返回Google圖書搜尋

A Review of Instrumental Variables Estimation in the Applied Health Sciences

出版	SEDAP Research Program, McMaster University, 2007
URL	http://books.google.com.hk/books?id=wiOK0AEACAAJ&hl=&source=gbs_api

註釋In this paper, I review examples of the application of IV in the health and social sciences, I show how the IV estimator works, I discuss the factors that affect its performance, I review how the interpretation of the IV estimator changes when treatment effects vary by individual, and consider the application of IV to nonlinear models. [...] The estimate of ? is of primary interest - it reflects the impact of a one unit change in ? on the mean of ? , holding constant the influence of ? . Matching compares the values of ? among subjects with different levels of ? but who share common values of all of the variables in ? . A defect of both these techniques is that the analyst might fail to adjust for pertinent confounding variables, [...] It is clear that the treatment effect parameter ? 1 is in fact a ratio of two unknown quantities: the numerator is the correlation between the instrument and ? , the denominator is the correlation between the instrument and ? . The weak instrument problem adversely affects this denominator. [...] I also reproduced the results for the generalized IV estimator that controls for ? ? ? , and uses ? ? as an instrument, shown in Figure 1. The kernel-smoothed histograms of the ? treatment effect estimates are displayed in Figure 2. As expected, the sampling distribution of the DIM estimator, ? ? ? 1 , estimated using the observational data (labeled as DIM OBS in the figure) is well to the left [...] Large values of ? × ? 2 from this regression indicates that the instruments in ? explain some of the variation in ? ? ? , which is a violation of Assumption 2. If the assumption is satisfied, this test statistic is distributed ? 2 with the number of degrees of freedom equal to the number of instruments minus one.