Search found 49 matches

by Harvey Goldstein
Wed Jan 10, 2018 11:16 am
Forum: Realcom user forum
Topic: calculation of variance explained
Replies: 1
Views: 15220

Re: calculation of variance explained

Dear Lorraine If the multiple imputation has done its job properly, i.e. te assumptions are reasonable (MAR) then you dont need to have the same no of records for null modle and imputed model in order to study the relative amounts ofvariance explained in both cases since you should be getting unbias...
by Harvey Goldstein
Sun Oct 29, 2017 8:17 pm
Forum: MLwiN user forum
Topic: Mlwin supplement: autocorrelated errors in continuous time
Replies: 2
Views: 4827

Re: Mlwin supplement: autocorrelated errors in continuous time

That's right. If you have a binary response you need to move to a multivariate model. see
Barbosa and Goldstein (2000), Quality and quantity, 34, pp323-330
by Harvey Goldstein
Fri Feb 17, 2017 4:08 pm
Forum: Realcom user forum
Topic: How to impute a two-level model with repeated measures
Replies: 5
Views: 25520

Re: How to impute a two-level model with repeated measures

As I understand it, your model is a repeated measures model where you have some responses missing for some individuals at some occasions (months). If you just set this up as a 2-level model then there is typically no problem since the level 2 units (individuals) just have different numbers of level ...
by Harvey Goldstein
Tue Jan 10, 2017 3:17 pm
Forum: runmlwin user forum
Topic: Specifying Weights
Replies: 2
Views: 4597

Re: Specifying Weights

If you only have weights at 1 level then leave the weights at the other level as 'equal'. To use standardised weights just specify the weights column and check 'use standardised weights' in the menu box.
Harvey
by Harvey Goldstein
Thu Apr 28, 2016 10:45 am
Forum: MLwiN user forum
Topic: Constraining the variance to zero.
Replies: 2
Views: 7273

Re: Constraining the variance to zero.

My view would be that all associated covariances should be constrained to zero also.
harvey
by Harvey Goldstein
Tue Feb 02, 2016 12:32 pm
Forum: MLwiN user forum
Topic: Multinomial logistic & VPC
Replies: 2
Views: 5051

Re: Multinomial logistic & VPC

What you suggest seems quite sensible to me and would apply whether you used IGlS or MCMC estimation. There is an analogous procedure for ordered data. You also might consider partitioning the covariances (correlations).
by Harvey Goldstein
Thu Jan 07, 2016 10:11 am
Forum: Realcom user forum
Topic: Realcom-imput / imputation model with missing values in explanatory variables
Replies: 3
Views: 15840

Re: Realcom-imput / imputation model with missing values in explanatory variables

You can impute several variables at a time - but it does not use chained equations. Explanatory variables in imputation model must have no missing values - if any have make them responses in imputation model.
by Harvey Goldstein
Mon Sep 28, 2015 5:41 pm
Forum: MLwiN user forum
Topic: cross-level interaction with Level-1 moderator
Replies: 4
Views: 21078

Re: cross-level interaction with Level-1 moderator

If I understand your model it seems to me that what you call the level 1 classification (am/pm) is actually a predictor variable and if you treat it as such then your analysis is straightforward, isn't it?
by Harvey Goldstein
Fri Sep 25, 2015 1:32 pm
Forum: MLwiN user forum
Topic: cross-level interaction with Level-1 moderator
Replies: 4
Views: 21078

Re: cross-level interaction with Level-1 moderator

Ignore previous post - should have read that it is inappropriate to predict a level 2 using a level 1 variable - this seems implicitly what is happening in your model.
by Harvey Goldstein
Fri Sep 25, 2015 1:26 pm
Forum: MLwiN user forum
Topic: cross-level interaction with Level-1 moderator
Replies: 4
Views: 21078

Re: cross-level interaction with Level-1 moderator

It typically doesn't make sense to predict a level 1 outcome using something defined at level 2. It can be specified but what happens is that such a model effectively induces dependencies among level 1 residuals. Also not easy to interpret.
Harvey Goldstein