2011 ; 19 ( 8 ): 1574 – 1579 . Google Scholar CrossRef Search ADS PubMed 36. Stok FM , De Vet E , Wardle J , Chu MT , De Wit J , De Ridder DT . Navigating the obesogenic environment: How psychological sensitivity to the food environment and self-regulatory competence are associated with adolescent unhealthy snacking . Eat Behav . 2015 ; 17 : 19 – 22 . Google Scholar CrossRef Search ADS PubMed 37. Bassett R , Chapman GE , Beagan BL . Autonomy and control: The co-construction of adolescent food choice . Appetite . 2008 ; 50 ( 2–3 ): 325 – 332 . Google Scholar CrossRef Search ADS PubMed 38. Deci EL , Ryan RM . Self-determination theory: A macrotheory of human motivation, development, and health . Can Psychol . 2008 ; 49(3) : 182 . Google Scholar CrossRef Search ADS 39. Dalen J , Brody JL , Staples JK , Sedillo D . A Conceptual framework for the expansion of behavioral interventions for youth obesity: A family-based mindful eating approach . Child Obes . 2015 ; 11 ( 5 ): 577 – 584 . Google Scholar CrossRef Search ADS PubMed 40. Forman EM , Butryn ML . A new look at the science of weight control: How acceptance and commitment strategies can address the challenge of self-regulation . Appetite . 2015 ; 84(1) : 171 – 180 . Google Scholar CrossRef Search ADS 41. Johansson G , Wikman A , Ahrén AM , Hallmans G , Johansson I . Underreporting of energy intake in repeated 24-hour recalls related to gender, age, weight status, day of interview, educational level, reported food intake, smoking habits and area of living . Public Health Nutr . 2001 ; 4 ( 4 ): 919 – 927 . Google Scholar CrossRef Search ADS PubMed 42. Macdiarmid J , Blundell J . Assessing dietary intake: Who, what and why of under-reporting . Nutr Res Rev . 1998 ; 11 ( 2 ): 231 – 253 . Google Scholar CrossRef Search ADS PubMed 43. Schultes B , Ernst B , Wilms B , Thurnheer M , Hallschmid M . Hedonic hunger is increased in severely obese patients and is reduced after gastric bypass surgery . Am J Clin Nutr . 2010 ; 92 ( 2 ): 277 – 283 . Google Scholar CrossRef Search ADS PubMed 44. Subar AF , Freedman LS , Tooze JA et al. Addressing current criticism regarding the value of self-report dietary data . J Nutr . 2015 ; 145 ( 12 ): 2639 – 2645 . Google Scholar CrossRef Search ADS PubMed © Society of Behavioral Medicine 2018. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Journal

Annals of Behavioral MedicineOxford University Press

Published: Feb 7, 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