(not for long though). As usual, I was trying to
make my useless data explain or say something about
what I am trying to study.
Working with data is extremely cumbersome. I will
leave for another post some of the data problems,
today, I just want to concentrate on how you make
data say something which is usually called,
the “data torture”.
This data torture might have some “classical”
requirements. Sometimes, you need to take logs;
that’s a simple one. I have also heard that it is
useful to get rid of all outliers (which as a
drawback the reduction of the generally small
datasets). If you are working with time series,
you might need to differentiate even twice to get a
decent dataset. To the raw data you could add a few
“dummy” variables (which are basically binary
variables, 1 and 0). Lately I learned a new trick:
if you have variables with several (numbered)
options, you should derive a dummy variable of
each of the options, otherwise “it does not make
much sense (why? Not very clear to me yet, as long
as the numbered options have some inner logic, I
think is fine). In order to capture marginal
decreasing effects of a variable, you can also add
the square of the variable. Never ending, isn’t it?
You perform so many changes on the data; you torture
it so much that at the very end, something must make
sense. In the process, however, you have not only
tortured the data, though: also yourself. If you are
like me, a beginner, just learning the process is quite
painful, as usual, a trial-and-error sort of thing.
Before you are able to come up with a reasonable
equation (or regression) you must spend hours
trying different combinations of variables and tortures.
Just to find out that your reasonable equation
DOES NOT EXPLAIN MOST OF THE DATA.
The amount of work behind each single equation on
an economics paper is immense, I had never thought about
that.
* Do I really what to spend my life doing such things? *
It is really frustrating to spend so much time
looking for an explanation that simply does not
come, because behind those equations lies your
own theory, your contribution to the science. And
you treat your own theories like your babies: anything
that happens to them is like they are doing it to you
(that is also a good topic for another post). Sadly,
most of the times even if the explanation fits perfectly,
the equation itself does not matter at all. No-one’s
life is improved by that. I studied economics to
make a difference in the world, or at least in my
country (Colombia).
*Is this the way? Is there a way?*
Today, Economics loses 1 nil
