Dear Research Gate Members,
I have a query regarding comparing the two groups on the basis of the regression models
For Instance:
I have two sector companies i.e. Public Sector and Private Sector
I have tested whether their CSR Score in Annual Reports affected by various factors or not like the size, profitability etc. For that, I have run two different multiple regression models for both sectors individually.
Thus, can anybody suggest me any statistical tool or technique available to compare those two regression models?
If you want to compare the goodness-of-fit you can use R^2, but it's not clear as to what you would like to compare regarding the two models you described. You can also compare coefficients and corresponding CIs.
Yes, the method is called "general linear models". You can use categorical and continuous predictors in the same model, which may include coefficient for the difference in slopes between groups.
If you want to compare the goodness-of-fit you can use R^2, but it's not clear as to what you would like to compare regarding the two models you described. You can also compare coefficients and corresponding CIs.
Thank You Jochen Wilhelm for your response and @Andrey Davydenko I want to test the difference between two groups on the basis of their independent variable predictions because my dependent and independent variables are alike for both groups
I see. For your task please let me recommend the diagram we presented in this presentation, slide 16: Conference Paper Data Formats and Visual Tools for Forecast Evaluation in Cyb...
Dear Harmandeep,
One suggestion -: If the dependent variables and independent variables are continuous, and observations for each variable in one group are independent from observations of variables in another group (private vs public), then you can run two seperate regression models (one for each group) and assess the estimated coefficients by comparing the nature (positive or negative) and magnitude (statistically significant or insignificant) of the effect of each regressor in one group to the analogous effect it has in another group. In concurrence with Andrey Davydenko, you can further compare the adjusted R^2 values, Means of Dependent Variables, SE of regression, etc.
Just wanted to add to what Elvis Munyaradzi Ganyaupfu suggested: to compare coefficients you can see if CIs for coefficients overlap or you can use some formal tests like those described here: https://statisticsbyjim.com/regression/comparing-regression-lines/
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression