Bias

- Whenever you're working with data, there's always a risk that your bias could impact how you analyze the data or even how you frame the results. When you're working with a team there are a number of different kinds of bias that you and your team members should be aware of. The most important bias to be cognizant of is confirmation bias. This is when you approach data with a certain expectation and it can impact how you analyze results because you're looking for a certain outcome. The way to counter this risk, in my opinion is to try not to go into an analysis with any expectations to try a few different things, when you're working with data, without trying to make your analysis fit into a certain expectation. Another major risk when you're working with data is anchoring bias. This can be a particularly problematic bias when you're working with stakeholders. Anchoring bias is the notion that you heard something first and that's what you'll anchor to. So if analysis was done in the past or a consultant for many years ago, shared a certain concept and you've got that idea in your mind, you are anchored to that first idea and it might be very difficult to overcome that. This can also happen when you're working with data where maybe you see something in your initial analysis and you really anchor to that. But it's important to be adaptive and iterative whenever you're working with data. Another major bias to watch out for is availability bias. This means that when you're performing analysis, you might be biased around using whatever is available. One of the ways to counter this is to create data you need if it doesn't exist. Our fourth critical bias is recency bias. This is the risk that you put too much weight on the most recent past and that you're ignoring long-term trends. There's a saying in finance that "The trend is your friend", and that's true, both in the short-term and the long-term. So in order to counter the risks around this bias make sure you're looking at both the long-term trend and the recent data. Another critical bias to watch out for is false consensus bias. This is a risk that you think everyone else around you believes what you do. Maybe you have an opinion about the data or what the data should look like, or you have an opinion about the business you work in or something else and you believe everyone else surely thinks what I do and that may not be the case. This is really important when communicating implications after a data analysis, to make sure that everyone sees how you got there and why and not just assuming everyone believes the same thing. Of course, there are many kinds of bias and these are just some examples, but you should at least be aware of these and make your team aware of them whenever you're working with data and sharing implications.