Development and validation of a bean knowledge, attitudes, and behaviors survey in a college population

Despite recommendations for increased bean intake, high nutritional value as a low-cost source of protein, fiber, and other nutrients, and culinary versatility, very few college students regularly consume beans. This study developed and validated a survey to assess knowledge, attitudes, and behaviors related to bean consumption among college students.

Twenty-six items were generated from a literature review and health behavior theories. An expert panel (n = 6) reduced the pool to 20 items assessing students' bean knowledge, attitudes, and behaviors (content validation). Cognitive interviews (n = 7) with college students tested face validity. In all, 982 undergraduate students aged 18-24 were recruited via convenience sampling in fall 2023 through an introductory nutrition course. Exploratory factor analysis (EFA) was conducted using principal axis factoring and varimax rotation on a random subsample (n = 250). The factor structure was examined using confirmatory factor analysis (CFA) with maximum likelihood estimation (n = 732). Missing data was handled via Multivariate Imputation by Chained Equations. Internal consistency was assessed using Cronbach's alpha and McDonald's omega for multi-item subscales and construct level composite reliability. Cross-validation analysis using 70/30 split, network analysis, and group differences analyses were conducted. Content validity index (CVI) and discriminant validity were also evaluated.

Respondents were primarily female (65.6%), White (90.5%), aged 18-19 (69.6%), and nearly half (46.4%) were freshmen. EFA supported a 4-factor solution (Bean Consumption Behavior, Attitudes and Behavioral Patterns, Perceived Benefits, and Comparative Health Perceptions) explaining 56% variance (Kaiser-Meyer-Olkin (KMO) = .73). CFA showed poor initial model fit (Comparative fit index (CFI) = .840, Tucker-Lewis index (TLI) = .795, root mean square error of approximation (RMSEA) = .096, standardized root mean square residual (SRMR) = .059); after modifications, fit was acceptable (CFI = .914, TLI = .886, RMSEA = .071, SRMR = .047), a significant improvement (Δχ ²[2] = 239.07, p < .001). Cronbach's alpha for multi-item subscales ranged from .63 to .79 (total-scale alpha = .82; omega = .87). The hypothesized correlated four-factor model fit substantially better than one-factor (CFI = .629) and orthogonal (CFI = .706) alternatives. Factor loadings ranged from .338 to .844; composite reliability from .432 to .660. Cross-validation confirmed model stability across subsamples.

This research produced a validated tool to assess college students' knowledge, attitudes, and behaviors regarding bean consumption. Results can inform targeted interventions to promote bean intake, an important priority given increasing food insecurity and poor dietary quality among college students.