Gambler's Fallacy : Why balancing forces of nature is a myth

in psychology •  7 years ago 

0.jpg
Probability is one of the most widely used concepts of mathematics, known for it universality and highly intuitive nature. Even before the theorems of probability were formally established, we were able to sense its presence and have appropriately applied in our daily lives. However, probability can be misleading. We systemically come to wrong conclusions

A classic example that reveals our limited understanding of this subject is something known as "Gambler's fallacy".

Consider a very simple example of six consecutive coin tosses. Say, all of the first outcomes land on heads. If a person were to bet a million dollars on the next outcome, what would he choose?

As an interesting study of the human psyche, a survey conducted using the above data produced results where majority chose tails as their outcome over heads. In addition to their choice, each of the participants were asked to justify why they came to such conclusions, which was extremely important for the experiment. The majority chose tails over heads, but almost none had a "correct" reason for making such choice.
1.jpg

Those who chose tails said something along the lines of "balancing forces at play", and those who said heads said that the coin was rigged. Despite being the correct answer, it was nothing more than a speculation.

The biggest shock for them, was when the answer was revealed. It was "none" which was understandably disappointing, not because the coin itself is unbiased, but because we lacked the data set. Theoretically, any one of the two choices would produce same results.

In both of the above answers, the candidates left out a very important aspect of the coin's outcome - the base rate. Gambler's fallacy is a special case of a larger psychological bias known as Base Rate neglect. The base rate is simply a large number of outcomes that reveals an object's true behavior. The larger this data, the more accurate the prediction becomes. For example, if we were to toss a coin 100 times and observe 52:48 heads to tail ratio, we don't know for a fact that the probability of heads is 52/100. Toss it a million times, and we get a closer understanding of it.

The worse answer is to consider "the balancing forces at play" - our belief that things balance out eventually, and this is simply not true. Next time, if anything happens more frequently than others, don't expect things to change just for things to balance out. A coins die or anything of that nature is unaware of its previous outcome. We should try and understand the causes or analyse the whole set of data in order to draw better conclusions.

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

Hey! If you want to be accepted by the steemit science community, I recommend quoting your sources and crediting your pictures. If you want to meet other people from STEM fields join the #steemSTEM channel on steemit.chat :)

Thanks I did!

Isn't the probability 50/50 all things being equal no matter how many times you toss it?

Not really! You see, practical coins have a slight defect, which can be understood only using extremely large data sets. 50/50 is only applicable when its an ideal unbiased coin.

Ya that's what i meant by "all things being equal".

Another interesting use case of this is how a man recorded 15,000 roulette wheel spins and used the data to game the casinos. If I remember correctly it was called the Monte Carlo system.

Exactly! Probability is amazing. Make sure to upvote and follow me for STEM based articles. :-)
Thank you.

You telling me too ensures that I never will tbh. It's nothing personal, just don't like to be told what to do haha

But just this once ;)

Also learned a new anagram "STEM" :)

Yup, Monte Carlo Simulations are fantastic!

There's 2 ways to look at probability:

  1. the theoretical probability (ie. what you would expect; a coin is 50/50)
  2. the experimental probability (ie. actually flipping the coin and counting how many heads/tails)

The idea being, like xmachina mentioned, the more flips in the experiment, the closer we'd expect it to be to the theoretical probability. Then we can also use statistical tests to check the p-value and determine if the experimental results are significantly different than the expected results.

@stats-n-lats. There are three ways to look at probability.

  1. Idealistic probability
  2. True probability
  3. Experimental probability

The idealistic probability is nothing more than an approximation for our convenience. The odds 50-50 is more appealing than say 51-49. We consider a coin to be equal on both sides, because we aren't consider the difference in weights due to different engravings. True probability is what it actually is. The value of experimental probabillity is arrived at, by performing a set of experiments. Theoretically, if we could perform this experiment an infinite number of times, we would be getting its true probability.

Ok. See if I got this right.

1: Idealistic would be the way I'm thinking of the Theoretical? As in, in a perfect situation it would be 50-50, not accounting for any influences that may affect the events. Essentially it's an assumption.

ie. Pulling a king from a deck of cards theoretically/ideally is 4/52 probability

2: Then True would be where you'd start to account for different types of variables that would actually affect what we are assuming is the ideal probability, regardless if it a big or small factor?

ie. Maybe the kings in the deck are slightly larger than the other cards. Maybe there's a crease or a card is bent. Maybe the kings are 'sticky'. So the true probability would not necessarily be 4/52 because where going from ideal to real world.

3: And then experimental is obvious.

ie. randomly pull a card from a deck and count how many times you pull a king versus not pull a king.

Does that sound about right?

@stats-n-lats Thats correct! :-)