I am trying to visualize 3 categorical variables: a binary categorical variable between the treatment and control groups and before&treatmet. The main variable measures whether someone's views are sociotropic or egocentric, and below is my attempt in doing so using bar graphs, but I am open to other graphs to visualize the same variables.

data example:

print data example

dput(sample_n(socio_egotropic_graph, size = 5))

structure(list(date = structure(c(1346112000, 1335139200, 1318118400, 1349913600, 
1339891200), tzone = "UTC", class = c("POSIXct", "POSIXt")), 
sentiment_human_coded = c("negative", 
    "negative", "neutral", "negative", "negative"), economic_demand_complaint = c(1, 
    0, 0, 0, 0), collective_action = c(0, 0, 1, 0, 0), directed_to_whom = c("Private employer", 
    "N/A", NA, "N/A", "Private employer"), socio_egotropic = c("sociotropic", 
    "egocentric", "sociotropic", "egocentric", "egocentric"), 
    gender = c("female", "female", NA, "male", "female"), treatment_announcement = c("post", 
    "post", NA, "post", "post"), treatment_details = c("post", 
    "post", "pre", "post", "post"), treatment_implementation = c("pre", 
    "pre", "pre", "post", "pre"), month_year = structure(c(2012.58333333333, 
    2012.25, 2011.75, 2012.75, 2012.41666666667), class = "yearmon"), 
    group = c("treatment", "treatment", "control", "treatment", 
    "treatment")), row.names = c(NA, -5L), class = c("tbl_df", 
"tbl", "data.frame"))

graph code:

socio_egotropic_graph |> 
  drop_na() |> 
    filter(socio_egotropic != "N/A")  |>  
  select(socio_egotropic, treatment_details, group) |>     # we're only interested in socio_egotropic 
    group_by(socio_egotropic) %>%   # group data and
    add_count(treatment_details) |>       # add count of treatment_details
    unique() |>           # remove duplicates
    ungroup() |>            # remove grouping
    group_by(treatment_details) |>        # group by treatment_details 
    mutate(socio_egotropic_percentage = n/sum(n)) |>  # ...calculating percentage
    mutate(socio_egotropic = as.factor(socio_egotropic)) |>        # change to factors so that ggplot treats...
    mutate(am = as.factor(treatment_details))  |>   
  ggplot(aes(x = treatment_details, fill = socio_egotropic, y = socio_egotropic_percentage))  
    geom_bar(stat = "identity", position=position_dodge())   
    #scale_fill_grey()  
    xlab("Socio vs. egocentric emotions...")   
    ylab("Socio egocentric share")   
    theme(text=element_text(size=10))   
        scale_y_continuous(labels = percent_format(accuracy = 1))  
    theme(plot.title = element_text(size = 10, face = "bold"))  
    scale_x_discrete(limits = c("pre", "post"))  
  theme_bw()

Here is the output, while the code works, I am unable to show variation depending on treatment status, which is measured using the "group" variable.

CodePudding user response：

One approach:

Since the dput output failed on my side, first construct an example dataset:

n = 40 ## sample data size

get_dummy <- function(choices, n = 40) sample(choices, n, TRUE)

set.seed(4711)
df <- data.frame(month_year = as.Date('2022-03-15')   sample(-2:2, n, TRUE)  
                   30 * sample(0:1, n, TRUE),
                 socio_egotropic = get_dummy(c('sociotropic', 'egocentric')),
                 treatment_details = get_dummy(c('pre', 'post')),
                 group = get_dummy(c('treatment', 'control'))
                 )

summarize data as frequency table (retain 'egocentric' trait only):

  df_stats_egocentric <-  df |> 
    ## uncomment, if control values cannot be assigned to pre/post:
    ## mutate(treatment_details = ifelse(group == 'control',
    ##                                  'post', treatment_details)
    ##       ) |>
    count(socio_egotropic, treatment_details, group) |>
    group_by(group, treatment_details) |>
    mutate(prop = prop.table(n)) |>
    filter(socio_egotropic == 'egocentric')

bullet plot, if control observations can also be assigned to pre- and post-treatment periods:

df_stats_egocentric |>
    ggplot()  
    geom_col(data = . %>% filter(group == 'treatment'),
             aes(treatment_details, prop),
             alpha = .5)  
    geom_col(data = . %>% filter(group == 'control'),
             aes(treatment_details, prop),
             width = .01)

If control observations apply to both pre- and post-treatment effects, draw a horizontal reference line:

  df_stats_egocentric |>
    ggplot()  
    geom_col(data = . %>% filter(group == 'treatment'),
             aes(treatment_details, prop),
             alpha = .5)  
    geom_hline(data = . %>% filter(group == 'control'),
               aes(yintercept = prop)
               )