Why Statistics.... Why.....!!!!

Let say I'm going to have a party next Saturday at my house and you are all invited. But then you're all going to ask me, "Where's my address?" To be better in describing it, I'm going to draw a map for you, and I'll ask this to you, "How much do you want me to draw my map as close as to the reality?"

"100%" one of you might scream.
Then I start to draw my 100% close to reality map with 1:1 scale. Now, can you imagine how big it's gonna be and how big is my paper will be?

"90% is enough" some of you might say.
Okay, I wont draw at 1:1 scale, but then I draw the map to the very last detail up until the rocks, the plants and the trees. 

"80% is enough" you might say.
Well, I might still have to draw every names of the buildings and every turns that you have to make.

So on and so forth, and In the end I just draw some roads and some landmark you can use as a guidance to find my house. The result of the map would be so simple and focus in pointing my house and how to get there. In the process, I will (kind of) ignoring the fact that is exist in the reality and ignore it so it won't distract you to reach the goal, which is to find my house.

Well, isn't that how statistic works?? or all mathematical model works for that matter. We try to draw a simpler form of a complicated matters in order to find our goal and in the process, we hold other variables constant.

If I think that way, then I would know that statistic is actually meant good. But and it really is a very big BUT, Sometimes statistic is just simply like crap... Sorry for you guys who love statistic to death. Why? because statistic is no more than just a tool to help you count and not with the decision. 

I give you a sample,
if there's a data of the level of happiness of 20 people correlated to their income.
Darn it, I can't make a table here? so let's make it like this. pretend there's a table, with a heading : Subject | Happiness Level | Income
A | 100 | 1               K | 100 | 1
B | 200 | 2               L | 200 | 2
C | 300 | 3               M | 300 | 3
D | 400 | 4               N | 400 | 4
E | 500 | 5               O | 500 | 5
F | 100 | 1               P | 100 | 1
G | 200 | 2               Q | 200 | 2
H | 300 | 3               R | 300 | 3
I | 400 | 4                S | 400 | 4
J | 500 | 5                T | 500 | 5

As you can see with naked eyes, I mean without using any application, there's a connection between income and happiness and it is a very good and strong correlation. Every time income increase by 100 points, happiness level increase by 1 point.

Now, you can change subject T's level of happiness from 5 to 1. You can re-regress the data, the R squared will decrease from almost 1 into only around 50% something. Just because one anomaly where subject T has income of 500 but yet still uhnappy.

So, in the statistics, there are only one data that shows anomaly from the other data which means only 5%, the R squared value can decrease up to 50%. Well maybe because that one anomaly is very extreme. But this is to show you that Statistics not always picturing the truth model. For subject A until S there are a very strong relation, but because of subject T, it decreases and the statistics just made up a formula consist of Constanta of each variables.

Well, once again I am innocently stupid and a little bit selfish, so this is just a personal opinion of someone who never like to use the Quantitative Method. Nevertheless, it's true that statistics help you to serve a concrete evidence of a relation, but then again... to fully hold on to that, I think it's not quite right. It's just numbers. What we measure and interpret as a human being is far more precious than just the result of a statistics number.

The thing that I hate the most is that when you didn't actually see a connection then you push a connection to exist by creating so many arguments into it. That's what happen with my final assignment. Aaargh.... The data is like so random, and I can't seem to find a good correlation among the variables, but I feel like I have to force a correlation there.

There are so many other variables connected to happiness other than just income. let's say is that person married or not, how's the education, how's the family condition, how's the children, how's the environment and so on and so forth.
Back to the sample of how much or a reality that I want to portrait. If I want a 100%, I should incorporate all the variables. but then again, how long my formula will be. So then I should decide which variable that needs to be controlled in order to crate a good model to picture the reality as close as possible. The problem is, not that easy to choose from that various variables, I don't know which one should I control. 

SOMEONE HELP Meeeee.........!!!!

This modules really should be taught in one whole semester rather than just in 4 weeks.
Oh Statistics.... That is another science that's not align with my way of thinking... just like economics. Aaah, thinking about my economics exam makes me sad again...
I think i should really prepare to re-take this modules next semester...


image source:
http://www.wired.com/magazine/wp-content/images/18-05/st_thompson_statistics_f.jpg