Item analysis using Rasch models confirms that the Danish versions of the DISABKIDS® chronic-generic and diabetes-specific modules are valid and reliable

Table 2 Test-of-fit for two models using conditional likelihood ratio tests and comparing estimates of item thresholds in groups defined by different test criteria

Test criterion	A: Rasch model–six items			B: Graphical log linear Rasch model^a–five items
Test criterion	CLR	Df	P	CLR	df	p
Low and high score groups	47.6	22	.001	34.3	33	.407
HbA1c	54.5	44	.133	90.0	66	.026
Treatment	42.3	22	.006	49.8	33	.030
Age	68.7	44	.010	65.3	44	.020
Sex	32.3	22	.072	36.6	33	.306
Inclusion or follow-up	22.9	22	.405	35.5	33	.350

The pure Rasch model is rejected because of local dependency between items 1 and 3 and because it operates differently for different age groups and treatments; however, it is improved when item 6 is excluded and GLLRM is applied allowing for local dependency between items 1 and 3 and DIF in relation to age
^aThe model assumes that items 1 and 3 are locally dependent and that item five is affected by DIF depending on age

ISSN: 1477-7525