Kind request for advice on how to analyze lagged variables in regression - tsset?

[Matthew Berg]

Dear Statalist,

I have experimental data on how individual retail investors set up portfolios. Every participant completes rounds 1-4 in order. Now I would like to analyze whether the decisions in one round affect the decisions in another round. For that, I was thinking of the following linear regression: alloc round 4 = intercept + alloc round 1* beta1 + alloc round 2*beta2 + alloc round * beta3 + alloc round 4* beta4 + error term, clustered using vce(cluster participant ID).
Where round 4 could equal t, round 3 t-1, round 2 t-2, and round 1 t-4

I am unsure how to implement this in Stata.
I feel like this seems to be a time-series regression due to the lagged components of the round. I am beware of using tsset as some rounds will have to be dropped (due to illogical responses), which will create an unbalanced sample.

I am most grateful for any help or advice on this matter.

Thank you very much in advance!

[George Ford]

is the data long or wide?

[Matthew Berg]

The data was wide, now it has been reshaped to long, where each round has its own row. So four rows (= 4 rounds) per individual.

[George Ford]

it's easy in wide format.

reg y4 y3 y2 y1

if long, then

reg y l1.y l2.y l3.y

or xtreg , fe

[Matthew Berg]

Thank you very much, George! You have been most helpful!