The nested logit model has two different specifications, UMNL and NNNL, which affect estimation results differently. The UMNL form is preferred as it aligns with random utility theory. Simulation studies using NNNL software like PROC MDC in SAS(c) require assumptions of RUT-conformity, which can impact the accuracy of input parameters. Using different software packages can also influence estimation results in a nested logit model.