That is very useful, thank you. It seems it doesn't work with multiply imputed datasets though. I have some heteroskedasticity issues with my regression and I would need heteroskedasticity robust SE. However, the default is listwise.
Thank you very much for the video :) Do you know wether there is a macro for the the Panel Corrected Standard Errors according to Beck and Katz (1995), accounting for an AR(1) process. Surprisingly, as far as I know, only stata is offering this...
Thank you for your clear explanation. I am not familiar with macros and I would like to know if there is a way to verify the source. For example, for Dr. Daryanto's macro, it is from google, how can we be sure that it is the right one?
This site is the only one for his macro: sites.google.com/site/ahmaddaryanto/scripts/Heterogeneity-test. However, I have checked out results using this macro against those in other programs and they are the same. However, if you have SPSS version 25, then there is a built in route that will allow you to test for heteroskedasticity. Go to this video for demonstration: ruclips.net/video/TOY61BBbkVI/видео.html
Hi Pieter, the control variables are basically added into you model as additional predictors. Although you may conceptually think of your focal independent variables as different from your control variables, they are handled in the same way in your analysis. They are all treated as independent variables. Best wishes.
Hello Dr., thanks for the video it was very helpfull. However, I tried to execute the Haye's robust regression (new model) after spliting the file and it could not operate . How do we execute this operation with data grouped by a certain variable?
FYI - for the Hayes macro download and install worked great, but the Daryanto wouldn't install. I tried the version directly from his site and the version from the journal but neither would install.
Hi there. Thanks for letting me know. I wanted to mention too that I have found it possible to run a robust regression in SPSS without relying on the macros. You can do this using the following steps: (a) click on general linear model-->univariate, (b) specify the predictors in your model using the covariates box, (c) Under options, if you click Parameter estimates you get the model expressed in terms of regression parameters (the b's, the standard errors, t-tests), (d) under the options box click on Parameter estimates with robust standard errors (HC-3 is what I select). You will then get output containing all the information for your regression. (e) use the F-test and eta-squared from the corrected Model (this is the first line in the output) in the ANOVA summary table (the partial eta-squared will = R-square); (f) the Parameter estimates table will contain the regression slopes, standard errors, and tests. (g) A new box will show up in the output , Parameter estimates with robust standard errors. I know this is a lot more work than just using the macro. Just thought I'd offer it as an FYI. Cheers!
thank you
That is very useful, thank you. It seems it doesn't work with multiply imputed datasets though. I have some heteroskedasticity issues with my regression and I would need heteroskedasticity robust SE. However, the default is listwise.
Thank you very much for the video :) Do you know wether there is a macro for the the Panel Corrected Standard Errors according to Beck and Katz (1995), accounting for an AR(1) process. Surprisingly, as far as I know, only stata is offering this...
Thank you for your clear explanation. I am not familiar with macros and I would like to know if there is a way to verify the source. For example, for Dr. Daryanto's macro, it is from google, how can we be sure that it is the right one?
This site is the only one for his macro: sites.google.com/site/ahmaddaryanto/scripts/Heterogeneity-test. However, I have checked out results using this macro against those in other programs and they are the same. However, if you have SPSS version 25, then there is a built in route that will allow you to test for heteroskedasticity. Go to this video for demonstration: ruclips.net/video/TOY61BBbkVI/видео.html
Dear Mike, Thanks for the explanation. How do you include control variables into this macro?
Hi Pieter, the control variables are basically added into you model as additional predictors. Although you may conceptually think of your focal independent variables as different from your control variables, they are handled in the same way in your analysis. They are all treated as independent variables. Best wishes.
Hello Dr., thanks for the video it was very helpfull. However, I tried to execute the Haye's robust regression (new model) after spliting the file and it could not operate . How do we execute this operation with data grouped by a certain variable?
FYI - for the Hayes macro download and install worked great, but the Daryanto wouldn't install. I tried the version directly from his site and the version from the journal but neither would install.
Hi there. Thanks for letting me know. I wanted to mention too that I have found it possible to run a robust regression in SPSS without relying on the macros. You can do this using the following steps: (a) click on general linear model-->univariate, (b) specify the predictors in your model using the covariates box, (c) Under options, if you click Parameter estimates you get the model expressed in terms of regression parameters (the b's, the standard errors, t-tests), (d) under the options box click on Parameter estimates with robust standard errors (HC-3 is what I select). You will then get output containing all the information for your regression. (e) use the F-test and eta-squared from the corrected Model (this is the first line in the output) in the ANOVA summary table (the partial eta-squared will = R-square); (f) the Parameter estimates table will contain the regression slopes, standard errors, and tests. (g) A new box will show up in the output , Parameter estimates with robust standard errors. I know this is a lot more work than just using the macro. Just thought I'd offer it as an FYI. Cheers!
@@mikecrowson2462 Thanks for taking the time to answer, I appreciate that!
@@mikecrowson2462 can you make a video about it ? i quite new with spss, it will help me alot