A preference-based item response theory model to measure health: concept and mathematics of the multi-attribute preference response model

Table 4 Comparison of sum score, health-state estimates based on Rasch (holistic MAPR) model, MAPR model (LLTM). The absolute differences in outcome between the Rasch model (holistic MAPR) and the LLTM (MAPR) model are due to scaling and should be ignored

EQ-5D-3 L health state^a	Sum score	Rasch (Holistic MAPR) model	MAPR (LLTM) model
11111	2	^b	0.000
11211	57	4.761	−0.548
21111	60	4.555	−0.274
11121	63	4.358	−0.752
11112	76	3.575	− 1.527
12111	79	3.406	−1.626
11113	122	0.956	−3.006
11131	129	0.482	−3.548
11113	130	0.410	−3.352
13311	142	−0.605	−4.306
22222	142	−0.605	−4.728
32211	151	−1.733	−5.083
11133	152	−1.903	−6.900
32313	156	−2.780	−9.366
33323	157	−3.087	−12.045
23232	157	−3.087	−9.451
32223	159	−3.960	−9.187
33333	160	−4.742	−14.841

^aCode is representing the five attributes, each with three levels, of the EQ-5D-3 L instrument
^bNo estimate is obtained since the data matrix is ill-conditioned when this state is included

ISSN: 1471-2288