Tipultech logo


Richard Chambers

Calculations of Effect Size

Due to the relatively small sample size of this project (N = 40), statistical significance testing lacks power (Cohen, 1992). While significant results indicate strong effects in cases where there is reduced power, statistical significance testing may fail to detect meaningful effects. (It is important to distinguish here between statistical significance and meaningfulness in a broader sense.) Accordingly, measures of effect size will be reported for all results, along with traditional measures of statistical significance. Table 1 represents interpretations of effect sizes, consistent with Cohen (1992), for each statistical procedure employed in the following analysis.

Table 1

Critical Values for Interpretation of Effect Sizes for Independent Samples t-tests, Analysis of Variance (ANOVA) Procedures, and Correlations

Statistical Procedure Effect Size Measure Effect Size

Small Medium Large
Independent Samples t-test Cohen's d .2 .5 .8
ANOVA Partial Eta Squared, ╬À2 .1 .25 .4
Correlations Pearson's r .1 .3 .5

<< >>