Measuring Bullying Victimization in Albanian Schools: Ordinal Factor Analysis, Reliability, and Item Response Evidence from Seven Revised Olweus Indicators

Valid interpretation of school bullying surveys requires evidence about the indicators used and their scoring. This study examined the internal structure, internal consistency, and item-response properties of seven victimization-form indicators derived from the Albanian Revised Olweus Bully/Victim Questionnaire (ROBVQ). Archived survey records covered 2,377 students in Grades IV–XII at 144 schools; complete-case analyses used 2,148 students in 143 schools. A comparison with the 229 excluded students examined sex, grade, and urban–rural school location. Five indicators represented single questions and two combined question pairs. Ordinal one-factor confirmatory factor analysis (CFA) used weighted least squares mean- and variance-adjusted estimation (WLSMV), school-cluster-robust inference, and five original response categories; dichotomous and three-category models examined coding sensitivity. A three-category graded response model (GRM), internal-consistency estimates, and exploratory residual and subgroup diagnostics provided complementary evidence. The five-category CFA showed favorable fit, scaled ?²(14) = 20.76, p = .108, robust CFI = .993, TLI = .989, RMSEA = .048 (90% CI [.000, .084]), and SRMR = .020; standardized loadings were .768–.835. Five-category Cronbach’s ? was .835 (school-cluster bootstrap 95% CI [.790, .869]). GRM discrimination ranged from 2.25 to 2.68, while exploratory diagnostics suggested possible local dependence. These findings support coherent, internally consistent measurement of seven selected victimization forms in this Albanian sample, without establishing full-questionnaire validity, group invariance, or individual screening cutoffs.