problems creating histogram of discrete values

[ben earnhart]

I have a really nice .do file based primarily on putexcel that summarizes a bunch of six-value scale variables, including tiny little histograms. The survey instrument is generally administered to small groups at a time, maybe eight to twenty people. It's basically an employee satisfaction scale, so there are very few who admit to being really negative, and very few who choose the highest category.

As far as I can tell, if every category has a response, my histograms are fine. But if a category is skipped, the histograms don't come out right; especially that people get shifted to the highest category. Below I have posted some example data, and the problematic part of my command file. If it would be useful I can post the whole command file, but the rest of it (including placing the tiny little histograms into Excel) seems to work just fine.

I have tried all kinds of options with hist to no avail. Every time I think I have the histograms right, one or a few of them are mangled. The options chosen below are certainly not the right ones. Remember: these are generated on small sample sizes, often with truncated data.

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(q2_1 q2_2 q2_3 q2_4 q2_5 q3_1 q3_2 q3_3 q3_4 q3_5 q4_1 q4_2 q4_3 q4_4 q4_5 q5_1 q5_2 q5_3 q5_4 q5_5 q6_1 q6_2 q6_3 q6_4 q6_5)
2 1 6 4 1 2 0 1 2 6 6 3 2 4 2 5 3 2 4 3 2 2 2 0 0
3 2 3 2 2 3 2 4 1 2 3 1 4 4 6 2 4 2 3 2 3 4 3 1 3
1 4 3 3 3 2 2 3 2 5 2 2 0 3 2 5 2 2 0 3 1 0 2 0 4
4 4 4 3 3 4 0 0 2 3 2 3 3 3 4 3 3 0 0 2 2 2 2 2 1
2 4 2 0 0 5 4 6 4 2 2 2 6 4 3 4 2 1 4 0 6 2 2 0 4
6 2 2 1 3 2 2 0 1 3 2 2 1 4 2 0 2 4 2 2 2 2 3 3 2
1 3 6 2 3 6 3 0 3 6 3 2 4 5 4 1 3 4 2 3 2 4 3 1 4
end

generating histograms:

gen tmp=.
*======reverse-coding and generating histograms
foreach var of varlist q2_1-q6_5 {
    replace `var'=6-`var'
    replace tmp=`var'
    hist tmp, bin(6) color(blue) ylabel(none) xlabel(none) start(0)
    sum tmp
    graph export `var'.png, as(png) replace wid(100) hei(40)
 }
 drop tmp

[ben earnhart]

BTW -- based on the help file for hist, I tried the option discrete, but things just get worse, with values shifted even more.

[Nick Cox]

What I see here are values from 0 to 6 so in principle I understand there are 7 bins, not 6.

I didn't inspect them all carefully, but something like this seems to work.

hist tmp, discrete barw(0.8) xsc(r(0 6)) color(blue) ylabel(none) xlabel(none) star

Belay that: Here's a good way to see them all at once:

gen id = _n
reshape long q, i(id) j(which) string
replace q = 6 - q

hist q, discrete barw(0.8) xsc(r(0 6)) color(blue) ylabel(none) xlabel(none) by(which)

I don't know why you want to suppress the axis labels, but that isn't the question.

[ben earnhart]

Nick --

Thanks for the thoughts. I'm doing it in a loop instead of by going long since I save each histogram for later insertion into an Excel spreadsheet.

When I use your first specifications, the axis still seems to slide around, sometimes starting at 0, sometimes at 1, etc It sometimes mostly works, like the commands I'd been trying..

Your second way of doing it all at once seems to keep the axis the same across graphs, which is what I'm after. However, I can't place individual .pngs into a spreadsheet this way.

Hmm.