two-way bar graph adjustments

[Tracy Lam]

For the graph below, I would like to move the 1 on the y-axis down to where 0 is. How do I do that? Or should I switch to two way bar?

This is what my data look like:

var1 var2 var3 var4 var5
1 5.25 6.14 5.62 4
2 5.31 6.18 5.74 4.02
3 5.52 6.3 5.9 4.36
4 5.29 6.21 5.69 3.97
5 5.03 6.1 5.44 3.54
6 5.13 6.12 5.46 3.82


I have this code: graph bar var3 var4 var5, over(var1)

That generated this graph:

[William Lisowski]

When I run this code (note the use of dataex to present sample data)

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte var1 float(var2 var3 var4 var5)
1 5.25 6.14 5.62    4
2 5.31 6.18 5.74 4.02
3 5.52  6.3  5.9 4.36
4 5.29 6.21 5.69 3.97
5 5.03  6.1 5.44 3.54
6 5.13 6.12 5.46 3.82
end

graph bar var3 var4 var5, over(var1)

I obtain the following graph. I do not understand what it is you want to change, and unfortunately the graph in post #1 does not display for me.

[Tracy Lam]

Thanks, William.

I want to get rid of the 0 and make it start at 1 on the y-axis.

[William Lisowski]

Is this what you have in mind?

graph bar var3 var4 var5, over(var1) ylabel(1(1)6)

[Tracy Lam]

Yes, but how do I move the 1 on the y-axis down to where 0 would be?

[Andrew Musau]

graph bar var3 var4 var5, over(var1) ylab(1 (1) 6) exclude0

[Nick Cox]

The idea here will just evoke flak and/or puzzlement in most readers, more flak from experienced readers and more puzzlement from the others. If zero has meaning as a base or reference level for your variables, then bars should start at zero. If not, or you think that starting at zero is a waste of space, then use graph dot instead.

Bar heights (or lengths) encode distance or deviation from a reference level, usually but not necessarily zero. If you don't want that principle, use another principle, encoding value by position on a numeric scale.