Limitation of This Research

While the application of constrained optimization can eliminate boundary violations, it costs more time and computing capacity to execute the revised methods, especially with the maximum likelihood assumption. Given the scope of this article, a systematic assessment is yet to be done with regard to different methods's performance under various conditions. We expect more future works in simulations, as well as empirical studies to illuminate to what extent the current panel regression suffers from boundary violations, and to what extent the revised models can successfully eliminate these violations. By possessing this knowledge, scholars in this field can provide more definite criteria to prevent political science research from reporting an illogical out-of-bound finding.

Download [full paper] [supplementary material A] [supplementary material B] [all replication files]