On the occasion of the results of the first round of the presidential elections in Cyprus elections again started an extremely ungrateful and meaningless discussion about reliability polls.

Let’s start with the data. Polls for the Cypriot elections did not fall out. In Cyprus, there is a ban on the publication of public opinion polls a week before the elections. Thus, the latest published surveys did not cover all those who made a decision last week, and this is a considerable percentage.

Of the six most recent polls published before the ban, three gave Mr. Mavroyiannis second place and three gave Mr. Neofit, all with very small differences within the margin of error. However, without exception, the time series reflected a steady upward trajectory for Mr. Mavroyiannis, essentially making him the favorite to advance to the second round. This general trend was also confirmed in the last phase, although it was not recorded publicly due to the lack of publications of studies last week. No honest observer can consider this a failure.

Some of the assumptions associated with the surveys – of course, about serious companies in this area – now seem to be established. Let’s take a look at three of them.

First, poll predictions are usually confirmed. At least in terms of the ranking order of candidate parties and the statistical range of votes. Of course, there are cases of failure, but there are not so many of them to justify complete doubt about their reliability. They remain the most reliable tool for analyzing social trends and certainly more reliable than the social media “digital bubble”.

Secondly, The effect of public opinion polls is manifested in more than one direction. Over time, as a result of measurements of polls, various phenomena of electoral behavior were recorded.

The most common occurrence is the footboard effect. That is, a further strengthening of what precedes it. However, this is by no means a one-way street. For example, in the last US presidential election, Trump appeared to have lost in 10 of the 13 swing states. He ended up winning six of those 10, and three of them (which ultimately determined the outcome of the election) he lost by less than 1%.

The messenger is not to blame for the message. Both mythologization and demonization of the political analysis of the messenger cloud.

There is also the backlash effect, which describes the reaction of a subset of voters to the mood created by the polls.

A classic example is the June 2012 elections, when most of the voters of the parties of the then “intermediate space” switched to the ND at the last moment. (whose percentage grew by 11% in 40 days) with the main motivation “not to beat SYRIZA”, which gained impressive momentum after the snap elections in May 2012. ND win. in that election it was a backlash effect, not a bandwagon effect.

While there is also the underdog effect (loosely translated “trend in favor of outsiders”), which describes the voter phenomenon by looking at broader correlations, choosing a smaller candidate party to send a message or influence the outcome of an election. by ordinary voting.

The 2002 French presidential election is a classic example of this phenomenon. All the polls looked like the then center-right President Chirac and the center-left candidate Lionel Jospin made it to the second round. This resulted in many voters voting for smaller candidates in the first round, either to support them or to send a (low-cost) message of discontent. As a result, Chirac advanced to the second round with the lowest percentage ever received by an incumbent president, while Jospin – due to too weak a vote from left-wing voters who voted for Trotskyists, environmentalists, etc. – almost lost second place Jean – Marie Le Pen.

Third, surveys themselves, as an analytical tool, are developing. At one time, surveys were conducted only by landline phones. Now they are also carried out through mobile phones. They are often combined with online surveys to complete the sample and identify behavioral differences. The search for how surveys can continually improve by combining traditional methods with big data, more targeted sampling, smart weightings, etc. continues internationally. But this is mostly of concern to professionals in the field, as many politicians and journalists are simply trying to bring the debate to their own electoral needs. But they forget one simple thing: that the messenger is not to blame for the message. Both the mythologization and the demonization of the messenger are detrimental to substantive political analysis.

Mr. Eftichis Vardoulakis is a strategy and communications consultant.