Two business professors are radically revamping a decades-old academic tradition. Rather than vet a research paper for publication after it’s been written, they evaluate the research first – before it even begins.
For the first time in the fields of finance and accounting, two top journals are using this “registered reports” format to publish special issues, under the guidance of two professors at the Samuel Curtis Johnson Graduate School of Management.
“Our approach is, ‘Draw the target and then shoot,’” said Robert Bloomfield, the Nicholas H. Noyes Professor of Management and a guest editor at the Journal of Accounting Research (JAR). “What we’ve done until now is we shoot, and then we draw a target around it – and pretend we’re awesome. … Almost every paper that’s published in any of the top journals in finance or accounting make it sound like they are perfect archers and hit the target directly.”
Registered reports have been gaining ground in the life sciences and other fields. But until now it hasn’t cracked business journals.
“I wanted to turn the business of editing journals upside down,” said Andrew Karolyi, Johnson’s associate dean for academic affairs and executive editor of The Review of Financial Studies (RFS), among the top three finance journals.
In the current system, a scholar conducts research, writes a paper, then submits it to a journal. Papers often go through two to three rounds of revisions before they’re accepted and published.
But, Karolyi says, “People have gotten very, very skeptical about the integrity of research published in top journals.”
That’s because only high-impact, bigger-than-life results are likely to get published, Karolyi said. Reviewers routinely say a paper’s results may be correct but are not that exciting and suggest authors slice and dice their data until they get a more sensational result, he said.
It’s known as “HARKing,” short for “Hypothesizing After Results are Known.” And plenty of researchers are criticized for what are known as QRPs: questionable research practices.
“Every one of these papers … imply very large, economically important results, and we just know reality is not that clear. We are manipulating results so they look good to the outside world,” Karolyi said.
In contrast, the registered reports format requires a scholar to submit a research proposal to the journal, before she even begins the work. Editors and reviewers vet the proposal based on its potential to answer an important question, its research design, proposed data collection and the inferences the researcher would draw from her findings. The scholar must then defend her proposal in front of the review committee and editors. If the journal accepts the paper, it guarantees it will publish the paper – no matter what the scholar’s findings.
“The idea is we shift the risk from the researcher to the journal,” Karolyi said. “We are at risk of publishing a paper that has a nonresult. All we’re asking scholars to do is to be honest; tell us what you did. If they veer off the plan, which is perfectly OK, just tell us.”
In 2014 Bloomfield suggested JAR – among the top three accounting journals – adopt registered reports. It did, for the May 2018 issue, which they asked him to join as a guest editor. Under his leadership, JAR’s reviewers are vetting papers on empirical research related to accounting.
Registered reports encourage authors to be more ambitious in their data-gathering, Bloomfield said. “We have a paper where teams of researchers went through every single major newspaper article in most of the world that mentioned ‘scandal’ and ‘regulation’ since 1860!” he said. “Who is going to do that just on spec?”
But, Bloomfield, cautions the format comes with a cost.
“We’re applying the science of reporting – that is, accounting – to the reporting of science,” Bloomfield said. “And the first thing we learn in accounting is, no reporting system is perfect.”
The registered reports format does limit a scholar’s discretion to explore areas that emerge as the research progresses, Bloomfield said. “There can be a lot of useful stuff you didn’t anticipate at the beginning that now you have less incentive to really dig into it because you’re going to get published anyway,” he said. To compensate, JAR will allow authors to include this material in supplemental analyses, he added.
At RFS, Karolyi has organized two upcoming issues using registered reports on rarely studied themes: financial technology, in 2018, and climate finance, in 2019. The financial technology authors will present their papers March 16-17 at a conference held at Cornell Tech.
“People have been talking about this integrity problem for decades. It’s just nobody’s ever done anything about it,” Karolyi said. “And now we’re trying.”