Jump to content

Talk:Robust confidence interval

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

Original text copied over from article on confidence intervals, where it was in danger of being deleted. I think there may be something worthwhile here. Melcombe (talk) 10:33, 7 April 2008 (UTC)[reply]

I find this article incredibly confusing and self-contradictory. Why don't you just say that, if you use the sample standard deviation to compute confidence intervals, then if a single weighing is off by a large amount (maybe a decimal digit was completely omitted, for instance), then the computed confidence interval might be thrown off by a lot? And you could say that the purpose of robust confidence intervals is to ensure that no single error or outlier can throw off your results by a huge margin. That would be a lot more understandable than the current paragraph. Also, there are apparent errors. "weighing 1000 objects": I think you mean "weighing 100 objects, each one weighed ten times". "200 extra weighings": huh? do you mean "700 extra weighings"? 70.36.134.243 (talk) —Preceding undated comment added 01:13, 12 November 2010 (UTC).[reply]