search for




 

Turnover determinants with truncated count data model
Journal of the Korean Data & Information Science Society 2018;29:1595-604
Published online November 30, 2018
© 2018 Korean Data and Information Science Society.

Jangsik Cho1

1Division of Mathematics and Applied Statistics, Kyungsung University
Correspondence to: Professor, Division of Mathematics and Applied Statistics, Kyungsung University, Busan 48434, Korea. E-mail: jscho@ks.ac.kr
This research was supported by Kyungsung University Research Grants in 2017.
Received September 13, 2018; Revised October 29, 2018; Accepted October 30, 2018.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Abstract
In this paper, we analyze the determinants that affect the turnover frequency of college graduates who have experienced turnover. Since the number of turnover which is a dependent variable has only a positive integer value that does not include ‘0’, it has count data truncated at ‘0’. In the case of using the standard Poisson regression model or the negative binomial regression model for the data with truncated count data, the estimated statistic has a problem with the bias and inconsistent estimator. To solve this problem, we analyzed Poisson and negative binomial regression models using truncated count data model. The main results are as follows; First, we note that the truncated negative binomial regression model is most significant. Second, it can be seen that the turnover rate of college graduates is higher than that of vocational colleges. Third, the higher the grade point average and the higher the satisfaction of major and university, the lower the turnover frequency. Finally, as the salary and firm size increased, and the number of regular employees decreased, the number of turnover decreased significantly.
Keywords : Hetero-skedasticity, over-dispersion, truncated count data, truncated negative binomial model, truncated Poisson model, turnover.