 |
The 130th Annual Meeting of APHA |
|
The 130th Annual Meeting of APHA |
Pei-Tseng Kung, ScD, Graduate Institute of Health Administration, Taichung Healthcare and Management University, No. 11 Ln. 16 Sec. 3 Chungching Rd., Taya, Taichung, 428, Taiwan, 886-4-2560-3149, ptkung@seed.net.tw and Wen-Chen Tsai, Dr PH, Department of Health Services Management, China Medical College, 13 Ln.16 Sec.3 Chungching Rd, Taya, Taichung, 428, Taiwan.
Prediction is very popular and has been widely applied in many different scientific fields. Prediction becomes difficult when there is the high variability in the distribution of the observed values. This study used individual health care expenditures as the observed outcome to examine and compare the predictive accuracy among the linear regression model, the two-part model, and the sample selection model with various proportions of population with zero health care expenditures in a given year. The focus of this research is to determine the impact of the sample size and the proportion of medical care nonspenders on the predictive power of the model.
This study conducted the Monte Carlo simulations varying the proportions of nonexpenders in the sample for each iteration. The split-sample analysis was applied for the comparison of the predictive accuracy of models. The conclusions of this study are that either the two-part model or the sample selection model performs better than the linear regression model and that, for this data set, the two-part model and the sample selection model performed equally well. Given the relative complexity of the sample selection model, these results suggest use of the simpler and equally effective two-part model in predicting health care costs. Among the three performance criteria used for model comparison, the mean squared forecast error and the means of forecast bias ratio were better criteria than the mean forecast bias.
Learning Objectives: The participant in this session will be able to
Presenting author's disclosure statement:
I do not have any significant financial interest/arrangement or affiliation with any organization/institution whose products or services are being discussed in this session.