And yet the small-sample effect does hold: Reply to Juslin and Olsson (2005) and Anderson, Doherty, Berg, and Friedrich (2005)

Y. Kareev (2000) showed that samples of size 7 ± 2 increase the chances of detecting strong binary correlations. P. Juslin and H. Olsson (2005) fault that analysis for not taking into account posterior probabilities and for concentrating on signal trials only; they maintain that if that is rectified, the conclusions do not hold. R. B. Anderson, M. E. Doherty, N. D. Berg, and J. C. Friedrich (2005) mostly concur with them. Here the author argues that it is immaterial whether posterior probabilities are used, that careful consideration of benefits and costs of all outcomes justifies the original analysis, and that any apparent residue advantage of larger samples stems from overlooking the distinction between weak and strong correlations. Furthermore, Anderson et al.'s analysis of continuous variables in fact supports and extends the original claims.