Graph with a string variable

[annettepelkmans]

I have a string variable that I need to keep as the x variable. I tried transforming it to numeric, but the scale is now all messed up. I really need to keep string variable because it represents product codes. Thanks for the help.

[Nick Cox]

At a guess you need to map to a (well-behaved) numeric variable, and here is one way:

 
egen numvar = group(strvar), label

Illustration code:

 
clear
input str1 strvar 
"A"
"B"
"D" 
"G" 
"K"
end 
egen numvar = group(strvar), label 
label var numvar "my preferred variable label" 
d
l

[annettepelkmans]

Thanks Nick! My problem is that when I convert my product codes (which are string variables) into numeric, stata inputs the numeric code as chronological numbers, and this is why the scale of my X axis no longer reflects the scale of my original string variable.

pssc7 prodcode
0014101 00
0176002 01
0230001 02
.
.
9710300 97


When I graph the frequencies, however, my scale ends at 65 instead of 97.

[Nick Cox]

I don't see that you tried my solution. In any case you haven't showed us a reproducible problem. No code, no graph, ...

See e.g. http://www.statalist.org/forums/help Section 12 for how to ask a better question.

My wild guess is that you used a numeric variable but forgot to spell out that you want to see value labels. But why should we have to guess?

[Konrad Zdeb]

Thanks Nick! My problem is that when I convert my product codes (which are string variables) into numeric, stata inputs the numeric code as chronological numbers, and this is why the scale of my X axis no longer reflects the scale of my original string variable.

pssc7 prodcode
0014101 00
0176002 01
0230001 02
.
.
9710300 97


When I graph the frequencies, however, my scale ends at 65 instead of 97.

Hi, I would try:

encode prodcode, generate(prodcode_num)

If the syntax:

pssc7 prodcode
0014101 00
0176002 01
0230001 02

corresponds to the extract from the input command you should indicate that variable is a string:

input str2 prodcode

[annettepelkmans]

Thanks Konrad, this works!