tc

That's exactly why you need a number of tanks. All the "noise" - wind, drafting, tire pressures, fuel quality, etc. will cancel out then when the numbers are compared properly. The smaller your sample size, the bigger those "random" effects can be.

How the statistics work is to compare the mean and standard deviation and essentially look for overlap.

Let's assume for a second that my hypothesis (that there is no difference) is correct. You say hubs locked, you get an average over many tanks of 13.5 MPG. Let's say that the standard deviation is 1 MPG. Typically, a normal distribution would say that 99.5% of the time then, you will get between 10.5 MPG and 16.5 MPG. So, how do you know that the 14.8 MPG was due to the hubs, and not just a "good", but normally to be expected, tank's worth? You don't.

briholt

I hear what you are saying and in the long run we'll see who's right. You are correct in your statistical analysis, but you are forgetting to include the experimental factor of control. I'm saying that there is less "noise" wind, drafting, tire pressures, fuel quality, etc., by taking several smaller samples than larger ones, and therefore I've got more control over my 'samples.' And to dove tail this into your points about standard deviations, #1, you have no idea what the standard deviation is, and #2, having higher internal validity tends to reduce those very standard deviations.

It sounds like we are arguing over external and internal validity. You are arguing for more external validity, I'm arguing that I've got high internal validity.

aviator

I can say personally after driving several 1000kms in both states [locked/unlocked] over a variety of roads/wx/temps/loads/seasons (with a 22r/5spd/4wd IFS/longbox/no tonneau or cap/tailgate both up and down) the difference was in the neighborhood of 1mpg. Therefor I feel safe to say that the differences with an Xtra cab or 4runner both being slightly heavier would be slightly worse up to maybe 1.4mpg diff. Remembering that for 3.0 drivers the mileage is slightly worse to begin with the differences would still be of the same magnitude more or less.

tc

Internal validity is only accurate if you know, and can control, ALL the sources of variation. Since this is usually unlikely, and difficult/impossible to actually execute IF you do know all the sources, external validation, allowing the "real world" variability to be present in both sets of samples, generally to be more statistically reliable.

Anyway, control all the variables or none of them, post up several tanks with the hubs locked and unlocked, and let's run the statistics and see what the REAL difference is. I can about guarantee you're not going to get P<0.05 (or >95% confident there is a real difference)

Teuf

Why would you keep wheels locked? I could see locking them when there may be a hint of needing them, otherwise?

ToyotaDude

I check my fuel mileage every time I fill up and I try to keep my hubs unlocked as much as possible. My 4x4 is an 86 22r, there was one fill up I went around this winter with my hubs locked I got 11.8, just in town. Unlocked I average around 12.5. On the highway unlocked, since i dont drive on the highway with my hubs locked I get around 18.

briholt

You do realize we are talking about a truck here, and not the cure for cancer? There are many way to control variables, and one of the ways is to limit them by treating samples as similarly as possible. You don't have to list every single variable nor it's impact to control them. In fact, most research is done by my method because as you can imagine, listing and controlling every single variable and it's influence is a daunting task. But hey, if you want to measure the gravitational pull on your truck, making it lighter, during full moons and calculating that into your mileage, go for it.

By using short trips over the same route, day after day, I'm canceling out most variables (wind speed will be very consistent since I don't get up to high speeds and in the NW we don't have many wind storms; I drive the same route with the same stop signs and pavement and so on). My point is that you are assuming a wide range of variance when you say that the standard deviation is around 1 mpg from tank to tank. For my short trips I consistently get an average between 13.3 and 13.7.

And regarding your 'real' comment, the only 'real' mileage I get is completely dependent on whether I drive this route or other routes. The more varied the routes I drive, the more your point of view becomes 'real.' But the more I drive my route, the more 'real' my view becomes.

briholt

Good question. For me, I discovered that the previous owner didn't reassemble the passenger hub correctly. So even though it was set to free, the CV spun with the tire.

Teuf

Ah well that would do it, If you said that in the beginning, sorry, I should have read back.

aviator

To answer your ? Tuef driving in winter weather where road conditions can change from clear and dry on a main road to 1' deep snow on a side street with the turn of a corner. Or clear to blinding snow in the space of a mile it is helpful to be able to shift to 4hi without having to pull off the road and stop which might not be safe to do, forget about dang cold and inconvenient.
If you were staying around your local area and it has been clear for a few days then sure drive around unlocked you're probably going to be safe but otherwise lock 'im up. Winter is one season where having ADD can be a lifesaver. Locking up the hubs gives you that system's benefits albeit with that system's mileage hit as well.

Teuf

Aye Aye, long live 4wd