Introduction to multilevel linear models in Stata®, part 2: Longitudinal data
HTML-код
- Опубликовано: 31 янв 2013
- Explore the basics of using the xtmixed command to model longitudinal data using Stata.
Note that the xtmixed command was replaced by the mixed command in Stata 13. Also check out the xt suite of commands for panel data.
If you'd like to see more, please visit the Stata Blog: blog.stata.com/2013/02/18/mul...
www.stata.com
Copyright 2011-2019 StataCorp LLC. All rights reserved.
I am happy that I found this video, very helpful and well explained.
Great video (even five years later)! Thanks
I sometimes show the command syntax in the shorter videos but not as often in
the longer videos. If you would like to see examples of the syntax, click the
question mark at the bottom left of the dialog box and it will open the help
entry for that command. Every help entry has examples of the command syntax.
It would be great if you can make a tutorial about the use of post estimation for multilevel models!
Great video! I'm taking HLM class and I wish my professor showed some slides like yours, very clear. Thanks for sharing!
Thank you for the feedback! Did you see our videos on margins and marginsplot? We have a playlists on Using margins in Stata, and marginsplot is covered on the Profile plots and interaction plots in Stata playlist.
The command syntax is shown in detail at the Stata Blog. The title is "Multilevel Linear Models in Stata Part 2: Longitudinal Data" and the weblink is in the description of the video above.
You can perform a likelihood-ratio chi-squared test that the variance is equal to zero using the -lrtest- command after fitting the model both with and without the random coefficient of interest. You can see an example of using -lrtest- after -mixed- on pages 12-13 of the Mixed Effects Reference Manual available within Stata or the Support section of the Stata website.
Thank you for this video. It couldn't be clearer.
We are performing a test where the null hypothesis is on the boundary of the parameter space because the variance cannot be negative. Therefore, this test is conservative, as indicated in the note. For a detailed discussion, please refer to: Gutierrez, R., S. Carter, and D. M. Drukker. 2001. On boundary-value likelihood-ratio tests. Stata Technical Bulletin 60: 15-18.
SO HELPFUL!
Great video!tks
Thank you! The videos are very helpful. Are there any tutorials on mixed linear models with first-level and second-level variables, cross-level interactions as well as moderating effects between the first- and second-level variables?
Thank you!
Please note that there an error in the graph at approximately 1:00. The connection method in the graph should have been "connect(ascending)". You can do this by clicking Line Properties > Connecting Method > Ascending. You can see the correct graph at the Stata Blog.
Please provide more videos for multi-level models (including margins/marginsplot) to show how this might be prepared for dissemination/publication (i.e. how to present these models to others). Thanks!
Hi! Thank you so much for this blog, it helps a lot (I bought the book by RABE-HESKETH and SKRONDAL but couldn't find illustred top-down menu examples like you provide on RUclips). I was wondering if you also have a blog to help us navigate our way through "residual diagnostics for linear regression such as checking for outliers, influential observations, heteroskedasticity and normality"?
You can download the dataset used in this example by typing: use www.stata-press.com/data/r13/productivity.dta
Hi StataCorp LP, Amazing, thanks a lot. I have a small question, Can i apply this model also to my data where I expect the coefficients to be heterogeneous but I have 2 layers of data (full data, country) not like in your example (region and states)?
Thank so you much for this. I have a clarifying question. I am using Stata 15 and my output gives me the variance (var) coefficients, not the standard deviation. The command also changed from xtmixed to just mixed. Is this just a change in the format of the output that corresponds to the change in the command or could I be doing something wrong?
Please provide a description for a way to run the multilevel linear regression using ordinal responses?
Thanks a lot for this post. I have a question. How do I add controls of different levels of the model. I understand that if I want to add controls in the first level I have to type in stata xtmixed y x z, || municipality:, var. If I want to add municipal controls, should I simply add them after z?
Great video! The data has 4 levels(region, state, car, year), the random part of the results only has 3 (intercept) levels, don’t you need to add another level?
Hello. Anyone that has a video or code for displaying the result achieved by using the "xtmixed" command? I'm having a little trouble visually presenting them
please can statacorp lp put video on logit and probit model thank you
Hi
I am using mixed effects model. I had one small query. What is the reason for taking log values of independent variables and dependent variables? Can I proceed without taking log values?
전원우공부해
Hi
I am using mixed effects model. I had one small query. What is the reason for taking log values of independent variables and dependent variables? Can I proceed without taking log values?
Hi
I am using mixed effects model. I had one small query. What is the reason for taking log values of independent variables and dependent variables? Can I proceed without taking log values?
Log
transformations are sometimes used with skewed variables in order to fit a linear model. Whether you should use log-transformed variables depends on your data.