Wikipedia and Plagiarism

Spo22a writes "Daniel Brandt found the examples of suspected plagiarism at Wikipedia using a program he created to run a few sentences from about 12,000 articles against Google Inc.'s search engine. He removed matches in which another site appeared to be copying from Wikipedia, rather than the other way around, and examples in which material is in the public domain and was properly attributed. Brandt ended with a list of 142 articles, which he brought to Wikipedia's attention.... 'They present it as an encyclopedia," Brandt said Friday. "They go around claiming it's almost as good as Britannica. They are trying to be mainstream respectable.'"

Re:That doesn't seem like alot

[tomhudson]

... and after an investigation of some of those by Wikipedia, it was found that some were in the public domain, some were culled from government sites, and some were copied from the wiki, and not the other way around. Of those 12,000, we can now say that the wiki is at least as clean as Ivory soap (99.44%).

Re:ok methodology, bad analysis

[Skippy_kangaroo]

12,000 is easily enough to be statistically effective. Election polling gets acceptable results with samples of about 1,000.

Assuming that it is a binomial distribution then p=142/12000=0.0118, q=0.9882, n=12000 which means the standard error is sqrt(npq)=11.5 (approximately). Thus a 95% confidence interval is that the true number of plagiarised articles in the sample lies between 165 and 119.

And this is only plagiarism from on-line sites that are indexed by Google. Plagiarism from dead tree sources could well be significantly more.

This has got nothing to do with faith-based science and low analytical quality. I am once again amazed at how little people seem to know or care about proper statistics and just say "I don't believe it" if something doesn't accord with their preconceived notions.