EPA Vehicular Emissions Study

Best and Worst: no surprises!

2017 Hyundai Ionic Electric

• 150 MPG city
• 122 MPG highway
• 136 MPG combined

1986-1990 Lamborghini Countach

• 6 MPG city
• 10 MPG highway
• 7 MPG combined

• Supports common consumer beliefs about vehicle types, design, and fuel efficiency.
• Supports claim that electric cars have lower highway MPG than city MPG.
• BUT – only one example of each vehicle in the EPA dataset!

Limitations: sampling issues

• Many models have only one example in data set.
• Older cars appear more common in data set.
• For the sake of fair comparison and relevance to contemporary automotive market, we include only vehicles manufactured from 2010-17 were considered in statistical analysis.
• Comparisons within categories sometimes have very small sample populations. Ex: electric car brand comparisons.

Central limit theorem

Limitations: human factors

• Are individual driving habits a factor?
• Is the performance of the manufacturer consistent?
• Did the EPA do their math correctly?

Limitations: time, demand, creativity

• Myriad variables included in dataset: myriad possible tests.
• Ignoring diesel, etc.
• What are some of these variables (GHG score, economy score, etc)?
• Do the magazine or its subscriber base have specific questions for us to investigate within this data set?

Inquiries and Hypotheses

• Which brands produce the most gas efficient vehicles?
• We compared Toyota and Honda, two manufacturers known for efficiency.
• We compared within the ‘midsize cars’ category.
• Which brands produce the most energy efficient electric cars?
• We compared Volkswagen and Ford.
• We compared within the ‘compact cars’ category.

​

• Does 4 wheel drive really use more energy than 2 wheel?

​

• Are hybrid cars really more energy efficient than gas cars?
• Are electric cars really more energy efficient than hybrid cars?

​

​

• All inquiries use a two sample t-test assuming unequal variances.
• All tests operate with a null and alternative hypothesis:
• Null – No difference in means exists between the two groups being tested
• Alternative – a statistically significant difference between means exists.
• All tests were done at a 95% level of confidence.

Handling the data

• Data comes from a 2017 report from the Environmental Protection Agency.
• For each test, we chose a variable from this data accordingly:
• Test 1: Combined MPG (FT1)
• Test 2: Combined electricity consumption
• Test 3: Combined MPG (FT1)
• Test 4: Annual fuel cost (FT1)
• Test 5: Annual fuel cost (FT1)
• We compare the means of these variables across two groups of vehicles.
• All tests exclude vehicles before 2010, however we will include descriptive statistics of the entire data set below. This includes entries all the way back to 1984.
• It appears that this data does not always show normal distributions as a whole, but it does when you separate categories.

Test 1: Toyota vs Honda, gas cars

• 95% confidence
• Tested on midsize cars
• Large variance within each group
• Are personal driving habits a factor?
• Have we come a long way since 2010?
• Midsize cars from 2010-7 may be built for a variety of purposes and price ranges.
• Test shows significant difference, suggesting that Toyota outperforms Honda on gas mileage.

Toyota: Range from 23-56 MPG (33)

Honda: Range from 21-48 MPG (27)

​

Confidence interval: 2.1-7.7 MPG

Test 2: Volkswagen vs Ford, electric cars

• 95% confidence
• Cannot slam Tesla – no comparable class vehicles by competing manufacturers!
• Tested on compact cars
• Severely limited sample
• Tested for electronic fuel consumption
• Data suggests that Volkswagen’s electric compact models are less costly to run than than those by Ford.

Confidence interval: 2.2-3.1 kW-h

Test 3: Drivetrain and mileage

• 95 % confidence
• Tested on all vehicle types
• Testing 4wd, AWD, and combined drivetrains against 2wd front-wheel and rear-wheel drive
• Could there be further differences within these types?
• Significant difference found, supporting the claim that supplying power to all four wheels consumes more gas, which is widely accepted as fact in the industry.

Confidence interval: 2.9-3.7 MPG

Test 4: Gas vs Hybrid

• 95% confidence
• Tested on yearly running costs
• Tested on regular, midgrade, and premium fuel measurements vs gas and electric, gas or electric, regular and electric, regular or electric
• Large difference in sample sizes
• Test shows significant difference, which would support the consumer conception that hybrid cars cost less to operate than gas cars.

Confidence interval: $408.20-$631.55

Test 5: Hybrid vs Electric

• 95% confidence
• Tested on yearly running costs
• Tested on aforementioned hybrid types vs “electricity” fuel type
• Surprisingly, there were more electric models included in the EPA study than hybrid models.
• Analysis found significant difference in mean costs, suggesting that electric cars are indeed cheaper to run than hybrid cars.

Confidence Interval: $518.91-$745.56

Implications

• Our findings suggest that conventional knowledge generally holds true in the car market.
• The automotive industry as a whole is improving in being environmentally conscious.
• Your readers are getting accurate and relevant information from your publication.
• Everyone has “their” car and your publication can be a way for individuals to find and connect with it.

Further study

• Testing within make/model over the years to see if things are improving.
• Testing within make/model/year to determine if they are made consistently. This may also be an indicator of factory performance.
• Testing the improvement over the years between manufacturers to see if one is beginning to eclipse the other.
• Auditing EPA data.
• Incorporating new data sources.
• Is progress so fast that we need to compare vehicles within a single year? If so, we certainly need more cars to test on!
• Technical specs:
• Transmission type
• Engine type
• Correlation between emissions and fuel efficiency
• Cargo and passenger capacity vs running expense
• Charging times
• Range
• Any specific questions you may have for us.
• Any two brands can be compared within sub-classes.

Conclusion

• Magazines have continued to thrive as other forms of print media dwindle in readership.
• Car and Driver has successfully made the transition into the digital age.
• 87.8% of audience is male.
• 2015 comScore//GfK MRI Media + MMx Fusion (08-15/S15)
• Stats available on company website.