Talk:Statistical noise

The phrase "noise" when used in discussing the statistical reliability of a test tends to mean that differing results between test cells are as likely to be random variation rather than a statistically reliable result, usually as a result of a very small sample or result set. For example, when performing an A/B (two equal sized cells) test on two versions of a web page and measuring resulting orders, version A might result in 12 orders, while version B results 8 orders. In this case someone who is not trained in the appropriate use of statistics might be tempted to proclaim version A "the clear winner, with 50% more orders than version B," while in reality the result set is so small that the difference between the versions could quite likely be the result of normal variation, or "noise."

Change to redirect?
This article has only one substantive sentence, and it’s been that way since May 2004. Any objection to my changing it to a redirect to Errors and residuals in statistics? Loraof (talk) 17:28, 22 February 2018 (UTC)

Redirected to Unexplained variation. Loraof (talk) 15:25, 24 February 2018 (UTC)