Effect sizes are more accurate when calculated from a large sample. Thus, caution should be taken when interpreting effect sizes from small samples (<30 students). One should also take particular caution when the sample group has outliers i.e. students with exceptionally high or low scores. When calculating an effect size, it is best to compare your results including and excluding outliers to see if it makes a difference. When comparing pre-test and post-test scores, it is more useful to ensure that all students are tested and that scores from the same group of students are compared. Finally, one should always look at the context when interpreting effect sizes.

### Articles in this section

- Why does the Visible Learning research use effect sizes?
- Why do you use an effect size of d=0.40 as a cut-off point and basically ignore effect sizes lower than 0.40?
- What is the preferred timescale over which an effect size can be calculated?
- Is there a bias when using effect sizes in favor of lower achieving students?
- What caution should I take when calculating an effect size?
- Why are effect sizes used when conducting meta-analysis?
- Why can an effect size of 0.40 be gained in a shorter timeframe?
- Can effect sizes be added (or averaged)?
- How accurate are the conclusions drawn from meta-analysis?
- How can the variability associated with each influence be evaluated?