Building a simple logic model for a counseling program

START HERE

The essential idea

Build a logic model whose arrows contain explicit assumptions. Separate resources, activities, service counts, and changes in participants. Use a failed link to choose a program revision and an evaluation question.

BUILD THE REASONING · 01

Start with a defined need and a decision

A logic model is a compact explanation of how a program expects its work to produce an outcome. It is useful only if the components are specific enough to question. “Provide support so people feel better” hides the service, the proposed mechanism, and the conditions needed. Begin with the people the program intends to serve, the problem they identify, the context, and the decision that the model will inform. A needs statement should not assume that participants lack motivation when resources or access may be the limiting factors.

Consider an original fictional six-week program for adults who recently became family caregivers. Participants report difficulty dividing tasks, locating respite options, and protecting time for their own health appointments. The program proposes brief group learning and individual navigation support. Its initial decision is whether this combination helps caregivers obtain a workable support arrangement, and what needs revision before another cohort. The model is not evidence that these activities already work; it is an argument to test.

Invite perspectives from intended participants, including people who cannot attend, and from relevant service partners. Caregivers may define useful support differently from program staff. One wants transport, another wants help negotiating a task, and another wants information about an existing benefit. A shared model can retain different pathways while naming the common aim. It should also identify the boundaries: the program is not promising to supply respite places that local services do not have.

BUILD THE REASONING · 02

Write every component so it can be observed

Inputs are resources available to do the work: staff time, skills, accessible space, an accurate directory, and coordination capacity. Activities are what staff and participants do with those resources. Outputs are direct products or counts, such as sessions held or navigation contacts completed. Outcomes are changes in knowledge, skill, behavior, access, functioning, or conditions. Time alone does not distinguish an output from an outcome; twelve weeks of attendance remains attendance.

The fictional program’s active learning component is a structured task-sharing conversation: identify one needed task, formulate a specific request, negotiate a feasible arrangement, and review what happened. Navigation support addresses resource information and contact barriers. These components target different mechanisms. If the model merely calls both “support,” the evaluator cannot tell whether a disappointing result reflects unclear requests, unavailable services, weak navigation, or some combination.

Write the desired outcome without assuming every participant has the same goal. A workable support arrangement might be one agreed task covered by another person, an accessible service appointment, or a revised plan that reduces an unrealistic burden. Record the participant’s chosen goal and what counts as progress. A generic satisfaction item can describe experience but cannot establish that a support arrangement exists or is usable.

BUILD THE REASONING · 03

Put assumptions and alternative explanations on the arrows

Inputs do not automatically become activities: trained staff may lack protected time. Activities do not automatically produce outcomes: participants may not understand the examples or have a chance to rehearse. A skill does not automatically create support: relatives or services may lack capacity. Write at least one assumption between each major link and a question that could reveal its failure. This transforms the diagram from a promotional flowchart into an evaluation tool.

For the task-sharing pathway, the assumption is not that every relative will agree. It is that the participant can identify an appropriate person or service, make a feasible request, and use the response to create or revise an arrangement. Cultural meanings, family power, safety, work schedules, language, and service eligibility may change what is feasible. A refusal may be useful information about capacity rather than evidence that the participant performed the skill incorrectly.

Identify external influences and possible unintended effects. A new public benefit could improve support independently of the program; a service closure could worsen access despite good delivery. Participation could also add travel burden or create pressure to disclose family details. Include such possibilities in the model and evaluation. A program that measures only the expected success path will be poorly equipped to explain a mixed or harmful result.

BUILD THE REASONING · 04

Use the model to choose measures and decisions

Select measures at the links most important to the current decision, rather than measuring everything because it appears on the diagram. An early cohort may need reach, component delivery, acceptability, new-case skill, and short follow-up on usable support. Longer-term health outcomes may be important but unrealistic for a short feasibility phase. Explain the phase-specific scope so a successful pilot is not advertised as proof of long-term clinical impact.

For each indicator specify who or what is counted, when, the data source, who collects it, and how missing information is handled. Use a participant-defined goal alongside a common skill rubric so the evaluation can compare the taught process without erasing individual priorities. Obtain the required institutional oversight determination, minimize collected private information, and plan how participants can report negative experiences without feeling that services depend on praise.

At review, locate the earliest weak link and examine downstream consequences. If people cannot attend because there is no respite during group time, improving the handout is unlikely to solve reach. If attendance and skill improve but resource access does not, revise navigation or partnerships within authorized scope and reassess available capacity. A model can be updated when evidence changes; retain a dated record of why the revision was made instead of redrawing the original theory after the result to make it look correct all along.

FROM CONCEPT TO DECISION

The group taught the skill, but the support was unavailable

1. 1

   Separate the measured links

   The new-case rubric supports a proximal skill finding among attendees. It does not demonstrate that services have capacity or that every participant has a safe, willing support network. The outcome problem occurs after the skill link.

2. 2

   Use successes and failures to refine the model

   The two task-sharing successes suggest one route that can work for some participants. Waiting lists identify a resource constraint on another route. Do not generalize the successful family arrangement to people without comparable contacts.

3. 3

   Choose a targeted revision

   Update the directory with real availability, make eligibility and waiting time explicit, and add planning for alternatives where appropriate. Preserve useful rehearsal while avoiding the claim that more practice alone will create respite capacity.

4. 4

   Set the next evaluation question

   Ask whether the revised navigation information increases feasible attempts and usable arrangements without adding burden. Track whom it still fails to reach and keep longer-term sustainability separate from an initial contact.

MAKE THE REASONING YOURS

Work through a new situation

READ WITH A PURPOSE

Best references

These readings deepen the specific models, evidence, and decisions in this lesson.

1. 01
   CDC Program Evaluation Framework, 2024

   Use the framework to connect context, program description, evaluation design, evidence, conclusions, and use. Its public-health examples require adaptation to a counseling setting.

2. 02
   CDC: Developing and Using a Logic Model

   Use the official logic-model guidance to inspect resources, activities, outputs, outcomes, and obstacles in the original caregiver example.

3. 03
   AHRQ: Elements of an Evaluation Plan

   Translate the model into focused questions, criteria, data collection, and decisions that fit available resources.