On Tue, 19 Jan 2016 14:22:28 +0000, chris wrote:

The study make several explicit regression models to test whether
different factors have an affect on car crash related hospitalisations.
They found that texting bans, handheld bans, seatbelt laws and graduate
licensing laws all had a measurable and significant decrease in the
hospitalisation rates.

What's strange is that they found no relationship whatsoever to
accidents.

Methinks there is a smell emanating from the data.

For example, just having a cellphone could mean that you can call for
an ambulance which will take you to the hospital even if you were
just scratched up a bit.

If you didn't have the cell phone, you wouldn't easily have that
ambulance, which means you wouldn't have that ride to the hospital.

You might just walk home, or drive home, or take a longer time
to get "official" help (like from police or ambulance).

So, BECAUSE they can't find ANY relationship to accidents,
they can only find a relationship to hospital visits, but
that could just be BECAUSE it was convenient.

Any study that can't find any relationship to accidents is
nearly worthless.