Item analysis of the Eating Assessment Tool (EAT-10) by the Rasch model: a secondary analysis of cross-sectional survey data obtained among community-dwelling elders

Table 5 Fit statistics during removal of misfit items from J-EAT-10 with scoring structure 01122

Scale analysis	% Extr.	Fit residuals mean (SD)		Item-trait interaction		Reliability		Individual item-fit
Scale analysis	% Extr.	Item	Person	χ² (df)	P	PSI / α	Power of fit	Misfit items
4.a. Remove item 6	42%	−1.58 (3.96)	−0.36 (0.95)	382.42 (36)	< 0.001	0.61 / 0.79	Reasonable	Item 1 / FR = 5.5^a,b Item 3 / FR = −4.3^a,b Item 4 / FR = −6.2^a,b Item 8 / FR = − 3.2^a,b Item 10 / FR = −6.4^a,b Item 2,5,7,9 ^a,b
4.b. Remove item 10	42%	−1.28 (3.53)	− 0.34 (0.93)	299.55 (24)	< 0.001	0.53 / 0.74	Reasonable	Item 1 / FR = 4.6^a,b Item 3 / FR = − 4.5^a,b Item 4 / FR = − 6.2^a,b Item 8 / FR = − 3.1^a,b Item 2,5,7,9^a,b
4.c. Remove item 4	43%	−1.02 (2.85)	−0.33 (0.87)	230.68 (21)	< 0.001	0.40 / 0.66	Low	Item 1/ FR = 3.5^a,b Item 3 / FR = − 4.9^a,b Item 8 / FR = − 2.6^a,b Item 2,5,7 ^a,b
4.d. Remove item 3	43%	−0.69 (2.15)	− 0.32 (0.85)	196.29 (18)	< 0.001	0.24 / 0.55	Too low	Item 7 / FR = − 2.8 ^a,b Item 8 / FR = − 2.6 ^a,b Item 1,2,5 ^a,b
4.e. Remove item 7	45%	−0.39 (1.63)	− 0.28 (0.75)	163.76 (10)	< 0.001	0.02 / 0.38	Too low	Item 1,2,5,8 ^a,b
*Optimal values < 15%*		*0 (< 1.00)*	*0 (< 1.00)*		*> 0.05*	*> 0.70*		*FR < ±2.5, or non-significant fit statistics*

In analyses 4.c. to 4.e, item 9 displayed satisfactory fit
J-EAT-10 Japanese version of the Eating Assessment Tool, % Extr. percentage of extreme scores, SD standard deviation, χ² Chi-square, df degrees of freedom, PSI Person Separation Index, α Cronbach’s alpha, CI confidence interval
^aSignificant χ² (Bonferroni adjusted)
^bSignificant F statistics (Bonferroni adjusted)

ISSN: 1477-7525