[R-sig-ME] mixed effect models where time ordering is important

[Emmanuel Curis]

I'm confused by the second model, the LogDose = 0 if control seems
strange to me. How does it distinguish, in the intercept term, the
« control » (0 + 0) and « dose = 10^-1 » (1 + -1) groups? And if the
intercept is monotonic with dose, wouldn't be an annoyance to have the
« dose = 0 » just somewhere between other points, depending on the
exact doses used (and the log basis and the unit used...)?

On Wed, Apr 22, 2015 at 09:16:09PM +1000, Steve Candy wrote:

« To model the dose response component and assuming a common time response
« shape on the log scale you might want to consider; 1). specify a dummy (1,0)
« for Control (0) vs Dosed (1) as "NotContr_d" (i.e. the intercept gives the
« control intercept), 2). LogDose  is set to zero for the control and is the
« log of the actual dose (low, medium, high) for the dosed treatments, and
« define Control_f <- as.factor(NotContr_d)  then 3). replace the above gamm
« with
« 
«  
« 
« gamm_01 <- gamm(formula = log(Weight) ~ NotContr_d + LogDose + s(time,
« by=Control_f, bs="cr"), data=data, random=list(subject=~1),  
« 
«      correlation=corCAR1(form= ~ time | subject))
« 
«  
« 
« The log transforms used above are just suggestions which can be compared to
« using raw Weights or Doses.
« 
«  
« 
«  
« 
« > Hi all,
« 
« > 
« 
« > I have repeated measures weight data on rats who were in a 28-day toxin
« study. I have one control group and the toxin was administered at one of
« three doses (low, medium, high). This is not a cross-over design, so (for
« example) the rats who were in the low dose group always got the low dose
« over the course of the study.
« 
« > 
« 
« > The rats were not fully grown when the study started. Body weights were
« measured every fourth day.
« 
« > 
« 
« > The interest is in seeing if the toxin has an influence on body weight. I
« am looking at using lmer to analyse this data, however I am unsure how to
« handle the ordering of time, as this will be correlated with increasing body
« weight.
« 
« > 
« 
« > If I did not have to worry about time ordering, I thought this model would
« work:
« 
« > 
« 
« > Weight ~ dose + (1|subject)
« 
« > 
« 
« > The doses are being treated as fixed effects as I am not wanting to
« extrapolate the impact of dose beyond what was administered in the study.
« 
« > 
« 
« > I was wondering if the appropriate model for my data would be:
« 
« > 
« 
« > Weight ~ dose * time + (1|subject)
« 
« > 
« 
« > However, time is measured as days from initial dose administration (e.g.
« day 1 = first day of dosing). While the rats are all very similar in age, I
« do not believe they were all born on the same day, and so I am unsure about
« time as a proxy for age (assuming an intercept in the model). And day is
« measured discontinuously (every fourth day). I feel that omitting day will
« remove one obvious explanatory variable from the model, which may bias the
« results as well as producing a model that poorly fits the data.
« 
« > 
« 
« > I have tried to find an example of a toxicology study that uses a mixed
« effects model in R on repeated measures, that specifies the model. I have
« been unable to locate one.
« 
« > 
« 
« > I would appreciate any advice/recommendations on how to handle this data.
« I have already advised that a series of separate ANOVAs are not
« statistically defensible given that the weights are likely to be
« auto-correlated and the statistical analysis needs to account for this.
« 
« > 
« 
« > Cheers
« 
« > Michelle, note: I do not work Fridays
« 
«  
« 
« Dr Steven G. Candy
« 
« Director/Consultant
« 
« SCANDY STATISTICAL MODELLING PTY LTD
« 
« (ABN: 83 601 268 419)
« 
« 70 Burwood Drive
« 
« Blackmans Bay, TASMANIA, Australia 7052
« 
« Mobile: (61) 0439284983
« 
«  
« 
« 
« 	[[alternative HTML version deleted]]
« 
« _______________________________________________
« R-sig-mixed-models at r-project.org mailing list
« https://stat.ethz.ch/mailman/listinfo/r-sig-mixed-models

-- 
                                Emmanuel CURIS
                                emmanuel.curis at parisdescartes.fr

Page WWW: http://emmanuel.curis.online.fr/index.html