🚀 go-pugleaf

On Sat, 8 Jul 1995, Frank F Kroger wrote:

> It is a given that the global economy has been increasing while at the
> same time deforestation has also increased. How exactly are you going
> to LINK these two phenomena?
>
> When 2 variables are changing in a constant direction during the same time
> are they therefore  necessarily LINKED?

Here's a sobering experiment.

Take your computer, and construct 100 (or more) observations each of x
and y as random walks.  That is, select x(1) as a random normal, make
x(2) by adding a random normal to x(1), and so forth.  Pick a new random
normal for y(1), and use a new set of random numbers to repeat the
process to build the sequence of y's.

Then regress x on y.  Repeat this a few times.

You will get significant effects more often than not.  If you do it ten
times, you'll probably get some t-statistics in the neighborhood of 5, or
7, or higher.

Do this on a computer in front of a class that's learning regression, and
they may manage to take away some important understanding about why you
can't just run regressions to show that two time series are linked.

Dave
* David A. Scocca                     CB#3265, Hamilton Hall      *
* UNC Political Science Department    UNC-Chapel Hill             *
* scocca@gibbs.oit.unc.edu            Chapel Hill, NC 27599-3265  *