I’m on a mission to warn the innocent of the dangers of mean imputation. Perhaps that’s a bit dramatic, but mean imputation (also called mean substitution) really ought to be a last resort in most situations. There are many alternatives to mean imputation that provide much more accurate parameter estimates, so there really is no excuse to use it.
This post is the first in a series explaining the many reasons not to use mean imputation (and to be fair, its advantages).
First, a definition: mean imputation is the replacement of a missing observation with the mean of the non-missing observations for that variable.
Problem #1: Mean imputation does not preserve the relationships among variables.
True, imputing the mean preserves the mean of the observed data. So if the data are missing at random, the estimate of the mean remains unbiased. This is the original logic involved in mean imputation.
If all you are doing is estimating means (which is rarely the point of research studies), mean imputation will not bias your parameter estimate. (It will bias your standard error, but I will get to that in another post).
Since most research studies are interested in the relationship among variables, mean imputation is not a good solution. The following graph illustrates this well:
This graph illustrates hypothetical data between X=years of education and Y=annual income in thousands with n=50. The blue circles are the original data and the solid blue line indicates the best fit regression line for the full data set. The correlation between X and Y is r=.53.
I then randomly deleted 12 observations of income (Y) and substituted the mean. The dataset with the mean imputations is shown by the red dots. Blue circles with red dots inside them represent non-missing data. Blue circles that are empty represent the missing data. If you look across the graph at Y=39, you will see a row of red dots without blue circles. These represent the imputed values.
The dotted red line represents the new best fit regression line with the imputed data. As you can see, it is less steep than the original line, and the new correlation is now r=.39. It is still significant, but it is clearly biased. The real relationship is quite underestimated.
One note: if X were missing instead of Y, mean substitution would actually artificially inflate the correlation.