In other words, our models do not use the difference between pre-election and post-election system support as the dependent variable.
During the pre-election interview respondents were asked if they would be willing to participate in the post election survey.
For 1996, we used the general government trust pre-election measure to control for pre-election trust.
Evidence of political manipulation would be a departure from the trends in an expansionary direction during the relevant pre-election periods.
Adjustments to the minimum pension were not so concentrated in pre-election months, in part because the minimum pension was adjusted far more often.
This hypothesis predicts that the pre-election surge in personal income growth is a monotonic decreasing function of the incumbent government's ex ante re-election prospects.
Thirdly, the strategic environment in the pre-election year is different from that of the election year.
Rather, change is modelled by controlling for pre-election attitudes in the model predicting post-election attitudes.
