Browse Teach Curate

Test-Retest Reliability of a Voice Measure

Reliability makes measurement error empirical: record the same people twice and see how much of the spread is real between-person difference rather than noise. This is the quantity that bounds every downstream statistic, including any heritability estimate, since a trait cannot correlate with anything — genotype included — beyond how reliably it was measured.

Questions this Concept answers

Why does low reliability place a ceiling on any correlation the measure can show with another variable?

jeremy Text

Test-Retest Reliability via Intraclass Correlation

Test-retest reliability asks how much of the variation between two administrations of the same measurement reflects real, stable differences between people rather than random measurement error. The r…