Exploratory Factor Analysis of a Survey on Corporate Training in Software Engineering

This study investigates the internal structure validity evidence of a 27-item instrument designed to measure the perceived quality of corporate training in Software Engineering, applied to 282 Brazilian professionals. The package contains the following artifacts: Instrument construction matrix — Full documentation of the four development stages: extraction of 106 items from 20 literature sources, semantic grouping (40 items), Likert scale adaptation (30 items), and cognitive load reduction (27 items). Includes the mapping of each item to the subcategories of the Salas et al. (2001, 2012) framework. Complete questionnaire — Instrument with 9 sociodemographic questions, 27 closed-ended Likert-scale items (5-point), and 2 open-ended questions. Dataset (Likert items) — Aggregated responses from 282 participants to the 27 Likert items used in the factor analysis. FACTOR report — 1-factor solution — Full output from FACTOR software (v12.06.07) using RDWLS extraction, Robust Promin rotation, and number of factors determined by Parallel Analysis (95th percentile). Includes polychoric correlation, KMO, Bartlett's test, MSA, eigenvalues, Parallel Analysis, unidimensionality indicators (UniCo, ECV, MIREAL), factor loadings, communalities, fit indices, and reliability. FACTOR report — 5-factor solution (sensitivity analysis) — Full output from FACTOR using RDWLS extraction forced to 5 factors (Kaiser criterion), Robust Promin rotation, and BCa bootstrap confidence intervals (500 resamples). Includes rotated loading matrix and interfactor correlations.