Expert: Surajit Basu - 9/2/2008

Question
Hi there, I would  really appreciate your help on the following query:  I am looking at quarterly data from the last three years which I am going to use to benchmark other experimental data.  However the quarterly benchmark data is subject to change due to seasonality and other factors.  If i wanted to use this information as accuaretly as possible what would you suggest.  a) using the average of all quarters as one benchmarking quarter or b) the average of all similar quarters (giving four benchamrking quarters)?  i would be very grateful for any suggestions.  many thanks, Naheed

Answer
hi

it is tough to make a generic statement.

if the seasonality is high, e.g. with sales curves which peak in one quarter, then it is wise to take that into account and plan for experimental data which is based on specific quarters(b)

if the seasonality is low, and the data you will collect will be annual-based, i would suggest (a)

in general, one should go with b
and  analyse as part of the experiment if seasonality exists.