February 12, 201412 yr Hi, There was a Time study carried out in one of the back office processes. The objective of the Time study was to establish a baseline which we could propose to the client as our initial target in terms of Average Handling Time for each transaction. However, the data turned out to be non-normal, a distribution which was skewed towards the right with majority of the data points on the left. Would it be logical to use the mean as the baseline measure in this case and propose the AHT to the client. Or should it be Median since the data is not normal and most data points aren't around the mean, can it also be the 3rd quartile instead of median? I'm looking at a logical explanation on how to arrive at a baseline figure when it comes to any data that's non-normal, including bimodal or other distributions.
April 2, 201412 yr Dear Vikram, You have options. If the client is looking for AHT, I do not know if they will be happy looking at median though it is a better representation of your data. You may transform the data for baselining purposes and calculate the Z value. For hypotheses however, it is better you consider non parametric tests with original data. Hope this helps.
Create an account or sign in to comment