Cross posted to sci.stat.consult, where the same original text was posted.
On 11 Jun 2003 19:43:01 -0700, [EMAIL PROTECTED] (Mohammad Ehsanul Karim) wrote: [ ... ] > I am working on some regression project and would love to know that > if there are any other assumptions of Regression analysis other than > the following unlucky 13 assumptions. > Also please let me know if any of these are inappropriate as any > assumption in general in regression. I think 13 is far too many for me to keep track of. However these are mostly for some need other than regression "in general". You can perform regression so long as you (generally) have more cases than variables; and the scores can't all be the same. I don't know if I've seen a totally precise statement of that. You do need to meet assumptions in order to trust the statistical tests. You need to meet some additional assumptions in order to trust the implications of the regressions coefficients. [ ... ] > Please send a copy of your kind response in my e-mail address at > [EMAIL PROTECTED] It is easier if that is your Reply address. I've added it. > > ----------------------------------------------------------- > Assumptions that we make in Regression Analysis > ----------------------------------------------------------- [ snip: commentary that reminds me of Box's observation, something like, "All models are wrong; some models are useful."] > The least squares fitting procedure can be used for data analysis as > a purely descriptive technique. - right - > However, the procedure has strong > theoretical justification if a few assumptions are made about how the > data are generated. - "justification" - for what, in particular? It is easier to have tests than it is to draw valid conclusions from the tests. > In the context of scientific investigation, > "assumptions" are not just devices to simplify mathematics, they are > supposed to be a reasonable mathematical representation of the > data-generating process. Some assumptions are mathematical. Some assumptions are ontological (about origins) or epistemological (about knowledge). [snip, Assumptions 1 - 8] > (i) Assumption 9: Model Specification > > There should be no specification bias in the model used in empirical > analysis, that is, the regression model should be correctly specified > about > - Variables > - Functional form > - Assumption [ snip, rest] It does seem to me that if the model is correct in variables and functional form, that precaution includes much of what is detailed in the other 12 assumptions. I would discard the first half of the sentence with the word "bias", above, and "assumption" in the list. You get into certain sorts of problems if you mis-specify the variables. There are other potential problems if you miss the functional form. You can mess those up in different ways - some are less serious than others, and oftentimes "it depends" on the exact details. When they do matter, they can mess up tests, or mess up inferences despite the legitimate tests. Hope this helps. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." Justice Holmes. . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
