|WikiProject Statistics||(Rated Start-class, Mid-importance)|
|WikiProject Mathematics||(Rated Start-class, Mid-importance)|
This is not the same thing as Log-odds. There should be a separate article for that. --Scottieb 20:02, 2 October 2006 (UTC)
- What do you take to be the difference? The logit of a number p is the log-odds of any event whose probability is p. The difference would seem to be that the logit is the logit of a number, whereas the log-odds is the log-odds of an event. Is that what you had in mind? Michael Hardy 20:40, 2 October 2006 (UTC)
Could someone fix up the image size for the logit-plot. I don't have the time right now. SW
- I use this function quite often without doing logistic regression - better to merge logistic regression with Probit model --Henrygb 01:24, 23 April 2006 (UTC)
- How about before merging probit and logistic regression, we clarify on both pages what the relationship is between them? Currently neither page explains the other. Cazort 23:03, 24 October 2007 (UTC)
This article has improved much over the last few days. Two things I might suggest: 1) use the deleted hospital example as an example in the logistic regression article (the current Ax+B example is nothing more than an example of calculating the odds); 2) In History, distinguish between the logit proper, and the logit model, which I assume is another word for logistic regression. Comments welcome and encouraged. Baccyak4H (Yak!) 16:09, 13 June 2007 (UTC)
Needs more Explanation
This article does not give adequate explanation/justification for the use of the logit function. Why is it used? What compelling mathematical, physical, or philosophical reasons are there for its use, for example, its use in logistic regression? I think this page can only be viewed as a stub until it contains this sort of material so I am going to mark it as such. Cazort 18:47, 24 October 2007 (UTC)
- I agree. Furthermore it would be interesting to know why it is called "logit". -- 188.8.131.52 (talk) 08:31, 6 April 2009 (UTC)
- OR, but the presumed analogies are probit, logarithm and logic --Rumping (talk) 11:21, 9 October 2009 (UTC)
Comparison with Probit
"As a result, probit models are sometimes used in place of logit models because for certain applications (e.g. in Bayesian statistics) implementation of them is easier."
Could probit and logit be reversed in this sentence? Logit is almost always going to be computationally easier than probit, and Bayesian statistics is computationally demanding, even today.
Another important difference with probit that is often considered, i.e. when deciding between regression models, is that the logistic distribution has thicker tails than the normal distribution. This is visible in the graphic but strikes me as worth of a comment, as it offers reasons to use a logit model beyond computational ones. — Preceding unsigned comment added by 184.108.40.206 (talk) 05:18, 1 August 2013 (UTC)