Graphing power curves given simulation results

[CEdward]

I recently ran some simulation analyses for a multi-level randomized control trial design in a separate programming software and generated the results. Briefly, the design has a level of clustering at the individual-level and cluster-level. I calculated the power given different combinations of number of clusters and individuals.

Number of clusters	Cluster size	Intervention coefficient	Estimated power
3	5	5	0.4474474
3	10	5	0.7377377
3	15	5	0.8538539
3	20	5	0.9489489
3	25	5	0.982983
6	5	5	0.7127127
6	10	5	0.956957
6	15	5	0.995996
6	20	5	0.997998
6	25	5	1

What I am wondering is whether it would be possible to plot a power curve given those results, with power on the y-axis, number of clusters on the x-axis, and separate curves for the number of individuals per cluster.

I took a look at some of the commands that are available and the power command seems like a potential candidate: https://blog.stata.com/2019/01/10/ca...-1-the-basics/. However, I am not sure that it accommodates my current situation where I am bringing in results from external to Stata.

Is this something that is workable?

Version: Stata 15.1

[Maarten Buis]

clear
input N_clust clust_size coef power
3   5   5   0.4474474
3   10  5   0.7377377
3   15  5   0.8538539
3   20  5   0.9489489
3   25  5   0.982983
6   5   5   0.7127127
6   10  5   0.956957
6   15  5   0.995996
6   20  5   0.997998
6   25  5   1
end

separate power , by(clust_size) veryshortlabel


twoway line power? power?? N_clust,   ///
    xlab(3 6) ytitle(simulated power) ///
    xtitle(number of clusters)        ///
    legend(subtitle("cluster size")   ///
           cols(3))                   ///
    lcolor("254 240 217"              ///
           "253 204 138"              ///
           "252 141  89"              ///
           "227  74  51"              ///
           "179   0   0")             ///
    scheme(s1color)

[CEdward]

This is perfect, Dr. Buis. I didn't realize it was as simple as using twoway line so I clearly over thought it.

[CEdward]

I was playing around with this further to add some extra details and I am not sure why I get an error. Likely it is a syntax problem, but I can't spot it:

twoway line Power? Power?? clusters,
xlab(3 6 9 18 36) ///
ytitle(Power (1 - β)) xtitle(Number of Clusters), ///
legend(subtitle("Cluster Size") ///
cols(1) ///
ring(0) ///
bplacement(seast)) ///
lpattern(solid dash dash_dot shortdash longdash) ///
lcolor(black black black black black) ///
region(lcolor(white))) ///
scheme(s1color)

The goal here is to get a white plot region with no border around the legend. Everything else works fine when I remove region(lcolor(white))), but I am not sure why I get this error

option region() not allowed r(198);

Does anyone understand why this happens? Likely a syntax error?