Statistics

Computers can churn through data, identify trends and patterns that a human couldn’t comprehend, and spit out statistical probabilities with precision. Yet, we
seem to still prefer that ol’ human touch.

Studies published in the Journal of Experimental Psychology show that we are more willing to trust the judgment of a person over a statistical model. This confidence remains even when participants are shown evidence that the person had made almost twice as many errors as the model.

I wish I read this study prior to mentioning the term ‘structural equation modelling’ at work. After conducting this advanced analysis on extensive data, I was puzzled when the data failed to convince but, rather, seemed to do the opposite. Years later and other statistics are still quoted from senior leaders, but these finely tuned statistics conducted under controlled conditions are largely forgotten.

Perhaps what underlies these outcomes is a subconscious distrust of computers? Often films depict futuristic computers with artificial intelligence, such as Hal from 2001: A Space Odyssey, as cold, cunning, and diabolical. Maybe we prefer to trust the flexibility and creativity of the human mind?

The findings published in the journal direct us to be cautious when confronted by a critical decision. How often do we make a decision based on the data, compared to doing so based on the advice from someone who seems to know what they’re talking about?

More importantly, even if you have some important facts to communicate, you may find that people are unwilling to listen. Perhaps better to leave out the stuff about the modelling and tell them you have a hunch!