Hello Stack Community,
My question is in regards to the command
error.bars.by()
from the package error.bars.by{psych}.
I am familiar with how the command works:
error.bars.by(
mydata[3],
mydata[1],
by.var=TRUE,
x.cat=TRUE,
legend=5,
eyes=FALSE,
pch=10,
cex=2,
alpha=.05,
density=10,
)
My question is whether or not this command can handle two grouping variables.
If so, how would I code it?
The data table
Group Subgroup Measure
1 A 1 0.234213
2 A 1 0.046248
3 A 1 0.391376
4 A 2 0.911849
5 A 2 0.729955
6 A 2 0.991110
7 A 2 0.378422
8 A 3 0.898037
9 A 3 0.258884
10 A 3 NA
11 A 3 0.057631
12 A 3 0.745202
13 A 3 0.121376
14 B 1 0.385198
15 B 1 0.484399
16 B 1 0.115034
17 B 1 0.073629
18 B 1 0.456150
19 B 2 0.336108
20 B 2 0.845458
21 B 2 0.267494
22 B 3 0.536123
23 B 3 1.331731
24 B 3 0.505114
25 B 3 0.843348
26 B 3 0.827932
27 B 3 0.813351
28 C 1 0.095587
29 C 1 0.158822
30 C 1 0.392376
31 C 1 0.284625
32 C 2 0.898819
33 C 2 0.743428
34 C 2 0.298989
35 C 2 0.423961
36 C 3 0.868351
37 C 3 0.181547
38 C 3 1.146131
39 C 3 0.234941
Try <something like>(TM):
error.bars.by( mydata[[3]], interaction(mydata[[1]], mydata[[other_col_name]]), by.var=TRUE,
x.cat=TRUE, legend=5, eyes=FALSE, pch=10, cex=2, alpha=.05, density=10, )
The interaction function creates a new single column that can be used when you want a "long" version of a cross-classification.
Edit: I took out the by.var=TRUE since it wasn't documented on the help page for the function and you didn't describe what it was doing.