How to create my statistical summary table manually

January 13, 2023

mydata<-structure(list(Weight = c(66.2, 65.2, 69.8, 63.4, 67.4, 66.3, 
                         63.8, 67.8, 66.7, 66.2, 61.9, 66.9, 69.4, 60.8, 64.1, 62.8, 62.5, 
                         60.9, 61.3, 67.8), Age = c(68, 67, 65, 65, 63, 64, 68, 65, 65, 
                                                    71, 64, 65, 68, 61, 65, 62, 60, 66, 62, 58), 
               Sex = c("H", "H", 
                        "H", "H", "H", "H", "F", "F", "F", "F", "H", "H", "H", "F", "F", 
                        "F", "F", "F", "F", "F"),
               Group = c("G1", "G1", "G1", "G1", 
                          "G1", "G1", "G1", "G1", "G1", "G1", "G2", "G2", "G2", "G2", "G2",
                          "G2", "G2", "G2", "G2", "G2")), row.names = c(NA, -20L), 
          class = "data.frame")

I want to summarize my data by creating my table manually. My goal is to compare variables between two groups. I don’t know of any software that allows me to have a confidence interval of the difference of the mean and the p-value, in a table format. I have to export my data with Rmarkdown in word format, so I should have it in table format.

I created all the parameters like this:

confInt<-paste(round(t.test(mydata$Weight~mydata$Group)$conf.int[1],2),
               round(t.test(mydata$Weight~mydata$Group)$conf.int[2],2),sep = ";")
p.value<-round(t.test(mydata$Weight~mydata$Group)$p.value,3)

mean1<-mean(mydata$Weight[mydata$Group=="G1"])
mean2<-mean(mydata$Weight[mydata$Group=="G2"])

mean_diff<-(mean(mydata$Weight[mydata$Group=="G1"])-
mean(mydata$Weight[mydata$Group=="G2"]))

The goal is to create these parameters for each of my numeric variables, via a loop or a function.
First for the variable Weight:

Then via a rowbind , bind the statistics of each variable

>Solution :

We can create a function that takes in data mydata, numeric column col and grouping column group:

summary_val <- function(mydata,col,group){
  x <- mydata[[col]]
  group_data <- mydata[[group]]
  
  confInt<-paste(round(t.test(x~group_data)$conf.int[1],2),
                 round(t.test(x~group_data)$conf.int[2],2),sep = ";")
  p.value<-round(t.test(x~group_data)$p.value,3)
  
  mean1<-mean(x[group_data=="G1"])
  mean2<-mean(x[group_data=="G2"])
  
  mean_diff<-(mean(x[group_data=="G1"])-
                mean(x[group_data=="G2"]))
  diff <- paste0(mean_diff,"[",confInt,"]")
  return(data.frame(var=col,G1=mean1,G2=mean2,`Diff.CI.`=diff,`p.value`=p.value))
}

summary_val(mydata,"Weight","Group")

     var    G1    G2         Diff.CI. p.value
1 Weight 66.28 63.84 2.44[-0.01;4.89]   0.051

Then we can use the following to extract the names of numeric columns:

num_var <- names(mydata)[unlist(lapply(mydata, is.numeric))]
num_var
[1] "Weight" "Age"

And get the summary output via for loop:

mysummary <- data.frame()
for(var in num_var){
  mysummary <- rbind(mysummary,summary_val(mydata,var,"Group"))
}
mysummary
     var    G1    G2                    Diff.CI. p.value
1 Weight 66.28 63.84            2.44[-0.01;4.89]   0.051
2    Age 66.10 63.10 2.99999999999999[0.43;5.57]   0.025

or using do.call+lapply

summary_val2 <- function(col,mydata,group){
  summary_val(mydata,col,group)
}

do.call(rbind,lapply(num_var,summary_val2,mydata,"Group"))
     var    G1    G2                    Diff.CI. p.value
1 Weight 66.28 63.84            2.44[-0.01;4.89]   0.051
2    Age 66.10 63.10 2.99999999999999[0.43;5.57]   0.025