www.phidot.org

I was wondering if anyone could tell me whether this makes sense or not:

If I want to compare whether a drought had an effect on survival, does it make sense to compare a model with a linear trend on survival across all sampling occasions, to a model where I fit a separate trend to the pre-drought and post-drought model estimates? Thus, if the single trend model is better than the multiple trend model, there is no support for a drought effect.

Eurycea wrote:I was wondering if anyone could tell me whether this makes sense or not:

If I want to compare whether a drought had an effect on survival, does it make sense to compare a model with a linear trend on survival across all sampling occasions, to a model where I fit a separate trend to the pre-drought and post-drought model estimates? Thus, if the single trend model is better than the multiple trend model, there is no support for a drought effect.

The is analogous (very) to the ubiquitous 'dipper flood analysis' which is presented in much detail in chapter 6. I'd suggest you study that chapter carefully.

So you are saying just add drought as a categorical variable in the model, such as the dipper flood example, right? Hmmm. yeah that probably makes more sense, doesn't it? Thanks.

Eurycea wrote:So you are saying just add drought as a categorical variable in the model, such as the dipper flood example, right? Hmmm. yeah that probably makes more sense, doesn't it? Thanks.

If you read chapter 6, you will see there are at least a couple of ways to handle this -- in MARK. But, the question of categorical or continuous covariate is something *you* need to think about -- the flood example presumes that a given year can be categorized as either flood or non-flood; i.e., two levels of a factor by which each year is classified (so, single classification level ANOVA with 2 levels of the classification factor). If you believe that any drought year is the same as any other drought year, such that you simply dichotomize between drought and non-drought (making the similarity with the flood, non-flood dipper example very obvious), then what you do next is straightforward. If you think there are interesting temporal relationships mediated by whether or not there is a drought, you'd do other things.

From the sounds of it, you're best advised to pretend your organism is a dipper, and simply substitute 'drought' for 'flood'.