I am looking after many research fellows.
I am interested in best books or papers on data and econometric methods and techniques.
Can you recommend?
Dear Shaozhong Shi -
I worked primarily in finite population prediction (not forecasting) for survey statistics (Official Statistics), but here are econometrics books I used, though more recent editions or other textbooks may be better:
Granger, C.W.J., and P. Newbold (1977), Forecasting Economic
Time Series, Academic Press.
(Apparently there is a more recent version by Granger.)
Griffiths, W.E., Hill, R.C., Judge, G.G. (1993), Learning and Practicing Econometrics,
Wiley.
Kutner, M., Nachtsheim, C., and Neter, J.(2004), Applied Linear Regression Models, 4th
ed., McGraw-Hill/Irwin.
Maddala, G.S. (1977), Econometrics, McGraw-Hill.
Maddala, G.S. (2001), Introduction to Econometrics, 3rd ed., Wiley. (A 4th ed was made with Lahiri - but I'm not as familiar with it.)
In econometrics texts, one usually sees heteroscedasticity as a "problem" to be solved. My approach is to treat it as the natural phenomenon that it is, occurring in the model error structure. For some notes on this, see the following:
https://www.researchgate.net/publication/263036348_Properties_of_Weighted_Least_Squares_Regression_for_Cutoff_Sampling_in_Establishment_Surveys
That generally deals with one regressor, where the size measure in the regression weight is x, but for multiple regression, the size measure could be a preliminary prediction-of-y.
Many want to "eliminate" or "remove" heteroscedasticity to be able to use hypothesis tests, but confidence intervals based on other distributions (or perhaps the Chebyshev inequality) may be far more meaningfully interpretable. Further, transformations themselves may confuse interpretation of results. If you must transform, here is a good book:
Carroll and Ruppert(1988), Transformation and Weighting in Regression, Chapman & Hall, Ltd. London, UK.
Hypothesis tests though, are often misleading. A common bad practice is to set thresholds at 0.05 everywhere, when sample size and effect size should have been considered. I noted this problem of interpretation long ago:
https://www.researchgate.net/publication/262971440_Practical_Interpretation_of_Hypothesis_Tests_-_letter_to_the_editor_-_TAS
Since then, problems with misuse of hypothesis tests caused at least one journal to recently ban them, and then the American Statistical Association finally issued a press release, just months ago, noting the problem.
You may find better texts than those above, but the ones given here may be useful for you. At any rate, they were useful to me, perhaps Maddala(2001) in particular. (More recent textbooks should be available as well.)
Best wishes - Jim
Article Practical Interpretation of Hypothesis Tests - letter to the...
Article Properties of Weighted Least Squares Regression for Cutoff S...
- Badges
- Science topic
- Similar topics
- Books