What are the usually variables in level 2 of multilevel model

[David Wong]

Dear everyone,

I am working on one paper of which data I collected from online survey from US citizens. I launched this survey through the platform of Qualtrics, in terms of random sampling. Now I have around 400 valid individual answers. I visualized the location of these participants and graph indicates that their distribution is dispersive. My data include basic demographic variables, such as gender, income, employment status, and also my key variables in individual level.

My question is that

"What variables usually are used as level 2 variable?"

As far as I know, in the literature, we usually use country, company, university, school as the level 2 variables, is it right? Is there any paper using gender, education, employment status as the level 2 variable?

I am totally new to multilevel model but I have a strong interest on it. Many thanks for any help and suggestion.

David

[Marcos Almeida]

Hello David,

Welcome to the Stata Forum / Statalist.


Your question relates to the core-knowledge concerning mixed models and I strongly recommend you take a close look in the basic information before delving with such type of model.

That being said, in a 2-level mixed model, for example, level 2 is the 'grouping' variable or cluster variable. By 'level 2', so as the model works appropriately, you should expect a variable with many groups. Some say at least 30, some say at least 10. Sex, usually, presents just 2...

With regards to the 'explanatory' variables, you may add level-1 as well as level-2 variables in the 'left' parcel of the - mixed - command.

I hope that helps.

[David Wong]

Dear Marcos,

I am sorry to be late to reply. Yes, your answer is very helpful, which is exactly what I expect. I have just one follow-up question. Is there any reference, in terms of textbook or paper, discussing the minimum number of groups in this 'grouping' variable which we can add into level 2? Thank you so much.

Best regards,
David

[Marcos Almeida]

Hello David,

I recommend you start with Stata Manual so as to get an introductory grasp of the matter.

As a suggestion, you may start the 'caveats' by reading: Regression analysis of country effects using multilevel data: a cautionary tale. ISER Working Paper Series 2014-2014.

[David Wong]

Dear Marcos,

Thank you for time and kind help. I have downloaded this paper which is very useful.

Best regards,
David

[Marcos Almeida]

Hello, David,

Great that you liked it.

Sorry , I forgot to write the names of the authors: Bryan, M and Jenkins, S.J.

I'm on a trip and typing it all in the smartphone is not that easy.