@Article{HuangGalinskyGruenfeldGuillory2011,
author = {Huang, L. and Galinsky, A.D. and Gruenfeld, D.H. and Guillory, L.E.},
title = {Powerful Postures Versus Powerful Roles: Which Is the Proximate Correlate of Thought and Behavior?},
journal = {Psychological Science},
year = {2011},
volume = {22},
number = {1},
pages = {95-102}
}
The DOI for this article is: 10.1177/0956797610391912
Huang, Galinsky, Gruenfeld, & Guillory ran three experiments to test the interaction between posture and role on established correlates of power. Following prior research (French & Raven, 1959; Keltner, Gruenfeld, & Anderson, 2003; Magee & Galinsky, 2008) [p95], the authors define power as control over resources in a relationship. While a study by Hall et al. (2005) [p96] found that power vested in an individual by their hierarchical role is typically associated with having an expansive posture, posture and role can also diverge. As an example, the authors point to a New Yorker cartoon from December 5th, 2005, in which then-President George Bush adopts a constricted posture associated with having a lower role, and Vice President Dick Cheney adopts an expansive posture, which is associated with having a higher role [p96, Fig. 1]. The representation suggests that posture can provide more information about the dynamics of a relationship than actual role. The authors' hypothesis was as follows: posture may have a greater effect than role on two behaviours typically associated with being in a powerful role: tendency to take decisive action (Anderson & Berdahl, 2002; Galinsky, Gruenfeld, & Magee, 2003) [p96] and tendency to think abstractly (Magee, Milliken, & Lurie, 2010; Smith & Trope, 2006) [p96].
The authors manipulated two main variables in their experiments: posture, and role. The posture manipulation used for all three experiments was performed using a mock "marketing test on ergonomic chairs" [p97], in which participants were instructed to sit in either a constricted or expansive posture for several minutes. Role manipulation was performed in the first two experiments by telling the participants, as they held their pose, that they would be either the manager or the subordinate in a later task [p97]. In the final experiment, role manipulation was performed by having participants recall being in either a position of high or low power [p100].
The first experiment measured participants' implicit activation of power using a word-completion task, in which they were given seven word fragments to fill in with the first word that came to mind [p97]. If the participants provided a power-related word (such as "power," "lead," or "command" [p97]) it was considered an implicit activation of power. Afterwards, their explicit sense of power was measured with a short survey.
Participants in the expansive-posture condition generated a moderately larger amount of power-related words (F(1,73) = 7.08, p = .01, d = .60) [p97], but the main effect of role did no reach significance (F(1,73) = 0.78, p = .38, d = 0.17) [p98]. Role had a moderately large effect on participants' explicit sense of power (F(1,73) = 10.49, p < .01, d = 0.69) [p98], while posture had a somewhat moderate effect (F(1,73) = 5.46, p = .02, d = 0.48) [p98].
The second experiment measured two effects of power on behaviour: abstraction in thought (Magee et al., 2010; Smith & Trope, 2006) [p98] and action in behaviour (Galinsky et al., 2003; Magee et al., 2007) [p98]. The task for measuring action was based on a simulated blackjack game, in which the dealer has 10 points and the player has 16 points. If the player took a card, taking the risk of bringing their total above 21, it was coded as taking action. Abstraction was based on a variation of the Gestalt Completion Task (Bowers, Regehr, Balthazard, & Parker, 1990) [p98]. Participants were shown fragmented pictures and asked to identify the object contained within, requiring abstract thought on their part. Responses were scored on three measures: whether they had identified any images at all, whether any individual image was identified correctly, and incorrect guesses were rated by two blind judges on the level of abstraction [p98].
The effects of posture on action were statistically significant, but the effect size was small (χ2(1, N = 77) = 5.28, p = .02, φ = .25) [p98]. In addition, as mentioned in the second footnote found on page 101, the effects showed an strong influence of gender: male participants showed no statistically significant effect from the manipulation of posture on their tendency to act (χ2(1, N = 34) = 0.05, p > .82, φ = .04), while the manipulation had a moderate effect on female participants (χ2(1, N = 43) = 9.92, p = .01, φ = .48).
Posture did have a moderate effect on correctly identified objects in the abstraction task (F(1,73) = 3.96, p = .05, d = 0.45), as participants in the expansive posture condition correctly identified more objects than those in the constricted-posture condition. With regards to the participants' incorrect guesses, posture had a large effect on generation of superordinate categories (F(1,73) = 11.51, p < .01, d = 0.74) and a moderately large effect on generation of subordinate categories (F(1,73) = 7.30, p < .01, d = 0.64). Participants in the expansive-posture condition generated more superordinate categories in their incorrect guesses, while participants in the constricted-posture condition generated more subordinate categories [p99]. As in the previous experiment, role had a large effect on participants' explicit sense of power (F(1,72) = 11.36, p < .01, d = 0.76), but posture did not have a significant effect (F(1,72) = 1.97, p = .16, d = 0.28) [p99].
Based on prior research that recalling an experience of being in a high-power role can lead individuals to act as if they were presently in that role (Galinsky et al., 2003; Guinote, 2007; Smith & Trope, 2006) [p99], the third experiment used a role manipulation based on recall. Instead of being placed in a manager or subordinate role, participants were prompted to describe either being in a position of high or low power [p100]. In this experiment, the dependent variable was the participants' decisions in three simulated scenarios in which they could either take a decisive action (e.g., speaking first in a debate [p100]).
Posture had a significant effect on participants' likelihood to take action, with a moderately large effect size (F(1,53) = 5.41, p = .03, d = 0.63), with participants in the expansive-posture condition acting more frequently (M = 1.82, SD = 0.67) than participants in the constricted-posture condition (M = 1.41, SD = 0.63). However, the effect of recall did not reach statistical significance (F(1,53) = 0.39, p = .53, d = 0.21) [p100].