In article , "Ed Huntress" wrote:

"Doug Miller" wrote in message
. net...
In article , "Ed Huntress"
wrote:

Think of this as something like teaching a man to fish. Now you know where
the data is, and how neatly they lay it out for you. The thing to do with
it
is to plunk it into Excel and let it draw a curve. If there's a kink in
the
curve soon after the CCW law was enacted, something happened. If there's
no
kink, it means that the CCW law had no effect.

Phooey. A simplistic single-variable analysis like that means nothing of
the
sort, either way. All it means is that _if all other factors were equal_
then
the CCW law had 'x' effect (or failed to have 'y' effect).


No, Doug, on two counts. First, the claim we're testing is that a single
variable (enactment of a CCW law) had a measurable effect. The authors of
the quote Gunner posted made that claim.

Right, and without looking at other variables that may have influenced the
outcome one direction or the other, it's not possible to determine whether it
did or did not have the claimed effect.

If there were other, confounding events going on at the same time, the curve
would have short-term lumps or kinks in it.

Nonsense. Without knowing what other events might have been going on at the
same time, you have no way to know what effect they would have on the shape of
the curve.

It does not. Furthermore, you
can test it further (I did) by overlaying it with national and/or regional
data, where there were no new CCW laws in the other states. If the CCW law
had an effect in Texas you would see a kink or lump in the curve and no
comparable lumps in the other states.

What about other factors specific to Texas, or more prominent in Texas than
elsewhere? I can imagine that illegal immigration, for example, may have had a
disproportionate effect in Texas compared to, say, Oklahoma.

Second, if there were countereffects from other variables, they would
produce lumps, as well.

Without knowing what they were, you don't know what effect they would have.

The chance that one variable would *exactly
counterbalance* the CCW law, and thus produce the *illusion* of a smooth
curve, is so vanishingly small that you'd might as well forget it. One
effect would have to turn on just as the other one turned off, and to
exactly the same degree and with the same trend. Your chances of winning the
lottery are greater.

Sure, but just how smooth is that curve, anyway?

This is standard curve-testing stuff in economics. It's a first cut, and you
have to look closer before putting your money on a bet. But it's a very
effective way to see if something happened or it didn't. In this case,
nothing happened.

That conclusion may be correct -- but I don't agree that it's supported by the
(slim) data provided.

BTW, I ran the same data for Florida over the past 15 minutes, around the
year 1987, when they enacted their CCW law. But I'll leave that as an
exercise for anyone who's interested. d8-)


--
Regards,
Doug Miller (alphageek at milmac dot com)

It's time to throw all their damned tea in the harbor again.