Method
Procedure and sample . A random‑digit‑dialing telephone survey was conducted in Spring 1991 among adults living off‑campus in a university town in the Midwest. Out of the 813 valid attempts (excluding business numbers and no answers), there were 615 completions and 198 refusals, for a 75.6% completion rate. The sample was 45.1% male and ranged in age from 17 ‑ 93 (M  = 36.27, SD  = 17.01). The average respondent had completed 14.45 years of education (ranging from 8 ‑ 20 years, SD  = 2.45). Hollingshead’s two‑factor social position index (Miller, 1983) measured occupational level (lower scores reflect higher occupational levels) and ranged from 11 ‑ 73 (M  = 46.50,SD  = 18.52).
Television exposure . Respondents indicated how many hours they viewed ”yesterday” and ”on a typical day.” Averaged television exposure ranged from 0 to 17 hours a day (M  = 2.98, SD  = 2.24).
New technology use . Because the focus of this second study was on the impact of new television technology use and channel repertoire on cultivation effects of television, we considered six media use variables in this study: Cable subscription, VCR ownership, RCD ownership, channel changing, and two types of channel repertoire.
Of the sample, 68.7% subscribed to cable television (comparable to national figures of 59%). More than three‑quarters of the sample (76.9%) had access to a VCR where they lived (comparable to the national average of 73%). Of the sample, 73.5% had access to a RCD where they lived (once again, comparable to the national average of 77%).
Channel repertoire (CR) is defined as the number of channels that a respondent typically watches (Heeter, 1985; Ferguson, 1992a). For this study, we created two CR measures. Broadcast channel repertoire (BCR) was defined as the sum of the broadcast channels and the cable channels that are nearly identical to broadcast channels for which at least some response (in hours) was given using unaided recall (Ferguson, 1992a; Ferguson & Perse, 1992). These included network affiliates, independent stations, superstations (e.g., WTBS), and cable networks (e.g., USA Network). Broadcast channel repertoire represents exposure to channels that offer much the same content as network programming (”more of the same”). Broadcast channel repertoire had a possible range of 0 to 9 channels, given the capacity of the only cable system in the sampling area during the survey. Broadcast channel repertoire ranged from 0 ‑ 8 channels (M = 3.59, SD = 1.22).
Cable channel repertoire (CCR), on the other hand, was operationalized as the sum of all cable networks for which at least some response (in hours) was given using unaided recall (Ferguson, 1992a; Ferguson & Perse, 1992). Cable channel repertoire reflects exposure to channels that offer more specialized content that differs from network programming (e.g., FNN, The Weather Channel, CNN, MTV, etc.). The possible range was from 0 to 31 channels, given the capacity of the only cable system in the sampling area during the survey. Cable channel repertoire ranged from 0 ‑ 20 (M  = 2.76, SD  = 2.36).
Channel changing . Ferguson (1992b) reported on the unreliability of asking respondents to indicate how many times per hour they change channels. Such mundane behavior is difficult to recall precisely. Instead, respondents with RCDs described their frequency (1 = never, 2 = seldom, 3 = often, 4 = very often) to the question ”How often do you flip channels?” The mean score was 2.37 (SD = 0.81).
Cultivation measures . Respondents indicated their agreement (1 = strongly disagree, 4 = strongly agree) with three statements about their fear of crime and three statements concerning interpersonal mistrust.3 These items were drawn from earlier cultivation analyses (Gerbner et al., 1978; Gerbner et al., 1979). Item responses were summed to create scale scores. Fear scores ranged from 3 ‑ 12 (M = 7.62, SD = 1.80, alpha = .78). Mistrust scores ranged from 3 ‑ 12 (M = 7.06, SD = 1.15, alpha = .60).
Statistical analysis . There were two stages to data analysis. First, Pearson correlations were computed to explore the bivariate relationships between the variables of study. Then, hierarchical multiple regression was used to test the hypotheses of the study. In all cases, demographics were entered on the first step to control for any variance they might contribute to the equation. Television exposure was entered on the second step, followed by the new technology use variables entered on the third step.