We propose a latent variable model for binary responses that account
for outliers or over-represented response patterns.
Outliers are considered to be those response patterns that are not
fitted by the hypothesized model.
Outliers are expected to be generated by secondary response strategies.
The proposed model is an extended latent trait model that models the
guessing mechanism through an unobserved pseudo-item.
The modeling approach proposed estimates simultaneously the parameters
that describe both the primary and the secondary response strategies.
Covariates are used to identify the demographic characteristics of the
guessers. The model is estimated using an E-M algorithm.