0

Given two confidence bands, it seems natural to consider the confidence band of the product of these two distributions. Does this have a name? Are there any reference that might be helpful regarding this notion?

Edit: Assume that the errors of the first, say, regression curve are not correlated with the errors of the second.

Lepidopterist
  • 712
  • 5
  • 23
  • Threads about the [delta method](http://stats.stackexchange.com/search?q=delta+method) are likely to provide appropriate solutions. – whuber Oct 24 '13 at 05:37
  • You might need to a bit more specific, @Gregorian Funk. Did you mean, say, the confidence interval of a variable that is the product of two variables whose confidence intervals are known? – abaumann Oct 24 '13 at 19:07
  • Yes, I am asking for a new confidence band of the product of the two curves. Assume that each curve is derived from some simple regression, and it is assumed that the errors are uncorrelated. – Lepidopterist Oct 24 '13 at 21:20
  • As @whuber said, the delta method might be an answer. This might also interest you: http://stats.stackexchange.com/questions/62916/confidence-interval-for-the-product-of-two-parameters – abaumann Oct 25 '13 at 06:44

0 Answers0