One of the things about being in a position of organizational power – whether you’re an internal unit head or a library director – is that people will be constantly trying to convince you to do certain things. Those things may be micro (“I need a raise”; “We need more student employees”) or  macro (“The library needs a new remote-work policy”; “The library needs to do more to promote diversity, equity, and inclusion”), or they may be somewhere in between. But you will regularly be dealing with arguments both in favor of and against doing particular things, and the people advancing those arguments are likely to offer data in support of their contentions.

And here’s something I’ve noticed about how people present data: when they use percentages but don’t tell you raw numbers, or use raw numbers but don’t tell you percentages, you should be cautious in accepting the data at face value. Because in many contexts, either the raw numbers or the percentages – when presented alone – will be misleading.

For example, consider the following statements:

• “Circulation of books on chemistry has increased by 200% in the past year.”
• “The price of Database X went up by over $1,000 with this renewal.”
• “Interlibrary loan transactions are down 20% over the past five years.”
• “Twenty of our student employees have quit, this semester alone.”

In each of the above cases, you’ve been presented with either a percentage or a raw number; and in each of those cases, your first question should be about the data point that wasn’t offered. For example:

• “How many chemistry books circulated last year, as compared to this year?”
• “What percentage increase is represented by the new price for Database X?”
• “How many ILL transactions have we conducted in each of the past five years?”
• “What percentage of our total complement of student employees is represented by those twenty?”

Why are these follow-up questions important? Because in any of the cases above, either the raw number or the percentage by itself may give you a distorted impression of what’s going on. If Database X went from $5,000 to $6,000 in price, that $1,000 increase represents a 20% jump in cost; but if went from $30,000 to $31,000, it’s a 3% increase. A thousand dollars is a thousand dollars, of course, but it does matter whether the database provider is imposing a huge price hike or a modest one.

Or consider the chemistry books. If 30 of them circulated in 2023 and 90 of them circulated in 2024, that’s a very big percentage increase – but because the real numbers can only represent the behavior of a small number of patrons (perhaps only one or two), this is a data point that offers pretty limited information on which to make collection development decisions. On the other hand, if the 200% increase in circs was an increase from 3,000 to 9,000 circulations (which would almost certainly represent the behavior of a relatively large number of patrons), that would tell you something different.

To be clear, I’m not saying that whenever someone hands you raw numbers without percentages (or vice versa) they’re trying to mislead you. More likely, they’re just offering data that supports their argument and not thinking too hard about whether the data are fully complete. Gentle follow-up questions are the best way to get to a more complete and accurate picture – and to help teach your staff about the responsible use of data.

Takeaways and Action Items

• Are there items in your to-do list that are based on data you’ve been provided by advocates for one course of action or another? Review the data and make sure you can be confident in its completeness and reliability before proceeding with that action item.
• Ask yourself how you approach data and statistics in your own decision-making and case-making to the leaders to whom you report. Are you always providing the most accurate possible picture of reality, as opposed to the version of reality that best supports your desired course of action?
• Think about how you’ll respond when someone presents a case based on what looks like a partial and perhaps misleading data point. How will you ask for more and better data without making the other person feel criticized or attacked?