I estimated a Rasch model with the eRm-Package via the RM() function in R with a data set (w_demo) that contains four binary items (W11_s:W14_s; 0/1) and 1510 cases.
w_demo <- rent21 %>%
select(W11_s:W14_s)
# Rasch-Model (CML-estimator, conditional maximum likelihood)
rm.demo <- eRm::RM(w_demo)
tmp1 <- eRm::person.parameter(rm.demo)
pp_ml <- coef(tmp1)
In the default-settings, there should be an estimation of the person.parameter for raw scoures 0 and 4 (via extrapolation), but I get NAs for these cases (which is described in the documentation of the package). Can anyone explain why this is the case? (I didn't find any clues).
Thank you.
I tried to set the argument
extrapolated=TRUE
of the coef() function in case the default is changed, but the result was the same (NA for the cases with a raw score 0 or 4).
