13
Table 6 about here.
Table 6 contains the logit coefficients of regression analysis where the DV satisfaction
with democracy is coded as 1 if the respondent indicated they were very or fairly satisfied
with democracy, and 0 if they were not very or not at all satisfied with democracy. The
results contain three models. Model 1 tests the impact of winning and ideological
distance on levels of satisfaction. The model distinguishes between those who backed
local winners and national losers, those who backed national winners but in
constituencies where they did not gain seats, and those who voted for the national winner
in a constituency where it gained a seat. By eliminating total losers this precludes the
possibility of collinearity in the model. Model 1 also includes the ideological distance
from the national winner and from the local winner. Model 2 includes political predictors
found in prior research to be determinants of satisfaction. These include political
knowledge and partisan identification for two of the opposition parties. This will help to
determine whether those who backed more successful parties were more satisfied with
democracy. Last, model 3, the fully-specified model, includes standard socio-
demographic information cited in previous literature as predictors of satisfaction,
specifically gender, age, education and income. The results show that different predictors
drive satisfaction with democracy in each of the three case studies. In all three cases
model 1 does not account for much of the variation in the dependent variable. The Cox
and Snell, and the Nagelkerke R
2
were both below .1 for the first model of all three cases,
never exceeding .2 for the fully-specified model.
Immediately relevant is the variation in results across the three case studies. In Australia,
backing a national winner does not appear to drive satisfaction with democracy whereas
this feature is more relevant in Canada and Great Britain. In both of these cases backing
the eventual government in a constituency where it gained a seat prompted a significant
and positive increase in satisfaction with democracy, an effect that remains in the fully-
specified model. Backing a local winner only, however, does much less to improve
satisfaction. The effect is not significant in Great Britain, and significant only in Model 1
Convention