Skip to main content

Table 4 Coefficients and indices of the goodness-of-fit of VAS and TTO models

From: Similar responses to EQ-5D-3L by two elicitation methods: visual analogue scale and time trade-off

 

Model 1 (VAS)

Model 2 (TTO)

Model 3 (VAS)

Model 4 (TTO)

11111

1

1

1

1

constant

− 0.0299

− 0.0141

–

–

MO2

− 0.1154

− 0.1292

− 0.1158

− 0.1340

MO3

−0.2180

− 0.2058

− 0.2111

−0.2159

SC2

−0.1060

− 0.1329

− 0.1064

−0.1377

SC3

−0.2172

−0.2220

− 0.2103

−0.2321

UA2

−0.0879

−0.1170

− 0.0884

−0.1218

UA3

−0.2413

−0.2103

− 0.2345

−0.2205

PD2

−0.0767

−0.0673

− 0.0772

−0.0721

PD3

−0.2553

−0.1967

− 0.2484

−0.2068

AD2

−0.0646

−0.1096

− 0.0650

−0.1145

AD3

−0.2087

−0.1893

− 0.2018

−0.1994

N3

0.0439

−0.0318

–

–

Adj. R-square

0.9409

0.9499

0.9550

0.9664

AIC

− 5285.48

− 6918.29

− 5252.47

− 6876.14

BIC

− 5192.57

− 6825.38

− 5172.83

− 6796.50

MAE

0.0304

0.0269

0.0327

0.0310

# > 0.05

4

1

4

4

# > 0.1

0

0

0

0

r

0.9847

0.9879

0.9813

0.9867

Logical error num.

0

0

0

0

  1. All models and regression coefficients were significant (P < 0.05); Adj. R-square adjusted R-square, AIC Akaike information criterion, BIC Bayesian information criterion, MAE mean absolute error between observed mean and predicted value # > 0.05, number of MAE > 0.05 out of 18 states; # > 0.1, number of MAE > 0.1 out of 18 states, r,correlation coefficient between observed means and predicted values, Logical error num., number of inconsistencies among all predicted health state values