Is there a transform for computing dropout rate?

Is there a transform for computing the dropout rate in a group,
given only this information?

A. Imagine that you have a group where a lot of people come
at random times, and start attending meetings for a while,
and then they leave at various times, although a rare few stay
in the group for years.

B. The only data you have is a snapshot of the group, where
a survey was done one day. The members who had a year or
less of membership time were asked how many months they
had been attending.
And the answers, rounded to the nearest integer, were:
1 month: 19%
2 months: 13%
3 months: 10%
4 months: 9%
5 months: 8%
6 months: 7%
7 months: 7%
8 months: 6%
9 months: 6%
10 months: 6%
11 months: 6%
12 months: 5%

Note that many people will not even attend for a whole month.
They may only come for two or three days before dropping out.
Nevertheless, some of them would have been present on the
day of the survey.
The survey did not say how they scored those people -- as
having one month, or just not counting them at all.

So the question is, is there some kind of mathematical transform
that can derive the dropout rate -- preferably the annual dropout
rate -- from that data?

This is not a hypothetical question. There is a real-world application:
I'm trying to come up with another way to compute the A.A. dropout
rate:

http://www.orange-papers.org/orange-effectiveness.html#AA_dropouts


* Agent Orange *
* orange@xxxxxxxxxxxxxxxxx *
* AA and Recovery Cult Debunking *
* http://www.orange-papers.org/ *
Pi is an unending irrational number that was calculated
to 6,422,450,938 digits by a supercomputer in 1995.
The U.S. federal budget deficit is an unending irrational
number that is being expanded by a gibbering gang of
lunatic hairless monkeys at present.

.

Quantcast