How robust are the initial power posture findings?
While the empowering message in Cuddy’s communications was impressive and remains important, many researchers were sceptical whether the short and subtle intervention of adopting a posture for two minutes could really induce changes in hormone levels as large as the ones reported. Much more intense interventions, such as actual competitions or intense stress induction protocols including a public speech and mental arithmetic under time-pressure and negative social evaluation (Kirschbaum, Pirke, & Hellhammer, 1993) typically yield smaller effects on cortisol and testosterone levels (Dickerson & Kemeny, 2004; Geniole, Bird, Ruddick, & Carre, 2017). Could adopting a posture for two minutes really have the similarly large effects? It seemed too good to be true.
Non-replications of hormonal and risk-taking effects and the following debate on replicability
When a first attempt to replicate Carney et al.’s findings in a considerably larger sample of participants yielded null-effects on all measures except feelings of power (Ranehill et al., 2015), a debate on the robustness of their findings started to unfold that has continued until today. At first, Carney, Cuddy & Yap (2015) responded with a review of 33 studies reporting embodied effects of expansive and constrictive postures, in which they argued that small methodological differences could be responsible for the non-replication of their original findings. For example, participants in Ranehill’s posture held postures for a bit longer (6 minutes) and did not watch faces while doing so, whereas the initial study had presented faces and instructed participants to form an impression of the presented people (they refer to this as a social filler task).
Non-replications of power posture effects on job interview performance
Two highly powered pre-registered studies (Keller et al., 2017; Klaschinski, Schnabel, & Schröder-Abé, 2017) have failed to replicate the positive effect of expansive postures on subsequent performance and hireability in a mock job interview (Cuddy et al., 2015). In the initial study, effects seemed mediated by nonverbal presence, that is, ratings as confident, enthusiastic, captivating, and awkward (reverse-scored). In addition, raters in the replication by Klaschinski et al. (2017) judged interviewees along other indicators of dominance and indicators of social competence. The results refute that adopting a posture before public speaking tasks affects others’ impressions along any of those dimensions.
A small but possibly robust postural feedback effect on feelings of power
In contrast to postural feedback effects on hormones and risk-taking, those on feelings of power have been partially supported. When one considers single studies, their results seem to partially contradict each other: some studies report significantly higher feelings of power in participants who adopted an expansive as compared to a constrictive posture (Fischer, Fischer, Englich, Aydin, & Frey, 2011; Huang, Galinsky, Gruenfeld, & Guillory, 2011 study 1; Park, Streamer, Huang, & Galinsky, 2013; Rotella & Richeson, 2013; Teh et al., 2016; Turan, 2015), while others report no significant difference (Cuddy et al., 2015; Hao, Xue, Yuan, Wang, & Runco, 2017; Huang et al., 2011 study 2; K. M. Smith & Apicella, 2017), and yet others even report an effect in the reverse direction (Garrison, Tang, & Schmeichel, 2016). Nevertheless, a p-curve analysis across all these studies (Cuddy, Schultz, & Fosse, 2018) shows that all available evidence taken together actually supports a differential effect of expansive and constrictive postures on feelings of power (see Figure 6). However, because p-values do not provide information about the direction or size of effects, conclusions from this p-curve analysis cannot distinguish between studies that observed an increase or a decrease of power feelings after the expansive posture. As the pool of analysed p-values included contradictory findings, the result only implies that expansive as compared to constrictive postures have an effect, but not that this effect is necessarily in the desired direction. Nevertheless, the fact that size and direction of the effect vary between studies, but is positive in most, could hint at a small true effect of postures on feelings of power. Consequently, the only way to detect it reliably would be to use very large sample sizes.
Postural feedback effects on other feelings and behaviour
Although the “power posture” findings were the first to attain such popularity, they were not the first demonstrations of an impact of bodily expansion on feelings and behaviour. An independent body of research building on a several studies by Riskind in the 80ies used upright and slumped sitting postures as bodily feedback manipulations (Riskind, 1983, 1984; Riskind & Gotay, 1982). While most studies after Carney et al. (2010) consistently interpreted the whole body expansion exclusively in terms of power and dominance, interpretations of postures with an upright or slumped upper body postures vary more broadly. Their interpretations comprise different affective states related but not limited to power, such as strength, mood, valence, happiness vs. depression, stress vs. relaxation (Riskind, 1983, 1984; Riskind & Gotay, 1982), pride (Stepper & Strack, 1993) or confidence vs. doubt (Brinol, Petty, & Wagner, 2009). Given that the review and p-curve analyses by Cuddy and colleagues (Carney et al., 2015; Cuddy et al., 2018) have incorporated these studies, I also include them in the below review of all effects other than those on risk-taking, hormones and feelings of power, clearly labelling the type of posture used in each study.
Table of contents :
Chapter 1 Why adopting a body posture might affect the agent’s social perception and behaviour
1. Postural expansiveness as a social signal of power, dominance and status
2. How bodily actions affect social perception
2.1. Cognition is grounded in perceptual, motor and introspective experience
2.2. Cognition serves to control bodily action in specific contexts
2.3. Causal impact of bodily states on the perception of social signals
Chapter 2 A critical review of power posture effects on human cognition and behaviour
1. Initial findings on power posture effects on feelings of power, testosterone, cortisol, risk-taking and job interview performance
2. How robust are the initial power posture findings?
2.1. Non-replications of hormonal and risk-taking effects and the following debate on replicability
2.2. Non-replications of power posture effects on job interview performance
2.3. A small but possibly robust postural feedback effect on feelings of power
3. Postural feedback effects on other feelings and behaviour
3.1. Explicit mood and self-esteem
3.2. Implicit indicators of changes in mood and attitudes
3.3. Self-evaluation and self-concept
3.4. Physical strength and pain tolerance
3.5. Social behaviour
3.6. Abstract, logical and creative thinking
4. Postural feedback effects are context dependent
5. Disentangling true from false effects and possible reasons for the replicability debate
Chapter 3 Power posture effects on social perception and behaviour: The research questions and methodological approach of the present thesis
1. Objectives and research questions
2. General methodological approach
2.1. Posture manipulation
2.2. Faces as social signals
2.3. Varying the focus of attention
Chapter 4 Does your body affect what you see? No power posture effects on explicit recognition of threat-related facial expressions
Chapter 5 Assessing physiological mechanisms: Repeatedly adopting power postures does not affect hormonal correlates of dominance and affiliative behaviour
Chapter 6 Does your body affect your mental images? Facing a controversy: Assessing the robustness of power posture effects on mental representations of faces
Chapter 7 Does your body affect your actions? Power posture effects on approach and avoidance actions under social threat
Chapter 8 Do power postures influence social perception and behaviour?
1. Summary of findings
2. Potential determinants of posture effects on social perception and behaviour
2.1. Posture effects may only occur when social signals are processed implicitly or when faces are entirely unattended
2.2. Posture effects may only occur in the presence of actual action opportunities
3. Setting my studies in context: A critical discussion of the research on power postures
3.1. Contextual meaning influences bodily feedback effects
3.2. Determinants of postural feedback effects neglected in the review and meta-analysis of research on power postures
4. A personal note on the current replicability debate
5. Final conclusions
Article: Stimulus and observer characteristics jointly determine the relevance of threatening facial expressions and their interaction with attention