Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I even say who wrote them. If you have more recent summaries to add to this collection, send them my way I guess. Sorry for the ads; they cover the costs of keeping this online.
Green and Shachar. 2000. Habit formation and political behaviour: Evidence of consuetude in voter turnout. British Journal of Political Science 30: 561-573.
Voting in one election makes you more likely to vote in future elections; voting appears to be habit-forming. Even after controlling for everything else, voting breeds more voting and abstention breeds more abstention. See Gerber, Green, and Shachar (2003) for an updated--and much more persuasive--version of this argument.
"Consuetude" is simply a pretentious word for "habit."
Previous studies have controlled for as many variables as they knew to, but still found that controlling for whether a subject voted in the previous elections correlates with whether they voted in the recent elections. Thus, these previous studies have concluded that voting is habit-forming.
The problem, of course, is that these models might simply be underspecified. And if they are, then the "habit" variable is simply picking up other phenomena that went into both the previous and the current voting decision.
The authors attempt to get around this problem by using time series panel data (ANES) from 1972-1974 in an instrumental regression. First, they use 1972 controls to predict 1972 turnout. Then, they use 1974 controls to predict 1974 turnout. Then, they use all these variables as instruments in a final regression, which uses 1976 controls to predict 1976 turnout. Despite concluding a whole bunch of controls, they still find that previous turnout correlates with future turnout.
The authors also use an "experimental" design, though the experiment doesn't manipulate the variable of interest (voting in a previous election). Instead, the experiment manipulates a variable known to correlate with turnout: Whether you were contacted by pollsters in the weeks before the election. They are using data from an experiment conducted in the 1970s to determine whether contact from a pollster boosts turnout (it does), but they are taking the analysis further: They use this "treatment" as an instrument to control for prior turnout, then conclude that prior turnout nonetheless correlates with current turnout.
Recognizing that their data may not be persuasive, they nonetheless evaluate four possible explanations for these findings. They like the third.
The authors suffer from the same methodological challenge as earlier studies did. Though they may have reduced the problem, it is still perfectly possible that some uncontrolled variable is being picked up in the 1972 and 1974 turnout variables, which therefore correlate with 1976 turnout. In fact, their theoretical discussion (below) suggests this very interpretation. Moreover, the later article by Gerber, Green, and Shachar all but concedes this criticism; p 541.
Research by the same authors
Research on similar subjects