I'm an undergrad and my stats knowledge is passable, but the current project we're doing is a bit out of my depth and would appreciate some help in figuring out how to analyse the data.
We are trying to see how "personality" very's by size in cat's eyes (an intertidal snail). We recorded weight and size of our test subjects and did two tests on each to see how long they would hide in their shell after 1): being moved 2): being tapped on their shell
We timed how long it would take for them to open their operculum - a trap-door like seal they use to cover the entrance to their shell - and how long it would take them to subsequently emerge from their shell. We recorded if they hid in their shell, and if they closed their operculum, and recorded if they opened their operculum and if they emerged from their shell. Recording went for 5 minutes max, and some did not open their operculum or emerge at all in that time. If they didn't retreat or didn't open their operculum for the first test, they were not treated in the second test and were given an N/A for that one. If they didn't emerge or open their operculum, they were given an N, these were therefore the most fearful of the subjects, but I don't know how I should incorporate them in the rest of the continuous data.
So just looking for some help on figuring out what statistical tests I should use, any help is appreciated. Also, our professor showed us this site and suggested we use it if we're not confident in our stats, before anyone asks if I'm allowed to ask this here. Any help is appreciated, let me know if clarification is needed, cheers.
First, you have to establish the research question. One might presume the hypothesis to be tested is that snail size is associated with reduced "shyness," the reasoning being that larger snails are less likely to feel threatened. But it could be the other way around: larger snails could be more shy because only the most sensitive individuals survive long enough to grow to a particular size.
In any case, you have a situation where you have multiple measures of "shyness"--some of these are binary variables (did it retreat, did they close their operculum), and some are continuous (operculum opening time, emergence time). Then you have multiple measures of size: you have weight and operculum diameter. Finally, you have multiple interventions: relocation, and tapping on the shell.
All of these combined make for a very complex dataset, for which a meaningful investigation of the research question can be extremely challenging. Moreover, you have relatively few experimental units (i.e., snails), at least what I can see. This is probably your biggest issue with your experiment's ability to draw any meaningful inferences.
A statistical test applied to your data can be extremely simple: for example, perform multiple logistic regression on the binary outcomes using operculum diameter, weight, and intervention type as predictors. Or your test can be more like a traditional linear regression on the continuous outcomes.
Or you could create some kind of composite index for "shyness" based on some kind of scoring system, and test whether snail size is associated with this. The drawback with this approach is that you would need to justify how you constructed such a composite score.
Unfortunately, with just four snails, I sincerely doubt there will be much that can be done to make any meaningful inference, no matter how sophisticated the analysis. With so many measures for shyness, you would probably need at least ten times as many snails in your dataset.