The Plausibility of Probability

By now, we all have a really good understanding of what mathematical probability is; the likelihood of any event occurring within a specified series of constraints. We also know that changing any one of these variables can have a profound effect of the projected outcome and that we are best to leave a self-correcting algorithm to figure it out.

In this week's How Predictable, we will run through an compelling example of how your brain can confound plausible and probable and again, another experiment you can conduct to convince yourself of your own unreliability.

The Linda Problem

Again, Tversky and Kahneman make an appearance. Their famous problem displays what is called the Conjunctive Fallacy:

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.

Which is more probable?

1. Linda is a bank teller.
2. Linda is a bank teller and is active in the feminist movement.

For most people, the answer is easily #2. Her involvement in social justice in college and her demographic match it perfectly. Except #1 also captures all of the women in #2 as well. Adding a condition to a prediction can NEVER make the likelihood of it occurring MORE than the larger group without the condition. If you selected #2 without much thought, you now have a clear example of how your brain can confuse the two. Your bias is that more information makes you appear more correct, when in fact, more CONDITIONS lessens your predictive capability.

Your Own Private Experiment

Apologies to Gus Van Sant and his 1991 classic, but you can conduct your very own experiment on how you can be swayed. If you are not a big fan of the lottery or such, you can use that. Otherwise, select a brand of automobile that is aspirational for you but not the MOST (ie. if you love Lexus, select BMW or Genesis as your TEST)

Now rate your likelihood of WINNING the lottery on a scale of 1-10, or predict your likelihood of buying this vehicle in the next 12 months. Both should be pretty low.

Let's let the big data giants of online do their thing now! Google or Bing "Lottery winners in my area" and open the stories. Allow cookies and let every open the floodgates of media to you. Now for the car, simple ask "Why should I buy a BLANK over YOUR FAVOURITE" and read a few articles on it.

Now for the next 2-3 weeks, every time you get pushed a story or ad, you open it and read it. Let your brain take in the information without much debate.

Your Brain on Ads

You have now primed your brain with a whole bunch of new information. In the case of the lottery, you now know names and faces of local people who have won, and stores that have sold winning tickets. You start to see that people DO win, and that they are in your community. That COULD be you. Also of note is that your odds of winning remain unchanged, yet, you start to confound the two.

The automobile example will create a bit of dissonance in your thinking, that what you love is not what others are seeing, and that your approval rating for the TEST brand has certainly gone up. Your top brand may still be the top, but the competitor has closed ranks based on nothing but you hearing about their biased advertising, it becoming more available to you. It is why it costs to much to advertise; it works so very well on human minds.

In Data We Trust

You now have a pretty clear idea about just how unreliable your brain can be at predicting something. You fall prey to recency bias and conjunctive fallacies all the time, and are often your own worst noise-maker in your forecast. Your best course is to establish a very robust forecasting method that can account for as many dependent variables as possible, and let it do that very tough job of staying neutral in an unpredictable environment.