Culpability when publishing goes awry

From the dredges of academia… a recent study published in Science Magazine claimed to show that with regards to gay marriage, voters could be influenced by something as simple as a conversation with a gay vote canvasser (quick summary – but the full study is here).

The study is certainly interesting, but it appears that some of the data may have been falsified. This is still an unfolding story, so I don’t want to rush to judgment.

My issue here is in some of the commentary regarding the story. I was listening to NPR last night, and Kenneth Prewitt, the incoming head of the American Association of Political and Social Science, was interviewed. During the interview, he was asked if the faculty member (and 2nd author) on the study was culpable given that the graduate student (and 1st author) had been responsible for the data. Prewitt stated that the faculty member was NOT culpable, and was not responsible for the data.

The idea that a faculty member – or any author – is not responsible for knowing the data in a journal article strikes me as unacceptable. I believe that if you’re working on a journal article that is going to be published – particularly with graduate students – you have an obligation to understand all aspects of the data. Yes – that sets a high bar – but the bar should be high.

Faculty members who work with graduate students are responsible for teaching those graduate students about proper research protocol, the nuances of research and data analysis, and the core tenets of research publication. Data are often messy, and a lot can go wrong in the process of moving from raw data to publication. Moreover, the method of data collection matters as well, and can have a significant impact on the resulting analysis. To trust a co-author to take responsibility for these choices is setting the stage for problems further down the road.

All of this is not to say that Donald Green, the 2nd author, behaved in this way. My issue here is with Prewitt’s statement that Green is not culpable. He is an author on the study. He is culpable. What Green’s culpability means in the long run is a subject for debate and investigation, and I’m not passing judgement beyond stating that I believe he should be held responsible.

As I step off my soap box, I’ll close by reinforcing the growing call for transparency in all aspects of research. I’m starting to work more and more with large datasets and algorithmic data collection. My team will be publishing all of our code and raw datasets to github and a public data repository so that others can replicate our work. I hope the others will increasingly follow suit.