I'm trying to use pivot_longer with the repeated measures data. I'm struggling because they have a shared variable.
Basically, my data includes target stimuli and their corresponding responses, also an intervention condition with pre-post levels. There are 4 different targets but they were presented in mixed order. So the pre-post order of the targets is not identical.
SubjNo Group Target_Pre Response_Pre Target_Post Response_Post
122 Gr1 3 2,5 4 3,3
122 Gr1 2 1,8 1 2
122 Gr1 1 1,7 3 1,9
122 Gr1 4 3,5 2 2,3
123 Gr2 4 2,6 1 2
123 Gr2 1 1,4 3 1,5
123 Gr2 3 2,6 2 2,3
123 Gr2 2 1,5 4 3
The outcome I want is;
SubjNo Group Target Response Condition
122 Gr1 3 2,5 Pre
122 Gr1 2 1,8 Pre
122 Gr1 1 1,7 Pre
122 Gr1 4 3,5 Pre
123 Gr2 4 2,6 Pre
123 Gr2 1 1,4 Pre
123 Gr2 3 2,6 Pre
123 Gr2 2 1,5 Pre
122 Gr1 4 3,3 Post
122 Gr1 1 2 Post
122 Gr1 3 1,9 Post
122 Gr1 2 2,3 Post
123 Gr2 1 2 Post
123 Gr2 3 1,5 Post
123 Gr2 2 2,3 Post
123 Gr2 4 3 Post
So far, I can only manage to separate targets OR responses. For example, the code I've tried for the targets;
longdf <- pivot_longer( df,
cols = c("Target_Pre", "Target_Post"),
names_to = c("Condition"),
names_pattern = "Target_(.*)" , values_to = "Targets")
When I try the same code for responses, then I have two columns for conditions in a different order.
Is there a way to use pivot_longer for both targets and responses, but keeping the conditions in one column?
CodePudding user response：
We could use the names_sep argument of pivot_longer from tidyr package:
Or if you prefer names_pattern argument: Then the regex could be this way: names_pattern = '(.*)_(.*)'
library(dplyr)
library(tidyr)
df %>%
pivot_longer(
-c(SubjNo, Group),
names_to =c(".value","Condition"),
names_sep ="_"
)
SubjNo Group Condition Target Response
<int> <chr> <chr> <int> <chr>
1 122 Gr1 Pre 3 2,5
2 122 Gr1 Post 4 3,3
3 122 Gr1 Pre 2 1,8
4 122 Gr1 Post 1 2
5 122 Gr1 Pre 1 1,7
6 122 Gr1 Post 3 1,9
7 122 Gr1 Pre 4 3,5
8 122 Gr1 Post 2 2,3
9 123 Gr2 Pre 4 2,6
10 123 Gr2 Post 1 2
11 123 Gr2 Pre 1 1,4
12 123 Gr2 Post 3 1,5
13 123 Gr2 Pre 3 2,6
14 123 Gr2 Post 2 2,3
15 123 Gr2 Pre 2 1,5
16 123 Gr2 Post 4 3
data:
structure(list(SubjNo = c(122L, 122L, 122L, 122L, 123L, 123L,
123L, 123L), Group = c("Gr1", "Gr1", "Gr1", "Gr1", "Gr2", "Gr2",
"Gr2", "Gr2"), Target_Pre = c(3L, 2L, 1L, 4L, 4L, 1L, 3L, 2L),
Response_Pre = c("2,5", "1,8", "1,7", "3,5", "2,6", "1,4",
"2,6", "1,5"), Target_Post = c(4L, 1L, 3L, 2L, 1L, 3L, 2L,
4L), Response_Post = c("3,3", "2", "1,9", "2,3", "2", "1,5",
"2,3", "3")), class = "data.frame", row.names = c(NA, -8L
))
CodePudding user response：
library(dplyr)
library(tidyr)
# Create reproducible dataset
df <- tribble(~SubjNo, ~Group,~Target_Pre,~Response_Pre,~ Target_Post,~ Response_Post,
122, "Gr1", 3, "2,5", 4, "3,3",
122, "Gr1", 2, "1,8", 1, "2",
122, "Gr1", 1, "1,7", 3, "1,9",
122, "Gr1", 4, "3,5", 2, "2,3",
123, "Gr2", 4, "2,6", 1, "2",
123, "Gr2", 1, "1,4", 3, "1,5",
123, "Gr2", 3, "2,6", 2, "2,3",
123, "Gr2", 2, "1,5", 4, "3")
# Pivot longer
df |>
pivot_longer(cols = -c("SubjNo", "Group"),
names_to = c(".value", "Condition"),
names_sep = "_") |>
arrange(desc(Condition))
Output:
#> # A tibble: 16 x 5
#> SubjNo Group Condition Target Response
#> <dbl> <chr> <chr> <dbl> <chr>
#> 1 122 Gr1 Pre 3 2,5
#> 2 122 Gr1 Pre 2 1,8
#> 3 122 Gr1 Pre 1 1,7
#> 4 122 Gr1 Pre 4 3,5
#> 5 123 Gr2 Pre 4 2,6
#> 6 123 Gr2 Pre 1 1,4
#> 7 123 Gr2 Pre 3 2,6
#> 8 123 Gr2 Pre 2 1,5
#> 9 122 Gr1 Post 4 3,3
#> 10 122 Gr1 Post 1 2
#> 11 122 Gr1 Post 3 1,9
#> 12 122 Gr1 Post 2 2,3
#> 13 123 Gr2 Post 1 2
#> 14 123 Gr2 Post 3 1,5
#> 15 123 Gr2 Post 2 2,3
#> 16 123 Gr2 Post 4 3