The Problems of Opinion Polls, by Berta Barbet

Sunday’s results leave a scenario far from the clear victory of the People’s Party over the Socialist Workers’ Party indicated by many opinion polls. Less than it sounds, but far enough to open up one of the most classic debates of the post-election days: Why do polls fail?

Much has been said about the problems with surveys of profiles abstaining the most or voting for options that are socially frowned upon. A problem that clearly shows part of the error. Also from the need to understand the margins of error involved in predicting the distribution of seats for the 52 constituencies. But we often forget one of the main problems with surveys to hit election resultIt is difficult to measure what citizens will vote if citizens do not know.

It is difficult for a citizen to inform us well of a decision he did not make

For a long time now, political scientists have wondered how polls could differ so much if the factors that explain voting were, for the most part, stable and predictable months before an election. Elements that determine voting such as the economic management of the government, the groups polled by candidates or the popularity of representatives and decisions made throughout the legislature tend to change little during the 15 days of the campaign. So why are polls moving so far and voting so unpredictable?

The answer that political scientists Gilman and King came up with in 1993 is that campaigns are a period in which many voters learn and reflect on what happened during the legislature and incorporate it into the decision to vote. Motions at the polls capture this learning process of what the candidates and the groups they questioned did as it happened. A non-linear process because campaign messages do not always reach the same density or the same voters. Polls move because voters take time to gather and integrate the information the campaign gives them to make a decision. It’s so time consuming that an increasingly large percentage do it only hours before going to the polls, making it very difficult to predict from questions about voting intentions. It is difficult for a citizen to tell us well about a decision he did not make.

This does not mean that surveys are not reliable or useful. Surveys show changes resulting from information that citizens have already learned and managed. Nobody expected PP results similar to those in 2019, because space citizens had already included changes in space (disappearing Cs or regional and local PP-Vox conventions) in their decision and could thus inform the poll organizers. But this means that questions about voting intentions hardly give a stable and reliable picture of what will happen on their own. It is necessary to understand what has changed on the stage, what the voters have already entered, and what has not. If we do not correlate opinion polls with basic political movements, our reading will easily make us unable to anticipate the decision of the entire undecided people.