We have tried and true ways to track program attendance (who, how many), and some robust methods to gather information about the value of those programs for attendees (what was most helpful, what actions likely to be taken as a result). This is easy enough: no rocket science or hard-core statistical analysis required. This data is invaluable for helping us focus on what's working.
It seems critical to remember that this data is coming from a select audience: a group of individuals who have already opted in by making the effort to attend at least one of our events. The fact that someone shows up to a program or event suggests that they've already been "sold" to some extent by our offering. (Whether we deliver what was expected is a worthy subject for a separate post.)
How then to find out more about who is not engaging in a department's services, and why not? How do we as a programming office determine if we're A) even on the radar of potential audience, or if B) we're known, but nothing that the potential audience has seen has been engaging enough to prompt action (attendance).
It only makes sense to extrapolate data from a small population to a larger group if the smaller population is a reasonably representative sample. Knowing about who shows up doesn't necessarily help us understand who doesn't show up.