思索者 发表于 2016-10-2 17:21 
第六版有什么改动吗，楼主能否贴出New to This Edition那一部分看看
I have added new exercises to almost every chapter, including the appendices. Most of the new computer
exercises use new data sets, including a data set on student performance and attending a Catholic
high school and a time series data set on presidential approval ratings and gasoline prices. I have also
added some harder problems that require derivations.
There are several changes to the text worth noting. Chapter 2 contains a more extensive discussion
about the relationship between the simple regression coefficient and the correlation coefficient.
Chapter 3 clarifies issues with comparing R-squareds from models when data are missing
on some variables (thereby reducing sample sizes available for regressions with more explanatory
variables).
Chapter 6 introduces the notion of an average partial effect (APE) for models linear in the parameters
but including nonlinear functions, primarily quadratics and interaction terms. The notion of an
APE, which was implicit in previous editions, has become an important concept in empirical work;
understanding how to compute and interpret APEs in the context of OLS is a valuable skill. For more
advanced classes, the introduction in Chapter 6 eases the way to the discussion of APEs in the nonlinear
models studied in Chapter 17, which also includes an expanded discussion of APEs—including
now showing APEs in tables alongside coefficients in logit, probit, and Tobit applications.
In Chapter 8, I refine some of the discussion involving the issue of heteroskedasticity, including
an expanded discussion of Chow tests and a more precise description of weighted least squares when
the weights must be estimated. Chapter 9, which contains some optional, slightly more advanced
topics, defines terms that appear often in the large literature on missing data. A common practice
in empirical work is to create indicator variables for missing data, and to include them in a multiple
regression analysis. Chapter 9 discusses how this method can be implemented and when it will produce
unbiased and consistent estimators.
The treatment of unobserved effects panel data models in chapter 14 has been expanded to
include more of a discussion of unbalanced panel data sets, including how the fixed effects, random
effects, and correlated random effects approaches still can be applied. Another important addition is a
much more detailed discussion on applying fixed effects and random effects methods to cluster samples.
I also include discussion of some subtle issues that can arise in using clustered standard errors
when the data have been obtained from a random sampling scheme.
Chapter 15 now has a more detailed discussion of the problem of weak instrumental variables so
that students can access the basics without having to track down more advanced sources.
扫码加好友，拉您进群
全部版块
我的主页
