The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
In a recent post, co-blogger Eugene Kontorovich criticizes Israel's law restricting the publication of public opinion polls in the last few days before an election. I largely agree with Eugene's critique. The Israeli law (and similar legislation in many other countries) is an unjust infringement on freedom of speech.
But Eugene does not address an important possible defense of these types of laws. Late preelection polls could exacerbate the pernicious "bandwagon effect," which leads some voters to support a given candidate or party merely because it seems likely to win. I discussed bandwagon voting in this 2012 post:
A small but significant number of swing voters tend to support whichever side seems to be winning, partly because they want to be identified with a winner and partly because of a sense that whoever seems to be winning might well be the best person for the job for that very reason. Bandwagon voters are unlikely to make a decisive difference in an election where one side has an overwhelming edge to begin with. But they can be decisive in a closer race. They can also increase the winner's margin of victory, thereby adding to the perceived extent of his "mandate." For these reasons, candidates and their supporters routinely project greater optimism than they really feel.
The bandwagon effect is an inversion of the normative ideal of democracy. Instead of choosing the winner based on their perception of what would best serve the public interest, bandwagon voters modify their perception of the public interest based on who they think is likely to win. Worse, these voters are often among the key swing voters who decide electoral outcomes.
Well-informed voters and those with strong views on political issues are unlikely to change their minds because of the bandwagon effect. But political ignorance is widespread, and swing voters (the ones most likely to change their intentions at the last minute) are, on average, considerably more ignorant that those with stronger partisan commitments.
Restricting publication of last-minute polling results could potentially prevent relatively ignorant swing voters from deciding who to support for based on bandwagon effects, and thereby lead them to consider more substantive reasons for choosing one party over the others. In Chapter 2 of my book on political ignorance, I discuss some situations where voter ignorance might actually have beneficial results. Perhaps ignorance of late preelection polling results might be another such case.
But to be really effective, such laws would probably have to ban publication of polling results for many weeks prior to the election, not just the last few days. Polls often create a strong impression of who the likely winner weeks or even months before election day. That approach, of course, would restrict freedom of speech far more than current Israeli law does.
Even if polling publication restrictions could diminish bandwagon voting, it is not clear that the voters diverted from bandwagoning would choose better-informed reasons for voting. They might instead rely on a variety of other dubious and often misleading heuristics and information shortcuts. Sadly, the bandwagon effect is just one of many negative consequences of widespread political ignorance. We are unlikely to solve the problem by giving government more power to restrict the flow of supposedly harmful information to the public.
Moreover, as Eugene points out, late polls can sometimes provide valuable information to better-informed voters. In a multi-party system like Israel's, the decision to vote for a particular party reasonably depends not just on the party's own merits, but on the potential impact of an increase in that party's support for the configuration of a coalition government. And that effect, in turn, often depends on the extent of support for other parties. For example, a given Israeli voter might be willing to support Party X if giving that party an extra seat in the Knesset is likely to lead to a center-left coalition government, but not if it is more likely to lead to a center-right coalition. Admittedly, only unusually well-informed voters are likely to make such careful calculations effectively. But such people can sometimes make a difference in a close election. And if the election is not close, then there is also less need to worry about harmful effects of bandwagon voting.
Of course, one can try to justify restricting preelection polling not on the grounds that it prevents bandwagon effects, but because some polling results are biased or even deliberately manipulated to support one party or candidate. But the same can be said for a wide range of other political information and preelection commentary by pundits, political activists, and the media. If the danger of bias and manipulation justifies censoring publication of preelection polling results, it can easily justify censorship of most other types of political speech during election season, as well.
In sum, I have a measure of sympathy for concerns about bandwagon voting that underpin rationales for restricting the publication of preelection polls. But the supposed cure is unlikely to genuinely alleviate the disease, and might sometimes have negative side-effects.