Talk:Box–Pierce test

Obsolete ?

"Kiefer.Wolfowitz" would like to label this test as "obsolete". No reliable source has been given for someone calling this test "obsolete". Melcombe (talk) 10:24, 12 March 2010 (UTC)[reply]

The article on Portmanteau tests labels this test as "obsolete". Melcombe may check that I have never edited the Portmanteau test article.Kiefer.Wolfowitz (talk) 14:52, 12 March 2010 (UTC)[reply]

Given the article now says "In fact, the Ljung–Box statistic was described explicitly in the paper that lead to the use of the Box-Pierce statistic, and from which the statistic takes its name", there doesn't seem much point in keeping two separate articles and the best solution may be to do a merge into "Ljung-Box" with a moderate amuunt of detail about Box-Pierce retained. Whether it would be worth keeping Portmanteau test as well is another question. Melcombe (talk) 15:10, 12 March 2010 (UTC)[reply]

A merger is unwarranted, because the Box-Pierce test is often discussed.

Does Melcombe think that every scientific theory in the category "Obsolete scientific theory" needs to have a reference containing the word "obsolete" or does it suffice that such theories be described as "replaced by superior theories" in (nearly all) reliable sources (that condescend to mention the obsolete theory)?

Apparently all reliable sources discussing both the Box-Pierce test and the Box-Ljung test observe note that the Box-Peirce test has been replaced by the Box-Ljung test.

See page 202 (in the paper according to Google Scholar, and page 204 in the book) of the invited Biometrika centenary issue, which describes the flawed algebra of the original paper, and it replacement by Box-Ljung, etc.

* Howell Tong. A personal journey through time series in Biometrika. Biometrika, vol. 88, 195-218.

Kiefer.Wolfowitz (talk) 15:15, 12 March 2010 (UTC)[reply]

The present article gets the history backwards. See page 204 in the book version of Tong. IMHO, this renders the merger discussion moot.

This is as usual total nonsense. We can ask ourselves ... who would know? Answer: Ljung&Box. Since the Ljung&Box reference explictly states that the test statistic derives from the Box&Pierce paper, that would seem to be definitive.Melcombe (talk) 16:25, 12 July 2010 (UTC)[reply]