I summarized the common misinterpretions of statistical results due to biases after reading Thinking Fast and Slow by Daniel Kahneman.
According to Daniel Kahneman, people may misinterpret statistical results due to some instinctive biases. Since statistics is the basic subject of many subjects and itself represents fact and truth, the misinterpretation of statistical results may obstruct both the decision-making process and the development of science. It highly shocks me that, however, there are so many common biases related to the interpretation of statistical results. As a result, I summarized those biases as following to help myself better recognize them and therefore avoid them.
1. Cause explanation of chance event
Many statistical results are due to chance, including accidents of sampling. However, our system 1 automatically and effortlessly identifies causal connections between events, even if the connection is spurious. In fact, causal explanations of chance events must be wrong.
2. The law of small numbers
If the sample size is not big enough, extreme outcomes are very likely to be found. Those extreme outcomes is just a statistical result, but people has an instinct to assign some reasons to explain the results and misunderstand the randomness.
3. The source of the information
Naturally, people pay more attention to the content of messages than to information about their reliability. In addition, system 1 is not prone to doubt and system 2 can maintain incompatible possibilities at the same time although it is capable of doubt. As a result, people may easily trust some information without justifying the source of that information.
4. Neglected statistical base rate
System 1 can deal with stories in which the elements are causally linked but it is weak in statistical reasoning. For predictions, people need to consider the base rate related to a case. However, if the base rate is just a statistical fact, people tend to underweight or neglect the statistical base rate because they find nothing to serve their hungry for causal stories. In contrast, statistical results with a causal interpretation have a stronger effect on our thinking than non-causal information.
5. Individual case trump general statistical result
General statistical fact will not change long-held beliefs rotted in personal experience. However, surprising individual cases have a powerful impact because the incongruity must be resolved and embedded in a causal story. “You are more likely to learn something by finding surprises in your own behavior than by hearing surprising facts about people in general.”
6. Regression to the mean
Whenever the correlation between two factors is imperfect, there will be regression to the mean. However, people usually confuse mere correlation of a statistical result with causation, and adopt a problematic reward system.
Kaitan Sun Jun 30, 2018
תגובות