Hi. Thanks. To fix my example, there is only one item that is to be reverse-scored, and should be represented in my example as a column among several columns:
Yes, it ends up not mattering for correlation (reliability) computations because each column is ultimately converted to standardized scores (standardized separately for each column), but still mattering for other computations.
Inconsistency in scale calculations
Re: Inconsistency in scale calculations
i assume this can all be solved by adding the missing levels with the level builder, and using the "retain unused levels in the analysis" option?
jonathon
jonathon