Always on My Mind: Exploring How Frequent, Recent, and Vivid Television Portrayals Are Used in the Formation of Social Reality Judgments

Abstract

Prior research has found consistent support for the heuristic processing model of cultivation effects, which argues that cultivation effects can be explained by the availability heuristic. The present study represents an experimental test of the heuristic processing model and tests the impact of frequency, recency, and vividness on construct accessibility and social reality beliefs. 213 students participated in a 2 × 2 × 2 prolonged exposure experimental design varying the frequency of exposure to violent television programs, the level of vividness in the programs, and recency of exposure. Dependent measures were accessibility and social reality beliefs. Results showed that reaction times were largely unresponsive to the independent variables. Although there were no main effects for frequency on social reality beliefs, there was a significant interaction between frequency and vividness on beliefs: People watching vivid violent media gave higher estimates of the prevalence of crime and police immorality in the real world in the 3× viewing condition than those in the 1× viewing condition. In concluding, it is argued that this study has important implications for the heuristic processing model, cultivation theory, and research into vividness effects.

Notes

1. There was also a no exposure control group that led to an experimental design that was not fully nested. Because there were only 10 participants in this control group, the no exposure control group was not included in any of the analyses.

2. Initially, skew and kurtosis values for 11 of the prevalence estimate questions fell outside acceptable ranges (Kline, 2005). Tabachnick and Fidell (2001) argue that standardized scores greater than 3.29 standard deviations from the mean are potential outliers. To be extra cautious in the present study, only scores that were greater than four standard deviations from the mean were recoded as missing data. After removing these extreme outliers (n = 35), the revised frequency distributions for nine of the questions were still skewed positively. A square root transformation reduced the skewness problem for four of the items, and a logarithmic transformation reduced the skewness problem for four more items. The remaining two prevalence estimate items were removed from analysis when no transformations could reduce the skewness problem.

3. Based on the factor analysis, the amount of variance explained by these three factors was 45%.

4. The range of possible responses for the general crime and personal susceptibility scales ranged from 0 to 100. If the means were not transformed into z scores, the raw data averages would be as follows: general crime belief scale (M = 20.58, SD = 14.84), personal susceptibility scale (M = 9.70, SD = 10.11).

5. To determine whether or not any differences occurred when watching episode A or C last, data from participants in the vivid and nonvivid conditions were subjected to three separate independent samples t tests. Order of presentation was the independent variable, and prevalence estimates about crime in general, police morality, and personal beliefs were the three separate dependent variables. In the vivid condition, the order of presentation had no impact on participants' scores when employing prevalence estimates about general crime beliefs or police morality as the dependent variable. There was, however, an effect for presentation order that was nearing significance on prevalence estimates about personal crime susceptibility, t(107) = −1.87, p = .064. An analysis of means showed that prevalence estimates about personal crime susceptibility were higher when participants viewed episode A last than when they saw episode C last. In the nonvivid viewing conditions, there was an effect nearing significance for order on prevalence estimates about general crime, t(102) = −1.87, p = .065. Once again, prevalence estimates were higher when participants viewed the nonvivid episode A last. There was also a significant effect for presentation order on prevalence estimates of police immorality, t(102) = −2.82, p < .01. Once again, prevalence estimates were higher when participants watched the nonvivid episode A last. There was no effect for presentation order on the personal crime susceptibility scale.

6. The heuristic-processing model, more specifically, argues that accessibility mediates the relationship between television exposure and social reality beliefs. Because there were no effects for accessibility at all, no tests were performed to determine whether or not it mediates the relationship between exposure and social reality beliefs.