How to improve this auto. trans. acceleration test

[savoF3]

Today I tried running an acceleration test in the minivan with automatic transmission.

The method:
Engine is warmed up from previous driving.

Reset manufacturer's avg economy display and punch in a throttle setting from a dead stop at a stop sign turning right and holding constant throttle to 40mph noting TPS setting from scanguage --

Then coast engine on to the same landmark each time.

Note mpg from manufacturer's display while rolling past landmark.

Easy to feel different levels of acceleration by butt gauge. Unfortunately I decided that since trying to maintain a constant 40mph on a constant slope and coasting should BOTH use constant GPH I would just coast because it's easier. But I after running it realized the results would be skewed by the different times coasting at at different average speeds.

Other sources of error:
1) The manufacturer's display updates every 2 seconds (same as Scanguage on "normal" setting, right?) and each run is fairly short.

2) I didn't record coasting time or final coasting speed so I can't mathematically correct this part.

3) I started with engine hot so I could do all the runs in a row without this variable. BUT I'd really like to know if it makes a difference starting cold, especially from this particular stop.

I didn't worry about rpms etc. since it isn't a stick. I just want to get a good idea of how the acceleration should feel while keeping my eyes on the road.

Anyway here's the results such as they are:

TPS  MPG
24   17.9  this feels pokey
28   18.4  this feels about normal "moderate"
32   20.0  \__ these feel like accel. "smartly" but not peeling out
33   20.0  /
39   19.2  FAST, just under spinning at the stop, but not 
               continuing to push the pedal down to WOT
               after pulling away.

I'm going to try keeping a log of individual cold starts to 40mph and then maintaining 40mph to my landmark over the next several days on normal trips. I'll note temperature as well.