P-curving A More Comprehensive Body of Research on Postural Feedback Reveals Clear Evidential Value For 'Power Posing' Effects: Reply to Simmons and Simonsohn
Forthcoming in Psychological Science
17 Pages Posted: 19 Oct 2017 Last revised: 7 Dec 2017
Date Written: November 6, 2017
Abstract
We conducted a series of p-curve analyses, following Simonsohn et al.’s rules of p-curving and using a systematically selected, comprehensive and updated set of published studies of “power posing” (i.e., postural feedback), comprising 55 studies (vs. 34 studies), which yield starkly different results and conclusions from those of Simmons and Simonsohn (S&S):
(1) evidential value for postural feedback across aggregated effects;
(2) evidential value for a clearly specified single effect – feelings of power – which was omitted from the p-curve analyses presented by S&S; and
(3) remarkably strong evidential value for a well-defined, theoretically-important category of effects from the same set of studies identified in our systematic review - all measures of “feelings,” including emotions, affect, mood, and evaluations, attitudes, and feelings about the self.
This third p-curving analysis of the set of emotions-related effects, which include methods and variables that are more resilient to demand characteristics, either because responses were implicit or otherwise difficult for participants to control, responses were embedded in a broader survey instrument, or, as demonstrated in recent research on demand effects in survey research, participants likely varied in their orientation such that some would have wanted to confirm the hypothesis, some to disconfirm it, and others would have been indifferent, also demonstrate that power posing effects are not merely demand effects. How did two groups of researchers reach such divergent findings and conclusions about the same area of research?
Our analyses reveal two of the practices that contributed:
(1) sample selection decisions that may lead to an incomplete or non-representative set of studies and/or effects for inclusion in the analysis; and
(2) undifferentiated aggregation of disparate effects (i.e., the apples-and-oranges problem).
Keywords: nonverbal behavior, postural feedback, power posing, meta-analysis, p-curving, emotion, affect, self-evaluation
Suggested Citation: Suggested Citation