Thank you very much for the video. I am doing a multiple logistic regression for my study now and this really helped deal with the confounders. However my SPSS output still states ''Warnings: Due to redundancies, degrees of freedom have been reduced for one or more variables". I read that this is could be due to multicollinearity between the independent variables. all my exposure variables are categorical and some are non binary. I repeated the regression analysis using dummy variables (to check for interactions) and I ran the collinearity diagnostics (using the linear regression though my variables were categorical). The tolerance values and VIf did not show any significant collinearity. But the warning still pops up and spss drops out one of my categories from a certain variable. Wondering if you could help with multicolleniarity in logistic regression or if you have already uploaded a video for the same?
Thx! It's quite helpful. The topic of my dissertation is exposure to pro- and anti-smoking media messages and their association with intention to smoke among adolescents. I planned to use logistic regression. But I am wondering if it makes sense to use hierarchical multiple regression and put pro- and anti-smoking media messages in the second block. Is it suitable to do that?
Thank you for the video. What if the F change value in the Model Summary table is not significant (greater than .05) for model 2 and 3, but the F value in the Anova table is significant? Should one be interpreting models 2,3 and their corresponding beta coefficients?
I saw many HMR in the research papers contain 3 models: model 1- a main IV; model 2-added some more IVs and model 3- their interactions. If my research doesn't want to investigate the main effects of IV added in the model 2, I just wanna know if there is any moderating effect on the relationship in model 1, can I just combine model 2 and 3. i.e.: model 1: a main IV, model 2- added moderators and the interaction between the main IV and moderators Thanks !!!
hi, super useful video, but there is one question I don't understand. In the hierarchical regression, if I have category variables : -.5 = public high school and +.5 = private high school. What could be some of the reasons why I need to code like this?
Hi thank you for this. I really never understood what was meant when my professor told me to "control for the demographic variables" in my model. :) Finally found it!!! A question though, I have up to 32 independent variables ( due to dummy coding of my demographic variables => f.e. almost 14 dummy codes of NGO names ) and I see when doing a multiple hierarchical regression analysis some of the variables get excluded. Given its 3 dummy codes of my demographic variable, is this really a problem? What should I do or can I just interpret the results with no issues. Thanks so much!!!
Hi there. Thanks for the video, it was very helpful. I hate a question regarding repeated measures as a dependent variable. Would this work the same way if my dependent variable has multiple (i.e. 16) time points it has been measured at? I've transformed it into long data, so it just comes up as one variable now. Would it still work that way?
Hi, I'm translating a text (outside my field of specialty!) that mentions "linear hierarchical multiple regressions" and I'm having a hard time understand. Is a hierarchical regression necessarily a multiple linear regression? Or does "hierarchical" refer simply to the order in which the variables are entered in a linear multiple regression? Many thanks to anyone who can help!
No it does not always include multiple predictors. It simply refers to the sequential placement of variables in the prediction model and being able to measure the effect of the variable entered in the first block.
This was exactly what I needed! Thank you!
Are you able to do a results writeup vid as well?
you can add results section how to report this steps according to APA-6 with my best wishes . Also, sample table is a good bet.
So great you are :) Thank you so much for your valuable clip!
amazing! super useful!
Happy to hear that!
Thank you very much for the video. I am doing a multiple logistic regression for my study now and this really helped deal with the confounders. However my SPSS output still states ''Warnings:
Due to redundancies, degrees of freedom have been reduced for one or more variables". I read that this is could be due to multicollinearity between the independent variables. all my exposure variables are categorical and some are non binary. I repeated the regression analysis using dummy variables (to check for interactions) and I ran the collinearity diagnostics (using the linear regression though my variables were categorical). The tolerance values and VIf did not show any significant collinearity. But the warning still pops up and spss drops out one of my categories from a certain variable. Wondering if you could help with multicolleniarity in logistic regression or if you have already uploaded a video for the same?
Thx! It's quite helpful. The topic of my dissertation is exposure to pro- and anti-smoking media messages and their association with intention to smoke among adolescents. I planned to use logistic regression. But I am wondering if it makes sense to use hierarchical multiple regression and put pro- and anti-smoking media messages in the second block. Is it suitable to do that?
Could you please post a video using a step-wise logistic regression? :)
Thank you for the video. What if the F change value in the Model Summary table is not significant (greater than .05) for model 2 and 3, but the F value in the Anova table is significant? Should one be interpreting models 2,3 and their corresponding beta coefficients?
That means that the model is a significant predictor but the change in the model is not significantly different or the change is not significant.
I saw many HMR in the research papers contain 3 models: model 1- a main IV; model 2-added some more IVs and model 3- their interactions.
If my research doesn't want to investigate the main effects of IV added in the model 2, I just wanna know if there is any moderating effect on the relationship in model 1, can I just combine model 2 and 3. i.e.: model 1: a main IV, model 2- added moderators and the interaction between the main IV and moderators
Thanks !!!
Yes, that is correct.
hi, super useful video, but there is one question I don't understand. In the hierarchical regression, if I have category variables : -.5 = public high school and +.5 = private high school. What could be some of the reasons why I need to code like this?
I'm not sure you do need to code that way. You should be able to use whatever values you want. Personally I would avoid using negative numbers.
Hi thank you for this. I really never understood what was meant when my professor told me to "control for the demographic variables" in my model. :) Finally found it!!!
A question though, I have up to 32 independent variables ( due to dummy coding of my demographic variables => f.e. almost 14 dummy codes of NGO names ) and I see when doing a multiple hierarchical regression analysis some of the variables get excluded. Given its 3 dummy codes of my demographic variable, is this really a problem? What should I do or can I just interpret the results with no issues.
Thanks so much!!!
SyrahStormx Based upon your explanation you should be able to run the analysis and interpret the result normally.
Hi,
I'm wondering how many variables can you control for at once when you want to look at one IV's effect on the dichotomous DV? Is there a limit?
To the best of my knowledge there is not a limit.
Hi there. Thanks for the video, it was very helpful. I hate a question regarding repeated measures as a dependent variable. Would this work the same way if my dependent variable has multiple (i.e. 16) time points it has been measured at? I've transformed it into long data, so it just comes up as one variable now. Would it still work that way?
+cruz11 Technically it should but I would be concerned that conclusions you make might be biased due to the aggregation of the outcome.
Hi, I'm translating a text (outside my field of specialty!) that mentions "linear hierarchical multiple regressions" and I'm having a hard time understand. Is a hierarchical regression necessarily a multiple linear regression? Or does "hierarchical" refer simply to the order in which the variables are entered in a linear multiple regression? Many thanks to anyone who can help!
No it does not always include multiple predictors. It simply refers to the sequential placement of variables in the prediction model and being able to measure the effect of the variable entered in the first block.
how many variables can you put in model 1 and model 2. Is there a limit? When would you use a 3rd or even 4th model?
There is not limit to the number of variables you can use or the number of models.
thank you.
I like your vid. Very handy.