One year ago I wrote about Bamako’s “lone pollster,” an independent, unaffiliated Malian applied economic statistician doing his own surveys of public opinion in Mali’s capital city. Mr. Sidiki Guindo is still at it, and has just come out with results of a poll asking 2600 residents of four cities in central and southern Mali (Bamako, Sikasso, Segou and Koutiala) about their preferences for scheduled presidential elections. Neither he nor I can claim that the results are representative of public opinion nation-wide; they offer a sometimes murky glimpse of what certain segments of the Malian population are thinking. As I wrote last month, it’s unclear whether the scheduled timetable for July elections can be maintained. As we wait for Mali’s electoral process to take shape, let’s consider Mr. Guindo’s preliminary findings. My slightly abridged translation appears below (click here for the original French version).

Sidiki Guindo

After more than a year of unprecedented crisis, Mali and its allies want to organize presidential elections in July 2013. This presidential election is singular for many reasons: first, it comes on the heels of (if not amidst) a very bitter crisis, the blame for which is widely shared; second, it also comes at a time when the most senior candidates seem to be giving it their last shot. Therefore, Malians and political parties are more determined than ever to choose a president. In light of this situation, every Malian has the responsibility to work for the total success of these elections. For our part, as a statistical economist of this country, we plan to carry out a series of opinion surveys to predict the results of this electoral contest. These surveys are motivated by three factors:

1. Imitating other countries: Almost everywhere else, polling methods play a major role in predicting the results of various elections. There is no reason why Mali should be excluded from this trend.
2. Motivating our leaders to pay more attention to statistics: While political parties spend millions on their electoral campaigns, none of them seem to be taking stock of their strengths and weaknesses, neither during nor after the election. Today, even the government seems to lack the culture of quantifying things before acting. In Mali, the significance ascribed to statisticians and to statistics leaves much to be desired.
3. Taking on an interesting challenge for a young statistician: For an Ingénieur Statisticien Économiste (ISE) [applied economic statistician], predicting the results of this election by statistical methods is a good problem to take up.

Who commissioned this survey? This survey is neutral with respect to political parties, current and former governments, juntas and international organizations. It is a scientific exercise aimed at helping our country get out of its crisis. To this end, detailed reports of the results will be made available online in the coming weeks.

The questions asked: We sought answers for an array of questions such as, What are Malians’ greatest concerns? What does the population think about organizing elections before liberating Kidal? How would each political party fare in the first round of the vote? Will there be a second round? In the event of a second round, how would votes be divided? Who will be the next president of Mali? What image do these different leaders have with the people?

Methodology: The poll was carried out from 10-12 May 2013 with a sample of 2600 people aged 18 and older. This first stage of our polling was limited to four cities: Bamako, Segou, Sikasso and Koutiala. We chose these cities because of their importance in choosing the next president. They act as a testing site for the next stages. The results obtained in these cities might be a  predictor of national results for the large parties. The size of these cities is significant and the results will be related to the overall country results. Nonetheless, a representative survey at the national level must give more precise results (and will undoubtedly be undertaken before the vote).

The theoretical conception of this study respects all the theories for such a poll. Different tools were used to analyze the data and we checked the relevance of each tool before applying it. The survey utilized the quota sampling method (the most widely used in opinion polls around the world). The quota variables used were sex and age. Education level is taken into account during analysis. We surveyed 2600 people divided among the four cities (860 in Bamako, 710 in Sikasso, 650 in Segou, 380 in Koutiala).

Our sample size respects the set of hypotheses to verify in parameter estimation, corresponding to a 2% margin of error. We preferred, however, to provide upper and lower estimates of the scores (confidence intervals), taking account of certain realities on the ground. These details are very important in judging the quality of the results. We invite our statistical colleagues and research firms seeking to predict the election results to provide details on the size and distribution of their samples, the area of study and type of survey used. Without these details, poll results are only worth so much.

RESULTS

Are the people motivated to vote?

To the question “Are you ready to vote in presidential elections?”, 81.95% of respondents said they were ready to vote; this was the proportion of people registered under RAVEC [the Recensement Administratif à Vocation d’Etat Civil, a national ID and voting registration campaign conducted in 2010-2011], and having the intention of voting on election day. We therefore can hope for a relatively high level of voter turnout. We must however note that there may be a significant divide between these stated intentions and actual behavior. Among those who said they did not intend to vote, 44.6% think there is no trustworthy candidate, and 14.3% say the election will be rigged in advance.

Should we hold elections before liberating Kidal?

We asked, “If we are not able to liberate Kidal before the date set for elections, do you think election day should be postponed, or should the vote be held in the rest of Mali without Kidal?” To this question, the vast majority of respondents, 82.8%, preferred postponing elections if Kidal is not liberated beforehand.

Candidates and parties

To the question, “If elections are held next month, for whom would you vote?” we found different results in different cities (even if one trend was visible throughout the area of study). We also noticed that certain confidence intervals overlap (in which case, our conclusion is based on the average).

In Bamako, voters will turn out for two candidates: Ibrahim Boubacar Keïta of the RPM and Soumaïla Cissé of the URD. Keita is in the lead with a score between 35.3% and 42.5%. He is followed by Cissé with a score between 10.8% and 15.8% (with an average of 13.3%). These two candidates will garner at least 46% of the votes in Bamako. Third place is contested between three candidates: Moussa Mara of the Yelema party (10.2%), Dramane Dembelé of ADEMA (9.2%) and Modibo Sidibé of FARE (9.2%).

In Segou, Keïta leads with a score between 39.6% and 48.4%. Dembelé of ADEMA is in second place with an average score of 11.2%. Third place is a toss-up between Soumaïla Cissé (8.6%) and Moussa Mara (8%). The confidence intervals overlap between these two. Mountaga Tall of CNID garnered 5.5% of the vote in Segou.

In Sikasso,one of the key cities in these elections, there is a wide gap between the top two candidates and the others. Keïta emerges with a score between 37.4% and 45.5%, followed by Cissé with a score between 15.2% and 22.6%. Third place is too close to call between ADEMA’s Dembelé (9.7%) and Housseini Amion Guindo of CODEM (9.0%), given overlapping confidence intervals.

In Koutiala, Keïta is again in the lead with a score between 26.0% and 35.9%. He is followed by Oumar Mariko of SADI with between 18.5% and 27.5% (averaging 23.01%). Dembelé of ADEMA and Cissé of URD garnered 10.9% each.

In the entire survey area (all four cities), one trend is noticeable: Ibrahim Boubacar Keïta comes in first place (35.7%-42.1%), with Soumaila Cissé in second (10.4%-15.4%).Third place is held by Dramane Dembelé of ADEMA (6.1%-11.7%).

Average tallies across all four cities

The case of Modibo Sidibé: During the design of this survey, before collecting our data, we had the hypothesis that four candidates would poll noticeably above the others: Dramane Dembélé of ADEMA, Soumaïla Cissé of URD, Ibrahim Boubacar Keïta of RPM and Modibo Sidibé of FARE. According to our results, the latter’s support never exceeded 10% (9.2% in Bamako, 4.1% in Segou, 5.0% in Sikasso and 7.4% in Koutiala).

The case of Ibrahim Boubacar Keïta: One of our more remarkable findings is the support for the RPM’s candidate [known as “IBK”]. In all four cities, he leads by a clear margin. Everywhere but Koutiala [which happens to be his birthplace!], his score is twice that of the second-place candidate.

The case of ADEMA and the URD: The URD is in second place in Bamako and Sikasso; in Segou, second place is held by Dramane Dembelé of ADEMA, while in Koutiala it is held by Omar Mariko of SADI. Neither ADEMA nor the URD has yet secured enough support to assure getting through to the second round.

Will there be a second round of the presidential election?

Looking at the results from these four cities, we see that no candidate has more than 50% of the vote. Supposing that results elsewhere are similar, we can say that there will inevitably be a second round of the presidential election. This conclusion could change if “IBK” gains more support. The URD seems more likely than ADEMA to make it to the second round.

How will the support for candidates who lose the first round be divided? How much weight will their endorsements carry? Who will be the next president of Mali?

In countries like ours, a great many candidates do not enter elections to win, but rather to negotiate government posts through alliances in the second round, or to test their popularity for subsequent elections. Therefore, the distribution of their supporters will depend in large part on the alliances that are struck after the first round, and on the weight of their endorsements. During our survey we asked questions about these different aspects. The analysis of the responses will be part of the second phase of this project (in the coming weeks).

The limitations of the survey: Like all scientific work, ours has its limitations. This first poll is limited in two respects: it concerns only four cities, and was conducted well before the scheduled election date. These two limitations are mitigated by the fact that the cities selected are known for their weight in the choice of president, and the fact that the populations these cities are reasonably well informed and choose their candidates well in advance (this latter hypothesis is not yet verified for rural areas).

My closing remark: As Mr. Guindo indicates above, such information must be interpreted with caution. Not having participated in the data-gathering and analysis process, I cannot personally vouch for the accuracy of these data. I am actually rather conflicted about Mr. Guindo’s call for statistics to play a greater role in Malian public life: in my view, while these tools can be useful, we Americans have taken our obsession with quantification and polling too far. But for whatever they’re worth, I think his results make interesting reading.

Postscript, 30 May: Bamako’s Le Pretoire newspaper has published an editorial denouncing “the publication of complicit polls whose scientific value is close to zero, in the eyes of nearly all experts.” The editorial does not identify Mr. Guindo by name, nor does it identify any of the experts who have allegedly found his or any other recent survey worthless. This is the first time I’ve seen polling as the subject of invective in the Bamako press — which means it must be attracting at least some attention.

Postscript two, 7 June: Maliweb has posted an item claiming to be by the “Société Malienne de Sondages” (Malian Survey Company) reporting the results of a different poll, which give Soumaïla Cissé 40% support among those surveyed, and IBK only 10% support — effectively inverting the order reported by Sidiki Guindo. No details are provided on the sampling frame, on where or when the survey was conducted, etc., and the polling company in question appears to be a very recent creation (a Google search reveals no mention of it until this week). The Bamako newspaper La Nouvelle Patrie has also run an article about this supposed poll, not even attempting to mask its position in favor of Soumaïla Cissé’s candidacy.