Why "group=1" is not linking discrete variables together using geom_line?

Question

To link discrete variables together, we use group=1 as recommended here ggplot: line plot for discrete x-axis and here Using `geom_line()` with X axis being factors. However, this is not working in my case. Can someone help with that please ?

My starting code is the following:

ggplot(df, aes(x = cohort, y = est,color=name)) +
  geom_point(position=position_dodge(width=0.3)) + 
  geom_errorbar(aes(ymin = lower, ymax = upper), width = 0,position=position_dodge(width=0.3))

I get the following plot:

Now the aim is to link the red dots using a line, and the same applies for the blue dots.

If I use the following code:

ggplot(df, aes(x = cohort, y = est,color=name,group=1)) +
  geom_point(position=position_dodge(width=0.3)) + 
  geom_errorbar(aes(ymin = lower, ymax = upper), width = 0,position=position_dodge(width=0.3)) + 
  geom_line(position=position_dodge(width=0.3))

I get this:

Here is my data:

df=structure(list(est = c(-0.0584741125344448, 0.021444947354496, 
0.0283447339408831, 0.040105980055471, 0.0374816335738256, -0.00876274041718853
), se = c(0.0139423923119975, 0.0116604369679277, 0.012226641006313, 
0.0224145123205577, 0.0207370700447159, 0.0214173453687987), 
    cohort = structure(c(1L, 2L, 3L, 1L, 2L, 3L), .Label = c("S", 
    "B", "G"), class = "factor"), upper = c(-0.0332526819060428, 
    0.0416835581439528, 0.0495098759734735, 0.0840384242037641, 
    0.0781262908614688, 0.0332152565056569), lower = c(-0.0836955431628468, 
    0.00120633656503907, 0.00717959190829267, -0.0038264640928221, 
    -0.0031630237138175, -0.0507407373400339), name = structure(c(2L, 
    2L, 2L, 1L, 1L, 1L), .Label = c("Low-skilled working class", 
    "Upper-middle class"), class = "factor")), row.names = c("Silent generation...1", 
"Baby boomers...2", "Generation X...3", "Silent generation...4", 
"Baby boomers...5", "Generation X...6"), class = "data.frame")

In your case you have to group by `name`, i.e. do `group=name`. group = 1, well, treats all obs as one group. — stefan, Mar 17 '22 at 14:40

score 1 · Accepted Answer · answered Mar 17 '22 at 15:24

You could set group in your line to the variable name. I set an alpha for the error bars to make the plot more clear. You can use this code:

 ggplot(df, aes(x = cohort, y = est,color=name)) +
  geom_point(position=position_dodge(width=0.0)) + 
  geom_errorbar(aes(ymin = lower, ymax = upper), width = 0,position=position_dodge(width=0.0), alpha =0.5) +
  geom_line(aes(x = cohort, y = est, group = name))

Output:

Why "group=1" is not linking discrete variables together using geom_line?

1 Answers1