Index plots

df %>%
  dplyr::filter(!is.na(performance)) %>%
  ggplot + 
  geom_bar(aes(x = performance))

df %>%
  dplyr::filter(!is.na(supervision)) %>%
  ggplot + 
  geom_bar(aes(x = supervision))

To make it even clearer, we can take a look at the data in a table:

table(df$performance)
#> 
#>    CD    UP    IP    PR    EP 
#>    38  4346 14986  8554   817

table(df$supervision)
#> 
#>    ST    AH    PH    SO 
#>  3166 13914 10074  4753

df %>%
  dplyr::group_by(supervision, traineePGY) %>%
  dplyr::filter(!is.na(supervision)) %>%
  dplyr::summarise(n = dplyr::n()) %>%
  dplyr::mutate(rate = n/sum(n)) %>%
  ggplot() +
    geom_bar(aes(x = supervision, y = rate, fill = traineePGY), stat = "identity", position = "dodge")
#> `summarise()` regrouping output by 'supervision' (override with `.groups` argument)

df %>%
  dplyr::group_by(performance, traineePGY) %>%
  dplyr::filter(!is.na(performance)) %>%
  dplyr::summarise(n = dplyr::n()) %>%
  dplyr::mutate(rate = n/sum(n)) %>%
  ggplot() +
    geom_bar(aes(x = performance, y = rate, fill = traineePGY), stat = "identity", position = "dodge")
#> `summarise()` regrouping output by 'performance' (override with `.groups` argument)

Number of complete paired evaluations

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired) %>%
  dplyr::group_by(subjectID) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  ggplot() +
  geom_histogram(aes(x = count), bins = 75)
#> `summarise()` ungrouping output (override with `.groups` argument)

Plotting number of completed evaluations vs. performance as rates

getquant <- function(x) {
  ifelse(x < 5, 1,
  ifelse(x < 10, 2,
  ifelse(x < 20, 3, 4)))
}

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired,
                !is.na(performance)) %>%
  dplyr::group_by(subjectID, performance) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  dplyr::mutate(totalcount = sum(count),
                rate = count/sum(count),
                quant = getquant(totalcount)) %>%
  ggplot(aes(x = quant, y = rate, group = quant)) +
  geom_boxplot() + 
  facet_wrap(~performance)
#> `summarise()` regrouping output by 'subjectID' (override with `.groups` argument)

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired,
                !is.na(supervision)) %>%
  dplyr::group_by(subjectID, supervision) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  dplyr::mutate(totalcount = sum(count),
                rate = count/sum(count),
                quant = getquant(totalcount)) %>%
  ggplot(aes(x = quant, y = rate, group = quant)) +
  geom_boxplot() + 
  facet_wrap(~supervision)
#> `summarise()` regrouping output by 'subjectID' (override with `.groups` argument)

Plotting number of completed evaluations vs. performance as ratios

getquant <- function(x) {
  ifelse(x < 5, 1,
  ifelse(x < 10, 2,
  ifelse(x < 20, 3, 4)))
}

getRatio <- function(count, performance) {
  if(length(performance) == 2) {
    count[performance == "Independent"]/count[performance =="Not independent"]
  } else 0
}

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired,
                !is.na(performance)) %>%
  dplyr::mutate(performance = 
                  forcats::fct_collapse(performance, Independent = c("PR","EP"), `Not independent` = c("CD", "UP", "IP"))) %>%
  dplyr::group_by(subjectID, performance) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  dplyr::mutate(ratio = getRatio(count, performance),
                totalcount = sum(count),
                quant = factor(getquant(totalcount), levels = c(1,2,3,4), ordered = TRUE)) %>%
  dplyr::filter(ratio != 0) %>%
  ggplot(aes(x = quant, y = ratio)) +
  geom_boxplot()
#> `summarise()` regrouping output by 'subjectID' (override with `.groups` argument)

getquant <- function(x) {
  ifelse(x < 30, 1, 2)
}

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired,
                !is.na(performance)) %>%
  dplyr::mutate(performance = 
                  forcats::fct_collapse(performance, Independent = c("PR","EP"), `Not independent` = c("CD", "UP", "IP"))) %>%
  dplyr::group_by(subjectID, performance) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  dplyr::mutate(ratio = getRatio(count, performance),
                totalcount = sum(count),
                quant = factor(getquant(totalcount), levels = c(1,2), ordered = TRUE)) %>%
  dplyr::filter(ratio != 0) %>%
  ggplot(aes(x = quant, y = ratio)) +
  geom_boxplot()
#> `summarise()` regrouping output by 'subjectID' (override with `.groups` argument)

Plotting number of completed evaluations vs. autonomy as ratios

getquant <- function(x) {
  ifelse(x < 5, 1,
  ifelse(x < 10, 2,
  ifelse(x < 20, 3, 4)))
}

getRatio <- function(count, supervision) {
  if(length(supervision) == 2) {
    count[supervision == "Independent"]/count[supervision =="Not independent"]
  } else 0
}

df %>%
  dplyr::filter(traineeType == "CATEGORICAL",
                subjectRole == "Trainee",
                raterRole == "Attending",
                paired,
                !is.na(supervision)) %>%
  dplyr::mutate(supervision = 
                  forcats::fct_collapse(supervision, Independent = c("SO","PH"), `Not independent` = c("ST", "AH"))) %>%
  dplyr::group_by(subjectID, supervision) %>%
  dplyr::summarise(count = dplyr::n()) %>%
  dplyr::mutate(ratio = getRatio(count, supervision),
                totalcount = sum(count),
                quant = factor(getquant(totalcount), levels = c(1,2,3,4), ordered = TRUE)) %>%
  dplyr::filter(ratio != 0) %>%
  ggplot(aes(x = quant, y = ratio)) +
  geom_boxplot()
#> `summarise()` regrouping output by 'subjectID' (override with `.groups` argument)