When I was in college (exercise science) we were always told to not use any research older than 5 years old for a reference because it might not be accurate due to advances in understanding and technology.

Is this a valid thought in fields that have to reference research? Why or why not? It always seemed ignorant of me to not look into a study just based on the date it took place.

Depends upon the research and the field of study.

Older research which has been continually revalidated in subsequent studies is still current and citable, but it does require a thorough study of all related research in the field in order to determine if there's anything newer than or contradictory to the (older) research being cited.

Basically, you can't just cite an older study without researching to see if it's still current...

Field matters.  Dental materials take a 5-10 year run to see whats going on so it's very common for us.   Always new stuff coming out but it takes another 10 years to see if that generation of zirconia would microfracture and so on.

It all depends on the subject.  In Astronomy and Biology, I would never use data much older than 2 years and in Astronomy, jeeze 6 months old could get you in trouble.

My first year astronomy professor changed data on the size of the visible universe being taught in class, mid chapter because new data came out that replaced the old data.  He did not hesitate updating data if it was properly vetted and peer reviewed.

Depending on your time available, you may not want to completely ignore old research. Potentially the (in)accuracies are due only to the technology of the day.

Damn, that makes doing background research a lot easier than in archaeology.  When I review manuscripts and proposals I always check the bibliography first to make certain that there are references that predate 1948 ... my birth year!

I disagree with the biology bit. We routinely reference old papers, which we actually refer to as the 'pillars of immunology.' Biological research methods may get more sophisticated over time, but that doesn't invalidate older papers. Watson & Crick will always be valid despite their lack of advanced methodology.

About the same in a lot of engineering.

There remain certain fundamental things that are as valid as when they appeared.

If you get into new technology all bets are off though.

Remember what color the bleeding edge of technology is.

Green.  The color of money.

Huge amounts of semiconductor information are guarded by companies as 'proprietary.'

It is not patented since that would require revealing how to do it.

Better to keep you competitors guessing.

It is pretty rare in semiconductors that having the finished device will reveal HOT it was fabricated.

A SiGe-CMOS (a special type of Bi-CMOS) device can take around 1,000 process steps to fabricate.

It interweaves two different technologies on the same device.  SiGe is operable at incredible frequencies.

CMOS has power and density advantages and it runs at 50 GHz.

The steps have to be done in such a way as to prevent one process from interfering or damaging the other.

Need to operate at 60 GHz? Real switching speed not slower things in parallel.

The answer, of course, is that it depends. An analysis of agricultural data taken in the Census of 1850 done in 2010 is invalid? I think not.

Realize the "Research" generally means academia. If no research is valid except the last five years, then you get to write the same papers over and over again (or assign them to your grad students, many of whom aren't really capable of independent research), and for yourself you can apply for the grants to fund them. It gives the grant administrators an out to give away their money without depending on these academes to actually come up with something new. Follow the money, as usual.

If research more than five years ago is invalid, should not Ph.Ds become invalid after 5 years, since they are based on obsolete technology and research?

Depends on the subject area.

Hehhe, I find this quite entertaining. Most of my references are from the 70's. "New" references are mid-90's. 

My field is Aero propulsion/turbine integration.

I used to regularly write those "references."

Radiation testing at cyclotrons of integrated circuits for use in satellites.

Brookhaven for the win.

You can test parts in  packages without exposing the die.
Tungsten is used for beam shaping and shielding.