SIOP 2015: Reflections Part 1, on Big Data

Somehow, every year, SIOP rises a little bit in intensity for me. There are a few more sessions I need to attend, a few more people I need to connect with, and a few more drinks I need to space out better throughout each evening. This particular year, things reached a fever pitch when I got a bit of long-term news mid-conference: I have been granted tenure at ODU! I suppose that means this blog wasn’t such a horrible idea after all.

One of the reasons for this ramp-up in conference intensity for me is the increasing focus at the SIOP conference on technology. It is a change I honestly was skeptical would ever happen; in fact, in our primary journals, it really still hasn’t. But at our annual conference, at least, there is a growing recognition and appreciation for technology research related to workplace behavior. This year, the focus was clearly Big Data, with about a dozen presentations on Big Data specifically and another dozen or so on related approaches, such as workforce analytics. One thing that became apparent quite quickly was that Big Data presentations all had essentially the same content, which I can summarize for you here: Big Data is complicated, Big Data provides answers to questions we didn’t know we had, Big Data can’t answer causal questions, and Big Data is so complicated that I/Os don’t really understand how it works. Depending upon whom you ask, this last part is either a huge problem or a huge opportunity.

The problem is obvious, but the opportunity is complicated. On one hand, there is a perception there is an opportunity for I/O to “take over” Big Data. There were many presentations about how Big Data is just an extension of what we already do, how we’ve already been doing Big Data for years, how Big Data is nothing new, etc. These are all based upon false assumptions, primarily driven by a misunderstanding of what Big Data actually involves. On the other hand is the viewpoint I endorse,which is that Big Data is an opportunity either 1) for I/O to enter a new era of interdisciplinarity, cooperating with computer/data scientists in a way we never have before or 2) for I/O to begin training graduate students in computer science so that can eventually be in a place to contribute meaningfully. Or maybe both.

There is some movement on both fronts. In terms of I/O training, one of the questions asked at one of the panels I was on was if computer programming should be taught to new I/O PhD students. I think it should, and I’ve been teaching it for a few years now. But one doctoral I/O program is not enough to make a dent in Big Data, and definitely not any time soon.

In terms of interdisciplinarity, businesses are beginning to realize that Big Data is not the solution to every problem, and this is a very promising shift for our field. As evidenced in part by this article in the New York Times, Big Data must be paired with might be called “small data” in order to fully understand any particular organizational phenomenon. The definition of “small data” could very well be “what I/O psychologists do.” It involves carefully validated surveys, a strong command of the research literature, and the interpretation of data based in expertise surrounding both. If we take that approach – that we are the meaning-makers for the Big Data folks – I think I/O will find a very comfortable position within the Big Data movement, one which neither sacrifices how we define ourselves nor pretends that change isn’t in the water.