Malte Friese, David D. Loschelder, Karolin Gieseler, et al., “Is Ego
Depletion Real? An Analysis of Arguments.” Personality and Social
Psychology Review, published online, March 29, 2018.
• The ego depletion hypothesis maintains that self-control is a construct that applies across (essentially?) all domains…
• …and the exertion of self-control increases the chance of a failure of self-control in a subsequent (temporally proximate) task.
• The standard experimental test for ego depletion is to conduct two consecutive self-control tasks. Ego depletion holds that folks who were induced to exert more control on the first task will show relatively low self-control on the second task.
• The ego depletion literature frequently goes beyond the “depletion” being induced by self-control; any effortful behavior, including IKEA shopping, might lead to a depleted state and lessened self-control.
• Some meta-analyses of the ego depletion literature conclude that the average effect might be zero; a preregistered large-scale replication of a previous ego depletion study also found a zero average effect.
• Standard depletion-inducing exercises include variations on “e-crossing” and the Stroop test. Do these manipulations really work?
• Journals traditionally have not been as interested in publishing null results than publishing articles that find statistically significant results. Researchers presumably respond to this situation by not bothering to write up and submit experiments that produce null results. Are the hundreds of published articles that find ego depletion matched or overmatched by countless unpublished studies that found no impaired self-control?
• Another way that researchers might respond to the (perceived?) difficulty in publishing null results is to ensure that they do not get null results. One way to do that is to run many different empirical specifications (of a regression, say), but only report the “best” one. If you do this, the usual tests for statistical significance are meaningless. (See Shiny Apps p-hacker.)
• We don’t see many studies with reverse depletion effects, and if the truth were that the effect is null, we might(?) see as many negative as positive studies.
• The unpublished studies might have to be legion to completely “offset” the many positive studies: that is, the truth might well be a positive effect (but of what size?)
• Don’t we see evidence of ego depletion in everyday life?
• The solution lies in better science! Pre-register studies; increase sample sizes; use improved controls; report all results; collaborate with your intellectual opponents; develop theoretical understandings; check the quality of manipulations.
• Friese et al. suggest that the burden of proof on the existence of ego depletion is now on the proponents of the phenomenon.
• The ego depletion hypothesis maintains that self-control is a construct that applies across (essentially?) all domains…
• …and the exertion of self-control increases the chance of a failure of self-control in a subsequent (temporally proximate) task.
• The standard experimental test for ego depletion is to conduct two consecutive self-control tasks. Ego depletion holds that folks who were induced to exert more control on the first task will show relatively low self-control on the second task.
• The ego depletion literature frequently goes beyond the “depletion” being induced by self-control; any effortful behavior, including IKEA shopping, might lead to a depleted state and lessened self-control.
• Some meta-analyses of the ego depletion literature conclude that the average effect might be zero; a preregistered large-scale replication of a previous ego depletion study also found a zero average effect.
• Standard depletion-inducing exercises include variations on “e-crossing” and the Stroop test. Do these manipulations really work?
• Journals traditionally have not been as interested in publishing null results than publishing articles that find statistically significant results. Researchers presumably respond to this situation by not bothering to write up and submit experiments that produce null results. Are the hundreds of published articles that find ego depletion matched or overmatched by countless unpublished studies that found no impaired self-control?
• Another way that researchers might respond to the (perceived?) difficulty in publishing null results is to ensure that they do not get null results. One way to do that is to run many different empirical specifications (of a regression, say), but only report the “best” one. If you do this, the usual tests for statistical significance are meaningless. (See Shiny Apps p-hacker.)
• We don’t see many studies with reverse depletion effects, and if the truth were that the effect is null, we might(?) see as many negative as positive studies.
• The unpublished studies might have to be legion to completely “offset” the many positive studies: that is, the truth might well be a positive effect (but of what size?)
• Don’t we see evidence of ego depletion in everyday life?
• The solution lies in better science! Pre-register studies; increase sample sizes; use improved controls; report all results; collaborate with your intellectual opponents; develop theoretical understandings; check the quality of manipulations.
• Friese et al. suggest that the burden of proof on the existence of ego depletion is now on the proponents of the phenomenon.
No comments:
Post a Comment