Ken Aston science forum beginner
Joined: 07 Jul 2006
Posts: 1

Posted: Fri Jul 07, 2006 2:18 am Post subject:
Test of significance of a mean value?



Hello,
Is there any method to calculate the significance of a mean value?
Example:
n customers i visited some store count_of_j times in one year and spent
x_ij minutes in the store.
For each customer I calculate the mean value of minutes per visit
x_mean_i. Also I calculate the overall mean value for all customers
x_mean.
Now I want to estimate how much time a customer will spend in the store
in the future. If the customer had visited the store fairly often his
x_mean_i is a reliable estimator. But if he visited only a few times,
for example twice, x_mean_i should not be used. Instead I would like to
use the overall x_mean in these cases.
My question is, how can I calculate a statistic which supports me in my
decision of using x_mean_i or x_mean? Ideally I would like to know how
significant the calculated x_mean_i is and in case of sig<0.05 reject
x_mean_i and use x_mean instead.
Thanks for any help!
Ken Aston 
