[R-sig-ME] Multiple binary responses per ID and time
Lize van der merwe
lizestats at gmail.com
Mon Mar 21 17:29:40 CET 2016
Thank you for the quick response, Thierry.
Yes, unfortunately the response n_it_j does not equal the predictor n_it_j.
I will go the INLA route.
From: Thierry Onkelinx [mailto:thierry.onkelinx at inbo.be]
Sent: Monday, 21 March 2016 17:30
To: Lize van der merwe
Cc: r-sig-mixed-models at r-project.org
Subject: Re: [R-sig-ME] Multiple binary responses per ID and time
So the response for individual i at time i is (x_it, n_it). Does the predictor has the same amount of trials as the responses (y_it, n_it)? If so, do you have information on the n_it Bernouilli trials of both the response and the predictor? If that is the case then you can model the individual Bernoulli trials.
If you don't have the information at that detail, then you have to turn the binomial predictor into a proportion. With a Bayesian hierarchical model you can first model the predictor and then uses this modelled proportion as a predictor for the response. There is an example in the INLA FAQ: http://www.r-inla.org/faq#TOC-Can-I-have-the-linear-predictor-from-one-model-as-a-covariate-in-a-different-model-
ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
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
2016-03-21 15:33 GMT+01:00 Lize van der merwe <lizestats at gmail.com <mailto:lizestats at gmail.com> >:
Please advise. I cannot get my head around modelling this data.
Study involves 200 individuals with several (not always the same number)
dichotomous outcomes, at 10 different times. The predictor also has several
(not the same as each other, nor the same as what the individal has at that
time-point) dichotomous outcomes for the same individuals at the the same
timepoints. There are time-level covariates and also individual level
covariates to include.
How do I model these? Not even sure how to lay out the data.
Binomial pair (x,n) outcome, for each individual and each time and another
binomial pair for the predictor?
Lize van der Merwe
R-sig-mixed-models at r-project.org <mailto:R-sig-mixed-models at r-project.org> mailing list
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models