Daytime Television Talk Shows and the Cultivation Effect among U.S. and International Students

Article excerpt

Cultivation analysis is one of the more influential research traditions that has examined the long-term effects of television viewing. The typical cultivation study begins by identifying the most recurrent and stable patterns in television content, emphasizing consistent images and portrayals. The next step generally consists of a survey that explores the extent to which television watching contributes to audience members' perceptions of the real world (Morgan & Shanahan, 1997).

One programming genre that seems a likely candidate for cultivation study is the daytime television talk show (such as those hosted by Jerry Springer, Jenny Jones, and Ricki Lake) that specializes in sensational topics. These shows have increased in popularity the last several years, and along with the increased popularity has come increased criticism about the sensationalized and outrageous nature of many of these programs. Donovan (1997) labeled them "carnival sideshows." Rubin and Step (1997) called them "cultural oddities," while White (1993) labeled them "warped and weird." These particular programs typically focus on relationship problems and behaviors that many in the audience find aberrant, bizarre, deviant, or, at best, unconventional. (1) Critics have suggested that these programs may have negative effects on viewers because, as Senator Joseph Lieberman (quoted in Alter, 1995, p. 47) put it, these shows "make the abnormal normal." By focusing on bizarre behaviors and dysfunctional relationships, these programs might prompt audience members to think that such behaviors are typical in the United States and, in turn, prompt more skeptical and negative judgments about human relationships in American society.

These criticisms take on added significance for foreigners living in the U.S. who have relatively little experience in their host society. As Tamborini and Choi (1990) point out, limited understanding of the culture of a host society, difficulties in language proficiency, and lack of direct contact with the lifestyles of the host society could result in social perceptions about the host society that are more easily influenced by stereotyped behavior contained in television programming. Consequently, it is possible that, when compared to U.S. residents (who might also hold stereotypical perceptions), foreigners who are heavy viewers of daytime television talk shows might be more apt to hold distorted perceptions and skeptical attitudes about interpersonal relationships and overestimate the frequency of certain deviant or not widely accepted behaviors in U.S. society.

Theoretic Background

Many studies of cultivation (see Morgan & Shanahan, 1997) have provided evidence of a small but persistent relationship between television viewing and beliefs about the social world that are similar to or plausibly implied by the images in television programs. Nonetheless, it is still unclear how this process occurs. Shrum and O'Guinn (1993) offer one possibility based on the "bin model" of memory (Wyer & Srull, 1989). This model suggests that the human memory resembles a file cabinet. When new information about a topic is acquired, a copy of that information is placed at the front of file. If asked to make a judgment about a particular topic, a person will use that information which is most accessible. Two factors, among others, contribute to the greater accessibility of information: frequency and recency. When a person retrieves information about the topic, the contents of the file are searched from the front to the back. Thus, information that has been frequently repeated and recently acquired has the best chance of being remembered. A person who regularly watches a large amount of television programming that presents a consistent view of a topic might group a large number of television images at the beginning of the file. When asked to make a judgment about social reality, these images may be the most accessible. Consequently, the person might base his or her judgment of social reality on them.

Applying Shrum and O'Guinn's (1993) reasoning to the situation of the international student suggests that television exposure might be an even more influential part of the file. International students may not have much contact with people in the host society. Their understanding of the culture of the host society might be limited and influenced by previously acquired stereotypes. Moreover, limited proficiency in the English language might make it difficult to acquire information about the host society. Thus, compared to U.S. students, television exposure may become more influential for foreign students, thanks to its ubiquity and ease of access. In addition, the visual nature of the medium makes it easy to follow even for those with below average language skills. In short, in contrast to U.S. students, whose files should contain information provided by numerous sources in addition to television programming, the files of international students might consist primarily of information and perceptions gleaned from scores of images presented by television.

Moreover, Shrum (1999) describes another factor in the process--source discounting. In order for repeated exposure to televised images to have an impact on perceptions and attitudes, individuals must consider the portrayals useful and relevant. Shrum further suggests, however, that these retrieved images are not normally connected to the source unless viewers are first prompted to consider the source. Questions about talk show viewing might serve as such a prompt and increase the saliency of the source. Once reminded of the source, it is possible that American students, having more direct experience with U.S. residents and U.S. culture and being more aware of the contrived and sensationalistic formula used by talk shows to attract an audience, are more likely to perceive talk show content as atypical and discount it as a source of relevant information about society. On the other hand, international students, with less direct experience and less knowledge of American television formulas, might be less apt to source discount. This notion is consistent with Tyler and Cook (1984), who reported that media had little influence on judgments of topics with which respondents had more direct experience. On the other hand, media had more impact concerning judgements where there was presumably less experience.

Literature Review

There is a large body of cultivation literature (see Morgan & Shanahan, 1997). This review examines only those areas most relevant to the current study. The first area concerns daytime television talk shows.

In an early study, Greenberg and Smith (1995) examined 200 episodes of daytime talk shows and discovered that the most frequently discussed topics were sex, abuse, drug addiction, and criminal activity. In a later study, Greenberg, Sherry, Busselle, Hnilo, and Smith (1997) content analyzed a sample of daytime talk shows and found that most concerned problematic relationships among primary groups (parents and children, siblings, spouses, friends, people in romantic relationships) and that most shows focused on family issues, sexual activity, and dating. Woo (1998) analyzed 115 episodes of daytime talk shows and found that 6 in 10 concerned abnormal human relationships, and most of these portrayed problems associated with marriage, romance, and the family. In short, these studies suggest that this genre presents a recurrent pattern of social images that might lead to a cultivation impact.

In fact, the results of a survey among U.S. high school students (Davis & Mares, 1998) indicated that viewing these programs did result in a limited cultivation effect. They found that teens who were heavy viewers of talk shows overestimated the frequency of undesirable behaviors that were commonly discussed on these shows (e.g., teens running away from home, girls pregnant before age 18, boys and girls who have had teenage sex). Also not supported were hypotheses that heavy viewing of these programs would desensitize viewers to the suffering of victims of immoral acts and would prompt viewers to trivialize important issues.

There are relatively few cross-national studies of cultivation that look at the cultivation impact of American television on foreigners who are living in the United States. Choi and Tamborini (1988) found little evidence of a cultivation effect concerning fear of victimization and perceptions about the frequency of crime among a sample of Korean immigrants living in several different cities in the United States. Similarly, Tamborini and Choi (1988) examined Korean students attending college in the United States. The authors found that those heavily exposed to television were not more likely to fear being victimized by crime. The results also suggested that direct experience was more important than television viewing in predicting social perceptions. Despite the lack of results supporting the cultivation hypothesis with regard to perceptions and attitudes about crime, Tamborini and Choi (1990) point out that these studies suggest that cultural diversity and experiences are important factors in understanding the relationship between television viewing and perceptions of social reality. The current study also looks at foreigners living in the U.S. but examines an area less provocative and less dramatic than perceptions of crime and physical violence.

Hypotheses

Drawing upon the literature review and the conceptual rationale presented above, the current study was designed to test one main hypothesis and three sub-hypotheses.

The main hypothesis dealt with differences in the cultivation effect between international students and U.S. students. In general terms:

H1: International students will display more of a cultivation effect than U.S. students.

Specifically,

H1a: The relationship between viewing daytime television talk shows and the estimation of the frequency of certain improper or undesirable behaviors in the United States will be stronger among international students than American students;

H1b: The relationship between viewing daytime television talk shows and holding negative attitudes toward human relationships in the U.S. will be stronger among international students than among American students; and

H1c: The relationship between viewing daytime television talk shows and having more negative perceptions about human relationships in primary groups in the U.S. will be stronger among international students than among American students.

Methods

Sample

Questionnaires were administered to a total of 320 American and international students at a large southeastern university. The American students were enrolled in introductory journalism and mass communication courses. About 80% of these U.S. students were non-majors or beginning majors who were enrolled in their first mass media course. The remainder were students majoring in public relations, advertising, or print journalism, academic areas that don't emphasize cultivation effects. All in all, it was unlikely that many of these students had prior knowledge of cultivation analysis or were overly sensitive to media effects.

The international students were recruited from various settings, including American Language Program classes, an international coffee hour sponsored by the university's Office of International Activities, and an e-mail listserve for international students. About 69% of the international students came from Pacific Rim countries. They had been in the U.S. an average of 1 year and 3 months. After eliminating incomplete and unusable surveys, there were 152 respondents in the American student group and 143 in the international student group. The two samples did not differ significantly in terms of gender (58% of the U.S. sample were females compared to 50% of the international sample), but the international student sample was older than the American sample and more likely to be graduate students. (2)

The American students watched more overall television than did the international students, averaging about 2 hours and 20 minutes compared to 1 hour and 30 minutes (p < .001). The two groups did not differ on talk show viewing frequency.

Independent Variables

One of the issues in cultivation analysis involves identifying the most appropriate measurement for the television exposure variable. Indeed, the choice of a measurement schema is fundamentally related to the conceptualization of cultivation itself. Gerbner, Gross, Morgan, and Signorielli (1986) argue that total time spent with television is the proper measure because, when considered at an abstract level, the world of television presents a message system that is consistent over time. Constant exposure to these messages is associated with a tendency to believe that the television world reflects what is actually the case in the real world. As a consequence, Gerbner and his associates assert that it is cumulative exposure to the whole world of television that is important (Morgan & Shanahan, 1997).

On the other hand, there seems to be growing dissatisfaction with a total viewing measure on both conceptual and empirical grounds. Potter (1993), for example, argues that the notion of consistent messages was more defensible during the 1970s when cultivation analysis was being developed. Television programming was provided primarily by three major networks operating in an oligopoly situation that put pressure on programmers to duplicate the offerings of competitors (Dominick & Pearce, 1976). Network domination has been declining the last two decades, and the emergence of 200-channel cable systems has meant more diversity of content for viewers.

Moreover, there is growing empirical evidence that a total viewing measure may not be as helpful in detecting a cultivation effect as more program-specific measures. When the total television exposure measure is compared to other program or genre-specific measures, the global scale has turned out to be less predictive. (See, for example, Allen & Hatchett, 1986; Carveth & Alexander, 1985; Hawkins & Pingree, 1980; Hoover, 1990; Pfau, Mullen, Deidrich, & Garrow, 1995; Potter & Chang, 1990).

Potter and Chang (1990) suggested that two other measures might be useful. One is a measure of exposure to a specific type of program genre; the second is a dominance measure based on the ratio of specific genre viewing to total television watching. Accordingly, in addition to measuring total television viewing time, the survey also measured exposure using both a genre-specific measure and a viewing dominance measure. Total television viewing time was gauged by the item "How many hours do you watch television per day?" with eight response categories, including "0" to "less than an hour" to "1 hour," "2 hours," etc., to "more than 5 hours." The genre-specific measure was constructed by giving respondents a list of nine daytime talk shows (the actual shows were Jenny Jones, Ricki Lake, Sally Jesse Raphael, Jerry Springer, Leeza, Maury Povich, Montel Williams, Oprah Winfrey, and Forgive or Forget) and were asked to report if they watched each show "never," (scored 0); "rarely," (scored 1); "sometimes" (scored 2); or "everyday," (scored 3). (3) Respondents' scores were summed and the resulting distribution was trichotomized to form three viewing groups: light, medium, and heavy talk show viewers. (4) The viewing dominance measure was created by asking respondents to report how many hours they watched television talk shows in the previous week and dividing this answer by the respondents' total weekly television viewing time. Total weekly television viewing time was calculated by taking the respondents' reported daily viewing time and multiplying by seven. All of the television viewing measures appeared on the first three pages of the survey instrument before the dependent variables were presented. (5)

Dependent Variables

Three different dependent variable measures were used. All asked about attitudes and perceptions concerning behaviors in the United States. The first involved perceptions of the frequency of certain inappropriate or undesirable behaviors among members of the U.S. population--a "first order" measure as defined by Gerbner et al. (1986). Respondents were asked to indicate on an 11-point scale with five-percent increments ranging from "less than 1%" (coded as 1%) to "50% or more" (coded as 50%) the percentage of people who engage in six activities commonly discussed on daytime talk shows, including infidelity, premarital sex, and running away from home. (6) The six deviant behavior items were taken from Davis and Mares (1998).

A second measure examined attitudes toward four areas of human relationships in the U.S.: family relationships, relationships between married couples, romantic relationships, and friendship. For each area, respondents filled out seven-point rating scales consisting of nine-pairs of polar adjectives (dishonest-honest; disloyal-faithful; untruthful-truthful; bad-good; aggressive-peaceful; unstable-stable; superficial-substantial; unfriendly-friendly; unselfish-selfish). These particular evaluative adjectives were drawn from research by Prothro and Keehn (1969) and Tan (1982). (7)

Confirmatory factor analysis revealed that for each of the four areas, these scales formed a single factor. (For family relationships, alpha = .95 with 74% variance explained; for relationships between spouses, alpha = .95 with 83% variance explained; for romantic relationships, alpha = .95 with 78% variance explained; and for friendship, alpha = .94 with 74% variance explained.) Consequently, for each of the four areas, scores were summed across scales and divided by nine to form a single evaluative score. The third area concerned how respondents perceived human relationships in three areas: (1) general relationships among people; (2) relationships within the family; and (3) relationships between spouses. Two indexes were taken from previous research to measure the first two areas. The first measure, examining perceptions about general human relationships in primary groups, was derived from the "faith in people scale" (also called a "misanthropy" scale) developed by Rosenberg (1957). There were eight items in this scale (e.g., "Most people can be trusted in America," "Many people in America are friendly only because they want something from you") with three items phrased in such a way that a "strongly agree" response indicated a negative perception and five phrased so that the same response would indicate a positive perception. The second scale consisted of six items and dealt with perceptions of relationships among family members in the U.S. This scale was based on items derived from "the family scale" (see Shaw & Wright, 1967) and from "the traditional family ideology scale" (Levinson & Huffman, 1955). Sample items included: "In the U.S., when planning for the future, parents are given first consideration;" and "In the U.S., parents usually treat children fairly and sensibly." Similar to the first index, there were five response categories to all items: "strongly agree," "agree," "neutral," "disagree," "strongly disagree." Alpha values for both scales were .89 and .81, respectively. Scores were summed across items and averaged for both indexes. (8)

One additional scale was developed for this study to measure perceptions of relationships between married couples. There were two items in this scale: "Husbands trust their wives completely in America," and "In America, most couples in a marriage are fair in their dealings with one another." The same response categories were used for this scale. The correlation between the two items was .47. Although this two-item scale seemed internally consistent and appeared to possess at least face validity, no systematic evaluation of its reliability and other validity dimensions was conducted. Consequently, results on this index should be treated with caution.

To sum up, there were three different types of dependent measures used: the "first order" measure of perceived frequencies of various behaviors, and two "second order" measures--one examining attitudes toward relations among family members, marriage partners, romantically involved couples, and friends; and a second measure that looked at perceptions of general human relationships, family relationships, and marriage relationships. (9)

Results

A series of regression analyses was performed first to examine which of the three measures of television exposure was most closely associated with the dependent variables. Along with the demographic variables of age, gender, and year in the university, (freshman, sophomore, etc.), the measures for total television viewing, television talk show viewing dominance, and total viewing of talk shows were entered into the regression equation. (10) Regression equations were computed separately for the international and American sample for the six estimates of deviant behavior, the four attitude scales, and the three perception scales for a total of 13 estimates for each sample.

Of the three demographic variables entered into the equation, gender and year in the university did not correlate with the dependent variables. Age was not correlated with any of the six estimates comprising the first order measure, but it was correlated with three of the four scales of the attitude measure. In the U.S. sample, age was positively correlated with attitudes toward family and romantic relationships, while among international students age was negatively correlated with both these items as well as with the item measuring attitudes toward friendship. Finally, among American students, age was positively correlated with the index measuring perception of family relationships in the U.S.

The results on the television viewing variables varied according to nationality. For the American sample, total television talk show viewing was the strongest predictor of all the dependent measures. All 13 of the standardized beta weights were significant. Only two of the possible 13 beta weights for the television talk show dominance variable were significant. Only two of the beta weights for total television viewing were significant.

For the international sample, the talk show dominance variable and total television talk show viewing were the strongest correlates. Ten talk show dominance viewing beta weights out of a possible 13 and 8 total television talk show viewing beta weights out of 13 were significant. Similar to the U.S. sample, only one of the beta weights for total television viewing was significant. Partialing out age and length of time in the U.S. reduced the correlations, but the beta weights remained significant.

In contrast to the U.S. sample, among international students, viewing dominance was more strongly related to the six estimates comprising the first order measure than was total television talk show viewing. Four of the possible six beta weights were significant for the dominance measure compared to two of six for the total television talk show viewing variable. For American students, all six of the total talk show viewing betas were significant for the first order measures compared to only one of the dominance measures. (11)

In sum, it appeared that the program-specific measure of total television talk show viewing was the best predictor of cultivation, with significant betas on 21 of 26 equations. The dominance measure was next in line. Total television viewing time was the weakest predictor. Since the measure of television talk show viewing was the strongest predictor, it was used in subsequent analyses. In order to test the main and sub-hypotheses, a series of MANOVAs were performed on the dependent variables using total television talk show viewing as the independent variable. Hypothesis 1a predicted that international students who are heavy viewers of daytime television talk shows would overestimate the frequency of certain improper or undesirable behaviors in the U.S. more than American students who are heavy viewers. Table 1 contains the means of each sample for the six items that called for behavior estimates.

Bartlett's test of sphericity for these six items was significant at the .001 level, indicating that the use of MANOVA was appropriate. The overall influences of nationality and talk show viewing were tested by Wilk's Lambda. The Lambda showed that estimates of deviant behavior were significantly affected by nationality (p < .001) and television talk show viewing (p < .001). Of more relevance to the hypothesis, however, is the interaction effect that was significant at the .05 level. Univariate ANOVAs performed on the six measures revealed that talk show viewing had a significant main effect on all six of the behavior estimates while the main effect for nationality was significant on five of the six measures. The interaction term, however, was significant in only two of the six items. The lack of more significant interaction terms in the univariate situation appears to be the result of high intercorrelations among the six items. (see Stevens, 1996). A post hoc examination of the means in Table 1 using Tukey's t test reveals that in every case the highest estimate was given by the international students in the high talk show viewing cell as suggested by the hypothesis.

Hypothesis 1b predicted that international students who are heavy viewers of daytime television talk shows would have more negative attitudes toward human relationships in the U.S. than American students who are heavy viewers. Table 2 contains the mean scores for each sample on the four scales that examined attitudes toward family relationships, marriage, romantic relationships, and friendship in the U.S.

Bartlett's sphericity test was significant at the .001 level, indicating that a MANOVA was appropriate. Similar to the results on the first order measure, the Lambdas for both the main effects for nationality and talk show viewing were significant at the .001 level. In addition, there was a strong interaction effect as well (p < .001). A univariate analysis of variance disclosed that the main effects were significant on all four measures and the interaction effect was significant for three of the four. Only the scale gauging attitudes of friendships had a non-significant interaction effect. Once again, as suggested by hypothesis 1b, the scores of the international students in the high talk show viewing cell were the most extreme.

Hypothesis 1c made the same prediction concerning perceptions of human relationships in the U.S. Table 3 contains the mean scores for each sample on the three scales that measured perceptions of relationships.

Bartlett's sphericity test was again significant at the .001 level. MANOVA results revealed that Wilks' Lambda was significant for the nationality and television talk show viewing main effects (p < .001 for both). In addition, the interaction term was also significant (p < .05). The univariate ANOVAs revealed a main effect for television talk show viewing on all scales and a main effect for nationality on the scales that measured perceptions of relationships among family members and between married couples. The interaction effect was also significant on these two scales as well. Table 3 provides a summary table that displays differences between the means using Tukey's t test. Once again, for two of the three scales (perception of family relationships and perceptions of spousal relationships), the international heavy talk show group has the most extreme score.

Discussion

In general, these results underscore the notion that cultivation is a complicated process that is made even more complicated when looking at cross-cultural differences.

One main finding is that program-specific measures and viewing dominance measures were more predictive of a cultivation effect than was the total television viewing time measure. This replicates the finding of Davis and Mares (1998) who found that overall viewing was not significantly related to their measures of behavior frequency estimates. This result seems understandable given the bizarre, unconventional, and sensationalized topics that are the hallmark of these programs. Those who favor overall viewing as the best measure (Morgan & Shanahan, 1997) argue that the total world of television manifests coherent, complimentary, and, at the broadest level, consistent portrayals and messages. The rather bizarre, outlandish, and unusual topics that permeate the daytime talk genre, however, seem to be themes that mainstream television programming generally ignores. In fact, perhaps part of the "side show" appeal of these shows may lie in the fact that they are the only place viewers can be exposed to this unconventional content. In short, exposure to the potential messages contained in this sort of content may only occur by watching this specific genre.

On a more general level, however, is it appropriate to call the results of this study (or any study that uses a genre-specific measure) evidence for a cultivation effect? Gerbner's original conceptualization proposed that understanding the impact of television as a cultural institution required looking at exposure to the totality of its message system. The similarities across different genres create a distinctive symbolic world. Examining the impact of exposure to specific genres does not fit this model. Nonetheless, this study and others suggest that there is some relationship between exposure to a specific type of program and specific views of social reality. Strictly speaking, however, cultivation may not be the best term to use to describe this relationship.

There were differences in impact by nationality. The program-specific measure produced the most significant beta weights overall, but the viewing dominance measure was more predictive among the sample of international students than among American students. The explanation for this difference merits further study.

Second, the results supported the main hypothesis: International students showed more of a cultivation effect than did American students on both first-order and second-order measures of cultivation. The findings mentioned above suggest one possible explanation for this. The data revealed that international students and American students watched the same amount of talk shows but that American students watched more television overall. Thus, talk show viewing accounted for a greater proportion of the viewing done by international students. The "bin model" of memory discussed earlier notes that frequency of information is one factor that contributes to what information is most remembered. Since the proportion of total television viewing accounted for by talk shows is greater among international students, talk show images concerning interpersonal relationships in the U.S. are relatively more frequent than among the American students. The American students would presumably have images of interpersonal relationships drawn from direct experience or from the increased viewing of other television programs that would make the talk show images less dense and relatively less frequent in their memories. Along those same lines, American students have more direct experience with relationships that may lead them to agree with Donovan (1997) and discount these programs as "carnival sideshows."

Additionally, the analysis of variance disclosed a cultivation effect among American students, a finding that also supports the Davis and Mares (1998) results. One unexpected finding from the data was the fact that international students held consistently more negative attitudes and perceptions about interpersonal relationships in America than did the U.S. students regardless of their television viewing habits. The reasons behind this discrepancy are unclear. Perhaps some international students were exposed to anti-U.S. information in their native country and already possessed some negative stereotypes before coming to the U.S. Or perhaps the political position between their mother country and the U.S. shaped their attitudes. Or perhaps international students hold more negative and skeptical attitudes in general.

This study, of course, had its limitations. As is the case with all survey research, these data contain no evidence about cause and effect. The theoretic base for cultivation analysis assumes that television viewing cultivates various attitudes and perceptions, but it is at least logically possible that causation might run the other way. Students with negative perceptions of interpersonal relationships might be drawn to these sensational talk shows simply because the shows reinforce their already existing attitudes.

Second, the samples were not randomly selected. Both samples depended on volunteers. In addition, the samples were recruited differently. Further, the international students came from a variety of countries and cultures and were older than their U.S. counterparts. It is possible that the samples were not typical of American and international students in general. Subsequent replications of this study are needed to establish the generalizability of these results.

Finally, there is considerable debate about the most appropriate way to analyze cultivation data. This study made choices regarding how it measured exposure and the criterion variables that seemed appropriate but that other researchers might challenge. Nonetheless, the results of this analysis highlight the importance of studying the cultivation phenomenon of different television genres across different nationalities and cultures.

Table 1
Multiple and Univariate Analysis of Variance Results for Estimates
of Frequency of Certain Behaviors

MANOVA

   Source            Wilks' Lambda      F        p

Talk show viewing         .80          5.28    <.001
Nationality               .75         15.20    <.001
Interaction               .92          1.97    <.05

UNIVARIATE ANOVA

                          Daytime Television Talk Show
                              Viewing (Cell means)

    Item                  Light                  Medium

Percentage of husbands' infidelity
  American          22.[4.sub.ef] (N = 58)     27.[4.sub.f] (N = 46)
  International     29.[9.sub.f] (N = 57)      32.[7.sub.ae] (N = 36)

Percentage of wives' infidelity
  American          16.[9.sub.f] (N = 58)      20.[7.sub.f] (N = 46)
  International     24.[1.sub.f] (N = 57)      27.[2.sub.f] (N = 36)

Percentage of teenagers pregnant
  American          15.[6.sub.ef] (N = 58)     16.[6.sub.ef] (N = 46)
  International     20.5 (N = 57)              28.[0.sub.ab] (N = 36)

Percentage of boys premarital sex
  American          39.8 (N = 58)              44.4 (N = 46)
  International     38.5 (N = 57)              40.3 (N = 36)

Percentage of girls premarital sex
  American          33.[4.sub.e] (N = 58)      39.3 (N = 46)
  International     37.7 (N = 57)              39.0 (N = 36)

Percentage of teen runaways
  American          12.[1.sub.ef] (N = 58)     17.[5.sub.f] (N = 46)
  International     20.[7.sub.f] (N = 57)      25.[5.sub.a] (N = 36)

                     Daytime Television Talk Show
                              Viewing (Cell means)

    Item                  Heavy

Percentage of husbands' infidelity
  American          29.[3.sub.f] (N = 47)
  International     42.[9.sub.abcde] (N = 48)

Percentage of wives' infidelity
  American          24.[8.sub.f] (N = 47)
  International     40.[6.sub.abcde] (N = 48)

Percentage of teenagers pregnant
  American          24.5 (N = 47)
  International     32.[2.sub.ab] (N = 48)

Percentage of boys premarital sex
  American          45.4 (N = 47)
  International     47.0 (N = 48)

Percentage of girls premarital sex
  American          41.3 (N = 47)
  International     46.[1.sub.a] (N = 48)

Percentage of teen runaways
  American          21.[2.sub.f] (N = 47)
  International     33.[5.sub.abcd] (N = 48)

    Item                   ANOVA Summary

Percentage of husbands' infidelity
  American          Talk show viewing (p < .001)
  International     Nationality (p < .001)
                    Interaction (p < .05)

Percentage of wives' infidelity
  American          Talk show viewing (p < .001)
  International     Nationality (p < .001)
                    Interaction (p < .05)

Percentage of teenagers pregnant
  American          Talk show viewing (p < .001)
  International     Nationality (p < .001)
                    Interaction (n.s.)

Percentage of boys premarital sex
  American          Talk show viewing (p < .001)
  International     Nationality (n.s.)
                    Interaction (n.s.)

Percentage of girls premarital sex
  American          Talk show viewing (p < .001)
  International     Nationality (p < .05)
                    Interaction (n.s.)

Percentage of teen runaways
  American          Talk show viewing (p < .001)
  International     Nationality (p < .001)
                    Interaction (n.s.)

Note: For these and the following tables, post hoc tests of mean
differences are reported in the following manner. Cells in the
top row of the matrix from left to right are given the letters a, b,
and c. Cells in the bottom row from left to right are given the
letters d, e, and f. The subscript letter within each cell
identifies significant differences from other cells at the .05
level via a Tukey's t test. In the above table, for example,
the mean in cell "a" is significantly different from the means
in cells "e" and "f"; the mean in cell "b" is significantly
different from the mean in cell "f," etc.
Table 2
Multiple and Univariate Analysis of Variance Results
for Attitudes toward Human Relationships

MANOVA

Source              Wilks' Lambda     F       p

Talk show viewing        .73        11.92   <.001
Nationality              .75        23.26   <.001
Interaction              .88         4.14   <.001

UNIVARIATE ANOVA

                             Daytime Television Talk
                            Show Viewing (Cell means)

    Item                     Light                   Medium

Attitude toward family relationships
  American               3.1[6.sub.f] (N = 57)   3.0[6.sub.f] (N = 47)
  International          3.5[3.sub.f] (N = 54)   3.5[2.sub.f] (N = 32)

Attitude toward spousal relationships
  American               3.3[5.sub.f] (N = 57)   3.3[8.sub.f] (N = 47)
  International          3.8[2.sub.f] (N = 54)   3.8[4.sub.f] (N = 32)

Attitude toward romantic relationships
  American               2.8[7.sub.f] (N = 57)   2.5[8.sub.ef] (N = 47)
  International          3.7[3.sub.f] (N = 54)   3.8[4.sub.bf] (N = 32)

Attitude toward friends
  American               2.5[6.sub.f] (N = 57)   2.5[9.sub.f] (N = 47)
  International          3.57 (N = 54)           3.61 (N = 32)

                           Daytime Television Talk
                            Show Viewing (Cell means)

    Item                     Heavy

Attitude toward family relationships
  American               3.9[3.sub.f] (N = 47)
  International          5.2[1.sub.abcde] (N = 47)

Attitude toward spousal relationships
  American               4.1[2.sub.f] (N = 47)
  International          5.8[3.sub.abcde] (N = 47)

Attitude toward romantic relationships
  American               3.5[7.sub.f] (N = 47)
  International          5.8[3.sub.abcde] (N = 47)

Attitude toward friends
  American               3.3[4.sub.f] (N = 47)
  International          4.3[9.sub.abc] (N = 47)

    Item                    ANOVA Summary

Attitude toward family relationships
  American               Talk show viewing (p < .001)
  International          Nationality (p < .001)
                         Interaction (p < .05)

Attitude toward spousal relationships
  American               Talk show viewing.(p < .001)
  International          Nationality (p < .001)
                         Interaction (p < .05)

Attitude toward romantic relationships
  American               Talk show viewing (p < .001)
  International          Nationality (p < .001)
                         Interaction (p < .001)

Attitude toward friends
  American               Talk show viewing (p < .001)
  International          Nationality (p < .001)
                         Interaction (n.s.)

Note: Means based on a seven-point rating scale (1 = strongly
positive; 2 = quite positive; 3 = slightly positive; 4 = neutral;
5 = slightly negative; 6 = quite negative; 7 = strongly negative).
Different subscripts reflect significant differences between cell
means using Tukey's t test.
Table 3
Multiple and Univariate Analysis of Variance Results
for Perceptions of Human Relationships

MANOVA

  Source             Wilks' Lambda      F       p

Talk show viewing        .72         12.41   <.001
Nationality              .85         15.08   <.001
Interaction              .92          2.34   <.05

UNIVARIATE ANOVA

                         Daytime Television Talk Show
                            Viewing (Cell means)

     Item               Light                     Medium

General perception of human relationships *
  American           2.85 (N = 57)             2.61 (N = 46)
  International      2.74 (N = 54)             2.71 (N = 35)

General perception of family relationships **
  American           2.7[1.sub.f] (N = 57)     2.6[0.sub.cf] (N = 46)
  International      2.90 (N = 54)             3.05 (N = 35)

General perception of spousal relationships **
  American           2.6[1.sub.f] (N = 57)     2.6[1.sub.f] (N = 46)
  International      2.7[8.sub.f] (N = 54)     2.9[2.sub.f] (N = 35)

                       Daytime Television Talk Show
                            Viewing (Cell means)

     Item                 Heavy

General perception of human relationships *
  American           2.66 (N = 46)
  International      2.50 (N = 48)

General perception of family relationships **
  American           3.1[8.sub.bf] (N = 46)
  International      3.9[6.sub.bc] (N = 48)

General perception of spousal relationships **
  American           3.0[7.sub.f] (N = 46)
  International      4.0[6.sub.abcde] (N = 48)

     Item                ANOVA Summary

General perception of human relationships *
  American           Talk show viewing (p < .001)
  International      Nationality (n.s.)
                     Interaction (n.s.)

General perception of family relationships **
  American           Talk show viewing (p < .001)
  International      Nationality (p < .001)
                     Interaction (p < .01)

General perception of spousal relationships **
  American           Talk show viewing (p < .001)
  International      Nationality (p < .001)
                     Interaction (p < .01)

Note: Based on a five-point scale (5 = strongly disagree; 4 = disagree;
3 = neutral; 2 = agree; 1 = strongly agree). Different subscripts
represent significant differences between means using Tukey's t
test. * the lower the mean, the more negative the perception.
** the higher the mean, the more negative the perception.

Notes

(1) For example, a sampling of show titles taken from the 1998-1999 season included the following: "Multiple Infidelities," "Animosity between Mothers and Daughters," "Wanting to Date a Friend's Ex," "Infidelity in Relationships," "Disapproving of Mate's Relationship with his Mother," "Sexual Addiction," "Teens Harassed for Being Geeks," "Video Voyeurism," "Revealing Harmful Secrets to a Fiance," "Teen Girls Harassed by Classmates," "Incest," and "Transsexualism." It is true, of course, that daytime talk shows are not the only programs on television that contain portrayals of the behaviors listed above. Nonetheless, the elements of apparent unscripted reality, audience response, and the unrelenting nature of the presentations make daytime talk programs distinctive. Consequently, this genre seems an appropriate candidate for a cultivation analysis.

(2) More specifically, 64% of the international sample were graduate students. The demographics and country of residence of this sample were similar to the demographics and country of residence of the population of international students at the university.

(3) Scores on this measure varied from 0 to 27. The average was 6 with a standard deviation of 6. The mean of the viewing dominance measure was .10 with a standard deviation of .05. The total television viewing measure had a mean of 2.9 and a standard deviation of 1.3. All measures were moderately skewed to the right.

(4) As Potter (1994) correctly points out, the choice of cut points is an important decision in cultivation analysis. In accordance with his recommendations, total talk show viewing scores were plotted against the dependent measures. (Scattergrams of the data are available from the authors upon request.) The distributions on the first-order measures were linear. For the second-order measures, however, the distributions were generally non-linear and resembled a reclining backward "J," or what Potter described as an exponential curve. Low and medium viewers showed much less of a cultivation effect than did high viewers. To see if the three cut points used in the subsequent analyses created artificial results, the data were analyzed with the respondents divided into quartiles. The pattern of results was the same: the lowest three groups showed little cultivation impact, with the high group demonstrating the most. In addition, following the example of Potter (1991), we divided the distribution into quintiles and re-did the MANOVAs. The pattern of significant results was the same. The first three quintiles were generally not different from one another, while the upper two differed from the rest. (These results should be treated with caution because some of the cell Ns were less than 25.) We next divided up the sample according to the standard deviation of the talk show viewing variable and did the MANOVAs again. Scores less than one S.D. below the mean were re-coded as "low," scores greater than one S.D. above the mean were coded "high," while the rest were coded "medium." All of the main effects for nationality and talk show viewing remained the same, but several of the interaction effects were rendered non-significant. (Again, these results should be treated with caution because some of the cell means had Ns of less than 25.)

(5) Shrum (1995) suggests that gathering television data first might actually reduce or erase a cultivation effect. If so, then the significant results obtained in this study are even more impressive. On the other hand, Morgan and Shanahan (1997) report that gathering viewing data first seems to result in an increased cultivation effect.

(6) Although used in many studies, an open-ended estimate of percentages as a cultivation measure has been the subject of some controversy. Potter (1994) endorses them but Morgan and Shanahan (1997) question their validity. This particular scale was a close-ended scale with response options: "<1%," "5%," "10%," etc., up to "> than 50%."

(7) It is possible that differences on this and the other measurement scales might have resulted from different cultural response patterns. Americans, for example, might make use of the extreme scores on a scale (e.g., "1" and "7") more often than international students. To check this possibility, the number of "1" and "7" responses on these seven-point scales by each sample were compared. U.S. students were more likely to use the extreme response options, but this difference was not significant.

(8) Some minor editing of these scales was done to make the language more current. Although several decades old, these measures are still considered useful (see Shaw & Wright, 1967; Wrightsman, 1991).

(9) Preliminary analysis indicated that there was some intercorrelation between the two second-order measures. In fact, the seven-step rating scale for attitudes toward family relationships correlated .60 with the five-step Likert scale that measured perceptions of family relationships. In like manner, the seven-step scale that measured attitudes toward relationships between marriage partners correlated .62 with the five-step Likert scale that measured perceptions of the same phenomenon. Consequently, it is possible that both sets of the second order measures were gauging the same underlying concept. Although this finding limits the breadth of the results, it also increases their reliability since similar results were found using two different measurement techniques.

(10) The correlations between total television viewing and the dominance measure was -.21 and .29 with the genre-specific measure. The correlation between dominance and the genre measure was substantially higher at .67. Conley and Pollard (1998) suggest that multicolinearity is a problem only when correlations among predictor variables exceed .7.

(11) Frequent talk show viewers might note that some shows in our sample are more confrontational and argumentative than others. To test if this made a difference, we removed three shows that might be considered more "benign" than some of the others (specifically, "Oprah," "Montel Williams," and "Leeza") and re-did the regression analysis using only the viewing scores from the other six shows. The betas that were significant using the original measure were still significant using this alternative measure.

References

Allen, R., & Hatchett, S. (1986). The media and social reality effects. Communication Research, 13(1), 97-123.

Alter, J. (1995, November 6). Next: The revolt of the revolted. Newsweek, 126, pp. 46-47.

Carveth, R., & Alexander, A. (1985). Soap opera viewing motivations and the cultivation process. Journal of Broadcasting & Electronic Media, 29(3), 259-273.

Choi, J. H., & Tamborini, R. (1988). Communication-acculturation and the cultivation hypothesis. Howard Journal of Communication, 1(1), 57-74.

Conley, D., & Pollard, D. (1998). Model comparisons: Direct and indirect effects. [Online]. Available: www.yale.edu/soc119a.htm.

Davis, S., & Mares, M. (1998). Effects of talk show viewing on adolescents. Journal of Communication, 48(3), 69-86.

Dominick, J. R., & Pearce, M. (1976). Trends in network prime-time programming, 1953-1974. Journal of Communication, 26(1), 70-80.

Donovan, S. (1997, December 15). New highs--and lows--in daytime. Broadcasting and Cable, 127, pp. 42-54.

Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1986). Living with television: The dynamics of the cultivation process. In J. Bryant & D. Zillmann (Eds.), Perspectives on media effects (pp. 17-40). Hillsdale, NJ: Erlbaum.

Greenberg, B., Sherry, J. L., Busselle, R. W., Hnilo, L. R., & Smith, S. (1997). Daytime television talk shows. Journal of Broadcasting & Electronic Media, 41(3), 412-426.

Greenberg, B., & Smith, S. (1995, November). The content of television talk shows. Report prepared for the Kaiser Family Foundation.

Hawkins, R., & Pingree, S. (1980). Some processes in the cultivation effect. Communication Research, 7(2), 193-226.

Hoover, S. (1990). Television, religion, and religious television. In N. Signorielli & M. Morgan (Eds.), Cultivation analysis: New directions in media effects research (pp.123-140). Newberry Park, CA: Sage.

Levinson, D., & Huffman, P. (1955). Traditional family ideology and its relation to personality. Journal of Personality, 23(2), 251-273.

Morgan, M., & Shanahan, J. (1997). Two decades of cultivation research. In B. R. Burleson & A. W. Kunkel (Eds.), Communication yearbook 20, (pp. 1-45). Thousand Oaks, CA: Sage.

Pfau, M., Mullen, L., Deidrich, T., & Garrow, K. (1995). Television viewing and public perception of attorneys. Human Communication Research, 21(3), 307-330.

Potter, W. (1991). The linearity assumption in cultivation research. Human Communication Research, 17(4), 562-583.

Potter, W. (1993). Cultivation theory and research: A conceptual critique. Human Communication Research, 19(4), 564-601.

Potter, W. (1994, October). Cultivation theory and research: A methodological critique. Journalism Monographs, 147, 1-34.

Potter, W., & Chang, I. (1990). Television exposure measures and the cultivation hypothesis. Journal of Broadcasting & Electronic Media, 34(3), 313-333.

Prothro, E. T., & Keehn, J. D. (1969). Stereotypes in semantic space. In J. G. Snider & C. E. Osgood (Eds.), Semantic differential technique (pp. 441-453). New York: Aldine Publishing.

Rosenberg, M. (1956). Misanthropy and political ideology. American Sociological Review, 21(6), 690-695.

Rubin, A. M., & Step, M. M. (1997). Viewing television talk shows. Communication Research Reports, 14(1), 106-115.

Shaw, M. E., & Wright, J. M. (1967). Scales for the measurement of attitudes. New York: McGraw-Hill.

Shrum, L. J. (1995). Assessing the social influence of television. Communication Research, 22(4), 402-429.

Shrum, L. J. (1999). Television and persuasion: Effects of programs between the ads. Psychology & Marketing, 16(2), 119-140.

Shrum, L. J., & O'Guinn, T. C. (1993). Processes and effects in the construction of social reality. Communication Research, 20(3), 436-471.

Stevens, J. (1996). Applied multivariate statistics for the social sciences. Hillsdale, NJ: Erlbaum.

Tan, A. (1982). Television use and social stereotypes. Journalism Quarterly, 59(1), 119-122.

Tamborini, R., & Choi, H. J. (1988). The cultivation of intercultural viewers. Unpublished paper.

Tamborini, R., & Choi, J. (1990). The role of cultural diversity in cultivation research. In N. Signorielli & M. Morgan (Eds.), Cultivation analysis (pp. 157-180). Newbury Park, CA: Sage.

Tyler, T., & Cook, F. (1984). The mass media and judgments of risk. Journal of Personality and Social Psychology, 47, 693-708.

White, D. (1993, June 2). The warped, weird world of TV talk shows. Boston Globe, p. 31.

Woo, H. J. (1998). Topics of TV talk shows. Unpublished paper. Department of Telecommunications, University of Georgia.

Wrightsman, L. (1991). Faith in people. In J. P. Robinson, R. S. Phillips, & L. S. Wrightsman (Eds.), Measures of social psychological attitudes (pp. 50-53). San Diego, CA: Academic Press.

Wyer, R., & Srull, T. (1989). Memory and cognition in its social context. Hillsdale, NJ: Erlbaum.

Hyung-Jin Woo is a doctoral candidate in the Grady College of Journalism and Mass Communication at the University of Georgia. His research focuses on international communication and computer mediated communication.

Joseph R. Dominick (Ph.D. Michigan State University) is a Professor in the Department of Telecommunications in the Grady College of Journalism and Mass Communication at the University of Georgia. His research interests are in media content and social effects of mass communication.

An earlier version of this paper won first place in the open competition of the Multicultural Studies Division at the 2000 convention of the Broadcast Education Association.

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.