This is the relevant paragraph from the link Casey offered:

When ELTs were mandated in 1973, most GA aircraft were equipped with an
ELT that transmits on the 121.5 MHz frequency, the designated
international distress frequency. The original ELTs were manufactured
to the specifications of an FAA technical standard order (TSO-C91A) and
have an activation rate of less than 25 percent in actual crashes and a
97 percent false-alarm rate. In 1985, a new TSO-C91A ELT was developed,
which substantially reduces or eliminates many problems with the
earlier model. The TSO-C91A provides improved performance and
reliability (with an activation rate of 73 percent in actual crashes)
at a reasonable cost to users ($200-$500 including installation). Since
then, an even more advanced model of ELT has been developed - the
C126 ELT (406 MHz). This newest model activates 81-83 percent of the
time, but the current cost is $1,500 or more per unit, not including
installation.

Please note that these are AOPA rendered statistics. AOPA has a very
strong position against mandatory use of the 406 MHz units based on
member financial impact. (In fact, little of AOPA writes should be
accepted at face value. They are an advocacy group, and not always in
the best interests of the majority of pilots.) Therefore, they have
painted a picture that shows little value in moving from the older
technology to the new and have omitted some important facts regarding
time to acquisiton of signal, ability to verify whether the signal is
an actual emergency, accuracy of first pass position resolution,
reduction in false alerts, and time to arrival of emergency personnel
on scene. They do, however, acknowledge that as of 2009, satellites
will no longer monitor the old bandwidths. This means no repsonse
unless someone watched you go in.

The rhetoric used by the AOPA is interesting. Note the lack of
parallelism between the first two examples. A false alarm rate in C91A
models is not cited. This is always a sign of a potentially flawed
argument, typically presented intentionally to drive the reader to
specific conclusion.

In response to Casey's point, I'll repeat my earlier concerns that put
me on the unpopular side of this discussion: the units aren't for our
safety but for the safety of those who might one day have to come
looking for us. Search and rescue is a dangerous business. Consider
Utah skiers killed in avalanches this past weekend. NONE wore beacons
while skiing off piste. That has put many dozens of searchers in harms
way for much longer than was necessary. Finally, recognize that we are
talking about racing. Pilots typically fly more aggressively, cover
larger distances at high speeds, often traversing unlandable terrain.
Organizers are asking racers to make their jobs a little easier. For
the cost-conscious, they can satisfy the requrest for under $300. For
the value-concscious, $1K will provide an even greater degree of
assurance.