Bias control: Confirmation bias

Zev Burton
Oct 27, 2022 05:28PM UTC
This post is the first blog post in our newest mini-series, “Bias Control,” where we will explore how certain biases can impact our forecasts and how to mitigate them.

Confirmation bias is a tendency to search for and favor information that already confirms our beliefs. Researcher David Perkins aptly nicknamed the concept “myside bias,” referring to a preference for “my” side of an issue. This bias has been shown to influence experts in all disciplines, contributing to poor judgment and decision-making. 

In practice, how does confirmation bias affect experts? Social scientist Philip Tetlock studied the predictions made by political experts over a two-decade period and found they were barely better than chance. This shortcoming was attributed mainly to confirmation bias, specifically the political experts’ inability to utilize new information that went against the theories on which they had built their careers. It’s not hard to see how confirmation bias can find its way into policymaking. In fact, this is an area where INFER aims to support policymakers – to minimize the risk of confirmation bias through the diverse perspectives of the crowd. Although we have different biases individually, the crowd is better calibrated and more accurate than experts since different biases cancel each other out.     

That being said, confirmation bias can significantly impact the accuracy of our individual forecasts. So, how does one spot it? Suppose you are looking for research to help you forecast the percentage of the U.S.’s renewable energy consumption that will come from biofuels in 2023. You are a strong advocate of climate reform and support the concept of alternative energy, so you start to only look for research on the adoption of biofuels. That may lead you to overestimate the percentage of energy coming from biofuels next year (what you want to have happen) and underestimate the possibility of what you don’t want to happen. This is a problem for forecasting because you only see part of the picture; you haven’t sought opposing perspectives that may contribute to the outcome. Research on confirmation bias shows that even if the evidence is shaky or inconclusive, we’re more likely to believe information that supports our preexisting beliefs and discount legitimate information if it doesn’t. 

Confirmation bias in forecasting is not limited to the information-gathering stage of analysis, and it also comes into play when recalling information as you’re making your forecast. You are up to 25% more likely to remember the information that aligns with your previous beliefs than any information that contradicts them. 

Now that you know what to look for, how do you combat confirmation bias? Since it is subconscious and ingrained into our thought process, you’re not likely to completely get rid of confirmation bias, but here are tips to mitigate it.

  1. Talk to people about your ideas. Spend time with people with diverse perspectives, and be prepared to hear opposing viewpoints. If there aren’t opposing viewpoints on an issue, appoint a team member to play "devil's advocate" when making important choices. Here, constructive disagreement will be helpful. If you have a team on INFER, bounce ideas with them– if not, consult others’ rationales and initiate discussion threads with other forecasters.
  2. Actively search out conflicting or opposing viewpoints. Typically get your information from media sources that lean politically in one direction? Try looking at a few sources with alternative perspectives on the same issue. If you need help understanding the bias of your sources (or finding sources with alternative viewpoints), check out this U.S. media bias chart for options.
  3. Think about why you may be wrong. Often, the best way to find out where our biases are is to attack them head-on by asking ourselves how our initial forecast might be wrong. On INFER, we have a designated area for each forecast to encourage you to challenge your initial forecast and explore alternative viewpoints that could subconsciously impact your forecasts.
With this knowledge in mind, you can improve your judgment and forecasting process. Good luck, and happy forecasting!

Tip: Mention someone by typing @username