Appendix
Coding for Control Variables and Descriptive Statistics
The analyses incorporate the demographic controls that are typically present in research on both media effects and criminal justice attitudes. These include respondents’ gender (Female = 1), race (White =1), Age, Education, Income, political ideology (Conservative = 1), and region of residence (South =1). In Samples 1 and 2, Age is measured ordinally, such that 1= 18 to 24 years, 2 = 25 to 34 years, 3 = 35 to 44 years, 4 = 45 to 54 years, 5 = 55 to 64 years, 6 = 65 to 74 years, and 7 = 75 or over. For Samples 3 and 4, Age is measured continuously (in years). For Samples 1, 2, and 4, the response categories for the Education variable range from 1 = no high school diploma to 5 = graduate degree. In Sample 3, the response categories for this variable range from 1 = high school degree or less to 4 = graduate degree.
For Samples 1 and 2, Income is measured on a six point scale ranging from 1 = less than $15,000 to 6 = more than $100,000. In Sample 3, Income is measured on a five-point scale ranging from 1 = up to $24,999 to 5 = $150,000 or more. In Sample 4, Income is measured on a 14-point scale ranging from 1 = less than $10,000 to 14 = $150,000 or more. To avoid losing respondents solely due to nonresponse about income, for each sample, we imputed missing values for the measure using scores on the other explanatory variables. This did not appreciably alter the results.
The data from Samples 1 and 3 also allow us to control for whether anyone in the respondents’ family has been the victim of a crime in the past five years (Household Victim = 1). And the data from Sample 3 permit us to account for whether the respondent personally has been the victim of a crime in the past five years (Victim =1). The data from Sample 4 allow us to control for whether the respondent reports being born again religiously (Born Again =1), and for the respondent’s attendance at religious services (Religious Attendance: 1 = never, 9 = more than once a week). Including these controls in the models for Sample 4 is important because studies show that religious beliefs influence views about the death penalty (Unnever et al. 2006).
Finally, in the analyses for Sample 3, we are also able to control for respondents’ Concern about community social deterioration and Neighborhood Disorder. Concern is measured with the following four-item question: “In your community, how much has each of the following decreased or increased in the past five years? 1) the sense of belonging among residents; 2) the sense of trust among residents; 3) the sense of right and wrong among residents; 4) the sense of shared responsibility for the community among residents.” The original response options ranged from 1 = decreased greatly to 5 = increased greatly. We reversed the responses so that higher values indicated negative social changes, and then averaged across the four items (alpha = .88). Neighborhood Disorder is equal to the average across respondents’ responses to five questions that asked how big of a problem (1 = not a problem, 4 = a big problem) each of the following are in their neighborhoods: 1) litter and trash, 2) graffiti, 3) run-down houses, 4) noisy neighbors, and 5) teenagers hanging out on corners (alpha = .81).
Table A1 about here
Table A1. Descriptive Statistics
Sample 1 / Sample 2 / Sample 3 / Sample 4Variables / Mean / SD / Mean / SD / Mean / SD / Mean / SE
Dependent Variables
Police powers / 4.01 / 2.61 / — / — / — / — / — / —
Punitive attitudes‡ / 5.94 / 2.05 / 5.60 / 2.51 / — / — / — / —
Perceived risk‡ / 3.76 / 2.17 / 2.97 / 2.29 / 17.35 / 19.05 / — / —
Favors death penalty / — / — / — / — / — / — / .71 / .01
Independent Variables
Hours of Exposure
National TV news (ln) / 1.51 / 1.05 / — / — / — / — / — / —
Local TV news (ln) / 1.53 / .90 / 1.78 / .98 / — / — / — / —
TV crime programs (ln)‡ / 1.17 / 1.11 / 1.14 / 1.07 / — / — / — / —
Internet news (ln)‡ / 1.03 / 1.06 / .78 / .92 / — / — / — / —
Days of Exposure
National TV news / — / — / — / — / 2.05 / 2.48 / — / —
Local TV news‡ / — / — / — / — / 3.13 / 2.65 / 2.41 / .03
TV crime programs / — / — / — / — / 1.82 / 2.13 / — / —
Internet news‡ / — / — / — / — / 4.61 / 2.42 / 3.22 / .03
Control Variables
Female / .56 / .50 / .56 / .50 / .53 / .50 / .49 / .01
White / .78 / .42 / .85 / .36 / .85 / .35 / .79 / .01
Age‡ / 3.94 / 1.58 / 4.70 / 1.63 / 49.56 / 15.79 / 49.72 / .19
Education‡ / 3.16 / 1.13 / 3.37 / 1.12 / 2.81 / .99 / 3.06 / .02
Income‡ / 3.67 / 1.64 / 3.61 / 1.60 / 2.92 / 1.30 / 8.12 / .04
Household victim / .28 / .45 / — / — / .28 / .45 / — / —
Victim / — / — / — / — / .18 / .39 / — / —
Concern / — / — / — / — / 3.10 / .67 / — / —
Disorder / — / — / — / — / 1.48 / .56 / — / —
Born again / — / — / — / — / — / — / .36 / .01
Religions attendance / — / — / — / — / — / — / 4.52 / .03
Conservative / .41 / .49 / .48 / .50 / .30 / .46 / .41 / .01
South / .36 / .48 / .38 / .49 / .28 / .45 / .36 / .01
NOTES: ‡Means and standard deviations are not comparable across the samples due to differences in measurement or response categories.
ABBREVIATIONS: SD = standard deviation, SE = linearized standard error.