I did a test with my drivers window up/down/up on three consecutive drag runs.
I posted these trap speeds:
1. window up - 108.14 MPH
2. window down - 107.68 MPH
3. window up - 108.69 MPH
All three runs were between 13.8-14 seconds
So the drag created by the windows does play a measureable role but I would only say by maybe 0-1.5 mph or so depending on if all the windows were down versus one or two.
These times of mine will be topped when I get my new engine project back together of course!