All the assumptions on the error terms have been stated as conditional on given X. Proofs of unbiasedness and the Gauss-Markov Theorem are initially done with X given and then extended to the case where it is a random variable.
Example 4.2 is new. Rest are moved up by one. Added DATA4-16 and DATA4-17.
Added a section on generating the forecast and confidence interval for the dependent variable in a multiple regression model.
Added a section on important comments about interpreting regression coefficients, especially those for variables measured as percentages or proportions.
Added a discussion and a table summarizing the interpretation of regression coefficients in models involving logarithms of variables.
Added DATA7-24, DATA7-25, and DATA7-26.
Added a section on the modified interpretation of dummy coefficients in the case of log-linear models.
Replaced Example 10.5 by a new one and added Example 10.6. Remaining examples have been renumbered. Added a section on VAR models and another on panel data. Example 10.9 is new.
Extends the forecast combination section to include serial correlation, ARCH effects, and time-varying weights.
Added a section on index numbers. Added section on Econ Lit software.
Data sets are now available in a variety of formats; ASCII (that is, text), B34S, EVIEWS, EXCEL, GRETL, PcGive, and SHAZAM. The appendix also describes where these data sets are available and provides a brief introduction to the econometrics software packages B34S, EVIEWS, GRETL, PcGive, and SHAZAM, with a link to other software packages.
Added descriptions of new data sets.