TY - JOUR
AU1 - Hetsroni,, Amir
AU2 - Tukachinsky, Riva, H.
AB - Abstract This study proposes a new scheme for cultivation based on measures of television viewing and the relationship between TV-world estimates and real-world estimates as they are examined in three topics—criminality prevalence, the share of violent crimes, and the number of old people. Content analysis of prime-time and off prime–time programming (210 hours) and a survey of viewers ( N = 591) form the data set. A model that covers 85% of the respondents, and is composed of five groups of viewers, is suggested. The groups are specified as simple cultivation (when estimation of the real world is biased but does match a correct estimation of the television world), over cultivation (when the real world is seen as a replica of the TV world but estimation of television reality is exaggerated), double distortion (when TV reality and the real world are both exaggeratedly estimated), simple no cultivation (when both the real world and the TV world are correctly estimated), and distorted no cultivation (when estimation of the real world is correct but TV reality is incorrectly estimated). The groups are differentiated by the amount of television viewing as the heaviest viewers are in the overcultivation group, and the lightest viewers are in the distorted no cultivation group. These results hold when demographics and consumption of media other than television are controlled for. Since cultivation theory was introduced by Gerbner and his associates in the 1970s (Gerbner, 1972; Gerbner & Gross, 1976; Gerbner, Gross, Jackson-Beeck, Jeffries-Fox, & Signorielli, 1978; Gerbner et al., 1977), it became the subject of a heated public debate and a high-pitched academic discourse. Several replications, attacks, and responses have been performed (see, e.g., Doob & McDonald, 1979; Hirsch, 1980; Hughes, 1980; Newcomb, 1978; Potter, 1986, 1988, 1991a, 1991b). The most consistent conclusion that stems from a meta-analysis of over 80 studies is a confirmation of Gerbner's proposition that television viewing is correlated with a line of distorted estimates of social reality (Morgan & Shanahan, 1997; Shanahan & Morgan, 1999; Weimann, 2000). Evidence of distortion has been detected in a wide range of topics. Among them are estimates of the rate of crime and violence (Gerbner et al., 1977; Ogles & Sparks, 1989; Potter, 1991a, 1991b, 1991c); estimates about personal victimization (Gerbner & Gross; Hawkins & Pingree, 1980; Morgan, 1983; Ogles & Sparks; Weaver & Wakshlag, 1986); estimates of life risks posed by lightning, flooding, and terror attacks (Gunter & Wober, 1983); estimates about the share of people employed as policemen (Gerbner & Gross) and lawyers (Pfau, Mullen, Dietrich, & Garrow, 1995); estimates of the number of women in professional occupations (Carveth & Alexander, 1985); estimates of divorce rates (Carveth & Alexander; Potter, 1991b); estimates of affluence prevalence (Fox & Philliber, 1978; Potter, 1991b); and estimates about the number of old people who live in the world (Gerbner, Gross, Signorielli, & Morgan, 1980). Early cultivation studies took place in the United States, but replications have been successfully conducted in Argentina (Morgan & Shanahan, 1992), Australia (Hawkins & Pingree), Canada (Gosselin, DeGuise, & Paquette, 1997), Hong Kong (Cheung & Chan, 1996), Israel (Cohen & Weimann, 2000; Weimann, 1984), South Korea (Kwak, Zinkhan, & Dominick, 2002), Taiwan (Morgan & Shanahan), Brazil, China, Hungary, the Netherlands, Russia, and Trinidad (Shanahan & Morgan, 1999, p. 4). Even some of the theory's more skeptical reviewers admit that cultivation works, when it comes to positive correlations between television viewing and distorted estimates of the likelihood of occurrences in the real world, commonly known as first-order effects (Weimann, 2000). Evidence in support of the other type of cultivation effects, second-order effects, according to which television viewing does not only cultivate distorted estimates but also contributes to the formation of a suspicious conservative outlook at the world, is weaker and more inconsistent (Doob & McDonald, 1979; Hawkins, Pingree, & Adler, 1987). Researchers are still in doubt whether second-order effects are, indeed, directly related to television viewing, or if they stem from first-order effects (see Potter, 1991c). In the past two decades, the mainstream of cultivation research searched for cognitive explanations how first-order effects occur in order to give a more detailed account of the mechanism than Gerbner's notion that “people simply internalize content from a medium with which they spend so much time” (Shanahan & Morgan, 1999, p. 173). The most often cited models have been Hawkins and Pingree's learning and construction (Hawkins & Pingree, 1980, 1982) and Shrum's heuristic processing (Shrum, 1995, 1996, 2002; Shrum, Burroughs, & Rindfleisch, 2004). These models are divided over the question whether people actively learn facts from the screen—as suggested by Hawkins and Pingree (1982)—or just encode information while viewing in a much unelaborated manner—as claimed by Shrum and O’Guinn (1993, p. 466). Shrum (1995, 1996), actually, insists that reality estimates may be constructed only at the time the judgment is required (e.g., when respondents are asked to give estimates in a study) because little active memorizing occurs when people watch television. However, there is no dispute over the fact that even if some viewers do not intend to use television programs for learning, they still incidentally learn few facts from the programs they watch. Furthermore, reliance on this incidental learning (which is encoded as information in the long-term memory), when estimating the likelihood of occurrences in the real world, is the source of cultivation's first-order effects (Shrum et al., 2004). This effect occurs when viewers, who are asked to estimate the likelihood of occurrences in the real world, give estimates that resemble the reality of typical TV content. In other words, they ascribe TV-world characteristics to the real world. This happens more often to heavy viewers (Shanahan & Morgan, 1999, p. 26). But do television viewers (be they heavy or light in the amount of viewing) really know the TV world well enough to estimate it correctly and later project this estimation on the real world? Potter explains why it is important to see how many of the viewers choose the exact TV-world estimate (known also as the “TV answer”) when asked about the real world: It is critical for cultivation research that the ‘television answer’ be correctly identified, because the analysis requires that the amount of television viewing be associated with the selection of the ‘television answer’. (Potter, 1991a, p. 96) Mapping the range of accurate and inaccurate estimations of the TV world and the real world and recognizing the relationships between the two is crucial, if we want to know to what extent viewers are actually cultivated. Yet, only three works—Hawkins et al. (1987), Potter (1988, 1991a)—attempted to measure viewers’ estimates of TV reality in relation to cultivation. From a methodological standpoint, the three works made an acute flaw by including the two sets of questions (about the television world and the real world) in the same questionnaire since using information obtained from watching television to estimate the frequency of occurrences in the social surrounding is probably an unconscious process (Shanahan & Morgan, 1999; Shrum et al., 2004). The affinity of questions about the television world and the real world may minimize and even eliminate the cultivation effect (Shanahan & Morgan). The three works yielded inconsistent, sometimes peculiar, results. Hawkins et al. detected a positive correlation between TV viewing and real-world estimates and a similar relationship between real-world estimates and TV-world estimates; however, they also found a negative correlation between TV viewing and TV-world estimates. Their interpretation of this finding was that TV-world estimates are unrelated to cultivation and that the positive correlation between TV-world estimates and real-world estimates is just an artifact of asking the two sets of questions so close together (Hawkins et al., 1987, p. 560). Potter's first study found a null correlation between estimates of violent acts on television and estimates of violent acts in the real world (Potter, 1988), whereas his later work discovered a positive correlation between the two estimates, which he interpreted as a sign that viewers do use their estimation of on-screen occurrences to construe their perception of the real world (Potter, 1991a). Beyond the puzzle of the contradicting results, the studies left the accuracy of the estimates obscured. None of them bothered to see whether any of the two types of estimates were true to the objective measures of census data or matched the findings of a content analysis of TV programming, which they did not conduct. The possibility that viewers may hold a biased estimation of the TV world, and that this biased estimation (and not a correct one) is the cultivating factor, is not considered in any previous work. No study ever looked at a situation in which viewers’ estimation of TV reality is biased to begin with. Logically, if this is the case, then coming up with a television answer, when asked about the real world, as demonstrated in so many cultivation studies, should not be identified as a cultivation effect in the traditional sense of the concept because viewers’ estimation of the social world does not resemble their perception of the TV world. If, on the other hand, cultivation works as predicted, which means that the estimates of TV reality and the real world do fall in the same direction, but the perception of TV reality is, nonetheless, erroneous—then the TV-world estimate does not really reflect TV reality, as it is etched in the minds of viewers, even if it does statistically reflect the findings of an objective message system analysis of television content. In such cases, traditional cultivation studies would have failed to detect the exact effect since for them the answer that reflects an accurate estimation of the TV world (the television answer) is the statistical approximation of television reality, even if viewers perceive this “reality” differently.1 For instance, viewers who estimate the TV world as more violent than it really is according to the findings of content analysis, and are cultivated, would estimate the real world as much more violent than it is in practice because they would project their erroneous estimation of the TV world on to the real world, which is far less violent to begin with (Potter et al., 1995). This could happen, for example, to viewers who are primed by the lingering critique of TV violence that comes from politicians (“Violence, irresponsibility and vulgarity are the staples of prime time”—Senator Sam Brownback, quoted in “Three U.S. Senators Speak Out,” 1999, p. 33), religious groups (“More than 3,000 studies over the past 30 years offer noncontradictable evidence that violent programming has a harmful effect on young minds”—“An Enemy in Your Living Room,” 1993), and academic experts (“I’m a researcher who's tired of documenting the problems of TV violence and sleaze and wants to start doing something about it”—Lichter, 1999, p. 37) to believe that the typical television fare is more violent than it really is. And what about viewers who are more inoculated to the aforementioned critique, who do correctly estimate television reality, but are nonetheless cultivated and so project what they see on television on to the real world? These viewers may perceive the real world as violent, however less violent than viewers who hold an exaggerated estimation of the TV world, because their base rate for evaluation is lower. When we bring into the picture viewers who are not cultivated (even Gerbner, Gross, Morgan, & Signorielli (1994) agree that cultivation does not have the same effect on everyone, see p. 27), further potential relationships and combinations of TV-world estimates and real-world estimates merit exploration, as we shall do in this work. (RQ1): Our first research question is therefore: What is the pattern of the relationship between TV-world estimates and real-world estimates: Do different combinations of estimates classify groups of viewers consistently across various topics and under multiple controls? How will the relationship between TV-world estimates and real-world estimates be influenced by the amount of viewing? There is extensive evidence that heavy viewing correlates with a stronger cultivation effect because heavy viewers have more accessible TV information to rely on and project from (Gerbner & Gross, 1976; Gerbner et al., 1994). This effect decreases somewhat, but by no means dies out entirely, when demographic variables and consumption of media other than television are controlled for (Shanahan & Morgan, 1999, pp. 126–135). We, too, predict viewing among cultivated viewers, who see the real world as a replica of the TV world and do estimate TV reality correctly, will be heavier than among noncultivated viewers, who still estimate TV reality correctly. It is harder to predict the amount of viewing of those viewers who do not estimate TV reality correctly (e.g., those who overestimate the TV world) since this condition is dealt with for the first time in this study. Logically, one might have expected that the more people viewed, the more accurate their perception of what they saw on television would have been. Yet, this logical expectation may not always be true since the TV diet of heavy viewers is often a mix of very different formats and programs that form together a sort of clutter (Weimann, Brosius, & Wober, 1992). Heavy viewers could be persons for whom an open TV set is a tapestry, which they absorb with only limited attention. Television could still, unconsciously, influence their social perception, but they would be likely unable to estimate correctly what they often see on the screen. In light of all that, our second research question (RQ2) is: Does the amount of television viewing relate to the classification of TV-world estimates and real-world estimates from RQ1? More in particular: RQ2a: Is there a stable (across topics) relationship between the amount of viewing and the different combinations of estimates? RQ2b: Does this relationship hold when we control for demographic variables and consumption of media other than television? Method The study consists of two phases: content analysis from which we determine the estimates of TV reality and a survey composed of two distinct questionnaires (one about TV reality and one about the real world). Content analysis Cultivation studies (see Gerbner, 1972; Gerbner et al., 1978) and other content analyses of prime-time programming (Greenberg, 1980; Lichter, Lichter, & Rothman, 1994; Signorielli, 2001) provide a rather consistent picture of the occurrences of violence and the demography of characters since the 1970s. Therefore, we had several sources from which to draw our estimates of TV reality. Nevertheless, because those studies analyzed the programming of American networks, and even though the programming of the major Israeli stations is abundant with imports of American films, programs, and formats (Israel Audience Research Board, 2004; Weimann, 1984), we have conducted a new set of content analysis of Israeli programming in order to be more certain of the accuracy of the TV answers. Most cultivation studies have concentrated on violence and crime. We, too, measure that, yet following Potter's (1991b, p. 572) suggestion to include a wider range of topics in order to increase the generalization of the findings, we also code the appearance of the elderly—a topic that has come into sight infrequently in cultivation studies and yielded less than consistent results (Passuth & Lomax-Cook, 1985). The three items we code are (a) the percent of violent crimes like murder, rape, robbery, and aggravated assault from among all the crimes, (b) the percent of characters who are criminals or have a criminal record, and (c) the percent of people aged 65 or more. The data were obtained from a content analysis of early-evening, prime-time, and late-night programs broadcast over 3 weeks in November 2003 on channel 2 and channel 10—Israel's two most highly watched networks with a combined share of over half of the households. We recorded all the programs that were broadcast between 19:00 and 24:00 for 21 consecutive days to obtain a total of 210 hours (105 from each network). Commercials and sport events were excluded from the analysis. The unit of analysis in coding old people and criminals was a character shown on the screen and having a speaking role. In coding violent versus nonviolent crimes, the unit of analysis was an antisocial act that is prohibited by the law and shown on the screen. These definitions follow those of Greenberg (1980) on which many TV content–coding schemes are based (see, e.g., Potter et al., 1995, 1997). The coding was performed by four students, who were trained for 3 hours each and were not privy to the goals of the research. The coders worked alone on the tapes to secure coding independence. Each of the 210 hours of programming was coded by two different coders. Intercoder reliability was computed for the overall sample and was measured through Krippendorff's alpha coefficient. The computation of the coefficient in violent crimes (α= 0.809) and in the number of criminals (α= 0.838) was based on dichotomous coding decisions. The age of characters, from which the data to determine the number of old people were taken, was a polytomous category (less than 18, 19–44, 45–64, 65 and more). The reliability coefficient for the relevant age group (α= 0.945) was higher than the category's overall reliability (α= 0.869). All the above values point at an impressive level of reliability. No significant differences were found between the two networks or between the prime-time hours (8:00–11:00 p.m.), the early-evening shows (19:00–20:00 hours), and the late-night programs (11:00–12:00 p.m.). Therefore, we report the combined figures. Of the total of 1,221 crimes shown on screen, 25.5% were violent (compared to 23% in American programs; Lichter et al., 1994). Of the total of 3,892 characters with a speaking role, 12.5% were criminals (compared to 10% on American television; Greenberg, 1980) and 4.2% were elderly (in comparison to 3% in America; Signorielli, 2001). The findings of our content analysis, thus, seem rather close to the figures obtained from analyses of American programming, which were the basis to determine the “TV reality answer” in cultivation questionnaires. Survey Subjects The respondents (N= 591) were freshmen in an Israeli public college. We made sure that none of them had learned about cultivation previous to the study. The respondents’ age ranged from 19 to 28 with a median of 23. The majority (64%) were females. Their number reflects the share of women in the student body. The ethnic distribution of the sample is nearly similar to that of the overall Israeli population (Central Bureau of Statistics, 2003): Eighty-six percent of our respondents were Jews (82% in the population) and 14% were Arabs (18% in the population). Instrument and procedure Two separate sets of questionnaires were administered in a 2-week interval during February 2004. Each set included estimates of either the real world or the TV world, measures of TV exposure (total viewing and viewing of particular television formats), newspaper reading, Internet use, age, sex, and ethnicity (control variables). Answers were anonymous, but we used the respondents’ date of birth to match the responses from the two questionnaires. To prevent a bias by putting together the questions about the television world with the questions about the real world (as stated previously), the sample was randomly divided into two groups. In the first questionnaire administration, one group answered the real-world questionnaire, whereas the second was asked about the TV world. Two weeks later, the group that had previously answered the questions about the real world estimated the television world, whereas the group that had formerly estimated the television world answered the questions about the real world. A comparison of the answers obtained from the two groups shows that the order of answering the questions about the two worlds had no effect on the pattern of the results: χ2(2)= 3.1, p=.21 for estimating the percent of violent crimes in the real world and χ2(2)= 3.0, p=.22 for estimating this percent in the TV world, χ2(2)= 2.3, p=.32 for estimating share of criminals in the real world and χ2(2)= 2.5, p=.29 for estimating this share in the TV world, and χ2(2)= 1.3, p=.52 for estimating the number of old people in the real world and χ2(2)= 1.7, p=.42 for estimating this number in the TV world. Measures Television viewing.  Each respondent was asked “on an average day, how many hours do you personally watch television?” This open-ended question was used in previous cultivation studies to measure total viewing (Carveth & Alexander, 1985; Gerbner et al., 1978; Gross & Jeffries-Fox, 1978; Morgan, 1983, 1984, 1986; O’Keefe, 1984; Signorielli, 1990; Tan, 1982; Volgy & Schwarz, 1980). We also asked about daily viewing of dramatic series and news programs as these shows achieve high ratings in the country where the study took place (Israel Audience Research Board, 2004), and—probably in light of their content—produce significant cultivation effects in violence and crime (Cohen & Weimann, 2000; Romer, Jamieson, & Aday, 2003). Real-world and television-world estimates.  Three multiple-choice questions were presented. Two of them, which measure the estimation of the share of violent crimes and the prevalence of criminality, are part of the original reliable cultivation index (see Rubin, Palmgreen, & Sypher, 1994, pp. 154–158) and were also used previously in studies of cultivation in Israel (Cohen & Weimann, 2000). The third item that measures viewers’ estimation of the number of old people in the world has also been used previously in cultivation studies and found to be reliable (Gerbner et al., 1980). The same wording was used when asking about the real world and the TV world. Each question had three optional answers (see Appendix A for the exact wording): A real-world answer, which is a close approximation of the social reality, based on the last edition of the Israeli Statistical Abstract (CBS, 2003). A TV answer, closely reflecting the television portrayal of reality, as determined from previous studies (Gerbner et al., 1980; Greenberg, 1980; Lichter et al., 1994; Signorielli, 2001) and from our own content analysis. An over-TV answer, where the biased TV answer (in relation to the real world) is even more exaggerated. The three answers were structured to create an arithmetic series, where the TV answer is the middle value. A few of the numbers have been rounded to look more reasonable in the eyes of “suspicious” respondents (for an extended discussion of this rationale, see Shanahan & Morgan, 1999, pp. 53–54). To prevent a potential response set bias, the optional answers were randomly reordered within every item and throughout all the questionnaires. We used the forced-choice format as a large number of cultivation studies have done (see, e.g., Cohen & Weimann, 2000; Gerbner et al., 1978; Morgan, 1983; Rössler & Brosius, 2001; Slater & Elliott, 1982) for a number of reasons. First, many respondents find it hard to come up with estimates on their own in topics where their prior knowledge is limited (Shanahan & Morgan). Studies about estimates and guestimates reveal a tendency of respondents not to answer open-ended questions and leave the researcher with an ipso-facto smaller and, possibly, less representative sample (Tanur, 1992). Second, forced-choice questions reveal systematic biases in a way that open-ended questions are unable to because some of the potential responses come into mind quickly, whereas others need a trigger (Campbell, 1950). Third, there is even indication that open-ended questions that ask people to give social reality estimates have their indigenous response set bias because few of the respondents tend to give higher or lower estimates in all the items—regardless of the topic (see Shanahan & Morgan, 1999, p.68). In contrast, letting viewers choose between thoughtfully crafted and randomly reordered answers may yield a range of responses that is not only as varied as that obtained from open-ended questions, but is also a closer reflection of the variance amongt the population (Schwartz, 1999). Tanur recommends providing three to seven optional answers to each question in order to improve the chances of reflecting the population's true variance. Our questionnaire offers three answers because this is the simplest option that gives the respondents, few of them may only have some very limited interest in the subject matter, a clear choice between a TV answer, a real-world answer, and an over-TV answer. The results should be able to inform not only how many of the viewers overestimate the real world and the TV world but also how many estimate them correctly. Results RQ1 To examine RQ1, let us look at the relationships between TV-world estimates and real-world estimates across different topics. The cross-tabulation for the respondents’ estimates appears in Table 1. Table 1 Estimates of the Number of Old People, the Share of Violent Crimes, and Criminality Prevalence in the Real World and the Television World (N= 591) Real-World Estimates . Television-World Estimates . Real-World Answer (%) . TV Answer (%) . Over-TV Answer (%) . Old people (χ(4)2= 36.1**, V= 0.264)  Real-world answer 13.1 28.5 13.8  TV answer 4.6 13.1 19.2  Over-TV answer 1.5 0.0 6.2 Violent crimes (χ(4)2= 14.2*, V= 0.146)  Real-world answer 3.1 9.3 11.6  TV answer 0.8 17.1 28.7  Over-TV answer 2.3 7.0 20.2 Criminality prevalence (χ(4)2= 26.4**, V= 0.226)  Real-world answer 9.3 8.5 3.1  TV answer 3.9 27.9 14.7  Over-TV answer 0.8 10.9 20.9 Real-World Estimates . Television-World Estimates . Real-World Answer (%) . TV Answer (%) . Over-TV Answer (%) . Old people (χ(4)2= 36.1**, V= 0.264)  Real-world answer 13.1 28.5 13.8  TV answer 4.6 13.1 19.2  Over-TV answer 1.5 0.0 6.2 Violent crimes (χ(4)2= 14.2*, V= 0.146)  Real-world answer 3.1 9.3 11.6  TV answer 0.8 17.1 28.7  Over-TV answer 2.3 7.0 20.2 Criminality prevalence (χ(4)2= 26.4**, V= 0.226)  Real-world answer 9.3 8.5 3.1  TV answer 3.9 27.9 14.7  Over-TV answer 0.8 10.9 20.9 * p <.01. ** p <.001. Open in new tab Table 1 Estimates of the Number of Old People, the Share of Violent Crimes, and Criminality Prevalence in the Real World and the Television World (N= 591) Real-World Estimates . Television-World Estimates . Real-World Answer (%) . TV Answer (%) . Over-TV Answer (%) . Old people (χ(4)2= 36.1**, V= 0.264)  Real-world answer 13.1 28.5 13.8  TV answer 4.6 13.1 19.2  Over-TV answer 1.5 0.0 6.2 Violent crimes (χ(4)2= 14.2*, V= 0.146)  Real-world answer 3.1 9.3 11.6  TV answer 0.8 17.1 28.7  Over-TV answer 2.3 7.0 20.2 Criminality prevalence (χ(4)2= 26.4**, V= 0.226)  Real-world answer 9.3 8.5 3.1  TV answer 3.9 27.9 14.7  Over-TV answer 0.8 10.9 20.9 Real-World Estimates . Television-World Estimates . Real-World Answer (%) . TV Answer (%) . Over-TV Answer (%) . Old people (χ(4)2= 36.1**, V= 0.264)  Real-world answer 13.1 28.5 13.8  TV answer 4.6 13.1 19.2  Over-TV answer 1.5 0.0 6.2 Violent crimes (χ(4)2= 14.2*, V= 0.146)  Real-world answer 3.1 9.3 11.6  TV answer 0.8 17.1 28.7  Over-TV answer 2.3 7.0 20.2 Criminality prevalence (χ(4)2= 26.4**, V= 0.226)  Real-world answer 9.3 8.5 3.1  TV answer 3.9 27.9 14.7  Over-TV answer 0.8 10.9 20.9 * p <.01. ** p <.001. Open in new tab Few things are noticeable from the table. First of all, the significant chi-square values across all the items indicate that there is a nonrandom relationship between the two types of estimates. Second, most of the respondents do not estimate television reality correctly. They do not choose the TV answer in questions that ask about the TV world. In the case of violent crimes, approximately two thirds choose the wrong estimates. With the elderly, nearly 60% give erroneous answers. In estimating criminality prevalence, the share of correct estimates is higher; yet, the majority of respondents (52.7%) still do not pick up the right TV answer. Next, we come to the heart of the cross-tabulation—what goes with what. Respondents who give the TV answer to questions about the television world and the real world make 17% of the sample in violent crimes, 28% in criminality prevalence, and 13% in the number of old people. They are cultivated in the traditional Gerbnerian sense, for the reason that they view the real world as a replica of the TV world (which they estimate correctly). We suggest calling this group of viewers simple cultivation. A second group of cultivated respondents are those who give over-TV answers to any of the questions—whether they are about the TV world or pertain to the real world. These viewers make approximately one fifth of the sample in violent crimes and criminality prevalence but only 6% in the elderly estimate. Their distorted estimates of the real world resemble their view of the television world. This makes a case for a cultivation effect. However, the exaggerated estimation of both worlds, to an extent that goes beyond the objective measures of TV reality (as determined from content analysis), merits a special consideration. Traditional cultivation studies, which search only for a general bias toward TV-world figures in viewers’ real-world estimates and do not make the distinction between TV answers (which truly represent television reality) and over-TV answers (that pose an exaggeration of that reality), could have misclassified these viewers as simple cultivation. We call them overcultivation for the reason that their estimates exaggerate what we recognize as a simple cultivation effect. Now, we come to the respondents who give over-TV answers to questions about the TV world and TV answers to questions about the real world. Traditional cultivation studies would have, probably, classified these viewers as simple cultivation (because they give a TV answer when asked about the real world), but our data suggest that this is, by all means, not the correct classification. These viewers (28.7% in violent crimes, 14.7% in criminality prevalence, 19.2% in old people) do distinguish the real world from the TV world. Otherwise, they would have estimated both worlds the same way. They are aware of the fact that television programs often feature more violence, more criminals, and fewer old people than there are in the real world. However, these viewers’ estimation of the TV world is so biased to begin with that it distorts their perception of the real world. As a result, the real world appears to them as more violent, more full of criminals, and emptier of old people than it actually is, but still not as violent, full of criminals, and empty of old people as the home screen is. We suggest calling this group of viewers double distortion for the reason that both of their estimates (that of the real world and that of the television world) are distorted without showing a sign of an ordinary cultivation effect. Three additional groups of viewers are not cultivated, nor are they characterized by double distortion. These viewers in the upper row cells of Table 1 are characterized by a correct estimation of the real world. One group is composed of respondents who give TV answers to questions about the TV world and real-world answers to questions about the real world. They are perhaps the kind of viewers one has in mind, when wishing for high media literacy and good knowledge of the social surrounding. Yet, they make only 9% of the viewers in violent crimes and criminality prevalence, and only 28% in the number of old people. We call them simple no cultivation. The second group of noncultivated viewers is composed of respondents who give a real-world estimate when asked about the real world and an over-TV answer when asked about the television world. These respondents make 3%–14% of the sample, depending on the question. The third group is composed of viewers who give a real-world answer to the questions on both worlds. Their share fluctuates between 3% and 13% of the respondents. We term the last two groups distorted no cultivation because their estimation of the TV world is distorted, but their view of the real world is not cultivated. The two cells of distorted no cultivation comprise together 12% of the respondents in criminality prevalence, 15% in violent crimes, and 27% in the number of old people. Later, we will see that the two cells of distorted no cultivation resemble one another but differ from simple no cultivation in the amount of TV viewing. Table 2 summarizes the different combinations of TV-world estimates and real-world estimates that we have just reviewed. This classification of respondents into groups and the numerical ratios between the groups do not change, when we control for sex and ethnic origin variables (tables which detail males’ estimates, females’ estimates, Jewish respondents’ estimates, and Arab respondents’ estimates are available from the authors). Age and education are practically already under control as we use a sample of college students. Table 2 Combinations of Real-World Estimates and Television-World Estimates Real-World Estimates . Television-World Estimates . Real-World Answer . TV Answer . Over-TV Answer . Real-world answer Distorted no cultivation Simple no cultivation Distorted no cultivation TV answer Simple cultivation Double distortion Over-TV answer Overcultivation Real-World Estimates . Television-World Estimates . Real-World Answer . TV Answer . Over-TV Answer . Real-world answer Distorted no cultivation Simple no cultivation Distorted no cultivation TV answer Simple cultivation Double distortion Over-TV answer Overcultivation Open in new tab Table 2 Combinations of Real-World Estimates and Television-World Estimates Real-World Estimates . Television-World Estimates . Real-World Answer . TV Answer . Over-TV Answer . Real-world answer Distorted no cultivation Simple no cultivation Distorted no cultivation TV answer Simple cultivation Double distortion Over-TV answer Overcultivation Real-World Estimates . Television-World Estimates . Real-World Answer . TV Answer . Over-TV Answer . Real-world answer Distorted no cultivation Simple no cultivation Distorted no cultivation TV answer Simple cultivation Double distortion Over-TV answer Overcultivation Open in new tab The combined scheme we suggest as an answer to RQ1—simple cultivation, overcultivation, double distortion, simple no cultivation, and distorted no cultivation—covers about 85% of the respondents in criminality prevalence, nearly 95% of them in the number of old people and around 90% in violent crimes. We cannot explain, at least for the time being, why a small number of viewers (5% of the respondents in the number of old people, 10% in violent crimes, and less than 15% in criminality prevalence—see the empty cells in Table 2) believe that the real world resembles the TV world more than the TV world resembles itself. Perhaps they did not take the questionnaire seriously. An important issue is the stability of the typology. Do the respondents who are classified to a certain group by their estimates in a specific topic keep their group membership on each and every topic? To examine this issue, we look at the correlations between the typologies for each pair of topics. Because the typology is composed of categories Cramer's V coefficient is used. The values (V= 0.712 for violent crime and criminality prevalence, V= 0.624 for violent crime and the number of old people, V= 0.580 for criminality prevalence and the number of old people) are high by all means. We calculate the overall consistency of the typology as a second indicator of stability. The value of Fleiss’ kappa agreement coefficient (κ= 0.605), a measure of concordance between more than two nominal variables (Kaplan, 2004), is just slightly above the lower boundary for acceptable agreement. However, let us not forget that it is not a measure of agreement between experts (a purpose for which Fleiss’ kappa was created in the first place) but indicates the stability of social estimation. For this objective, our figures pass the most rigid tests used to determine the stability of personality traits and social perceptions (see Matthewes, Deary, & Whiteman, 2003). RQ2 To examine RQ2, let us look at Table 3, which gives the daily mean amount of TV viewing (in hours) for each of the groups (overcultivation, simple cultivation, double distortion, simple no cultivation, and distorted no cultivation), as it was calculated in each topic (violent crimes, criminality prevalence, the number of old people). Table 3 Mean Amount of Daily Television Viewing (in hours) for Distorted No Cultivation, Simple No Cultivation, Simple Cultivation, Double Distortion, and Overcultivation in Different Topics TV Viewing . Characteristics . Group . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Light Correct real-world estimates with incorrect TV-world estimates Distorted no cultivation (underestimation of the TV world) M= 1.97, SD= 1.44 M= 1.94, SD= 1.18 M= 1.99, SD= 1.21 Distorted no cultivation (overestimation of the TV world) M= 1.88, SD= 1.61 M= 1.90, SD= 1.31 M= 1.73, SD= 1.30 Low medium Correct TV-world estimates with correct real-world estimates Simple no cultivation M= 2.20, SD= 1.26 M= 2.26, SD= 1.37 M= 2.18, SD= 1.48 High medium Correct TV-world estimates without a distinction between the real world and the TV world Simple cultivation M= 2.45, SD= 1.45 M= 2.35, SD= 1.32 M= 2.35, SD= 1.26 or incorrect TV-world estimates with a distinction between the TV world and real world Double distortion M= 2.41, SD= 1.33 M= 2.30, SD= 1.35 M= 2.39, SD= 1.42 Heavy Incorrect estimates of the TV world and a lack of distinction between the TV world and the real world Overcultivation M= 2.73, SD= 1.19 M= 2.42, SD= 1.51 M= 2.84, SD= 1.21 TV Viewing . Characteristics . Group . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Light Correct real-world estimates with incorrect TV-world estimates Distorted no cultivation (underestimation of the TV world) M= 1.97, SD= 1.44 M= 1.94, SD= 1.18 M= 1.99, SD= 1.21 Distorted no cultivation (overestimation of the TV world) M= 1.88, SD= 1.61 M= 1.90, SD= 1.31 M= 1.73, SD= 1.30 Low medium Correct TV-world estimates with correct real-world estimates Simple no cultivation M= 2.20, SD= 1.26 M= 2.26, SD= 1.37 M= 2.18, SD= 1.48 High medium Correct TV-world estimates without a distinction between the real world and the TV world Simple cultivation M= 2.45, SD= 1.45 M= 2.35, SD= 1.32 M= 2.35, SD= 1.26 or incorrect TV-world estimates with a distinction between the TV world and real world Double distortion M= 2.41, SD= 1.33 M= 2.30, SD= 1.35 M= 2.39, SD= 1.42 Heavy Incorrect estimates of the TV world and a lack of distinction between the TV world and the real world Overcultivation M= 2.73, SD= 1.19 M= 2.42, SD= 1.51 M= 2.84, SD= 1.21 Open in new tab Table 3 Mean Amount of Daily Television Viewing (in hours) for Distorted No Cultivation, Simple No Cultivation, Simple Cultivation, Double Distortion, and Overcultivation in Different Topics TV Viewing . Characteristics . Group . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Light Correct real-world estimates with incorrect TV-world estimates Distorted no cultivation (underestimation of the TV world) M= 1.97, SD= 1.44 M= 1.94, SD= 1.18 M= 1.99, SD= 1.21 Distorted no cultivation (overestimation of the TV world) M= 1.88, SD= 1.61 M= 1.90, SD= 1.31 M= 1.73, SD= 1.30 Low medium Correct TV-world estimates with correct real-world estimates Simple no cultivation M= 2.20, SD= 1.26 M= 2.26, SD= 1.37 M= 2.18, SD= 1.48 High medium Correct TV-world estimates without a distinction between the real world and the TV world Simple cultivation M= 2.45, SD= 1.45 M= 2.35, SD= 1.32 M= 2.35, SD= 1.26 or incorrect TV-world estimates with a distinction between the TV world and real world Double distortion M= 2.41, SD= 1.33 M= 2.30, SD= 1.35 M= 2.39, SD= 1.42 Heavy Incorrect estimates of the TV world and a lack of distinction between the TV world and the real world Overcultivation M= 2.73, SD= 1.19 M= 2.42, SD= 1.51 M= 2.84, SD= 1.21 TV Viewing . Characteristics . Group . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Light Correct real-world estimates with incorrect TV-world estimates Distorted no cultivation (underestimation of the TV world) M= 1.97, SD= 1.44 M= 1.94, SD= 1.18 M= 1.99, SD= 1.21 Distorted no cultivation (overestimation of the TV world) M= 1.88, SD= 1.61 M= 1.90, SD= 1.31 M= 1.73, SD= 1.30 Low medium Correct TV-world estimates with correct real-world estimates Simple no cultivation M= 2.20, SD= 1.26 M= 2.26, SD= 1.37 M= 2.18, SD= 1.48 High medium Correct TV-world estimates without a distinction between the real world and the TV world Simple cultivation M= 2.45, SD= 1.45 M= 2.35, SD= 1.32 M= 2.35, SD= 1.26 or incorrect TV-world estimates with a distinction between the TV world and real world Double distortion M= 2.41, SD= 1.33 M= 2.30, SD= 1.35 M= 2.39, SD= 1.42 Heavy Incorrect estimates of the TV world and a lack of distinction between the TV world and the real world Overcultivation M= 2.73, SD= 1.19 M= 2.42, SD= 1.51 M= 2.84, SD= 1.21 Open in new tab In all the items, the groups can be classified into four levels according to their amount of viewing: light viewers (distorted no cultivation), low-medium viewers (simple no cultivation), high-medium viewers (simple cultivation, double distortion), and heavy viewers (overcultivation). This order effect is significant for the number of old people (F(1,587)= 15.9, p <.001) and for violent crimes (F(1,587)= 3.8, p <.05). It misses the significance threshold just by a margin in criminality prevalence (F(1,587)= 3.2, p=.07). The correlations between the television viewing and the typology of viewers, which expresses the different combinations of real-world estimates and TV-world estimates, appear in Table 4. These are high correlations by cultivation standards that set the norm at 0.1 (see Shanahan & Morgan, 1999, pp. 110–115).2 However, post hoc Scheffé tests find significant differences in all the items only for the comparison between distorted no cultivation and overcultivation (I−J= 0.74, p <.05 in the number of old people; I−J= 0.51, p <.1 in violent crimes; I−J= 0.74, p <.05 in criminality prevalence). The answer to RQ2a is, therefore, that the heaviest viewers are members of the overcultivation group, followed—in decreasing order of viewing—by double distortion and simple cultivation, simple no cultivation, and finally distorted no cultivation. The relative magnitude of the differences is not minute. In violent crimes, distorted no cultivation viewers (the lightest viewers) watch approximately 55% less than overcultivation viewers (the heaviest viewers). In the number of old people, distorted no cultivation viewers watch 40% less than overcultivation viewers. Even in criminality prevalence, the average amount of viewing of distorted no cultivation viewers is nearly 25% smaller than the amount of viewing of overcultivation viewers. Table 4 Analysis of Variance (ANOVA) Results and Effect Size for Group Differences in the Amount of Television in Different Topics ANOVA Results . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Total viewing  F ratio for group differences F(3,587)= 5.9** F(3,587)= 2.7* F(3,587)= 3.0*  Eta correlation η= 0.251 η= 0.139 η= 0.135  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.148 η= 0.123 η= 0.131 News viewing  F ratio for group differences F(3,587)= 4.6** F(3,587)= 5.4** F(3,587)= 4.4*  Eta correlation η= 0.170 η= 0.326 η= 0.242  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.160 η= 0.310 η= 0.230 Dramatic series viewing  F ratio for group differences F(3,587)= 3.4* F(3,587)= 1.2 F(3,587)= 3.9*  Eta correlation η= 0.176 η= 0.126 η= 0.237  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.141 η= 0.109 η= 0.216 ANOVA Results . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Total viewing  F ratio for group differences F(3,587)= 5.9** F(3,587)= 2.7* F(3,587)= 3.0*  Eta correlation η= 0.251 η= 0.139 η= 0.135  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.148 η= 0.123 η= 0.131 News viewing  F ratio for group differences F(3,587)= 4.6** F(3,587)= 5.4** F(3,587)= 4.4*  Eta correlation η= 0.170 η= 0.326 η= 0.242  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.160 η= 0.310 η= 0.230 Dramatic series viewing  F ratio for group differences F(3,587)= 3.4* F(3,587)= 1.2 F(3,587)= 3.9*  Eta correlation η= 0.176 η= 0.126 η= 0.237  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.141 η= 0.109 η= 0.216 * p <.05. ** p <.005. Open in new tab Table 4 Analysis of Variance (ANOVA) Results and Effect Size for Group Differences in the Amount of Television in Different Topics ANOVA Results . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Total viewing  F ratio for group differences F(3,587)= 5.9** F(3,587)= 2.7* F(3,587)= 3.0*  Eta correlation η= 0.251 η= 0.139 η= 0.135  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.148 η= 0.123 η= 0.131 News viewing  F ratio for group differences F(3,587)= 4.6** F(3,587)= 5.4** F(3,587)= 4.4*  Eta correlation η= 0.170 η= 0.326 η= 0.242  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.160 η= 0.310 η= 0.230 Dramatic series viewing  F ratio for group differences F(3,587)= 3.4* F(3,587)= 1.2 F(3,587)= 3.9*  Eta correlation η= 0.176 η= 0.126 η= 0.237  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.141 η= 0.109 η= 0.216 ANOVA Results . Old People, N= 591 . Criminality Prevalence, N= 591 . Violent Crimes, N= 591 . Total viewing  F ratio for group differences F(3,587)= 5.9** F(3,587)= 2.7* F(3,587)= 3.0*  Eta correlation η= 0.251 η= 0.139 η= 0.135  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.148 η= 0.123 η= 0.131 News viewing  F ratio for group differences F(3,587)= 4.6** F(3,587)= 5.4** F(3,587)= 4.4*  Eta correlation η= 0.170 η= 0.326 η= 0.242  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.160 η= 0.310 η= 0.230 Dramatic series viewing  F ratio for group differences F(3,587)= 3.4* F(3,587)= 1.2 F(3,587)= 3.9*  Eta correlation η= 0.176 η= 0.126 η= 0.237  Fifth-order partial eta correlation—controlling for sex, ethnic origin, newspaper reading, and Internet use η= 0.141 η= 0.109 η= 0.216 * p <.05. ** p <.005. Open in new tab RQ2b asked if the above-mentioned pattern remains stable through topics, when we control for demographics and consumption of media other than television or when we base our calculations on viewing of specific television formats rather than total viewing. Table 3 shows that the ranking of groups by their amount of TV viewing keeps the same order through different topics. Table 4 shows that this effect (indicated by correlations and partial correlations between the amount of viewing and the group typology) remains significant when the basis for computation is total viewing, viewing of dramatic series, or viewing of news programs and when we control for newspaper reading, Internet use, sex, and ethnic origin (age and education level are practically already under control because the respondents are college students). In the case of news programs, the effect for criminality prevalence is slightly higher (as found in other studies as well; see Cohen & Weimann, 2000), perhaps because the topic is a cornerstone of news reports (Potter et al., 1997). Discussion The study expands on the idea of cultivation in general and the concept of first-order effects in particular. It classifies the viewers into five groups: overcultivation (estimating of the real world as a replica of the TV world and overestimating the TV world), simple cultivation (estimating the real world as a replica of the TV world and correctly estimating the TV world), double distortion (overestimating the TV world and incorrectly estimating the real world in a way that does not imitate the TV world), simple no cultivation (holding a correct estimation of both the real world and the TV world), and distorted no cultivation (holding a correct estimation of the real world and an incorrect estimation of the TV world). Our results do concur with the general idea of cultivation theory that a misguided perception of the social world is statistically related to television viewing (Gerbner & Gross, 1976; Gerbner et al., 1994). However, adding the accuracy of viewers’ TV-world estimates to the scheme reveals a picture that is far more complicated than the one traditional cultivation suggests. Cultivation in its traditional form, simple cultivation in our typology, appears to be only one node in an intricate network of relationships that connect viewers’ estimation of TV content and their perception of the real world. Simple cultivation characterizes only a minority of the viewers (13.1% in the number of old people, 17.1% in criminality prevalence, and 27.9% in violent crimes). In contrast, the composite share of the three distorted perceptions in this study (simple cultivation, overcultivation, and double distortion) is as large as two thirds of the viewers in violent crimes and criminality prevalence and 40% of them in the number of old people. Furthermore, in comparison with double distortion and overcultivation, simple cultivation is not only a statistical minority, but also a conceptually tamer distortion, because viewers retain a correct estimation of the TV world and do not go beyond that estimation when making judgments about the real world. Not surprisingly, simple cultivation viewers view less television than overcultivation respondents whose reality estimation is more extremely biased. Together, the classified types comprise a large majority of the viewers (at least 85%), but this is not the only reason why our discussion of the various relationships between real-world estimates and TV-world estimates typifies groups of viewers rather than analyzes correlations. TV-world estimates and real-world estimates are statistically related (see the values of Cramer's V coefficient in Table 1), but only the typified groups enable to distinguish between accurate and inaccurate estimates and differentiate smaller distortions from larger ones. For example, simple cultivation viewers and overcultivation viewers resemble one another in the identical estimation of the TV world and the real world, but while in simple cultivation the TV world is accurately estimated and estimation of the real world is only slightly exaggerated, in overcultivation both worlds are inaccurately and exaggeratedly estimated. Before we continue, although our analysis emphasizes findings that cross over topics and predominate the data set, there are also a few content-specific findings that should be mentioned. The three distortions—simple cultivation, double distortion, and overcultivation—appear to be, altogether, stronger in crime and violence but weaker in estimating the number of old people. This should not come as a surprise, as previous studies found that the cultivation effect is very consistent in estimating occurrences of violence and crime (Shanahan & Morgan, 1999, p.127), and not as stable in estimating the number of old people (Passuth & Lomax-Cook, 1985). The reason for this difference might be that the ubiquitous appearance of crime and violence is more vivid than the absence of the elderly. This vividness strengthens the cultivation potential (Shrum, 1995). Classifying groups of viewers by their amount of viewing The lightest viewers, who are members of the distorted no cultivation group, estimate the real world correctly but make mistakes in estimating the TV world. A sensible conclusion about these viewers is that limited viewing not only correlates with a minimal effect on reality perception but also comes with some incapability to identify the right TV answer. A slight increase in viewing appears to fix this distortion as simple no cultivation viewers, who watch a bit more than distorted no cultivation viewers, estimate correctly both the real world and the TV world. As viewing continues to increase, both estimations are gradually impaired. The next level of viewing features two groups—simple cultivation and double distortion—that in comparison with simple no cultivation, either lose the ability to distinguish one reality from another, but still estimate TV reality correctly (in the case of simple cultivation), or still make the distinction between the real world and the TV world, but lose the capability to estimate TV reality correctly (in the case of double distortion). Finally, the heaviest viewers, who are concentrated in the overcultivation group, no longer distinguish the real world from the TV world and are also unable to estimate correctly either one of the worlds. Both receive overestimates from this group. Heavier viewing does lead to a more accurate estimation of television reality when the baseline is low—as in the case of simple no cultivation versus distorted no cultivation. When viewing is heavier to begin with, increasing it further would not yield more accurate estimates of TV reality. In fact, the heaviest viewers (overcultivation) are characterized by extremity of perception and a loss of distinction between the real world and the TV world. These patterns hold when we control for sex, ethnicity, newspaper reading, and Internet use, or use exposure to particular formats (dramatic series, news programs) as a measure of viewing instead of total daily viewing. This robustness of effects, typical of first-order cultivation (Potter, 1993), might stem from the typically similar metanarratives and metalessons, which characterize many programs regardless of the genre (Shanahan & Morgan, 1999, pp.191–195). This work concentrates on extending the what (what happens in cultivation) and leaves the how (how it happens) for future endeavors. However, we will try to suggest few reasoned speculations. First, the reason that that the lightest viewers (distorted no cultivation) give incorrect estimates of the TV world while correctly estimating the real world could be that they do not know the TV world well enough to estimate it correctly. The relatively minimal exposure of these viewers to the programming does not provide sufficient opportunity for incidental learning from the screen to occur. Their limited acquaintance with television reality is not strong enough to influence their real-world judgments. Unsurprisingly, their perception of the real world is not similar to what they think of the TV world. The content of the programming does not frequently project itself on their view of the real world. Viewers from the overcultivation group, on the other hand, may go through what Gerbner, Gross, Morgan, and Signorielli (1980) call resonance—a situation in which somewhat similar impressions that are made by the two realities (the real world and the television world) strengthen one another and stimulate similar overestimates. When it comes to violence and crime, overcultivation respondents may go through resonance, if they grew up in violent boroughs or live now in neighborhoods abundant with crime. They may believe that the typical television reality is violent and full of criminals, more than it really is according to objective measures, since the intensive TV portrayal of violence and crime matches the reality of their personal surrounding. As for the elderly estimate, overcultivation viewers may go through the process of resonance, if their living surroundings (e.g., college dorms) are relatively empty of old people. This matches the under-representation of the elderly in television programs. Another potential explanation why heavy viewers do not give the most accurate estimates of television reality is that watching television in heavy doses impairs one's sense of measurement. Heavy viewers are no longer able to make fine distinctions between different programs because the overabundance of different shows forms a sort of clutter (Weimann et al., 1992) in which the most extreme exemplars (most violent, most filled with crime, etc.) stand out, engrave the most vivid impression (Brosius & Bathelt, 1994), and are most likely to be retrieved (O’Guinn & Shrum, 1997). The fact that the TV diet of heavy viewers is often characterized by larger portions of “action-adventure” programs, full of criminal characters and loaded with violent scenes (Van der Voort, 1986), may also hint why the estimates they give in these topics are exaggerated. Finally, the loud criticism of commercial television for annihilating older people (Davis & Davis, 1985), and for showing too much violence (Grossman & DeGaetano, 1999), may also have a share in bringing people to believe that the programming is extremely violent, full of criminals, and hardly ever features old people. The fact that this criticism is often heard on TV news and in talk shows (Davis & Davis) makes heavy viewers particularly exposed to its message. As the likelihood of choosing the TV answer just by chance out of three possible answers is 33.3%, our respondents, who have picked the right TV answer 47.3% of the times in criminality prevalence, 33.4% of the times in violent crimes, and 41.6% of the times in the number of old people, have performed quite poorly. What may characterize the minority of viewers who do correctly estimate TV reality is a more profound encoding of television information due to higher involvement with the programs, as suggested by Ward & Rivadeneyra (1999). In any case, the multitude of incorrect TV-world estimates shows that for a substantial number of viewers, content analysis findings appear to constitute a bias of what they see as typical TV reality (Potter & Tomasello, 2003). Study limitations and suggestions for future research A few words are due about the deficiencies of the study. First, conceptually, our claims are delimited to first-order judgments and leave for future investigations second-order measures, which have produced in the past significantly smaller effects (Hawkins & Pingree, 1982; Potter, 1991c). Second, from a methodological standpoint, although all our findings converge into a coherent framework that classifies viewers into groups according to their social reality judgments and TV reality estimates and while this typology consistently holds under multiple controls and meaningfully correlates with the amount of viewing, we should warn that a few of the intergroup differences are not significant—when tested post hoc. The differences between heavy viewers (overcultivation) and light viewers (distorted no cultivation) are significant; yet, the differences between heavy or light viewers (overcultivation or distorted no cultivation) and low-medium (simple no cultivation) or high-medium (simple cultivation, double distortion) viewers are not large enough to pass a post hoc test. Whereas relying on a sample of students, as we do here, does not add much to external validity, detecting consistent effects of TV viewing on social perception among viewers who are expected to be more critical and better educated than the man on the street indicates that our findings represent actual trends and are not caused by a measurement error. Using a sample of students can even contribute to internal validity because it eliminates alternative interpretations of the results based on socioeducational factors. One may still claim that it is impossible for viewers to give conscious estimates of the TV world because these estimates are sometimes not so highly thoughtful (Shrum & O’Guinn, 1993) and because viewers’ involvement with the subject matter can be low. However, if this is the case, then the capability to measure real-world estimates and relate them to television viewing—the essence of cultivation—is also in doubt because real-world estimates are frequently unconscious to viewers until the time they are required to give them, and the subjects of these estimates are of low involvement to most of the people (Shrum, 1995). If we accept that real-world estimates can be measured directly (and this is what cultivation studies do), then there is no reason why estimates of the TV world cannot be measured just as well. The most appropriate way to measure them is through a survey because we speak about the long-term influence of media consumption and not about a sudden effect of attending to any particular text. Apart from increasing the diversity of the respondents and extending the span of topics, future studies may possibly also probe deeper into the conceptualization of our scheme. One line of investigation might be to search for models that explain what stands behind the distortions—what causes some viewers to show overcultivation when others exhibit double distortion or simple cultivation. Psychological factors such as learning skills, personality traits, and the extent of involvement with the programs might prove to play a key role here. Another territory yet to be conquered is the wider array of media effects. Our findings signal a consistent trend; however, the field we cover—first-order cultivation effects—is limited. Future works may possibly examine whether our typology correlates with other media effects and— in particular—with cultivation's second-order effects. Acknowledgments The following people gave us inspiration during the work on this project: James Shanahan (Cornell University), Michael Morgan (University of Massachusetts, Amherst), Jonathan Cohen (University of Haifa), and Larry J. Shrum (University of Texas, San Antonio). We are also indebted to Einat Shaul for her careful proofreading. Notes 1 " The idea that viewers’ interpretation of the programs could be a factor in cultivation was initially proposed by Newcomb (1978). However, Newcomb's remark did not stimulate research perhaps because his words were targeted as a “humanistic critique” of the cultivation school and did not suggest ways to improve the theory through empirical studies. 2 " Over the years, cultivation studies have used a multitude of correlation coefficients to measure the strength of the association between television viewing and reality perception (see Potter, 1991b, for examples). Our rationale for choosing eta is that this coefficient uses as much information as can be produced from a data set in which one of the variables (the amount of television viewing per day) is metric and the second (five groups of viewers) is categorical. One may criticize our decision as treating the amount of viewing as the dependent variable rather than vice versa. We believe that in a correlative study, especially one that aims to examine long-term effects, it is impossible to determine the ultimate independent factor precisely from within the variables. Statisticians recommend, in such circumstances, to compute the coefficient that would use as much information as possible from the data set and not to bother too much with the alleged symmetry or asymmetry of the association (Kaplan, 2004). We adopt this recommendation. However, to challenge a potential criticism that only the choice of an asymmetric coefficient that treats TV viewing as the dependent variable was responsible for the consistent results, we also computed two alternative coefficients: Cramer's V (symmetric categorical coefficient) and lambda (asymmetric categorical coefficient for which we predetermined that the amount of viewing would be the independent variable). To compute these coefficients, we divided the respondents into four groups according to their amount of viewing: light viewers (less than 2 hours per day), low-medium viewers (2–2.5 hours per day), high-medium viewers (2.5–3 hours per day), and heavy viewers (more than 3 hours per day). This grouping puts the sample's mean (2.4 hours per day) approximately in the midpoint between low-medium and high-medium viewers. The values of Cramer's V run from 0.167 (for violent crimes), through 0.178 (for the number of old people), and up to 0.195 (for criminality prevalence). Lambda values run from 0.033 (for the number of old people), through 0.046 (for violent crimes), and up to 0.052 (for criminality prevalence). Because the value of lambda is equal to the proportion of variance accounted for by the coefficient, whereas the values of Cramer's V and eta still need to be raised by the second power in order to obtain an assessment of the size of the effect (Kaplan, 2004); the conclusion is that the values of three different coefficients (eta, lambda, Cramer's V) are at the same level and that the pattern of the results holds beyond the choice of any particular coefficient. Appendix Appendix A Please note: The order of answers was randomly reordered in all the questions throughout the questionnaires. Violent Crimes Real world measure What percent of all crimes involve violence, like murders, rapes, robbery and assault? (a) 15% (Real world answer) (b) 25% (TV answer) (c) 35% (Over-TV answer) Television world measure What percent of all crimes shown on television involve violence, like murders, rapes, robbery and aggravated assault? (a) 15% (Real world answer) (b) 25% (TV answer) (c) 35% (Over-TV answer) Criminality Prevalence Real world measure What percent of the population are criminals, or have a criminal record? (a) 2% (Real world answer) (b) 10% (TV answer) (c) 20% (Over-TV answer) Television world measure What percent of all television characters are criminals, or have a criminal record? (a) 2% (Real world answer) (b) 10% (TV answer) (c) 20% (Over-TV answer) Old People Real world measure What percent of the population are at the age of 65 or older? (a) 10% (Real world answer) (b) 5% (TV answer) (c) 2% (Over-TV answer) Television world measure What percent of all television characters are at the age of 65 or older? (a) 10% (Real world answer) (b) 5% (TV answer) (c) 2% (Over-TV answer) References Brosius , H. B. , & Bathelt , A. ( 1994 ). The utility of exemplars in persuasive communications . Communication Research , 21 ( 1 ), 48 – 78 . Google Scholar Crossref Search ADS WorldCat Campbell , D . ( 1950 ). The indirect assessment of social attitudes . Psychological Bulletin , 47 , 15 – 38 . Google Scholar Crossref Search ADS WorldCat Carveth , R. , & Alexander , A. ( 1985 ). Soap opera viewing motivations and the cultivation hypothesis . Journal of Broadcasting & Electronic Media , 29 , 259 – 273 . Google Scholar Crossref Search ADS WorldCat Central Bureau of Statistics (CBS) . ( 2003 ). Statistical abstract of Israel (No. 53) . Jerusalem : Author . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Cheung , C. K. , & Chan , C. F. ( 1996 ). Television viewing and mean world value in Hong Kong's adolescents . Social Behavior & Personality , 24 ( 4 ), 351 – 364 . Google Scholar Crossref Search ADS WorldCat Cohen , J. , & Weimann , G. ( 2000 ). Cultivation revisited: Some genres have some effect on some viewers . Communication Report , 13 ( 2 ), 99 – 114 . Google Scholar Crossref Search ADS WorldCat Davis , R. H. , & Davis , J. ( 1985 ). TV's image of the elderly: A practical guide for change . Lexington, MA : Lexington Books . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Doob , A. N. , & McDonald , G. E. ( 1979 ). Television viewing and fear of victimization: Is the relationship causal? Journal of Personality and Social Psychology , 37 ( 1 ), 170 – 179 . Google Scholar Crossref Search ADS WorldCat An enemy in your living room . ( 1993 , July 6). Christian Science Monitor , pp. 2 – 3 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Fox , W. S. , & Philliber , W. W. ( 1978 ). Television viewing and the perception of affluence . The Sociological Quarterly , 19 ( 1 ), 103 – 112 . Google Scholar Crossref Search ADS WorldCat Gerbner , G . ( 1972 ). Communication and social environment . Scientific American , 227 ( 3 ), 152 – 160 . Google Scholar Crossref Search ADS WorldCat Gerbner , G. , & Gross , L. ( 1976 ). Living with television: The violence profile . Journal of Communication , 26 ( 2 ), 173 – 199 . Google Scholar Crossref Search ADS PubMed WorldCat Gerbner , G. , Gross , L., Eleey , M. F., Jackson-Beeck , M., Jeffries-Fox , S., & Signorielli , N. ( 1977 ). TV violence profile no. 8: The highlights . Journal of Communication , 27 ( 2 ), 171 – 180 . Google Scholar Crossref Search ADS WorldCat Gerbner , G. , Gross , L., Jackson-Beeck , M., Jeffries-Fox , S., & Signorielli , N. ( 1978 ). Cultural indicators: Violence profile no. 9 . Journal of Communication , 28 , 176 – 207 . Google Scholar Crossref Search ADS PubMed WorldCat Gerbner , G. , Gross , L., Morgan , M., & Signorielli , N. ( 1980 ). The “mainstreaming” of America: Violence profile no. 11 . Journal of Communication , 30 ( 3 ), 10 – 29 . Google Scholar Crossref Search ADS WorldCat Gerbner , G. , Gross , L., Morgan , M., & Signorielli , N. ( 1994 ). Growing up with television: The cultivation perspective. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (pp. 17 – 41 ). Hillsdale, NJ : Erlbaum . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gerbner , G. , Gross , L., Signorielli , N., & Morgan , M. ( 1980 ). Aging with television: Images on television drama and conceptions of social reality . Journal of Communication , 30 ( 1 ), 37 – 47 . Google Scholar Crossref Search ADS PubMed WorldCat Gosselin , A. , DeGuise , J., & Paquette , G. ( 1997 ). Violence in Canadian television and some of its cognitive effects . Canadian Journal of Communication , 22 ( 2 ), 143 – 160 . Google Scholar Crossref Search ADS WorldCat Greenberg , B. C . ( 1980 ). Life on television: Content analyses of U.S. TV drama . Norwood, NJ : Ablex . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gross , L. , & Jeffries-Fox , S. ( 1978 ). What do you want to be when you grow up, little girl? In G. Tuchman, A. K. Daniles, & T. Benet (Eds.), Hearth and home: Images women in the mass media . New York : Oxford University Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Grossman , D. , & DeGaetano , G. ( 1999 ). Stop teaching our kids to kill: A call to action against TV, movie & video game violence . New York : Crown . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Gunter , B. , & Wober , M. ( 1983 ). Television viewing and public trust . British Journal of Social Psychology , 22 , 174 – 176 . Google Scholar Crossref Search ADS WorldCat Hawkins , R. P. , & Pingree , S. ( 1980 ). Some processes in the cultivation effect . Communication Research , 7 , 193 – 226 . Google Scholar Crossref Search ADS WorldCat Hawkins , R. P. , & Pingree , S. ( 1982 ). Television's influence on social reality. In D. Pearl, L. Bouthilet, & I. Lazar (Eds.), Television and behavior: Ten years of scientific progress and implications for the eighties (Vol. 2 , pp. 224 – 247 ). Rockville, MD : U.S. Department of Health and Human Services . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hawkins , R. P. , Pingree , S., & Adler , I. ( 1987 ). Searching for cognitive processes in the cultivation effect . Human Communication Research , 13 , 553 – 577 . Google Scholar Crossref Search ADS WorldCat Hirsch , P. M . ( 1980 ). The ‘scary world’ of the nonviewer and other anomalies: A reanalysis of Gerbner et al.'s findings on cultivation analysis . Communication Research , 7 , 403 – 456 . Google Scholar Crossref Search ADS WorldCat Hughes , M . ( 1980 ). The fruits of cultivation analysis: A re-examination of the effects of television watching on fear of victimization, alienation and the approval of violence . Public Opinion Quarterly , 44 , 287 – 302 . Google Scholar Crossref Search ADS WorldCat Israel Audience Research Board . ( 2004 ). Retrieved October 3, 2004, from http://www.midrug-tv.org.il/scripts/public1.asp Kaplan , D . ( 2004 ). The Sage handbook of quantitative methodology for the social sciences: Quantitative methodology for the social sciences . Thousand Oaks, CA : Sage . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Kwak , H. , Zinkhan , G. M., & Dominick , J. R. ( 2002 ). The moderating role of gender and compulsive buying tendencies in the cultivation effects of TV show and TV advertising: A cross cultural study between the United States and South Korea . Media Psychology , 4 ( 1 ), 77 – 111 . Google Scholar Crossref Search ADS WorldCat Lichter , R . ( 1999 ). Things we can do now to clean up television . The American Enterprise , 10 ( 2 ), 36 – 37 . OpenURL Placeholder Text WorldCat Lichter , S. R. , Lichter , L. S., & Rothman , S. ( 1994 ). Prime time: How TV portrays American culture . Washington, DC : Regnery . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Matthewes , G. , Deary , I. J., & Whiteman , M. C. ( 2003 ). Personality traits (2nd ed.). Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Morgan , M . ( 1983 ). Symbolic victimization and real world fear . Human Communication Research , 9 ( 2 ), 146 – 157 . Google Scholar Crossref Search ADS WorldCat Morgan , M . ( 1984 ). Heavy television viewing and perceived quality of life . Journalism Quarterly , 61 , 499 – 504 . Google Scholar Crossref Search ADS WorldCat Morgan , M . ( 1986 ). Television and the erosion of regional diversity . Journal of Broadcasting & Electronic Media , 30 ( 2 ), 123 – 139 . Google Scholar Crossref Search ADS WorldCat Morgan , M. , & Shanahan , J. ( 1992 ). Comparative cultivation analysis: Television and adolescents in Argentina and Taiwan. In F. Korzenny & S. Ting-Toomey (Eds.), Mass media effects across cultures (pp. 173 – 197 ). Newbury Park, CA : Sage . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Morgan , M. , & Shanahan , J. ( 1997 ). Two decades of cultivation research: An appraisal and meta-analysis. In B. Berelson (Ed.), Communication yearbook (Vol. 20 , pp. 1 – 45 ). Thousand Oaks, CA : Sage . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Newcomb , H . ( 1978 ). Assessing the violence profile of Gerbner and Gross: A humanistic critique and suggestion . Communication Research , 5 , 264 – 282 . Google Scholar Crossref Search ADS WorldCat Ogles , R. M. , & Sparks , G. G. ( 1989 ). Television violence and viewers’ perceptions of criminal victimization . Mass Communications Review , 16 ( 1 ), 2 – 11 . OpenURL Placeholder Text WorldCat O'Guinn , T. C. , & Shrum , L. J. ( 1997 ). The role of television in the construction of consumer reality . Journal of Consumer Research , 23 , 278 – 294 . Google Scholar Crossref Search ADS WorldCat O'Keefe , G. T . ( 1984 ). Public views on crime: Television exposure and media credibility. In R. N. Bostrom (Ed.). Communication yearbook (Vol. 8 , pp. 73 – 95 ). Newbury Park, CA : Sage . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Passuth , P. M. , & Lomax-Cook , F. ( 1985 ). Effects of television on knowledge and attitudes about older adults: A critical reexamination . The Gerontologist , 25 ( 1 ), 69 – 77 . Google Scholar Crossref Search ADS PubMed WorldCat Pfau , M. , Mullen , L., Dietrich , T., & Garrow . K. ( 1995 ). Television viewing and public perceptions of attorneys . Human Communication Research , 21 ( 3 ), 307 – 330 . Google Scholar Crossref Search ADS WorldCat Potter , W. J . ( 1986 ). Perceived reality and the cultivation hypothesis . Journal of Broadcasting & Electronic Media , 30 ( 2 ), 159 – 174 . Google Scholar Crossref Search ADS WorldCat Potter , W. J . ( 1988 ). Three strategies for elaborating the cultivation hypothesis . Journal of Mass Communication Quarterly , 65 , 930 – 940 . OpenURL Placeholder Text WorldCat Potter , W. J . ( 1991a ). Examining cultivation from a psychological perspective: Component sub-processes . Communication Research , 18 ( 1 ), 77 – 102 . Google Scholar Crossref Search ADS WorldCat Potter , W. J . ( 1991b ). The linearity assumption in cultivation research . Human Communication Research , 17 , 562 – 584 . Google Scholar Crossref Search ADS WorldCat Potter , W. J . ( 1991c ). The relationships between first and second order measures of cultivation . Human Communication Research , 18 ( 1 ), 92 – 123 . Google Scholar Crossref Search ADS WorldCat Potter , W. J . ( 1993 ). Cultivation theory and research: A conceptual critique . Human Communication Research , 19 ( 4 ), 564 – 601 . Google Scholar Crossref Search ADS WorldCat Potter , W. J. , & Tomasello , T. K. ( 2003 ). Building upon the experimental design in media violence research: The importance of including receiver interpretations . Journal of Communication , 53 , 315 – 329 . Google Scholar Crossref Search ADS WorldCat Potter , W. J. , Vaughan , M. W., Warren , R., Howley , K., Land , A., & Hagenmayer , J. C. ( 1995 ). How real is the portrayal of aggression in the television entertainment programming? Journal of Broadcasting and Electronic Media , 39 , 496 – 516 . Google Scholar Crossref Search ADS WorldCat Potter , W. J. , Warren , R., Vaughan , M. W., Howley , K., Land , A., & Hagemeyer , J. C. ( 1997 ). Antisocial acts in reality programming on television . Journal of Broadcasting and Electronic Media , 41 ( 1 ), 69 – 89 . Google Scholar Crossref Search ADS WorldCat Romer , D. , Jamieson , K. H., & Aday , S. ( 2003 ). Television news and the cultivation of fear of crime . Journal of Communication 53 ( 1 ), 88 – 104 . Google Scholar Crossref Search ADS WorldCat Rössler , P. , & Brosius , H. B. ( 2001 ). Do talk shows cultivate adolescents’ views of the world? A prolonged-exposure experiment . Journal of Communication , 16 ( 1 ), 143 – 163 . Google Scholar Crossref Search ADS WorldCat Rubin , R. B. , Palmgreen , P., & Sypher , H. E. (Eds.). ( 1994 ). Communication research measures: A sourcebook (pp. 154 – 158 , 226–229). New York : Guilford Press . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Schwartz , N . ( 1999 ). Self reports: How the questions shape the answers . American Psychologist , 54 ( 2 ), 93 – 105 . Google Scholar Crossref Search ADS WorldCat Shanahan , J. , & Morgan , M. ( 1999 ). Television and its viewers: Cultivation theory and research . Cambridge, MA : Cambridge University Press . Google Scholar Crossref Search ADS Google Scholar Google Preview WorldCat COPAC Shrum , L. J . ( 1995 ). Assessing the social influence of television: A social cognition perspective . Communication Research , 22 ( 4 ), 402 – 429 . Google Scholar Crossref Search ADS WorldCat Shrum , L. J . ( 1996 ). Psychological processes underlying cultivation effects: Further tests of construct accessibility . Human Communication Research , 22 , 482 – 509 . Google Scholar Crossref Search ADS WorldCat Shrum , L. J . ( 2002 ). Media consumption and perceptions of social reality: Effects and underlying processes. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (2nd ed., pp. 43 – 67 ). Hillsdale, NJ : Erlbaum . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Shrum , L. J. , Burroughs , J. E., & Rindfleisch , A. ( 2004 ). A process model of consumer cultivation: The role of TV is a function of type of judgment. In L. J. Shrum (Ed.), The psychology of entertainment: Blurring the lines between entertainment and persuasion (pp. 177 – 192 ). Mahwah, NJ : Erlbaum . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Shrum , L. J. , & O'Guinn , T. C. ( 1993 ). Processes and effects in the construction of social reality: Construct accessibility as an explanatory variable . Communication Research , 20 , 436 – 471 . Google Scholar Crossref Search ADS WorldCat Signorielli , N . ( 1990 ). Television's mean and dangerous world: A continuation of the cultural indicators perspective. In N. Signorielli & M. Morgan (Eds.), Cultivation analysis: New directions in media effects research (pp. 85 – 106 ). Newbury Park, CA : Sage . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Signorielli , N . ( 2001 ). Aging on television: The picture in the nineties . Generations , 25 ( 3 ), 34 – 38 . OpenURL Placeholder Text WorldCat Slater , D. , & Elliott , W. R. ( 1982 ). Television's influence on social reality . Quarterly Journal of Speech , 68 ( 1 ), 69 – 79 . Google Scholar Crossref Search ADS WorldCat Tan , A. S . ( 1982 ). Television use and social stereotypes . Journalism Quarterly , 59 ( 1 ), 119 – 222 . Google Scholar Crossref Search ADS WorldCat Tanur , J. M . ( 1992 ). Questions about questions . New York : Russell Sage Foundation . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Three U.S. senators speak out: Why cleaning up television is important to the nation . ( 1993 , March/April). Academic Enterprise , 2 , 32 – 34 . Van der Voort , T. H. A . ( 1986 ). Television violence: A child’s-eye view . Amsterdam : North-Holland . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Volgy , T. J. , & Schwarz , J. E. ( 1980 ). TV entertainment programming and sociopolitical attitudes . Journalism Quarterly , 57 ( 1 ), 150 – 155 . Google Scholar Crossref Search ADS WorldCat Ward , L. M. , & Rivadeneyra , R. ( 1999 ). Contributions of entertainment television to adolescents’ sexual attitudes and expectations: The role of viewing amount versus viewer involvement . Journal of Sex Research , 26 , 237 – 249 . Google Scholar Crossref Search ADS WorldCat Weaver , J. , & Wakshlag , J. ( 1986 ). Perceived vulnerability to crime, criminal victimization experience, and television viewing . Journal of Broadcasting & Electronic Media , 30 ( 1 ), 141 – 158 . Google Scholar Crossref Search ADS WorldCat Weimann , G . ( 1984 ). Images of America: The impact of American TV in Israel . International Journal of Intercultural Relations , 8 ( 2 ), 185 – 197 . Google Scholar Crossref Search ADS WorldCat Weimann , G . ( 2000 ). Communicating unreality: Modern media and the reconstruction of reality . Thousand Oaks, CA : Sage . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Weimann , G. , Brosius , H. B., & Wober , M. ( 1992 ). TV Diets: Towards a typology of TV viewership . European Journal of Communication , 7 , 491 – 515 . Google Scholar Crossref Search ADS WorldCat © 2006 International Communication Association
TI - Television-World Estimates, Real-World Estimates, and Television Viewing: A New Scheme for Cultivation
JF - Journal of Communication
DO - 10.1111/j.1460-2466.2006.00007.x
DA - 2006-03-01
UR - https://www.deepdyve.com/lp/oxford-university-press/television-world-estimates-real-world-estimates-and-television-viewing-uy0v26oftl
SP - 133
VL - 56
IS - 1
DP - DeepDyve
ER -