Comedy as a Gateway
Page 19
than did no-comedy viewers. The initial advantage in attention observed in Figure 2 for Daily
Show viewers relative to Leno/Letterman viewers was not, however, discernible at levels of
statistical significance.
In addition to its higher intercept, another characteristic of the Daily Show viewers’ time
series that appeared, in Figure 2, to distinguish it from that of the other two sub-groups was its
stable linear trend. That is, the pattern of increasing attention seen in the trend line for Daily
Show audiences looks to be unaffected by specific campaign events like the Iowa caucuses.
Indeed, the effect of the Iowa caucuses on The Daily Show time series was not found to be
significant (see Table 2). There is some concern, however, that due to the relatively smaller
daily sample sizes and wider sampling variability of The Daily Show audience compared to the
other two groups, statistical power would not be sufficient to detect an effect of the Iowa
caucuses. For this reason, a power analysis was conducted (Power & Precision Version 1.2)
based on the R
2
of the initial curve fit for The Daily Show model. This analysis indicated that
there was enough power to detect an effect roughly half the size of the Iowa effect observed in
the Leno/Letterman model. Thus, although we can’t say with absolute certainty that there was
no impact of the Iowa caucuses on Daily Show viewers’ attention to campaign news, if there was
an effect, it was substantially smaller than it was for the other two groups.
To test whether watching late-night comedy increases the rate at which people pay
attention to campaign news on television, we focused on the differences between the
Leno/Letterman time series and the non-viewer time series that were observed in Figure 2. A
comparison of the coefficients for the Iowa effect in Model 2 (0.227 vs. 0.216) failed to yield a
significant difference, t(418)=0.38, p=.71. Thus, the spike in news attention prompted by the
Iowa caucuses was no greater for viewers of Leno and Letterman than it was for individuals who
watch no late-night comedy. We also tested the difference between the overall linear trends
Convention