Skip to content

Irrationality 12,13: Making the Wrong Connections, Mistaken Connections in Medicine

01/22/2013

Chapter 12: Making the Wrong Connections

Covers incorrectly inferring causation or association. Not only do we need to know how often two things occur together, we need to know how often neither occurs, and how often one occurs and the other doesn’t. Otherwise we can’t infer a real association. Also, there may be many things that are confounding variables. I’m thinking gun debate (“In Puerto Rico, there are no guns, and people kill each other all the time!” “In South Korea, there are no guns, and it is so peaceful!”) Yup. No group escapes this craziness.

moral
1. If you want to determine whether one event is associated with another, never attempt to keep the co-occurrence of the events in your head. Maintain a written tally of the four possibilities set out on page 115.
2. Remember that A is only associated with B if B occurs a higher percentage of the time in the presence of A than in its absence.
3. Pay particular attention to negative cases.
4. Be careful not to associate things together because of your expectations or because they are unusual.
5. Flee any psychologist or psychiatrist who asks you to do a Rorschach test: he does not know his job (2419).

Chapter 13:  Mistaken Connections in Medicine

He doesn’t name it as such, but this is Bayesian probability theory. If a breast cancer test has a 90% chance of detecting the existence of the cancer, and 90% chance of showing no cancer, a positive result does not mean that the chances of having cancer are 90%. Doctors screw this up all the time. If the base rate of breast cancer is 1/1000, then there will be one true positive and 100 false positives if all were tested. 1/101 is about 1%, not 90%. The guess was almost 100 times off.

moral

1. If you are a doctor, learn some elementary probability theory.
2. If you are a patient, set your doctor a simple test on elementary probability theory.
3. In order not to foul up medical research even further, nobody without a good knowledge of statistics, probability theory and experimental design should become the editor of a medical journal, even at the expense of a vast reduction in the number of journals (2575).

Advertisements
Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: