The “pool” Function in R

• Package: mice

• Purpose: Combine results from multiple imputed datasets.

• General class: Imputation

• Required argument(s):

• object: An object of class mira (multiply imputed repeated analyses) or a list with mipo (multiple imputation pooled outcomes) objects.

• Notable optional arguments:

• dfcom: Degrees of freedom for the complete data.

• rule: “rubin1987” is the default used for missing data, you can also choose “reiter2003” for synthetic missing data.

• Example:

• # Load the required library
  library(mice)
  
  # Load the nhanes dataset
  data(nhanes)
  
  # Perform multiple imputation on the nhanes dataset
  imputed_data <- mice(nhanes, m = 5, maxit = 50, meth = 'pmm', seed = 500)
  
  # Fit a linear model on each of the imputed datasets
  fit <- with(imputed_data, lm(bmi ~ hyp + chl))
  
  # Pool the results of the linear models
  pooled_results <- pool(fit)
  
  # Print the summary of the pooled results
  summary(pooled_results)

• In this example, the mice package is used to handle missing data through multiple imputation. First, the mice function creates multiple imputed datasets. Next, the with function applies a linear model to each of these datasets. Finally, the pool function combines the results from these multiple models to provide a single set of estimates and inferential statistics. The summary function is then used to display the pooled results.