lagging for gllamm

[Alyssa Beavers]

I am going to be running multilevel (longitudinal) logistic regression using gllamm. I need to lag most of my variables by one year prior to performing this analysis. Important to note is that I need to lag by subject, which is Garden_ID_ . My data is currently in the "long" format. For simplicity, in this example I will need to lag Var 1 and Var2, but not var 3.

Do I need to make new lagged variables to run gllamm, and if so what's the easiest way to do that?

Garden_ID_	Year	Var1	Var2	Var3
1	2012
1	2013
1	2014
1	2015
2	2012
2	2013
2	2014
2	2015

Many thanks,
Alyssa

[Andrew Musau]

My data is currently in the "long" format. For simplicity, in this example I will need to lag Var 1 and Var2, but not var 3.

Do I need to make new lagged variables to run gllamm, and if so what's the easiest way to do that?

Yes, if I recall well, gllamm (SSC) does not support time-series operators.

local vars "Var1 Var2"
xtset Garden_ID_ Year
foreach var in `vars'
gen L`var'= L.`var'

where LVar1 and LVar2 define the lagged variables.

[Andrew Musau]

There is a loop in #2, so I am missing the brackets

local vars "Var1 Var2"
xtset Garden_ID_ Year
foreach var in `vars'{
     gen L`var'= L.`var'
}

[Alyssa Beavers]

Hi Andrew,

Thanks so much! It appears to have worked. Question on the syntax: Stata recognizes "L." to mean lag?

[Jorrit Gosens]

Yes, see the documentation on time series operators: https://www.stata.com/manuals13/tstsset.pdf

[Alyssa Beavers]

Thank you Jorritt!