N/A
Posts: 0

# Measuring Incremental impact

Hi all,

I'm fairly green when it comes to statistical procedures in SAS (and to be honest some stats in general). I have a dataset of customers that were used to conduct a test. The data was split into 3 segments, and then random control cells (~10%) were taken from each of these 3 segments.

I have a dataset that lists all the customers (segments and controls), the segment they were in, and whether or not they were part of the treatment or control. I have then appended data (left join) where the customer exhibited the desired behaviour.

So my dataset looks like the following example.
CUSTOMER | SEGMENT | GROUP | RESPONSE | RESPONSE DATE
--------------------|-------------------|----------------------|---------------------|-----------------------------
1234567 | SEG-A | TREATMENT | |
2225874 | SEG-A | TREATMENT | ABC123 | 12Jan2009
6657894 | SEG-A | CONTROL | ABC123 | 12Jan2009
3215870 | SEG-A | TREATMENT | ABC124 | 12Jan2009
--------------------|-------------------|----------------------|---------------------|-----------------------------

Please keep in mind this is not a complete sample and there are other segments that would/should be indicated here.

What I am wanting to do is to calculate is 2 things. Which groups had a statistically significant response (to 95% confidence level) and the number of incremental responses in each group.

It would be great to be able to determine this for each group over time (so I can indicate when the results became statistically robust).

Firstly, is there a SAS procedure that will allow me to do this easily, or am I going to have to do manual calculations to determine this?

Any help would be GREATLY appreciated. Thanks in advance.

DW
Frequent Contributor
Posts: 77

## Re: Measuring Incremental impact

Hi DWW,

Here are my few cents.

Use PROC FREQ to calculate frequencies and test if there is association between your column and row variables. I suggest first doing something like this:

proc sort data=;
by DATE;
proc freq data=;
by DATE;
table GROUP*RESPONSE/all;
run;

This will produce frequency counts and association tests for each RESPONSE DATE.

From your example I can not determine if you have observations on the same individuals at more than one DATE. If you do, then PROC GENMOD is one the procedures that can be used for analysis of longitudinal data with categorical response.
Frequent Contributor
Posts: 140

## Re: Measuring Incremental impact

You write
> What I am wanting to do is to calculate is 2 things.
> Which groups had a statistically significant
> response (to 95% confidence level) and the number of
> incremental responses in each group.
>
> It would be great to be able to determine this for
> each group over time (so I can indicate when the
> results became statistically robust).
>

I am not sure what you want in the first request. Statistically sig.
response to what?

The second request can be met with a DATA step using first.
and last. variables. I don't have time now to do this,
but searching through SAS-L archives will find it, or, if
you have Ron Cody's book on longitudinal data, he shows
several ways to do this

Peter
Discussion stats
• 2 replies
• 91 views
• 0 likes
• 3 in conversation