I am analysing a small number of patients (n=28) and want to do Spearman rank correlation. For one variable there are two people with missing data. Is it inappropriate to do a mean imputation? The reason the data is missing is entirely my own fault as I missed the question from the questionnaire by accident.