opsr_simulate function

Simulate Data from an OPSR Process

Simulates data from an ordinal probit process and separate (for each regime) OLS process where the errors follow a multivariate normal distribution.

opsr_simulate(nobs = 1000, sigma = NULL)

Arguments

• nobs: number of observations to simulate.
• sigma: the covariance matrix of the multivariate normal.

Returns

Named list: - params: ground truth parameters.

• data: simulated data (as observed by the researcher). See also 'Details' section.

• errors: error draws from the multivariate normal (as used in the latent process).

• sigma: assumed covariance matrix (to generate errors).

Details

Three ordinal outcomes are simulated and the distinct design matrices (W and X) are used (if W == X the model is poorely identified). Variables ys and xs in data correspond to the selection process and yo, xo to the outcome process.