This minor release fixes a small but longstanding bug in the drawing of contingency tables. The expected count for a cell should be its row total times its column total divided by the total of all column totals (or row totals; they are the same number). However, somebody mistakenly divided by the number of stories in the data set. We never noticed the error before because the answer count for most questions is not far below the number of stories. But when one or both of the graphed questions had few answers, the red expected-count circles were drawn smaller than they should have been. This error is now fixed. Note that this was only a graphing error; it did not affect statistical results.
Posted in Releases.