Adonis Diaries

Posts Tagged ‘Survivorship Bias

Research are plagued with studies ignoring cases that didn’t survive

Samples of population should include the cases that failed to survive, the failed projects, the patients who died taking the drug under testing and evaluation…

Otherwise, the study’s best resting place is the waste bin.

In survivorship bias you see only the survivors in any experiment, project or drug testing on patients.

In the intention-to-treat error, the failed projects show up in the wrong category that is not used to sample from.

For example, test subjects who didn’t “survive or failed” in the experiment tend to vanish from the sample.

Risk or Uncertainty? Does the difference make any difference in our behavior?

Risk involve known probabilities of what we decide to do or gamble on. For example playing in casinos, tossing a coin and studying probability in textbook.

Uncertainty is total ignorance of the outcome, and thus, people prefer to work within known probability.

Consider this experiment known as the Ellsberg Paradox:

We have tow boxes A and B.

Box A contains 50 red balls and 50 black balls. Box B contains 100 balls but the number of exact color-type is unknown.

Given the choice, people will select a red ball from box A. After this first selection, the subject will also prefer box A to select a black ball in the second trial.

Logically, if you opted to select a red ball from box A in the first trial, this should mean that you assumed black balls to be more numerous in box B. However, box A is still the preferred box to select a black ball. Why?

Are the subject not assuming anything? Did they not understand what to do?

Most probably, people have aversion for uncertainty. This is called “Ambiguity Aversion

Mostly, we lead our life navigating oceans of uncertainties, even though we have strong aversion for uncertainty.

Statistics and probability are hard to comprehend and master or even to recall their tenants after a short while.

Consequently, even incases of known probabilities (risky events), we tend to treat cases as “under uncertainty

In any case, statistics don’t stir us: people do.

A thousand of people died of famine? That’s terrible, but famine tragedy are getting pretty common.

Show the face of hungry kid dying of famine and donations pour in.

Give the story a face

Estimation is another part of risk, and we tend to systematically over estimate our chances of success.

If we think that we have talents, we start believing we are going to become a star musician, actor, painter, photographer, design model, sports…

After all, the media consistently display only stars and you think that the stars are the vast majority and pretty common to become a star  with the adequate skills and talents: “I have enough talents, and with a little luck, I’ll be  a star very shortly…

What you missed in that reasoning is that for every star, maybe 10,000 failed in their attempts. So much luck needs to be distributed for the few lucky ones, and mostly for those in developed nations and a variety of opportunities.

We don’t try to take a good look at the graveyards of all these once-promising talented hopefuls. This is called the” Survivorship Bias

Note: Daniel Ellsberg is mostly known as the one who leaked the top-secret Pentagon Papers to the press, leading to the downfall of President Nixon.





March 2021

Blog Stats

  • 1,462,104 hits

Enter your email address to subscribe to this blog and receive notifications of new posts by

Join 802 other followers

%d bloggers like this: