On Wed, 19 Jan 2005 02:57:09 GMT, "carl mciver"
wrote:
snip
Now you see why states that have high minimum wage laws have the highest
unemployment.
snip
Problem is that while this conjecture is plausible it is by no
means proven.

For anyone that is interested in this conjecture, one of the ways
to examine the relationship is to gather state unemployment rates
and minimum wages across several years and perform what is called
an analysis of variance or f-ratio test. This will show both the
probability of a relationship and estimate the size of the affect
of the relationship.

There are several possible outcomes.

(1) There is no statistically significant relationship at which
point you can move on to something else.

(2) There is a statistically significant relationship but the
practical effect is not significant. To use made up numbers for
an example, it can be that increasing the minimum wage does
indeed increase the unemployment rate compared to other states,
but that a $1.00 increase increases the unemployment rate by
0.001%, which is not practically significant, at which point you
can move on to something else.

(3) The third possibility is that there are both significant and
practical correlations. The problem here is that high
correlation does *NOT* prove causality. For example, North
Dakota with the federal minimum wage may indeed have a lower
unemployment rate than California with a state minimum wage $1.50
higher than the federal minimum, but only because all the
unemployed people in North Dakota moved to California.
Disentangling cause and effect, confounding and mediating
variables is frequently the hardest part of any real statistical
analysis.

This particular branch of statistics and economics is called
econometrics if you are interested and want to do a Google
search.

You can include the political party of the president and
composition of the houses of congress if you desire. The
analysis is then called political econometrics. By "shifting"
the data forward and backward in time compared to the variables
your are interested in, such as unemployment rate, inflation
rate, current accounts balance of payments deficit, etc. you can
determine if political parties make any difference, or if
unemployment rates, etc. affect election outcomes.

I have generally found then the affect of time is included in the
analysis, such as using the year as a variable, it generally
"swamps" everything else, leaving no variance to be explained by
any of the other variables, including political party control.

This indicates most of the current discussions and political
activity are simply "Punch and Judy" shows, with no real effect
on the results or outcomes, although these may be highly
entertaining to the participants and spectators.

GmcD