Changing the positions in a bar chart

[Lisa Klinkenberg]

Hi everyone,
I would appreciate some help by designing my bar graphs! I have the following datatset

* Example generated by -dataex-. To install: ssc install dataex
clear
input int ID_1 byte Ability_homogen float Performance1_total4 double Performance2_total
201 0 122 244
202 0 135 147
203 1 223 175
204 1 208 192
205 1 131 105
207 0 233 190
208 1  89  90
209 0 180 161
210 0 228 146
211 0 267 357
212 0 153 190
213 1 255 155
214 0 194 240
215 0 228 272
216 0 216 293
217 0 207 204
218 0 223 219
219 1 145 169
220 0 167 132
221 1 130 168
222 0 164 189
223 0 174 228
224 1  96  89
225 0 148 189
end

and want to get a bar graph which displays in 2 seperate areas of the graph Performance1_total for both types of Ability_homogen and in the other area of the graph two bars for Performance2_total with both types of Ability_total
I tried with:
graph bar Performance1_total4 Performance2_total, by(Ability_homogen),
but my goal is not to have two charts, one for Ability_homogen=0 an one for Ability_homogen=1, but for Performance1_total4 and Performance2_total.

Thanks a lot!

[Nick Cox]

See statplot from SSC.

statplot Performance*, over(Ability) recast(bar)

may be closer to what you want.

[Lisa Klinkenberg]

Awesome, thank you so much Nick!!

[Lisa Klinkenberg]

Hey Nick,

maybe you can help me once again? In a different dateset I have 3 different groups (Treatment==0; Treatment==1&Performance_div==0 and Treatment==1&Performance_div==1). How can I graph those in a barchart such that I have the bars grouped over those 3 on the x-axis (instead of 2 groups like before) and the Performance Variable is my y-axis?

Thanks a lot!

[Nick Cox]

Data example and code you've tried so far, please.

[Lisa Klinkenberg]

* Example generated by -dataex-. To install: ssc install dataex
clear
input int ID_1 float Treatment byte Performance1_total4_div float Performance1_total4
  1 0 . 171
  2 0 . 251
  3 0 .  97
  4 0 .  99
  5 0 . 145
  6 0 .  90
  9 0 .  99
 11 0 . 138
 14 0 . 140
 15 0 . 103
 16 0 .  72
 17 0 .  91
 18 0 .  53
 24 0 . 146
 28 0 .  86
 31 0 . 113
 32 0 . 104
 33 0 . 136
 36 0 .  60
 37 0 . 132
 38 0 .  73
 44 0 .  43
 46 0 . 304
 47 0 .  80
 51 0 . 193
 52 0 . 167
 53 0 . 185
 54 0 . 117
 58 0 .  30
 60 0 . 131
 61 0 . 135
 63 0 . 146
 65 0 . 149
 66 0 .  85
 67 0 . 127
 68 0 .  96
 69 0 . 119
 70 0 . 203
 72 0 . 130
 73 0 .  66
 74 0 .  81
 75 0 .  46
 76 0 .  95
 78 0 .  42
 79 0 .  78
 84 0 . 135
 85 0 .  70
 87 0 .  97
 88 0 . 151
 89 0 . 109
 91 0 . 149
 94 0 .  34
 95 0 . 194
 97 0 . 177
 98 0 .  51
 99 0 . 171
101 0 . 137
104 0 . 102
201 1 0 122
202 1 0 135
203 1 1 223
204 1 1 208
205 1 0 131
207 1 1 233
208 1 0  89
209 1 1 180
210 1 1 228
211 1 1 267
212 1 0 153
213 1 1 255
214 1 1 194
215 1 1 228
216 1 1 216
217 1 1 207
218 1 1 223
219 1 0 145
220 1 0 167
221 1 0 130
222 1 0 164
223 1 0 174
224 1 0  96
225 1 0 148
end

The Treatment-group is also now divided in Performance1_total4_div==0 or ==1.
I have tried the following:
statplot Performance1_total4, over(Treatment) over(Performance1_total4_div) recast(bar)
or
twoway bar Performance1_total4 Treatment || bar Performance1_total4 Treatment==1 & Performance1_total4_div==0|| bar Performance1_total4 if Treatment==1 & Performance1_total4_div==1


Thank you!

[Nick Cox]

I guess the major problem here is that the missing values should be recoded, say to 2, and then you have 3 non-missing categories.

[Nick Cox]

See also https://www.statalist.org/forums/for...rent-variables in which the focus is on plotting 3 means and one suggestion is that you have space and scope to be much more interesting and informative.