Page 240 - Information and American Democracy Technology in the Evolution of Political Power

P1: IBE/IRP/IQR/IRR
August 13, 2002
CY101-Bimber
12:12
CY101-05
0 521 80067 6
Information Technology and Political Engagement
The finding for attending a political event is more readily explained, since
many candidates in 2000 used the Internet to attempt to organize and an-
nounce campaign events, as we saw in the last chapter. The voting effect
suggests either that Internet-based information was a factor in turnout or
that some other factor not captured directly in the model influenced both
votinganduseoftheInternetforpoliticalinformation.Eitherway,theact
ofseeingpoliticalinformationontheInternetisassociatedinaverysmall
48
way with voting. The magnitude of this influence is tiny, even though
statisticallysignificant.Inthevotingmodel,whichissimilartotheothers,
the Internet variable improves overall predictive power by a little under
1 percent. Among 726 people included in the analysis, data on whether
each had seen political information on the Internet improves the num-
ber of correct predictions in the model about who reported voting for
49
just eleven individuals. (See Table 5.6.) This hardly counts as a revolu-
tion in political engagement, but it is just large enough to be difficult to
dismiss entirely without further evidence from future elections.
In interpreting this finding, it is important to note the NES data suffer
from several limitations. For one, the Internet measure is dichotomous.
By measuring only whether people obtained any information from the
Internet, the survey failed to capture useful information about frequency
or intensity of use. Second, the data lump together all forms of Internet
use, so it is impossible to differentiate among the huge number of ways
that people use the Internet politically: to read news from traditional
sources, to read alternative sources of political information, to watch
video of candidates, to communicate with others, to receive messages and
solicitations of various kinds, and so on. The problems with this tech-
nique are obvious in data from the 2000 General Social Survey (GSS),
which was fielded in the spring. This study was the first major schol-
arly survey to recognize that different forms of “Internet use” might be
conceptually distinct. The designers of the Internet module in the GSS
employed about a hundred Internet variables. These differentiate activ-
ities such as number of hours per week spent using a home computer
for commerce from hours spent using a work computer for electronic
mail. The GSS data show that the bivariate correlation between the two
major categories of Internet use – time spent using electronic mail and

48
Also since 1996. The author’s analysis of 1996 data was consistent with the 1998 results
here. See Bimber, “Information and Political Engagement in America: The Search for
Effects of Information Technology at the Individual Level.”
49 2
The Nagelkerke r value goes from 0.399 without the Internet variable to 0.404 with
it, and the chi-square value from 241 to 245.
223