Under-dispersed Health Insurance Count Data

Published: 17-07-2019| Version 9 | DOI: 10.17632/z7wznk53cf.9
Olumide Adesina


The data set was obtained from Ace-Medicare Clinics, Ota, Ogun State, Nigeria, comprising of National Health Insurance Scheme (NHIS) data with no zero count. A sample of 1647 patients under National Health Insurance Scheme was obtained from July 2016 to July 2017. The data set was used in the article titled "Bayesian Models for Zero Truncated Count Data" with link http://www.journalajpas.com/index.php/AJPAS/article/view/30105 Response variable was (Nencounter) that is, number of encounter (visit to the doctor). The class (Eclass) indicated whether a patient was ever on admission for the period or not, that is, (in-patient=1, out-patient= 0). Another predictor is (follow-up), indicating whether a patient was on regular check-up or not, (follow-up=1, no follow-up=0). Gender (sex) of patients; (male=1, female=0). Another predictor was Ndiagnosis, which represented the number of diagnosis a patient had for the period of encounter. The last predictor is biological age of patient. Following the dispersion test, the data is under-dispersed with dispersion parameter =0.7806.