This paper had the following two data handling decisions. At this point, we do not think they were optimal, but they do not impact the qualitative pattern of results,

Though these decisions do not impact the qualitative pattern of results, relative to taking subsequent attempts on a problem into account and treating student attempts on scaffolding problems as correct in some situations, further work using this data set (e.g. San Pedro et al., 2014, 2015) uses the policy of taking subsequent attempts on a problem into account and treating student attempts on scaffolding problems as correct in some situations.

An updated version of these results incorporating these alternate decisions is currently under review by a journal.