Donald Klein’s response to Martin Katz’s reply
Table 1 reproduces Katz's findings as presented in my Comment on “Onset of antidepressant action”.
It was a mistake not to spell out the simple algebra that allows the reconstruction of the original 2x2 data tables, from the indices in Table 1.
The predictor is the group of those subjects who had at least 20% improvement on a specific variable vs the others, during the first two weeks of the trial. The outcome variable is the group of those subjects who had at least 50% improvement on that variable at trial end vs the others.
The cells of the 2X2 are usually designated
Adding up to N. The number called positive at outcome is (A+C) The number called positive at 2 weeks is A+B Since we don't know the Total Number positive at outcome ( =True positives +False Positives) we stipulate it as N-X therefore the number called negative at outcome is X.
That allows the simple formula: TP*(N-X) +TN*X = C*N. Note the variable C is not the cell label ”C” but rather TP + TN . Entering the given values determines X, which allows filling in the entire 2X2.
The results are not exact integers due to rounding in Katz's values. These figures are displayed in Table Two of my initial comment.
The important issue is that they support every negation of all of Katz’s hypotheses, as previously detailed. Since my analysis is now understandable. Marty can address the fact that his findings do not support his theories.
Donald F. Klein
March 12, 2015