SPSS tutorials for beginners PART 6 - Logistic regression models

[@Nzayaba]

Great explanation. Thanks for video

[@henrykitunzi4130]

Thanks Nienke, this was helpful especially the categorical variable defining procedure and confounder adjustment. Could you please have another tutorial that covers the selection variable/ subgroups. Thank you in advance.

[@nienkedeglas_mdphd]

Thank you for the suggestion, that is a very good idea. I will do so!

[@humbertorodriguez3336]

Hi Nienke ! Thanks so much for your video! I am trying to understand which method should I choose for stepwise LR in SPSS. It offers Wald, LR and Conditional . Could you explain this please?

[@rawiahnaoum4382]

This is very helpful. I have a question: I want to make two logistics model one for female and one for male, but the only variable that I have is Gender ? How can I do that ?

[@iphonehelpcenter1333]

Thank Dr for doing this for free, this was really helpful. By the way if you can tell me how can I find/access the data you just used in this video or share them in your description box that would be very much helpful so that we follow along. Thanks again.

[@nienkedeglas_mdphd]

Thanks for the compliment! Unfortunately I cannot share my data as they are real data from patients and therefore bound to restrictions / regulations. I do think there are example datasets available if you search on google. Good luck!

[@iphonehelpcenter1333]

@@nienkedeglas_mdphd Understood and i honestly appreciate you explaining to me why you couldn’t. Waiting more of SPSS videos from this channel and other packages if possible, keep it up Dr and I promise i’ll share this channel with my fellow epidemiologists and statisticians as well.

[@nienkedeglas_mdphd]

@@iphonehelpcenter1333 Thank you, that would be great!

[@moraguardamagna1338]

Thank u so much!!! It is so clear :) Amazing

[@zztzztzzttzztzzt]

Thank you for the video; at 1:31 you describe the variable "Age" with 5 or so different categories within; how did you make multiple categories within the same variable?

[@nienkedeglas_mdphd]

please take a look at my video on how to make variables (also part of this series), I explain it there :)!

[@fabdoc]

Thank you for great explanation. I have a question regarding interpretation of CI. What if the CI span 1 but the p value is

[@nienkedeglas_mdphd]

that is not possible in logistic regression, the CI's and the p-values are directly linked with each other.

[@Manchahegeet]

Brilliant. Thanks

[@leticiaalvarado9401]

Hi Nienke, amazing videos!! Would you be so kind to explain if age can be used as a continuos variable instead of categorical? Also can ordinal independent variables be used as continuous? like for example ECOG (1,2,3,4). How is the output different if instead of using enter we use forward or backward? What do you recommend. Thanks for these very instructive videos

[@nienkedeglas_mdphd]

thanks for your feedback! It is a choice on what you want to see. If you choose to use a continuous variable, you will get an odds ratio for every increase in age (so in this case every year). This means that the odds ratio can be for example 1.01. If you want to show in larger steps what happens in different (larger) categories, it is sometimes more informative to use categorical variables like I did in this example. Personally I am more a fan of this but both ways are fine!

[@aracelimateosdiaz6775]

Thank you for your videos. Very clear explanation and very helpul to me. I have two questions: a) If you have one logistic regresion model without interactions and another with interactions, do you have to calculate ROC curve for both or just for the model without interactions? Many thaks

[@nienkedeglas_mdphd]

this is a choice to make depending on the importance of the interactions. If the effect of interactions is very large, I would go for that model, and if it is not, I would leave it out completely. There is no right or wrong here, but I think I would choose just one model to present in your paper.

[@aracelimateosdiaz6775]

@@nienkedeglas_mdphd Thank you very much for the answer