Work out test retest reliability instantly with clear inputs, formula shown and shareable results.
Test-retest reliability is the correlation between two administrations, and it converts into a standard error of measurement of sd times sqrt(1 - r). At r = 0.82 with a 12 point standard deviation the SEM is 5.09, so a change must exceed about 14.1 points to be distinguishable from measurement noise at 95% confidence.
Standard error of measurement
SEM = sd x sqrt(1 - reliability)
Minimal detectable change
MDC = 1.96 x sqrt(2) x SEM
Long enough to prevent memory of the answers, short enough that the trait has not genuinely changed - usually two to four weeks.
Because two measurements each carry error, so their difference has sqrt(2) times the error of one, and the 95% criterion adds the 1.96 factor.