search

bquistorff / synth_runner

A tool to run a pool of synthetic controls, conduct inference, and produce visualizations.
42 stars 27 forks source link

Error message - uno() invalid -- invalid numlist has too many elements #32

Open tricaudc opened 5 years ago

tricaudc commented 5 years ago

Dear Brian,

Thanks for the very useful synth_run package. I am using it on Stata (Stata version: 15.1 MP). My setting is the following: I have a balanced panel consisting of 16,362 units of which 1,297 are treated and 15,065 are control and I am following them over a period of 15 years (from 2004 to 2018). The treatment period begins in 2010. I am using 16 predictor variables. Note that the outcome variable I am considering is never missing.

When running the code I obtain the following error: "Error estimating treatment effect for unit 1075". After reading your answer in stackoverflow (https://stackoverflow.com/questions/48552687/error-estimating-treatment-effects-synth-runner), I used "trace" and recover the issue, the error message concerning this unit is the following: "uno() invalid -- invalid numlist has too many elements".

Do you know how I could solve this issue? Would it imply reducing my control group first, but then by how much?

Thanks in advance for your help

Preliminaries

Before submitting an issue, please check (with x in brackets) that you:

sannanphan commented 3 years ago

Hi @tricaudc, did you find a solution to your problem?

liqizheng0328 commented 1 year ago

struggle with same question, it does not work in multiple treated group

tricaudc commented 1 year ago

Sorry just seeing this, I did not find the solution unfortunately

On Wed, Oct 5, 2022 at 9:29 AM liqizheng0328 @.***> wrote:

struggle with same question, it does not work in multiple treated group

— Reply to this email directly, view it on GitHub https://github.com/bquistorff/synth_runner/issues/32#issuecomment-1268662898, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMUVUVHZ33WKMMEV4ESD4UTWBWUHJANCNFSM4IEZFT2A . You are receiving this because you were mentioned.Message ID: @.***>

Jolijn99 commented 1 year ago

Hi! Does anyone have a solution for this issue? @tricaudc @sannanphan @liqizheng0328 @bquistorff @janvavra That would help me out a lot :)

liqizheng0328 commented 1 year ago

Hi! Does anyone have a solution for this issue? @tricaudc @sannanphan @liqizheng0328 @bquistorff @janvavra That would help me out a lot :)

I suggest that you use gsynth (Yiqing Xu, 2017) package in r

bquistorff commented 1 year ago

If there's a reproducible example, that would help diagnose the issue. It's likely caused by a limit internal to Stata numlists (ref).

joannarmathias commented 4 months ago

I think this error is from having too many controls to choose from. I resolved it by limiting my sample through dropping very dissimilar potential control units.