Hypothesis testing is not bulletproof

Design thinking process has evolved over the years and most of the practitioners use data driven approach to continuously validate their approach.

When it comes to lean startup approach, we go one step further to do hypothesis testing.

Hypothesis testing is great tool, however it is not bullet-proof. Various biases can lead Lean startups to fail, despite heavily using hypothesis testing. It is very difficult to recognize our biases and somehow correlation takes a huge way in our hypothesis testing.

To understand more about co-relation and causation, let's consider a simple example in the diagram below.

Sometimes cartoons gets the message quickly.

Thus in most Lean startups, the founders are confident about their hypothesis testing; blinded to the fact that most of the times, correlation has steamrolled causation.

Thus it is very important to test several hypothesis together to validate the link of correlation or causation.

Credits:

• https://zoommetrix.com/analytics/design-thinking-process-data-driven-optimization/
• https://www.slideshare.net/ideacto/lean-ux-vs-design-thinking-lang-eng
• https://amplitude.com/blog/2017/01/19/causation-correlation
• https://scienceornot.net/2012/07/05/confusing-correlation-with-causation-rooster-syndrome/
• https://psychologenie.com/meaning-of-correlation-does-not-imply-causation-explained
• https://kaiserscience.wordpress.com/physics/the-scientific-method/correlation-and-causation/
• https://allabouthealthychoices.wordpress.com/2016/03/17/associationcorrelation-vs-causation/
• https://conversionsciences.com/correlation-causation-impact-ab-testing/