Hello! I have data for a sample of 74 people who answered a 97-question assessment utilizing a likert scale. Looking at the data visually, we noticed a few responses that look like outliers/careless responses and we are considering removing them from the dataset. We want to do a formal outlier analysis to justify this. I read that Mahalanobis distance can be used to identify outliers but I don't know how to do it in SAS and with my variables. The 97-item assessment is broken up into subscales with items belonging to only one subscale. Would I calculate the means for each of the subscales and then use those values to calculate Mahalanobis' distance between the individual's mean on the subscales and the sample's means? What would the specific syntax be for this?
