2
$\begingroup$

I have a dataset of 500 people and am trying to fit a prediction model using both quantitative and categorical variables. I have adjusted the dataset as much as possible, but still have one variable (which is important in the analysis) with 19 missing cases. What would you say is the best approach/program to use to replace the missing values/instead of deletion? Would the EM estimator from SPSS provide good enough replacement? Seeing that I have important categorical predictors that will be used in the analysis, I understand that I can't use them in Amos. What other programs could be suited for this type of model?

Thank you!

$\endgroup$

1 Answer 1

3
$\begingroup$

There is not one best approach. All missing data methods come with assumptions, which may or may not be realistic for your data.

However, the impact of the chosen missing data method on the result highly depends on the amount of missing data. You have 500 cases and one variable with just 19 missing values. In that case, the proportion of incomplete records is so low (<4%) that it probably does not matter what you do. So unless the 19 missing cases are very special, the simplest is to delete the 19 incomplete records. This is called complete-case analysis, and the default in many software packages. For higher proportions of missing data, you need to be more careful though.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.