Search found 156 matches

by billb
Thu Nov 15, 2012 5:14 pm
Forum: MLwiN user forum
Topic: Compare DIC
Replies: 4
Views: 8403

Re: Compare DIC

There is no theory based fixed difference e.g. 10 that should be compared when comparing the DIC of two models although 10 might have been a rule of thumb suggested by someone - in theory the model with the lowest DIC is better but of course there is stochastic noise when using MCMC so a very small ...
by billb
Thu Nov 15, 2012 5:08 pm
Forum: MLwiN user forum
Topic: Changing values in Variance- Covariance matrix
Replies: 3
Views: 6678

Re: Changing values in Variance- Covariance matrix

Dear hdilk0507, I saw your question and have a quick answer. It really makes no sense to allow a level 2 variable to vary across level 2 units. Consider 1 level 2 unit then the predictor associated with the intercept will take value 1 for all individuals in the unit and the level 2 variable will als...
by billb
Thu Oct 11, 2012 3:05 pm
Forum: MLwiN user forum
Topic: Residuals prediction and covariance-variance
Replies: 1
Views: 3741

Re: Residuals prediction and covariance-variance

Hi Econsic1976,
This is because the residuals are shrunken residuals - I think the answer might be quite technical and available in Harvey Goldstein's book or you could email him.
Regards,
Bill.
by billb
Thu Oct 11, 2012 3:02 pm
Forum: MLwiN user forum
Topic: Zero variance question
Replies: 2
Views: 4798

Re: Zero variance question

Hi Mdommers,
I could take a look if you send me your worksheet (william.browne@bristol.ac.uk).
Regards,
Bill.
by billb
Tue Sep 11, 2012 2:08 pm
Forum: MLwiN user forum
Topic: predint.txt problems
Replies: 1
Views: 4264

Re: predint.txt problems

Hi Sian, My predint.txt macro is going to be tricky to replicate for bigger datasets - basically the following is done: 1) The residual chains have been stored stacked up in c200 - there are 130 of them as there are 65 schools (intercepts and slopes) 2) The school ids go from 1 to 65 which is import...
by billb
Wed Jul 18, 2012 2:39 pm
Forum: MLwiN user forum
Topic: Variance disappearing in shift from 2- to 3-level model
Replies: 4
Views: 6954

Re: Variance disappearing in shift from 2- to 3-level model

Hi Matt, I am still a little puzzled. If as you say there are 13 categories for each GOR then a nested 3 level model should have 12 level 3 with 12*13 categories = 156 level 2 units. The only way that categories should be 13 would be if you had a cross-classified model. Am happy to take a look if yo...
by billb
Thu Jun 28, 2012 2:23 pm
Forum: MLwiN user forum
Topic: Variance disappearing in shift from 2- to 3-level model
Replies: 4
Views: 6954

Re: Variance disappearing in shift from 2- to 3-level model

Hi Mattoftheday, your email puzzles me a little. When you say you have a structure of 12 > 13 > 84 do you in fact mean 12 GORS with 12*13 Categories within the 12 i.e. categories within GOR or do you mean you have a cross-classified structure with GOR crossed by Category ? Either way these sorts of ...
by billb
Mon May 21, 2012 8:50 am
Forum: MLwiN user forum
Topic: one-tailed test for a predictor in MCMC
Replies: 2
Views: 7091

Re: one-tailed test for a predictor in MCMC

Hi Sian, I think you are a little confused here. The DIC diagnostic is an information criterion that can be used to compare models in a similar way to the AIC and BIC in a non-Bayesian/MCMC setting. What it is not is equivalent to 'a two-tailed test at 5% significance level'. The closest thing in th...
by billb
Mon May 21, 2012 8:42 am
Forum: MLwiN user forum
Topic: question about a cross-classified model
Replies: 2
Views: 5711

Re: question about a cross-classified model

Hi Jelle, Yes this sounds appropriate for the analysis you intend to do. I would firstly look at each of the 2 ratings separately and fit a straightforward single response cross-classified model as in the chapter in my MCMC book. To fit a MV model you will need to follow the example in the MV Normal...
by billb
Mon Aug 15, 2011 1:23 pm
Forum: MLwiN user forum
Topic: Cross-classified multinomial problems
Replies: 2
Views: 6953

Re: Cross-classified multinomial problems

This shouldn't be too bad in MCMC. Basically set up your model as in the MCMC manual chapter for multinomial then add on the extra levels as you would anyway. Ignore any error messages from IGLS. Switch to MCMC and tick the cross classified box as in the cross classified chapter in the MCMC manual. ...