I posted before explaining a problem I am having replicating the way a computer system computates the overall score for managers. If a manager has 4 employees, the system adds up all of the individual answers to each question, and then divides by the total number of answers recieved.

I have the raw data set and need to aggregate by manager ID. If I combine all of the items to create an overall score, such as Compute an overall variable like, MEAN(q1,q2,q3,q4..), then I get an average overall score for each individual employee, and when I aggregate by manager ID, it computes the mean of those employee means, by manager. This results in slightly different numbers than when all of the employee items under one manager are added together, and divided by number of items.

Previously, it was suggested that when I aggregate, instead of getting the mean, I combine the items into a sum, and then divide by the number of items mulitiplied by the N_Break variable. The problem with this, I found out, is that not everything is being divided by a consistent number, because items that were not answered (system missing) are not included in the overall count.

I essentially need a way to group output by manager id, where all items are added together and divided by the number of items that were added.

Anyone have any ideas?