Bar graphs in a long data format-CodePudding

I would like to create a bar graph using data in long format.

Here is my code:

library(data.table)
library(dplyr)
library(ggplot2)

dv1 = runif(n = 100, min = 1, max = 7)
dv2 = runif(n = 100, min = 1, max = 7)
dv3 = runif(n = 100, min = 1, max = 7)
country <- rep(c("India", "US", "Poland"), length.out = 100)

df <- data.frame(country, dv1, dv2, dv3)

df$casenum <- seq.int(nrow(df))

df2 <- df %>% select(casenum, country, dv1, dv2, dv3)

df.melt <- data.table::melt(setDT(df2), id = 1L,
                               measure = list(c(3,4,5)),
                               value.name = c("dv"))
df.melt2 <- df2 %>%
  select(casenum, country)  

df.melt.final <- dplyr::left_join(df.melt, df.melt2, by="casenum")

ggplot(df.melt.final, aes(fill=variable, y=dv, x=country))   
  geom_bar(position="dodge", stat="identity")

The bar graph looks like this, but the means on the graph do not correspond with the actual means in the data. What can I do with it?

CodePudding user response：

You'll want to calculate the mean before plotting. E.g.

library(ggplot2)
library(dplyr)

df.melt.final |>
  group_by(country, variable) |>
  summarise(dv = mean(dv)) |>
  ggplot(aes(fill=variable, y=dv, x=country))   
    geom_bar(position="dodge", stat="identity")

Output:

Check means:

# A tibble: 9 × 3
  country variable    dv
  <chr>   <fct>    <dbl>
1 India   dv1       3.97
2 India   dv2       3.97
3 India   dv3       3.91
4 Poland  dv1       3.97
5 Poland  dv2       3.73
6 Poland  dv3       4.18
7 US      dv1       4.16
8 US      dv2       4.16
9 US      dv3       4.02

CodePudding user response：

Another option is saying stat = "summary" and fun = "mean" in your geom_bar like this:

library(data.table)
library(dplyr)
library(ggplot2)
set.seed(7)
dv1 = runif(n = 100, min = 1, max = 7)
dv2 = runif(n = 100, min = 1, max = 7)
dv3 = runif(n = 100, min = 1, max = 7)
country <- rep(c("India", "US", "Poland"), length.out = 100)

df <- data.frame(country, dv1, dv2, dv3)

df$casenum <- seq.int(nrow(df))

df2 <- df %>% select(casenum, country, dv1, dv2, dv3)

df.melt <- data.table::melt(setDT(df2), id = 1L,
                            measure = list(c(3,4,5)),
                            value.name = c("dv"))
df.melt2 <- df2 %>%
  select(casenum, country)  

df.melt.final <- dplyr::left_join(df.melt, df.melt2, by="casenum")

ggplot(df.melt.final, aes(fill=variable, y=dv, x=country))   
  geom_bar(position="dodge", stat = "summary", fun = "mean")

# This are the means to show
df.melt.final %>%
  group_by(country, variable) %>%
  summarise(dv = mean(dv))
#> `summarise()` has grouped output by 'country'. You can override using the
#> `.groups` argument.
#> # A tibble: 9 × 3
#> # Groups:   country [3]
#>   country variable    dv
#>   <chr>   <fct>    <dbl>
#> 1 India   dv1       4.18
#> 2 India   dv2       3.97
#> 3 India   dv3       4.34
#> 4 Poland  dv1       4.14
#> 5 Poland  dv2       4.25
#> 6 Poland  dv3       4.28
#> 7 US      dv1       3.84
#> 8 US      dv2       4.66
#> 9 US      dv3       3.66

^{Created on 2022-08-26 with reprex v2.0.2}