I am currently doing statistical analysis of a knockout vs control mouse strain, investigating a gene that codes a protein that is hypothesized to be involved in learning and memory.
To do the test, we ran mice in a morris water maze, which involves placing mice in a little pool, the water is made opaque with milk, and a platform that they can stand on is hidden just under the water. Markers are placed at the four cardinal directions, in the form of a star, circle, triangle, and square, so that ideally, the mouse will learn where the platform is after being placed in the pool a few times, and swim there immediately (because like many mammals, they can swim, though they prefer not to)
I am measuring two dependent variables, swim speed and path length. We put the mice in 7 days and measure their progress. I made two figures using Statistica, which were made using ANOVA, which broke them down into strains and had the timepoints (x axis) as days, and y axis as pathlength, and the next figure was the same except y axis was swimspeed.
Statistica allows you to select adding 'standard error', which puts up vertical bars on the time points which correspond to Confidence Interval of 95%. My boss said he wants standard error instead of confidence interval. I was wondering what the difference is. Is standard error traditionally 1 standard deviation while the confidence interval I selected was roughly equal to 2? If I want to change the CI to SE, should I change the CI parameter from .95 to about .66?
I want to check out a few more possibilities, like perhaps there is greater variability within particular mice of a particular strain. As in, perhaps the knockout strain is affected unevenly, and *some* mice are very strongly affected, while others not, what statistical tests would you do this with?
Another thing I want to analyze is the rate of change perhaps, and the degree of change. Perhaps these parameters are different in the different strains, maybe the knockouts learn less quickly even though they might have similar means on the initial days, or vice versa. Any ideas on what statistical tests to do this with?
This one is not a big deal, I already truly truly appreciate your help if you have read up to this point, this one is just if you have any ideas for other analyses that might be of interest, feel free to suggest some, but like I said, that's more my job so please don't feel obligated to help me here (of course you are not obligated to help me *at all*, but you know what I mean I hope).