PDA

View Full Version : Credibility Theory

schaden
03-04-2002, 03:24 PM
Please help with this example. I have a school population of 30381 students where 5.2% are suspended. Of the 30381 students, there are 1584 minorities in which 9.8% are suspended. How many students do I need in the subset in order to say the data is credible?
Anonymous
03-04-2002, 03:49 PM
roughly 32 observations are needed to assume a normal distribution.

GG
General Kenobi (ret.)
03-04-2002, 03:49 PM
It's not a credibility question, it's a difference of means question for a binomial distribution.

Try this page. (http://www.cate.org/sms99/apst99/ymmsum99/ymm11112.htm)
Anonymous
03-04-2002, 03:50 PM
OB Wan - 32 and you can approximate a normal, which should work for C.I. (confidence interval)
General Kenobi (ret.)
03-04-2002, 03:52 PM
Good point. Just make sure to estimate the parameters correctly.
Jonas Grumby
03-04-2002, 05:25 PM
Obi - thanks for providing some 'schadenfreude'.
Anonymous
03-04-2002, 05:31 PM
Pleasure derived from the misfortunes of others