Multiple regression with Robust Standard Errors in SPSS (February 2021)

This video provides a general overview on how to obtain robust standard errors (and test statistics) in SPSS when performing multiple regression analysis.
A copy of the data can be downloaded here: drive.google.com/file/d/1-o78...
Link to article referenced in video: citeseerx.ist.psu.edu/viewdoc/...

[@Deborah28277]

You are the best! Thank you taking the time to include a little of the background as a reminder.

[@mikecrowson2462]

You are so welcome!

[@sofyanyamin7316]

thanks you so much, i get much information about statistics from your channel.

[@couragee1]

thank you!

[@geofaultlline]

Thanks so much for your videos, they are really helping me figure out my model! Question though- when reporting the results from the robust standard errors, and I essentially using the same format I would have used with a typical multiple linear regression? And include the comparison of regular vs adjusted?

[@atsushikanayama8977]

Thank you so much for the excellent insights! May I ask you how we can produce F statistic and the p value estimated using robust standard errors for R squared (and R squared change) if possible?

[@OmerFKocak]

Thank you so much!

[@mikecrowson2462]

You're welcome!

[@luisarango6190]

Hi Mike. First of all, thank you very much for your content; it's been extremely useful to me. I have a question and been searching for an answer in several places, like Researchgate , or even in books about multiple regression (e.g. the PLS SEM book by Hair et al), but with no success. It is about multiple moderation. I have an IV and a DV and want to test the effect of two discrete moderators, individually and jointly, on this relationship. My idea is that each moderator individually moderates the IV-DV relationship, but when combined the moderation effect is even bigger. I don't know what kind of analysis should I perform to test this last part (the combined effect of the moderators); I am not even sure if this makes complete sense. If you could help with a resource or a non-technical answer on how to approach this, I would really appreciate it. :)

[@kritijain8397]

Thank you for the video sir. Could you please explain how the coefficients will be calculated after getting robust standard errors? Thank You

[@helenakothner8758]

Dear Mike! Thank you for this Video! it's very helpful. I still don't understand how I can calculate my multiple regression with the robust standard errors after calculating them. I heard SPSS cannot do this and that I'll need Macro Process. If there is any way to do it without macro process, it would be very helpful!

[@pietislekker]

Is it possible to use pairwise deletion within General Linear Model Univariate?

[@anianimations]

Thanks for the video. But I can’t find same panel under options. I am using spss 21 and in my options available, only parameter estimates, not the robust and hetero test options. Please explain which spss version you are using.

[@mikecrowson2462]

Hi PakChina, I'm using version 26 in this video. I think this option started being included in later versions of SPSS than what you are using.

[@louis1505]

Hi I'm runnign a hierarchical regression and want to use robust std errors, do I enter all my predicotrs, regardless of level into the covarites box at the same time?

[@myatsu8227]

I'm also facing the same problem. Have you figured it out? If yes, could you plz share? Thanks.

[@mataempat290]

How can the Sig. With robust is different beetween the first nd second run? (Before checklist on whitetest and after checklist)

[@chrisoneill6553]

Hello! Thanks for the video, it's perfect. My question is how would you find the standardized coefficients using this robust method? Would you take them from the standard linear regression analysis you did first, or would that be inappropriate as the robust standard errors have changed? Basically I'd like to know how to find the effect sizes on this robust version. Thanks!

[@mikecrowson2462]

Hi Chris, that is correct. You can just re-run your analysis through the standard regression option and use the standardized coefficients. Cheers!

[@chrisoneill6553]

@@mikecrowson2462 Excellent. Thank you! Is there a way to do blocks with this method? Or do you have to do it manually by repeating the test and adding in IVs?

[@mikecrowson2462]

@@chrisoneill6553 Unfortunately, there's no way to enter variables in blocks using options available through this menu. You would have to add them in separate blocks.
My understanding is that the heteroskedasticity consistent errors applies to the regression coefficients and not the F-test proper. So I would imagine you could use blocks through the regression menu if you are wanting the change in R-square, and then report on the tests of the regression coefficients in your models using the robust standard errors after specifying each model under GLM. (Or you could report on both the unadjusted and adjusted test results, the latter accounting for heteroskedasticity). Cheers!

[@chrisoneill6553]

@@mikecrowson2462 Wonderful, I really appreciate you taking the time to help. Thank you. Keep up the good work!

[@manzoorwanigeo2335]

I am not seeing anything after selecting paramete estimates

[@mahiderevlogs9329]

It is only work on the last version of spss software

[@mikecrowson2462]

Hello. I just ran the data from this video using SPSS version 28 and it worked fine.

[@wewinwiracorneliswahid6602]

What estimator is used in this? or whos theory even?

[@mikecrowson2462]

The hc3 estimator for the standard error was used in the video. Read more about it here (there is a link to the article on this site): link.springer.com/article/10.3758/BF03192961