# Which Statistical Device do I use

For my University coursework, my lecturer asked us to send Tracert packets to a designated nameserver. We are then to record the response time. We are to take two readings per 24 hour period. One during off peak times, one during peak times. I first thought to use a Chi Square test to test if there was a relationship between these numbers. The issue I ran into was because the table looked like this:

          Expected      Not Expected
Peak
Off Peak


Where Expected peak packets were Peak packets with longer round trip times than it's off peak counterpart, and vise-versa for expected off peak packets. The issue is that Expected is dependant upon off peak, and these aren't two completely different sets of data. They're members of the same set!

So my question is what method should I be using to test for a trend with these values?

-