confidence interval of coefficient of variation

[Jeong-Kweon Park]

Hi! All,
How can I calculate the 95% confidence interval of coefficient of variation?

[Oded Mcdossi]

You can use bootstrap command to calculate the CI. here is a small program to do that:

cap program drop coef_v
program define coef_v, rclass
quietly summarize `1'
return scalar CV =  r(sd)/ r(mean)
end

sysuse auto, clear
bootstrap r(CV), reps(200) bca: coef_v price
estat bootstrap, all

[Jeong-Kweon Park]

Thank you so much, Oded!
I have a question.
Could you tell me the meaning of 'reps(200)'?
Default is 50.

[Jeong-Kweon Park]

Hello, Oded!
I understood the meaning of 'reps(200)'.
I have another question.
What is `1'?

[Oded Mcdossi]

The `1' in Stata programming language refers to the first variable in the program. the program above is based on results from summarize command, so only the first variable after the name of the program, in this case coef_v, is passed to the summarize command.

[Jeong-Kweon Park]

If I could use 'bootstrap r(CV), reps(200) bca: coef_v price mpg rep78 headroom' command?

[Nick Cox]

No; the program only takes a single variable. You would need to loop over variables.

sysuse auto, clear  
foreach v in price mpg rep78 headroom {
      bootstrap r(CV), reps(200) bca: coef_v `v'    
      estat bootstrap, all
}

[Jeong-Kweon Park]

Thank you so much, Nick!