Hello Mike. I have been following you video tutorials for quite some time now. This particular one makes me feel statistically moved... and it is significant if you ask. Thanks for sharing. I have learnt a lot from you.
Thank you for this very helpful video! I have a question -- I ran a forward stepwise binary logistic regression with 5 possible variables. I ended up with 3 out of the 5 variables included in the model. The model was significant compared to step 2 and compared to the null model; however, in the "Variables in the Equation" table, step 3 shows one of the variables in the model to be insignificant, while the other two are. How do I interpret this?
Good job and well done; I'm sure you're aware and probably sick of hearing it by now but one of those pen pads would be a big improvement for adding the handwritten text - one that can keep both the pen pad and the mouse going at the same time.
Hi Wendy, thank you very much. By the way, I have a video from last February that I put together or hierarchical multiple regression at ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-RyDteu6E7HY.html You can also download a Powerpoint and the data to play with by following the link under the video description, which you might find helpful. Cheers.
Dear Mike, thank you for the informative video. Can I ask what are the steps if I want to perform forward, backward or stepwise multinomial logistic regression since my dependent categorical variable has more than 2 groups?
Hi, i am testing 11 independent variables for binary regression analysis, following your explained method of variable selection when i am applying Forward LR method only Block 0 appears ? could you please explain why? (note the backward option is working as u explained) thanks
Thanks Mike for the good explanation, I got a question: what if the homsmer and lemeshow test is significant for the first model?? I got 2 models in total
Hi Desiree. When you say "significant for the first model", do you mean the first of two hierarchically-organized models (where the first model contains a subset of predictors relative to a second model)? If that's the case (and the second HL test is not significant for the second model), then I assume that the addition of predictors in the second model yielded an improvement in fit. By the way, in case you might be interested, I have a 2019 video on logistic regression as well (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-cpWSSJHuT2s.html). Cheers!
@@mikecrowson2462 Yes Mike. I got a 3 step hierarchical model and the first model with only non-modifiable factors is not calibrated (HL test significant).. while with the modifiable factors added in step 2, HL test becomes not significant..can I say that only non-modifiable factors are not adequate in predicting my outcome variable? Thanks so much for your response!
Hey Mike. If we have nominal and scale predictor variables, can we add all those simulataneously as you have shown or is there any other method available for that?
Hi Neha, thanks for your email. If you have nominal or ordinal variables as two levels (i.e., they are binary), then you can add them in this way. If you have categorical variables with more than two levels, you'll need to recode them into a set of dummy or effect coded variables. I talk about dummy coding here (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-XGlbGaOsV9U.html) and effect coding here (ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-Kse6t-a9s2U.html). Also, just an FYI, you might check out two new videos I put together on the topic from this video: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-RyDteu6E7HY.html and ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-lLsIanDaKRw.html . Cheers!
@@mikecrowson2462 those are linear regressions that I understand. But if we want to do backward stepwise logistic regression for suppose 3 categorical and 5 scale predictor variables with categorical (having 2 values) dependent variable, then how to add those all predictors in one model? Or we have to put them separately in block 1 and block 2 respectively.
