Which statistical test to use

suchi

New Member
#1
I have a 2 test groups and 1 control group and my goal is to understand if a particular email campaign is successful or not.The kpi under test is cancellation rate which is a continuous variable.I want to understand:

1.If the email campaign for the test groups results in a low cancellation rate?
2.Which test group test 1 or test 2 performs better?

Is running 2 independant t-tests a possible solution to the problem.Need ideas
 

Karabiner

TS Contributor
#2
I suppose that cancellation yes/no is your variable, not cancellation rate.
Cancellation rate is a summary statistic.
If you have n observations, each with outcome cancellation or
no cancellation, and you have 3 treatments, then this could be a
case for a 3x2 table with Chi2 test.

If your study design is different from what I supposed, then please
describe it in more detail.

With kind regards

Karabiner
 

suchi

New Member
#3
Thanks a lot for replying.

So the study contains 2 test groups with 200 observations each who receive emails and the control group has 2400 observations and dont receive emails.The aim of the email is to reduce cancellations and hence retain customers.This experiment is going to run over a time period and the cancellation % is calculated every week.

I believe the Chi square test would work if I am looking at 2 categorical variables and a possible relationship between them.

I need to determine the following:

1.Is the email campaign working ie.Does sending the email drive less cancellation? Compare the test groups and Control group
2.Which email message (Test 1 or Test 2) drives less cancellation. Compare two test groups

Let me know if this information helps
 

Karabiner

TS Contributor
#4
You could have told us before that this is a homework question.

You have 2 categorical variables, "group" with 3 levels
(or 2 levels, respectively, depending on the precise research
question) and "cancellation" (yes/no).

With kind regards

Karabiner
 

suchi

New Member
#5
I am fairly new to the platform.Will make sure to check previous threads.Also I am fairly new to statistical testing and this is my first assignment in conducting an experiment.

Well this makes sense to me know.A 3*2 contingency table should help.


Thanks a ton for all your help.

Regards,
Suchi
 
Last edited:

Karabiner

TS Contributor
#6
It is not 100% clear how to deal with the first research question.
It could be a 3x2 table, but maybe they wish that you collapse
both test groups and to build and analyse a 2x2 table.

With kind regards

Karabiner
 

suchi

New Member
#7
Thanks a lot for that advice.1 last follow up question.

The control group is quite small and the test group is much larger.Will the test results be sensitive to this difference.Please find the numbers attached in the file.

Thanks a lot for your time
Suchi
 

Attachments