Recommending a specific quantification strategy might be premature at this point as none have been substantially validated as measures of aggressive behavior. If one, or even a few, versions of the CRTT measure were established as most appropriate for use as measures of aggression for research, and these versions were actually used consistently by researchers, concerns about post-hoc "cherry-picking" of measure versions to support hypotheses predicting effects of stimuli on agression would be alleviated. Therefore, three clear recommendations can be derived from the current flexibility in the use of the CRTT:
1. Preregistration of Analysis Plan
When running a study that includes the Competitive Reaction Time Task as a measure of aggression, the analysis plan
, which includes the chosen quantification strategy, should be preregistered
before data are collected. There are multiple ways of achieving this, e.g. by publishing a time stamped IRB protocol on a personal repository, or by using services such as the Open Science Framework
. You can still present results gained with other quantification strategies (doing so might in fact help understanding which measures converge and which do not), but they need to be clearly marked as exploratory.
2. Preregistration of Decision Tree
It is possible that even with a preregistered protocol, data cannot be analyzed in the planned manner because of skew, kurtosis, or other problems that might require data transformation. While not every eventuality might be considered a priori, a preregistered analysis plan should be accompanied by a preregistered decision tree
outlining procedures that will be applied in case the data are affected by specific characteristics that would severely limit the planned analysis without prior transformation
3. Defense of Selected Quantification Strategy
In the preregistered analysis plan, be very specific why
you selected a quantification strategy (over another). Offer the reader a substantial justification
that makes sense theoretically and metrically and before having seen the data.
If these simple steps are adopted by laboratory aggression researchers, the next step is a validation of the quantification strategies most researchers agree on using in their preregistered analysis plans.