Just to check I understand this.
- You are exporting all the HTTP requests.
- Deleting the ones with a response time of 0ms (yes, it's milliseconds)
- taking an average.
- summing the 3 averages for the 3 machines.
What do you think this number is telling you? It sounds pretty confusing to me.
What I do, is set up a timer around every requst for a page, e.g. I have a timer called homepage, that starts before the request for the homepage, and stops once all the images, stylesheets etc. have downloaded.
This number gives me the user response time.
As for your test numbers being different to what you expected, I would be first suspicious of the test.
Can you run the test several times under the same conditions and get the same results? If not then you need to find out why.
You can't really do any comparisons until you know you can run the same test twice and get the same results.
My advice would be to use page timers, and get to a point where you can run the same test consistently time after time, then you'll be able to do some comparisons.