Suggest to use median MPG for a group of cars
 

It is reasonable to use average MGP (mean value) for one car. However, median value makes more statistic sense when calculating MPG for a group of cars. I saw people entered unrealistic MPG for their car. It may significantly pull the average MPG up for their car model. By using median, the unrealist MPG is just one sample point. If all other samples are real, it won't impact the overall result much.

Good idea, but most cars have around 3 or 4 different engine options, of those, some have more HP than others, some are diesel other petrol, and there's usualy 3 different gearboxes you can get too. All of the above use different amount of fuel so if you ask me, there are way too many options/varients for anyone to work out an average. Just my opinion though :)

I look up my car, filtering for year and engine (fuelly won't select for transmission type) and then start comparing. And even then I see many cars with incredibly unrealistic MPG listed for at least one entry. From that point on, I can no longer trust their entries and I disregard them in my personal comparisons.

Despite the fact that I really enjoy how this site helps me track and compare my mileage, we are still dealing with subjective data, as there are no objective controls that can be put in place. We do the best we can sort through it, but this information is only as good as the entries of the members, which are all over the place. In the fuel talk section I repeatedly tried to tell someone that they were never going to get truly accurate MPG unless they started filling their tank and as far as I can tell he still isn't doing that.

All that said, I agree, median mileage would be a better gauge than average (mean) MPG for vehicles.

Like Draig said earlier, I think it is meaningless unless separated by engine / fuel type etc.

If there are enough vehicles... it really won't make much of a difference.

What bothers me, is that there are people with my vehicle that strictly use it on the highway, while I almost strictly drive in the city... making me look like an idiot.

But that's minor! haha

The city/highway percentages are always going to be subjective on this website anyway.

I agree - I am one of those people posting horribly unrealistic MPG numbers for my car - but after an engine swap to diesel and some careful driving and LRR tires - that's what it's getting. I don't mean to mess up the stats, but I am doing just that.

Agree on using the median value for fuel economy. There's too many ways in which the mean value can get skewed. Biggest source of error would be the city vs. hwy driving as it's been mentioned.

Yes especialy where I live, we dont have any highways (motorways in the UK) and my nearest city is some 65 miles away too!

Thanks for the feedback, we'll take this suggestion into consideration as we plan the coming iterations of the site.

Also, we're planning to add a VIN decode function which should allow us to have more granular taxonomy to organize the vehicles. This will allow for more detailed reports by make, engine and transmission. This will be one of our top priorities since we think it's very important to capture this data about the vehicles participating in Fuelly so we can build a better data set for everyone to benefit from our crowd sourced results.

Keep the great feedback coming!

I agree. As a quality engineer who knows all about statistics, we know that Average is a very bad and un-representative statistic of a sampleset for the population where as Median is a very good statistic. Please change this as there are also bogus values posted on fuelly to purposely skew the data for the population.