Teaching and hospital production: The use of regression estimates
研究了美国退伍军人事务部数据中测量误差和多重共线性对医院成本函数回归估计的影响，特别是用于教学医院间接医疗教育成本支付的调整因子。
Medicare's Prospective Payment System pays U.S. teaching hospitals for the indirect costs of medical education based on a regression coefficient in a cost function. In regression studies using health care data, it is common for explanatory variables to be measured imperfectly, yet the potential for measurement error is often ignored. In this paper, U.S. Department of Veterans Affairs data is used to examine issues of health care production estimation and the use of regression estimates like the teaching adjustment factor. The findings show that measurement error and persistent multicollinearity confound attempts to have a large degree of confidence in the precise magnitude of parameter estimates.