Participants
The current study used a cross-sectional design, with undergraduate respondents affiliated with four universities in three GCC countries: Oman, Qatar, and Saudi Arabia (NOM = 238, NQA = 215, NSA = 237). The sample comprised 340 male and 350 female participants (mean age = 21.16 ± 2.44). The participants responded to a web-based survey that included a front cover explaining the nature of the study and clarifying that participation is anonymous and voluntary.
Materials
In addition to the demographic information, the study used four measures, all of which were self-evaluated:
Warwick-Edinburgh Mental Well-being Scale (WEMWBS):
A team of researchers from Scottish universities [11] developed this scale. It consists of 14 positively phrased statements to assess mental well-being by using a 5-point response scale (1 = none of the time to 5 = all of the time). Therefore, the total score on the scale ranges from 14 to 70. Respondents must read each statement carefully and determine the extent to which it applies to them during the past two weeks. Thus, higher scale scores indicate higher levels of psychological well-being.
In 2012, the first author of this study translated the scale to Arabic after obtaining official approval from its authors. The internal consistency of the Arabic version of the scale reached 0.91 in the current study.
In 2009, Stewart-Brown et al. [12] developed a shortened version of the Warwick-Edinburgh Mental Well-being Scale (SWEMWBS). This version consists of seven items of the original WEMWBS’s 14 items, namely, items 1, 2, 3, 6, 7, 9, and 11. The total score of the SWEMWBS scale ranges from 7 to 35. In our current study, the internal consistency of the SWEMWBS was high (Cronbach’s α ˃ 0.89).
Basic Psychological Need Satisfaction–Frustration Scale (BPNSFS):
The validated Arabic version of the Basic Psychological Need Satisfaction-Frustration scale [17] was translated into Arabic by the guidance of the second edition of the international test commission for translating and adapting tests (following the guidance of the second edition of the International Test Commission guidance for translating and adapting tests) [18]. The scale consists of two subscales: The first covers need satisfaction and consists of 12 items, while the second covers need frustration and consists of 12 items. The two subscales have three dimensions—autonomy, relatedness, and competence—derived from self-determination theory [6]. Official permission was obtained from the scale’s creators before translating it into Arabic. The internal consistency of the Arabic version of the BPNSFS varied from α ˃ 0.81 [17] to α ˃ 0.91 in our current study.
Beck Depression Inventory-II (BDI-II):
This scale is widely used to assess depressive symptoms. It consists of 21 groups of items representing all aspects of depressive symptoms. In each group, the respondent has to choose one of the four answers as follows: the first illustrates no depressive symptoms present, the second indicates that mild symptoms are present, the third indicates that medium symptoms are present, while the fourth answer illustrates that severe symptoms are present. The internal consistency of the Arabic version of BDI-II ranged from an average of α ˃ 0.86 [19] to α ˃ 0.85 in our current study.
Procedures
Participants in the current study were recruited from universities in three GCC countries. Paper questionnaires were distributed to the participants. The questionnaires included demographic questions and three instruments: the WEMWBS, BDI-II, and BPNSFS. The response period took approximately 20 minutes.
Data analysis
Using SPSS 25, descriptive statistics, including means, standard deviations, skewness, and kurtosis, were calculated for the WEMWBS and SWEMWBS. We utilized skewness and kurtosis to determine whether the data were normal. When both findings fall within the range of + 2 to -2, normality is declared [20]. Cronbach's alpha coefficients corrected item–total correlation, and the Spearman item correlation matrix was produced to evaluate the internal consistency of the complete and shortened versions of the scale. To evaluate how accurately the scale measures the outcome it was designed to measure, convergent validity was evaluated by investigating the correlation coefficients of the scores on the full and shortened scales with other related and well-established instruments. The need satisfaction, need frustration, and depression scales were all used to cross-check the original WEMWBS and SWEMWBS.
Furthermore, the constructor factorial validity of the WEMWBS and SWEMWBS was investigated using confirmatory factor analysis on the entire sample. Measurement invariance was carried out across countries and genders to ensure that the WEMWBS and SWEMWBS were comparable. To test measurement invariance and establish cross-group comparisons, a series of iterative processes were used. We looked at the construct's stability across countries and genders through configurable invariance. Metric invariance was also used to assess the comparability of item factor loadings across countries and gender subgroups. Metric invariance compares the strength of the relationships between the WEMWBS and SWEMWBS items and their underlying components across groups. Scalar invariance was further examined by restricting item intercepts across country and gender subgroups to be equal. This level of scale measurement invariance is essential to compare the latent means among the intended study groups [21]. AMOS-20 was used for the confirmatory factor analyses. The model fit was investigated using the cut-off values for the following indices: Tucker‒Lewis index (TLI), comparative fit index (CFI; model fit is good when it exceeds 0.95 and acceptable when it exceeds 0.90) [22], and root mean square error of approximation (RMSEA; when values are less than 0.08, the fit is acceptable; when values are less than 0.05, the fit is good). RMSEA and CFI values less than or equal to 0.015 and 0.01, respectively, were employed as model comparison criteria [23].
Table 1
Descriptive statistics and reliability indices for the WEMWBS and SWEMWBS
Items
|
Mean
|
Std. Deviation
|
Skewness
|
Kurtosis
|
Corrected Item–Total Correlation
|
Cronbach's Alpha if Item Deleted
|
MH1
|
3.90
|
1.158
|
-0.851
|
-0.134
|
0.522
|
0.859
|
MH2
|
3.96
|
1.095
|
-0.902
|
0.054
|
Corrected
|
0.855
|
MH3
|
3.52
|
1.124
|
-0.395
|
-0.444
|
0.592
|
0.855
|
MH4
|
3.63
|
1.214
|
-0.509
|
-0.664
|
0.464
|
0.862
|
MH5
|
3.48
|
1.142
|
-0.406
|
-0.478
|
0.556
|
0.857
|
MH6
|
3.77
|
1.032
|
-0.600
|
-0.180
|
0.513
|
0.859
|
MH7
|
3.53
|
1.100
|
-0.389
|
-0.509
|
0.490
|
0.860
|
MH8
|
4.06
|
1.038
|
-1.009
|
0.498
|
0.584
|
0.856
|
MH9
|
3.72
|
1.112
|
-0.644
|
-0.184
|
0.499
|
0.860
|
MH10
|
4.09
|
1.025
|
-0.991
|
0.456
|
0.539
|
0.858
|
MH11
|
4.06
|
1.036
|
-1.055
|
0.642
|
0.512
|
0.859
|
MH12
|
3.89
|
0.986
|
-0.625
|
-0.106
|
0.495
|
0.860
|
MH13
|
4.12
|
1.014
|
-0.987
|
0.281
|
0.438
|
0.863
|
MH14
|
3.69
|
1.141
|
-0.656
|
-0.195
|
0.541
|
0.858
|
Descriptive Statistics And Internal Consistency
The means, standard deviations, skewness, kurtosis, and corrected item–total correlations for each of the 14 WEMWBS items (N = 690) are displayed in Table 1. For the 14-item WEMWBS, the corrected item-to-total correlations varied from 0.438 to 0.594. It is worth noting that none of the items have severe skewness or kurtosis since all of the values are between 2 and − 2, indicating that all of the items' responses are normal [20]. Furthermore, the Spearman correlation matrix table that shows the relationships between all 14 items is presented in Table 2. The majority of the correlation coefficients were greater than 0.300, which was suitable for factor analysis [24]. The WEMWBS and SWEMWBS Cronbach's alphas were 0.867 and 0.772, respectively, which indicates good reliability. Additionally, the Cronbach alpha coefficients were comparable to the original Cronbach alpha value of the WEMWBS (0.890) and other relevant research [25, 11, 26]. The WEMWBS and SWEMWBS correlated with a Cronbach alpha value of 0.943 (p < 0.001), which was close to the 0.954 reported by the original developers [12].
Table 2
Spearman correlation matrix for the WEMWBS items
Items
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
8
|
9
|
10
|
11
|
12
|
13
|
14
|
MH1
|
1.00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
MH2
|
0.47
|
1.00
|
|
|
|
|
|
|
|
|
|
|
|
|
MH3
|
0.41
|
0.40
|
1.00
|
|
|
|
|
|
|
|
|
|
|
|
MH4
|
0.27
|
0.32
|
0.29
|
1.00
|
|
|
|
|
|
|
|
|
|
|
MH5
|
0.32
|
0.33
|
0.40
|
0.37
|
1.00
|
|
|
|
|
|
|
|
|
|
MH6
|
0.21
|
0.30
|
0.32
|
0.24
|
0.45
|
1.00
|
|
|
|
|
|
|
|
|
MH7
|
0.25
|
0.30
|
0.36
|
0.24
|
0.40
|
0.50
|
1.00
|
|
|
|
|
|
|
|
MH8
|
0.36
|
0.45
|
0.46
|
0.24
|
0.28
|
0.40
|
0.34
|
1.00
|
|
|
|
|
|
|
MH9
|
0.25
|
0.31
|
0.26
|
0.38
|
0.34
|
0.29
|
0.28
|
0.28
|
1.00
|
|
|
|
|
|
MH10
|
0.37
|
0.40
|
0.36
|
0.27
|
0.24
|
0.30
|
0.26
|
0.45
|
0.29
|
1.00
|
|
|
|
|
MH11
|
0.27
|
0.38
|
0.34
|
0.22
|
0.25
|
0.34
|
0.33
|
0.35
|
0.31
|
0.46
|
1.00
|
|
|
|
MH12
|
0.31
|
0.34
|
0.30
|
0.36
|
0.24
|
0.23
|
0.29
|
0.32
|
0.37
|
0.34
|
0.34
|
1.00
|
|
|
MH13
|
0.33
|
0.34
|
0.25
|
0.23
|
0.24
|
0.26
|
0.29
|
0.31
|
0.27
|
0.30
|
0.27
|
0.34
|
1.00
|
|
MH14
|
0.33
|
0.34
|
0.48
|
0.27
|
0.48
|
0.33
|
0.27
|
0.38
|
0.35
|
0.24
|
0.24
|
0.26
|
0.30
|
1.00
|
Convergent Validity
The WEMWBS was found to have substantial moderate to high positive correlations with domains such as positive affect, life satisfaction, and overall health, as well as a significant negative correlation with symptoms of anxiety and depression in earlier investigations [11, 27, 28, 29, 30]. This study reported the correlation between the overall WEMWBS and SWEMWBS, as well as the individual items and other construct-related measures (see Table 3). A significant positive correlation was observed with needs satisfaction (RWEMWBS = 0.552, RSWEMWBS = 0.544, p < 0.01) and a significant negative correlation with needs frustration (RWEMWBS = -0.389, RSWEMWBS = -0.359, p < 0.01) and depression (RWEMWBS = -0.373, RSWEMWBS = -0.337, p < 0.01). The correlation coefficients of the WEMWBS and the SWEMWBS items with the three constructs were consistent with their counterparts at the level of overall WEMWBS and the SWEMWBS scores, and they ranged between low and moderate relationships, as shown in Table 3.
Table 3
WEMWBS and SWEMWBS item correlations with other construct-related scales
Items
|
Needs Satisfaction
|
Needs Frustration
|
Depression
|
MH1#
|
.437**
|
− .315**
|
− .142**
|
MH2#
|
.378**
|
− .310**
|
− .197**
|
MH3#
|
.375**
|
− .322**
|
− .331**
|
MH4
|
.217**
|
− .117*
|
− .250**
|
MH5
|
.241**
|
− .176**
|
− .228**
|
MH6#
|
.349**
|
− .169**
|
− .231**
|
MH7#
|
Full-scale
|
− .140**
|
− .205**
|
MH8
|
.362**
|
− .339**
|
− .233**
|
MH9#
|
.210**
|
− .163**
|
− .214**
|
MH10
|
.440**
|
− .338**
|
− .268**
|
MH11#
|
.422**
|
− .221**
|
− .225**
|
MH12
|
.373**
|
− .255**
|
− .291**
|
MH13
|
.352**
|
− .218**
|
− .128**
|
MH14
|
.233**
|
− .210**
|
− .215**
|
WEMWBS
|
.552**
|
− .389**
|
− .373**
|
SWEMWBS
|
.544**
|
− .359**
|
− .337**
|
Note: * p < 0.05; ** p < 0.01 |
Factorial Validity Of The Full-scale And Short Scale
Table 4 displays the results of the confirmatory factor analysis performed on the 14-item WEMWBS and the 7-item SWEMWBS. The whole sample was used to conduct CFA for both the full and shortened versions. Model 1 assessed the WEMWBS on its whole scale, with no relationships between measurement errors. Model 1's findings indicated that the scale did not fit the model well, as indicated by the values of χ2 (77) = 462.38, p < 0.001, RMSEA = 0.085, TLI = 0.834, CFI = 0.860 and SRMR = 0.057. Some covariances between the error factors reported in the notes of Table 4 were added to Model 2. With χ2 (68) = 240.36, p < 0.001, RMSEA = 0.061, TLI = 0.916, CFI = 0.937 and SRMR = 0.042, the findings showed good model fit.
CFA was performed on Model 3 for the shortened version of the 7-item scale without associating the error terms. As the values of χ2 (14) = 69.47, p < 0.001, RMSEA = 0.108, TLI = 0.812, CFI = 0.874 and SRMR = 0.063 showed, this scale did not meet the cut-off values. Model 4 used the error correlations based on the modification indices to evaluate the SWEMWBS. The findings indicated that the SWEMWBS may be used as a shortened version of the WEMWBS. The values of χ2 (12) = 15.38, p = 0.221, RMSEA = 0.029, TLI = 0.987, CFI = 0.992 and SRMR = 0.030 showed a good level of model fit. In Model 4, the covariances between the error factors were between MH1 & MH2 and between MH6 & MH6. The results demonstrated that both the WEMWBS and SWEMWBS had a generally satisfactory match for one underlying structure after post hoc adjustment.
The factor loadings of the items in both WEMWBS and SWEMWBS are presented in Table 5. The factor loadings were greater than 0.40 for the WEMWBS and SWEMWBS items in the three countries and the gender groups. These results also support the construct validity of the full and shortened versions.
Table 4
Confirmatory factor analysis of WEMWBS and SWEMWBS
Model
|
\({{\chi }}^{2}\)
|
DF
|
p
|
CFI
|
TLI
|
RMSEA
|
90% CI for RMSEA
|
SRMR
|
WEMWBS
|
|
|
|
|
|
|
|
|
1
|
462.38
|
77
|
0.000
|
0.860
|
0.834
|
0.085
|
0.078–0.093
|
0.057
|
2*
|
240.36
|
68
|
0.000
|
0.937
|
0.916
|
0. 061
|
0.052–0.069
|
0.042
|
SWEMWBS
|
|
|
|
|
|
|
|
|
3
|
69.47
|
14
|
0.000
|
0.874
|
0.812
|
0.108
|
0.084–0.134
|
0.063
|
4**
|
15.38
|
12
|
0.221
|
0.992
|
0.987
|
0.029
|
0.000-0.066
|
0.030
|
* The adjusted model includes the covariance between the error terms for the items MH1 & MH2, MH3 & MH5, MH6 & MH7, MH10 & MH11, MH8 & MH10, MH5 & MH14, MH3 & MH14, MH5 & MH6, and MH4 & MH12.
** Includes the covariance between the error terms for the items MH1 & MH2 and MH6 & MH7.
Table 5
Factor loadings from CFA of WEMWBS and SWEMWBS
Items
|
WEMWBS
|
SWEMWBS
|
OM
|
SA
|
QA
|
Male
|
Female
|
OM
|
SA
|
QA
|
Male
|
Female
|
MH1
|
0.56
|
0.51
|
0.54
|
0.55
|
0.61
|
0.42
|
0.48
|
0.43
|
0.45
|
0.52
|
MH2
|
0.66
|
0.60
|
0.64
|
0.64
|
0.68
|
0.60
|
0.64
|
0.58
|
0.61
|
0.65
|
MH3
|
0.64
|
0.60
|
0.62
|
0.57
|
0.66
|
0.51
|
0.61
|
0.72
|
0.60
|
0.60
|
MH4
|
0.48
|
0.50
|
0.48
|
0.58
|
0.43
|
|
|
|
|
|
MH5
|
0.60
|
0.49
|
0.53
|
0.59
|
0.51
|
|
|
|
|
|
MH6
|
0.58
|
0.48
|
0.54
|
0.44
|
0.58
|
0.60
|
0.37
|
0.62
|
0.45
|
0.62
|
MH7
|
0.57
|
0.46
|
0.53
|
0.46
|
0.55
|
0.64
|
0.40
|
0.59
|
0.46
|
0.63
|
MH8
|
0.69
|
0.61
|
0.63
|
0.55
|
0.72
|
|
|
|
|
|
MH9
|
0.58
|
0.49
|
0.52
|
0.63
|
0.46
|
0.64
|
0.46
|
0.37
|
0.59
|
0.43
|
MH10
|
0.64
|
0.49
|
0.62
|
0.64
|
0.53
|
|
|
|
|
|
MH11
|
0.60
|
0.50
|
0.55
|
0.54
|
0.56
|
0.67
|
0.56
|
0.53
|
0.56
|
0.64
|
MH12
|
0.55
|
0.51
|
0.58
|
0.61
|
0.48
|
|
|
|
|
|
MH13
|
0.54
|
0.41
|
0.50
|
0.50
|
0.50
|
|
|
|
|
|
MH14
|
0.61
|
0.53
|
0.52
|
0.47
|
0.61
|
|
|
|
|
|
Cross-cultural Factorial Invariance Of The Full And Short Scales
For the Arabic versions of the WEMWBS and SWEMWBS, multigroup CFA comparisons were carried out to assess the degree of measurement invariance among the study subgroups. AMOS-22 was used to conduct the series of comparisons, and the outcomes are shown in Table 5.
Measurement invariance of the WEMWBS
For the configural across-country invariance that was tested in Model A1, the results, χ2 (204) = 413.73, p < 0.001, RMSEA = 0.039, TLI = 0.900 and CFI = 0.924, indicated acceptable model fit, which demonstrated that the construct can be assumed to be the same across countries. Therefore, the pattern of loadings of items on the latent factor does not differ across cultures. The same results were indicated for configural invariance, as the fit indices were at acceptable levels, χ2 (136) = 333.66, p < 0.001, RMSEA = 0.046, TLI = 0.908 and CFI = 0.931.
Following testing of the configural invariance, the metric invariance was also tested by constraining the loadings of the items on the constructs to be equivalent in the culture and gender subgroups. For the metric invariance, the factor loadings can be assumed to be the same across countries, and across gender, \(\varDelta\)CFI and \(\varDelta\)RMSEA were found to be acceptable (\(\varDelta\)CFI < 0.01 and \(\varDelta\)RMSEA < 0.015).
Additionally, as shown in Table 6 bel, the item intercepts are limited to being equal across the country and gender subgroups to test for scalar invariance. Full scalar invariance across countries and across gender was not caused by the difference between fit indices for scalar invariance, which was slightly outside of acceptable levels for Models A3 and B3 (A3: \(\varDelta\)CFI = − 0.025 and \(\varDelta\)RMSEA = 0.003; B3: \(\varDelta\)CFI = − 0.052 and \(\varDelta\)RMSEA = 0.011). Therefore, partial scalar invariance across cultures and across genders was tested by g the intercepts of items that exhibited a great difference among the subgroups. The intercepts of Items 1, 2, and 12 were found to have the highest change across country groups, and the intercepts of Items 1 and 2 were found to differ across gender subgroups. Therefore, we iteratively relaxed the constraints on intercepts of MH1, MH2, and MH12 to be free across countries and the intercepts of MH1 and MH2 to be free for males and females. For a cross-country subgroup, the improvement in fit indices supports the partial scalar invariance across countries (A3a: \(\varDelta\)CFI = − 0.008 and \(\varDelta\)RMSEA = 0.000). Across gender subgroups, \(\varDelta\)CFI = − 0.012, indicating that partial scalar invariance was not met.
Table 6
WEMWBS and SWEMWBS measurement invariance across countries and gender groups
Scales/Level of MI
|
Model fit indices
|
Model comparison indices
|
\({\chi }^{2}\)
|
DF
|
CFI
|
TLI
|
RMSEA
|
\(\varDelta {\chi }^{2}\)
|
df
|
CFI
|
RMSEA
|
WEMWBS
|
|
|
|
|
|
|
|
|
|
Cross Country
|
|
|
|
|
|
|
|
|
|
A1: Configural
|
413.73
|
204
|
0.924
|
0.900
|
0.039
|
-
|
-
|
-
|
-
|
A2: Metric
|
455.28
|
230
|
0.919
|
0.903
|
0.038
|
41.55
|
26
|
-0.005
|
-0.001
|
A3: Scalar
|
551.86
|
258
|
0.894
|
0.888
|
0.041
|
96.58
|
28
|
-0.025
|
0.003
|
A3a: Partial Scalar (1, 2, 12)
|
498.66
|
252
|
0.911
|
0.904
|
0.038
|
43.38
|
22
|
-0.008
|
0
|
Cross Gender
|
|
|
|
|
|
|
|
|
|
B1: Configural
|
333.66
|
136
|
0.931
|
0.908
|
0.046
|
-
|
-
|
-
|
-
|
B2: Metric
|
373.05
|
149
|
0.922
|
0.905
|
0.047
|
39.39
|
13
|
-0.009
|
0.001
|
B3: Scalar
|
534.98
|
163
|
0.870
|
0.855
|
0.058
|
161.93
|
14
|
-0.052
|
0.011
|
B3a: Partial Scalar (1, 2)
|
442.08
|
158
|
0.901
|
0.886
|
0.051
|
69.03
|
9
|
-0.021
|
0.004
|
SWEMWBS
|
|
|
|
|
|
|
|
|
|
Cross Country
|
|
|
|
|
|
|
|
|
|
C1: Configural
|
30.73
|
30
|
0.999
|
0.998
|
0.006
|
-
|
-
|
-
|
-
|
C2: Metric
|
54.25
|
42
|
0.987
|
0.981
|
0.021
|
23.52
|
12
|
-0.012
|
0.015
|
C2a: Partial Metric (9)
|
47.94
|
40
|
0.992
|
0.987
|
0.017
|
17.21
|
10
|
-0.007
|
0.011
|
C3: Scalar
|
113.94
|
56
|
0.940
|
0.933
|
0.039
|
66
|
16
|
-0.052
|
0.022
|
C3a: Partial Scalar (1, 2)
|
69.62
|
50
|
0.982
|
0.975
|
0.024
|
21.68
|
10
|
-0.010
|
0.007
|
Cross Gender
|
|
|
|
|
|
|
|
|
|
D1: Configural
|
23.36
|
22
|
0.999
|
0.997
|
0.009
|
-
|
-
|
-
|
-
|
D2: Metric
|
35.88
|
28
|
0.992
|
0.988
|
0.020
|
12.52
|
6
|
-0.007
|
0.011
|
D3: Scalar
|
86.85
|
35
|
0.949
|
0.939
|
0.046
|
50.97
|
7
|
-0.043
|
0.026
|
D3a: Partial Scalar (1, 2)
|
57.46
|
33
|
0.983
|
0.970
|
0.033
|
21.58
|
5
|
-0.009
|
0.013
|
Measurement invariance of the SWEMWBS
The findings of Models C1 and D1 asserted that configural invariance was met for both country and gender subgroups, as indicated by the findings that χ2 (204) = 413.73, p < 0.001, RMSEA = 0.039, TLI = 0.900 and CFI = 0.924, and χ2 (204) = 413.73, p < 0.001, RMSEA = 0.039, TLI = 0.900 and CFI = 0.924. Therefore, the pattern of loadings of items on the latent factor does not differ in the three cultures or for males and females.
For the metric invariance, the factor loadings can be assumed to be equal across genders, as the \(\varDelta\)CFI and \(\varDelta\)RMSEA were found to be within the acceptable range (D2: \(\varDelta\)CFI = − 0.007 and \(\varDelta\)RMSEA = 0.011). Moreover, the partial metric invariance was met across country subgroups by relaxing the factor loading of the MH9 to be free across the three countries, which improved the values of the comparison fit indices to fall within the acceptable levels (C2a: \(\varDelta\)CFI = − 0.007 and \(\varDelta\)RMSEA = 0.011).
Additionally, the item intercepts were limited to being equal across the country and gender subgroups to test for scalar invariance. Full scalar invariance across countries and across gender was not met because the difference between fit indices for scalar invariance Models C3 and D3 were out of acceptable levels (C3: \(\varDelta\)CFI = − 0.052 and \(\varDelta\)RMSEA = 0.022; D3: \(\varDelta\)CFI = − 0.043 and \(\varDelta\)RMSEA = 0.026). Therefore, partial scalar invariance across cultures and genders was tested by relaxing the intercepts of items that exhibit a great difference among the subgroups. After the intercepts of MH1 and MH2 were relaxed to be free across countries and gender, the results showed an improvement in fit indices that support the partial scalar invariance across the country and gender subgroups (C3a: \(\varDelta\)CFI = − 0.010 and \(\varDelta the\)RMSEA = 0.007; D3a: \(\varDelta\)CFI = − 0.009 and \(\varDelta\)RMSEA = 0.013).