"observation number out of range" problem

[Fengyi Liu]

Hello everyone,
I am working on an Econometrics problem with a dataset. My command lines are as follows:

// Set number of observations
eststo clear
set obs 1000
// Generate variables
/// y = beta0 + beta1 * x + u.
g id = _n
g x = runiform()
g u = rnormal()
g y = 3*x + u
// Estimate the model using OLS
reg y x
eststo m1 //store the results

// Replicate each observation by 4
expand 4, gen(dupind)

// Estimate the model using OLS and the new sample
reg y x
eststo m2
// Cluster the standard error
reg y x, cluster(id)
eststo m3

// Show all results
esttab m*, se(3)

log close

but it always reminds me that
"observation number out of range
Observation number must be between 3,010 and 281,474,976,710,654. (Observation numbers are
typed without commas.)"
when it comes to "set obs 1000"

But my dataset contains exactly 3010 data points, and I really don't know where went wrong.

Thanks a lot to any help!

[Maarten Buis]

You have data in memory which has 3010 observations (rows), than you tell Stata to set the number of observations to 1000. That is obviously impossible without drastically changing the data, so Stata won't allow that. If you want to drop observations, you have to do so explicitly with either the drop or keep commands.

[Fengyi Liu]

My teacher provided this reference file and seems I didn't really understand the course...... Thank you very much! I will try it out.

[Maarten Buis]

Be careful, you need a really good reason for dropping observations. "Somebody gave me a file where they want a 1000 observations" is not a good reason.

[Nick Cox]

The code seems to imply that students are expected to create their own fake dataset. if so, then

clear 

to drop all data in memory is not just fine but expected. But you should clarify the problem with your teachers, not with us.

[Fengyi Liu]

okay，thank you all very much!