What This Book Actually Is
Introduction to Econometrics Brief Edition is the condensed version of the fuller James Stock and David Andrews textbook that most undergrads get assigned in introductory econometrics courses. It covers the same core material OLS regression, hypothesis testing, multiple regression, dummy variables, instrumental variables, time series basics but in a tighter package. The brief edition cuts some of the more advanced chapters and condenses the applied examples. If your syllabus says you need it, you need it. Period. I taught an intro econometrics section using this edition for about four years. The difference between the full version and the brief is mostly in the later chapters. You lose the deeper treatments of panel data methods and some of the more quantitative micro applications. For a first course, that cut is usually fine. Most professors don't get there anyway.
Getting Introduction To Econometrics Brief Edition
The publisher is Pearson. You can find it on Amazon, the Pearson website, Book Depository, or any campus bookstore that hasn't completely sold out yet. The ISBN for the latest printing is 978-0138096749. You can also grab a digital copy through MyLab Economics if your professor requires access codes, though that stuff is annoying and costs more than the book itself. I always tell students to buy the used physical copy when they can. The formulas are the same either way and you won't need to highlight like three quarters of every page if you're reading ahead. It starts with simple regression and builds from there. Chapter 1 through 3 cover the probability and statistics refreshers you'll need, then it moves into simple linear regression, multiple regression, and hypothesis testing. The statistical inference chapters are where most students hit their first wall. The book does a decent job explaining confidence intervals and p-values without drowning you in measure theory, which is more than I can say about some of the competing texts. The real problem comes in Chapter 6 and beyond when OLS assumptions get tested. Endogeneity, omitted variable bias, simultaneity these are not intuitive concepts and the book presents them in a fairly formal way. I had students struggle for weeks with the intuition behind why OLS coefficients become inconsistent when you have correlation between the regressor and the error term. The algebra in the book is straightforward enough, but the conceptual leap is harder than the math suggests.
One thing the book handles well is the applied side. Each chapter has datasets and empirical exercises. The data typically comes from real economic sources wage data, GDP growth, housing prices. Running the regressions in the textbook exercises with whatever software your class uses (Stata, R, Python) is where it actually clicks for most people. Reading the theory without doing the problems is basically useless.
Get the Full Details
What It Does Not Cover Well
The brief edition drops several topics that come up in actual research. There is minimal coverage of difference-in-differences, regression discontinuity designs, or synthetic controls. If you are taking this class to prep for a research assistant position or a grad program, you will need supplementary material. The full Stock and Watson version covers some of this, but even that does not go deep enough for modern applied micro work. Another gap is that the treatment of time series is quite basic. Auto-regressive models, unit roots, cointegration get a chapter each but they are surface level. I had a student who needed to model quarterly inflation data for a thesis and kept running into spurious regression problems. The book mentions the issue but does not give you the tools to actually fix it. You would need to go to something like Enders or Hamilton for that.
Practical Issues I Ran Into
Here is a specific problem that came up repeatedly in my sections. Students would run a regression, get a weird coefficient, and immediately assume the model was wrong. The issue was almost always specification, not computation. For example, I had someone regress wages on education and experience and get a negative coefficient on experience that made no intuitive sense. The fix was not a different estimator. It was that the dataset had coded experience as years remaining until retirement rather than years worked. The variable was mislabeled in the source data. This happened in maybe three or four sections per year. Another edge case is multicollinearity. The book explains the formula for the variance inflation factor and shows you the math. But when students actually encounter it in the empirical exercises, they often miss it because the coefficients still look reasonable. I developed a quick check where I have students run each regressor against all the others individually before running the full model. If any pairwise correlation is above 0.8, they flag it. This catches the problem before the regression output becomes misleading. It takes about two minutes and saves a lot of confusion later.
How to Use This Book Effectively
Do not read it cover to cover. Read the chapter before the lecture, skim the derivations, and focus on the main results and the intuition paragraphs. Then do the empirical exercises. That is the sequence that works. The derivations are useful for understanding why things work but you do not need to reproduce every proof to pass the exam or use the material. The problem sets at the end of each chapter are where the real learning happens. Some are computational, some are theoretical. The computational ones are more valuable for building intuition. Run the same regression with different specifications. Drop variables. Add them back. See how the coefficients move. That exercise teaches you more than any number of re-reads of the theory sections. Pair the book with free online resources if you are struggling with a concept. The MIT OpenCourseWare notes on econometrics supplement this text well. The Khan Academy videos on regression are fine for the early chapters. There is no point in paying for extra tutoring when the same material is available for free, and the Stock and Watson problems are already challenging enough on their own.
Who Should Skip It
If you are already comfortable with regression analysis from another discipline or you have taken a quant methods course in a different field, you might not need this book. The mathematical level is accessible but not trivial. You need some calculus and basic linear algebra. If you have neither, spend a week on the prerequisites before touching this text. The probability review chapters are not remedial. They assume you have seen this before. If your program is heavily quantitative, consider the full edition instead. The brief version saves you about a hundred pages but those hundred pages contain material that matters if you plan to do empirical work beyond an introductory course. The panel data chapters in the full version alone are worth the extra cost if that is where your interests lie.
