Arguments abound regarding the best oil change frequency for your
automobile.  Some go by the manufacturer's recommendation, some change
more frequently, some less.  I have searched for studies that would
help me decide for myself when to change my oil, but have not had
satisfactory results.

Let me express what I am NOT looking for:
manufacturer warranty info. 
survey results (opinion polls)
studies based on small samples (you'll know what "too small" means)

Studies that only establish a correlation between hours of duty and
oil "dirtiness", by whatever measure of dirt, unless the study also
investigates a correlation between dirtiness and frequency of failure.

What I DO want:
What I want is a controlled study that attempeted to determine the
degree of correlation between oil dirtiness and frequency of repair. 
. A good study would have directly measured oil quality - a lesser
study would impute oil quality from hours in operation, operating
conditions, etc. And of course it would be a poor study that didn't
assure that the repairs included were only that could reasonably
attributed to oil hygiene.

A good friend in the motor trade tells me that modern oils never need changing.

Of course, no one in the know is going to share this with the punters.

Oops, what have I said?

Likely an honest study has never been done. My guess is changing the
oil your self at the recomended interval will save you $1000 (average)
per decade, unless you put a dollar value on your time. Any warenties
you may have could be voided if you change the oil less often than
recomended. Receipts showing a date of purchase of the correct number
of quarts of oil will likely satisfy any warenty disputes that arise.
  If no warenty is envolved, you are paying someone to change your
oil, my guess is 1/2 as often as recomended will save you $2 (average)
per decade. You will likely regret not changing your oil at all.  Neil

Thanks for your comments.  probono, thank you for the anecdote. 
Neilzero, I am amused that you can be so specific about the $ savings,
having (as you admit) no reliable data as a basis.  I am more
optimistic regarding the existence of the study I seek:  it seems like
a no-brainer that large fleet operators such as FedEx or the USPS
would have figured out the optimum.

The truth is out there...