# Thread: Confidence and prediction intervals

1. ## Confidence and prediction intervals

If we have an extremely large sample, 95% and 99% confidence
intervals for the mean response will be very narrow indeed. Discuss
whether, in the same situation, it also becomes possible to do very
accurate predictions of individual values – that is, to obtain very
narrow prediction intervals.

(Hint: the amount of random variation associated with individual
values does not change as information on other individuals becomes
available.)

2. ## Re: Confidence and prediction intervals

In order for larger sample populations to reduce the variance of the sample mean you have to make the assumption that members of the population are independent. Thus the conditional distribution of any parameters you might be interested in about the individual is equal to their unconditional distribution. Information about the rest of the sample population provides no information about that individual.