13 April 2022 |
Automatically Impute Missing Values
Sometimes you have some missing data and you want to work out what people would have said. Maybe it’s a political poll and they said “Don’t know”. Maybe it’s income and they refused to give you the answer. The technical name for estimating people’s missing data is imputation. Q’s now got an automated way of imputing, which uses a methodology known as MICE (Multiple Imputation by Chained Equations).
The way it works is:
- On the Variables and Questions tab, you select the variable or variables with the missing values.
- Automate > Browse Online Library > Create New Variables > Impute Missing Data. This will automatically generate missing values just based on the selected variables. But, you can improve this, by following the next few steps.
- Right-click and select Edit R Variable
- Choose any variables that you feel will improve the prediction as Auxiliary variables.
- Press Update R Variable.
In the example below, you can see that the first person to say I’d rather not say has been estimated to have an income of $200,001 or more, and the second person $150,001 to $200.000.