The availability heuristic is a mental shortcut that occurs when people make judgments based on the ease with which examples come to mind. Although this heuristic can be beneficial, such as when it helps us recall examples of a dangerous activity to avoid, sometimes it leads to availability bias, where we’re affected more strongly by the ease of retrieval than by the content retrieved.

In his thought-provoking book “Thinking, Fast and Slow,” Daniel Kahneman explained how availability bias works by recounting an experiment where different groups of college students were asked to rate a course they had taken the previous semester by listing ways to improve the course — while varying the number of improvements that different groups were required to list.

Counterintuitively, students in the group required to list more necessary improvements gave the course a higher rating, whereas students in the group required to list fewer necessary improvements gave the course a lower rating.

According to Kahneman, the extra cognitive effort expended by the students required to list more improvements biased them into believing it was difficult to list necessary improvements, leading them to conclude that the course didn’t need much improvement, and conversely, the little cognitive effort expended by the students required to list few improvements biased them into concluding, since it was so easy to list necessary improvements, that the course obviously needed improvement.

This is counterintuitive because you’d think that the students would rate the course based on an assessment of the information retrieved from their memory regardless of how easy that information was to retrieve. It would have made more sense for the course to be rated higher for needing fewer improvements, but availability bias lead the students to the opposite conclusion.

Availability bias can also affect an organization’s discussions about the need for data quality improvement.

If you asked stakeholders to rate the organization’s data quality by listing business-impacting incidents of poor data quality, would they reach a different conclusion if you asked them to list one incident versus asking them to list at least 10 incidents?

In my experience, an event where poor data quality negatively impacted the organization, such as a regulatory compliance failure, is often easily dismissed by stakeholders as an isolated incident to be corrected by a one-time data cleansing project.

But would forcing stakeholders to list 10 business-impacting incidents of poor data quality make them concede that data quality improvement should be supported by an ongoing program? Or would the extra cognitive effort bias them into concluding, since it was so difficult to list 10 incidents, that the organization’s data quality doesn’t really need much improvement?

I think that the availability heuristic helps explain why most organizations easily approve reactive data cleansing projects, and availability bias helps explain why most organizations usually resist proactively initiating a data quality improvement program.

This post originally appeared at OCDQ Blog.