It's been awhile now, partly because I moved again and have been internet-less for weeks. Not to mention that I have a long list of to-do things from work. But more importantly, I have to catch up with all the cat videos ever since I got back my intertubez.
I don't normally follow any elections which is a behaviour I'm trying to change but the recent US presidential elections kinda caught my attention a bit. But it wasn't the politics that I was interested in but the predictions made by one Nate Silver. He is a statistician who used to work out the career trajectories of baseball players based on a whole heap of data. Only recently he turned his attention to politics, employing the same techniques he used in his previous work. He predicted in 2008 that Obama would win the elections and the correctly called who would win in 49 out of 50 states. Now in the 2012 elections, he also called that Obama would win and not only that he was given a ridiculous 90% chance of winning over Romney. Again he has already correctly called 49 states with Florida still hanging (though he is most likely going to be right).
You can imagine what happens in the Romney camp when they heard that he called Obama to win the elections. They all say he does not know anything. Ironically, Nate Silver correctly predicted wins for the Republicans in other state elections and has praised him for that. I'll comment on this a little later on. Back to the story, virtually most pundits call this election to be a 50-50 race so for them to hear that Obama has a 90% chance of winning the election, most people dismissed him. Of course that doesn't matter now considering that Obama has won. Post election and pretty much all the pundits who slammed him for "bad" predictions sheepishly said that he was right.
On a slightly related note, when the technician came over to our place when we needed that telephone cable to be connected, I was talking to him. You know, about how long he has been in the business etc etc. One of the things he mentioned that the way things work around there has changed quite a lot. It doesn't matter how many jobs you do but as long as you press the right buttons and that's how you get paid, he said. He further commented that it's a stats driven world now.
The point from both these scenarios is that of the technician that came in to install our phone cable, it IS a stats driven world. People still have this idea that statistical predictions are pretty much synonymous with voodoo especially when they correctly predict something seemingly complex with a very small error to boot. Take for example Hurricane Sandy that hit most of the East Coast of the US. Statistical data combined with sophisticated weather models helped predict the trajectory of the hurricane down to within 50 miles. The only thing that they didn't quite get it right was the intensity of the hurricane. The gap between what scientists can do (with statistics included) and what the general public knows about them has gotten larger. And as what I can gather from these two scenarios, even if the public generally accepts that it is a stats driven world, it is viewed as a negative thing. They have resigned to the fact that "voodoo" now powers the world instead of what they would view as objective, measurable and somewhat naive data.
You don't have to go far to see that stats do indeed power the world. Apart from the elections and Hurricane Sandy, things like SEO, data mining for more effective targeted advertising, supermarket aisle rearrangements, behavioural economics and more. All of this is to tap into the subconscious behaviour of a collective and leverage on that. And it only can be done via a powerful grasp of statistics.
I once had a lecturer that said, we are now moving from the continuous age to the discrete age and it is only with a strong grasp of statistics that we can start to understand the strange phenomena around us. Mathematics that started in the 1700s all the way to about the early 1900s all had the idea that things in nature can be described as a continuum, everything is smooth, predictable and contains only one answer*. From the 1900s, we get things like quantum mechanics , statistics, probability, data communications and modern economics where there they don't work with exact answers^ but trends. And even then, it's trends given the right data and assumptions. It will take a long time before we get our head around the fact that statistics is a powerful tool to use and more importantly how to use it properly.
I did say I was going to comment from an earlier paragraph. Despite all of this, we as humans still practice the method that we accept what we want to hear. It is usually from the lack of understanding about the underlying process or just plain ignorant about the data. After many thousands of years, we humans still have the same problem, pride. Even as we advanced as the most technologically advanced species on the planet, our moral compass is still ultimately flawed.
*this is not technically correct but it is the general view
^again this is not technically correct but you get the idea already
Subscribe to:
Posts (Atom)
Posts
