how do I give an unbiased recommendation to my analysis?
Illustration by Catherine Madden from storytelling with data: Let’s Practice!
Through virtual and in-person workshops around the globe, we have taught tens of thousands of people how to communicate effectively with data. This series captures some of the noteworthy questions we hear during those sessions—and our answers.
How do I avoid misleading my audience—and introducing bias—when providing a recommendation to my audience?
When shifting from exploratory to explanatory analysis, you should always think about what you want your audience to do with the information. If you can’t clearly articulate a recommended action, then pause to think critically about whether you truly need to communicate in the first place.
This can certainly be uncomfortable territory. Many of us were told—especially early in our careers—that the audience knows better than us, and therefore it’s not “our place” to provide a recommendation. Or perhaps your field of study emphasized the importance of being “objective” above all other considerations, and that anything less would mean that you were misleading or inappropriately editorializing when communicating with data.
The reality is that unbiased data is a myth. Throughout the entire analytical process—from identifying, retrieving, cleaning, organizing, and analyzing data—we inherently introduce bias just by the nature of the decisions we make in each step.
When communicating with data, much of our responsibility is to point out what aspects of that information will be meaningful to our specific audience. Furthermore we want to help start the subsequent conversation about what action needs to be taken. To accomplish that, we have to identify some potential next steps. Those options can be presented alongside the data in a variety of ways, and the flow of that communication will depend on your audience and your specific scenario. Omitting any recommendation at all, however, for fear of seeming biased, is misguided.
Providing a recommendation is not the same thing as misleading your audience. In your communication, you still have to establish credibility and be able to back up your findings with data and answer their questions.
Dr. Steven Franconeri, director of the Visual Thinking Laboratory at Northwestern University, agrees and instills the same opinion in his students. Hear him elaborate in this short clip from a recent podcast episode with Cole:
In summary, resist the urge to skip a recommendation out of fear that you’re introducing bias or being misleading. Instead, add value by saying “here’s what I think you should know or conclude.” Your audience may disagree, but at least they’ve engaged with the data and started a conversation. After all, the end goal of your hard work is to drive action and positive change; if the conversations that inspire those actions never get off the ground, then all that effort could be for naught.
Illustration by Catherine Madden from storytelling with data: Let’s Practice!
Here are some additional resources to help you build your skills in identifying the action for your audience and tailoring your communication to their needs:
Exercise: identify the action for YOUR project
Exercise: put the data into words
Video: communicating to technical vs non-technical audiences