linear regression using dataset with missing values

Question

I have data on the effect sizes for 14 variables (var1-var14). Each value is the effect size of a specific treatment on a certain variable. Missing values are due to that some articles did not consider certain variables. A positive value show promoting while a negative value shows the inhibiting effect of that treatment on the variable. I want (1) to do a pairwise linear regression that runs through each and every variable and compare if there is an association between variables, (2) consider var1 as the dependent variable and var2-var14 all as independent variables to find the best-fit model (maybe using glmulti package?) and show changes in which variables are most important for change in var1.

Here is a sample data:

set.seed(123)

**# Create the dataset with effect sizes and missing values**

mydata <- data.frame(
  Var1 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var2 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var3 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var4 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var5 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var6 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var7 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var8 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var9 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var10 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var11 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var12 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var13 = sample(c(-20:14, NA), 64, replace = TRUE),
  Var14 = sample(c(-20:14, NA), 64, replace = TRUE)
)

**# Set more than 50% missing values in each column**
for (col in 1:14) {
  missing_indices <- sample(1:64, size = 32)
  mydata[missing_indices, col] <- NA
}

Is it possible to do all this with such dataset (i.e., missing values)? Thanks!

Answer 1

d being your example data:

d <- 
  paste0('Var_', 1:14) |>
  Map(f = \(.) sample(c(-20:14, NA),
                      size = 64,
                      prob = c(rep(.49/35, 35), .51),
                      replace = TRUE
                      )
      ) |>
  as.data.frame()

... you get the pairwise associations in terms of the correlation matrix like so:

d |> cor(use = 'pairwise.complete.obs')

... and a basic column-wise imputation (replacing NA with the mean value) this way:

d_imputed <- d |>
  apply(2, \(var) replace(var, is.na(var), mean(var, na.rm = TRUE)))

Finally you can obtain the regression coefficients of the predictors (columns) for each column like so:

d_imputed |> 
  apply(2, FUN = \(var) coef(lm(var ~ ., as.data.frame(d_imputed))))

A word of caution: above is just a technical answer to your literal question. For a statistically sound solution, I'd recommend researching over at Cross Validated about imputation, dimensionality reduction, predictor selection and such (see Ben Bolker's comment).