We've been tryin t'tell you Andy...Toyotas rule!  :)

Keep in mind that without examining both vehicles and measuring results with
you driving on a dynamometer, all anyone can give you are guesses.

I'll give my guesses.

The odometer on one or both cars could be incorrect.  Check accuracy by
using the mile markers on highways over a 5 or 10 mile stretch.  Inaccuracy
can be caused by differences in tire diameter as well as by an internal
problem with the odometer. If I were to guess, I'd say that the Camry's
odometer is overstating miles traveled.

Another possibility is that the driving style that has resulted in very good
fuel mileage in your Camry does not neessarily work as well for the Matrix.
The Matrix has a smaller engine and so later shifts may be necessary to get
the same performance and/or fuel economy as the Camry because the Matrix's
engine is lugging a little bit.

Did their estimates really change that much in 5 years?

The methodology for determing fuel economy has not changed that much in the
past 5 years but I believe that it will change soon to more closely reflect
real-world dirving conditions and will probably result in lower fuel economy
estimates.

Very rarely in a Toyota.  In fact, I have never run across one, although
accuracy can very as much as 10% over or under.

I did put on new tires, but I calculated the

Try measuing the circumference of the tires instead of calculating them.
Your suggested test is a good one.

K&N and other "high performance" air filters may improve fuel economy and
performance, but bear in mind that in order to improve air flow, the
passages are not as fine so they have a tendency to allow more contaminants
to pass than a conventional filter.  More contaminants means shorter engine
life and more frequent oil changes required.  The filters that are soaked
with oil can also have an adverse affect on MAP/MAF sensors.  For this
reason, I am not a fan of "performance" air filters for normal street use.

Andy:  When I bought my Corolla new back in 2003, it got considerably less
than EPA mileage for the first 2,000 miles.  The car didn't start getting
close to EPA until it had 6k miles.  And it wasn't until nearly 15k miles
that it would meet EPA mileage on occasion.

With the manual transmission, you DO need to limit engine rpm to no more
than 2500 and keep your foot out of the throttle to get those EPA mileage
figures.

So why did installing a K&N Air Filter result in the mileage increasing? How
does the engine tell the difference between the restriction of a paper air
filter + throttle plate and the restriction of a K&N air filter plus
throttle plate? With modern fuel injected engines, the amount of fuel
injected is adjusted by the PCM based on the oxygen sensor reading. If less
air gets into the engine less fuel will be injected. The amount of air
inducted into the engine is controlled by the air restriction in the intake
tract. This restriction is a combination of the air restrictions of the
intake tract, air filter, and throttle plate. If you lower the restriction
presented by the air filter, you'll just close the throttle a little more to
arrive at the same overall restriction. The PCM has no idea whether the
restriction is a result of the air filter or the throttle. There is no basis
to believe that a K&N filter should increase fuel economy for a modern
feedback controlled fuel injected engine. In the old days of carbureted
engines, air filter restrictions would affect fuel economy because the
increased restriction would act like a choke and richen the mixture.

So why do people report increases in fuel economy with K&N air filter. I
have a few theories:

1) Wishful thinking - they want it to show an improvement, so it does
2) Poor fuel economy measurement - most likely cause
3) Induction noise changes because of the change in filter and people drive
more slowly because they confuse noise with power
4) K&N was installed as part of a general tune-up
5) People modify their driving style after the installation because they
installed it as part of an overall desire to improve fuel economy
6) The reduced restriction does actually screw up the sensors and this
results in a leaner mixture which actually improves fuel economy. This seems
very unlikely unless things really get out of whack, The PCM should adjust
the fuel trim based on the O2 output, but I suppose minor improvement might
be possible - but certainly not the 10+% you are claiming. If Ford, or GM,
or Toyota thought they could increase fuel economy by 10% by just changing
air filters, they would. And they would not have to use a K&N to do it -
just a much larger paper filter. From calculations I have done in the past,
most US vehicles already come with air filter substantially larger than
necessary to provide a minimum air filter restriction.