User talk:Marcomio

Welcome...

Hello, Marcomio, and welcome to Wikipedia! Thank you for your contributions. I hope you like this place and decide to stay. Here are some pages that you might find helpful:
 * Introduction
 * The five pillars of Wikipedia
 * How to edit a page
 * Help
 * How to write a great article
 * Manual of Style

Please sign your name on talk pages using four tildes ( ~ ); this will automatically produce your username and the date. If you need help, check out Questions, ask me on my talk page, or place  on your talk page and ask your question there. Again, welcome! — Q uantling (talk &#124; contribs) 00:19, 24 February 2011 (UTC)

P-value
Thank you for your contribution to the p-value page. However, I wish to alert you that I substantially reversed your edit. My reasoning: The p-value is computed conditioned upon the assumption that the null model is true. As such it may say absolutely nothing about a scenario where the null model is false; those scenarios simply aren't considered in the calculation of the p-value.

If you think I have erred, please discuss the issue here (on your talk page) or on the Talk:p-value page. Of course you are permitted to edit the p-value article directly, but I urge you to instead talk about the edit first. Thank you — Q uantling (talk &#124; contribs) 00:11, 24 February 2011 (UTC)


 * My understanding is that "statistically significant" is an arbitrary decision that the data are unlikely to have occurred by chance if the null hypothesis is true. It's arbitrary in the sense that "unlikely" in conventionally defined as a p of .05 -- but we could have defined it as .01 or .001 or .10.  Your result is either in the rejection region of the null distribution, or it's not.  It's not more or less in the region.


 * I teach stats and have always taught my students that it is correct to say "more unlikely if the null is true" but that to say "more statistically significant" is like saying "more pregnant" or "more dead" or "the light is more on" -- it's an either/or and does not admit of degree.


 * Granted, this is a quibble, but I want people to understand that "statistical significance" has an arbitrary decision criterion, and thus once you have reached that criterion, you can say that it's significant. A p-value of, say, .00001 would suggest "very unlikely to have occurred by chance," but it's no more "significant" than a p of .0499999 (if .05 is the criterion).


 * So that's what I was thinking. It's a semantic thing, not a statistical thing.


 * Marcomio (talk) 20:24, 1 March 2011 (UTC)

I have taken the liberty of removing the duplicate section title, and in indenting your response. These are common Wikipedia conventions, but by no means required, especially since this is your talk page. If these changes are not what you want, please accept my apologies.

Now, back to the discussion at hand. I see the distinction you are making, but I have found that many, including myself, do not make that distinction. For us, "more statistically significant" and "less probable given the null" are exactly synonymous. Perhaps we could put a sentence or two on the inconsistency of this usage somewhere in the article. (Which doesn't help us resolve which convention should appear in the lead paragraph, but it's a start!) Best — Q uantling (talk &#124; contribs) 19:01, 2 March 2011 (UTC)


 * That sounds like a good idea: mentioning the usage in the article. Do you have an idea where you would put it? Marcomio (talk) 20:26, 2 March 2011 (UTC)

I changed the introductory paragraph to avoid the use of "more statistically significant". I suppose a discussion about whether it is appropriate to even use that phrase would belong in the article on statistical significance rather than in the article on p-values. — Q uantling (talk &#124; contribs) 15:51, 3 March 2011 (UTC)


 * I'll take a look at that page as soon as I get a chance. That does seem the best place for the discussion.  Thanks!  Marcomio (talk) 20:28, 4 March 2011 (UTC)