Structural Validation of a Sun-Exposure and Photoprotection Questionnaire for Physically Active Adults in Latin America

Background: People who train outdoors accumulate higher ultraviolet doses than the general population, yet the instruments used to quantify their photoprotection were developed in high-latitude, English-speaking settings and are typically scored as undifferentiated summative indices whose internal structure is never examined. We aimed to test the internal structure of a Spanish-language sun-exposure and photoprotection questionnaire for physically active Latin American adults and to refine its item pool through a convergent principal component analysis (PCA) and k-means procedure. Methods: This is an instrumental cross-sectional study. Of 297 responses, 257 physically active adults from Costa Rica, Colombia, Mexico and Cuba (71.6% men) provided valid consent. The 14 items, which shared a five-point ordinal metric, were analyzed. Components were retained by parallel analysis (1000 random matrices) and rotated with varimax, and item-to-component assignment was corroborated by k-means clustering of the row-normalized loading vectors, with k fixed by the silhouette index. Sensitivity analyses used polychoric correlations, a common-factor model with oblique rotation and multiple imputation. Reliability used α, ordinal α and ω with 95% confidence intervals (CIs); structural replicability used 500 random split-halves and Tucker congruence coefficients. Results: Sampling adequacy was satisfactory (KMO = 0.71; Bartlett χ2 (91) = 1041.1, p < 0.001). Kaiser, scree, and parallel analysis converged on four components (61.5% of variance). k-means selected k = 4 (silhouette = 0.731) and reproduced the PCA assignment exactly (adjusted Rand index = 1.00). Polychoric-based and common-factor solutions reproduced the same structure (Tucker φ ≥ 0.96). One hybrid item was removed (cross-loading 0.568/0.413; silhouette 0.252). The 13-item solution explained 62.5% of variance across four dimensions: pro-photoprotection attitude (α = 0.78, 95% CI 0.73–0.82; ω = 0.78), topical photoprotection and skin surveillance (α = 0.66, 95% CI 0.58–0.72; ω = 0.74), pro-tanning attitude (α = 0.79, 95% CI 0.74–0.83; ω = 0.82) and physical barriers and accessories (α = 0.65, 95% CI 0.56–0.72; ω = 0.71). Median congruence across split-halves ranged from 0.936 to 0.966. Inter-dimension correlations were low (ρ ≤ 0.24), and associations with sex, country, discipline, peer norms and perceived tanning desirability provided initial evidence of criterion validity. Conclusions: The 13-item instrument provides initial evidence of a four-dimensional structure, stable across random split-halves of a pooled multinational sample, that separates attitude from behavior and topical photoprotection from physical barriers. Content validity, confirmation in an independent sample, test–retest reliability and cross-country measurement invariance remain to be established before the instrument can be considered fully validated, and the behavioral dimensions are currently suitable for group-level rather than individual-level use.