Wednesday, February 22, 2006

if you torture the data...

"If you torture the data enough, they will confess"

What?! Torturing data and they’ll confess?!

Gosh! Why they love econometrics so much? Yes, what I mean by ‘they’ are economists-wannabe and especially economists-wannabe from underdeveloped/developing countries. I have witnessed this back in my university in Indonesia before, but I realized that it also happen here in my university in Norway.

Friends, mostly from Africa and South Asia, are questioning me, "Why don't you take the STAT330?”(a.k.a Analysis of Categorical Data course) Please note that this course actually statistics course, but since many Development and Resource Economics master student take that course so it becomes some kind of the second compulsory econometric course for us and the teacher adapt the material and cases since then to fulfill our economics theory background. We already have 2 econometric courses in first and third semester, and the last semester we should write our thesis. But taking another course – in econometrics – seems not enough? Uh! Are you going to become econometrician or economist?

Well, my first reason not taking this course is because I have difficulties to learn more about this subject. What I mean is that I need to learn more basic things in econometrics, just like what I have got so far from the previous two courses. I really love the teacher, Kyrre Rickertsen, because he makes this stuff simpler and more fundamental instead of showing it as like the “devil” of derivations or formula (like my teacher in bachelor, Mr.HNP). Guess what, I start to love econometrics in some ways for sure. Unfortunately, Kyrre is looks like one of a kind. Another teacher (the one that give lecture on STAT330) does not proof me the same feeling as with Kyrre, by stating the above quotation in his first handout. So, that quote initiates my second reason for not taking more course on econometrics.

As far as my concern with data, my query is what kind of confession we will get? My mentor, Pak Jack (John W. Molyneaux), once said to be very careful in “utilizing” (not torturing!) data, especially survey data. There are several small but important factors we need to understand before we may ‘interrogates’ them to get confession. Not mention that these data have a lot of technical problems such as stack of questionnaires to read and piles of files with unique format to be merge. They may only consist of numbers and codes, but they will talk different if we question differently. Or if you use another secondary data, which is quite ready use like data from publication, how far we can explore them since there are so many problems may happen before we get the data on hand, e.g. data collection or tabulation-printing problem.

Anyway, the quote and friends question looks like to remind me to finalize the data for my thesis immediately. “Talking” with data seems very enjoyable instead of interrogating or even torturing them since they have so many stories and facts. But I also realized that they probably delude me as well certainly if I am stupid by trusting them too much. I hear they whisper… Alas, I don’t understand at all. They use binary language!!

And for econometrics?... Just like Def Leppard song titled "When love and hate collide"


Anonymous said...

Suppose to be: data will speak if you treat them correctly. "Nip and tuck" of the regression model might make it more efficient and comply with theories, but we will not get any empirical evidence, which should be the goal of doing econometric analysis.
It reminds me of Pak Haidy who loves to "torture" econometric's students :-)

Berly said...

Greene's econometric is too broad and un-reader friendly (I used it for two courses).

Highly recommended is

Econometric Methods
by Jack Johnston & John Dinardo
which know how to explain the essence.


Econometric Analysis of Cross Section and Panel Data by Jeffrey M. Wooldridge
for more technical but still very readable manual

'za said...

i feel u bro'

but i do understand why we need to be intimate with data ... for that, ...let's struggle with facts and figures ...chayo!