R2MLwiN news archive
04 May 2016: A pre-print of R2MLwiN article available
If you wish to cite the article (available here), please do so as:
- Zhang Z, Parker RMA, Charlton CMJ, Leckie G, Browne WJ (in press). R2MLwiN: A Package to Run MLwiN From Within R. Journal of Statistical Software.
12 Apr 2016: The latest version of R2MLwiN (0.8-2) has been released on CRAN.
This includes the following new features (and also note bug fixes):
- Use "get" function for clearing "foreach" package globals and updating the formula to satisfy CRAN checks.
- If a denominator is not specified generate a column of ones and use that instead.
- Run multiple MCMC seed example as parallel chains.
- Add group size summary information for non-missing data to the hierarchy slot.
- Return imputed datasets as a list of dataframes (in the imputations slot). This matches the input required by the mitools package.
- Add option to turn on simple notation when in debug mode.
- Turn off reserved column overwrite warning in generated macro.
- Allow specification of default gamma prior parameters.
- Check data for Binomial, Poisson and Negative-binomial responses to ensure it is appropriate.
- Add S3 wrappers to allow functions such as linearHypothesis to work again.
- Add an option to allow saving the MLwiN worksheet to a file after the model is run.
- Add support for Negative-binomial in MCMC (requires MLwiN version 2.33 or higher).
- Add support for multiple MCMC chains with MLwiN.
10 Aug 2015: A bug in the update function can result in the intercept being removed.
Bug discovered when using generic update function to modify specification of formula resulting in the intercept being inadvertently removed. See the bugs page for further details.
21 Apr 2015: The latest version of R2MLwiN (0.8-1) has been released on CRAN.
The latest version of R2MLwiN (0.8-1) has been released on CRAN. Note that, as detailed on the bugs page, the MCMC options "hcen", "paex" and "resi.store.levs", as well as the IGLS "reset" option, now use level numbering as defined by the R2MLwiN command, as opposed to MLwiN, so please bear this in mind if you have used these options in previous versions of R2MLwiN. Also note that version 0.8-0 (now corrected in 0.8-1) had a typo in the help file for runMLwiN, namely in the table detailing the options for the Formula argument in the runMLwiN help file. The table indicated that the correct way to specify the random part of the model was via (<L2>|1), (<L3>|1[<common>]), etc.; this was not correct, the level ID (<L2>, <L3>, etc.) should be on the right of the vertical bar, i.e.: (1|<L2>), (1[<common>]|<L3>), etc. Apologies for any confusion this caused.
04 Mar 2016: The latest version of R2MLwiN (0.8-0) has been released on CRAN.
The latest version of R2MLwiN (0.8-0) has been released on CRAN. A number of wide-ranging changes have been introduced in this version, including a new model-fitting syntax more in keeping with that conventionally used in R. Whilst back-compatibility has been maintained, the old syntax is deprecated and users are therefore encouraged to use the new syntax which allows for models to be more parsimoniously specified. For more details, see the help files and demos distributed with R2MLwiN (e.g. typing ?R2MLwiN returns the R2MLwiN-package help file which summarises the main changes).