Why not?... Suppose 10,000 samples (say), each of size n, were taken from a normal population, and 10,000 CIs were constructed using the formula for t-interval. Then, 95% of the 10,000 CIs would contain the population mean, even though we don't know which 95% of them would contain the mean.
When we construct a CI based on a random sample, it is reasonable to expect that the sample is from one of those containing the population mean.. Krishnamoorthy ----- Original Message ----- From: "User968758" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Monday, November 18, 2002 7:39 AM Subject: Re: Two questions > "So if you take a good random > sample and compute a 95% confidence interval, there is a 95% chance > that the true population parameter is within the computed interval." > > Absolutely not true. > . > . > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at: > . http://jse.stat.ncsu.edu/ . > ================================================================= > > . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
