One-Way ANOVA, Nonverbal Displays Affect Neuroendocrine Levels

#1
Hi,

So im doing a critique review on this paper:

Power Posing: Brief Nonverbal Displays Affect Neuroendocrine Levels and Risk Tolerance by Dana R. Carney et al.

Anyway, the experiment they are conducting involves two groups of individuals high; power posers and low-power posers. Each individual is measured for testosterone and cortisol levels before they pose and after they pose. The results state that they have used one-way analysis of variance to examine postmanipulation hormones and baseline hormones.

From my little knowledge of ANOVA (sorry):

The one-way Analysis of Variance (ANOVA) is used with one categorical independent variable and one continuous variable. The independent variable can consist of any number of groups (levels).

So from the above, is the categorical independent variable the poses and the continuous variable the level of hormone?

Im just getting confused on where the before and after hormon levels come into play. Is this the two levels?

Thanks for any help.

Arj
 
#2
I couldn't understand the meaning of "power posing" and that when any manipulation had been happened. From your post, I could understand that they have manipulated something in subjects (the experiment for example) and then have tested their hormone levels, right? (and also they have tested their hormone levels before manipulation?)

If this is the case, a repeated-measures ANOVA should have been used in the first place.

Im just getting confused on where the before and after hormon levels come into play. Is this the two levels?
They should have performed a repeated-measures ANOVA for that purpose. Have they done so?
 
#3
Thanks for the reply victorxstc,

Power poses are to do with body language. So in the experiment 23 subjects were grouped into low power pose and 23 subjects as high power pose (ive attached two pictures for what low and high power means). After they have done these poses/body actions for 1 minutes there saliva was tested for hormone levels. Also there saliva was tested before the one minute as well.

Yes I have to agree with you with the idea that maybe one-way anova was not meant to be used here. Can you please explain to me why repeated measure should have been used? They only did the one way ANOVA - and its a published paper! I guess I have more to talk about now.

thanks
 

Karabiner

TS Contributor
#4
http://faculty.haas.berkeley.edu/dana_carney/power.poses.PS.2010.pdf
Unfortunately, descriptions of the statistical analyses in
that article tend to be unprecise or even a bit misleading.
Since they have only 2 groups, they probably would not use
a oneway ANOVA but a t-test (these are equivalent, but
anyway, the usual test would be t-test). So, what did they
actually do?

p. 1366:
"All hormone analyses examined changes in hormones
observed at Time 2, controlling for Time 1." Which seems
to say, they calculated differences between Time2-Time1,
and then analysed them using ANOVA with (again) Time1
measurements as covariate, which looks silly (and wasn't
actually what they've done, I suppose).

"One-way analyses of variance examined the effect of power
pose on postmanipulation hormones (Time 2), controlling for
baseline hormones (Time 1). " So they actually did NOT perform
oneway ANOVA, but an analysis of covariance with baseline
measurement as covariate. Which is an alternative to repeated
measures ANOVA, at least in randomized trials.

Reviewers of "Psychological Science" should have asked the
authors to properly describe their analyses, but seemingly
that journal doesn't mind to publish papers of limited
methodological quality http://andrewgelman.com/2013/05/29/another-one-of-those-psychological-science-papers/

With kind regards

K.
 
#5
Thanks a lot for the reply Karabiner.

Do you mind sharing you thoughts on why it is a 'analysis of covariance with baseline
measurement as covariate' instead, im not really familiar with this type of test.

Thanks
Arj
 

Karabiner

TS Contributor
#6
Because they wrote "controlling for..." and what they controlled
for were the Time 1 (baseline) measurements.

With kind regards

K.