It all depends on what you mean by statistics, and safer.

Harry Enfield (a UK comedian): "Flying is safe, you don't want to worry
about flying; it's crashing that's dangerous."

Winch launching and aerotowing are safe, it's the consequence of
mismanaging an abnormality that's dangerous.

Winch launches yield few accidents in themselves (e.g. occasional ground
loops); a competent pilot should have been trained in, and able to cope
with, a variety of abnormal situations such as cable breaks, too slow,
gradual power failure, etc. All of these are manageable with competent
handling. If badly handled, however, any can lead to a bad, even fatal,
accident. E.g. turning too low and slow after a cable break, leading to
a classic spin. But is that counted as a winch accident, or a failure to
monitor speed while turning close to the ground, i.e. a flight or
pre-landing accident? You may find that statistics are not compiled in
comparable ways over time, or between one club (or organisation, or
country) and another.

Aerotowing is more likely to kill the tug pilot than the glider pilot if
it goes wrong, i.e. the "tug upset" accident which seemed to start in
the 1970's. In terms of accidents per 100,000 launches, however, these
are rare. Oversimplifying to some extent, the only really dangerous
part of an aerotow where the competent glider pilot has little chance of
avoiding a bad accident is a rope break or power failure over bad
terrain, which is only applicable to a small proportion of sites (in my
experience).

Either method can lead very rarely to harm to second or third parties. I
know of one fatality to a ground handler associated with cable launch
operations (UK, 1970's). I know of other injuries from falling cables,
including one dead cow. There has probably been at least one instance
of somebody walking into a tug propellor, somewhere in the world. I
recall one tug accident where a child was a passenger in the tug and was
killed when it spun in (there is an argument for never carrying extra
people in the tug unless required for operational or training reasons -
it adds to the risk by reducing climb performance and increasing the
period of greatest hazard, as well as adding to the number of people at
risk in the operation - but many clubs seem to regard that as
acceptable).

There are some underlying causes of accidents that are more likely to be
found in the winch launch phase that in an aerotow - but could anyway
happen in the subsequent flight. A loose seat back leading to unintended
pitch up, and structural failure, are two I have heard of - winch
launches usually involve greater stresses and more extreme attitudes
than aerotows, unless the latter involve rotor or other adverse
conditions. But the subsequent flight could also have included
turbulence or other stress-inducing circumstances with equally tragic
results, if the problem had survived the launch phase - so should those
be counted as launch accidents?

Beware of the raw statistics.

Chris N.