Jennifer McGee, now at Colorado, performed a detailed student of the effect of different student behavior on their test average and conceptual gain. A detailed measurement of the actions a student performs in an introductory physics class is presented.
The correlation between various student actions and hourly test average and normalized gain is calculated.
The correlation analysis shows that reading has a disproportionate influence on conceptual gain.
A linear regression analysis shows that only 20% of the variation in student performance is explained by student behavior. Click here for the thesis. This research was extended to more students covering both versions of UPII. The results were presented as a talk at the Summer 2006 AAPT meeting. Click here for the talk.