Keywords: Tunisia, pre election polls, vote intention, expectation, accuracy

Title:
What are the factors that lead to better accuracy of pre-election polls in the Tunisian context?
Analysis of comparative experiences between the 2019 and 2024 elections

In Tunisia, since 2011, we have been faced with several electoral exercises.
It must be said that from that date to this day we have been able to conduct almost all kinds of researches and surveys on the subject: political barometer, pre-election surveys, electoral simulations, exit polls, etc.
If, moreover, we were very precise in our voting estimates during the exit polls, it must be recognized that we were less so during the pre-election polls, which were carried out a few weeks, or even two or three months before election day.
Indeed, for this type of survey, and in order to predict the winners of the elections, both legislative and presidential, our questions were mainly formulated in the form of voting intentions: if the elections were to take place tomorrow, who would you vote for?
Why is it so difficult to accurately predict the results of the elections in a country like Tunisia? What makes this exercise complex?
To understand this phenomenon, we conducted several research studies to analyze the causes and factors that contribute to this gap (between the forecasts and the final results of the elections).
We must not forget that Tunisia is a relatively "young" country in terms of holding democratic and transparent elections. We do not have, moreover, a sufficiently consistent history in terms of survey data or other research conducted on this subject.
On the other hand, it has been shown, according to several research studies conducted in the United States and Europe, that asking the question on expectations such as: who do you think will win the next elections? Rather than on voting intentions would give better precision.
We have, in fact, examined this approach during the 2019 presidential elections in Tunisia, but the results did not confirm this thesis. On the contrary, the results of the polls asking the question on voting intentions were closer to reality.
After repeating the exercise during the last presidential elections in Tunisia (2024), we were able to confirm that, ultimately, the question on expectations gave better precision than the one relating to voting intention.
What changed the observation?
The comparative analysis of the pre-election surveys conducted in 2019 and 2024 respectively was able to demonstrate the existence of a multitude of major factors that could have changed the situation.
This is what we will try to demonstrate following a series of research projects during these two electoral exercises.