PhineasFellOff

I wondered about the wind direction, too. But I don't know how to get that information on a real-time basis. So my data doesn't include it. However, I do know that the interstate 10 and 20 is slightly and very gradually uphill going in the east direction and therefore gradually downhill going in the west direction, which affects mileage accordingly.

PhineasFellOff

No sarcasm. I'm definitely keeping both cars.

SiMplyBluE

Wanted to bump this thread up, since it has great info. Great read and data collection!

TheColonel

I think the lurking variables in this case are driving style, road difference (diff. routes, etc.) and just variation between vehicles. Some cars are just better put together than others. (I.E. Jack just did a better job installing that one part on his line than frank did on his).

Even though this is data is far from a simple random sample, and arguably doesn't represent the population, I'll do a statistical analysis anyway and see what comes out.

Kudos to you Phineas.

Smoker

thanks for bumping this up. I totally missed this post from last month. Very informative indeed.

ABG

Excellent write up and thanks for taking such detailed notes!

PhineasFellOff

Yeah, thank Simply Blue for bumping this up, as there are plenty of people who would be encouraged to see the data.

PhineasFellOff

I don't think my car is anything but typical. In the original data post and also a subsequent post, I mention that in city driving DURING the trip, I got 18.0, 19.4, and 20.1 mpg, consistent with others and what I normally get for abundant-city driving.

I also point out in the Conclusion section (point #9), that a long-distance trip strictly on cruise control is highly idealized, since the mileages I got on the trip are higher than the mileages for my regular Los Angeles fwy commutes, although these are not bad.

*** Edit ***

Let me add that it even appears completely normal when many people have reported getting 14-16 mpg while driving the hell out of the car in virtually 100% city driving. I am convinced that if these people drove the way I do or did a long-distance trip on cruise control on the same roads as I did, they would get comparable mileages.

The only thing that appears to be abnormal is when a very few people have reported getting 12-16 mpg while driving like a granny. In those cases, there is definitely something wrong with their cars.

crumpmd

Well I tend to agree with you. I drive mine in the city almost all the time. On long trips when I can set the cruise on a reasonable number I see mpg go up to 18-19 range. But mostly it is 14-17 in the city.

DaveT

My recent trip compares favorably with Phineas' data. Round trip Knoxville to West Palm Beach using I-40 to I-26 to I-95.
Total miles = 1,912. Total fuel = 87.368 gallons. Overall = 21.883.
Best tank = 23.13
Worst tank (putzing around off Interstate) = 20.05
Car had 2,200 miles at start. Most of the interstate driving was between 75 and 80, with 1 or 2 tanks in the 80+ area.
Caris 6-speed and has the "M" flash. Cruise control used about 60% of the time. Also running AC about 85% of the time.
Used exactly 3/4 quart of oil.
Tires at 32 psi cold. People load = 300#. Luggage/Cargo = 200#.
Comfort level = excellent. Seats = great. Problems = none.

TheColonel

I'm not exactly disagreeing with you. I've heard a lot of people complaining that they attempted to drive like a grandma, kept he rpms low and still get 16mpg on mostly highway driving. I'm saying I think people's definition of what's grandma driving and what's beating the hell out of the car can vary. That's why I say the data may not be representative of the population, as some people may think they're being conservative when they are not, and so they wonder why they get lower mileage on the highway than you do since you supposedly drive the same.

Also, I've had exams all week, so if I come across as hostile or like I'm trying to correct you, or if I say something stupid that is obviously wrong, I apologize. I'm a little mentally drained.

d0 Luck

at what RPM were you cruising; 60 and 75??

and was it on 6th gear?

TheColonel

*EDITED FOR IDIOTIC CONTENT*

I realize this analysis is rather juvenile in the statistical world. If you are knowledgeable in the area of statistics, be kind in telling me that I probably did this all wrong.


ONE LAST NOTE: The RX-8 is an awesome fun car, and as much as good mileage is good for my wallet, I’m buying one regardless!

PhineasFellOff

First, welcome to the club.

Second, it just seems some of your comments are out of context because you came in at the eleventh hour. But's that no big deal.
See, this is one of about three or so major threads on MPG, which comprise two different lines of analysis. A statistics grad student or ph.d. name Norton started a major thread, in which he collected a ton of variables and identified which factors are most related to variance in mpg. I'm not sure you've checked that thread out. It is important to understand that for Norton's analysis, variation is a must, the more the better. Norton's analysis, however, was not designed to determine, once and for all, what mileage the RX8 CAN get. Thus, bickering about the discrepancy between what people were actually getting and the EPA rating continued and still continues. (1) Is the EPA hwy estimate of 24 mpg attainable? (2) Can virtually any RX8 attain it? And if so, (3) what kinds of driving styles and driving conditions would attain it? This last one is critical because many people in the early months of the RX8's production were assuming that in order to get 24 mpg, one had to drive unreasonably slow, like at 55 or 60 mph on the fwy.

This thread is my conscious attempt to answer those questions. As I noted in Norton's thread, whereas variation is desired in his analysis, it is not desired in my analysis. Whereas statistical analysis is important in his, it is not in mine, because I simply ask WHAT CAN BE ATTAINED AT WHAT SPEEDS (the way you phrase the question decides what statistics can be applied). It's a simple but precisely done observational analysis. Even though Norton's plentiful data was used by many people to determine what kinds of hwy mileages could be gotten, it also brewed a lot of arguments, e.g., people calling each other liars or assuming someone did not calculate mpg correctly. A large source to this problem is precisely because everyone drove differently and no one could be sure EXACTLY what was their average fwy speed. Indeed, his analysis even brought about a conclusion that driving style is a major source of variance for mpg (of course, right?). My thread avoids this problem simply by calculating everything (or just about everything) on CRUISE CONTROL, thus reducing a huge source of random error. Norton's analysis was elegant and complex. But this one is golden in its simplicity because I seek to answer a much simpler question.

The inferences that a combined approach allows us to make are cool and require a little "hop skip and jump," which seems questionable but scientists do it all the time. For example, if at least several of us have been able to attain a certain fairly high fwy mpg, say 23 mpg, by driving however we each drive the freeways, fine. We at least know that 23 mpg is attainable by much more than just one person, that it is no fluke. We determined this looking across different subjects. If I am part of the 23 mpg "club," I can do a follow-up examination/observation using ONLY MY CAR (within subjects). If I am able to achieve unusually high mileage, say the 27.7 mpg that I actually did get, and do it more than once under certain fairly precise conditions that are identified, such as at 60 mpg on constant cruise control at 75 degrees Fahrenheit going slightly downhill, then I CAN infer that others or many others in the group should also be able to match that mpg under similar conditions. The only thing I did not do was reproduce the 27.7 mpg, but as was stated in my analysis, I had another tank that appeared to be headed even higher than that. Now a double inferential jump that you will just hate. At least four others have reported getting between 29 and 31 mpg. This bolsters my 27.7 mpg and says that it is not a fluke even though I have not yet reproduced it (and that of course is because my trip came to an end). The final inferential leap is that my car is nothing special, as the city driving mileages of 18 to 20 mpg under less than crazy driving attest, ranges that many many people get. I assure you that these inferential leaps are reasonable and done all the time in various domains of science.

Finally, because I have identified fairly precisely the conditions under which I got my mileages for various speeds, ANYONE from this forum can then mirror those conditions (e.g., cruise control at 70 mph) and see what mileages they can get. And of course, share them with the rest of us in this thread, so we can make more inferences.

PhineasFellOff

Jesus Christ, colonel. You can't really use my data for this kind of analysis. Just think about your question or hypothesis. You are asking: Are Phineas' data significantly different from 24 mpg?. But you must also look at what's behind my data. What if I had done all 20 tanks at 80 mph? There's no way I would have gotten as high as 24 mpg on average. Then what would your result have been? Similarly, most of my data is at 70 mph. You know what I'm saying? Another way is to look at your histogram. The shape of the histogram is meaningless because it contains a bunch of different speeds. The direction of skew is totally irrelevant, as is the histogram.

What you need to do is do a histogram (i.e.,collect a separate set of data) for EACH speed, one at 60 mph, one at 65 mph, one at 70 mph, and one at 75 mph. But for this, you would need a bunch of mpg measurements for EACH speed! You can then look for spread of the MILEAGES, hoping they come out close to normal distribution, which is a required assumption of inferential statistics. LOL. Then you could run ANOVA (analysis of variance) for the mpg's at each of the four speeds compared to the EPA rating.

Come on, colonel!!! Focus on the fried chicken!

TheColonel

Well Phineas, you obviously didn't read my post very well as you missed the 2nd to last paragraph!

So hey... like I said... I probably did this all wrong... :D

I was bored, all my things are packed as I finished my exams, I have nothing to do. I will pull my post since it is such a haneous distortion of anything resembling science.

TheColonel

And by the way, I'm a Poli. Sci. major, so I'm not used to having to actually make any sense. :D

PhineasFellOff

Oh yeah, and I even missed something. You can't even compare my data to EPA's 24 mpg because we don't have their data and we can assume their conditions were very much different from mine in this thread. You can't even just throw in twenty 24 mpg's in for EPA in order to do ANOVA!!!


Hey this is early for final exams, which means you go to a private college. What school do you go to?

TheColonel

Well I didn't do ANOVA. (Probably because I don't know what it is... ) Since I never even suggested it, I can’t see how you can pick at “my” ANOVA comparison.

Another thing… and for once here, I can say, you’re wrong!

How can you assume that just because I've taken my exams that I attend a private school? Tisk, tisk, for someone so skilled in finding flaws in others assumptions, you make a rather precarious assumption with little to no basis other than perhaps your own preconceived notions of when exams are to be taken...

I go to public school, TCNJ.

So, anything else you'd like to point out that's wrong with my crappy little analysis that I really only wrote to amuse myself? Or can we stop knit picking the poor political science major who has never received an A in a math course and reminding him how horrible he is when it comes to math?

(I also hope any discussion of my, and I use the term loosely, “analysis” is in jest at this point. If not, then I apologize for any sarcasm, as hostility has never been my intention.)

PhineasFellOff

Just work on the fricking recipe and secret spices, Colonel.

TheColonel

Yeah, actually my name originally had nothing to do with KFC... I don't even really like KFC. Unfortunately, most interpret it as you have...

PhineasFellOff

1295 average Freshman SAT is not bad.

TheColonel

It's a good school and becoming harder to get into. They're raising the bar each year.

RodsterinFL

Whoa! Excellent data.

as far as the comment goes on "driving like a grandpa" or conservatively, this is what I and some others on the forum have discovered:


A conservative drive in an RX8 consists of shifting gears under 3600 RPM - period.

Note too the data on 75mph (over 3600 rpm in 6th gear)

That defines it best and levels the field of understanding. The reason for this 3600 cutoff is due to the injector utilization - minimal use below 3600 RPM.

You are right though in saying that what conservative is to one person is different to another.

Most of us don't drive the conservative way but it is great to know the best way to glean the most MPG out of the car if you want it.