May 30, 2012
New project aims to double-check international aid program resultsHow much do adult male circumcision programs in Africa reduce HIV rates [...if at all]? Does group lending, which encourages repayment by putting people in support groups, make it easier for peasants in Thailand to pay back loans? There are high-profile studies that answer those questions, said Benjamin Wood, a researcher at the International Initiative for Impact Evaluation's Washington, D.C. office. But his organization thinks there's still a piece missing – studies that double-check the results from aid program research. "It's a bit of a hole in the field," he said. So the initiative is looking to fund repeat studies, which have been perpetually unpopular in almost any field of science.
Getting the same results twice or several times over is an important part of showing a study is true. Repeating experiments can also resolve debates about studies some experts think were poorly done. They help the U.S. government, which spent $15 billion on international aid in 2011, and private donors, who spend billions more, decide how to spend their budgets. Yet repeats are rarely published because researchers don't have any incentive to perform them.
One problem is that scientists build their reputations by publishing new, exciting papers. "Academics have to get published. That's their job," Wood said. Academic journals are less likely to publish a repeat, however, especially if the repeat that finds the original researchers did everything right, Wood told InnovationNewsDaily. Papers that point out something is wrong are more popular, he said.
That leads to another problem. Because repeat studies tend to be negative, researchers are reluctant to release their original data for replication.
So a new project Wood is leading will give researchers money to do repeat studies. His organization will publish preliminary results on their website, so good work doesn't get lost just because it's a repeat. The initiative won't favor negative results, so original researchers shouldn't worry a repeat request will necessarily hurt them.
In fact, original researchers could benefit from cooperating on a replication study, he said. A positive repeat will bring positive publicity to the original study and make it seem more reliable, so more people will read it. "These are all things that are helpful for the original researchers," Wood said.
The initiative will also post on their site if the researchers won't give up their data within three months of getting a request. The listing isn't exactly meant to embarrass researchers, according to Wood. "I think it's just kind of an identification," he said, "and maybe if you're unable to produce your original dataset, it starts to raise questions."
Wood is now looking for experts to nominate studies to get repeated. He wants influential or controversial studies, especially.... The International Initiative for Impact Evaluation may perform in-house a repeat on a study about male circumcision for HIV transmission prevention.