There is so much to consider when collecting and analyzing data. We can't force subjects to answer every question so sometimes they skip some answers. When they do this it leads to the problem of MISSING DATA.
I went on a quest to figure out which is the right thing to do when this happens because of a doc student's questions and found out that, as always in this business, it all depends.
1. LITTLE'S MISSING DATA ANALYSIS
The first step is to figure out if data are missing and if so if it is random. Using SPSS, you can click Analyze --> Missing Data Analysis, and select the variables that are in your study.
2. IF DATA ARE MCAR
If data are missing at random (MCAR) then you can use imputation to fill in the holes. Simple imputation seems to have issues if there isn't some way to figure it out from another variable (like SES from income). But multiple imputation, if complicated, seems to be a pretty accurate way to fix the problem.
Next I'll post the step by step for multiple imputation...
I went on a quest to figure out which is the right thing to do when this happens because of a doc student's questions and found out that, as always in this business, it all depends.
1. LITTLE'S MISSING DATA ANALYSIS
The first step is to figure out if data are missing and if so if it is random. Using SPSS, you can click Analyze --> Missing Data Analysis, and select the variables that are in your study.
2. IF DATA ARE MCAR
If data are missing at random (MCAR) then you can use imputation to fill in the holes. Simple imputation seems to have issues if there isn't some way to figure it out from another variable (like SES from income). But multiple imputation, if complicated, seems to be a pretty accurate way to fix the problem.
Next I'll post the step by step for multiple imputation...
No comments:
Post a Comment