The Timeline of Elections: A Comparative Perspective is now available on Early View, and the authors summarize the article here:
How do voters’ preferences come into focus in between elections? Do they evolve in a patterned way? Does the structure and evolution of preferences vary across countries? In the paper we consider differences in political institutions. We examine government institutions, especially differences between presidential and parliamentary systems, and electoral institutions, such as differences between proportional representation and single-member district plurality systems.
Polls and the Vote over the Election Timeline in 45 Countries
For this research, we amassed over 26,000 vote intention polls in 45 countries since 1942. This data covers 312 discrete electoral cycles in total. Our analysis relates Election Day vote shares to the support for parties or candidates in the pre-election polls day-by-day over the election timeline. Figure 1 below illustrates how the vote and the polls line up at various points during the election cycle.
In the upper left-hand panel, using all polls available 900 days before the election, there already is a discernible correspondence, though there also is a good amount of variation. As we turn to polls later in the election cycle, moving horizontally and then vertically through the figure, a clearer pattern emerges; the poll share and final vote share line up. It is not highly surprising, but it is as one would expect if voters’ preferences change and a nontrivial portion lasts. But how much do preferences evolve? And, how do political institutions matter?
To provide a definitive assessment, we estimate a series of equations predicting the vote share for parties or candidates in different elections across countries from vote intentions in the polls on each day of the timeline. These allow us to show whether and how preferences come into focus over time.
Our analyses reveal considerable differences between presidential and parliamentary elections. This can be seen in Figure 2, which plots the root mean squared errors (RMSEs) from each of the daily vote-poll regressions over the final 200 days of the election cycle. At the beginning, polls are much more informative in parliamentary elections, with an RMSE of 4 percentage points by comparison with 6 points for presidential elections. The difference is statistically significant; as can be seen, the confidence intervals do not overlap. The gap narrows over time, with preferences for presidential elections coming increasingly into focus, and the two are virtually indistinguishable on Election Day. There thus are important differences in the structure and evolution of preferences in presidential and parliamentary elections. Other analysis shows little difference in the pattern for legislative elections in presidential and parliamentary systems, however. The main difference relating to government institutions is between presidential elections on the one hand and legislative elections on the other.
We find less evidence of the influence of proportional and plurality systems on how electoral preferences evolve over the timeline. Figure 3 plots the RMSEs for regressions relating the polls and the legislative vote in proportional and single member district plurality systems. This suggests that polls are slightly more predictive in proportional systems. The differences are very small, however, and only intermittently statistically significant. If nothing else, the results indicate that preferences do not come into focus earlier and more completely in proportional systems, though additional analysis does suggest that party-centric electoral rules that are often associated with PR systems do matter.
Conclusion and Implications
Voter preferences evolve in a systematic way over the election timeline in a wide range of representative democracies. But the pattern is not precisely the same in all countries. Political institutions structure the evolution of voters’ preferences, though especially the difference between presidential and legislative elections. In this paper we only scratch the surface of the variation in context, however. Political institutions differ in ways that we have not considered. There are other differences in context that we have not even begun to explore. Some of the differences relate to countries themselves. Other differences relate not to political institutions or the countries themselves, but to characteristics of political parties. Clearly, much research remains to be done, and our methodology can help guide the way.