[R-sig-ME] nested mixed model with covariate and missing data

[ONKELINX, Thierry]

Dear Carlos,

Use baseline as an offset factor. That is equivalent of subtracting the baseline from the activity prior to analysis.
(1|subject/neuron) gives you a random effect for both the subject level as the neuron within subject level.
Don't use (1|time) in combination with time as a fixed effect unless time is continuous and you have a fairly large number of different timepoints.
Missing data is not a problem as long as it is missing at random.

So your model looks like this
activity ~ offset(baseline) + treatment*time + (1|subject/neuron)

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
+ 32 2 525 02 51
+ 32 54 43 61 85
Thierry.Onkelinx op inbo.be
www.inbo.be

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


-----Oorspronkelijk bericht-----
Van: r-sig-mixed-models-bounces op r-project.org [mailto:r-sig-mixed-models-bounces op r-project.org] Namens Carlos Gias
Verzonden: woensdag 2 januari 2013 13:48
Aan: r-sig-mixed-models op r-project.org
Onderwerp: [R-sig-ME] nested mixed model with covariate and missing data

Hi,

I am new to mixed models and not sure how to design a model for my data to use with the lmer function.


I am trying to study the effect of a drug (treatment) in subjects along time. For every subject we measure activity from a (variable) number of neurons at different time points. Therefore, the neuron measurement is nested within subject. I would also like to use the baseline activity measurement as a covariate for each neuron. This is a small sample of the dataset:

subject treatment neuron baseline time activity
1 1 1 3.06 1 7.02
1 1 1 3.06 2 6
1 1 1 3.06 3 9
1 1 2 3 1 5
1 1 2 3 2 6
1 1 2 3 3 9
2 2 3 4.77 1 3
2 2 3 4.77 2 2
2 2 3 4.77 3 1
3 2 3 2.14 1 2.03
3 2 3 2.14 2 2
3 2 3 2.14 3 1.5

I was wondering if the following would be an appropriate model.

activity ~ baseline + treatment*time + (1|subject:neuron) + (1|time)


I am also wondering how I could deal with the problem of missing data. Is there a function/package that could be helpful in this design?

I hope you can help.

Best regards,

Carlos
        [[alternative HTML version deleted]]

* * * * * * * * * * * * * D I S C L A I M E R * * * * * * * * * * * * *
Dit bericht en eventuele bijlagen geven enkel de visie van de schrijver weer en binden het INBO onder geen enkel beding, zolang dit bericht niet bevestigd is door een geldig ondertekend document.
The views expressed in this message and any annex are purely those of the writer and may not be regarded as stating an official position of INBO, as long as the message is not confirmed by a duly signed document.