A note on the use of confidence intervals in psychiatric research.

The emphasis on hypothesis testing, which characterizes the majority of medical research, is often inappropriate because it focuses on statistical significance (the viability of the nil hypothesis) rather than substantive significance (the magnitude of the treatment effect). An alternative approach, the reporting of a treatment effect with the corresponding confidence interval, may better reflect the researcher's intent since these are the issues that are relevant to an informed clinical decision. Additionally, results reported in this format are less likely to be misinterpreted than results presented as p values since the precision with which the effect size is estimated would be reported separately from the effect size itself.