1. ## Before/after surgery comparison

Hi!
I am currently doing experiments on song birds looking at pre- and post surgery data. I have 10 male birds each singing at least 100 songs both before and after a muscle surgery. What I would like to be guided with is how I set this up statistically - how to properly investigate for possible significant changes in fx amplitude, frequence and so on.

10 males
100 songs from each male before surgery
100 songs from each male after surgery

Check for differences in amplitude (dB), frequency (kHz), duration (ms/s) and more.

Thank you

2. ## Re: Before/after surgery comparison

What is the hypothesis, differences in values will not equal zero?

3. ## Re: Before/after surgery comparison

Yes, indeed, the hypothesis is that the differences in values will not equal zero. I would like to test for a shift in the mean of the normal distribution, and I would like to test for each bird because of the variation that is between birds - meaning I want to look at the hypothesis for one bird at a time, not collecting all "bird values" for e.g. amplitude in one box.

4. ## Re: Before/after surgery comparison

I'm thinking a simple t-test as I only have two samples with 100 values for each (before and after) and 10 observations (10 males) - consider that a question...

5. ## Re: Before/after surgery comparison

Oh, males means bird, so you have 10 unique birds or 10 groups of birds?

So the 100 pre and 100 post measures are for the same bird or a group birds for a specific type?

And you have three variables of interest you want to compare differences between the pre and post measures on, so 200 amplitude, frequency, and duration values?

6. ## Re: Before/after surgery comparison

Yes, I'm gonna have data from around 10 male birds (females don't sing), and yes I have 10 unique birds - NOT groups of birds.
So, I wanna have a look at what happens when comparing 100 songs from a male BEFORE the surgery with 100 songs from the SAME male AFTER the surgery, and then do that 10 times (one time for each male). I should mention that it could well be e.g. 120 songs before surgery that I would like to test against e.g. 160 songs after surgery!

I have different variables that I want to look at but the three mentioned (amplitude, frequency, duration) might very well be of interest.

In total, I do the song recording for one male at a time, and I want to test for possible significant differences in the mentioned variables (and some more), one variable at a time. So, when I figure out how to test one variable (e.g. amplitude before and amplitude after) all I have to do is change it to the values of another variable of interest (e.g. frequency before and frequency after). Finally, for each variable, I will have a look at whether the results are consistent in each bird or not.

7. ## Re: Before/after surgery comparison

SOLUTION: I'm gonna change my approach a bit and instead of looking at birds individually I will do a paired t-test, which is possible if I take the average of all values before surgery for Bird 1 and compare with the average of all values after surgery for the same bird, and then do that for 10 birds. I don't have many "observations" (birds) but each observation builds on a large amount of data, which I think justifies the statistical aspects.

End.

8. ## Re: Before/after surgery comparison

You are going to have weak power which I am sure you know, but you should consider how this influenced the results if you don't reject the null (and be prepared to defend this part of the design).

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts