Faded Box Plot

[Donald Likosky]

I am using Stata/IC 14.2. When I develop box plots, sometimes one of the box plots appear faded. I have enclosed an example of what I am speaking of (see far right box plot).

In the code below, I am graphing a continuous variable ("volume") across categories of a nominal variable ("randomnum").

Any thoughts would be welcomed.

Donny

*Prime Volume - RAP Volume*
capture drop median
egen median= median(volume), by(randomnum)
sum volume, det
di r(p50)
local x=r(p50)
local x: di %9.1f `x'
graph box volume if volume>=100, yline(`r(p50)') ylabel(, angle(0)) over(n_no_circ, label(angle(rvertical)) sort(median)) nooutsides ytitle("Prime Vol - RAP Vol (ml)") title ("Net Prime Volume (ml) by Center") note("Not Using Circulatory Arrest" "Net Vol=Prime Vol - RAP Vol" "Only Includes Cases Matched to Surgical Records" "Red line:median value across the database is:`x' ml")

[Paul T Seed]

You may be trying to draw boxes where there is no data (or very little). If so, combining boxes may be the solution.
You can start with

tab n_no_circ
bys n_no_circ : su volume

and then combine the empty or near-empty boxes.

Another possibility is that if you have too many groups when using over() with boxplots, the later boxes are not always correctly drawn. The obvious thing to do is reduce the number of groups, but that is not really a solution. This looks like a problem within Stata, so I suggest hat you contact Stata help & see what they can say. Perhaps, .