main vignette: minor cleaning

Author: paul-buerkner

inst/doc/brms.ltx

@@ -359,7 +359,7 @@ Trace and density plots of the model parameters as produced by \code{plot(fit4)}
 
 \section[Comparison]{Comparison between packages}
 
-Over the years, many \proglang{R} packages have been developed that implement MLMs, each being more or less general in their supported models. Comparing all of them to \pkg{brms} would be too extensive and barely helpful for the purpose of the present paper. Accordingly, we concentrate on a comparison with four packages. These are \pkg{lme4} \citep{bates2015} and \pkg{MCMCglmm} \citep{hadfield2010}, which are possibly the most general and widely applied \proglang{R} packages for MLMs, as well as \pkg{rstanarm} \citep{rstanarm2016} and \pkg{rethinking} \citep{mcelreath2016}, which are both based on \pkg{Stan}. As opposed to the other packages, \pkg{rethinking} was primarily written for teaching purposes and requires the user to specify the full model explicitly using its own simplified \pkg{BUGS}-like syntax thus helping users to better understand the models that are fitted to their data. However, this may be a rather time-consuming (and sometimes hard to debug) process as compared to the classical \proglang{R} formula syntax used in the other packages. 
+Over the years, many \proglang{R} packages have been developed that implement MLMs, each being more or less general in their supported models. Comparing all of them to \pkg{brms} would be too extensive and barely helpful for the purpose of the present paper. Accordingly, we concentrate on a comparison with four packages. These are \pkg{lme4} \citep{bates2015} and \pkg{MCMCglmm} \citep{hadfield2010}, which are possibly the most general and widely applied \proglang{R} packages for MLMs, as well as \pkg{rstanarm} \citep{rstanarm2016} and \pkg{rethinking} \citep{mcelreath2016}, which are both based on \pkg{Stan}. As opposed to the other packages, \pkg{rethinking} was primarily written for teaching purposes and requires the user to specify the full model explicitly using its own simplified \pkg{BUGS}-like syntax thus helping users to better understand the models that are fitted to their data. 
 
 Regarding model families, all five packages support the most common types such as linear and binomial models as well as Poisson models for count data. Currently, \pkg{brms} and \pkg{MCMCglmm} provide more flexibility when modeling categorical and ordinal data. In addition, \pkg{brms} supports robust linear regression using  Student's distribution, which is also implemented on a GitHub branch of \pkg{rstanarm}. \pkg{MCMCglmm} allows to fit multinomial models that are currently not available in the other packages.
 

inst/doc/brms.pdf

@@ -359,7 +359,7 @@ Trace and density plots of the model parameters as produced by \code{plot(fit4)}
 
 \section[Comparison]{Comparison between packages}
 
-Over the years, many \proglang{R} packages have been developed that implement MLMs, each being more or less general in their supported models. Comparing all of them to \pkg{brms} would be too extensive and barely helpful for the purpose of the present paper. Accordingly, we concentrate on a comparison with four packages. These are \pkg{lme4} \citep{bates2015} and \pkg{MCMCglmm} \citep{hadfield2010}, which are possibly the most general and widely applied \proglang{R} packages for MLMs, as well as \pkg{rstanarm} \citep{rstanarm2016} and \pkg{rethinking} \citep{mcelreath2016}, which are both based on \pkg{Stan}. As opposed to the other packages, \pkg{rethinking} was primarily written for teaching purposes and requires the user to specify the full model explicitly using its own simplified \pkg{BUGS}-like syntax thus helping users to better understand the models that are fitted to their data. However, this may be a rather time-consuming (and sometimes hard to debug) process as compared to the classical \proglang{R} formula syntax used in the other packages. 
+Over the years, many \proglang{R} packages have been developed that implement MLMs, each being more or less general in their supported models. Comparing all of them to \pkg{brms} would be too extensive and barely helpful for the purpose of the present paper. Accordingly, we concentrate on a comparison with four packages. These are \pkg{lme4} \citep{bates2015} and \pkg{MCMCglmm} \citep{hadfield2010}, which are possibly the most general and widely applied \proglang{R} packages for MLMs, as well as \pkg{rstanarm} \citep{rstanarm2016} and \pkg{rethinking} \citep{mcelreath2016}, which are both based on \pkg{Stan}. As opposed to the other packages, \pkg{rethinking} was primarily written for teaching purposes and requires the user to specify the full model explicitly using its own simplified \pkg{BUGS}-like syntax thus helping users to better understand the models that are fitted to their data. 
 
 Regarding model families, all five packages support the most common types such as linear and binomial models as well as Poisson models for count data. Currently, \pkg{brms} and \pkg{MCMCglmm} provide more flexibility when modeling categorical and ordinal data. In addition, \pkg{brms} supports robust linear regression using  Student's distribution, which is also implemented on a GitHub branch of \pkg{rstanarm}. \pkg{MCMCglmm} allows to fit multinomial models that are currently not available in the other packages.