Is it possible to apply Multilevel models when data set consists of 3 IVs of categorical nature and 1 or more continuous DVs ?.......... My data set consists of 3 categorical IVs (1st IV consists of 2 categories, Second IV also consists of 2 Categories and 3rd IV consists of 3 categories).
Hello Dr. Crowson, please why did you group-mean centered the variable gender (genderCWC) and standardized reading test scores (standLRTcwc) instead of using the variables in their original forms? Slides 3 and 9 The Level 1 predictors include gender (genderCWC) and standardized reading test scores (standLRTcwc) - both group-mean centered. [Prior to centering, the gender variable was coded 0=male, 1=female] The Level 2 predictors include compositional variables at the school level: (a) proportion of females within schools and (b) the average standardized reading test score within schools.
Thank you so much for sharing Dr Mike! I would like to ask you a question. If I have standardised my predictors, then should I still need to create the centring-within-cluster variables? Thank you very much! And I am looking forward to your reply.
It is because, after the standardisation, my grand mean values of predictors will be "0", and the predictors with the centre within the cluster will remain the same. By extension, my level 2 predictor will be "0" as well. This weird situation perplexes me......
Thanks for a great, very clear video! I have a question about the level 2 equation. I've been told that the Level 2 effects are independent of the Level 1 effects so including or excluding the Level 1 variables should not affect the Level 2 relationships. I was told this by people who use R to analyze their MLMs. When I exclude the Level 1 variables, the level 2 effects are the same as I would get from a regular multiple regression (as I would expect). However, when I include the Level 1 predictors, the Level 2 relation ships are changes. Your thoughts?
Hello, Dr. Mike.Thanks for your video.I am a doctoral student in Japan. I have a question about simple slope analysis. What I would like to ask is, if we centered the moderators(binary variables), do we use absolute values such as regression coefficients when we enter them in the preachers website? We ran a two-level multilevel analysis and entered between-level interaction items. We did the analysis by centered the moderator of the binary variable, which I think is the variable w2 in case2 on Preachers website. The regression coefficients for the binary variable and the interaction term were negative value. However, if I type in case2 as is, I get different results than I expected. Referring to your video on how to use jamovi, I had to entered in absolute values in order to get the same results as when I did the same analysis in jamovi. Is this correct? If you know, I would appreciate it if you could let me know.
Thank you for your video. Do you have any resources to help in multilevel modelling using SPSS on longitudinal data? I have data from two Waves and wish to include time as a level 1 (time-varying) predictor with my IVs and co-variates as Level 2 (time-invariant) predictors. I'm just not sure how to create my 'Time' variable to represent rate of change. Is it simply T2-T1? Any resources or information would be much appreciated. Thanks in advance
Hi Dr. Crowson, my level 1 outcome variable is composite score of three survey items with likert scales (1-5), so the outcome variable range is from 3 to 15. The outcome variable is not left-skewed. Is it still considered continuous? I checked GLMM but am not sure which target distribution and link will be appropriate for my data? Thank you.
Hello Dr. Mike, thanks for the video. I have a slightly similar data sample but a little different variable. We have data from different classrooms and want to check if the class composition (in this case, portion of non migrant and migrant students in each class) has an influence in our DV national identity of students. My supervisor informed that this cannot be done with SPSS, only with a multilevel analysis on R or Mplus. Is this correct? Could I use something from this video or do you have perhaps another one showing how to do it? I like your way of teaching and it is also easy to understand as the examples are similar to my data set. Thank you!
Hi there, Beatriz. SPSS can be used for multilevel analysis - and actually this video is an example of that. I also have another video demonstrating multilevel analysis in SPSS: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html Here is the link to an entire book devoted to multilevel modeling in SPSS (which you should check out): www.routledge.com/Multilevel-and-Longitudinal-Modeling-with-IBM-SPSS/Heck-Thomas-Tabata/p/book/9780415817110 In your scenario, classroom serves as a level 2 unit whereas your students (level 1 units) are nested within those classroom. It is possible to model classroom-related predictors (those associated with the Level 2 units/classroom level) as predictors of a level 1 outcome; and that is demonstrated in this video (and the other referenced above). R and MPLUS are programs that also allow you to perform multilevel modeling (here's an example I made using R: th-cam.com/video/8r9bUKUVecc/w-d-xo.html). But they are not the only ones that allow you to do the analysis. Keep in mind that if you have only a small number of classrooms you are working with, that could pose a problem - as multilevel analysis generally requires a fairly large number of level 2 units (even though I have some literature indicating that perhaps a smaller number is ok; I'm not well-versed on that literature, however). I believe that you should have somewhere in the neighborhood of 30 level 2 units (e.g., classrooms) to use maximum likelihood estimation (the standard approach) to estimate model parameters; however, 15 or greater may be ok when estimating parameters with restricted maximum likelihood (which is actually the initial default in SPSS). I hope this is helpful to you. Best wishes!
Hi there. I have a video that demonstrates how these can be computed here: th-cam.com/video/NSFCibkocRA/w-d-xo.html Also, I have a much more 'up-to-date' video that provides a lot more details on multilevel modeling with SPSS here: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html . It also includes a powerpoint link as well. And there are more resources here: sites.google.com/view/statistics-for-the-real-world/contents/multilevel-modeling-and-panel-regression Hope this helps!
Hi Chris, yes you can certainly do that using SPSS. I cover slopes as outcomes in the video: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html Although I don't use the term 'slopes-as-outcomes', the cross-level interaction tested in the video is addressing whether the level 1 slope is a function of a level 2 predictor. I hope this helps!
Hi Adonis, the Generalized linear mixed model allows you to perform multilevel modeling where your level 1 outcome variable can be continuous or categorical (e.g., ordered categorical or nominal, count, etc., which generalizes to multilevel logistic, Poison regression etc). For example, here is something I put together awhile back on multilevel logistic regression using the generalized linear mixed model approach: th-cam.com/video/roWTULimNPk/w-d-xo.html Cheers!
Hi there. There is no global goodness of test of per se (as you might be used to in the context of SEM). However, you can evaluate the fit of a model by comparing it against others. Multilevel analysis tends to be carried out in a series of steps - often proceeding from a null model to increasingly complex models. When you have nested models, you can test whether the fit of the more complex model represents a significant improvement in fit relative to the reduced model using a chi-square difference test. I have a video (and calculator) for doing this here: th-cam.com/video/bzqMFnl3IV4/w-d-xo.html You could also theoretically compare AIC values with competing models, with the preferred model among two being the one with the lowest AIC. [Although AIC was originally developed to compare non-nested models, it can still be used with nested models.] By the way, I have a more recent video on multilevel modeling in SPSS here: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html I hope this is helpful to you. Best wishes.
Hello Mike, Many thanks for your tutorials. I am very new to HLM and I have a question. My results indicate that ICC based on Model 1 is 0.010150305 and Intercept [subject = Industry] Variance is insignificant (p=0.545 which is 0.273 even if considered as one-tailed), while residual is significant. Now, should I continue further with that particular dataset to perform HLM? Or should I just infer that the response (dataset) does not show evidence to perform HLM as this concern is raised by my manuscript's reviewer? Your response is much appreciated. Thanks!
Hi Szeshing, I just saw your question. You can can compute pseudo-r-squares for within and between components using the Level 1 and Level 2 variance components [see table containing 'Estimates of covariance parameters' at the bottom of your output; look under Estimate]. For the between-groups pseudo- R-square, you can divided the variance of the Level 2 intercepts by [variance of level 1 residuals + variance of Level 2 intercepts]. You could compute a proportion of unexplained variance in y in a similar manner: variance at level 1 divided by variance of level 1 residuals+variance of Level 2 intercepts. I hope this helps! Another option might be to generate a pseudo R-square reflecting the proportionate reduction in a variance component associated with a full model against a reduced model (such as an intercept only model). For example, the proportionate reduction in variance in intercepts (i.e., between variance) can be computed as: 1 - [Level 2 variance full model / Level 2 variance reduced model ] I hope this helps!
Thank you for this video. I was wondering, how did you convert gender to be a continuous variable for the purposes of this analysis? I have a grouping variable as the predictor, and I am wanting to run a series of simulations to assess the power of my study. However, the generalised linear mixed models approach does not allow for split file specifications so I can view output from each simulation. The approach (above) does, but doesn't work for categorical variables.
Gender is technically not a continuous variable. However, since it is a binary variable it can be treated 'as if' it is continuous in the regression model (whether it is OLS regression or even multilevel regression, a binary predictor can be included in the model). The coding of 0 and 1 is based on a dummy coding system. The codings of 0 and 1 facilitate interpretation of the intercept in regression models. You might also check out a more recent video too: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html
Is it possible to apply Multilevel models when data set consists of 3 IVs of categorical nature and 1 or more continuous DVs ?..........
My data set consists of 3 categorical IVs (1st IV consists of 2 categories, Second IV also consists of 2 Categories and 3rd IV consists of 3 categories).
Hello Dr. Crowson, please why did you group-mean centered the variable gender (genderCWC) and standardized reading test scores (standLRTcwc) instead of using the variables in their original forms?
Slides 3 and 9
The Level 1 predictors include gender (genderCWC) and standardized reading test scores (standLRTcwc) - both group-mean centered. [Prior to centering, the gender variable was coded 0=male, 1=female] The Level 2 predictors include compositional variables at the school level: (a) proportion of females within schools and (b) the average standardized reading test score within schools.
Wao! Excellent! Dr. Crowson Thank you!
Thank you Leslie for visiting! I'm glad you found the video useful!
Hello. Do you offer specific help for projects? Could I contact you?
Thank you so much for sharing Dr Mike!
I would like to ask you a question. If I have standardised my predictors, then should I still need to create the centring-within-cluster variables?
Thank you very much! And I am looking forward to your reply.
It is because, after the standardisation, my grand mean values of predictors will be "0", and the predictors with the centre within the cluster will remain the same. By extension, my level 2 predictor will be "0" as well. This weird situation perplexes me......
Thank you for making this video! It is really helpful.
Thanks for a great, very clear video! I have a question about the level 2 equation. I've been told that the Level 2 effects are independent of the Level 1 effects so including or excluding the Level 1 variables should not affect the Level 2 relationships. I was told this by people who use R to analyze their MLMs. When I exclude the Level 1 variables, the level 2 effects are the same as I would get from a regular multiple regression (as I would expect). However, when I include the Level 1 predictors, the Level 2 relation ships are changes. Your thoughts?
Hello Mike, the link containing data and PowerPoint is unavailable due to technical issues. Can you please look into it?
Hi Qaiser, The links to the data and powerpoint should now be functional. Thanks for letting me know of the problem. Cheers!
Hello, Dr. Mike.Thanks for your video.I am a doctoral student in Japan.
I have a question about simple slope analysis. What I would like to ask is, if we centered the moderators(binary variables), do we use absolute values such as regression coefficients when we enter them in the preachers website?
We ran a two-level multilevel analysis and entered between-level interaction items. We did the analysis by centered the moderator of the binary variable, which I think is the variable w2 in case2 on Preachers website. The regression coefficients for the binary variable and the interaction term were negative value. However, if I type in case2 as is, I get different results than I expected. Referring to your video on how to use jamovi, I had to entered in absolute values in order to get the same results as when I did the same analysis in jamovi.
Is this correct? If you know, I would appreciate it if you could let me know.
Hi Mike! Is it possible to get an R^2 for a multi-level analysis? Hope to hear from you!
Thank you for your video. Do you have any resources to help in multilevel modelling using SPSS on longitudinal data? I have data from two Waves and wish to include time as a level 1 (time-varying) predictor with my IVs and co-variates as Level 2 (time-invariant) predictors. I'm just not sure how to create my 'Time' variable to represent rate of change. Is it simply T2-T1? Any resources or information would be much appreciated. Thanks in advance
Very useful, thank you very much!
Hi Dr. Crowson, my level 1 outcome variable is composite score of three survey items with likert scales (1-5), so the outcome variable range is from 3 to 15. The outcome variable is not left-skewed. Is it still considered continuous? I checked GLMM but am not sure which target distribution and link will be appropriate for my data? Thank you.
Thank you so much
Hello Dr. Mike, thanks for the video. I have a slightly similar data sample but a little different variable. We have data from different classrooms and want to check if the class composition (in this case, portion of non migrant and migrant students in each class) has an influence in our DV national identity of students. My supervisor informed that this cannot be done with SPSS, only with a multilevel analysis on R or Mplus. Is this correct? Could I use something from this video or do you have perhaps another one showing how to do it? I like your way of teaching and it is also easy to understand as the examples are similar to my data set. Thank you!
Hi there, Beatriz. SPSS can be used for multilevel analysis - and actually this video is an example of that. I also have another video demonstrating multilevel analysis in SPSS: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html
Here is the link to an entire book devoted to multilevel modeling in SPSS (which you should check out): www.routledge.com/Multilevel-and-Longitudinal-Modeling-with-IBM-SPSS/Heck-Thomas-Tabata/p/book/9780415817110
In your scenario, classroom serves as a level 2 unit whereas your students (level 1 units) are nested within those classroom. It is possible to model classroom-related predictors (those associated with the Level 2 units/classroom level) as predictors of a level 1 outcome; and that is demonstrated in this video (and the other referenced above). R and MPLUS are programs that also allow you to perform multilevel modeling (here's an example I made using R: th-cam.com/video/8r9bUKUVecc/w-d-xo.html). But they are not the only ones that allow you to do the analysis.
Keep in mind that if you have only a small number of classrooms you are working with, that could pose a problem - as multilevel analysis generally requires a fairly large number of level 2 units (even though I have some literature indicating that perhaps a smaller number is ok; I'm not well-versed on that literature, however). I believe that you should have somewhere in the neighborhood of 30 level 2 units (e.g., classrooms) to use maximum likelihood estimation (the standard approach) to estimate model parameters; however, 15 or greater may be ok when estimating parameters with restricted maximum likelihood (which is actually the initial default in SPSS).
I hope this is helpful to you. Best wishes!
Please tell me how you calculated the gender_mean and gender CWC, so that I would be able to derive the same in my research data.
Hi there. I have a video that demonstrates how these can be computed here: th-cam.com/video/NSFCibkocRA/w-d-xo.html
Also, I have a much more 'up-to-date' video that provides a lot more details on multilevel modeling with SPSS here: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html . It also includes a powerpoint link as well. And there are more resources here: sites.google.com/view/statistics-for-the-real-world/contents/multilevel-modeling-and-panel-regression Hope this helps!
Hi Dr. Mike, can SPSS be used to conduct slopes-as-outcome model of the hierarchical linear modeling, without using the HLM software package?
Hi Chris, yes you can certainly do that using SPSS. I cover slopes as outcomes in the video: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html
Although I don't use the term 'slopes-as-outcomes', the cross-level interaction tested in the video is addressing whether the level 1 slope is a function of a level 2 predictor.
I hope this helps!
thanks for this Sir! I am just wondering if this analysis is similar to the Generalized linear mixed model (GLMM?
Hi Adonis, the Generalized linear mixed model allows you to perform multilevel modeling where your level 1 outcome variable can be continuous or categorical (e.g., ordered categorical or nominal, count, etc., which generalizes to multilevel logistic, Poison regression etc). For example, here is something I put together awhile back on multilevel logistic regression using the generalized linear mixed model approach: th-cam.com/video/roWTULimNPk/w-d-xo.html Cheers!
Do you know how you can test for goodness of fit for a model of this type?
Hi there. There is no global goodness of test of per se (as you might be used to in the context of SEM). However, you can evaluate the fit of a model by comparing it against others. Multilevel analysis tends to be carried out in a series of steps - often proceeding from a null model to increasingly complex models. When you have nested models, you can test whether the fit of the more complex model represents a significant improvement in fit relative to the reduced model using a chi-square difference test. I have a video (and calculator) for doing this here: th-cam.com/video/bzqMFnl3IV4/w-d-xo.html
You could also theoretically compare AIC values with competing models, with the preferred model among two being the one with the lowest AIC. [Although AIC was originally developed to compare non-nested models, it can still be used with nested models.]
By the way, I have a more recent video on multilevel modeling in SPSS here: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html
I hope this is helpful to you. Best wishes.
Hello Mike,
Many thanks for your tutorials. I am very new to HLM and I have a question.
My results indicate that ICC based on Model 1 is 0.010150305 and Intercept [subject = Industry] Variance is insignificant (p=0.545 which is 0.273 even if considered as one-tailed), while residual is significant. Now, should I continue further with that particular dataset to perform HLM? Or should I just infer that the response (dataset) does not show evidence to perform HLM as this concern is raised by my manuscript's reviewer?
Your response is much appreciated. Thanks!
I think you should not proceed with HLM further because the ICC value is below the 0.05 threshold.
Hi Mike, is it possible to obtain within-group and between-group R-square in SPSS? xxx
Hi Szeshing, I just saw your question. You can can compute pseudo-r-squares for within and between components using the Level 1 and Level 2 variance components [see table containing 'Estimates of covariance parameters' at the bottom of your output; look under Estimate]. For the between-groups pseudo- R-square, you can divided the variance of the Level 2 intercepts by [variance of level 1 residuals + variance of Level 2 intercepts]. You could compute a proportion of unexplained variance in y in a similar manner: variance at level 1 divided by variance of level 1 residuals+variance of Level 2 intercepts. I hope this helps!
Another option might be to generate a pseudo R-square reflecting the proportionate reduction in a variance component associated with a full model against a reduced model (such as an intercept only model). For example, the proportionate reduction in variance in intercepts (i.e., between variance) can be computed as: 1 - [Level 2 variance full model / Level 2 variance reduced model ]
I hope this helps!
@@mikecrowson2462 It does help, thanks.
Thank you for this video. I was wondering, how did you convert gender to be a continuous variable for the purposes of this analysis? I have a grouping variable as the predictor, and I am wanting to run a series of simulations to assess the power of my study. However, the generalised linear mixed models approach does not allow for split file specifications so I can view output from each simulation. The approach (above) does, but doesn't work for categorical variables.
Gender is technically not a continuous variable. However, since it is a binary variable it can be treated 'as if' it is continuous in the regression model (whether it is OLS regression or even multilevel regression, a binary predictor can be included in the model). The coding of 0 and 1 is based on a dummy coding system. The codings of 0 and 1 facilitate interpretation of the intercept in regression models. You might also check out a more recent video too: th-cam.com/video/x5Z5KYODwNk/w-d-xo.html
@Mike Crowson thanks for the helpful work you always share. what's more, is multilevelling model same as hierarchical regression?
thanks in advance!
Thanks for your lecture
You're very welcome! best wishes
super helpful, thank you!!
I'm so glad the video was helpful. Thank you for visiting!