How data analysts experience cognitive biases and how to recognise them: Part 1

29 January 2018

Cognitive biases: We all have them and we regularly encounter them in everyday life. But how do data analysts actually experience these biases? And how can you recognise them? In this series we will discuss cognitive biases in the context of analysts. Each blog picks out 3 biases from 20 categories.

In this blog, the 3 biases are from the “Too Much Information” spectrum: Self-relevance effect, Confirmation bias and the Bias blind spot.

Self-Relevance effect

Basically, this bias means that the more you can relate something to yourself, the more you notice it. Let’s say you are a huge cat lover and you subscribe to all kinds of animal publications. If you had to make a churn model, e.g. for a publisher, then you will notice the variable “Owns a cat” more quickly. After all, you have a cat yourself and you subscribe to these publications too. Therefore you are more likely to see this variable as an important variable, although other variables, such as “Owns a dog”, may be just as influential on churn probability. However, these variables are less relevant to you and so you notice these variables less.

Confirmation bias

If you believe that imputing missing values based on the average is better than imputing missing values based on the median, then you will tend to look for information that confirms this rather than information that refutes your argument.

Or if you think Access is a better program than SQL Server, you will look for information to confirm your argument. If you end up with information that doesn’t confirm your ideas or you end up with information that contradicts what you know already, you are less inclined to accept it and tend to disregard this information.

Bias blind spot

This bias says that you can’t actually recognise biases in yourself. It is much easier to recognise that others are suffering from a bias than see a bias in yourself. If you notice that one of your fellow analysts always chooses Access or to impute missing values based on the average, you attribute it to the confirmation bias. Or if you notice that your cat-loving colleague has put the variable “Owns a cat” in his analysis, you would attribute that to the self-relevance effect.

However, the fact that you only ever do decision trees – “because they’re just better” – or that you put the variable “travel distance between home and work” in your model – since you have to travel for 2 hours to get to work every day – of course that’s not bias. No, these are informed decisions.

Also read the other parts:

Part 2: Availability heuristic, Contrast effect and Clustering illusion.

Part 3: Essentialism, the Positivity Effect and the Appeal to Probability Fallacy.

Part 4: Curse of Knowledge, the Hindsight Bias and the Hard-Easy Effect.

Part 5: Sunk Cost Effect, the IKEA effect and the Google Effect.

Part 6: Ambiguity Bias, Appeal to Novelty and False-Consensus Effect.

Contact

Do you want to know more about this subject? Please contact us on +31 (0)33 258 28 30 or info@cmotions.nl.