Talk:stignificance

OK, you've caught me. I've just made up this word, BUT it is very important that this word gets some currency and some acceptance.

Scientific and medical literature is currently bedevilled by use, misuse and ab-use of the word "significance". To the researcher and statistician, this word has been co-opted from regular English to denote a research result linking two or more dependent variables, the correlation between which is highly unlikely to have been random, generally by convention a less-than-5% chance (p<0.05). Consequently the significance of a "significant" result is as dependent on study size, or numbers of subjects studied, as it is on the import of the result. Small studies of major phenomena might not have the statistical power to meet "significance" and conversely huge studies can have the statistical power to detect trivial differences between two study arms and thus report a result that is highly "significant" but actually so minor as to be meaningless.

If "statistical significance" as a concept can be denoted instead by "stignificance" then we can remove the ambiguity of the root word, use the word "stignificance" with certainty of meaning, and allow "significance" to signify significance again.

So, please be kind and don't destroy this contribution to the language and to accuracy in scientific communications (and, I might add, honesty in communication between Pharma and doctors and lay persons).

Use it.

Cheers

Dr John Mahony


 * We don't add words here until they have already gained currency, but you can use WT:LOP. Thanks. Equinox ◑ 13:23, 17 September 2013 (UTC)