What model should I use??? I have daily repeated measures data. It has multiple dependent presence absence variables, (of which, I have collapsed into a CA with continuous variables of CA1 & CA2....or I can choose to not collapse the data and leave each dependent variable as a binary variable). My independent variables are weather data that was collapsed into a PCA (or I can use each individual weather measurement), 2 categorical variables, and time (or a covariate of time). I am wanting to create a model that allows me to solve for time (or the covariate of time). I have been looking at SEM latent growth models, logistic regression, time series, or mixed regression models. The software that is available to me is SAS, SPSS, PCord, and Canoco.

"Gung":More information: CA: correspondence analysis. Presence and absence of insects on decomposing organisms that were measured periodically, mostly daily for the first 2 months and then periodically after that as insect presence greatly decreased. Two organisms were put out in the spring at different times, then some more in the summer, then fall, then winter. The time component that I am wanting to predict is how long that the organism has been decaying...so once I have a model of my data, I could put in all of the measurable variables of a decomposing organism that I find, and solve for time (decomposition day: how many days that the body has been decomposing). By covariate of time, which I think would actually be a better predictor is accumulated degree days (ADD). This is the sum of the daily min and max temp divided by two since the start of decomposition. This represents the accumulated heat units of the organism, and has been shown to influence different species of insects. EX) typically, flies are the first to a decomposing organism, so at smaller ADD values. Later, beetles appear at large ADD values. All of my independent variables have had literature published that suggest that they influence insect occurrence. As a second part, I would like to look at the rate of change in community structure (or insect occurrence) over time...or meaning as decomposition progresses. Thank you so much for any advice!!

  • $\begingroup$ What do you mean by CA? I'm not sure what that stands for, is it correspondence analysis? Is this 1 long time-series, or many short series running simultaneously? By "covariate of time", do you mean a function of time? Are you only / primarily interested in estimating the beta for time? $\endgroup$ – gung - Reinstate Monica Oct 25 '12 at 0:42
  • $\begingroup$ I edited my description by adding more information. $\endgroup$ – Michelle Oct 25 '12 at 5:58

I now think that my data may be applied to a Generalized Estimating Equations (GEE) model. This is an extension of a generalized linear model that allows me to account for repeated measures. Does anyone disagree?


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.