rdplyr

How to get t.test and dplyr group_by/summarise to be recognized within a function in R?

I created a function so I can perform multiple statistical procedures for several values in a dataset: 1) paired t-test, 2) create a simple boxplot, 3) get summary statistics of patient measures before and after intervention.

Here is my code so far:

stat_func<-function(df, var){
  
  t.test(data=df, var ~ timepoint, paired=TRUE)
  
  
  boxplot(data=df,
          var ~ timepoint, 
          col = c("pink", "#EFC000FF"),
          xlab="",
          ylab="Percent Time Between 70-140")
  
  df%>%
    dplyr::group_by(timepoint)%>%
    dplyr::summarise(min=round(min(var), 2), mean=round(mean(var), 2), max=round(max(var), 2), sd=round(sd(var), 2))
  
  
}

When I run this code, RStudio doesn't show the output for the t-test, it DOES create the boxplot, and the output for dplyr::summarise is identical. It doesn't recognize the "before" versus "after" for the timepoint variable.

Here is example data for one of my variables of interest:

have<-as.data.frame(structure(list(subjectid=structure(c(1, 1, 2, 2, 3, 3, 4,4)), 
                    timepoint=structure(c("before", "after", "before", "after", "before", "after", "before", "after")),
                    estimated_a1c=structure(c(10, 7.5, 10.5, 9.4, 9.8, 7, 9.9, 7.3)))))

When I run my function and then write stat_func(have, have$estimated_a1c) to get the output, in the RStudio console, I don't see any t-test output, it generates the boxplot, and then my summarise results are this but they shouldn't be identical:

  <chr>     <dbl> <dbl> <dbl> <dbl>
1 after         7  8.93  10.5  1.41
2 before        7  8.93  10.5  1.41

Any suggestions for how to get t-test & correct group-by/summarise output would be much appreciated.

Thank you!

Solution

  • Here's a shot at what I think you're getting at.

    library(dplyr)

df <- as.data.frame(structure(
  list(
    subjectid = structure(c(1, 1, 2, 2, 3, 3, 4, 4)),
    timepoint = structure(
      c(
        "before",
        "after",
        "before",
        "after",
        "before",
        "after",
        "before",
        "after"
      )
    ),
    estimated_a1c = structure(c(10, 7.5, 10.5, 9.4, 9.8, 7, 9.9, 7.3))
  )
))

# Define function
stat_func <- function(df, var, time) {

  # print boxplot to console
  boxplot(
    formula = as.formula(paste(var, "~", time)),
    data = df,
    col = c("pink", "#EFC000FF"),
    xlab = "",
    ylab = "Percent Time Between 70-140"
  )
  
  # results list for t.test and summary df
  results <- list()
  
  # save t test to list
  results$t.test <- t.test(
    df[[var]][df[[time]] == "before"],
    df[[var]][df[[time]] == "after"],
    paired = TRUE
  )
  
  # save summary df to list
  results$summary <- df %>%
    dplyr::group_by(!!sym(time)) %>%
    dplyr::summarise(
      min = round(min(!!sym(var)), 2),
      mean = round(mean(!!sym(var)), 2),
      max = round(max(!!sym(var)), 2),
      sd = round(sd(!!sym(var)), 2)
    )
  
  return(results)
}

    stat_func(df = df, var = 'estimated_a1c', time = 'timepoint')

$t.test

    Paired t-test

data:  df[[var]][df[[time]] == "before"] and df[[var]][df[[time]] == "after"]
t = 5.7934, df = 3, p-value = 0.01023
alternative hypothesis: true mean difference is not equal to 0
95 percent confidence interval:
 1.014025 3.485975
sample estimates:
mean difference 
           2.25 


$summary
# A tibble: 2 × 5
  timepoint   min  mean   max    sd
  <chr>     <dbl> <dbl> <dbl> <dbl>
1 after       7     7.8   9.4  1.09
2 before      9.8  10.0  10.5  0.31

    The boxplot should appear in the plots pane, while a list containing the t.test and summary df should print to the console. If you want the boxplot included in the list to save, I have better luck with saving ggplot figures to lists than base boxplots, but I'm sure there's a way.

    A couple of things to note: