Issues arising from the use of effect sizes in analysing and reporting research.

Abstract

Effect sizes may be seen as an alternative - or supplement - to the use of statistical significance tests. A number of arguments for their use in this way are put forward. These include the potential of effect sizes to emphasise amounts, not just directions, to avoid inappropriate dichotomies, to reduce confusion over the meaning of 'significance', to draw attention to issues of statistical power, to lessen the 'file drawer' problem, to promote synthesis rather than disagreement and to allow the accumulation of knowledge through the combination and comparison of the results of different studies. However, the use and interpretation of effect sizes are not without problems, including questions of which type of measure to use, sensitivity to violations of assumptions and interpretations in cases of restricted range or variance, non-normality or unreliable measures.