Adding a timing light to the “tool box”: Stephen DesJardins, Brian McCall, Dennis Ahlburg, and Melinda Moye

Read here

This 2002 paper by Stephen DesJardins et al. is a response to Clifford Adelman's "Answers in the Toolbox Study" (see a summary of that paper in a previous post) with a focus on adding the element of time to their analysis. The researchers believe that different factors will have greater or lesser effect as time goes on. Their model asks, if a subject has not graduated from a 4-year college by time t-1, how does a given factor correlate with graduation. While they mainly use the same factors outlined by Adelman, and agree with many of his conclusions, they show that the effect of different factors change when time is taken into account.

DesJardins et al. use the same data set as Adelman, and start by replicating his results. Afterwards, they start testing various factors, and how they change correlation to gradation over time. In their final model, for example, being male is initially negatively correlated with gradation. But over time, the effect becomes less negative, and by around year 7, being male becomes positively correlated with graduation. Their biggest critique of Adelman was his academic resources factor, which was and index of high school intensity/quality, high school gpa/class rank, and high school test scores. They found the index became less predictive overall when college gpa was included, a fact Adelman acknowledged in his paper "The Toolbox Revisited." Much more interestingly, when they separated academic resources into its three factors, they all had relatively similar, positive prediction power at year one, gpa/rank and test scores declined in influence as time went on, eventually becoming negative, while intensity/quality grew and became much more significant as time passed.

Overall, this study had many of the same findings and limitations as Adelman's "Answers in the Toolbox" study, such as the fact that it has a hard time measuring the effect of student aid due to problems with reporting. Despite its similarities, it adds more credibility to many of Adelman's arguments while highlighting areas where Adelman may be wrong.