I don't understand how Fuelly.com and other community-based statistic sites can maintain non-skewed results. If it's just a matter of the "honor system" that's fine and dandy for a small-time operation. But when well-respected car sites start referencing fuelly.com in their reviews, that starts causing tremendous effect in revenue for some manufacturers (ie, Ford, Hyundai). Yeah, there would be serious PR issues at stake for car companies if they were caught trying to "doctor" results on sites like Fuelly, but it just seems too damn tempting for them not to try something... Maybe I just don't understand how Fuelly works? Someone please explain.
Well anything can be abused. Sending "ringers" to a car mag for a review. Gearing to take advantage of EPA ratings. Fake reviews on Yelp or Amazon for a product. Ultimately it is on the consumer to do their due diligence. Personally, I looked at several sources before buying a car. Fuelly was one. Car reviews another. Feedback on here or other similar forums. Fueleconomy.gov user stats was another source.
So what you're saying is the reliability of these sites like Fuelly.com depend upon them not becoming popular enough as to be the sole-referenced sites in reviews? I could buy that if it weren't so easy to fix results on multiple community-driven sites. If I were a car company, and were desperate enough, I'd hire a full-time team of internet nerds to just sit there and create hundreds of unique Fuelly profiles and keep them updated so as to make them appear legit, and also make numerous posts on forums confirming their findings. Let me approach this from another angle: why do the car companies seemingly let their MPG numbers on Fuelly drop [sometimes well] below what they advertise in the first place? It seems like the ability to mitigate that is both too easy to implement and too tempting to ignore.
I'd think sheer number of participants would help fuelly.com. A community based...even internet community based website is prone to misinformation. But IMO Fuelly is just an optional tool. For whatever degree you want to participate or use the site. Ultimately, the mileage you get with your vehicle is going to be the mileage you get with your vehicle regardless of what Fuelly says. What accepted public estimate is 100% accurate? All of them come with the caveat "Your Mileage Will Vary". Be it EPA estimates or fuelly.com. If respected car sites start referencing fuelly, it's still up to owners and consumers to be honest with themselves. And approach any presented estimate knowing it could be potentially flawed. Usually if estimates become too out of synch with the presented reality the public will let that be known. I don't bother participating at fuelly, but I've visited the site. Not so much for information on my Prius but because my Dad bought a Rav 4 a couple years back and was disappointed in the mileage he was getting. I went to fuelly to look at the mileage for the year Rav4 my dad had bought, and fuelly represented that the mileage most people were getting was lower than other years. Didn't really help my dad, but at least he knew he wasn't suffering alone. Anyway, my bottom line is it appears to me fuelly.com is a well organized and most often valid site that presents fair information. But yes, it's a community based internet tool, could it be misused or corrupted? Potentially. But just like Wikipedia, I avoid sourcing presented information that doesn't have citation or reference. How we perceive, accept and use any information we get off the internet is still the responsibility of ourselves. Hey, not every bit of information I've gotten on Prius Chat has been golden.
Like Wikipedia, Fuelly moderators can look at a log of when accounts were made and where from. So an attempt to shift results one way or another can be simple to trace, and it will take many fake accounts to shift an average with any popular model. A model with a low number of Fuelly users isn't going to give a statistical relevant number to begin with. Plus, the low user number makes it easier to spot the outliers or delve deeper on variables. An extreme example is the new Accord hybrid. Last time I checked there were only 2 registered, and one of them was being driven in Canada with temperatures far below zero. So the 30 something mpg average isn't going to be representative of the car overall. If another car shows up that is doing much better without so tell as to why that is so, then that is going to raise suspicions. I don't see the car companies attempting anything like this because it is little gain for the risk, and it will eventually get out. Even before then, skewing the Fuelly numbers could have unintended consequences. Not only will unsatisfied customers complain about not meeting EPA with a poor performing model, they can now point to the higher number on Fuelly when they take the car to the dealer to have them waste time checking for a fault to explain why their car isn't doing well.
I am sure Fuelly and other sites use preventative measures to avert "ballot stuffing" they probably do this by IP address detection along with a sudden rise in particular vehicles being reviewed
Couldn't agree more. If you single-source efficiency estimates before doing the coin-toss on a new vehicle purchase then you might be disappointed with the results. Generally speaking, if you regard all of the data available then you'll get a somewhat accurate picture of what a vehicle is capable of producing WRT efficiency. The numbers for Priuses are fairly consistent when you look at all of the data that is available. You can (and should) expect different testing results from C&D than you would get from Mother Earth News. Fuelly is a Wikefficiecy site, and whether you consider the data to be accurate or garbage is irrelevant for two reasons: it's a single data point, and YOUR MILEAGE MAY VARY. If you're down for the whole green thing (either $$$ or the bunny hugging kind of green) then it ought to warm your heart that people are getting into daffy irrelevant arguments about fuel efficiency measurements instead of 0-60 times and lateral acceleration (skidpad) results. YMMV!
