Abstract
Background
University dropout remains a major challenge in higher education, particularly in health sciences programs, where academic demands are high and retention is critical for workforce sustainability. Psychometrically sound instruments for assessing constructs associated with the dropout process remain limited, especially in Latin American contexts. This study aimed to evaluate the psychometric properties and measurement invariance of a
Screening Instrument for Students At-Risk of Dropping Out
among health sciences students from public and private universities.
Methods
A cross-sectional methodological study was conducted using secondary data from two universities in Cali, Colombia (public:
n
= 320; private:
n
= 437; total
n
= 757). Confirmatory factor analysis (CFA) was performed to assess the three-factor structure of the instrument (satisfaction with education, academic burnout, and dropout intention). Internal consistency was evaluated using McDonald's omega (
ω
) and ordinal alpha (
α
). Convergent and discriminant validity were assessed through average variance extracted (AVE) and heterotrait-monotrait ratio (HTMT), respectively. Measurement invariance (configural, metric, and scalar) was examined across institutional context, gender, and working status. Latent mean differences were subsequently estimated.
Results
The three-factor model showed an acceptable fit (CFI = 0.964, TLI = 0.954, SRMR = 0.057, RMSEA = 0.080). Factor loadings were significant and substantial (
λ
= 0.706–0.964). Internal consistency was high across subscales (
ω
= 0.863–0.978;
α
= 0.860–0.978). AVE values exceeded 0.50, and HTMT values were below 0.85, supporting convergent and discriminant validity. Measurement invariance was supported across all grouping variables (
Δ
CFI ≤ 0.006;
Δ
RMSEA ≤ 0.005). Latent mean comparisons indicated that students from the participating public university reported higher levels of satisfaction with education, academic burnout, and dropout intention than those from the participating private university. Female students showed lower dropout intention than males.
Conclusions
The instrument demonstrated adequate psychometric properties and measurement invariance across key groups, supporting its use for comparative analyses. These findings support its use for assessing and comparing satisfaction with education, academic burnout, and dropout intention among health sciences students. However, these constructs, particularly dropout intention, should not be interpreted as equivalent to actual dropout risk. Further longitudinal research is needed to determine its ability to predict actual dropout and support individual-level screening decisions.