Date: Mon, 22 Mar 2004 18:00:58 +0100
Reply-To: Fair Shannon <s.fair@CRIFGROUP.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Fair Shannon <s.fair@CRIFGROUP.COM>
Subject: Re: Statistical Modeling Question
Content-Type: text/plain; charset="us-ascii"
Hi Nick - Usually the number of times mailed is highly correlated with
their response. You can use a variable in the model, say 'times
mailed,' and see what happens. In general, they're more likely to
respond if they've never been solicited or have only been solicited once
before, regardless of slight variations in product. If you know which
product they responded to, maybe you could see if the product is causing
them to respond. You could create interaction variables as well to see
if it is the combination of product & times mailed that is driving the
response.
Hope this helps...
Shannon Fair
-----Original Message-----
From: Nick I [mailto:ni14@MAIL.COM]
Sent: Monday, March 22, 2004 11:24 AM
To: SAS-L@LISTSERV.UGA.EDU
Subject: Statistical Modeling Question
Hello SAS forum
I have a statistical modeling problem I need some guidance/ideas on.
Say I have a mortgage with a bank. As you know the bank will send me, at
some point, information on some of its other products (insurance,
bank-related products, etc.) for me to sign on.
Here is an example:
CUSTOMER_1
In Mar04, they campaign to customer_1 product A. He gets it in the mail,
he reads it, he tosses it out.
In May04, they campaign to customer_1 product A. He gets it in the mail,
he reads it, he tosses it out.
In Sep04, they campaign to customer_1 product A. He gets it in the mail,
he reads it, he tosses it out.
In Dec04, they campaign to customer_1 product A. He gets it in the mail,
he reads it, he tosses it out.
CUSTOMER_2
In Mar04, they campaign to customer_2 product A. He gets it in the mail,
he reads it, he tosses it out.
In May04, they campaign to customer_2 product A. He gets it in the mail,
he reads it, he accepts. (Bank will no longer campaign to CUSTOMER_2
this product since they bought it, so what's the point, right?)
CUSTOMER_3
In Mar04, they campaign to customer_3 product A. He gets it in the mail,
he reads it, he accepts. (Bank will no longer campaign to CUSTOMER_3
this product since they bought it, so what's the point, right?)
CUSTOMER_4
This customer is like CUSTOMER_1.
CUSTOMER_5
In Mar04, they campaign to customer_5 product A. He gets it in the mail,
he reads it, he tosses it out.
In May04, they campaign to customer_5 product A. He gets it in the mail,
he reads it, he tosses it out.
In Sep04, they campaign to customer_5 product A. He gets it in the mail,
he reads it, he tosses it out.
In Dec04, they campaign to customer_5 product A. He gets it in the mail,
he reads it, he accepts. (Bank will no longer campaign to CUSTOMER_5
this product since they bought it, so what's the point, right?)
The most I have seen a customer being campaigned to is 4 times in the
space of a year.
My question is:
How can I use SAS to model the response in this problem?
CUSTOMER_1 never responded
CUSTOMER_2 responded the ***2ND*** time (i.e. in the 2nd try)
CUSTOMER_3 responded the ***1ST*** time (i.e. in the 1st try)
CUSTOMER_4 never responded
CUSTOMER_5 responded the ***4TH*** time (i.e. in the 4th, and last try,
for this year 2004)
In this problem I have kept the PRODUCT A the same for simplicity.
In real life, and this is happening to me right now with a credit card
company (Citi bank), product A will vary slightly from campaign to
campaign. For example, in my case, the first time I received their
credit card offer was for an APR of 8.99%. (I did not respond.) A few
days ago (i.e. about a month later) I received their second offer except
the APR was 5.99% this time, all else was the same. Anyway. You see the
problem I am trying to solve. (I am still not responding bec., I don't
like certain other things although the APR is great.) By the way. If you
think you can also offer me suggestions on how to solve this more
realistic problem where product A changes sligthly, that will be great.
I am affraid though that when you send different products--even if they
are slightly different--to the same person and the person responds the
3rd time, you don't really know if they responded bec. of the # of times
they have been touched, or bec. they received a slightly different pr
oduct and they liked it better and therefore responded. I just wanted to
keep the product the same so the statistical theory will be easier, I
think????
What I am trying to understand is if there is a connection between the
number of campaigns and response. Say I am CUSTOMER_5 and that I was in
a 'bad mood' when I got the mailings for product A the previous 3 times,
but on the 4th time, for whatever reason, I responded. (NOTE: I am
keeping the PRODUCT A the same so we don't have too many things
changing, but, in reality, say, I responded this time bec. their offer
got sweeter and more to my liking.) That's what I want SAS to help me
understand.
You help is appreciated as always
nick
--
___________________________________________________________
Sign-up for Ads Free at Mail.com
http://www.mail.com/?sr=signup
<http://mail01.mail.com/scripts/payment/adtracking.cgi?bannercode=adsfre
ejump01>
