Methodologic training appears to aggravate or induce cognitive and motivational biases in statistical inferences. The extent of the problem is exhibited by persistent misinterpretations of "nonsignificant" hypothesis tests and confidence intervals as proving null hypotheses. Teaching correct definitions seems to have been insufficient to address this problem, and it can be argued that concepts from cognitive psychology are needed to reduce statistics abuse. Further improvements may include developing and teaching of correct interpretations for inferential statistics that do not rely on probability for their understanding. This session will present and discuss proposals for basic statistics based on these ideas, including coverage of biases implicit in conventional methodology and interpretations, and information-theoretic interpretation of inferential statistics.