ArticlePDF Available

Abstract and Figures

Objective: The Mental Health Continuum – Short Form is a brief scale measuring positive human functioning. The study aimed to examine the factor structure and to explore the cross cultural utility of the MHC-SF using bifactor models and exploratory structural equation modelling (ESEM). Method: Using multigroup confirmatory analysis (MGCFA) we examined the measurement invariance of the MHC-SF in 38 countries (university students, N = 8,066; 61.73% women, mean age 21.55 years). Results: MGCFA supported the cross-cultural replicability of a bifactor structure and a metric level of invariance between student samples. The average proportion of variance explained by the general factor was high (ECV = .66), suggesting that the three aspects of mental health (emotional, social, and psychological well-being) can be treated as a single dimension of well-being. Conclusion: The metric level of invariance offers the possibility of comparing correlates and predictors of positive mental functioning across countries; however, the comparison of the levels of mental health across countries is not possible due to lack of scalar invariance. Our study has preliminary character and could serve as an initial assessment of the structure of the MHC-SF across different cultural settings. Further studies on general populations are required for extending our findings.
Content may be subject to copyright.
For Peer Review
The Mental Health Con
tinuum
-
Short Form: The structure
and application for cross-cultural studies – a 38 nation
study
Journal:
Journal of Clinical Psychology
Manuscript ID
JCLP-16-0234.R3
Wiley - Manuscript type:
Research Article
Keywords:
Mental Health Continuum-Short Form, measurement invariance, cross-
cultural study
John Wiley & Sons
Journal of Clinical Psychology
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Abstract
Objective: The Mental Health Continuum – Short Form is a brief scale measuring positive
human functioning. The study aimed to examine the factor structure and to explore the cross-
cultural utility of the MHC-SF using bifactor models and exploratory structural equation
modelling (ESEM). Method: Using multigroup confirmatory analysis (MGCFA) we examined
the measurement invariance of the MHC-SF in 38 countries (university students, N = 8,066;
61.73% women, mean age 21.55 years). Results: MGCFA supported the cross-cultural
replicability of a bifactor structure and a metric level of invariance between student samples. The
average proportion of variance explained by the general factor was high (ECV = .66), suggesting
that the three aspects of mental health (emotional, social, and psychological well-being) can be
treated as a single dimension of well-being. Conclusion: The metric level of invariance offers
the possibility of comparing correlates and predictors of positive mental functioning across
countries; however, the comparison of the levels of mental health across countries is not possible
due to lack of scalar invariance. Our study has preliminary character and could serve as an initial
assessment of the structure of the MHC-SF across different cultural settings. Further studies on
general populations are required for extending our findings.
Keywords: Mental Health Continuum-Short Form, measurement invariance, cross-
cultural study
Page 1 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
The Mental Health Continuum – Short Form: The Structure and Application for Cross-Cultural
Studies – a 38-Nation Study
Emerging adults are frequently exposed to the challenges of transition into adulthood
(low personal finances, entering the workplace, changes in personal relationships) (Arnett, 2000;
Roberts, Golding, Towell, Reid, & Woodford, 2000) and are at risk for various mental health
problems (Eisenberg, Gollust, Golberstein, & Hefner, 2007). The mental health of university
students is often found to be worse than that of the general population (Stock et al., 2008;
Mikolajczyk et al., 2008; Vaez, Kristenson, & Laflamme, 2004), perhaps due to the additional
challenges and risks facing them, such as increased financial worries, costs, and debt associated
with university, academic pressure, moving away from home, changes in sources of emotional
support, dealing with new environments, and increased exposure to drinking and drug-taking
culture. However, despite the variety of studies reporting low mental health scores among
students around the globe (see Boot, Donders, Vonk, & Meijman, 2009; Kurré, Scholl, Bullinger,
& Petersen-Ewert, 2011; Stewart-Brown et al., 2000; Vaez, Kristenson, & Laflamme, 2004), still
little research has been done regarding comparisons between different countries. This may be
due to the lack of a valid standardized measure which could be used in cross-cultural studies
aimed at assessing well-being and could consequently guide the assessment and mental health
promotion of university students across the globe.
The problem of mental health and its assessment is one of the broadest and most complex
issues in psychology (see Sirgy, 2012), given the many different sources and factors contributing
to well-being (Keyes, Ryff, & Shmotkin, 2002; Keyes, 1998; 2002; Ryff, 1989). Therefore, it
seems crucial to develop a single valid and reliable instrument that could be used to study,
assess, and promote the students’ mental health. To achieve these goals we focus on exploring
Page 2 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
the cross-cultural utility of a measure that assesses several theoretical domains of well-being: the
Mental Health Continuum – Short Form (MHC-SF; Keyes, 1998).
The MHC-SF is based on the concept of positive mental health proposed by Keyes (2002)
and is an abbreviated form of the 40-item MHC-LF (Keyes, 2002). It is an effort to integrate
hedonic and eudaimonic aspects of well-being. Specifically, the Mental Health Continuum is
regarded as a syndrome encompassing three broad aspects: emotional well-being (EWB: positive
emotions along with life satisfaction), social well-being (SWB, based on the definition offered
by Keyes, 1998, comprised of: social coherence, social acceptance, social actualization, social
contribution, and social integration), and psychological well-being (PWB, based on a model by
Ryff, 1989, including: self-acceptance, positive relationships with others, autonomy, purpose in
life, environmental mastery, and personal growth).
Comprising 14 items, the MHC-SF (Keyes et al., 2008) measures all three dimensions:
emotional, psychological, and social well-being. It can be used both for research purposes (as an
indicator of positive functioning of individual) and for diagnosis of the levels of positive
functioning (Keyes, 2002). The MHC-SF captures three categorical diagnoses: flourishing,
languishing, and moderate mental health. Flourishing is diagnosed when someone reports having
experienced at least one of the three hedonic well-being symptoms (items 1–3) and at least 6 of
the 11 positive functioning symptoms (items 4–14) “every day” or “almost every day” within the
past month. Languishing is diagnosed when someone reports having experienced at least one of
the three hedonic well-being symptoms and at least 6 of the 11 positive functioning symptoms
“never” or “once or twice” in the past month. Individuals who are neither “languishing” nor
“flourishing” are considered “moderately mentally healthy” (Keyes, 2002).
Page 3 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Currently, there are several versions of this scale in different languages (see Keyes, 1998,
and Karaś, Cieciuch, & Keyes, 2014, for review), including Korean (Young-Jin, 2014), Serbian
(Jovanović, 2016), Italian (Petrillo, Capone, Caso, & Keyes, 2015), and Polish (Karaś et al.,
2014). Therefore MHC-SF seems to be a perfect tool for cross-cultural research studies of well-
being among university students. These studies could focus on searching for the risk factors and
factors important for increasing the mental health, as well as on diagnosing the number of
languishing, flourishing, and moderately healthy individuals within different populations.
As the MHC-SF has been used in a number of countries, one could expect that this tool is
well-validated in different cultural contexts and there are no controversies and/or obstacles to
implementing it in the cross-cultural surveys. Nevertheless, despite the work of Keyes (1998)
that assumes a three-factor structure of the MHC-SF, some researchers suggest that the proposed
three-factor structure of the MHC-SF scale is problematic, both theoretically, as it fails to
provide the information needed to justify the calculation of a general well-being index (deBruin,
& duPlessis, 2015), and empirically, as it often produces only marginally acceptable fit indices
(Jovanovich, 2015).
To address these issues, two alternative, more flexible models, were proposed for MHC-
SF. First, a bifactor model, in which each items loads on a general factor (reflecting a common
construct underlying all the items) and on one of uncorrelated specific, or “group” factors (which
capture the content similarity of homogeneous groups of items forming the subscales). This
approach was suggested as particularly useful for composite models of subjective well-being,
because it allows to separate the general well-being dimension from specific factors related to
particular life domains or aspects of human functioning (Sirgy, 2012) and it has been applied
successfully with MHC-SF (Jovanovich, 2015). The advantage of this model is that it allows to
Page 4 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
separate the item variance associated with the general factor and specific factors and evaluate the
reliability of a general score and the discriminant validity of subscale scores using a range of
indices discussed below (Reise, 2012; seel also Chen, West, & Sousa, 2006, for a comparison of
correlated-factor and bifactor models).
Another analytic solution recently proposed for MHC-SF is Exploratory Structural
Equation Modeling (ESEM). Unlike conventional CFA models (termed “independent-cluster
model”, or ICM-CFA), ESEM models allow for non-zero cross-loadings, addressing the issue of
imperfect indicators. Advocates of this approach have argued that the assumption of the absence
of cross-loadings in the conventional CFA may be unrealistic, resulting in overestimation of
factor covariances; some of the MHC-SF items were indeed found to show statistically
significant cross-loadings (Joshanloo, Jose, & Klepikowski, 2016; Joshanloo & Jovanovic,
2016). However, ESEM models have some important drawbacks. First, they can be viewed as
more data-driven, because strong item loadings on non-target factors may affect the theoretical
interpretation of factors. Second, they include a much larger number of free parameters (i.e.,
loadings), compared to conventional CFA models, which may result in increased sample size
requirements and entail convergence difficulties. Finally, a ESEM model with three correlated
factors, which was previously applied for MHC-SF (Joshanloo, Jose, & Klepikowski, 2016;
Joshanloo & Jovanovic, 2016), may not be optimal due to the presence of a common factor,
whose variance could contribute to item cross-loadings, resulting in overestimation of factor
overlap. Theoretically, a bifactor ESEM model (Morin, Arens, & Marsh, 2016) should be more
relevant for MHC-SF, but may entail identification or convergence problems due to its
complexity.
Page 5 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Past cross-cultural studies using MHC-SF did not take advantage of bifactor or ESEM
approaches and have compared only a limited number of national samples (Joshanloo, Wissing,
Khumalo, & Lamers, 2013). Because the MHC-SF seems a promising brief measure of positive
mental functioning, the issues of structure and measurement invariance of MHC-SF need to be
studied in diverse cultural contexts to reveal the possibilities and limitations of this measure for
multicultural projects.
The Present Study
The aim of the present study is, firstly, to examine the structure of MHC-SF in different
cultural contexts, comparing the fit of the bifactor model and that of the three-factor solution
and, secondly, to examine the applicability of MHC-SF to cross-cultural studies using multi-
group CFA analyses with data from 38 countries. This aim is not only theoretical, but could
allow to address applied issues, i.e., the comparability of findings on mental health (in terms of
conceptual invariance, findings on predictors and correlates, and, finally mean scores) among
youth populations from different cultures.
Method
Sample and Procedure
The sample included 8,066 university students (61.7 % women), ranging in age from 16
to 50 (M = 21.55, SD = 4.37), originating from 38 countries (see Table 1 for details). The
students filled out the MHC-SF as part of a broader research project on entitlement and well-
being. In addition to the MHC-SF, the study included other measures of subjective well-being:
Personal Well-being Index (Cummins, Eckersley, Pallant, Van Vugt, & Misajon, 2003),
Satisfaction with Life Scale (Diener, Emmons, Larsen, & Griffin, 1985) Positive and Negative
Affect Schedule (Watson, Clark, & Tellegen, 1998), and two scales measuring attitudes:
Page 6 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Entitlement Attitudes Scale (Żemojtel-Piotrowska et al., 2015) and belief in life as zero-sum
game (Różycka-Tran, Boski, & Wojciszke, 2015). Data were collected in paper-pencil form and
also online (presented in Table 1) between March 2015 and March 2016. The students were
recruited at universities during their classes by the members of the research team and did not
receive any financial remuneration for participation. The students participated in the study
voluntarily and informed consent was obtained from all study participants. The registered data
was alphanumerically coded, ensuring anonymity. The study has been conducted according to
the principles expressed in the Declaration of Helsinki. All procedures were approved by each
participating University Ethics Committee.
The selection of participating countries aimed to reflect cultural diversity in the most
comprehensive way possible. In terms of cultural regions, we included countries representative
of all Huntington (1996) cultural groups (i.e., Western, Orthodox, Confucian, Japanese, Latin
American, Hindu, Buddhist, Islamic, African, and Sinic) and, in terms of religion, we had
countries representing all main world religions. In the current study, we included data from:
Europe (16), Asia (13), Africa (3), and Latin America (6). Former studies indicate the
importance of cultural, political, and economic factors related to subjective well-being. For
instance, subjective well-being is related to income inequalities (Berg & Veenhoven, 2010),
values (Sagiv & Schwartz, 2000; Welzel & Inglehart, 2010), and religion (Donahy, Lewis,
Schumaker, Akuomah-Boateng, Duze, & Sibiya, 1998). Therefore, our aim was to include
countries with different levels of affluence, cultural values, and religion in order to indicate
usefulness of the MHC-SF in measuring mental health as a multi-dimensional construct.
(Table 1 about here)
Measure
Page 7 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
The Mental Health Continuum – Short Form (MHC-SF; Keyes, 2013) comprises 14
items that represent various aspects of well-being (the items were chosen from the longer version
of this tool, as the most prototypical for each aspect of well-being). The response scale consists
of 6 points, which describe the frequency of experiencing various well-being symptoms during
the past month, ranging from 1 = never to 6 = every day. The MHC-SF allows two kinds of
assessments, of the level of well-being (and its three dimensions: social, psychological, and
emotional) and a categorical assessment of mental health status, with three categories:
flourishing (i.e., high levels of well-being), languishing (i.e., the absence of mental health), and
moderate mental health (located between these two extremes).
We used translation and back-translation procedure to obtain versions of the scale in
different languages. The resulting back-translated versions were discussed with the author of the
MHC-SF, Corey Keyes. We do not report the results of validation of the MHC-SF, as they would
go beyond the scope of the present paper. However, in different countries we have found a
consistent pattern of negative correlations of MHC-SF with revengefulness and belief in life as
zero-sum game, as well as positive correlations of MHC-SF with other scales measuring
subjective well-being.
Data Analysis
The analyses were conducted using SPSS 20 and Mplus 7.4. The robust MLM estimator
with Satorra-Bentler-scaled chi-square resulted in fewer convergence problems and inadmissible
solutions for the bifactor model, compared to the ML and MLR estimators. Unfortunately, the
MLM estimator in Mplus currently does not handle missing data. Because the percentage of
missing responses was quite small (0.28%) and the data were missing at random, we used
Page 8 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Expectation Maximization (EM) imputation in SPSS to impute the missing values, in order to
take advantage of the MLM estimator.
Preliminary confirmatory factor analyses. First, we performed a confirmatory factor
analysis (CFA) in each sample separately. These analyses were aimed at finding the best
measurement model of the MHC-SF to be used as a basis for cross-cultural comparison.
We identified the models by fixing the latent factor variances to 1 and freely estimating
the factor loadings. Given the known limitations of the chi-square test of overall model fit
(dependence on sample size making the results incomparable across samples and reliance on the
null hypothesis of exact overall fit which is too stringent to be informative in evaluating the
usefulness of a model: see West, Taylor, & Wu, 2012), we used practical fit indices (CFI,
RMSEA, and SRMR) to assess model fit. We followed the guidelines proposed by Hu and
Bentler (1999), i.e., the values of CFI close to .95 or above, RMSEA close to .06 or below,
SRMR close to .08 or below as indications of good fit, using these indices in combination
(Brown, 2015). In order to compare the fit of nested models in individual samples, we relied on
the scaled chi-square difference test (Satorra & Bentler, 2001).
Based on theory and previous findings (Jovanovich, 2015; Karaś, et al., 2014), we tested
four different CFA models of the MHC-SF: (1) a single-factor model, in which all 14 items load
on one underlying dimension of well-being; (2) a two-factor model with two correlated
dimensions of well-being – hedonic well-being (comprising EWB; items 1 through 3), and
eudaimonic well-being (comprising both SWB and PWB; items 4 through 14); (3) a three-factor
model with three correlated dimensions of well-being – hedonic well-being (items 1 to 3),
eudaimonic social well-being (items 4 to 8), and eudaimonic psychological well-being (items 9
to 14); and (4) a bifactor model (Reise, 2012), with a general factor and three uncorrelated
Page 9 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
“group factors”, capturing specific variance or hedonic, social, and psychological well-being. We
did not test the hierarchical model with a single second-order factor separately, because a
hierarchical solution with three first-order factors is mathematically equivalent to the correlated-
factor model.
The advantage of the bifactor model is that it makes it possible to separate the general and
specific variance. To evaluate the reliability of the general dimension and the subscales, we
calculated the omega coefficient (Reise, 2012), which is similar to the alpha, as it reflects the
proportion of total item variance explained by the model, with joint contribution of the general
well-being factor and group factors. To separate the effects of the general well-being factor and
those of the group factors, we calculated coefficients ω
H
and ω
S
(Reise, 2012), the former
reflecting the share of total variance explained by the general factor and the latter reflecting the
unique share of variance explained by each group factor (excluding the contribution of the
general factor). We also calculated the Explained Common Variance (ECV) coefficient (Reise,
Scheines, Widaman, & Haviland, 2013), which measures the relative strength of the general
factor to the group dimensions.
Measurement invariance analyses. The second aim of study was to evaluate the
measurement invariance of the MHC-SF and establish non-equivalent parameters using multi-
group bifactor CFA model. To evaluate the absolute model fit, we used the same criteria for
practical fit indices as described above. Because the chi-square difference test is known to be
overly sensitive in large samples (Chen, 2007), we relied on practical fit indices to compare the
nested models, using the CFI and RMSEA cutoff values of .010 and .015, respectively, as
indicators of pronounced difference in fit between the nested models (Chen, 2007; Cheung &
Rensvold, 2002). Because of the large number of parameter constraints tested, we only relied on
Page 10 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
modification indices significant at p < .05 with Bonferroni correction in order to prevent false
positives. We relaxed the parameter constraints sequentially (Yoon & Kim, 2014), one at a time,
after which the model was re-estimated.
There are three levels of measurement invariance that are most commonly used to
establish whether a measure is equivalent. Configural invariance indicates that the general factor
structure of the measure is the same across different groups. At this level, the construct is
measured by the same set of indicators in different samples. Metric invariance implies that the
factor loadings of items are similar across groups. At this level, the effects of correlates and/or
predictors of the measure may be compared across samples. Scalar invariance indicates that item
intercepts are equal across groups. At this level, mean scores may be compared between samples
(Davidov, Meuleman, Cieciuch, Schmidt, & Billiet, 2014). Scalar invariance is rarely found in
large cross-cultural comparisons (see Davidov et al., 2014), so we expected to find metric
invariance of the MHC-SF. To examine the structure of the scale and its cross-cultural
replicability, however, only configural invariance is required. Since most cross-cultural studies
focus on examining predictors and correlates of subjective well-being, the metric level of
invariance is sufficient.
ESEM analyses. We also performed single-group ESEM analyses based on a model with
three correlated factors and a bifactor model. However, because of complexity of this model,
which resulted in convergence issues, we could not use the ESEM model as a basis for
multigroup comparison and we present these results as supplementary findings.
Results
Preliminary Confirmatory Factor Analysis
Page 11 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Single-group analyses. The single-factor model did not fit the data well, with at least 2
out of 3 fit indices lying outside the acceptable ranges for all samples. Across the 38 countries,
the CFI ranged from .508 to .868 (M = .791, SD = .066), the RMSEA ranged from .079 to .144
(M = .112, SD = .015), and the SRMR ranged from .058 to .134 (M = .079, SD = .013). The two-
factor model (i.e., factors representing hedonic and eudaimonic well-being) showed a better fit,
with CFI ranging from .587 to .926 (M = .848, SD = .060), RMSEA ranging from .067 to .133
(M = .095, SD = .014), and SRMR ranging from .053 to .132 (M = .072, SD = .013). However,
based on the combination of fit indices, the fit was still unacceptable in all countries but one
(Ukraine).
The fit indices for the three-factor model and the bifactor model are shown in Table 2.
Based on the combination of indices, the three-factor model showed a good fit in 2 countries
(Ukraine and Uruguay) and acceptable fit in 14 countries (Azerbaijan, Belgium, Brazil, Czech
Republic, Estonia, Hungary, Indonesia, Japan, Kazakhstan, Malaysia, Portugal, Russia, South
Africa, and Vietnam). In most of the remaining cases, the fit was marginal. The correlations
between the factors were moderate to strong in all samples. The mean correlation between the
emotional and psychological well-being factors was .75, and social well-being was correlated at
.69 and .62 with psychological and emotional well-being, respectively.
The unrestricted bifactor model failed to converge in 10 out of 38 samples (Bulgaria,
Czech Republic, Hong Kong, Hungary, India, Latvia, Panama, South Africa, Vietnam, Uruguay)
due to a negative error variance. In 2 samples (Vietnam, South Africa) a proper solution could be
reached by adjusting the default starting values for group factor loadings. In five other samples
where the model converged one of the estimates of residual variances was negative, but not
significantly different from zero, suggesting normal sampling variation. An investigation of
Page 12 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
parameter estimates has shown that in many samples the loadings on one of the group factors
(typically, PWB) were generally quite low. Negative error variances are often found when
models with a relatively large number of free parameters and low empirical factor loadings are
tested in samples of modest size (Chen et al., 2001).
We explored two approaches to resolve this issue. Firstly, to improve the model
identification by ruling out inadmissible solutions, we introduced inequality constraints,
restricting the estimates of residual variances of observed variables to values above 0. As a
result, model convergence was obtained in all samples. The constrained bifactor model showed
good fit in 16 countries (Azerbaijan, Belgium, Brazil, Bulgaria, Czech Republic, Estonia,
Hungary, Indonesia, Japan, Latvia, Malaysia, Portugal, South Africa, Spain, Ukraine, and
Uruguay) and acceptable fit in all others, except Kenya and Iran, where the fit was marginal. The
investigation of modification indices revealed an unexplained error covariance of items 9 and 10
in the Kenyan sample. In Iran, we found no pronounced modification indices, but exploratory
factor analyses showed that items 4 to 8 failed to form a single dimension. To avoid the necessity
for model modifications, we opted to exclude these two samples from the multigroup model.
(Table 2 about here)
Secondly, to investigate the potential bias introduced by constraints, we simplified the
model by omitting the PWB group factor (Eid, 2000; Reise, 2012). The resulting incomplete
bifactor model (unconstrained) converged successfully in all samples. Predictably, the fit of the
incomplete bifactor model was significantly worse, compared to that of the full bifactor model
(scaled chi-square difference test, p < .05) in all but three samples (Hong Kong, Kazakhstan,
Malaysia). The differences in practical fit indices were quite small (average ∆CFI=-.025,
Page 13 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
∆RMSEA=.010, ∆SRMR=.008), generally favoring the full bifactor model. The fit indices for
this model are given in Supporting Information.
Bifactor structure analyses. Based on the bifactor models for each sample, we
calculated a set of indices to evaluate the reliability and dimensionality of the MHC-SF in each
sample. The results are shown in Table 3. The ω reliability coefficients, reflecting the proportion
of true score variance (with contribution of both the general factor and group factors), ranged
from .82 to .95 for the general well-being index and from .57 to .92 for the subscales, indicating
good reliability. The social well-being subscale showed somewhat lower reliability, compared to
the emotional and psychological well-being subscales.
The ω
H
coefficient, reflecting the proportion of total variance explained by the general
factor, ranged from .56 to .87, indicating a substantial contribution of the general factor. The
ECV index, reflecting the share of the general factor in the true score variance, ranged from .40
to .76 (M = .66). This suggests that, on average, two-thirds of the variance captured by the MHC-
SF is shared by the three scales, and only one-third is related specifically to emotional,
psychological, or social well-being. According to O’Connor’s (2014) recommendations, ECV
values above .70 indicate unidimensionality of scales. In Brazil, Colombia, Germany, Estonia,
Hong Kong, Iran, Kazakhstan, Malaysia, Poland, Russia, UK, and Vietnam, ECV exceeded this
value, suggesting that the general dimension of MHC-SF may be most relevant in these countries
as an indicator of overall mental health.
The residual reliability coefficients (ω
S
) reflect the proportion of true score variance of
each subscale excluding the contribution of the general factor. The psychological well-being
subscale reveals a comparatively small amount of unique variance (M = .12), indicating that the
variance it captures is mainly shared by all three subscales of MHC-SF. The emotional and social
Page 14 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
well-being subscales emerge as more distinct (M = .29 and .31, respectively), suggesting that
their associations with other variables may be different from those exhibited by the MHC-SF as a
whole.
The structural coefficients based on the incomplete bifactor model (provided in
Supporting Information) were convergent with those based on the full bifactor model, with ω
reliability in the .81-.94 range and ω
H
in the .55-.88 range. The ECV (ranging from .44 to .82, M
= .71) and the residual reliability coefficients for the two subscales modelled were marginally
higher (M = .33 and .36 for EWB and SWB, respectively), but led to the same substantive
conclusions.
These findings support the validity of the general index of the MHC-SF and the
discriminant validity of its individual subscales, particularly, EWB and SWB.
Measurement Invariance Analyses
We proceeded by investigating the measurement invariance of the MHC-SF based on the
full bifactor model for 36 countries (excluding Kenya and Iran). We failed to achieve
convergence of an unrestricted configural invariance model. To rule out inadmissible solutions,
we introduced 20 inequality constraints restricting residual variances of observed variables to
positive values. This allowed to obtain convergence of the configural invariance model, which
showed good fit. The metric and scalar invariance models converged successfully without any
constraints, suggesting that the model identification issues were due to a combination of model
complexity and modest size of individual samples.
The fit of the metric invariance model was acceptable. Using Bonferroni correction, we
established critical chi-square values to detect loading and intercept non-invariance (∆χ
2
= 16.46
based on N = 1008 for loadings and ∆χ
2
= 15.15 based on N = 504 for intercepts). We proceeded
Page 15 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
by searching for non-invariant loadings based on the metric invariance model. The complete list
of non-invariant parameters is given in Supplementary Information 1.
Only one loading revealed a strong non-invariance (∆χ
2
= 39.57), the loading of item 12
on the general well-being factor in Algeria, which was negative (λ = -.22). To prevent a negative
group factor variance in Algeria, we also relaxed the constraint for this loading on the
psychological well-being factor. The remaining four modification indices were marginal (∆χ
2
=
19 or below) and, when they were all addressed, the fit of the model did not change substantially
(∆CFI ≤ .001, ∆RMSEA ≤ .001), so we opted against including them into the model for the sake
of theoretical parsimony.
The non-invariance of intercepts was more pronounced. The fit indices of the model with
full scalar invariance were well outside the acceptable range. Based on modification indices, we
relaxed the equality constraints for 54 non-invariant intercepts (listed in Supplementary
Information 1). Although the target difference in practical fit indices was only reached for the
RMSEA, but not for the CFI (∆CFI = .024, ∆RMSEA = .007), the remaining modification
indices were all below the cut-off and exhibited no pronounced outliers.
The items tapping into social well-being turned out to be the most problematic, with 30
non-invariant intercepts (55.6%). The psychological well-being items were less biased, with 15
non-invariant intercepts (27.8%). Finally, emotional well-being items only revealed 9 instances
of intercept bias (16.7%), mainly confined to item 3 (“satisfied with life”, N = 6). After all the
relevant constraints were relaxed in the model, the resulting partial scalar invariance model
showed acceptable fit. There were no non-invariant intercepts in 10 countries (Chile, Colombia,
India, Kazakhstan, Nepal, Pakistan, Portugal, South Africa, UK, and Vietnam), suggesting that
MHC-SF data across these countries can be considered scalar-invariant.
Page 16 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
The parameters of the model were within acceptable ranges in all groups. The model-
based estimates of ω ranged from .84 to .96 (M = .93, SD = .02) for the general index and from
.67 to .93 (M = .84, SD = .05) for the subscales. The ω
H
values ranged from .73 to .85 (M = .82,
SD = .02). The ω
S
values ranged from .11 to .15 (M = .14, SD = .01) for hedonic well-being,
from .28 to .35 (M = .32, SD = .01) for social well-being, and from .14 to .19 (M = .15, SD = .01)
for psychological well-being. The ECV based on the multigroup model was .72. These bifactor
structure estimates based on the multi-group model were consistent with the results of single-
group analyses.
In our sample of countries, metric and scalar invariance were partially supported. The
comparison of practical fit indices between the nested models indicates that the non-invariance
of loadings is much less pronounced, compared to the non-invariance of intercepts. We only
found one strongly non-invariant factor loading (i.e., item 12 in Algeria), suggesting that metric
invariance can be assumed for all the other countries. These findings indicate that the effects
found using the MHC-SF can be safely compared across countries, but the comparison of mean
individual and group scores necessitates using latent factor scores based on the partial invariance
model.
We used the final partial invariance model to investigate the mean scores across
countries. We chose the Armenian group, whose scores were the closest to the grand mean, as
the reference group, setting its latent factor variances to 1 and latent means to 0. The results are
shown in Supplementary Information 2. We used a basic multilevel model to investigate the
associations of observed scores with latent score estimates based on the multigroup bifactor
model at the individual and group level. For the general factor, this association was very strong
at the individual level (r = .98), but moderate at the group level (r = .31). Similarly, the
Page 17 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
correlations of subscale scores with estimates of group factors were moderate to strong at the
individual level (.71, .68, and .62 for hedonic, social, and psychological well-being,
respectively), but weak at the group level (.26, .61, and .28). These findings suggest that
observed scores provide fairly good estimates of the general factor and group factors for
individual-level analyses, but country-level analyses may be biased, unless the non-invariance of
intercepts is accounted for.
To find out the possibility that the mode of administration could contribute to
measurement non-invariance, we conducted measurement invariance analyses across the mode
of administration. The differences in practical fit indices were well below the thresholds (CFI <
.003, RMSEA < .002), supporting scalar invariance. This suggests the absence of effects of
mode of administration independent of those of culture and language.
ESEM analyses
The results of single-group ESEM analyses are given in Supplementary Information. We
failed to obtain convergence of the 3-factor ESEM model in two samples, and the bifactor ESEM
model failed to converge in five other samples. Predictably, the fit of the 3-factor ESEM model
was generally better, compared to that of the ICM-CFA model (scaled chi-square difference test
significant at p < .05 in 34 out of 36 samples). However, the difference in the change in practical
fit indices showed a great variability, with ∆CFI ranging from -.090 to .107 (M = .032, SD =
.038), and ∆RMSEA ranging from -.033 to .042 (M = -.006 to SD = .016) across the samples.
Out of the 33 samples where the bifactor ESEM model converged, its fit, compared to
that of the bifactor model, was only significantly better in 21 samples, based on the scaled chi-
square difference test (p < .05). The change in practical fit indices was quite marginal, with ∆CFI
Page 18 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
ranging from -.097 to .087 (M = .015, SD = .031), ∆RMSEA ranging from -.060 to .054 (M = -
.004, SD = .022). These findings suggest instability of the ESEM model.
We failed to obtain convergence of the multigroup bifactor ESEM model in the 36-
country sample (excluding Kenya and Iran). The 3-factor ESEM model converged only after five
countries were removed, which contributed to negative residual variances (Hungary, India,
Colombia, Hong Kong, Pakistan). The fit indices of the 3-factor metric invariance model were
comparable to those of the bifactor metric invariance model, χ
2
(2602)=4896.49, CFI=.928,
RMSEA=.063 (90% CI: .061-.066), SRMR=.073. However, most of the cross-loadings were
weak (below .20), the only exception being the cross-loading of item 4 (“that you had something
important to contribute to society”) on the psychological well-being factor (in the .30-.40 range).
The factor intercorrelations remained strong, ranging from .44 to .86 across the samples, with
mean correlation of emotional and psychological well-being r = .69, and those of social well-
being .57 and .62 with emotional and psychological well-being, respectively, suggesting a strong
common construct.
Discussion
The current study aimed to examine the measurement invariance of the Mental Health
Continuum – Short Form across 38 countries. It is the first attempt to establish metric invariance
for the MHC-SF in a broad group of countries. Additionally, we examine whether the proposed
bifactor structure of the MHC-SF is cross-culturally replicable and whether it represents the
factor structure of this scale better in comparison to other competitive models, especially the
three-factor model proposed by Keyes (1998) and replicated in some cultural context (e.g.,
Joshanloo et al., 2013; Jovanović, 2015; Young-Jin, 2014), as well as three-factor and bifactor
ESEM models.
Page 19 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
We believe the findings reported in this study to have both theoretical and applied
significance. From the theoretical perspective, a bifactor model allows to examine the extent to
which specific (group) factors are independent from the general factor, and therefore may have a
differential association with other mental health predictors, correlates or outcomes. At the same
time, the bifactor approach also supports the validity of Keyes’s (1998) broad model of mental
health as comprised by three strictly related components (i.e., emotional, psychological, and
social). Regarding the applied perspective, it is useful to know whether the MHC-SF could be
used as a screening test measuring mental health in different cultural contexts. Given that
nowadays many young people study and work in different countries, it is necessary to have a
valid instrument to assess their mental health across countries. Finally, in cross-cultural studies
the issue of measurement invariance is crucial to evaluate the possibility of generalizing findings
across cultural contexts and comparing the levels of mental health across populations.
Our study has shown that a bifactor model provides a better approximation of the factor
structure of the MHC-SF than do alternative models, including a three-factor solution. The
model identification difficulties we encountered are to be expected in small samples, given that
bifactor models include a large number of parameters and some factor loading are expected to be
low (due to partitioning of the variance between two sets of factors). The correspondence of
findings between obtained using constrained full bifactor model and unconstrained incomplete
bifactor model indicates that inequality constraints, although theoretically debatable, turn out to
be a workable approach in these situations.
We have found that the bifactor model showed a good fit to the data in nearly all
countries (with the exception of Kenya and Iran). More in-depth analyses revealed substantial
differences in terms of common and specific variance captured by different MHC-SF subscales.
Page 20 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
In short, emotional and social well-being subscales capture a more substantial proportion of
specific variance, whereas the variance captured by the psychological well-being factor largely
overlaps with that of the general factor. These findings indicate that the effects obtained for the
psychological well-being subscale are most likely to be very similar to those obtained for the
total score and using this subscale on its own may be the best choice when a shorter instrument is
needed.
We also found differences across countries in the extent of common variance captured by
the general factor. Although the countries comprising our sample are mostly collectivist (with the
exception of Germany and the UK: see Hofstede, Hofstede, & Minkov, 2010), we found that in
countries with higher collectivism the general factor tended to be stronger, to the point of making
subscale scores redundant. There are two potential explanations. First, due to interdependent
self-construal present in collectivistic societies, individual and social well-being could be less
distinct domains of subjective well-being (Cross, Bacon, & Morris, 2000; Singelis, 1994).
Second, because there are no reverse-scored items, the effects of acquiescence, which are
stronger in collectivistic contexts (Harzing, 2006), may have contributed to the common factor
variance.
We also investigated differences in the invariance of items belonging to different well-
being domains. The data supported metric invariance of the MHC-SF in all countries except
Algeria, where partial metric invariance was found, as well as full scalar invariance in 10
countries and partial scalar invariance in 26 countries. Although the target ∆CFI was not
reached, recent studies suggest that more lenient cutoff criteria are optimal when the number of
groups is large (Rutkowski, & Svetina, 2014) and models based on a more realistic approximate
Page 21 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
invariance assumption may perform better in these conditions (Kim, Cao, Wang, & Nguyen,
2017).
The emotional well-being subscale emerged as the most universal in terms of item
invariance, whereas the social well-being subscale turned out to be the most problematic. There
are two possible explanations for these findings. First, the items measuring emotional well-being
have simple content and their translations are less likely to be biased, compared to those of more
complex social and psychological well-being items. Second, cultural differences may play a role:
while emotions appear to be universal (Frijda, 2016), social context is strongly culturally diverse,
as it is conditioned by the type of interpersonal relations in society (e.g., collectivism-
individualism, power distance), quality of social environment (as measured by functioning of
democracy or number of crimes), and social beliefs (such as interpersonal trust or societal
cynicism).
We found the application of ESEM models to MHC-SF to be problematic, for several
reasons. The instability of ESEM models we observed can be explained by their complexity and
by the presence of an underlying common construct (as a result, the factors in a correlated-factor
ESEM model are expected to correlate strongly, making it difficult to separate the shared
variance of items due to common construct from that due to indicator cross-loadings). Given the
presence of a common construct, a bifactor model ESEM model would be more appropriate.
However, its non-convergence is not surprising, given the similarity of bifactor ESEM models to
multitrait-multimethod models, where this issue is well-known (Marsh & Bailey, 1991). Also, in
our analyses, we found that the factor correlations based on correlated-factor ESEM models were
not much lower than those obtained using a conventional ICM-CFA model, the difference in the
fit indices between the correlated-factor ICM-CFA and ESEM models was minor, and most
Page 22 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
cross-loadings (except for the cross-loading of item 4) were quite weak. Taken together, these
findings suggest that indicator cross-loadings do not pose a serious issue in the case of MHC-SF
and that bifactor CFA model is a more optimal choice for this instrument.
In general, our analyses indicate that the MHC-SF is invariant at the metric level across
university student samples from most countries and is partially invariant at the scalar level.
Therefore, research findings on correlates, predictors, and consequences of mental health
measured by MHC-SF could be regarded as cross-culturally comparable among university
students, but the bias needs to be addressed whenever a comparison of mean scores is to be
performed. Because the participants in our study were sampled from 38 different countries with
different cultural traditions and socio-political situations, our findings concerning the metric
invariance suggest that MHC-SF can be used with confidence for the assessment and promotion
of mental health in university student samples around the globe. This finding has applied
importance, given the internationalization of universities at both undergraduate and postgraduate
level, as it suggests that health promotion campaigns encompassing emotional, psychological
and social well-being developed for home students may translate well for international students.
An important limitation of our findings is the inclusion of convenience samples, made up
of students, which reduces the level of representativeness. Therefore, future research should
prioritize the study of the validity of the MHC-SF in larger and more heterogeneous samples,
accounting for individual differences in age, sex, socio-economic status, educational level, and
exposure to stressful life events. Our study focuses only on measurement issues and does not
include the validity criteria (i.e., correlations of three MHC-SF factors with objective or
observational data or other established indicators of good vs. poor emotional and psycho-social
functioning), which could be a next step.
Page 23 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Other limitations of the present study include the potential systematic effects of language
(some languages are represented by samples from more than one country, i.e., English,
Portuguese, Russian, and Spanish) and data collection method (paper-based or online survey).
Although we found no uniform effects of the mode of administration on invariance across
countries, the effects of language and mode of administration could potentially interact with
those of culture. Specially designed future studies using parallel samples of respondents from the
same cultures filling out the questionnaire in different languages and using different modes of
administration could separate these effects reliably
.
Conclusion
The MHC-SF scores were found to be reliable and valid for comparative cross-cultural
research. Our results are congruent with those obtained in earlier studies (de Bruin & du Plessis,
2015; Joshanloo et al., 2013; Jovanović, 2015; Keyes, 1998; Young-Jin, 2014) and support the
cross-cultural utility of the MHC-SF and its scoring procedures. Moreover, our project extends
the previous findings to countries from different cultural regions (like Asia, e.g., Nepal, Vietnam,
and Korea; Africa, e.g., RSA or Kenya; and Latin America, e.g., Brazil, Chile, and Puerto Rico).
Despite the fact that the findings of our study suggest that the bifactor model is optimal across
different countries, we recommend the researchers to investigate the internal structure of the
MHC-SF in each new country and determine the best-fitting solution. More specifically, in
collectivist countries the general score of the MHC may be the most informative and bifactor
structure may be unstable due to low variance of domain-specific factors. Therefore, using the
MHC-SF for categorical diagnosis in collectivistic countries should be done with caution, as this
diagnosis is based on the distinction between hedonic (emotional) and eudaimonic (psycho-
social) functioning.
Page 24 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Our findings suggest that the differentiation between the indicators of emotional, social,
and psychological well-being is not very strongly pronounced. The emotional and social well-
being scales capture somewhat more specific phenomena, whereas the psychological well-being
subscale is the closest to the general construct (i.e., it has little to no unique contribution to the
general index of mental health in many countries). The psychological well-being subscale, which
has sufficient reliability for research purposes, could be used on its own when a brief indicator of
general positive functioning is required.
The results concerning the measurement invariance indicate that although effects can be
safely compared across cultures, the comparison of mean scores between countries may be
biased by non-invariant intercepts. This bias is more likely to be associated with the social well-
being items and is less likely to be pertinent to the emotional well-being items. In cases when a
cross-cultural comparison of raw scores is necessary, we recommend researchers to investigate
the level of score comparability in their samples.
We have found scalar invariance only for limited selection of countries, namely Chile,
Colombia, India, Kazakhstan, Nepal, Pakistan, Portugal, South Africa, UK, and Vietnam. It
means that cross-cultural comparisons in the levels of scores could be made only within this
group. Most of these countries are collectivistic. Therefore it is possible to examine a distribution
of different categories of mental health and search for cultural factors associated to the
differences in the levels of mental health in terms of risk factors and protective factors.
In sum, we believe that our study provides initial evidence showing that the MHC-SF
demonstrates good psychometric properties in student samples from 38 different countries. This
empirical evidence of its structural validity and reliability can contribute to the progress in the
study of mental health in cross-cultural perspective.
Page 25 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
References
Arnett, J.J. (2000). Emerging adulthood: A theory of development from the late teens through the
twenties. American Psychology, 55, 469–80.
Berg, M., & Venhoveen, R. (2010). Income inequality and happiness in 119 nations. In search
for an optimum that does not appear to exist. In: BentGreve (Ed.), ‘Social Policy and
Happiness in Europe’ (pp. 174-194). Edgar Elgar Cheltenham UK. ISBN 978-1-84844-
574-1
Brown, T. A. (2015). Confirmatory Factor Analysis for Applied Research. 2
nd
Ed. N. Y.: The
Guilford Press.
Chen, F., Bollen, K. A., Paxton, P., Curran, P. J., & Kirby, J. B. (2001). Improper solutions in
structural equation models: Causes, consequences, and strategies. Sociological Methods
& Research, 29(4), 468-508.
Chen, F. F. (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance.
Structural Equation Modeling, 14, 464-504.
Chetveryk-Burchak, A, & Nosenko, E. (2014). On assessing emotional intelligence as a
precursor of positive personality functioning and mental health. Advances in Social
Science Research Journal, 1, 33-42. doi: 10.14738/assrj.14.248
Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing
measurement invariance. Structural Equation Modeling, 9, 233-255. doi:
10.1207/S15328007SEM0902_5
Cross, S.E., Bacon, P.L., & Morris, M.L. (2000). The relational-interdependent self-construal
and relationships. Journal of Personality and Social Psychology, 78, 791–808. Doi:
10.1037/022-3514.78.4.791
Page 26 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Cummins, R. A., Eckersley, R., Pallant, J., Van Vugt, J., & Misajon, R. (2003). Developing a
national index of subjective wellbeing: The Australian Unity Wellbeing Index. Social
Indicators Research, 64, 159–190. doi: 10.1023/A:1024704320683
De Bruin, G. P., & du Plessis, G. A. (2015). Bifactor analysis of the mental health
continuum-short form (MHC-SF). Psychological Report, 116, 438-446. doi:
10.2466/03.02.PR0.116k20w6.
Diener, E., Emmons, R. A., Larsen, R. J., & Griffin, S. (1985). The Satisfaction with Life Scale.
Journal of Personality Assessment, 49, 71-75. doi:10.1207/s15327752jpa4901_13
Donahy, M.J., Lewis, C.A., Schumaker, J.F., Akuomah-Boateng, R., Duze, M.C., & Sibiya,
T.E. (1998). A cross-cultural analysis of religion and life satisfaction. Mental Health,
Religion, & Culture, 1, 37-43.
Eid, M. (2000). A multitrait-multimethod model with minimal assumptions. Psychometrika,
65(2), 241-261.
Eisenberg, D., Gollust, S.E., Golberstein, E., Hefner, J.L. (2007) Prevalence and correlates of
depression, anxiety, and suicidality among university students. American Journal of
Orthopsychiatry. 77, 534–42.
Frijda, N. (2016). The evolutionary emergence of what we call 'emotions'. Cognition and
Emotion, 30, 609-620.
Harzing, A. W. (2006). Response Styles in Cross-national Survey Research A 26-country
Study. International Journal of Cross Cultural Management, 6, 243-266.
Hofstede, G., Hofstede, G.J., & Minkov, M. (2010). Cultures and organizations: software of the
mind (rev. 3rd ed.). New York, NY: McGraw-Hill.
Page 27 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Hu, L., & Bentler, P.M. (1999). Cutoff criteria for fit indexes in covariance structure analysis:
Conventional criteria versus new alternatives. Structural Equation Modelling, 6, 1-55.
Huntington, S. (1996). The clash of civilizations and remaking of world order. New York:
Touchstone.
Joshanloo, M., & Nostrabadi, M. (2009). Levels of mental health continuum and personality
traits. Social Indicators Research, 90, 211-224.
Joshanloo, M., Wissing, M. P., Khumalo, I. P., & Lamers, S. M. A. (2013). Measurement
invariance of the mental health continuum-short form (MHC-SF) across three cultural
groups. Personality and Individual Differences, 55, 755–759. doi.org/ 10. 1016 /j.paid.
2013.06.002.
Jovanović, V. (2015). Structural validity of Mental Health Continuum – Short Form: The
bifactor model of emotional, social and psychological well-being. Journal of Personality
and Individual Differences, 75, 154-159.
Karaś, D., Cieciuch, J., Keyes, C. L. M. (2014). The Polish Adaptation of the Mental Health
Continuum–Short Form (MHC–SF). Personality and Individual Differences 69, 104-109.
Keyes, C. L. M. (1998). Social well-being. Social Psychology Quarterly, 61, 121-140.
Keyes, C. L. M. (2002). The mental health continuum: From languishing to flourishing in life.
Journal of Health and Social Behavior, 43, 207-222.
Keyes, C. L. M. (2013). Atlanta: Brief description of the mental health continuum short form
(MHC–SF). Available: http://www.sociology.emory.edu/ckeyes/. [On–line, retrieved:
November, 7th, 2013].
Keyes, C. L. M., Wissing, M., Potgieter, J., Temane, M., Kruger, A., & van Rooy, S. (2008).
Evaluation of the mental health continuum-short form (MHC-SF) in Setswana-speaking
Page 28 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
South Africans. Clinical Psychology and Psychotherapy, 15, 181–192.
doi.org/10.1002/cpp.572.
Kim, E. S., Cao, C., Wang, Y., & Nguyen, D. T. (2017). Measurement Invariance Testing with
Many Groups: A Comparison of Five Approaches. Structural Equation Modeling: A
Multidisciplinary Journal. doi: 10.1080/10705511.2017.1304822
Kurré, J., Scholl, J., Bullinger, M., Petersen-Ewert, C. (2011) Integration and health-related
quality of life of undergraduate medical students with migration backgrounds – Results of
a survey. Psycho-Social-Medicine, 8, 7.
Marsh, H. W., & Bailey, M. (1991). Confirmatory factor analyses of multitrait-multimethod
data: A comparison of alternative models. Applied Psychological Measurement, 15(1),
47-70.Mikolajczyk, R.T., Maxwell, A.E., El Ansari, W., Naydenova, V., Stock, C.,
Ilieva, S., et al. (2008). Prevalence of depressive symptoms in university students from
Germany, Denmark, Poland and Bulgaria. Social Psychiatry and Psychiatric
Epidemiology, 43, 105–12.
O’Connor Quinn, H. (2014). Bifactor models, explained common variance (ECV) and the
usefulness of scores from unidimensional item response theory analyses. A thesis
submitted to the faculty of the University of North Carolina at Chapel Hill in the
Department of Psychology.
Petrillo, G., Capone, V., Caso, D., & Keyes, C. L. M. (2015). The Mental Health Continuum -
Short Form (MHC-SF) as a measure of well-being in the Italian context. Social
Indicators Research, 121, 291-312. doi.org/10.1007/s11205-014-0629-3.
Reise, S. P. (2012). The rediscovery of bifactor measurement models. Multivariate
Behavioral Research, 47, 667-696.
Page 29 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Reise, S. P., Scheines, R., Widaman, K. F., & Haviland, M. G. (2013). Multidimensionality
and structural coefficient bias in structural equation modelling a bifactor perspective.
Educational and Psychological Measurement, 73, 5-26.
Roberts, S., Golding, J., Towell, T., Reid, S., & Woodford, S. (2000). Mental and physical health
in students: The role of economic circumstances. British Journal of Health Psychology, 5,
289–297.
Różycka-Tran, J., Boski, P., & Wojciszke, B. (2015). Belief in a zero-sum game as a social
axiom: A 37-nation study. Journal of Cross-Cultural Psychology, 46, 525-548. Doi:
10.1177/02202215572226
Rutkowski, L., & Svetina, D. (2014). Assessing the hypothesis of measurement invariance in the
context of large-scale international surveys. Educational and Psychological
Measurement, 74, 31–57. doi:10.1177/0013164413498257
Ryff, C. D. (1989). Happiness is everything, or is It? Explorations on the meaning of
psychological well-being. Journal of Personality and Social Psychology, 57, 1069-1081.
Sagiv, L., & Schwartz, S. H. (2000). Value priorities and subjective well-being: Direct
relations and congruity effects. European Journal of Social Psychology, 30, 177-198.
Satorra, A., & Bentler, P. M. (2001). A scaled difference chi-square test statistic for moment
structure analysis. Psychometrika, 66, 507-514.
Singelis, T.M. (1994). The measurement of independent and interdependent self-construals.
Personality and Social Psychology Bulletin, 20, 580–591.
http://dx.doi.org/10.1177/0146167294205014
Sirgy, M. J. (2012). The psychology of quality of life. Hedonic well-being, life satisfaction,
and eudaimonia. New York: Springer.
Page 30 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Stock, C., Kücük, N., Miseviciene, I., Guillén-Grima, F., Petkeviciene, J., Aguinaga-Ontoso, I.,
et al. (2003). Differences in health complaints between university students from three
European countries. Preventive Medicine. 37, 535–43.
Vaez, M., Kristenson, M., & Laflamme, L. (2004). Perceived quality of life and self-rated health
among first-year university students; a comparison with their working peers. Social
Indicators Research, 68, 221–34.
Watson, D., Clark, L.A., & Tellegen, A. (1988). Development and validation of brief
measures of positive and negative affect: The PANAS Scales. Journal of Personality and
Social Psychology, 54, 1063-1070.
Welzel, C., & Inglehart, R. (2010). Agency, values and well-being: A human development
model. Social Indicators Research, 97, 43-63. doi: 10.1007/s11205-009-9557-z
West, S. G., Taylor, A. B., & Wu, W. (2012). Model fit and model selection in structural
equation modeling. In R. H. Hoyle (Ed.), Handbook of Structural Equation Modeling (pp.
209-231). New York, N.Y.: The Guilford Press.
Yoon, M., & Kim, E. S. (2014). A comparison of sequential and nonsequential specification
searches in testing factorial invariance. Behavior Research Methods, 46(4), 1199-1206.
Young-Jin, L. (2014). Psychometric characteristics of the Korean Mental Health Continuum–
Short Form in an Adolescent Sample. Journal of Psychoeducational Assessment, 32, 4,
356-364. doi:10.1177/0734282913511431.
Żemojtel-Piotrowska, M., Piotrowski, J.,Cieciuch, J., Calogero, R. M., Van Hiel, A., Argentero,
P., Baltatescu, S., Baran, T., Bardhwaj, G., Bukowski, M., Chargazia, M., Clinton, A.,
Halik, M. H.J., Ilisko, D., Khachatryan, N., Klicperova-Baker, M., Kostal, J., Kovacs, M.,
Letovancova, E., Liik, K., Marganski, A., Michalowski, J., Nord, I., Paspalanova, E.,
Page 31 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Perez de Leon, P., Techera, J., Rochas, M., Rozycka, J., Sawicka, A., Seibt, B., Semkiv,
I., Tiliouine, H., Khanh Truong, H., van den Bos, K., & Wills-Herrera, E. (2015)
Measurement of Psychological Entitlement in 28 Countries. European Journal of
Psychological Assessment. Online first http://dx.doi.org/10.1027/1015-5759/a000286
Page 32 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Table 1
Descriptive Statistics for the 38 Countries
Country
N
Fema
l
e%
Age
M
(
SD
)
SES
M
(
SD
)
L
anguage
procedure
MHC
-
SF
M
(
SD
)
α
Algeria 240 61.25 19.54 (1.58) 4.13 (1.30) Arabic Paper-pencil 51.04(11.39) .79
Armenia 223 47.98 19.00 (1.17) 4.98 (1.20) Armenian Paper-pencil 55.74(10.31) .81
Azerbaijan 120 60.83 20.83 (1.95) 3.38 (0.99) Russian Online 51.55(12.19) .88
Belgium 232 74.14 19.74 (3.95) 4.63 (1.09) Flemish Online 53.99(10.27) .87
Brazil 223 63.68 20.94 (5.21) 4.38 (0.99) Portuguese Paper-pencil 51.44(12.25) .89
Bulgaria 200 66.00 23.59 (5.25) 4.66 (1.16) Bulgarian Paper-pencil 53.16(11.34) .87
Chile 241 52.28 22.00 (2.10) 4.34 (1.03) Spanish Paper-pencil 56.07(11.63) .90
Colombia 138 50.00 18.82 (1.72) 5.74 (0.90) Spanish Online 58.09(12.38) .92
Czech Republic 223 74.89 24.52 (7.75) 4.37 (1.23) Czech Paper-pencil 50.50(12.08) .89
Estonia 301 69.10 23.11 (6.05) 4.41 (1.23) Esti Online 53.84(11.24) .89
Germany 233 82.83 24.99 (6.53) 4.56 (1.29) German Online 54.51(13.02) .91
Hong Kong 172 68.02 18.82 (1.16) 4.31 (1.39) English Paper-pencil 53.17(12.03) .94
Hungary 206 68.93 21.00 (1.68) Hungarian Paper-pencil 56.92(10.46) .88
India 200 68.50 22.59 (1.45) 4.32 (1.07) English Paper-pencil 63.41(10.40) .86
Indonesia 200 50.00 21.38 (1.65) 4.70 (1.02) Bahasa Online 58.98(11.90) .90
Iran 201 50.25 21.28 (1.53) 4.46 (1.41) English Paper-pencil 49.25(11.80) .86
Japan 195 26.15 18.96 (1.13) 4.11 (1.33) Japanese Paper-pencil 42.55(12.81) .89
Kazakhstan 285 74.74 20.12 (2.32) 3.43 (0.89) Russian Online 58.02(13.95) .92
Kenya 162 53.09 23.49 (4.54) 4.07 (0.92) English Paper-pencil 58.09(9.47) .80
Korea (S) 212 54.72 22.20 (1.91) 3.90 (1.24) Korean Paper-pencil 45.81(10.97) .92
Latvia 221 72.40 27.80 (7.91) 2.97 (0.79) Russian Online 53.40(9.86) .90
Malaysia 199 50.25 21.96 (1.22) 4.02 (1.20) Malay Paper-pencil 55.91(11.30) .93
Nepal 203 49.75 22.70 (4.44) 4.08 (0.93) English Paper-pencil 55.34(10.22) .82
Panama 170 33.53 21.41 (5.08) 4.13 (1.00) Spanish Online 56.83(12.89) .90
Pakistan 200 49.00 21.50 (1.59) 4.97 (1.05) English Paper-pencil 54.36(10.13) .82
Page 33 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Poland 227 60.79 22.31 (4.14) 4.69(1.15) Polish Paper-pencil 49.83(13.11) .92
Portugal 193 77.20 22.18 (5.73) 4.11 (1.08) Portuguese Online 54.52(11.50) .90
Puerto Rico 300 42.67 20.26 (2.23) 4.14(1.24) Spanish Paper-pencil 55.67(12.76) .91
Romania 206 48.54 21.33 (3.47) 4.72 (1.13) Romanian Paper-pencil 58.45(11.68) .90
Russia 229 79.48 21.64 (4.13) 3.11 (1.04) Russian Online 49.90(13.54) .90
Serbia 205 60.98 22.46 (5.75) 3.77 (1.10) Serbian Paper-pencil 53.06(12.08) .90
Slovakia 202 71.78 21.13 (1.26) 4.76 (1.00) Slovak Paper-pencil 53.03(11.78) .90
Spain 196 50.51 21.02 (4.66) 4.01 (1.05) Spanish (Catalan) Online 56.29(11.57) .89
South Africa 186 67.20 20.17 (1.86) 4.49(1.25) English Paper-pencil 57.58(11.06) .86
Ukraine 171 80.70 19.86 (2.66) 3.21 (1.06) Russian online 53.00(12.28) .88
United Kingdom 303 80.86 19.53 (2.80) 4.21 (1.33) English online 54.50(13.40) .92
Uruguay 197 80.71 23.51 (6.14) 5.02 (1.00) Spanish Paper-pencil 56.81(10.17) .87
Vietnam 251 52.19 20.51 (2.68) 4.25(1.01) Vietnamese Paper-pencil 53.26(14.17) .92
Overall 8066 61.73 21.55 (4.37) 4.27 (1.25) 54.12(12.37) .89
Note. SES = subjective economic status of family (range 1-7).
Page 34 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Table 2
Fit indices for the 3-Factor and the Bifactor Model in 38 Countries
Country 3-Factor model Bifactor model
χ
2
(74)
CFI
RMSEA (90% CI)
SRMR
χ
2
(63)
CFI
RMSEA
(90% CI)
SRMR
Algeria 135.03
***
.899 .059 (.043-.074) .057 101.40
**
.936 .050 (.031-.068) .046
Armenia
177.02
***
.838
.079 (.064
-
.094)
.067
128.2
6
***
.898
.068 (.051
-
.085)
.058
Azerbaijan 116.43
**
.922 .069 (.044-.092) .069 77.46 .973 .044 (.000-.074) .051
Belgium 154.32
***
.922 .068 (.053-.084) .060 103.63
**
.960 .053 (.034-.070) .046
Brazil 156.93
***
.929 .071 (.055-.086) .058 96.90
**
.971 .049 (.028-.068) .038
Bulgaria 165.95
***
.887 .079 (.063-.095) .073 101.80
**
.952 .055 (.035-.075) .043
Chile 208.16
***
.889 .087 (.073-.101) .070 121.09
***
.952 .062 (.045-.078) .044
Colombia 164.27
***
.888 .094 (.075-.113) .066 81.50 .974 .046 (.000-.073) .041
Czech R. 149.00
***
.929 .067 (.052-.083) .056 98.32
**
.966 .050 (.030-.069) .039
Germany 211.84
***
.908 .089 (.075-.104) .074 114.26
***
.925 .059 (.041-.076) .054
Estonia
156.24
***
.941
.061 (.047
-
.074)
.048
109.46
***
.966
.049 (.033
-
.065)
.037
Hong Kong 160.06
***
.940 .082 (.065-.100) .049 126.53
***
.956 .077 (.057-.096) .039
Hungary 147.03
***
.912 .069 (.053-.086) .064 100.54
**
.955 .054 (.033-.073) .043
India
140.04
***
.898
.067 (.05
0
-
.084)
.059
115.30
***
.92
0
.064 (.045
-
.083)
.051
Indonesia 147.04
***
.921 .070 (.053-.087) .061 99.21
**
.961 .054 (.032-.073) .047
Iran 190.15
***
.862 .088 (.073-.104) .066 157.93
***
.887 .087 (.070-.104) .057
Japan
145.21
***
.918
.07
0
(.053
-
.087)
.066
86.73
*
.973
.044 (.016
-
.065)
.046
Kazakhstan 175.91
***
.938 .070 (.056-.083) .045 125.19
***
.965 .059 (.044-.074) .038
Kenya 191.15
***
.777 .099 (.082-.116) .103 136.36
***
.860 .085 (.065-.104) .086
Korea (S) 209.09
***
.902 .093 (.078-.108) .070 134.50
***
.948 .073 (.056-.090) .043
Latvia 182.14
***
.881 .081 (.066-.096) .081 106.35
***
.952 .056 (.037-.074) .046
Malaysia 142.83
***
.946 .068 (.051-.085) .051 84.20
*
.983 .041 (.010-.063) .047
Nepal 144.92
***
.865 .069 (.052-.085) .071 108.88
***
.913 .060 (.040-.079) .063
Panama 198.87
***
.869 .100 (.083-.116) .089 119.56
***
.941 .073 (.053-.092) .055
Pakistan 165.01
***
.843 .078 (.062-.094) .075 86.10
*
.982 .043 (.015-.064) .047
Poland
192.74
***
.915
.084 (.07
0
-
.099)
.061
142.04
***
.944
.074 (.058
-
.091)
.043
Portugal 139.51
***
.935 .068 (.050-.085) .058 92.01
**
.971 .049 (.025-.069) .039
Puerto R. 253.88
***
.892 .090 (.078-.102) .067 132.31
***
.958 .061 (.046-.075) .039
Romania
170.37
***
.911
.08
0
(.064
-
.095)
.063
133.72
***
.935
.074 (.056
-
.091)
.057
Russia 174.29
***
.921 .077 (.062-.092) .065 120.41
***
.955 .063 (.046-.080) .048
Serbia 223.02
***
.868 .099 (.084-.114) .074 111.47
***
.957 .061 (.042-.080) .042
Slovakia 221.42
***
.873 .099 (.084-.114) .095 112.10
***
.958 .062 (.043-.081) .042
S. Africa 135.29
***
.905 .067 (.049-.084) .061 95.32
**
.950 .053 (.029-.073) .048
Spain 177.14
***
.886 .084 (.068-.100) .077 98.31
**
.961 .053 (.032-.073) .047
Ukraine 109.55
**
.956 .053 (.030-.073) .056 74.20 .986 .032 (.000-.059) .038
UK
247.14
***
.915
.088 (.076
-
.1
00
)
.066
155.31
***
.955
.07
0
(.056
-
.083)
.038
Uruguay 118.98
***
.950 .056 (.036-.074) .063 82.37 .978 .040 (.000-.062) .041
Vietnam 187.83
***
.923 .078 (.064-.092) .057 124.38
***
.959 .062 (.046-.078) .041
Note. Satorra-Bentler χ
2
, *** p < .001, ** p < .01, * p < .05. CFI = Comparative Fit Index,
RMSEA = Root Mean Square of Approximation, SRMR = Standardized Root Mean Square
Residual. The comparisons between models are impossible as they are not nested models.
Page 35 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Table 3
Reliability and Dimensionality Indices for the MHC-SF in 38 Countries
Reliability, ω Variance explained
Country Gen
EWB
SWB
PWB
ω
H
ω
S
EWB
ω
S
SWB
ω
S
PWB
ECV
Algeria .82 .73 .63 .67 .68 .17 .30 .24 .61
Armenia .83 .74 .57 .82 .68 .42 .40 .07 .54
Azerbaijan .91 .81 .78 .82 .83 .33 .28 .02 .69
Belgium .90 .87 .74 .83 .78 .39 .35 .09 .62
Brazil .92 .85 .74 .86 .83 .31 .27 .10 .71
Bulgaria .90 .84 .73 .81 .80 .18 .27 .14 .66
Chile .91 .83 .80 .85 .79 .18 .32 .22 .65
Colombia .94 .89 .84 .87 .86 .12 .28 .09 .71
Czech Rep. .92 .85 .80 .83 .83 .23 .36 .06 .69
Estonia .91 .85 .76 .84 .83 .32 .31 .03 .70
Germany .94 .88 .81 .89 .87 .18 .33 .00 .72
Hong Kong .95 .90 .87 .92 .87 .27 .27 .11 .74
Hungary .90 .79 .78 .84 .80 .27 .30 .14 .62
India .89 .76 .79 .79 .76 .32 .38 .07 .57
Indonesia .92 .82 .81 .87 .83 .31 .30 .07 .69
Iran .88 .83 .67 .78 .78 .09 .15 .31 .71
Japan .92 .83 .81 .86 .84 .46 .17 .11 .66
Kazakhstan .94 .85 .83 .89 .87 .23 .28 .00 .76
Kenya .86 .75 .82 .78 .56 .50 .20 .68 .40
Korea (S) .94 .91 .83 .89 .85 .36 .35 .03 .69
Latvia .93 .88 .83 .87 .76 .18 .47 .28 .60
Malaysia .95 .87 .88 .89 .86 .40 .00 .28 .71
Nepal .86 .67 .75 .77 .71 .40 .43 .07 .52
Pakistan .86 .69 .75 .80 .71 .44 .50 .00 .51
Panama .93 .88 .83 .88 .83 .47 .35 .00 .64
Poland .94 .88 .83 .89 .85 .32 .35 .03 .72
Portugal .92 .86 .79 .87 .82 .29 .29 .14 .69
Puerto Rico .93 .87 .82 .86 .81 .15 .32 .20 .69
Romania .92 .87 .77 .89 .80 .49 .38 .06 .65
Russia .92 .86 .76 .88 .84 .30 .34 .04 .71
Serbia .92 .88 .77 .87 .82 .11 .35 .20 .67
Slovakia .93 .78 .81 .89 .82 .17 .36 .15 .66
Spain .92 .85 .77 .87 .83 .36 .34 .01 .66
South Africa .89 .76 .77 .79 .77 .27 .40 .01 .62
Ukraine .91 .84 .75 .84 .80 .35 .27 .15 .67
UK .94 .90 .86 .85 .85 .18 .29 .13 .72
Uruguay .90 .87 .72 .85 .78 .31 .38 .08 .63
Vietnam .94 .84 .86 .89 .86 .37 .11 .18 .71
M .91 .83 .78 .85 .80 .29 .31 .12 .66
SD .03 .06 .06 .05 .06 .11 .09 .13 .07
Note. Gen = general score; EWB = emotional well-being; SWB = social well-being; PWB =
psychological well-being.
Page 36 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Table 4
Fit Indices for the Multi-Group Models
Model S-B χ
2
df CFI RMSEA (90 % CI) SRMR
ICM-CFA models (36 countries)
Configural invariance 3990.57*
2268 .955 .060 (.057-.063) .045
Metric invariance 5875.03*
3108 .928 .065 (.062-.067) .081
Scalar invariance 9162.20*
3458 .851 .088 (.086-.090) .102
Partial metric invariance 5834.54*
3106 .929 .064 (.062-.067) .080
Partial scalar invariance 7047.52*
3402 .905 .071 (.068-.073) .084
Note. Satorra-Bentler χ
2
, * p < .001. CFI = Comparative Fit Index, RMSEA = Root Mean
Square of Approximation, SRMR = Standardized Root Mean Square Residual.
Page 37 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting Information 1
List of Non-Invariant Parameters
Loadings Intercepts
Algeria 12 3, 4, 5, 7
Armenia 3, 5, 14
Azerbaijan 8
Belgium 1, 6, 8, 11, 12
Brazil 6, 12
Bulgaria 8
Chile
Colombia
Czech Rep 4
Estonia 4, 8, 13
Germany 4, 14
Hong Kong 6, 12
Hungary 5, 6, 14
India
Indonesia 7
Japan 3, 5, 11
Kazakhstan
Korea 2, 13, 14
Latvia 3
Malaysia 4
Nepal
Pakistan
Panama 5
Poland 3
Portugal
Puerto Rico 12
Romania 6, 7
Russia 1, 13
Serbia 4, 6, 11, 14
Slovakia 5, 8
South Africa
Spain 5
Ukraine 4, 7
United King.
Uruguay 4, 9
Vietnam
Page 38 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting information 2
Estimated Latent Factor Parameters
Means Standard deviations
MHC EWB SWB PWB MHC EWB SWB PWB
Algeria 1.61 -3.12 -0.87 -1.72 1.31 1.03 1.21 0.80
Armenia 0.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00
Azerbaijan -0.23 -0.85 0.23 -0.54 1.43 1.09 1.16 0.04
Belgium -1.51 1.76 1.37 1.30 1.14 1.10 0.91 0.59
Brazil 0.20 -0.50 -1.22 -1.16 1.43 0.93 1.04 0.88
Bulgaria -0.03 -0.41 -0.29 -0.40 1.30 0.89 1.11 0.48
Chile 0.15 -0.19 -0.22 -0.15 1.30 0.57 1.28 0.83
Colombia 0.26 0.08 0.21 -0.30 1.45 0.41 1.24 0.44
Czech Rep. 0.41 -1.06 -0.50 -1.65 1.40 1.03 1.15 0.31
Estonia -0.30 0.00 0.21 -0.01 1.29 1.02 0.95 0.37
Germany 0.42 -0.69 -0.34 -0.64 1.58 0.85 1.30 0.08
Hong Kong -1.79 1.45 2.40 1.18 1.33 0.75 1.07 0.80
Hungary -1.16 1.03 1.01 1.32 1.14 0.86 1.14 0.45
India 0.08 0.18 1.52 0.81 1.09 0.88 1.27 0.34
Indonesia -0.29 0.04 1.49 0.29 1.31 0.86 1.08 0.78
Japan -2.33 -0.08 1.12 -0.05 1.46 1.34 0.96 0.68
Kazakhstan 0.34 -0.40 0.68 -0.57 1.58 0.86 1.16 0.64
Latvia -0.39 -0.77 0.88 -0.29 1.15 0.89 1.02 0.71
Malaysia -1.23 0.86 2.01 0.81 1.29 0.93 0.86 0.60
Nepal -0.98 0.27 1.58 0.77 1.08 1.06 1.11 0.54
Pakistan -0.69 -0.04 1.36 0.08 0.99 1.02 1.39 0.87
Panama 0.47 -0.29 -0.16 -0.50 1.47 1.22 1.41 0.53
Poland -1.38 0.19 0.80 0.58 1.49 1.19 1.06 0.70
Portugal 0.01 -0.04 -0.28 -0.38 1.29 0.81 1.04 0.79
Puerto Rico 0.26 -0.33 -0.24 -0.59 1.52 0.72 1.32 0.78
Romania -0.24 0.53 1.30 0.43 1.27 1.08 1.09 1.01
Russia -0.32 -0.48 -0.02 -1.03 1.59 1.09 1.22 0.80
S. Korea -0.74 -0.64 0.07 -0.83 1.27 0.91 1.02 0.57
Serbia 1.03 -1.86 -1.05 -1.31 1.39 0.99 1.21 0.57
Slovakia -0.43 0.15 0.08 -0.17 1.33 0.65 1.21 0.93
South Africa -0.44 0.15 0.70 0.90 1.23 0.87 1.27 0.32
Spain 0.26 -0.19 -0.43 -0.44 1.32 0.95 1.20 0.66
Ukraine 0.42 -0.92 -0.46 -1.21 1.45 1.25 1.04 0.68
United King. -0.64 0.70 0.59 0.14 1.51 0.94 1.33 0.51
Uruguay 1.09 -1.03 -1.19 -0.92 1.18 0.89 1.14 0.44
Vietnam -0.58 -0.45 1.27 -0.25 1.57 1.21 1.04 0.92
Note. EWB = emotional well-being; SWB = social well-being; PWB = psychological well-
being.
Page 39 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting Information 3
Fit Indices for the 1-Factor and the 2-Factor Model in 38 Countries
Country
1
-
Factor model
2
-
factor model
χ
2
(77) CFI RMSEA (90% CI) SRMR χ
2
(76) CFI RMSEA (90% CI) SRMR
Algeria 191.74*** .810 .079 (.065-.093) .067 173.94*** .838 .073 (.059-.088) .064
Armenia 262.19*** .709 .104 (.090-.118) .081 209.62*** .790 .089 (.075-.103) .075
Azerbaijan 171.10*** .826 .101 (.081-.121) .078 135.50*** .890 .081 (.058-.103) .070
Belgium 331.37*** .752 .119 (.106-.133) .084 21.13*** .869 .087 (.073-.101) .068
Brazil 272.83*** .833 .107 (.093-.121) .071 192.08*** .901 .083 (.068-.097) .062
Bulgaria 233.62*** .808 .101 (.086-.116) .074 208.93*** .837 .094 (.079-.109) .071
Chile 315.43*** .802 .113 (.100-.127) .079 271.54*** .838 .103 (.090-.117) .074
Colombia 189.73*** .860 .103 (.085-.122) .071 184.92*** .865 .102 (.083-.121) .070
Czech
Republic 260.76*** .825 .103 (.090-.117) .071 214.92*** .868 .091 (.076-.105) .065
Estonia 310.41*** .831 .100 (.089-.112) .067 215.37*** .899 .078 (.066-.090) .057
Germany 300.56*** .850 .112 (.098-.125) .074 259.15*** .878 .102 (.088-.115) .069
Hong Kong 330.83*** .823 .138 (.123-.154) .070 241.08*** .885 .112 (.097-.128) .061
Hungary 237.60*** .805 .101 (.086-.115) .071 195.25*** .856 .087 (.072-.102) .064
Page 40 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
India 221.87*** .777 .097 (.082-.112) .076 191.67*** .822 .087 (.072-.103) .071
Indonesia 247.49*** .817 .105 (.091-.120) .075 206.62*** .859 .093 (.078-.108) .069
Iran 253.47*** .790 .107 (.092-.122) .077 231.59*** .815 .101 (.086-.116) .075
Japan 246.01*** .806 .106 (.091-.121) .077 169.52*** .892 .079 (.063-.096) .066
Kazakhstan 295.09*** .868 .100 (.088-.112) .058 237.51*** .902 .086 (.074-.099) .053
Kenya 335.56*** .508 .144 (.128-.160) .134 292.93*** .587 .133 (.117-.149) .132
Korea (S) 397.63*** .769 .140 (.127-.154) .086 262.45*** .865 .108 (.094-.122) .072
Latvia 321.01*** .732 .120 (.106-.133) .104 275.42*** .781 .109 (.095-.123) .098
Malaysia 270.37*** .848 .112 (.098-.127) .067 196.09*** .906 .089 (.074-.105) .058
Nepal 216.67*** .735 .095 (.080-.110) .087 184.83*** .793 .084 (.069-.099) .081
Pakistan 284.95*** .642 .116 (.102-.131) .096 244.34*** .710 .105 (.091-.120) .088
Panama 337.66*** .728 .141 (.126-.157) .091 248.90*** .819 .116 (.100-.132) .078
Poland 365.67*** .794 .129 (.115-.142) .075 281.53*** .853 .109 (.096-.123) .072
Portugal 265.25*** .812 .113 (.098-.128) .074 187.80*** .888 .087 (.072-.103) .064
Puerto Rico 380.09*** .818 .115 (.103-.126) .078 311.90*** .858 .102 (.090-.114) .073
Romania 359.26*** .741 .133 (.120-.147) .090 23.06*** .858 .099 (.085-.114) .075
Russia 297.11*** .827 .112 (.098-.125) .077 24.31*** .871 .097 (.083-.111) .072
Page 41 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Serbia 276.36*** .823 .112 (.098-.127) .077 256.46*** .840 .108 (.093-.122) .076
Slovakia 314.88*** .795 .124 (.110-.138) .084 281.21*** .824 .116 (.101-.130) .083
South Africa 210.83*** .793 .097 (.081-.112) .074 189.59*** .824 .090 (.074-.106) .071
Spain 256.43*** .802 .109 (.094-.124) .081 197.59*** .866 .090 (.075-.106) .073
Ukraine 202.79*** .843 .098 (.081-.114) .073 135.21*** .926 .067 (.049-.086) .061
United
Kingdom 475.82*** .805 .131 (.120-.142) .073 376.56*** .853 .114 (.103-.126) .067
Uruguay 239.76*** .818 .104 (.089-.119) .080 171.11*** .894 .080 (.064-.096) .070
Vietnam 333.39*** .827 .115 (.103-.128) .071 24.59*** .889 .093 (.080-.106) .059
Note. Satorra-Bentler χ
2
, *** p < .001, ** p < .01, * p < .05. CFI = Comparative Fit Index, RMSEA = Root Mean Square of Approximation,
SRMR = Standardized Root Mean Square Residual. The comparisons between models are impossible as they are not nested models. 2-factor
model is comprised by hedonic and eudaimonic well-being.
Page 42 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting Information 4
Descriptive Statistics and Correlations of MHC-SF Subscales by Country
Country
N
Reliability, α
Mean (SD)
Pearson correlation, r
EWB SWB PWB EWB SWB PWB EWB-
SWB
EWB-
PWB
SWB-
PWB
Algeria
240
.71
.61
.63
3.98 (1.23)
2.76 (1.07)
4.22 (0.89)
.
44
.
49
.
41
Armenia 223 .73 .57 .78 4.42 (0.94) 3.13 (0.98) 4.47 (0.91) .33 .49 .39
Azerbaijan 120 .80 .76 .78 3.94 (1.09) 3.21 (1.03) 3.95 (0.96) .46 .64 .66
Belgium
232
.87
.72
.79
4.41 (0.93)
3.16 (0.90)
4.16 (0.84)
.
41
.
57
.
57
Brazil
223
.84
.72
.85
4.36 (1.04)
2.87 (0.92)
4.00 (1.08)
.
49
.
66
.
59
Bulgaria 200 .80 .73 .77 4.29 (1.04) 3.07 (0.96) 4.15 (0.90) .50 .68 .56
Chile 241 .83 .78 .84 4.50 (0.88) 3.24 (1.08) 4.40 (0.92) .56 .66 .57
Colombia 138 .85 .81 .84 4.70 (0.91) 3.54 (1.09) 4.38 (0.95) .67 .77 .65
Czech Republic
223
.85
.77
.8
0
4.19 (1.11)
3.09 (0.99)
3.75 (0.96)
.
53
.
69
.
59
Estonia 301 .85 .74 .81 4.35 (1.01) 3.18 (0.94) 4.15 (0.90) .49 .66 .60
Germany 233 .88 .77 .86 4.39 (1.11) 3.19 (1.04) 4.23 (1.03) .55 .78 .65
Hong Kong
172
.89
.85
.92
4.27 (0.89)
3.40 (1.00)
3.89 (0.96)
.
63
.
70
.
70
Hungary
206
.77
.75
.8
0
4.39 (0.92)
3.39 (0.91)
4.47 (0.84)
.
48
.
59
.
60
India 200 .74 .78 .73 4.64 (0.95) 4.06 (1.04) 4.86 (0.75) .47 .55 .52
Indonesia 200 .82 .80 .84 4.35 (1.00) 3.95 (1.04) 4.36 (0.93) .56 .62 .62
Iran
201
.81
.66
.77
3.59 (1.26)
2.73 (0.86)
4.14 (1.00)
.
59
.
61
.
46
Japan
195
.82
.77
.82
3.24 (1.11)
2.88 (1.02)
3.07 (1.08)
.
54
.
51
.
69
Kazakhstan 285 .84 .82 .86 4.47 (1.11) 3.77 (1.17) 4.29 (1.08) .61 .73 .69
Kenya 162 .73 .75 .78 4.22 (1.05) 3.95 (0.96) 4.28 (0.84) .39 .26 .24
Korea (S) 212 .91 .80 .87 3.63 (0.90) 2.94 (0.92) 3.37 (0.88) .52 .67 .64
Latvia
221
.87
.82
.85
3.97
(0.94)
3.50 (0.86)
4.00 (0.79)
.
52
.
63
.
47
Malaysia 199 .85 .84 .88 4.27 (0.96) 3.72 (0.93) 4.09 (0.88) .60 .66 .70
Nepal 203 .67 .72 .72 4.07 (1.00) 3.55 (0.98) 4.23 (0.85) .30 .45 .47
Pakistan
200
.68
.73
.74
4.11 (0.95)
3.58 (1.00)
4.02 (0.86)
.
28
.
41
.
43
Panama
170
.88
.79
.84
4.63 (1.09)
3.31 (1.12)
4.39 (1.02)
.
52
.
60
.
64
Poland 227 .86 .82 .87 3.88 (1.11) 2.98 (1.05) 3.88 (1.07) .57 .68 .64
Portugal 193 .86 .78 .85 4.51 (0.88) 3.18 (0.98) 4.19 (0.96) .53 .63 .62
Puerto Rico
300
.85
.78
.85
4.49 (1.06)
3.30 (1.08)
4.29 (1.02)
.
58
.
70
.
60
Page 43 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Romania
206
.87
.75
.87
4.67 (0.99)
3.52 (0.99)
4.48 (0.99)
.
45
.
56
.
58
Russia
229
.86
.75
.86
3.96 (1.21)
3.10 (1.07)
3.75 (1.14)
.
52
.
69
.
56
Serbia 205 .83 .71 .76 4.12 (1.12) 3.17 (0.93) 4.14 (0.99) .57 .72 .55
Slovakia 202 .76 .75 .88 4.35 (0.91) 3.11 (0.96) 4.07 (1.02) .55 .65 .59
South Africa
186
.74
.75
.76
4.33 (0.98)
3.40 (1.08)
4.60 (0.83)
.
47
.
59
.
53
Spain
196
.85
.72
.84
4.59 (0.99)
3.36 (0.96)
4.29 (0.96)
.
52
.
62
.
59
Ukraine 171 .84 .73 .82 4.27 (1.14) 3.22 (1.03) 4.02 (1.00) .45 .57 .60
United Kingdom 303 .89 .84 .84 4.52 (1.05) 3.27 (1.17) 4.10 (1.02) .63 .72 .66
Uruguay
197
.87
.7
0
.81
4.58 (0.91)
3.26 (0.91)
4.46 (0.83)
.
40
.
66
.
52
Vietnam
251
.84
.82
.87
3.92 (1.18)
3.57 (1.18)
3.94
(1.11)
.
61
.
58
.
73
Note. EWB = emotional well-being subscale, SWB = social well-being subscale, PWB = psychological well-being subscale. Correlations are
calculated on observed scores.
Page 44 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting Information 5
Fit Indices for the 3-Factor and the Bifactor ESEM Models in 38 Countries
Country
3
-
Factor
ESEM
model
Bi
factor
ESEM
model
χ
2
(52) CFI RMSEA (90% CI) SRMR χ
2
(41) CFI RMSEA (90% CI) SRMR
Kenya
120.13
***
0.862
0.09 (0.069
-
0.111)
0.051
71.8
4
**
0.937
0.068 (0.041
-
0.094)
0.036
United Kingdom
112.31
***
0.968
0.062 (0.046
-
0.078)
0.026
75.85
***
0.982
0.053 (0.034
-
0.071)
0.021
Serbia 106.57*** 0.948 0.072 (0.052-
0.091)
0.033 82.34*** 0.961
0.07 (0.048-0.092) 0.025
Nepal 65.91 0.972 0.036 (0-0.061) 0.036 39.43 1 0 (0-0.045) 0.024
Chile 110.76*** 0.948 0.068 (0.051-
0.086)
0.034 69.79** 0.975
0.054 (0.031-0.075) 0.024
Portugal 89.54*** 0.959 0.061 (0.039-
0.082)
0.033 66.24** 0.972
0.056 (0.029-0.081) 0.025
Belgium
78.72
*
0.973
0.047 (0.024
-
0.067)
0.031
NA
NA
NA (NA
-
NA)
NA
Hungary
103.0
8
***
0.936
0.069 (0.049
-
0.089)
0.041
65.64
**
0.969
0.054 (0.028
-
0.078)
0.032
Romania 123.31*** 0.932 0.082 (0.063-0.1) 0.038 87.26*** 0.956
0.074 (0.052-0.096) 0.029
Spain 94.03*** 0.95 0.064 (0.043-
0.085)
0.034 57.33* 0.981
0.045 (0.006-0.071) 0.024
Puerto Rico 106.15*** 0.965 0.059 (0.043-
0.075)
0.029 53.28 0.992
0.032 (0-0.054) 0.02
Indonesia 111.16*** 0.933 0.075 (0.056-
0.095)
0.038 51.02 0.989
0.035 (0-0.063) 0.021
India
174.75
***
0.808
0.109 (0.091
-
0.127)
0.05
154.22
***
0.823
0.118 (0.098
-
0.138)
0.036
Page 45 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Slovakia 114.17*** 0.943 0.077 (0.058-
0.096)
0.035 99.22*** 0.947
0.084 (0.063-0.105) 0.027
Bulgaria
116.61
***
0.916
0.079 (0.06
-
0.098)
0.04
NA
NA
NA (NA
-
NA)
NA
Iran
129.5
8
***
0.903
0.086 (0.068
-
0.105)
0.043
NA
NA
NA (NA
-
NA)
NA
Panama
107.51
***
0.937
0.079 (0.058
-
0.1)
0.034
NA
NA
NA (NA
-
NA)
NA
Japan 75.55* 0.971 0.048 (0.021-
0.071)
0.034 55.03 0.983
0.042 (0-0.068) 0.026
Russia 122.99* 0.94 0.077 (0.06-0.095) 0.035 82.50*** 0.965
0.066 (0.045-0.087) 0.027
Ukraine 74.61*** 0.97 0.05 (0.02-0.075) 0.032 46.31 0.993
0.028 (0-0.061) 0.023
Malaysia 96.28*** 0.962 0.065 (0.045-
0.086)
0.036 43.00 0.998
0.016 (0-0.052) 0.019
Poland 172.59*** 0.907 0.101 (0.085-
0.118)
0.041 124.50*** 0.936
0.095 (0.076-0.114) 0.031
Azerbaijan
NA
NA
NA (NA
-
NA)
NA
38.09
1
0 (0
-
0.056)
0.027
Latvia
91.3
5
***
0.953
0.059 (0.038
-
0.078)
0.036
53.41
0.985
0.037 (0
-
0.063)
0.024
Colombia 97.08*** 0.941 0.079 (0.054-
0.103)
0.039 99.41*** 0.924
0.102 (0.076-0.127) 0.029
Czech Republic 100.50*** 0.951 0.065 (0.045-
0.084)
0.034 69.41** 0.971
0.056 (0.032-0.078) 0.027
Kazakhstan 147.12*** 0.937 0.08 (0.065-0.096) 0.035 97.48*** 0.963
0.07 (0.052-0.087) 0.023
Hong Kong 122.28*** 0.946 0.089 (0.068-
0.109)
0.035 70.31** 0.978
0.064 (0.038-0.09) 0.024
Uruguay
75.39
*
0.973
0.048
(0.02
-
0.07)
0.031
58.51
*
0.98
0.047 (0.012
-
0.072)
0.024
Page 46 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Germany NA NA NA (NA-NA) NA 64.35* 0.983
0.049 (0.024-0.072) 0.019
Algeria 104.14*** 0.909 0.065 (0.046-
0.083)
0.04 NA NA NA (NA-NA) NA
Pakistan
173.6
2
***
0.779
0.108 (0.091
-
0.126)
0.05
52.0
2
0.98
0.037 (0
-
0.064)
0.027
Vietnam
118.2
6
***
0.951
0.071 (0.054
-
0.088)
0.034
84.08
***
0.968
0.065 (0.045
-
0.084)
0.025
Korea (S) 140.63*** 0.932 0.09 (0.072-0.108) 0.038 79.49*** 0.97 0.067 (0.044-0.088) 0.025
Armenia 118.11*** 0.89 0.076 (0.057-
0.094)
0.048 83.99*** 0.929
0.069 (0.047-0.089) 0.032
South Africa 91.50*** 0.935 0.064 (0.042-
0.085)
0.042 57.43 0.973
0.046 (0.007-0.073) 0.03
Brazil 103.42*** 0.953 0.067 (0.048-
0.085)
0.033 85.90*** 0.959
0.07 (0.049-0.091) 0.026
Estonia
118.65
***
0.948
0.065
(0.05
-
0.081)
0.036
80.67
***
0.969
0.057 (0.038
-
0.075)
0.026
Note. Satorra-Bentler χ
2
, *** p < .001, ** p < .01, * p < .05. CFI = Comparative Fit Index, RMSEA = Root Mean Square of Approximation,
SRMR = Standardized Root Mean Square Residual. The comparisons between models are impossible as they are not nested models. 2-factor
model is comprised by hedonic and eudaimonic well-being.
Page 47 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Supporting Information 6
Fit Indices and Structural Coefficients for the Incomplete Bifactor Model in 38 Countries
Country Fit indices Structural coefficients
χ
2
(74)
CFI
RMSEA (90% CI)
SRMR
ω
ω
H
ω
S
EWB
ω
S
SWB
ECV
Algeria
135.40***
0.890
0.063 (0.047
-
0.079)
0.055
0.81
0.71
0.26
0.33
0.68
Armenia
172.40***
0.838
0.082 (0.067
-
0.097)
0.065
0.82
0.70
0.45
0.37
0.65
Azerbaijan
178.12***
0.805
0.115 (0.094
-
0.136)
0.125
0.92
0.73
0.45
0.64
0.55
Belgium
125.91***
0.944
0.060 (0.043
-
0.076)
0.050
0.89
0.80
0.43
0.34
0.68
Brazil
116.46***
0.960
0.056 (0.037
-
0.073)
0.041
0.91
0.85
0.34
0.30
0.78
Bulgaria
124.93***
0.931
0.064 (0.045
-
0.081)
0.049
0.89
0.82
0.21
0.29
0.74
Chile
160.70***
0.924
0.074 (0.059
-
0.089)
0.051
0.91
0.82
0.27
0.34
0.72
Colombia
107.44**
0.945
0.064 (0.039
-
0.086)
0.050
0.91
0.83
0.22
0.37
0.75
Czech R.
187.54***
0.910
0.088 (0.073
-
0.103)
0.048
0.93
0.87
0.21
0.25
0.82
Germany
139.76***
0.949
0.058 (0.044
-
0.072)
0.042
0.90
0.83
0.32
0.31
0.75
Estonia
138.98***
0.898
0.066 (0.050
-
0.082)
0.058
0.86
0.71
0.40
0.51
0.57
Hong Kong
100.42**
0.961
0.051 (0.027
-
0.072)
0.047
0.89
0.79
0.31
0.41
0.69
Hungary
115.16***
0.944
0.057 (0.038
-
0.075)
0.046
0.90
0.82
0.31
0.31
0.71
India
131.43***
0.904
0.067
(0.050
-
0.085)
0.055
0.88
0.76
0.32
0.37
0.65
Indonesia
141.46***
0.922
0.072 (0.055
-
0.089)
0.061
0.92
0.83
0.33
0.35
0.73
Iran
192.99***
0.852
0.095 (0.079
-
0.110)
0.067
0.88
0.82
0.26
0.27
0.78
Japan
119.88***
0.941
0.061 (0.043
-
0.080)
0.057
0.91
0.84
0.50
0.23
0.74
Kazakhstan
133.59***
0.964
0.057 (0.043
-
0.072)
0.043
0.94
0.87
0.30
0.32
0.76
Kenya
172.86***
0.802
0.096 (0.079
-
0.114)
0.101
0.83
0.55
0.61
0.72
0.44
Korea (S)
157.30***
0.936
0.078 (0.062
-
0.094)
0.046
0.93
0.85
0.37
0.34
0.73
Latvia
102.45**
0.968
0.047 (0.026
-
0.065)
0.039
0.92
0.85
0.20
0.33
0.76
Malaysia
94.92*
0.969
0.043 (0.018
-
0.064)
0.044
0.91
0.83
0.34
0.31
0.73
Nepal
126.70***
0.891
0.064 (0.046
-
0.082)
0.066
0.85
0.72
0.40
0.42
0.60
Panama
130.92***
0.951
0.067
(0.049
-
0.084)
0.047
0.94
0.88
0.35
0.24
0.79
Pakistan
140.75***
0.925
0.078 (0.060
-
0.097)
0.058
0.93
0.83
0.43
0.36
0.69
Poland
163.02***
0.933
0.077 (0.062
-
0.093)
0.046
0.93
0.86
0.31
0.34
0.77
Portugal
117.30***
0.952
0.060 (0.041
-
0.079)
0.047
0.92
0.84
0.36
0.32
0.75
Puerto R.
165.39***
0.942
0.068 (0.055
-
0.082)
0.044
0.92
0.85
0.24
0.35
0.75
Romania
153.64***
0.922
0.077 (0.061
-
0.094)
0.059
0.92
0.81
0.51
0.37
0.70
Russia
137.39***
0.946
0.066 (0.050
-
0.082)
0.049
0.92
0.84
0.29
0.36
0.75
Serbia
168.30***
0.912
0.084 (0.068
-
0.100)
0.055
0.92
0.84
0.20
0.39
0.75
Slovakia
145.73***
0.934
0.074 (0.057
-
0.091)
0.054
0.92
0.83
0.27
0.42
0.72
S. Africa
118.42***
0.924
0.062 (0.042
-
0.081)
0.054
0.88
0.77
0.25
0.38
0.67
Spain
124.03***
0.939
0.064
(0.045
-
0.082)
0.052
0.91
0.83
0.35
0.33
0.71
Ukraine
89.96*
0.974
0.042 (0.006
-
0.065)
0.044
0.90
0.82
0.39
0.30
0.73
UK
179.29***
0.946
0.073 (0.060
-
0.086)
0.046
0.93
0.86
0.26
0.31
0.76
Uruguay
149.07***
0.941
0.077 (0.060
-
0.094)
0.043
0.93
0.87
0.16
0.32
0.79
Vietnam
184.32***
0.922
0.082 (0.067
-
0.096)
0.060
0.93
0.86
0.43
0.25
0.76
M
0.90
0.81
0.33
0.36
0.71
SD
0.03
0.07
0.10
0.09
0.07
Note. Satorra-Bentler χ
2
, *** p < .001, ** p < .01, * p < .05. CFI = Comparative Fit Index,
RMSEA = Root Mean Square of Approximation, SRMR = Standardized Root Mean Square
Residual. The comparisons between models are impossible as they are not nested models.
Page 48 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Peer Review
MEASUREMENT INVARIANCE OF THE MHC-SF ACROSS 38 COUNTRIES
Appendix
MHC-SF
Please answer the following questions are about how you have been feeling during the past month. Place a check
mark in the box that best represents how often you have experienced or felt the following:
During the past month, how often
did you feel …
never
once or
twice
about
once a
week
about 2 or 3
times a week
almost
every day
every
day
1. happy 1 2 3 4 5 6
2. interested in life
1
2
3
4
5
6
3. satisfied with life 1 2 3 4 5 6
4. that you had something important to
contribute to society
1
2
3
4
5
6
5. that you belonged to a community
(like a social group, or your
neighborhood)
1
2
3
4
5
6
6. that our society is a good place, or is
becoming a better place, for all people
1 2 3 4 5 6
7. that people are basically good 1 2 3 4 5 6
8. that the way our society works makes
sense to you
1
2
3
4
5
6
9. that you liked most parts of your
personality
1
2
3
4
5
6
10. good at managing the responsibilities
of your daily life
1
2
3
4
5
6
11. that you had warm and trusting
relationships with others
1
2
3
4
5
6
12. that you had experiences that
challenged you to grow and become a
better person
1
2
3
4
5
6
13. confident to think or express your
own ideas and opinions
1 2 3 4 5 6
14. that your life has a sense of direction
or meaning to it
1 2 3 4 5 6
Page 49 of 49
John Wiley & Sons
Journal of Clinical Psychology
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
... A typical example of these challenges can be seen with the Mental Health Continuum-Short Form [MHC-SF; (15)]. The MHC-SF is one of the most popular PPAMs aimed at measuring mental health, has shown to produce various factorial models ranging from a correlated three first-order factorial model (comprised of emotional-, psychological-, and social wellbeing) through to various types of bifactor models with varying ranges of reliability (16)(17)(18)(19). Zemojtel-Piotrowska et al. (19) also showed that the MHC-SF is not equivalent between cultures and required various modifications of the factorial model to ensure that partial invariance could be established. ...
... The MHC-SF is one of the most popular PPAMs aimed at measuring mental health, has shown to produce various factorial models ranging from a correlated three first-order factorial model (comprised of emotional-, psychological-, and social wellbeing) through to various types of bifactor models with varying ranges of reliability (16)(17)(18)(19). Zemojtel-Piotrowska et al. (19) also showed that the MHC-SF is not equivalent between cultures and required various modifications of the factorial model to ensure that partial invariance could be established. Further, the item relating to "positive relationships" on the psychological well-being sub-scale, is strongly related to the social well-being subscale and has shown to load on both constructs in several contexts [cf. ...
... Further, the item relating to "positive relationships" on the psychological well-being sub-scale, is strongly related to the social well-being subscale and has shown to load on both constructs in several contexts [cf. (19)]. In individualistic cultures, a clear distinction between these three factors is apparent, however in collectivistic cultures psychological-and social well-being seem to be tied more closely together (17,19). ...
Article
Full-text available
Critics of positive psychology have questioned the validity of positive psychological assessment measures (PPAMs), which negatively affects the credibility and public perception of the discipline. Psychometric evaluations of PPAMs have shown that various instruments produce inconsistent factor structures between groups/contexts/times frames, that their predictive validity is questionable, and that popular PPAMs are culturally biased. Further, it would seem positive psychological researchers prioritize date-model-fit over measurement quality. To address these analytical challenges, more innovative and robust approaches toward the validation and evaluation of PPAMs are required to enhance the discipline's credibility and to advance positive psychological science. Exploratory Structural Equation Modeling (ESEM) has recently emerged as a promising alternative to overcome some of these challenges by incorporating the best elements from exploratory-and confirmatory factor analyses. ESEM is still a relatively novel approach, and estimating these models in statistical software packages can be complex and tedious. Therefore, the purpose of this paper is to provide novice researchers with a practical tutorial on how to estimate ESEM with a convenient online tool for Mplus. Specifically, we aim to demonstrate the use of ESEM through an illustrative example by using a popular positive psychological instrument: the Mental Health Continuum-SF. By using the MHC-SF as an example, we aim to provide (a) a brief overview of ESEM (and different ESEM models/approaches), (b) guidelines for novice researchers on how to estimate, compare, report, and interpret ESEM, and (c) a step-by-step tutorial on how to run ESEM analyses in Mplus with the De Beer and Van Zy ESEM syntax generator. The results of this study highlight the value of ESEM, over and above that of traditional confirmatory factor analytical approaches. The results also have practical implications for measuring mental health with the MHC-SF, illustrating that a bifactor ESEM Model fits the data significantly better than any other theoretical model.
... Later evaluation studies, however, reported internal consistency of the MHC-SF total scale and its various subscales within the acceptable-to-high range (α = .74 to .94), in various countries such as Hong Kong, India, Japan, Malaysia, Netherlands, United Kingdom, and Vietnam [18,47,48]. ...
... Bifactor model of MHC-SF. Whilst the MHC-SF was developed by Keyes [23] under the assumption of a three-factor structure (i.e., EWB, SWB, and PWB), and prior evaluation studies have demonstrated that a three-factor model to be the best fit in comparison to the single and/or two-factor models, researchers are now suggesting that the three-factor structure may be problematic [e.g., 48]. ...
... A second-order factor analysis, however, does not allow for direct comparisons of strengths between the GWB factor and the group factors (i.e., EWB, SWB, and PWB), and thus is not able to measure any contribution of unique variance by the various subscales [47,52]. Consequently, numerous researchers including de Bruin and du Plessis [47], Jovanović [52], and Żemojtel-Piotrowska, Piotrowski [48] have incorporated a bifactor model in their analyses, where each item loaded on a general factor (i.e., GWB) and simultaneously on one of the uncorrelated group factors (i.e., EWB, SWB, and PWB). Bifactor model analysis has been widely utilised in understanding the structures of multidimensional constructs measures such as self-esteem [53], depression [54], and intelligence [55]. ...
Article
Full-text available
The Mental Health Continuum-Short Form (MHC-SF) is aimed at measuring the three dimensions of mental health; emotional, social, and psychological well-being. The purpose of the current study was to evaluate the psychometric properties of the MHC-SF within the context of Singapore and Australia. A total of 299 Singaporeans or permanent residents (59.2% female; mean age = 24.26, SD = 6.13) and 258 Australians or permanent residents (69% female; mean age = 23.95, SD = 8.66) completed the study. Confirmatory factor analyses were used to assess the structural validity of the MHC-SF. Internal consistency reliability was assessed via the Cronbach’s α and MacDonald’s ω reliability coefficients. Concurrent validity was examined against the World Health Organisation-Five Well-Being Index, discriminant validity using the Hospital Anxiety and Depression Scale, and criterion validity using a self-rated question of “Please rate your averaged level of mental health over the past month”, all via Pearson’s correlations. A bifactor model of the MHC-SF, where each item loaded on a general factor and simultaneously on their respective uncorrelated group factors, yielded the best fit to the data across both samples. Further investigations demonstrated that the general well-being factor accounted for majority of variances of the MHC-SF. Internal consistency reliability, concurrent validity, discriminant validity, and criterion validity were all demonstrated. In conclusion, the current study provided support for the bifactor model of MHC-SF and demonstrated evidence of good psychometrics across both samples. The results highlighted the unidimensionality of the measure, suggesting that it is more informative to interpret the aggregated score than scores of independent factors standalone.
... For the MHC-SF, it should be noted that the CFA for the 3-factor MHC-SF was run, and the GOF measures were below the acceptable standards. However, studies have repeatedly shown that the bifactor model for the MHC-SF outperforms the 3-factor model [35,36], and, accordingly, Table 3 reports on GOF measures of the bifactor model; for more details on the bifactor model, see the studies conducted by [35] and [36]. ...
... For the MHC-SF, it should be noted that the CFA for the 3-factor MHC-SF was run, and the GOF measures were below the acceptable standards. However, studies have repeatedly shown that the bifactor model for the MHC-SF outperforms the 3-factor model [35,36], and, accordingly, Table 3 reports on GOF measures of the bifactor model; for more details on the bifactor model, see the studies conducted by [35] and [36]. ...
Article
Full-text available
There has been a preponderance of studies on student mental health, wellbeing and flourishing during the COVID-19 pandemic. Few studies have compared data on student mental health and wellbeing before and during the pandemic. The purpose of the current study was to compare mental health and wellbeing in undergraduate students before and during the COVID-19 pandemic. Survey research was conducted with three groups of undergraduate students (n = 905) from diverse scientific fields at a large, urban university in South Africa. Data was collected by means of electronic surveys, combining full-scale items from three instruments, the Mental Health Continuum Short Form, the Flourishing Scale and the Fragility of Happiness Scale. Data was analysed by the Statistical Package for the Social Sciences (SPSS), the Analysis of Moment Structures (AMOS) and R software. The results indicate that while the mental health and wellbeing of students declined during the pandemic concerning their perceived ability to contribute to society, having supportive and rewarding social relationships and them being engaged and interested in their daily activities, it also improved in terms of their perceived ability to manage their daily lives (environmental mastery), being challenged to grow (personal growth) and in terms of their views that society was becoming better (social growth/actualisation).
... ,Setswana (Keyes et al., 2008), Portuguese(Machado & Bandeira, 2015), French(Doré et al., 2017), and Dutch(Lamers et al., 2010;Luijten et al., 2019). The original MHC-SF elaborated in the United States(Keyes, 2002; Keyes, 2005b) has been adapted for countries such as Netherlands(Lamers et al., 2010;Luijten et al., 2019), Argentina(Perugini et al., 2017), South Africa(Keyes et al., 2008), Brazil(Machado, & Bandeira, 2015),Canada (Doré et al., 2017), and cross-cultural(Joshanloo et al., 2013;Żemojtel-Piotrowska et al., 2018). Acceptable psychometric properties of the MHC-SF have been found in nationally representative sample of adolescents between the ages of 12 and 18(Keyes, 2005b;Keyes, 2009), school-based sample of adolescent between the ages of 11 and 17(Luijten et al., 2019), postsecondary students(Doré et al., 2017), university students(Żemojtel-Piotrowska et al., 2018), adults (Keyes et al., 2008Perugini et al., 2017;Machado & Bandeira, 2015), general populations ...
... The original MHC-SF elaborated in the United States(Keyes, 2002; Keyes, 2005b) has been adapted for countries such as Netherlands(Lamers et al., 2010;Luijten et al., 2019), Argentina(Perugini et al., 2017), South Africa(Keyes et al., 2008), Brazil(Machado, & Bandeira, 2015),Canada (Doré et al., 2017), and cross-cultural(Joshanloo et al., 2013;Żemojtel-Piotrowska et al., 2018). Acceptable psychometric properties of the MHC-SF have been found in nationally representative sample of adolescents between the ages of 12 and 18(Keyes, 2005b;Keyes, 2009), school-based sample of adolescent between the ages of 11 and 17(Luijten et al., 2019), postsecondary students(Doré et al., 2017), university students(Żemojtel-Piotrowska et al., 2018), adults (Keyes et al., 2008Perugini et al., 2017;Machado & Bandeira, 2015), general populations ...
Preprint
Full-text available
The current study intended to validate the Mental Health Continuum-Short Form (MHC-SF) and Flourishing Scale (FS) in the Malay language. The scales of MHC-SF and FS are used to measure emotional, social and psychological wellbeing. Both instruments have been employed in assessing flourishing mental health and positive human functioning of university students. A total of 131 undergraduate students (29 males and 102 females) from a public university in Sabah aged 19-26 years old participated in the study. Partial least squares-structural equation modeling (PLS-SEM) is used to generate the result of measurement model. The findings showed that the MHC-SF and FS in the Malay language demonstrated a sufficient convergent and discriminant validity. The level of internal consistency for MHC-SF and FS was at an acceptable level. Both Malay versions of MHC-SF and FS have been proven as valid and reliable instruments to be used in the contexts of public undergraduate students in Malaysia, particularly in the state of Sabah.