Multiple regression in SPSS: Using the Wild Bootstrap to obtain bootstrap confidence intervals

This video demonstrates how to generate bootstrap confidence intervals for regression coefficients in those cases where you have evidence of residual heteroskedasticity. Specifically, I demonstrate how to use the Wild bootstrap using SPSS version 27.
The data for this demonstration can be downloaded here: drive.google.c...
The Powerpoint referenced in the video can be downloaded here: drive.google.c...
Link to manual referenced in video: www.sussex.ac.u...

[@JosephineLindhout]

Wild bootstraps starts at 11:20. Thanks!

[@anacrongr]

Best wild bootstrap regression video so far, thank you Dr

[@mikecrowson2462]

You're very welcome!

[@arthurarantesdc]

Great video! Tks!

[@blackbunny109]

Thank you for this great demonstration. Do you by any chance know how SPSS does its wild bootstrap? Meaning what transformation SPSS uses for the residuals (i.e. dividing by (1 - h.t)) etc.?

[@jdh2]

Awesome!

[@talzabidi1569]

Hi Dr , thanks so much for your informative video
I have question , if my population is small so I need to take all people , in this case how can I run inferential statistic if I don't have sample?

[@mikecrowson2462]

Hi Talal, technically speaking if you have every observation in your population then you have population data. If you compute a mean, standard deviation (using the population formula), regression coefficients, etc. you have the population parameters at your disposal. So the need for inference becomes a mute point. That said, sometimes you will see folks perform significance tests using their population data 'as if' it is coming from a sample. I'm not really sure why because inferential statistics are about inferences concerning population parameters based on sample data. I guess they do it out of habit and because it 'feels weird' to not run a significance test. But again, technically speaking there is no need for a significance test if you have all observations from a population. If your population is small but you only have a subset of all possible observations, then the data will represent a sample from the population. In that case, you would need to use inferential statistics. I hope this clarifies things for you.

[@talzabidi1569]

Thanks so much Dr for your clarification.
I have heard that, we can do (N-1) by using this method will be able to make inferential statistics . Is it correct?