There is an important cognitive heuristic that people use to simplify complex decisions. The seeming irrationality in our behavior isn’t just a minor inconsistency, the sort of thing that shows up in laboratory experiments and breathless science journalism. It has real world importance.
Subjective estimates, disconnected from objective statistics
Consider the national debate about crime its prevention in the US justice system. Surveys of American adults have consistently shown that people think that crime is a serious problem in the in the United States and most importantly they also think that crime rates are on the rise.
Since the late 1980s almost every year a substantial majority of Americans have reported that they thought there was more crime than a year ago. Now, whether crime is a serious problem is a matter of opinion whether crime rates are on the rise is a matter of data and the data indicate that crime rates have dropped dramatically.
Since the 1980s per capita rates of both of violent crime and of property crime have dropped in half in the United States and this is part of a much much longer gradual decrease in crime rates over the decades and centuries. Stated simply people’s subjective estimates of crime rates are disconnected from the objective rates of crime.
What causes this disconnect?
Ask yourself the question: is there more crime in my area now then say 20 years ago? Now think for a moment about how you answer that question. You may have brought to mind recent crimes you’ve heard about on the news or from friends, a burglary or a similar property crime or even some violent crime like a murder or kidnapping.
Almost certainly some such crime case was featured in your news this week. News agencies know that frightening crimes like a child being kidnapped by a stranger and held for ransom will capture the public’s attention so they focus on those cases when they arise. But these very vivid crimes are also very rare. For instance each year only about 100 – 150 children in the United States are abducted in stereotypical kidnappings, that is taken away by some stranger who has malicious intent.
Such crimes seem more common because we can readily bring them to mind from memorable cases in the past from current news or even from her own instructions to our kids when we warn them about strangers.
You might argue that overestimating crime rates could be a good thing: crime is clearly a problem even at the current lower rates and fear of crime helps mobilize people, law enforcement and governmental agencies for taking action, but I want to reemphasize that our biases will affect some sort of probability judgments more than others.
So, crimes that are rare but very vivid are seen as more common than they really are so that they can trawl resources away from crimes are more common but less vivid. For example familial kidnapping is far more common than stranger kidnapping, poor security and bus stations is a more common problem than poor security at airports.
Why more resources are allocated to rare but vivid problems?
A similar sort of logic can be applied to many sorts of social problems. In many cases resources are devoted more to very rare and vivid problems compared to more common problems. I’ve talked so far about subjective probability judgments of rare events it’s now time to expand this to events of all probabilities.
We want to know for each objective probability 1%, 2%, 3% all the way to 100% what is its subjective probability weight. When we make decisions to retreat a 1% chance just like 1% or like something else? For simplicity’s sake I want to first break down the entire range of probability into three categories: low probability, medium probability, and a high probability, which I’ll call: could, might and should.
Let’s begin with low probability events, events that could happen but are likely
I’ll define this range as probabilities up to about 25%. As discussed previously in the context of insurance purchases and crime rates low probability events tend to factor more into decisions and they should.
That is an event that has an objective 1% chance of occurring could subjectively seem like it has a 5% chance of occurring, an event that has an objective 10% chance could seem like it has a 20% chance. We overestimate the likelihood of these low probability events. In general the smaller the probability the more we overestimate his likelihood.