Dotplot or similar to draw dots and confidence interval whiskers of specified length: surely this can't be difficult!

[Joshua Nealon]

Dear all,

There have been a few threads about problems with confidence intervals on graphs – I’m afraid I am in the midst of one.

I am trying to make a dotplot, with the dots indicating regression coefficients; and whiskers above and below, confidence intervals. These data originate from a series of logistic regressions, individually exported into a table of the format:

* Example generated by -dataex-. To install: ssc install dataex
clear
input str9 country float(coef ci_low ci_high)
"country_a" .205 .158 .267
"country_a" .081 .055  .12
"country_b" .094 .083 .105
"country_c" .105 .089 .124
"country_c" .056 .036 .086
end

The graph I would like to plot is therefore something like:

dotplot coef, over(country)

combined with:

twoway (rcap ci_low ci_high), over (country)

But of course the syntax for the second is incorrect because string variables [country] can’t be used with -twoway- graphs.

I feel this should be simple with a scatterplot or dotplot, but cannot find a way to simply specify the values of my dots and whiskers on a graph with a discrete x-axis variable [note - this is only a problem because I'm specifying CI values rather than calculating from a dataset - otherwise, -ciplot- would do this].

I want to generate a graph which looks something like the version below (generated by changing the string values of my countries to numbers) but there is surely a simpler solution than re-labeling the axis?

Thank you very much for any advice – I am using Stata v. 15.

Josh.

twoway (rcap ci_low ci_high count) (scatter coef country)

[Carole J. Wilson]

Rather than -dotplot-, -twoway scatter- is more helpful. See this post: https://www.statalist.org/forums/for...ng-only-values

[Joshua Nealon]

Thank you very much for advice, Carole, and sorry for asking an answered question. I did this in the end by manually adjusting the x axis (anyway, I needed to "jitter" a lot and found the easiest way was to manipulate the axis.
Have a good day.
Josh.