Skip to main content

Table 1 Variance components for the p x r x c design explained

From: Clinical assessment of transthoracic echocardiography skills: a generalizability study

Variance component

Explanation

σ2p

The variance in scores attributable to real differences in residents’ performances on TTE. This is known as the ‘universe score’ variance in GT. The equivalent in CTT is the ‘true score’ variance, σ2τ.

σ2r

The variance in scores attributable to rater differences in rating, e.g. differences in knowledge, skills and attitudes (e.g. ‘hawk’ or ‘dove’ attitudes) of raters.

σ2c

The variance in scores attributable to the case/patient. Some patients/conditions are easier to scan than others.

σ2pr

The variance in scores attributable to the interaction or ‘chemistry’ between physician scanning style and rater.

σ2pc

The variance in scores attributable to the interaction between residents and patients/cases. Different residents may perform differently based on the specifics of the case/patient.

σ2rc

The variance in scores attributable to the interaction between the rater and cases/patients. Different raters may rate differently based on the specifics of the case/patient.

σ2prc,e

The residual, which includes interaction between all effects (p, r and c) plus any systematic error variance not identified, as well as random error (e).