Table 1.
Established measures used to test concurrent and discriminant validity of the Unmind Index.
Reliability estimates are averages across USA and ANZ samples.
Fig 1.
The second-order factor structure used for the Unmind Index.
Table 2.
Fit indices for the second-order factor model fit to data from USA and Australia/New Zealand (ANZ).
Table 3.
Standardised item-to-factor loadings and residual variance for the second-order factor model, fit to data from USA and ANZ.
Table 4.
Mean and standard deviations of raw scores for each subscale and total score (from 0 to 5), and standardised loadings (± standard errors) of each subscale onto the second-order mental health and wellbeing factor.
Table 5.
Pearson correlation coefficients (± standard errors) for correlations between Unmind Index subscale scores.
Coefficients for USA are shown above the diagonal, and values for ANZ below.
Fig 2.
Correlation residuals greater than 0.1 in absolute value for the second-order (A) and correlated-factors (B) CFA models, for the USA sample. Large residuals reflect ways in which a model fails to fully capture the correlation between pairs of items.
Table 6.
Internal consistency (Cronbach’s α and McDonald’s ω) and test-retest reliability, ICC(C,1) for subscales and total scores, by location.
Values in brackets show 95% confidence intervals.
Fig 3.
Dis-attenuated absolute correlation coefficients between Unmind Index scores and existing measures for the USA sample.
Values in red show correlations with mental health and wellbeing measures predicted to correlate most strongly with the Unmind Index subscale in question. Values in blue show personality measures, which were expected to correlate most weakly with all scales. Error bars show standard error.
Fig 4.
Dis-attenuated absolute correlation coefficients between Unmind Index scores and existing measures for the ANZ sample.
Table 7.
Measurement invariance results by age and gender, within USA and ANZ.
Table 8.
Tests of measurement invariance for the Unmind Index and its individual subscales across locations (UK, USA, and ANZ).
Results indicate that strong measurement invariance holds for the full Unmind Index, and individually for each subscale.
Fig 5.
Unmind Index total scores by location, gender, and age, for Study 2.
A. Means (± standard errors) by location and gender, B. LOESS smoothed estimates (± standard errors) by age, gender, and location.
Table 9.
Benchmark means (± standard deviations) of total Unmind Index scores by location, gender, and age group, from Study 2.
Table 10.
Linear model coefficients for effects of location, gender, and age group on total Unmind Index score, Study 2.