After publication of the first edition of On the Success of Failure, Shepard and colleagues (1996, hereafter SSM) published a critical review of the book. Our initial response (Alexander 1998) sought to correct what seemed to us the review's chief flaw, its misrepresentations of our conclusions and interpretations (e.g., SSM's assertion that we express “outright enthusiasm for retention”). Since then numerous references to the first version of Success have included disclaimers citing SSM as identifying technical flaws and suggesting that SSM's “reanalysis” of the data yielded different conclusions (e.g., Hauser 1999; Rury 1999; Thompson and Kolb 1999; Viadero 2000). Shepard, Smith, and Marion (1998) also claim they reanalyzed the data in their published rejoinder to our response. Further, they say the original review “pointed out errors in [the] analyses that allowed [Alexander and coworkers] to see some benefits for retention that were not there once their data were rescaled and appropriately analyzed” (Shepard et al. 1998: 404). In hindsight, it was a mistake not to address their technical criticisms in our rejoinder. Their criticism rests mainly on interpretation, not new evidence: SSM did no reanalysis of the data in the book; nor did they discover flaws in our analyses.

To restore balance to the discussion surrounding the book, we turn now to SSM's specific criticisms. To do so, we return to their original review, pointing out areas of agreement and disagreement. For each subsection of SSM's review we first summarize their argument, then respond to their main points.