MerQuR+Statistics Glossary

Statistics Glossary → Cohen's d for paired data (d_z, d_av, d_rm)

Cohen's d for paired data (d_z, d_av, d_rm)concept

Effect sizes that standardize the mean difference between two measurements on the same people, differing in which standard deviation goes in the denominator.

d_z divides the mean difference by the standard deviation of the difference scores and equals t / √n; it is tied directly to the paired t statistic and is the one used in power analysis. However, as the correlation between the two measurements rises, the standard deviation of the differences shrinks, so d_z grows and cannot be compared directly with d values from independent groups.

d_av divides the mean difference by the average of the two measurements' standard deviations, while d_rm takes the correlation between measurements (r) into account: d_rm = d_z · √(2(1 - r)). These two are more comparable with an independent-groups d. Which one is used must always be reported, because on the same data their values can differ markedly.

An everyday example

To assess a reading course, the reading speed of 30 students is measured before and after. The average gain is 10 words per minute; the standard deviation of the differences is 8 (d_z = 1.25), while the standard deviation of each measurement is about 20 (d_av = 0.50). Because every student improved by a similar amount, d_z is large, but relative to the natural differences between people the gain is moderate.

In MerQur

The 'd_av', 'd_rm' and 'd_z CI, noncentral t' checkboxes of the Paired t-Test; the paired design of the Permutation Test and the 'Paired d' option of the Effect Size Calculator also report these measures.

Analyses where this concept appears

Paired t-TestPermutation TestEffect Size Calculator

Related concepts

Cohen's d (effect size)Hedges' g