The War of the Surveys

Semana Santa is over, and the latest bulletin from COE (Centro de Operaciones de Emergencia) reports that 40 people died during Easter week this year (16 more than last year). The majority died in traffic accidents, and the majority of dead in traffic accidents are motor cyclists (probably without a helmet). The last day of Semana Santa also brought the first dead of this electoral campaign as fighting turned into gunshots between PLD and PRD supporters during campaigning in Moca. One PRD supporter died and a PLD activist was wounded. During the easter week the candidates had promised not to do any campaigning, so Mejía and Medina went to the USA to campaign there (since campaigning abroad was not mentioned in the Pacto de Civilidad the candidates signed a week before Easter). The major parties' activists, however, campaigned both on Thursday and on Easter Sunday, when traffic in and out of the cities was the highest, creating a lot of extra traffic jams and frustration among people.

Now that Easter is over, the quiet week turns into a noisy last 40 days of electoral campaigning. I am taking a lot of taxis in the capital these days, and most taxistas say that it doesn't matter who wins as long as there only is one round of elections (In fact a political cartoonist from El Día argued today that Jesus Christ's suffering only lasted the Long Friday, while the Dominican people's suffering would last until May 20, ie. election day). A particular feat of the campaigns in the DR is "la guerra de las encuestas", the war of the Surveys, and today I got to experience a little part of that when I went to a press conference organised by Benenson Strategy Group (see also Wikipedia) who wanted to publicise their latest survey. Before I get to that survey, let me briefly display this "war" between the top two candidates.

The bandwagon effect is well known among behaviouralists and within political science and electoral studies. The idea is that people feel better if they support the winner, and that makes some voters vote for the candidate that seems most likely to win the election (independent of what they think of this candidate). The opposite effect would be the underdog effect, but of the two, the bandwagon effect has gotten more attention and seems to be more solidly supported by evidence and studies. The parties in the DR must think that the bandwagon effect is particularly strong here, because both major candidates put great effort into publishing that the surveys put them on top and a sure winner in the first round. Therefore walking the streets of Santo Domingo you can see that part of Danilo Medina's strategy is basically to show that a majority supports him, see for instance this poster (even though the guy you see on the photo in the lower right corner, and his three other friends (not in the photo) playing domino on a street off the Malecón in Santo Domingo on Saturday April 7, all said they supported "Papá", ie. Hipólito Mejía).

Several places in the city you'll see the posters showing that either 52 or 56% prefers Danilo or thinks that Danilo will win the election. Both parties also pay for full page ads in the major newspapers. On Wednesday 4 of April, Danilo reports that all surveys show him as winner in the first round, referring to ASISA's survey of March 27 and 28 which as Medina as winner with 52.8% against Mejía's 45% (others 2.2%), and NEWLINK Research's survey from March 31-April 1, which has the Danilo/Margarita ticket(-s) as winner with 49.1% against Mejía/Abinader with 44%. Today, on April 9, however, Hipólitio Mejía and the PRD retaliated with a full page in El Día, El Nacional, and Hoy (and potentially other newspapers) stating "change is coming" and that the surveys confirm it. Even though Medina said that all surveys point to him as winner, Mejía could present at least four surveys that disagree with Medina. According to Mejía's ad, Bendixen and Amandi's survey from March 23-31 has Mejía as winner with 51.4% against Medina's 39.7%, JZ analytics (31 March) has 49% for Mejía, 44% for Medina, Ana María Acevedo (27-30 March) has Mejía/Abinader with 51.8% and Medina/Cedeño with 44.7%, and Greenberg-Diario Libre has Mejía with 49% and Medina with 46%. My summary of these surveys is just below, and an average of all these surveys (plus the Benenson Strategy group survey) puts Mejía just above Medina with 47.2 against 46.3% (but bear in mind that this includes four surveys used in Mejía adds, two used in Medina adds, and the Benenson survey, which for reasons I deal with below, I am sure will be used in Medina adds soon. This gives Mejía a 4-3 edge in surveys for this average calculation).

So why are the surveys so important? Well, there are reasons to believe that the Bandwagon effect should be strong in the DR, although I have never seen any evidence that substantiates this. First of all, the DR is a clientelistic society and state. Literally thousands of people are promised jobs by the candidates during the campaign, and the candidates will deliver as best as they can if they win office. If you belong to the winner you increase your chances tremendously of getting a "botella", i.e. a state-job won through clientelism, or being put on the nomina which puts you on the payroll without you having to work; and if you are with the loser you are guaranteed not to get either. If you can demonstrate that during the campaign you worked for the winner, you own a party membership card, and so on, you have a shot. The bandwagon effect could be important for everyone trying to decide the last few months and wants to become active in the campaign with the motive of winning a state job. Second, as mentioned in an earlier blog, there are virtually no ideological differences between the two main parties, which both started out on the left, but are now relatively conservative parties. The less ideology means in an election, the more likely it is that voters may be swayed by other motives, such as a bandwagon effect. Third, many do not want a second round and may vote for the likely winner according to the surveys just to avoid a round number 2 (there is a ballotage if no candidate gets 50%+1 vote in the first round), which is likely to enhance a bandwagon effect. 

Most of these surveys cannot be taken seriously, and I do not trust any of these now that election day is getting closer. Historically it is Hamilton and Gallup that have produced the most reliable surveys here, but I have not seen any of these doing surveys in March. In February Hamilton reported that Medina had 50% vs. Mejía's 45% (and given the margin of error that is a technical draw). The latest example of a non-trustworthy survey is the one from the respectable US firm Benenson Strategy group. This is a firm that does surveys, but mainly to be used to provide consults to politicians, find political solutions, etc. See their webpage: http://www.bsgco.com. Benenson reports that Juan Manuel Santos, Barack Obama, UK Labour Party and other prominent politicians are and have been their clients. They sell information to politicians and give politicians advice (a bit like accounting firms/consulting firms doing accounting and consulting for the same companies, recent US history demonstrates that this was a bad combination). So Benenson is a serious firm, but normally hired by politicians for (among other things) survey-based advice, which in the DR context makes them unreliable given that potentially the best advice Benenson could give a candidate is to demonstrate strength in surveys, and then produce this survey.

Anyway, today, April 9, at 10 am Benenson called for a press conference at the relatively new hotel Holiday Inn in the posh area of Lincoln Avenue (fun fact is that while Holiday Inn is a moderately priced hotel/motel in the USA, in the DR Holiday Inn is a luxury, business hotel which is more expensive than the luxurious Inter-Continental hotel in Santo Domingo that belongs to the same chain of hotels). As demonstrated in the graph above, Benenson could report 48% for Medina and 40% for Mejía when all candidates were included; asked about the ticket (but excluding other candidates) 51% for Medina/Cedeño vs. 43% for Mejía/Abinader; it was pointed out that Medina/Cedeño was strong among women (55% vs. 37%); that the majority (51%) thinks Medina will win in the first round; that Mejía has the strongest rejection among voters, 39% say they would never vote for Mejía, only 27% say the same about Medina; that 77% say that the election will be decided in the first round and that among these 77%, 57% think that Danilo Medina will win (39% think Mejía would win); and that 54% were more satisfied with the Fernández government (2004-2012), while 43% less satisfied with the Fernández government compared to the Mejía government (2000-2004).

There are many things to say about the presentation of this survey, all of which convinces me that this survey is ordered by PLD, the Medina (and in particular the Cedeño) faction or companies close to the PLD/Medina/Cedeño. But, first I have to declare that since I was not a journalist I did not receive the CD with the background data material that the survey is based on, I only received the print-out of the PPT slides that the Benenson representatives showed at the press-release. Therefore these comments are based on my impression from that press-conference and from the PPT-slides. Even though I am quite confident that this survey is ordered by the PLD, or indirectly by the PLD, this does not mean that the numbers the survey shows are wrong. On that account I can only say that I have no way of knowing whether they are correct or wrong. Second, according to the PPT-slides the survey is based on 1200 respondents, which should make out a reprentative sample, and the margin is +/- 2.83 points.

There are five elements that convince me the PLD is behind this survey: 1) the questions, which all are centered around what seems to be Danilo Medina's message in this campaign; 2) the slides; 3) the presentation of the slides by Benenson; 4) the fact that Benenson would not say who hired them; 5) the type of firm Benenson is.

1) Danilo has centered his message on several key issues, most of which were covered with questions in this survey. Medina focuses on the fact that he shall win the women, and together with Margarita Cedeño he highlights the women's importance, arguing that Margarita gives Danilo a dimension that Abinader does not give Mejía. Second, as mentioned above as well, Medina argues not only that he will win, he will win in the first round, and that he is the candidate that most people think will win (in addition to being the candidate that most people would vote for). Third, Medina and the PLD works as hard as they can to link Mejía (rightfully) to his disastrous rule in 2000-2004, and that is the worst (and most disliked) politician of this century. Fourth, even though Medina offers slogans of "cambio", it is qualified as "cambio seguro", which points out the insecurity of the erratic Mejía. In addition, Medina points out that he will continue that which is good. All of these elements were covered by questions in the survey. The survey's questions thus fits Medina's message quite well, and as such, works to confirm what Medina is arguing day in and day out to the public. The survey first points out that Danilo is more popular than Mejía, then by adding the vice-presidency to the ticket (and by excluding the rest of the candidates), the survey suggests that Margarita brings more to the table than Abinader (Mejía's vice-presidential candidate), which is victory in the first round for Medina. Then it looks at gender-support for the ticket, demonstrating strong female support for Medina/Cedeño, before it builds on Medina's message that people no matter what they vote, think that he will win, and with another question asking the ones who knows who to vote for whether they think it will be decided in the first round and who they think will win (yes, Medina). Finally, supporting Medina's message of "cambio seguro" and continue what is good, it looks at the rejection rate of the candidates and compares Fernández with Medina. The questions (and sum of answers) all fit Medina's message too well.

All of this is done quite well, and are relatively weak circumstantial evidence for my hypothesis. Therefore, I turn the attention to the slides which includes a couple of, maybe, equally weak circumstantial evidence (the reader will have to judge this). First of all, the slides highlights Medina's positive results through headings, everything in the text is written as to underscore the strength of Medina instead of presenting dull, statistical survey results, and to convince us that Medina will win in the first round. Only Medina is mentioned in headings with one exception to highlight Mejía's high rejection percentage. Second, as a consequence of the first point, Mejía's name is almost never mentioned, all is about Medina and Margarita Cedeño, and strangely enough Margarita Cedeño is always mentioned with first and last name, Medina only with first name (is it because she is a woman?, or is there some other reason?). 3) On the gender-based breakdown of support for the candidates, the slide highlights and circles how more women than men support Medina, it does not highlight that more men than women support Mejía. Granted, this difference is smaller, but could have been mentioned. All in all, the slides seem created to convince that Medina is the sure bet, which fits well with a theory that you believe the bandwagon-effect is important.

The presentation in itself also supports, in my view, the tendencies I note above. The oral presentation seemed to aim to convince us Medina would win. But, given the fact that these guys are professional survey analysts, some strange corners were cut in the presentation (and on the slides) that made Medina look better. First is the strange assessment by the presenter (who I think was Giancarlo Sopo, but I am not sure, I arrived a bit late) that when including Cedeño in the question the survey demonstrates Medina would win in the first round (and implying her importance being greater than Abinader's). This is basically a white lie given the fact that this question only asked about preference of the two main tickets Medina/Cedeño and Mejía/Abinader and excluded the smaller parties. Therefore Medina reached the 50% threshold, it was not because Cedeño was added. In fact Benenson's own numbers give Medina a three point increase when adding Cedeño to the question (and excluding all minor parties/candidates), but the same is the result for Mejía/Abinader. Also, the oral presentation highlighted all the positive numbers for Medina/Cedeño and highlighted any negative numbers for Mejía without any caveats. Again, I repeat myself, my impression is that the presentation was made to convince us that Medina would win, not to present dry results.

When asked by a journalist who had commissioned this survey, Benenson answered "a group of private companies" whose identity they could not disclose. This is probably true. A group of companies may have paid for this on a promise of anonymity. For all we know these companies might be owned by Félix Bautista (see my previous blog). However, given this fact, I do not trust the survey. Surveys may be manipulated and I see no valid reason why some companies, whichever they are, should want to hide themselves if they order a survey done. Of the surveys listed above, we know who ordered one of seven. This is the one ordered by Diario Libre, and in my view is the most reliable of the ones I have registered in March (but a source in the PLD tells me that these figures were fixed before published in the Diario Libre...if that is true, I do not know, but what is clear is that Diario Libre did what they could to present Mejía in a positive light in their presentation of that survey, and that on the second day of releasing results from the survey it became clear that Mejía's lead over Medina was somewhat more qualified than what was presented the first day). Given the fact that the DR is a democracy, albeit imperfect, there is no reason to hold back who orders a survey unless you want to hide that it may have been ordered for political reasons (which I believe is the case here).

Finally, Benenson lives of giving advice to politicians and uses surveys to do so. The use of surveys is used to fine-tune politicians' message, not to report results and inform the public. Therefore I am not surprised that the questions fit Medina's message so well (this, however, is no reason for the survey company to cheat with the survey, but many results can be created by the formulation of questions, etc). Benenson in fact brags in their webpages of all the politicians they have worked for (and they are mostly politicians...can the PLD be considered a private company?), why would they suddenly want to keep this a secret now. It could be nice for Benenson to have the PLD and Leonel Fernández as clients. Fernández holds a very good name in Latin America, why not put that on their webpage? I believe this survey is made with the purpose of giving advice to Medina and Cedeño, which is what Benenson does. It is therefore not an objective survey, and I do not trust its results. It may be that the correct figures are presented, or it might be that the correct figures are only presented to the candidates to fine-tune the message, and that Benenson gave advice to "juice" the official stats to win more support. We cannot know. I can basically see no reason for why Benenson would harm their name in producing bad surveys, but then again, they will do what is best for whoever contracts them. If the presented results are false, I presume the candidates have seen the correct results. The only motive for not presenting the correct results must be the fear of not getting a contract renewal for another survey if the results are bad for the candidate. But, as mentioned, I have not seen the data material and cannot say anything about the veracity of the results presented. What I do think I have argued well for is that Benenson's message seemed like ordered by the PLD: The questions, the slides and the presentation fit Medina's message extremely well, they seem created to convince the public that Medina is a sure bet, and therefore seem biased and non-trustworthy.

I must finally add that this long story of a survey which I think is paid for by the PLD or companies that support the PLD, is only one of probably many such examples in this (and previous) campaigns. I will probably not write extensively about other politicised surveys, but I want to add that this could just as well have been a story about a survey ordered by the PRD. I happened to write about this one since I had the opportunity to go to this press-conference.