-
13. Correlates of Online Political Information in Seven Democracies
- Johns Hopkins University Press
- Chapter
- Additional Information
H aving considered how socioeconomic status, political culture, political preferences, political engagement, and mass media use correlate with voters’ propensity to acquire political information through the internet, we can complete the analysis by comparing the results of multivariate logistic regression models across the seven countries included in this study, highlighting the similarities and differences among these political systems. Table 13.1 shows the results of a qualitative meta-analysis that, for each independent variable included in the regression models, reports the direction of each country’s coefficient, thus showing whether the correlation was positive or negative, and indicates whether it was statistically significant at the conventional threshold of p ≤ .05.1 For ease of comparison, in countries where surveys for two general elections were available, I included the most recent election (Italy 2008 and Australia 2010). Readers who are interested in examining the individual coefficients can refer to the appendix (tables A.4 to A.10). Before presenting the results of this analysis, three caveats must be offered regarding the validity of the relationships that emerged from the data. CHAPTER THIRTEEN Correlates of Online Political Information in Seven Democracies Table 13.1. Meta-analysis of the Findings of Multivariate Regression Models Predicting Online Political Information in Each Country’s Last General Election Australia France Germany Italy Spain United Kingdom United States Gender (male) + + + (+) + + (−) Race (white) // (−) // // // (+) (+) Age − − − − − − − Education + + + + + + + Urban density + + (−) (+) + // (+) Income (+) (+) (+) // (+) + + Religious attendance − (−) (−) (+) (+) (+) (+) Interest in politics + (+) (+) + + + + Sense of efficacy + − + (+) + (+) + Political trust (+) (−) // (+) − + (−) Ideology None None None Left > right None // Lib > Cons Party identification None None + Other parties + Other parties + PSOE, No party + LibDems None Political discussion + (−) // // + + + Participation in rallies (−) // + (+) (+) + + Working for parties/candidates + // (−) // + (−) (+) Reading political news (+) // + (+) + (+) (−) Watching political news (+) (+) + (+) (+) + (+) Note: + indicates a statistically significant (p ≤ .05) positive relationship, (+) indicates a statistically not significant positive relationship, − indicates a statistically significant negative relationship, (−) indicates a statistically not significant negative relationship, and // indicates that the variable was not included in the model. [44.192.107.255] Project MUSE (2024-03-28 23:33 GMT) 192 Citizens and Digital Politics The first is that, as is well known, cross-sectional survey data such as those employed here can highlight correlations between variables but cannot fully establish causation. Doing so would require (at least) longitudinal data that are currently not available in comparative research on digital politics. Even if modeling use of the internet for political information as a dependent rather than independent variable leads to plausible arguments (see chapter 4), the causal patterns of influence suggested in this and the previous chapters should be treated as hypothetical rather than definitive. The second caveat is that the relationships that have been found here may not necessarily be stable over time. As shown by Bimber and Copeland (2013) with US data and by Bimber et al. (2012) with UK data, relationships between internet use and various forms of political participation are inconsistent and idiosyncratic because they are contingent on context and timing. Although Bimber and colleagues tested the inverse causal relationship (from internet use to political engagement), there is no reason to rule out that similar fluctuations may occur in the causal patterns tested here. It is my hope that the analyses and interpretations presented in this book will help us move toward better specified theories that allow us to explain some of the likely variations through time in the relationships between online and offline political engagement by taking into account contextual changes, to the extent that these have systematic effects. Shedding light on these issues, however, requires greater and sustained scholarly efforts, not least because a certain number of electoral campaigns need to take place before one can test the effects of different electoral contexts. Finally, the third caveat is that different patterns might emerge from the same causal models if other dependent variables measuring different aspects of online engagement were to be considered. As I argued in chapter 2, acquiring political information through the internet can be seen as the first step on a ladder of web-based political engagement that comprises multiple activities and endeavors. In this study, I have not moved beyond this first step, not least because of the lack of reliable measures that could travel across the seven countries analyzed here. Future research, however, should ask how different facets of online political engagement are shaped by systemic and individual factors in comparative perspective. Because the previous chapters have already discussed the results for the most relevant independent variables included...