Why Regression Equations Don’t Work

People want to make the perfect decision; Regression equations don’t do that.

That thought stuck in my head after we discussed it in class this week, and I shall indulge in my nerdy tendencies to discuss that today.

Regression equations are often used as a tool to make decisions, such as which stock to buy and what kind of policies to implement. Some people even use it to decide the house they buy, or the school to go to. It is billed as a tool that helps to make rational choices.

However, one fact about regression equations should make you think twice about using it: Regression ALLOWS room for error. Over a large set of decisions you make, you are guaranteed to have the least errors possible. However, any individual decision in that set can still go hopelessly far off the target. And if you’re buying stocks, one off target decision could kill you.

That’s why people are sometimes uneasy even when they are told that it is a best guess, or a good estimate. People want to be perfect. They want to be certain. It is not good enough to be told, “Although there are many good candidates, we think you have a high chance of getting into the program.” We NEED to hear, “You’re in!”

And that, is why regression equations don’t work.