Please see the attachment.
1. It is hard to evaluate these three rounding rule in the practical study. This simulation study can investigate bias because the true theta is known. in the practical data, we only can observe the intra-rater and inter-rater inconsistency but it dosen't associate with rounding rules.
2. this study doesn't need to do more number of replication, because there is no error term and random effect in it. it just calculate the p value for each item based on intended cut score, then round all the p value, and then obtain the estimate cut score based on those rounding values. You cant find any random number sampling in this simulation.