Interesting--that increased the variance again, but also pulled up the mean enough that we have a p-value of .0348; or 3.5%. If we'd done the experiment under rigorous conditions and documented it thoroughly, that'd be good enough for journal publication. Well, also if we'd used a more appropriate statistical test and had a lot more data.
Given those results, I just might have to get me some airtabs.
Right now, I'm recording the mpg of every highway-only trip I take, which is a pretty decent amount since I live next to a freeway onramp. As long as traffic allows me to drive ~70mph most of the way, I record the mpg right before I decelerate for the offramp. I figure that'll help isolate the effects of drag as much as possible.
If I do buy airtabs, I will write my conclusions and methods up in a post here. I will also probably point to that post in an argument against the useless A-B-A test method that somehow rules the ecomodder forum
