Frequency Distributions
Your year-end report says the organization ran 48 events last year with an average attendance of 45 people. That sounds like a solid baseline for planning next year's calendar. But when an intern pulls up every event and tallies how many had fewer than 20 attendees, how many had 20 to 40, how many had 40 to 60, and so on, a pattern jumps out. There's a cluster of events around 15 to 25 people (the regular community meetings nobody promotes) and another cluster around 60 to 80 (the flagship events with press coverage and paid ads). Almost nothing sits near 45. The "average event" doesn't exist.
What the intern built is a frequency distribution, a count of how many observations fall into each group. It's the most basic tool in statistics, and the one people skip most often. Before computing a mean, a median, or a standard deviation, a frequency distribution shows you what your data actually looks like.
The concept is straightforward. Take any collection of numbers, whether donation amounts, email open rates, or volunteer shift hours. Decide on a set of ranges, called bins. Then count how many values land in each bin. When you draw those counts as bars, you get a histogram. It looks like a bar chart, but the horizontal axis represents a continuous range, and each bar's height tells you how many data points fell in that slice.
The part that trips people up is the bins themselves. The number of bins you choose changes the picture dramatically. Narrow bins (say, €1 increments for donations) show every ripple and bump, which can make it hard to see broader patterns. Wide bins (€100 increments) smooth everything out, which can hide important detail. There's no single correct answer, but the goal is to find a bin width that reveals the structure without drowning in noise or papering over it.
Everything we've covered in the first four entries describes features of a frequency distribution. The mode is the tallest bar. The standard deviation measures how wide the bars spread. Percentiles mark positions along the horizontal axis. Skewness describes whether the bars trail off to one side. Those summary statistics are useful shortcuts, but they all try to compress what a frequency distribution shows you directly. When in doubt, look at the picture before reaching for a number.
In fundraising, plotting the frequency distribution of gift sizes is often the first step toward smarter segmentation. You might discover your donors form two distinct clusters rather than one smooth curve, which means a single "upgrade ask" strategy won't work for both groups. In digital campaigning, plotting the distribution of email open rates across all sends in a quarter can reveal whether your "average open rate" reflects steady performance or wild swings between viral hits and duds. In program evaluation, a frequency distribution of participant outcomes can expose whether a program lifts everyone a little or transforms some participants while leaving others unchanged. The average hides that distinction. The distribution does not.
Grant writers often report means because funders ask for them. But attaching a simple histogram alongside the average makes a report substantially more credible. A funder can see at a glance whether "average improvement of 12 points" means most people improved by 10 to 14, or whether it means half improved by 25 and half didn't change at all. Those are very different programs, and the frequency distribution makes the difference obvious.
Before you compute a single summary statistic, count what's there. A frequency distribution is the map. Everything else is a description of the map.
See It
Drag the slider to change the bin width. Watch how the same donation data tells different stories depending on how you group it.
Reflect
Think about a metric your organization reports regularly. Have you ever seen the full distribution, or do you only see the average? What might you learn if you plotted every value as a histogram?
When you choose how to group data in a report (by gift size, by age bracket, by attendance tier), how do you decide where the boundaries go? Are those groupings revealing patterns, or hiding them?