[R-sig-ME] toxicology dietary subchronic feed/weight analysis

[Steve Denham]

Really no need for imputation if feed intake is re-parameterized as cumulative feed intake to each weight date.  Granted this greatly reduces the number of observations (28 to 7) but it aligns each.
I do like the idea of a hierarchical model in this case as the best way to handle a continuously measure covariate. Steve Denham
Director, Biostatistics
MPI Research, Inc.
 
      From: Thierry Onkelinx <thierry.onkelinx at inbo.be>
 To: "Gosse, Michelle" <Michelle.Gosse at foodstandards.gov.au> 
Cc: "r-sig-mixed-models at r-project.org" <R-sig-mixed-models at r-project.org> 
 Sent: Monday, May 11, 2015 3:04 AM
 Subject: Re: [R-sig-ME] toxicology dietary subchronic feed/weight analysis
   
Dear Michelle,

The best solution for the next study is to measure weight as the same
frequency as the intake ;-)

I see two possible solutions for your current study. 1) Use multiple
imputation on the missing values of weight based on your weight model. 2)
Go Bayesian an fit both the weight and intake models simultaneously use a
hierarchical model.

Best regards,

ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and
Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey

2015-05-10 23:32 GMT+02:00 Gosse, Michelle <
Michelle.Gosse at foodstandards.gov.au>:

> Hi all,
>
> Still continuing on the dataset I have for the effect of two toxins on
> rats, I managed to fit very nice linear mixed effects models for the
> effects on the toxins on body weight over time, using weight = toxin * time
> + (time + 1 | subject.ID) as per Steven Pierce's suggestion.
>
> I didn't need to get into a nonlinear mixed effect model as the results
> were very nice staying within a linear framework.
>
> The final analysis I need to do is to examine feed intake and see how this
> is associated with body weight, time, and drug. I have feed intake measured
> daily, so 28 intake data points per subject, but only 7 weight data points,
> so while I have repeated measures for both, I do not have a fully linked
> intake-weight series.
>
> The research question is whether the toxin influenced feed intake (feed
> palatability issue). I'm interested in intake slopes/partial slopes, but
> obviously body weight should be the main driver of feed intake (heavier
> rats eat more).
>
> I'm thinking of an analysis similar to: intake = toxin*body weight*time
> (time +1|subject.ID)
>
> But I'm not sure I have the sample size to do a three-way effect, and I
> don't know that this is the correct model specification given that I have
> weight data which is not missing at random - all the rats were measured on
> specific days such as Day 1, Day 4, Day 7.
>
> Has anyone worked with a similar dataset to advise what model to fit.
>
> Cheers
> Michelle, note: I do not work Fridays
>
>
> **********************************************************************
> This email and any files transmitted with it are confide...{{dropped:12}}
>
> _______________________________________________
> R-sig-mixed-models at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models
>

    [[alternative HTML version deleted]]



_______________________________________________
R-sig-mixed-models at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models


   

	[[alternative HTML version deleted]]