Dear all,
I made a simple linear regression model with the same scale (i.e. X variables, and Y variable). I need to run this for each year. How to determine incremental significance of added IV's across determined number of years?
One suggested using, e.g., Fama-MacBeth t-stats. But I cannot find useful information for this subject. Have you ever encountered with this method?
Thanks a lot!
Not sure if I understand. Are the units being analysed the same each year (so either John gets measured in 1973 and 1974, or a higher order unit, like John's Hospital data are used each year)? Also, for what purpose are you trying to get a significance value. Is it one, so whether beta_x1 is "sig" pooling the data, or do you want a multiple comparison procedure so that you can say beta_x1 in 1973 was sig. Is there are reason you are not including year in your model?
this is a typical assigment with old style stepwise regression variable selection. I would plot change in adjusted r^2 vs your variable and again against year. as Daniel Wright indicated you were not entirely clear on how you were adding things. The instructor may be trying to teach you hierachical regression where variables are entered in blocks(blockwise) rather than 1 at a time. These are largely worthless techniques. Surprise him with some info on the modern approach to do this adaptive lasso which gives an optimal solution to the prediction problem. If you don't recognize some terms just Google them. I'll add some links and attachments that you may find useful. https://www.researchgate.net/post/Last_predictors_in_my_hierarchical_regression_resulted_in_an_insignificant_R_square_change_which_model_coefficients_do_I_report
https://www.researchgate.net/post/Multiple_regression_in_a_blockwise_manner_vs_simple_multiple_regression2
https://www4.stat.ncsu.edu/~boos/var.select/lasso.adaptive.html
Best Wishes, D. Booth
A stepwise time series regression will do since you are adding IV's across number of years (either backward or forward).
Daniel Wright David Eugene Booth Daglous Gesora
Thank you all for your answers / clarifying questions.
I decided to go with the Vuong test, which allows you to compare the adjusted R square from two linear models.
If you are interested in the case that could include Fama-MacBeth statistics, I will share the details privately.
Thanks again for your attention!
Daria
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics