Tuesday, October 25, 2016

How do predictions affect results?

Upon my first Google search, this question cracked open an immediate and oppressive flood of statistics.
Which was not something I was attempting to look into when I asked how predictions affect results. My mind was thinking in the humanities realm: how self-fulfilling prophecy seems to jeopardize the future or create an excuse for future actions.

I got my knowledge question from a situation I couldn't grasp: Trump claims that the election will be rigged in Hillary's favor. When I heard this, I thought how incredibly unfair of him to cry wolf before voting even begins. Just saying that he'll be defeated gives him lee-way to be able to walk away from the election no matter the results with his pride still intact. And that to me seems completely unjust. He is predicting that he'll lose, so when the results come out his victory will be glorified or his loss will already have it's blame on Hillary.

But when this question was discussed in TOK, most people leaned towards a mathematical interpretation of it, just like Google did. While sifting through articles on Google I came across the term predictive
Linear regression, showing a predictive model. 
modelling
, which is basically the statistical process of predicting outcomes. The goal of predictive modelling is to determine the probability of an outcome when given certain data inputs. One interesting application of this methodology is uplift modelling, also known as incremental modelling. This technique looks at how different input variables would change the model and change the outcome. So we can see how different actions taken can cause certain results, and we can set up actual models to run through different predictions over and over.

Statisticians use these things to analyze a whole slew of different things, but that's all still looking at
predictions and results in a linear, objective way. My question remains unanswered, because it calls for
something more than mathematical probability.

On a second Google search, I found something that hints a little nearer to an answer in an article about the psychology of self-prediction. The article discusses the psychological tracking of self-predictions and the actual results of them, based on a theoretical framework presented by Koehler and Poon in 2006. They said that people have a natural tendency to predict their future actions with an optimism-bias, to where the higher their intent the moment they predict, the higher the probability they will actually complete the task. But with that optimism-bias comes the risk of under-weighting possible obstructions, such has a loss of intent or situational barriers.

The deeper I read, the more and more I saw the same thread of statistical reasoning within the brain. It seems as though you could almost take a skeleton predictive model, and plug in our brains to one end, and out the other end you could see the psychology of self-prediction. Incredibly interesting, but still not exactly talking about how this thing happens and what it affects.