Talk:Fractional model
Add topic | This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | ||||||||||||||||||||
| |||||||||||||||||||||
Logistic regression as a link function
[edit]The distinction between "uses a log-odds transformation of y as a linear function of x_i" and "using the logistic regression as a link function" seems dubious to me but I do not have enough qualifications in statistics to feel confident in modifying this page. Both sound like they point to the same model, with the second one appearing to be a misunderstanding of using the logit model as fractional model. The link function in the logistic regression is the log-odds.
The following problems are raised with the log-odds link function: "This approach [log-odds transformation of y] is problematic for two distinct reasons. The y variable can not take on boundary values 1 and 0, and the interpretation of the coefficients is not straightforward.". These problems also seem to hold for the "logistic regression as link function" model, as it is proposed here.
No clear reference is given for the "logistic regression as a link function". Wooldridge, J. (2002): Econometric Analysis of Cross Section and Panel Data, MIT Press, Cambridge, Mass. is provided as a reference for the paragraph but it is extremely general. Pof & Co (talk) 12:21, 29 December 2025 (UTC)