A simulated data set where three different raters (rater1, rater2 and rater3)
assign ordinal ratings on different firms. rater3 uses a different rating scale
compared to rater1 and rater2, i.e., the number of threshold categories is different.
For each firm we simulate five different covariates X1, ..., X5 from a standard
normal distribution. Additionally, each firm is randomly assigned to a business sector (sector X, Y or Z), captured by the covariate X6. Furthermore, we simulate
multivariate normally distributed errors. For a given set of parameters we obtain the three rating variables for
each firm by slotting the latent scores according to the corresponding threshold parameters.
The IDs for each subject \(i\) of the \(n = 1000\) firms are stored in the column firm_id. The IDs of the raters are stored
in the column rater_id. The ordinal ratings are provided in the column rating and all the covariates in the remaining columns.
Overall, the data set has 3000 rows, for each of the \(n = 1000\) firms it has three rating observations.
data("data_mvord", package = "mvord")A data frame with 3000 rows and 9 variables
firm_id firm index
rater_id rater index
rating ordinal credit ratings
X1 covariate X1
X2 covariate X2
X3 covariate X3
X4 covariate X4
X5 covariate X5
X6 covariate X6 (factor)