BMC Medical Research Methodology

Table 1 RUMM2030 and R Output

From: A comparison of the polytomous Rasch analysis output of RUMM2030 and R (ltm/eRm/TAM/lordif)

			Item Fit Residual^a		Person Fit Residual^a		Chi-Square^b		PSI^c		UNID T-Test^d
			Mean	SD	Mean	SD	Value (DF)	p	With	without	UNID T-Test^d
RUMM 2030	PAIN	Initial	−0.8398	1.7301	−0.4845	0.9770	24.391 (20)	0.4969	0.9218	0.9135	4.6800%
	PAIN	Final	−0.5336	1.1746	−0.4492	0.9127	18.5992 (20)	0.5480	0.8781	0.8644	2.3500%
	SPECIFIC	Initial	−0.6052	2.1366	−0.4538	1.0702	54.0898 (30)	0.0045	0.9220	0.9033	4.7800%
	SPECIFIC	Final	−0.6266	1.5595	−0.4497	0.9686	50.3341 (20)	0.0002	0.9016	0.8699	3.6800%
	USUAL	Initial	−0.4119	1.7207	−0.4544	0.9432	23.1212 (20)	0.2829	0.9208	0.8304	6.1600%
	USUAL	Final	−0.3182	1.8380	−0.4642	1.0076	32.1662 (20)	0.0416	0.9217	0.8888	5.0700%
			Item Fit Residual^a		Person Fit Residual^a				Cronbach’s Alpha^c		Martin-Loef^d
			Mean	SD	Mean	SD			Cronbach’s Alpha^c		Value (DF)	p
R	PAIN	Initial	1.63322	1.6195	−0.0447	1.89779			0.922		370.171(599)	1.0000
	PAIN	Final	1.7807	1.6219	0.0067	1.7558			0.872		239.851(339)	1.0000
	SPECIFIC	Initial	1.6092	0.80491	0.12174	1.37694			0.92		392.039(899)	1.0000
	SPECIFIC	Final	1.7810	1.0349	0.1273	1.3960			0.8980		246.659(399)	1.0000
	USUAL	Initial	2.73459	1.43031	0.27866	1.84451			0.921		216.992(399)	1.0000
	USUAL	Final	2.9494	1.9276	0.2744	2.0210			0.9200		ERROR	n/a

^a The fit residuals should have mean of 0 ± 2.5 and a standard deviation of 1 ± 2.5
^b The Chi-Square statistic should be small and statistically non-significant
^c A Person Separation Index (PSI) or Cronbach’s Alpha should be > 0.70 to be statistically reliable
^d Unidimensionality is present if the percentage of statistically significant t-tests is < 5% or if the result of the Martin-Loef test is not significant

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com