A Thai team developed the SHL-US for undergraduates from a 50-item pool reviewed by 13 experts and piloted in 50 students, then used independent samples for exploratory factor analysis (n=300) and confirmatory factor analysis (n=1,235). The final 45 items cover access (12), understanding (12), appraisal (10) and application (11). Overall Cronbach's alpha was .976 and the four-factor model returned CFI .914, TLI .908, RMSEA .055 and SRMR .023. However, very high factor correlations and HTMT values of .868–.959 led the authors to recommend an overall score rather than treating four subscale scores as distinct constructs.
Key findings
- EFA retained 47 items and explained 53.17% of variance; CFA removed two items loading below .50, leaving a 45-item SHL-US across four domains.
- The content-validity index was .94, item-total correlations ranged from .284 to .755, and overall Cronbach's alpha was .976, indicating very high internal consistency.
- CFA indices were χ²=4150.084, df=931, χ²/df=4.458, GFI=.841, AGFI=.823, CFI=.914, TLI=.908, RMSEA=.055, NFI=.892 and SRMR=.023, meaning some indices were favourable while others fell below common conventions.
- Domain composite reliability was .919–.940 and omega .910–.935, but convergent validity was supported for only three domains because Access had AVE .486, below .50.
- Factor correlations of .884–.956 and HTMT of .868–.959 exceeded common thresholds, so the four domains lacked discriminant validity and support primarily an overall score.
Why this matters globally
A student-specific measure can give Thai sexual-health education programmes a consistent baseline and follow-up outcome, while adding Southeast Asian evidence to a field often reliant on Western instruments. Cross-country or cross-gender use still requires translation, invariance testing and external validation. Scores should not stigmatise individuals or substitute for evaluation of sexual-health services.
Thai researcher contribution
Khon Kaen University and Valaya Alongkorn Rajabhat University under Royal Patronage created a tool grounded in Thai undergraduate context, sampling public and private universities across four regions and providing the questionnaire as supporting information. This supplies an inspectable research infrastructure for further refinement.
Limitations to consider
The CFA sample included only one public and one private university per region; voluntary participation and unreported cluster counts may introduce selection effects. Ordinal items were estimated from Pearson rather than polychoric covariances/WLSMV, and clustering was not modelled. Discriminant validity failed. LGBTQ+ students comprised 23.23%, yet some content was heteronormative and no gender/sexual-orientation invariance or DIF was tested. Test–retest, criterion, concurrent, predictive and known-groups validity and social-desirability assessment remain absent. Cross-sectional scores therefore do not demonstrate behavioural prediction or responsiveness to intervention.