Statistical significance of mailing

#1
Hi All,

I'm needing help with identifying the statistical test to conduct in order to determine whether or not the results of my test are statistically significant.

I have done a mailing and split the audience randomly into two discreet groups. I then have data for response, cost per mailing (the groups got different valued packs) and purchase amount.

I know how to calculate whether the difference in response rates is significant, however I am not sure what test to use to calculate whether the difference in the purchase amounts for my two groups are significant.

It is important to be aware that the data doesn't follow a normal distribution.

The data I have in my file are:
URN
Mailing Audience (Treatment or Control)
Mailing Cost
Response Flag
Purchase Amount


I am learning R so if some guidance on how to do this in R is possible that would be super helpful, but I understand not all people use R so am happy with the help for just which test to use.

Thanks in advance
Dan
 

Karabiner

TS Contributor
#2
I know how to calculate whether the difference in response rates is significant, however I am not sure what test to use to calculate whether the difference in the purchase amounts for my two groups are significant.
Have a look at either independent samples t-test or Mann-Whitney U-test.
It is important to be aware that the data doesn't follow a normal distribution.
Usually not. For the t-test, each group (not the total sample)
should be from a normally distributed population. But with
n > 30 this assumption is considered negligible. If both
samples have markedly different sizes, then the standard
deviations should be equal (if not, one can use the t-test
with Welch correction).

With kind regards

K.