Re: [R-sig-phylo] to log or not to log

2017-01-26 Thread Jacob Berv
This makes sense to me too… thanks for your thoughts!
Jake

> On Jan 26, 2017, at 7:55 PM, Donald Miles  wrote:
> 
> I agree with Ted. The comparison should be how the transformation affects the 
> distribution of the response variable. As Ted mentioned, you should examine 
> the behavior of the residuals to determine whether they conform to the 
> assumptions of the statistical analysis after transformation.
> 
> Donald Miles
> 
> On Thu, Jan 26, 2017 at 7:49 PM, Theodore Garland  > wrote:
> I don't think you can compare models like this for different transforms of
> the dependent variable.  The likelihood, etc., values are not comparable,
> as your results suggest.  But I am sure someone will correct me if I am
> wrong!
> 
> Assuming this is some sort of regression model (i.e., you have one or more
> independent variables), then you CAN look at residuals to see if they are
> better/worse behaved (e.g., approximately normal, no nasty outliers,
> homoscedasticity).
> 
> Cheers,
> Ted
> 
> 
> On Thu, Jan 26, 2017 at 2:25 PM, Jacob Berv  >
> wrote:
> 
> > Dear R-sig-phylo,
> >
> > When analyzing comparative data in a PCM framework, is it be appropriate
> > to use an AIC score to advocate for log transforming input data? ie
> > model(data) vs model(log(data))
> >
> > I’m running a few OUwie models and it’s not entirely clear from the
> > biology whether or not the ‘standard’ log transformation makes sense. In my
> > case, the model(log(data)) results have much lower AIC values relative to
> > the model(data) results.
> >
> > Cheers,
> > Jake Berv
> > ___
> > R-sig-phylo mailing list - R-sig-phylo@r-project.org 
> > 
> > https://stat.ethz.ch/mailman/listinfo/r-sig-phylo 
> > 
> > Searchable archive at http://www.mail-archive.com/r- 
> > 
> > sig-ph...@r-project.org/ 
> 
> [[alternative HTML version deleted]]
> 
> ___
> R-sig-phylo mailing list - R-sig-phylo@r-project.org 
> 
> https://stat.ethz.ch/mailman/listinfo/r-sig-phylo 
> 
> Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/ 
> 


[[alternative HTML version deleted]]

___
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/

Re: [R-sig-phylo] to log or not to log

2017-01-26 Thread Donald Miles
I agree with Ted. The comparison should be how the transformation affects
the distribution of the response variable. As Ted mentioned, you should
examine the behavior of the residuals to determine whether they conform to
the assumptions of the statistical analysis after transformation.

Donald Miles

On Thu, Jan 26, 2017 at 7:49 PM, Theodore Garland 
wrote:

> I don't think you can compare models like this for different transforms of
> the dependent variable.  The likelihood, etc., values are not comparable,
> as your results suggest.  But I am sure someone will correct me if I am
> wrong!
>
> Assuming this is some sort of regression model (i.e., you have one or more
> independent variables), then you CAN look at residuals to see if they are
> better/worse behaved (e.g., approximately normal, no nasty outliers,
> homoscedasticity).
>
> Cheers,
> Ted
>
>
> On Thu, Jan 26, 2017 at 2:25 PM, Jacob Berv  com>
> wrote:
>
> > Dear R-sig-phylo,
> >
> > When analyzing comparative data in a PCM framework, is it be appropriate
> > to use an AIC score to advocate for log transforming input data? ie
> > model(data) vs model(log(data))
> >
> > I’m running a few OUwie models and it’s not entirely clear from the
> > biology whether or not the ‘standard’ log transformation makes sense. In
> my
> > case, the model(log(data)) results have much lower AIC values relative to
> > the model(data) results.
> >
> > Cheers,
> > Jake Berv
> > ___
> > R-sig-phylo mailing list - R-sig-phylo@r-project.org
> > https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
> > Searchable archive at http://www.mail-archive.com/r-
> > sig-ph...@r-project.org/
>
> [[alternative HTML version deleted]]
>
> ___
> R-sig-phylo mailing list - R-sig-phylo@r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
> Searchable archive at http://www.mail-archive.com/r-
> sig-ph...@r-project.org/

[[alternative HTML version deleted]]

___
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/

Re: [R-sig-phylo] to log or not to log

2017-01-26 Thread Theodore Garland
I don't think you can compare models like this for different transforms of
the dependent variable.  The likelihood, etc., values are not comparable,
as your results suggest.  But I am sure someone will correct me if I am
wrong!

Assuming this is some sort of regression model (i.e., you have one or more
independent variables), then you CAN look at residuals to see if they are
better/worse behaved (e.g., approximately normal, no nasty outliers,
homoscedasticity).

Cheers,
Ted


On Thu, Jan 26, 2017 at 2:25 PM, Jacob Berv 
wrote:

> Dear R-sig-phylo,
>
> When analyzing comparative data in a PCM framework, is it be appropriate
> to use an AIC score to advocate for log transforming input data? ie
> model(data) vs model(log(data))
>
> I’m running a few OUwie models and it’s not entirely clear from the
> biology whether or not the ‘standard’ log transformation makes sense. In my
> case, the model(log(data)) results have much lower AIC values relative to
> the model(data) results.
>
> Cheers,
> Jake Berv
> ___
> R-sig-phylo mailing list - R-sig-phylo@r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
> Searchable archive at http://www.mail-archive.com/r-
> sig-ph...@r-project.org/

[[alternative HTML version deleted]]

___
R-sig-phylo mailing list - R-sig-phylo@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/