@misstyrabanks,
misstyrabanks wrote:This is easy to visualize on a graph, but how can I say statistically that there is a trend?
You draw separate regression lines through the "before" and "after" data. Then you calculate the probability of producing the "after" data by chance, given the "before" trend line and the "before" standard deviation. If the probability is smaller than a certain value---medical trials would use 5%---you have a statistically significant change
misstyrabanks wrote:I have no control group, so there is nothing to compare this class to....
Then you're screwed. Even if you
have a change in trend, you have no empirical information on what caused it. Run a new experiment, and remember to include a control group to which the measure was not introduced. Make sure the control group is as similar as possible to the test group in all ways, other than this disciplinary measure, that may affect the outcome.
I'm sorry to be so harsh about i, but no amount of fancy analysis can extract information from your data if it isn't in there to be extracted. And your experimental setup did not produce the information you're asking about.
Stumble It!
Tweet This
Bookmark on Delicious
Share on Facebook
Share on MySpace