Critical n in order to reliably replace missing data with series mean?

#1
I am looking for a good source that I can use as argument to replace missing data reliably with the series mean. I have a dataset (n=118) from which three datapoints are missing.

  • 2 datapoints are missing within Variable X and 1 datapoint is missing within Variable Y.
  • 2 datapoints are missing within Subject 1 and 1 datapoint is missing within Subject 2.
Now, I think enough datapoints are left to reliably estimate the missing data points from the series mean, but I can't find a good source with a critical value of n.

Any thoughts about this?