Sunday, June 12, 2011

An analysis of Scantron testing

Below is a link to a paper (PDF file) I did on Scantron testing at my school. It was done for a course requirement:

It's written in an academic mode, as opposed to a critical or agit-prop mode, with lots of tables and basic statistics. Of some interest: mode effects overall appear small; a large number of students spoil tests; there is a large discrepancy between Scantron national percentile rankings and ISAT NPRs; there is a relatively high general correlation between school-level Scantron results and ISAT results, which breaks down when you look at individual classrooms. The final "recommendation" in the paper is that the tests should not be used for either teacher evaluation or student promotion.


No comments: