Probably not. Premium is for engines that run at a higher compression ratio, it prevents preignition (knocking). If your car runs ok on regular gas, it will not run better on premium. Only if the engine is knocking in the winter should you consider it.
The presumption is that Mini uses a system similar to Honda’s (as mentioned in the next link) with a float in the tank that contributes to their volume used calculation, which contributes to the Mini’s better mpg accuracy over number of squirts as the sole input to the volume used calculation method. The accuracy of the float level as an indicator of volume change however is impeded by the fact that size of the tank is in fact variable. (Neither 25% down of a 14 gallon capacity tank or of a 12.9 gallon capacity tank is the same volume of gas used as 25% down of the listed 13.2 gallon capacity.) No doubt they have algorithms that attempt to correct and control for those variations but the message seems to be that they are not exact corrections for real world changes.
Neither the author nor the guy from Mini seem to be idiots to me.
I defer to the car guys here but it seems to me that the counting the squirts method also needs to vary its algorithm with temperature as we’ve been told that the amount of fuel in each squirt varies to keep in step with denser cold air (giving more power and less efficiency).
Bottom line seems to be that the car generated mpg indicator is often off by 5 to 6% and that depending on the methods and algorithms used to calculate they be systematically more off during one or another season.
Note though, neither of these cites clarify if the distance measurement includes the zero movement while idling or only while the vehicle is in motion for all vehicles.
Do you check it a lot? Once you’ve got more than 50 miles or so on the trip counter, it’ll take several minutes of idling to change the cumulative reading 0.1 MPG. I only noticed it one time right after I’d filled up a rental car years ago, but have been kind of weirdly fascinated by it ever since so usually make a point to check it when driving a strange car. I think one of my cars does it… I’ll try to check over the next few days and report back.
I think the guy from Mini mentioned the gas tank volume thing because that’s a reason why measuring the MPG by hand might not be entirely accurate, since the point of the article was explaining the discrepancy between that and the dashboard MPG readings. (Although I do note that none of the reasons cited in that article would explain why dashboard MPG readings are systematically optimistic. I don’t think I’ve ever heard of one that read less than the measured MPG.)
Actually I do I like to know what kind of fuel mileage I actually get in whatever I drive so when I get into a new car I always reset the trip computer.
So I drive out of the rental car agency and watch the mileage climb until the first stop light then I watch it fall.
Personally I think the electronic gauge is way more accurate than looking at the pump and the odometer. The amount of fuel that goes into the tank can vary based on temp, the pump’s shut off sensitivity, the angle the car is parked at.
Until I see a scientifically done study where they control for those variables I would say the issue is still undecided.
I know when I have checked the pump against the dash it is always is inside of what I consider experimental error.
Well, in my 2013 Hyundai Elantra, the MPG reading from the trip computer is always off. It consistently reads high compared to the pump/odometer method. Curiously, Hyundai is in trouble with the feds (again) for unrealistic mileage claims.
I have driven numerous Elantras* for MPG complaints. Not only did the pump agree with the trip computer (+/- .1 MPG over 100 miles plus of driving) but I always exceeded the published MPG numbers.
:::shrug:::
I would be willing to bet I could do the same with your car also.
*I was the service manager of the largest Hyundai dealer in the western US
ETA: my Sonata exceeds the EPA numbers before the reduction.
Well, maybe I got a lemon. My odometer/pump technique works out to no more than one or two tenths of an mpg off compared to my other two vehicles’ computers.
If they’re counting injection events, then they’re also paying attention to injection durations, since there’s a big difference between a zero-throttle fuel injection quantity and a wide-open-throttle quantity. So it doesn’t really matter what the ambient temp is, only the fuel temp, since that affects fuel density - and even then, fuel density varies by only a couple of percent from 0F to 70F. That’s within the acceptable range of accuracy for a dashboard fuel economy meter, so they may not even incorporate fuel temperature, since that would require an extra temperature sensor; if they do incorporate it, then they probably just use intake air temp as a proxy, since that’s a sensor that every fuel-injected car has.
Over the course of an entire tank, either one is reasonably accurate for making sure nothing is terribly wrong with your car.
I did notice a bit of that on my old M3 under the filler cap once upon a time (same commute), and started making an effort to get the engine fully up to temp more often after that. Luckily the Subaru warms up a lot more quickly than the M3 (at least the coolant temp does, I don’t have an oil temp gauge) so hopefully it’s at least a bit less of a concern on this car.
I do change the oil very often in the WRX – even though it’s full synthetic, I change it once BETWEEN scheduled maintenance intervals, since I do it myself it’s relatively inexpensive and easy – and I haven’t seen the issue pop up yet.
It’s getting pretty close!
20 mpg city in the summer (I get into boost more often than the EPA testing folks, I guess) drops to under 14 when it’s really cold and I’m only making short trips. The price to pay for turbocharged AWD fun in snow-covered parking lots and living really close to work… 
My son and I have each rented Elantra’s for trips. Each of us has seen the dashboard computer say that we are averaging (over several hundred miles) 43 - 44 mpg. When refilling with gasoline, though, the numbers are smaller, around 40. Elantra doesn’t seem to me to be claiming anything higher than 38?
Rented Elantra’s what? 
I have spreadsheets for my last Prius and the current one, both 2005s. The displayed MPG is almost always less than what I calculate. The calculated mileage is currently an average of 3.8%. Example:
**
Date Odo Miles Gallons Ind MPG Calc MPG Difference Average MPG Avg. Error**
07/29/15 242.1 5.391 48.8 44.91 8.67% 43.81 3.79%
What numbers are you using to calculate mileage between fill-ups? You’d have to write down the number of gallons required to fill the tank to calculate how much fuel you actually used. If you just divide by the tank size the number you get will be wrong. Also, if you refill when it’s hot out the gasoline going in will have expanded by a significant enough amount to reduce apparent mileage.
For what it’s worth, I never got more than 35 miles per gallon out of my (2008) Elantra even on all-highway trips.
I make an effort to fill up at the same pump at the same time of day (usually around 0600).
I think I’ll just repeat what I said in January, which was that most modern cars don’t start out doing the full-tilt stoichiometric type operations where the MAF (Mass air flow sensor) and the engine computer exactly figure out the proper air/fuel ratio with feedback from the oxygen sensor(s).
Instead, they start out in what’s called “Open Loop” mode, where the engine computer only uses the data from the MAF and the coolant temp to calculate the air/fuel ratio, as the oxygen sensor isn’t warmed up, and the engine temp (as measured by the coolant temp) isn’t high enough yet- 105 is the temp I’ve read.
Once the coolant temp and the oxygen sensor are warmed up, and in most (all?) cars, some minimum amount of time has elapsed, then the car goes into “closed loop” mode, where the oxygen sensor, coolant temp and MAF all contribute to the calculation of the air/fuel ratio, and it’s much more accurate than open loop mode.
In practice, this means that while the car’s in open loop mode, it’s running rich, and using more fuel, as the oxygen sensor isn’t in play yet. When it’s colder outside, the car stays in open loop mode longer than in warmer weather, and the coolant temp is lower, so the engine likely runs even richer in open loop mode than it would in warmer weather.
I bet that’s the main difference between summer and winter economy; the MAF/O2 sensor feedback loop will take care of any density differences due to temp (that’s what a MAF does, you know) and not everyone idles their car for extended periods, but pretty much everyone sees an economy drop in the winter because they operate in open loop mode longer than in the summer.
If applicable, putting on winter tires with a more aggressive tread and softer rubber will increase rolling resistance.
Also, perhaps there are other external social factors. Someone mentioned more stop and go driving. Where I live, there is noticeably more traffic during winter months: increased paranoia during any types of precipitation, winter months = shorter days = people driving slower in low light, kids are in school = more school buses and their frequent stops. I’d imagine a significant cumulative effect of driving slower, ie. operating your automobile for a longer time at a lower efficiency just to travel the same distance.
I don’t know how you are calculating you numbers but the trip computer is right. It knows to the fraction of a cc how much fuel is used by the engine and the car knows exactly how far you have driven.
Filling the tank you have variations of fuel temp, angle of filler neck, did you top it off? How many times?
I work with Toyota technical department now. Their policy is on fuel mileage issues the trip computer is the gold standard. All other results are ignored as worthless.
Well, unless you fit bigger or smaller wheel/tire combinations. But presumably that won’t be the case for a rental, and that will throw off pump-based calculations by the same measure anyway.