Don’t rely on fancy statistics

They can obscure the structure of evidence.

Remain skeptical of all statistical and mathematical modeling procedures that pre-process evidence while leaving little trace of its original structure.  

Start out close to the data by scrutinizing frequency distributions, means, standard deviations, how variables are actually measured, degree of measurement error, zero-order correlations, and sample composition.

24 people saved this idea

Save it with our free app: