Example Code: predict()

Once the multiple response ordered probit has been calculated, the user should employ predict() to translate the cutpoints and coefficients into the proportional distribution of $C$ . The user can specify which of the two types of probabilities to calculate: unconditional probabilities based on the multiple response ordered probit (using the defaults), or conditional probabilities based on the multiple response ordered probit (by specifying the anchors output as C).

## Unconditional prediction, for each observation
predict(out, summary = FALSE)

## Conditional prediction, for each observation
predict(out, C = C, summary = FALSE)

By accepting the default summary = TRUE, all the observations are summarized into one convenient vector. If summary = FALSE, the predicted probablities for each observation are returned in a matrix indexed by (observations $\times$ categories).