" It’s a potential voter pool of more than one million citizens,” states Kais Ben Halima, a leading member of Hezb El Warka (Rolling Paper Party), a collective advocating for the decriminalization and commercialization of cannabis. The collective claims that multiple candidates have attempted to gain their support in order to access their electoral base.

Privately commissioned surveys could help explain why this advocacy group is receiving positive feedback from numerous candidates. Some no longer hesitate to pledge their support for the campaign. During the televised debate on September 9, 2019, Mehdi Jomaa, Abdelkarim Zbidi, and Said Aidi openly considered the decriminalization of cannabis and agreed to respond to questions from Hezb el Warka on the subject.

Thanks to polling institutes, these candidates know that such a public claim could help them gain a new segment of the electoral pool. Throughout the campaign period, pollsters are forbidden from releasing any survey results; nonetheless, requests for surveys continue to increase. Those who have the means do not hesitate to spend tens of thousands of dinars in order to better inform their communication strategies.

Survey results allow candidates to position themselves in relation to their competitors but also to get a glimpse of constituents’ opinions on a variety of pressing social issues. The cost of a single survey varies depending on the sample size and selection methodology. On average, a survey costs between 10,000 and 50,000 dinars. 

 INSIDE THE OPINION POLL MARKET

Downtown Tunis. A two-story villa on a small street in Cité Olympique houses the headquarters of Elka Consulting. On the top floor, Ikbel Elloumi examines documents at his desk. He chain smokes while waiting for his next meeting.

For several months now, the CEO of this polling firm has been focused on the elections.  With the launch of the presidential race, demands on his team are more pressing; polls have been conducted weekly. Elloumi is working with five different candidates.

Through these surveys, Elka Consulting has provided precious guidance and rankings for candidates. Ikbel Ellloumi lists off the kinds of questions their surveys look into. “ It’s not just about intention to vote. There are also specific questions so that the candidate knows what their public image is, their strong points, their weak points, and what people take away from them.”

More generally speaking, these surveys are carried out in order to know citizens’ priorities or to better understand their opinions on candidate platforms. “ For me, the question ‘Who do you plan on voting for?’ is very much secondary,” says the CEO.

When a campaign team requests a survey, the firm works with them to define the questionnaire and eventually analyze the results and even suggest improvements.

“ For instance, on the question of equal inheritance rights for women, nearly 75% of the population is against it. It’s a very polarizing topic, where a candidate can lose a large part of their electorate.” In this case, if a candidate supports it, “ they don’t change their position, but they try not to talk about it.”

Once the data is collected, the firm analyzes it and develops recommendations for communication strategies. “ We cannot conduct a data analysis without also giving a consultation,” Ikbel Elloumi insists.  

From just one polling institute, Mehdi Jomaa’s campaign team has already requested four studies. " For us, the confidence indicator is more important than a candidate's rank," explains Jomaa’s campaign manager, Mohamed Ali Toumi, especially “ when there is a majority of undecided voters,” he specifies. He believes that percentage rankings do not necessarily provide a reliable prediction of polling day results, given the number of unknown variables and the margin of error. 

“ We mainly look to understand the candidate’s popularity in specific regions to select locations for mass assemblies,” explains the campaign director. “ I’m not saying it’s a calculation, but you have to make some choices. The candidate can’t go everywhere.” 

Surveys can also impact the discourse of a presidential candidate. “ For example, knowing one’s popularity among retirees tell us what kind of discourse to take on. If it’s weak, we know what corrections to make.” 

 But Mohamed Ali Toumi believes that surveys are just a way to choose or “ correct” a strategy. “It’s what happens on the ground that makes the difference,” he asserts. 

A MARGIN OF ERROR THAT CAN CHANGE IT CALL

To carry out these surveys, polling firms and campaign teams define a sample of the population based on figures from the National Institute of Statistics. Generally, the sample size is at least 1,000 people, explains Ikbel Elloumi. The aim is to establish a representative sample of individuals - mixed by gender, age, place of residence, level of education, etc. - that reflects the makeup of the country on a national level.

The individuals to be surveyed are then randomly selected. Elka Consulting only does door-to-door surveying, while Emrhod Consulting also conducts telephone polling as a cheaper and faster alternative. 

Depending on the method, the margin of error can vary. “ For a sample size of 1,200 people, we have a 3% margin of error,” Ikbel Elloumi explains, which can mean considerable fluctuations in the forecasted results. 

For instance, if two candidates are within 6 points of each other in the polls, a 3% margin of error makes it impossible to confidently ascertain their respective standings in the race.

One way to reduce this margin of error is to increase the sample size. This is the approach that Hatem Boulabiar took in 2014, while he was managing all surveys for the Ennahdha Party. Many weeks before the legislative elections, he organized a survey with a sample size of 22,000 people - costing the party 200,000 dinars. According to him, the margin of error was only 0.5%. “ I had predicted Ennahdha’s defeat, but no one [at the time] wanted to believe it,” he recalls.

Ikbel Elloumi also mentions that certain variables are not factored into the final results but are nonetheless important for the campaign teams to consider. For example, a considerable number of respondents do not wish to reveal for whom they intend to vote. “ Right now, this group represents 10% [of the surveyed].”

This “hidden constituency” can lead survey results away from reflecting the reality of voter intentions. 

This limitation was observed in 2011 with the party lists led by Hechmi El Hamdi, who is currently a presidential candidate. No poll at the time had predicted the arrival of this massive political alliance. In order to calculate more precise forecasting, polling institutes need to draw on surveys from previous elections.

But " the political scene is not stable enough" in Tunisia, says Ikbel Elloumi. " This is what happened in 2014 with [one of the candidates]. It’s called a ‘shame vote.’ And now I expect the same [phenomenon] to happen with another candidate.” 

ALTERNATIVE DATA GATHERING 

In a posh café in Tunis’ Berges du Lac neighborhood, Hatem Boulabiar sits at ease. He smiles at the café’s clients and calls out to the waiter to bring him some olive oil to drink straight. “ I’ve been talking all day and am losing my voice. I need to clear up my throat a bit,” he says, laughing. 

The candidate is confident, pleased with his performance the day before during the presidential debate. After several years alongside Ennahdha, Boulabiar decided to run as an independent.

The man who oversaw Ennahdha’s opinion polls and meticulously analyzed their results can no longer afford such services.  But thanks to his network, the candidate is still able to get some insights from polling institutes. “ I’ve asked some of my friends who run polling institutes to sneak in a couple of questions about me when they conduct their clients’ surveys,” Boulabiar admits. This way, he can at least get an idea of his popularity rating.  

Boulabiar isn’t the only one using personal contacts to get surveys. Abdelkarim Zbidi and Elyes Fakhfakh – among others – also tap into their networks to get results from polling institutes. According to Khalil Ben Cherif, a member of Abedelkarim Zbidi’s communications team, some candidates receive survey results from their social networks “ out of goodwill.” He clarifies that they don’t receive the finer details of the polls, just the rankings of voting intentions.

“ In any case, polls are not how you win elections, they’re just an indicator to reassure yourself,” states Maher Laroui from Fakhfakh’s campaign team. 

Though they know the sample is not representative, Abdelkarim Zbidi’s communications team analyzes the reactions and comments on their social media accounts themselves. “ In terms of the audience, there is plenty of buzz, but does that translate into a vote?” asks Khalil Ben Cherif. 

To better understand reactions on social networks, Mohamed Abbou’s team has opted instead to use “WebRadar,” a content analysis tool for public, online information. For 1,500 dinars a month, WebRadar provides four weekly reports that indicate the most followed publications, a  comparison with other candidates, and internet users’ different reactions according to their “feelings” (positive, negative, etc.). 

But these results aren’t enough for Sami Habita, the digital manager of Abbou’s campaign team. “ We cross-check this information alongside other data that we collect ourselves,” he elaborates. It’s only once all this information is compiled that the candidate’s team issues their own conclusions. 

In parallel to private polls and other alternatives, numerous rumors and fake opinion polls circulate on social media, favoring one candidate or another, with the intention to manipulate public opinion. On September 11, 2019, a poll copying a template of Sigma Conseil (one of the main Tunisian polling institutes) was released. Hassen Zargouni, director general of the institute, quickly denied the veracity of this study. " Another candidate and his henchmen show how vile the procedures are for getting elected," he says with disappointment.