AEM414 - notes4_Notes

AEM414 - notes4_Notes - Overconfidence Socrates is credited...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
Overconfidence Socrates is credited with saying that the only thing he knew for certain was that he knew nothing. Unfortunately this is a truth that very few of us can come to grips with. In fact, if we are not forced to use rigorous statistical analysis to form beliefs and make decisions, we will form faulty beliefs and make decisions that can have dire consequences. Rational theory struggles with the notion of beliefs and learning. Absent a good way to measure beliefs in a natural setting, or to measure the costs or speed of learning, rational models tend to assume perfect and instantaneous learning. At any given time, this means that my beliefs should be a close reflection of reality. Sad experience tells us this is not the case. In fact, individuals don’t intuitively understand statistics or probability. The theme of this lecture (and at least one to come) will be that we need rigorous statistical analysis to make informed decisions. Without it, there are a host of biases that will crop up in our thinking. In statistics we learn to estimate values, and construct confidence intervals to represent our uncertainty about the true value of the parameter we are interested in. For example, we may wish to forecast the price of some commodity at some point in the future. If we wanted to know how uncertain we were about the price, we might want to construct the 95% confidence interval. If we were to independently estimate the 95% confidence interval an infinite number of times (with independent data each time) 95% of the intervals should contain the true value. So, we are 95% certain that the true value is contained in our confidence interval. Absent all of the data necessary to predict the price, we may make an educated guess, after reading several trade journals and learning of the market conditions that are expected in the coming months. We would then need to make
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
some similar assessment of confidence. So, we may hazard a guess, forming an interval that we think should only be exceeded 25 times out of 1000, or preceded 25 times out of 1000. In general individuals fail at this task. In fact, they fail miserably. Alpert and Raiffa conducted an informal (read: no rewards) experiment that illustrated one interesting bias in judgment. They questioned a group of Harvard MBA students, eliciting a bunch of confidence intervals. This was done in 1969 when the Vietnam War was still a big issue. They were asked to guess 98% confidence intervals (exceeded or preceded by 1%) for the following values 1. The percentage of first year students responding, excluding those who never drink, who prefer bourbon to scotch. 2. The percentage of first year students responding preferring draft deferments for all graduate students while in school regardless of field of concentration. 3.
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 02/20/2009 for the course AEM 4140 taught by Professor Schulze,w. during the Fall '08 term at Cornell.

Page1 / 11

AEM414 - notes4_Notes - Overconfidence Socrates is credited...

This preview shows document pages 1 - 3. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online