Hi all, I am working on a study of grant applications and selections. I have about 8 years worth of data, and am looking at the research term – PIs and CO-Is. The goal is to see if certain characteristics of the person (e.g. demographics, experience) predict the likelihood of getting a grant.
(Fyi, I previously posted about building a model for PIs only here: https://www.statalist.org/forums/for...model-question)
However, now I have data on BOTH PIs and the CO-Is. The grant is awarded to the whole team. Thus, I was thinking I have to now account for the team in the modeling. A person could be a PI on more than one grant, and be selected as a PI on more than one grant. A person could apply as a PI on one grant and be a CO-I on another grant, for example, and have both grants selected.
The dataset is unique on the unique identifier for the person (ID_person) and the team_id (ID_team)
I was thinking a crossed random effects model may be appropriate here, since it seems that observations are nested in a combination of individuals and teams.
Here is the code I came up with, but I’m not sure it’s correct (?). In particular, I am not sure about the random effects part.
melogit awarded independentvariables || all:R.ID_team || ID_person:, or
A couple of additional questions:
1. I have 8 years worth of data. There is interest in examining the data for each year and also for all years combined. Does it make sense to run the model separately for each year and also for all years combined?
2. I mentioned in the previous post that some of the specific grant areas a person applies to can give out more grants than others, and that it can depend on the year. Across the 8 years, there are about 150 year-grant area combinations. Since this model already has 2 random effects, it seems that it might be best that these are included as dummy variables (grant-area year) in the model. However, this seems like it would be a very complicated model.
3. Does it make sense to have a dummy variable for whether someone is a Co-I or PI?
Thank you in advance for any advice!
MJ
Related Posts with model with crossed random effects
convert excels files to .dta filesHello, I am very beginner in Stata, so sorry if my questions seem silly. I am reading my friend's …
exponential scale for plot axesI am trying to plot distributions of article citations. Due to the skewness of citations, I want the…
Comparing coefficient across subsamples in the presence of interactive effectsI am interested in estimating a relation between two variables across multiple subsamples but I am h…
Merge fileHello, I am reading instruction on merging files on stata in this link: https://www.stata.com/manua…
Grstyle graph and transparencyDear All, I have the following code: Code: sysuse bplong.dta, clear keep if when==1 reg bp agegr…
Subscribe to:
Post Comments (Atom)
0 Response to model with crossed random effects
Post a Comment