Friday, May 3, 2013
At AERA I attended a range of presentations, using data and analytic techniques from the historical to the quantitative. Many of the presenters struggled with shortcomings in their data, or at least limitations. One paper used the average wealth in zip codes to approximate student family income - problematic in some ways, but a commonly used tactic. Other data issues proved more intractable. For example, Brian Bridges used some data from an ACE survey on presidents to examine HBCU presidents, but the data wasn't broken down by public/private institutional control. Another paper used a survey conducted with the development office to predict young alumni giving, but the researcher wasn't allowed to ask about income - surely an important predictor. Of course, the alternative would have been no data at all. One takeaway I was reminded of was that people who work for higher ed organizations have some of the best access to pre-existing datasets, but not necessarily the time to mine them on their own - so this is a great place for scholars looking to collaborate.