Repeated measures ANCOVA

Question

I apologize in advance, my knowledge in statistics is pretty basic. I wonder if a repeated measures ANCOVA would be a good approach in my case.

I have two species of fish and I want to test if their activity response to a contaminant is different. So I will measure repeatedly over time (up to 3 hour-exposure) the activity of 5 individuals of each species in the presence of the contaminant. I guess exposure time will be the covariate. Activity is measured with telemetry tags that deliver an activity read every 5-10 minutes (so time points are not exactly the same for each individual fish).

Since I do not know whether there is a linear relationship between exposure time and activity, would be adequate to use a repeated measures ANCOVA?? And how should individual fish-effects assessed?

Answer 1

I know this is an old question but I am answering it for it may lead to some useful comments.

The data can be shown as in following table (with only top 4 rows shown).

It can be analyzed with regression using following "formula":

ActivityDifference ~ exposureTime * species + ActivityBasal

This will show relation of activity with exposure time taking into account baseline activity of species. It will also show if there is any interaction between species and activity following exposure to contaminant. This is very much plausible that one species is more sensitive to contaminant exposure than other species.

Alternatively:

ActivityDifference ~ exposureTime + species + ActivityBasal

Above can be used if an interaction is not thought to be important. This simplifies the procedure and the analysis.

Very often, "Percent change", i.e. 100* ActivityDifference / ActivityBasal is used. Since this takes into account basal level, the formula can be probably be simplified to:

ActivityPercentChange ~ exposureTime + species

or:

ActivityPercentChange ~ exposureTime * species

This last approach (with interaction) may be the best.