Abstract
This study examines the presence and causes of pluralistic media ignorance – incorrect estimation of media content in a way that creates the impression that popular broadcasting is overabundant with morally controversial material. Using the appearance of sexual content in television advertising in Israel as a case study, we compare survey estimates (N = 305) of the prevalence of this material with its actual level according to the findings of a content analysis of commercials (N = 752). Pluralistic media ignorance emerges as a predominant norm, when viewers are requested to estimate the prevalence of “soft core” content like kissing and partial nudity, but it is a minor phenomenon, when estimating the share of more explicit material such as intercourse, homosexuality and rape. The most prominent predictors of pluralistic media ignorance are disapproval of the inclusion of objectionable content in popular media and extended recall of media exemplars that contain such content. Religiosity, ethnicity and sex are only partly predicting the emergence of pluralistic media ignorance. Media consumption, fear of media influence and other demographics are not predicting it at all. Public policy implications are discussed and suggestions for future research are given.
Acknowledgements
The author thanks Revital Solomovich, who served as research assistant in this project and extends his indebtedness to the editor of Social Science Journal, N. Prabha Unnithan, and three anonymous reviewers the beneficial editorial comments.
Notes
1 The actual low share of sex in TV advertising (less than 0.5% for some of the indicators) prevented us from offering values that underestimate the prevalence of this material as optional answers. As shows, none of the items shows a tendency on the part of the respondents to choose in particular the scale's mid-value, which is an exaggerated estimate. This further strengthens the credibility of the estimates measures format.
2 Ordinal regression was chosen because the dependent variable – estimates of the prevalence of sexual content in TV advertising– forms a three-point ordinal scale. As suggested by CitationMcCullagh (1980) in his overview of this procedure, the link function chosen when the dependent measures were kissing and patting and female nudity was complementary log-log because in these variables the higher estimates have largest frequency. When modeling the regression on homosexuality, intercourse and rape, whose distributions have higher preponderance of lower estimates, the link function used was negative log-log. Since the test of significance of the regression coefficient in this model is based on the Wald statistic, which is square of the ratio of the coefficient to its standard error, large coefficients of variables that are greatly dispersed are not always significant.