Table 1.

Testing and statistical methods of psychometric testing and diagnostic accuracy test

Psychometric Properties	Methods of Testing	Statistical Method and Cut-Off Standard	Testing Samples
Reliability
Internal consistency	Cronbach’s method	Cronbach’s α statistic, > 0.7 = satisfactory	All 350 older people
Stability	Two-week test-retest reliability	Intra-class correlation coefficient (ICC), >  0.75 = satisfactory	A subgroup of at least 50 older people (Giraudeau & Mary, 2001)
Validity
Criterion-related validity	Concurrent validity: correlating GFI-C with the Fried’s frailty phenotype	Pearson moment–product correlation coefficient, r ≥ 0.7 & < 0.9 = satisfactory	All 350 older people
Construct validity
1. Known-groups method	Comparing the GFI-C of older people in the community and long term care facility	1. Independent sample t-test, significant result = satisfactory	All 350 older people
2. Hypothesis testing:	Correlating the frailty (GFI-C) with cognitive level (AMT) and physical ability (SBI)	2. Pearson moment-product correlation coefficient, r > 0.5 = satisfactory	All 350 older people
3. Factor analysis		3. Confirmatory factor analysis χ2/df < 5.0, TLI > 0.90, CFI > 0.90, GFI > 0.90, RMR < 0.05, RMSEA ≤ 0.08.	All 350 older people
Diagnostic accuracy test
Sensitivity and specificity analysis	Comparing GFI-C results with the Fried’s frailty phenotype results	The receiver-operating characteristic (ROC) curve, sensitivity and specificity > 0.70	All 350 older people
Discriminative properties of the diagnostic accuracy		The area under the curve (AUC), AUC > 0.70

GFI-C Groningen Frailty Indicator – Chinese, AMT Abbreviated Mental Test, SBI Simplified Barthel Index, TLI Tucker–Lewis Index, CFI Comparative Fit Index, GFI Goodness-of-fit Index, RMR Root Mean Square, RMSEA Root Mean Square Error of Approximation