I am estimating
standard error for
IRT-3PL models. For
IRT-2PL, some of the standard errors are equal to
NA while in IRT-3PL all standard errors are equal to
NA. To clarify more, I have around 30 items, which the standard error for 28 of them is correct while the standard error for two of them is NA (this is for IRT-2PL). In IRT-3PL, the standard error for all 30 items is equal to
What would be the reason for getting these NA?
Here is some statistic about the data:
Number of users: 144
Number of items:30
Number of total user-item transaction =2646
user-item response is binary (0 or 1), i.e. if a user could answer a question (item) correctly, the value is 1 otherwize 0.
FYI, I am using the
ltm package in
R to estimate the standard error.