Hi,
I am trying to run cross-classified IRT or CFA multi-level analysis with BRMS on ordinal data. My question is regarding the model fit assessment, how can I know if my modeI meets recommended guidelines and fulfill criteria of unidimensionality as in traditional analysis ? I sincerely welcome your suggestions and recommendations on running such complex model. Please find below the code for model specification.
Define the model formula with item-specific thresholds and discriminations
irt_formula ← bf(
value ~ 1 + (1 | id) + (1 | PROF) + (1 | INST) +
(0 + F1 + F2 + F3 + F4 + F5 | o | id) +
(0 + F1 + F2 + F3 + F4 + F5 | p | PROF) +
(0 + F1 + F2 + F3 + F4 + F5 | c | INST),
disc ~ 0 + item_id,
family = cumulative(“logit”, thres = “flexible”)
)