A fictional transport program
Suppose a community program offers rides to appointments. Process questions include how many rides were requested, how many were completed, how long people waited, and who could not access the service.
Outcome questions include whether missed appointments decreased or whether people found care easier to reach. These questions concern changes the rides were intended to support, rather than the number of rides alone.
A result can answer only part of the question
| Finding | What it supports | What remains unresolved |
|---|---|---|
| 400 rides completed | The service delivered recorded rides | Whether access improved for the intended population |
| Missed appointments decreased | The measured outcome changed | How much change was caused by the program |
| Users report easier travel | Users describe a benefit | Whether nonusers or excluded people had similar access |
CDC's evaluation framework distinguishes implementation and outcome questions and connects evaluation design to intended use. The table illustrates why those questions need separate measures.
When the result is disappointing
If few rides occurred, the issue might involve awareness, eligibility, scheduling, or demand. If many rides occurred but missed appointments did not change, the outcome measure, comparison, and program theory deserve examination.
Neither result alone identifies a single cause. The purpose of process information is partly to make an outcome result interpretable.
Add attribution only when the design supports it
A before-and-after decrease may coincide with a new clinic schedule, another transport service, or changes in who had appointments. An evaluation designed to estimate the program's effect needs a credible comparison and attention to those alternatives.
A useful report can therefore say both that delivery succeeded and that causal impact remains uncertain. Those are distinct findings, each potentially useful for deciding what to maintain, investigate, or change.
