Table 1.

Nonparametric rescaling of quality-of-life (QoL) variables through the use of anchoring vignettes

Observed order	Consistent with expected order	New variable Q possible responses
X > V1 > V2 > V3	Ordered	7
X = V1 > V2 > V3	Ordered	6
V1 > X > V2 > V3	Ordered	5
V1 > X = V2 > V3	Ordered	4
V1 > V2 > X > V3	Ordered	3
V1 > V2 > X = V3	Ordered	2
V1 > V2 > V3 > X	Ordered	1
X > V1 > V2 = V3	Tied	7
X > V1 = V2 = V3	Tied	7
X > V1 = V2 > V3	Tied	7
X = V1 > V2 = V3	Tied	6
X = V1 = V2 > V3	Tied	3, 4, 5, 6
X = V1 = V2 = V3	Tied	2, 3, 4, 5, 6
V1 > X > V2 = V3	Tied	3, 4, 5
V1 > X = V2 = V3	Tied	2, 3, 4
V1 = V2 > X > V3	Tied	3
V1 = V2 > X = V3	Tied	2
V1 = V2 > V3 > X	Tied	1
V1 = V2 = V3 > X	Tied	1
V1 > V2 = V3 > X	Tied	1

Vignette responses are used to determine individual thresholds. Rescaling of the QoL variables creates a new variable, free from scale bias caused by differences in rating behaviour