Is it possible to generate a new variable "phase" that is based on the time the participants were questioned? I generated this new variable by dividing the time into three phases (beginning, middle, end) and would like to use it as an independent variable in a panel regression.
If I would use the time itself as an independent variable, this would logically result in an error, but I would like to see if the time has an effect on the dependent variable. How can I do this in a panel regression?
Thanks in advance!
You can use logistic regression and give a code for the time like 1 for the beginning and 2 for the middle and 3 for the end.
Yes you can. But doing this often causes loss of information. AS Abdulmuhsin S. Shihab suggests if this is a DV doing so can change your type of regression. from what you suggest, it sounds like ordinal logistic regression., again if this new variable is your DV. Best, David Booth.
It is better to plot a graph of time versus the dependent variable, then suggest the weather regression model is better or the model will be better.
But of course, you may obtain a regression model. But the question of whether it will give you better predictions.
Dr. Kamran
I mean whether the regression model is better or any other model will be better for your data. It may be suggested after plotting and after considering the various factors of model building. The detail is given in the following reference
Elements of Forecasting
by Francis X. Diebold.
Regards.
Dr. Kamran
Here a paper using logistic regression
Using lowess to remove systematic trends over time in predictor variables prior to logistic regression with quantile categories
Craig B. Borkowf, Paul S. Albert, Christian C. AbnetJournal:Statistics in MedicineYear:2003
What is the type of error? How many cases you have for each level.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression