Fig. 2.
Selected results from the simulations used to evaluate the Carlisle Method. If the expected distribution from the sum of P values from a trials report’s “table 1” (x-axis) is the same as the observed distribution of P values (y-axis), the test values (colored lines) will fall directly on the diagonal black line. Departures from the black line indicate false-positives as the observed distributions exceed what is expected by chance. When the correlation between items is 0 (top row), the Carlisle Method performs well. However, as the degree of correlation (rows) increases, many false positives are observed. This effect is amplified with increasing number of items (columns) in the table.