3.3 Logic Models and Program Theory

Using a logic model to test whether the plan makes sense

A logic model is a planning tool that displays the expected relationship between what a program invests, what it does, what it produces, and what changes as a result. It is more than a grant attachment: it lets the planner confirm that assessment findings, objectives, strategies, and evaluation measures fit together before time and money are spent. CHES items often hand you a partial model and ask which component is missing or misplaced.

The standard columns

A logic model reads left to right: inputs → activities → outputs → short-term outcomes → intermediate outcomes → long-term outcomes.

• Inputs are resources: staff, volunteers, money, curriculum, space, partner relationships, transportation support, translation, and data systems.
• Activities are what the program does: training peer leaders, running skill-practice sessions, conducting outreach, revising a referral workflow.
• Outputs are direct products that count what happened: classes delivered, text reminders sent, families enrolled, sites trained. Outputs do not prove change; they show whether the planned work occurred at the expected dose and reach. The exam often uses outputs as distractors when the stem asks for an outcome.
• Outcomes describe change in participants, organizations, communities, or systems.

Short-term outcomes are awareness, knowledge, attitudes, skills, self-efficacy, intentions, or readiness. Intermediate outcomes are behavior change, service use, policy adoption, environmental support, or shifted norms. Long-term outcomes are reduced disease burden, better quality of life, fewer injuries, or lower risk indicators. The longer the time horizon, the more careful the planner must be about attribution — you cannot credibly claim a one-semester class lowered five-year mortality.

Program theory: the reasoning behind the arrows

Program theory is the explanation for why the activities should move the chosen determinants. If assessment shows new parents lack confidence with safe-sleep practices, program theory emphasizes demonstration, guided practice, feedback, and social reinforcement. If healthy food is unavailable at meetings, the theory points to environmental support and organizational policy rather than individual education alone.

Assumptions and external factors belong in the model too. Assumptions are beliefs the plan depends on — participants trust promotores, schools allow class time, clinics share referral data. External factors are outside program control — bus route changes, staffing shortages, weather, competing events, new state policy. Naming both leads to more realistic objectives and built-in contingencies.

Spotting weak alignment

A logic model exposes gaps. Suppose the goal is to raise colorectal cancer screening, but the only activity distributes a general cancer-awareness flyer. The model reveals a gulf between low-intensity activity and a behavior-change goal. A stronger plan adds patient navigation, reminder calls, provider prompts, transportation support, and culturally appropriate messaging — chosen from assessment findings.

The model also drives evaluation: process evaluation measures inputs, activities, and outputs; outcome evaluation measures short- and intermediate changes; impact evaluation addresses long-term health or system effects. Match the indicator to the component — attendance logs measure an output, a verified appointment record measures screening completion (an intermediate outcome).

When answering, ask where the proposed item belongs: staff time is an input; a workshop is an activity; number of workshops completed is an output; increased skill is short-term; sustained behavior is intermediate; improved population health is long-term. This single classification separates correct answers from plausible but misplaced choices.

Building a logic model from assessment findings

A logic model is usually built backward from the desired outcome, then checked forward. Start with the long-term outcome the goal implies, work back to the intermediate behavior change that would produce it, then to the short-term knowledge or skill change, then to the activities that could create that change, and finally to the inputs those activities require. Building backward keeps activities tied to outcomes instead of letting a favorite activity define the program.

Once drafted, read the model forward and ask at each arrow, "Is this enough to cause the next box?" If a single flyer is expected to produce sustained behavior change, the forward read exposes the gap.

Common logic-model errors the exam tests

• Output mistaken for outcome: counting attendees and calling it impact. Attendance is reach, not change.
• Missing intermediate step: jumping from a class directly to lowered mortality with nothing in between.
• Input listed as activity: treating "funding" or "staff" as something the program does rather than something it uses.
• Indicator mismatched to component: measuring a long-term outcome to judge a short-term objective, or vice versa.
• Ignored assumptions: the model only works if schools allow class time or clinics share data, yet those conditions are never named.

Logic models also support program improvement during and after a cycle. If outputs were met (sessions delivered, people reached) but short-term outcomes did not move, the issue is likely program theory — the activities do not change the determinant. If outputs fell short, the issue is implementation, not theory. This output-versus-outcome diagnosis is a recurring CHES theme and explains why building monitoring into the model from the start, in Area II, pays off when interpreting results in Area IV evaluation.