Timeline for Parametric modelling of variance of count data
Current License: CC BY-SA 3.0
14 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| May 11, 2017 at 22:16 | history | edited | kjetil b halvorsen♦ |
edited tags
|
|
| Dec 4, 2011 at 21:56 | history | notice removed | whuber♦ | ||
| Dec 4, 2011 at 21:56 | history | bounty ended | whuber♦ | ||
| Dec 2, 2011 at 18:36 | vote | accept | Brian Diggs | ||
| Dec 2, 2011 at 18:33 | comment | added | guest | Chapter 10 of McCullagh and Nelder discusses Joint Modelling of Mean and Dispersion, i.e. parameterizing both the mean and the variance. Extended quasi-likelihood is the main tool, but in some situations there can be concerns about that method | |
| Nov 28, 2011 at 0:24 | answer | added | timbp | timeline score: 9 | |
| Nov 28, 2011 at 0:20 | answer | added | jbowman | timeline score: 11 | |
| Nov 27, 2011 at 22:20 | history | notice added | whuber♦ | Draw attention | |
| Nov 27, 2011 at 22:20 | history | bounty started | whuber♦ | ||
| Oct 19, 2011 at 23:30 | comment | added | Karl | I think there's a chapter in McCullagh and Nelder, Generalized linear models, 2nd edition, that covers this (but my copy's at work)...there won't be a real likelihood, but you can use quasi-likelihood, and so that may be the title of the chapter. You apply iteratively reweighted least squares even though there's no probability model that corresponds. | |
| Oct 19, 2011 at 22:05 | history | tweeted | twitter.com/#!/StackStats/status/126781164416532481 | ||
| Oct 19, 2011 at 21:46 | comment | added | Brian Diggs | @whuber That's a fair point. For a single categorical predictor looking at the variance and mean of the sub-groups would be sufficient to detect overdispersion, but for a multivariate Poisson regression, it is not. For the sake of argument, let's assume both a Poisson and negative binomal regression have been done and the negative binomial shows a better fit via anova model comparison. That should indicate overdispersion. Given that, how could the variance/overdispersion be modeled parametrically rather than as a constant? | |
| Oct 19, 2011 at 21:31 | comment | added | whuber♦ | How do you know there is overdispersion without first doing the Poisson regression? After all, comparing the variance of the raw (response) values to their mean isn't relevant: what matters is the goodness of fit of the Poisson model (this is the analog of evaluating the distribution of residuals in a linear model compared to evaluating the distribution of the response variable). Another way to put this is that the link between the independent variables and the response can create the appearance of overdispersion even in a beautifully accurate Poisson model. | |
| Oct 19, 2011 at 21:21 | history | asked | Brian Diggs | CC BY-SA 3.0 |