CODEWITHSUNDEEP
rggplot2
AnnK
AnnK

Reputation: 189

ggplot: show graph of response variable paired by specific variable

This is a simple question, but I am trying to make a graph in ggplot and I am having a hard time trying to show the data in pairs according to the variable 'time'. My data looks like this:

id  season       sex    treatment   time    shift(Km)
1   Early dry   PRIDES  UNDISTURBED BEFORE  0.016
2   Early dry   PRIDES  UNDISTURBED BEFORE  0.016
3   Early dry   PRIDES  UNDISTURBED BEFORE  0.016
4   Wet         PRIDES  DISTURBED   AFTER   0.0972

I would like the final graph to have the exact layout as the graph below, but what I have not figure out yet is how to plot my response variable in pairs according to the variable 'time', that is, to show the before and after values paired for each 'treatment' group (disturbed/undisturbed), not on top of each other as it is showing in the graph below.

enter image description here

This is the code I have:

ex <- read.csv(file = "https://dl.dropboxusercontent.com/u/23723553/data.csv",
               header= TRUE, row.names =1)

ex$time <- relevel(ex$time, ref = "BEFORE")

ggplot(data = ex, mapping = aes(x = treatment,
                                        y = dist_shift.Km., 
                                        shape= time)) +
  stat_summary(fun.data = mean_se) +
  facet_grid(season ~ sex)+
  labs(x= "\nTreatment", y = "Shift of centroids (m)\n")+
  theme_bw()

Could someone help me modify my code?

UPDATE: plotting the time/treatment interaction as suggested by @lukeA gets me closer to what I am aiming for, but graphing the interaction plots side by side 'disturbed before' with 'undisturbed before' and then 'disturbed after' with 'undisturbed after'. I would like to plot instead side by side: 'disturbed before' with 'disturbed after', and 'undisturbed before' with 'undisturbed after'.

enter image description here

And, is there a way to change the names in the x axis so I can have only 'disturbed' 'undisturbed' instead of 'disturbed.before' etc. The time before/after is already explained by the shape and described in the legend.

Many thanks!

Upvotes: 2

Views: 385

Answers (2)

Didzis Elferts
Didzis Elferts

Reputation: 98489

You just have to add argument position=position_dodge() to stat_summary() to place before and after values close to each other.

ggplot(data = ex, mapping = aes(x = treatment,
                                y = dist_shift.Km., 
                                shape= time)) +
      stat_summary(fun.data = mean_se,position = position_dodge(width=0.2)) +
      facet_grid(season ~ sex)+
      labs(x= "\nTreatment", y = "Shift of centroids (m)\n")+
      theme_bw()

enter image description here

Upvotes: 1

lukeA
lukeA

Reputation: 54237

You could for example do

ggplot(data = ex, mapping = aes(x = interaction(time, treatment),
                                        y = dist_shift.Km., 
                                        shape= time)) +
  stat_summary(fun.data = mean_se) +
  facet_grid(season ~ sex)+
  labs(x= "\nTreatment", y = "Shift of centroids (m)\n")+
  scale_x_discrete(labels = function(x) sub(".*\\.", "", x)) +
  theme_bw() + 
  theme(axis.text.x=element_text(angle=90, vjust=0.5, hjust=1))

enter image description here

Upvotes: 1

Related Questions