Page 8 - IJPS-8-1

International Journal of
Population Studies Validity and reliability of Mini-Mental State Examination in older Chinese

MMSE is nowadays a widely used test of cognitive function the number of factors for the MMSE inventory and
among older adults. Its usability and applicability were well the main statistics used was the percentage of variance
noted in social and medical research as a short screening explained (e.g., Shigemori, Ohgi, Okuyama, et al., 2010).
tool, especially for older adults at risk of mild cognitive This statistic explains the level of variances that could be
impairment (Folstein, Folstein, and McHugh, 1975; extracted from an MMSE inventory, however, it does not
Lezak, Howieson, and Loring, 2004), screening the risk point out whether the number of dimensions is statistically
of dementia and Alzheimer’s disease (Burke, Grudzien, and theoretically appropriate. Choosing a higher level of
Burgess, et al., 2021; Mitchell, 2009; Arevalo-Rodriguez, variance explained will end in a higher dimension and
Smailagic, Roqué-Figuls, et al., 2021; Shigemori, Ohgi, vice versa. This statistic easily allows an arbitrary decision
Okuyama, et al., 2010). in determining the dimension. Theoretically, a prior
expectation about the dimensionality should be decided
The MMSE specifies a list of cognitive domain functions
as an inventory that contains items to measure them which before starting a study such as recall and visuospatial
capability which are most probably two separate factors.
include orientation, registration, attention, calculation, Using the appropriate and developed statistical procedures
spelling, recall, delayed recall, naming, repetition, verbal and that have been established in the measurement literature to
written comprehension, and visuospatial capability (Davey determine the dimensionality of MMSE were also absent
and Jamieson, 2004; Folstein, Folstein, and McHugh, 1975; from the MMSE validation literature. These developments
Reilly, Challis, Burns et al., 2004; Shulman, Herrmann, include the Velicer’s MAP criteria (Lim, Li, Xie, et al., 2019;
Brodaty, et al., 2006). The past studies that investigated the Velicer and Jackson, 1990), the Horn’s parallel analysis
factorial structure of MMSE remained unsettled about its (Dinno, 2009; Garrodp, Abad, and Ponsoda, 2013; Hayton,
dimensionality. These studies did not show the alignment Allen, and Scarpello, 2004; Horn, 1965), the very simple
between the domain functions to the MMSE items about the structure (VSS) criterion (Revelle and Rocklin, 1979),
dimensionality it is supposed to form and normally ended various information criteria, and approaches such as BIC
up with a lower dimension, not in line with the theoretically and sample adjusted BIC (Schwartz, 1978; Sclove, 1987).
expected dimension (Folstein, Folstein, and McHugh, 1975;
Park, Kwon, Jung, et al., 2012; Shigemori, Ohgi, Okuyama, EFA to establish the number of dimensions is
et al., 2010; Tinklenberg, Brooks, Tanke, et al., 1990). The first the preliminary step in a validation process. Further
study on MMSE factorial structure in the literature indicated incorporating confirmatory factor analysis (CFA) and
a two-factor structure with the first factor including attention/ the factorial invariance test into the validation procedure
concentration, language, and constructional praxis, and the were already established in the fields of measurement,
second comprising time-space orientation and delayed recall psychology, and education (Bollen and Lennox, 1991;
(Fillenbaum, Heyman, Wilkinson, et al., 1987). Three years Brown, 2006; Liau, Chow, Tan, et al., 2010; Lane, Anderson,
later, similar results of a two-factor solution were established Ponce, et al., 2012) and acknowledged as part of the
with the first factor including writing, naming, immediate validation process. That is, after EFA, CFA should be carried
memory, reading a sentence, and verbal comprehension, and out to confirm the structure of an inventory. Providing the
the second factor including constructional praxis, delayed fit of a hypothesized CFA model and comparing it with
recall, temporal orientation, attention/concentration, and competing CFA models are part of the validation procedure
spatial orientation (Tinklenberg, Brooks, Tanke, et al., 1990). to ensure that the best fitted CFA is chosen (Brown, 2006;
The non-alignment of the MMSE items to the cognitive Jackson, Gillaspy, and Purc-Steogebsib, 2009; Liau, Chow,
functions was clearly shown by these two studies that the Tan, et al., 2010). The factorial invariance further confirms
MMSE items within the two factors differed. The number of if the chosen CFA model is applicable in various controlled
dimensions increased in the later year publication to three conditions. The reliability and validity of measurements
factors (Shigemori, Ohgi, Okuyama, et al., 2010). While on the health of older adults were constantly a concern
(Gu, 2005). Due to the major health disparity between
the number of factors was not a straightforward solution young-old (aged 65 – 79 years old) and oldest-old (aged
empirically, the number of dimensions also differed within over 80 years old), it is desirable to check the factorial
a study between the control and experimental groups (Baek, invariance of these two major age groups. The factorial
Kim, Park, et al., 2016), and some ignored the determination invariance testing was carried out in this study to make
of dimensionality procedure and directly used the domain sure the MMSE psychometric properties for both the
MMSE summated subscores (e.g., Park, Kwon, Jung, young-old and old-old age groups were invariant. After
et al., 2012).
the best fitted confirmatory factor model was chosen, four
The exploratory factor analysis (EFA) was the main factorial invariance tests were carried out to ensure there
approach in the MMSE validation literature to determine is no inequality in their factor structure and form despite

Volume 8 Issue 1 (2022) 2 https://doi.org/10.36922/ijps.v8i1.1285