PValues

Abstract

The advantages of reporting exact P values rather than using conventional significance levels are outlined. While P values do not represent the probability of truth of the null hypothesis, they do provide an index of the strength of evidence against that hypothesis. The P value depends on whether the alternative hypothesis is one‐tailed or two‐tailed, with the two‐tailed P value generally being twice the one‐tailed P value. There is considerable debate about how P values for statistics with discrete distributions should be computed. In situations where the null hypothesis is known to be false, P values provide useful information about the probability of error in the establishment of the direction of the true effect.

Affirmative Action and Leadership Attitudes in Brazilian Women Managers

“…To assess the effects of perceiving the existence of AA on womens’ leadership attitudes ( N = 51) and the existence-justice interaction, we used hierarchical moderated regression analysis, composing the interaction from the product term (Aiken & West, 1991; Cohen, Cohen, West, & Aiken, 2003; see Table 2). Because our hypotheses were directional, we used a p value cutoff of .10 (e.g., Dracup, 2005). Neither AA presence ( M = 1.76, SD = 0.43; β = −0.09, p > .05) nor AA justice perceptions ( M = 2.22, SD = 0.36; β = 0.02, p > .05) were significantly related to self-evaluated leadership capacity ( M = 3.64, SD = 0.63), contrary to Hypothesis 3, although the interaction effect was significant ( β = 0.36, p = .02, Δ R 2 = .10).…”