How long to set a baseline?

[Floordford]

Ive got a gauge on the way and I was wondering, how long or how many tanks is a good representation of your baseline MPG? Id like to get that before I start messing with ideas.

[Bill in Houston]

Depends how consistent your results are. If you get 15.3, 15.0, 15.2, then that is a good baseline. If you get 13, 17, 15, well, not so much. The more consistent you are beforehand, the easier it is to see, and to statisticall prove, a change.

[gone-ot]

...mathematically, you need three data sets to average in order to obtain a meaningful "average" value (think of them as: one-too high, one-real, one-too low).

...statistically, you need 30 data sets to determine any (if they exist) trends or relationships.

[Bill in Houston]

look up "delta test" and figure it out for yourself. the thought of you putting in 30 bad tanks just because old tele man says to just kills me.

[gone-ot]

...uh, a "baseline" is a 'starting point' from which to compare 'new' data against, it's NOT used for trendline analyses!

...it's the "starting" line of a race, not the one-foot graduation marks around the field!

...I understood that he wants to establish his "starting" point mpg, so that he can accurately determine his ±mpg results, ie: what "worked" (produced +mpg) and what "didn't work" (produced -mpg).

...to determine if his rate of changes (ie: trendline!) is linear, curvilinear, etc., now *that's* when he'd need about 30 data sets.

[Bill in Houston]

so what would your answer be to his actual question?

[gone-ot]

Originally Posted by Bill in Houston

so what would your answer be to his actual question?

Originally Posted by Old Tele man

...mathematically, you need three data sets to average in order to obtain a meaningful "average" value (think of them as: one-too high, one-real, one-too low).

...statistically, you need 30 data sets to determine any (if they exist) trends or relationships.

...three for baseline, more (30) if he plans on making analyses or predictions (look-up student-t approximation to normal distribution)

...what he intends to do with the data in the future is the determining factor.

[euromodder]

Originally Posted by Floordford

Ive got a gauge on the way and I was wondering, how long or how many tanks is a good representation of your baseline MPG? Id like to get that before I start messing with ideas.

It might already be too late for a baseline.
The fact that you're registered here, is most likely already influencing your driving behaviour.
Seasonal changes can be rather large as well.

You probably have some idea of your mileage ?
Use that, and start messing right away
You'll see the MPG increase over time.
If you want to, you can then undo one of the mods and see how it influences your mileage.

Chances are it won't.
Removing my grille block didn't show up in the fuel log.

[Fat Charlie]

Originally Posted by euromodder

Removing my grille block didn't show up in the fuel log.

Did you reinstall it?

[Floordford]

Wow, im surprised at that. I hope I have different results. I plan on doing an upper grille block off, fog light area smoothing, headlight gap sealing and mirror removal.

My thoughts about 30 tanks is that it might spill over into the "winter blend" that I guess gas stations switch to. Don't know if that would effect anything.

Yes my driving habits have changed but I can easily go back to how I was driving before. I drive a Mustang GT and it is easy to wind it up to 3k before a gear change. Lately ive been shifting around 1,500-2,000. Plus skipping gears, neutral going down hill and even killing the engine in the neighborhood while using 5th gear with the engine off so I can still have power steering. Also doing 55-60 in 70mph highway areas. 5mph under most speed limits also. And because its free I took off my antenna and I keep my wipers in my car unless ran is predicted.