Professors Offer a Reality Check For Politicians By LYNNLEY BROWNING When Leonard Wantchekon set out to investigate what kind of campaign messages are most effective, he could have done the standard review of past political advertisements or interviewed candidates, consultants and voters. But Professor Wantchekon, who teaches politics at New York University and is an expert in game theory, wanted a more scientific approach. So he chose a novel but controversial approach: instead of observing an election, he participated in one. Born and reared in Benin, Professor Wantchekon decided to use the 2001 presidential primary in his small West African home country as his case study. As a younger man, he had helped found the Democratic Front of Benin, in the early 1980's the only opposition group to military rule in that country, and in 1988immigrated to Canada as a political refugee. With his previous political ties, he was able to convince Benin's four primary candidates to allow him and his researchers to write and disseminate campaign messages for each of them and then test them on villagers. Working hand in hand with the two opposition and two pro-government parties, Professor Wantchekon and his team created two types of messages. One used specific promises — to build schools, clinics and roads in a particular village, for example; the other invoked a much broader appeal to improve the nation's general welfare. The team enlisted tribal leaders, soccer stars, popular musicians and local officials to help deliver their messages. From December 2000 to the primary in March 2001, they campaigned in about 20 villages, each with some 700 inhabitants. Each village was randomly assigned to receive either the specific promise or the broad appeal. Some 4.1 million voters in other villages who received the candidates' regular campaign messages, and not ones prepared for the experiment, served as a control group. The research, known as a field experiment, was unusual, Professor Wantchekon said, because "it involved real candidates in real elections." But intervening in real elections creates problems, say some political scientists, who argue that potentially tampering with election results is unethical, as is having people unknowingly participate in an experiment. "There are some major ethical concerns with field experiments in that they can affect election results and bring up important considerations of informed consent," said Rebecca B. Morton, who also teaches politics at New York University. F. Christopher Arterton, dean of the Graduate School of Political Management at George Washington University, said he was not familiar with the study but cautioned that in general with field experiments, "one has to be careful not to cross the line between being a researcher and being an advocate." Professor Wantchekon, who still counts himself among the pro-democracy advocates in Benin, said such concerns were overblown. Villages were carefully screened to help ensure that the experiment would not influence the results, he said. Only villages where the votes were not close in the previous election were selected. (As it turned out, the incumbent, Mathieu Kérékou, a Marxist former military ruler, won). Professor Wantchekon also argued that it was perfectly reasonable not to inform villagers that they were participating in his experiment because doing so would have skewed the study's results. In the end, he emphasized, it would be the villagers who might eventually benefit from his findings. Professor Wantchekon said his results, which are under peer review, showed that specific promises worked well for incumbents, but not for challengers or underdogs, who won more votes when they used broad appeals. Women also preferred broad appeals. The findings, he added, could sharpen the political debate: "There are clear benefits to voters from this experiment." Candidates courting women, he said, might be motivated to work for more lasting changes in children's health care and education. To Professor Wantchekon and other proponents of field experiments, that method has a scientific rigor that is lacking in other types of political analysis. It is well accepted in medical research, for example: when scientists test new medical treatments, they compare a new drug to an existing one or to a placebo by testing it on people who are randomly chosen and who do not know which drug they are getting. Random selection gives researchers confidence that any difference in results between the two groups is due to the treatment. Similarly, field experiments, supporters say, promise to shed light on important political questions, from why people vote to what makes for successful welfare reform. "We see experiments in medicine and say, `We can do that,' " said Donald P. Green, who heads the Institute for Social and Policy Studies at Yale University, a research center that sponsors field experiments, including Professor Wantchekon's. Professor Green concedes that critics of the method raise important questions. But in his view, designing social programs around untested theories or allowing problems to go unfixed is worse than staging randomized experiments that can produce important findings. Some scholars point out that because the method takes the researcher into the real world with all its complexities, it offers less control over what is being measured, which makes the results less verifiable. Unlike the messy real world and all its hard-to-control variables, the laboratory is "refreshingly free of confounding influences," argues James E. Alt, director of the Center for Basic Research in the Social Sciences at Harvard University, who says he personally favors statistical analyses. But Professor Alt also acknowledges that the artificial environment of the laboratory can distort results. He said that while some researchers in recent years have sought to make laboratory experiments more realistic — setting up television sets in shopping malls, for example, where participants can view campaign commercials in a natural setting — laboratory experiments still lack the context and realism of the field. Field experiments began in the 1920's with agronomists, who wanted to improve crop yields — hence the method's name. An early experiment in politics, in 1927, looked at voter turnout in Chicago. Although a handful of political scientists tried the technique, their efforts were almost completely forgotten by the 1960's, as the statistical revolution swept through the social sciences, including political science and economics. A small but influential group of academics, including Professor Green, began resurrecting field experiments in the late 1990's in an effort to put the "science" back in political science. He began in his own backyard. He and Alan S. Gerber, another political scientist at Yale, wanted to find out which type of campaign was most effective in getting voters to the polls: mailings, telephone or door-to-door canvassing. During the 1998 local elections in New Haven, Professors Green and Gerber randomly surveyed 30,000 registered voters. They found that canvassing in person increased turnout by 9 percentage points. Telephone appeals had no effect. Mailings increased turnout by up to several percentage points, depending on how frequently they were sent out. The results, they said, could be used to create programs that promoted greater voter turnout. There are huge questions that field experimentation cannot answer, for obvious ethical and logistical reasons. For example, what conditions make for a successful coup d'état and what causes genocide? And while scholars might find it useful to identify the reason people don't bother to vote, most would consider it unethical to stage an experiment that kept voters away from the polls. Kathleen McGraw, who teaches political science at Ohio State University, said that while field experiments provided more rigorous findings that could be used to improve democracy, the ethics of a particular experiment must be considered on a case-by-case basis. "It all comes down," she said, "to whether in specific experiments, the potential benefits outweigh the potential harm" to participants. One place likely to feel the impact of field experiments, if they become more widely used, is the multimillion-dollar political-consulting industry, which thrives on creating campaign messages of unproven and untested value, said Professor Arterton of George Washington University. "It's an industry that is very skilled at talking about the factors that produce a result," he said, "but it's not very effective. Field experiments could make some political consulting firms out there very uncomfortable." >From: Momodou Camara <[log in to unmask]> >Reply-To: The Gambia and related-issues mailing list ><[log in to unmask]> >To: [log in to unmask] >Subject: Re: Do you want to win an election? >Date: Sat, 31 Aug 2002 10:42:19 +0200 > >On 31 Aug 2002 at 0:21, Ousman Gajigo wrote: > > > Here is an interesting piece from the New York Times on the science or >art > > of winning an election. > > > > http://www.nytimes.com/2002/08/31/arts/31FIEL.html > > > > If only one could find a proven mathematically formula for PDOIS... > > > > Ousman > > > >Hi Ousman, >Could you please send the piece? It seems one has to be a registered member >to >be able to access the information. > >regards, >Momodou Camara > >~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > >To unsubscribe/subscribe or view archives of postings, go to the Gambia-L >Web interface >at: http://maelstrom.stjohns.edu/archives/gambia-l.html >To contact the List Management, please send an e-mail to: >[log in to unmask] > >~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ _________________________________________________________________ Chat with friends online, try MSN Messenger: http://messenger.msn.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To unsubscribe/subscribe or view archives of postings, go to the Gambia-L Web interface at: http://maelstrom.stjohns.edu/archives/gambia-l.html To contact the List Management, please send an e-mail to: [log in to unmask] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~