Talk:Chow test

null hypothesis

what is the null? with break or without break? Jackzhp (talk) 02:55, 11 March 2008 (UTC)[reply]

the null hypothesis is no break, cf Rao et al, section 3.15.1, 'The Chow-Test as a Mean-Shift Test'. Shabbychef (talk) 20:40, 5 December 2011 (UTC)[reply]

Degrees of freedom

Should'nt the degrees of freedom for the Chow test be k and N₁+N₂-2k, when k are the number of parameters? —Preceding unsigned comment added by Radiohead67 (talk • contribs) 06:53, 13 June 2008 (UTC)[reply]

Well I think yes! But isn't it the same on the page? I also read k and N₁+N₂-2k,... Be carefull that there are actually two forms of chow tests in his seminal paper! EtudiantEco (talk) —Preceding comment was added at 12:33, 13 June 2008 (UTC)[reply]

The cited STATA sources agree with y'all and disagree with the March 16th edit to the degrees of freedom (which subtracted the constant 2). Unless that version is in the orginal paper (and is the version usually relied upon) I suggest reverting that edit. 24.155.242.66 (talk) 15:13, 23 September 2008 (UTC)[reply]

Added the part on the predictive Chow test used by SAS and checked the number of degree of Freedom. MBuisine (talk) 09:22, 24 June 2015 (UTC)[reply]

Need to improve

The global sum of squares (SSE) is often called the Restricted Sum of Squares (RSSM) as we basically test a constrained model where we have assumptions (with the number of regressors).

Wrong or very unclear sentence. And preceding symbols of the restricted (RSS_R) and unrestricted (RSS_U), and DoF not defined. If I understand instead of RSSM should be RSS_R. Pawel.jamiolkowski (talk) 15:04, 4 July 2022 (UTC)[reply]