Program Logic Model

Program Logic Model

Attribution: YourVisionYourCreation LLC — yourvisionyourcreation.com Doctrine class: YVYC original — nonprofit category

Universal So-What

Every program is a claim: if we do THIS, THAT will change. Most programs never state the claim — they run on the founder's conviction that the work matters, which is fuel but not steering. The logic model forces the claim onto paper where it can be tested, funded, improved, and defended. A program that cannot draw its own causal chain is asking funders and communities to take causation on faith.

Core Doctrine

1. The Chain — Five Links, Left to Right

The discipline is in the arrows, not the boxes: each arrow is a causal claim someone should be able to challenge.

2. The If-Then Test on Every Arrow

• Every arrow reads aloud as: "IF this, THEN that, BECAUSE..." — and the BECAUSE must invoke a mechanism (skill built, barrier removed, connection made, incentive changed), not a hope
• An arrow that cannot state its mechanism is where the program's logic is thinnest — mark it, do not decorate over it
• Steal mechanisms with pride: if evidence exists that this kind of activity produces this kind of change elsewhere, cite the lineage. A logic model standing on prior evidence outfunds one standing on conviction.

3. Assumptions and External Factors — On the Page

The two rows most logic models omit, and the two evaluators check first:

• Assumptions: what must be true for the arrows to hold? (Participants can attend; trust can be established; referral partners have capacity.) Every assumption is a risk wearing a calm face — list them, and flag the load-bearing ones.
• External factors: what outside forces could move the outcomes regardless of the program? (Economy, policy changes, other services.) Naming them at design time is what makes attribution claims honest at reporting time.

4. Measurement Hooks at Design Time

• Every output and every outcome gets its measure, source, and cadence AT the model, not at the evaluation scramble: what is counted, who counts it, how often, stored where
• Outcome measures scaled to program reality: validated instruments where feasible, structured self-report where not — an honest simple measure beats an aspirational one nobody administers
• Baseline before service begins, wherever humanly possible — change cannot be shown without a starting point, and the starting point cannot be reconstructed later
• Data burden audited from the PARTICIPANT side: every form is a cost paid by the person the program serves

5. The Dosage and Dropout Honesty

• The model states its intended dose: how much program, over how long, produces the claimed outcomes — one workshop does not produce durable change, and a model claiming otherwise is advertising
• Attrition is designed for: expected completion rates stated, outcomes reported against starters AND completers — reporting completers only is survivorship laundering

6. The Living Model Rule

• The logic model is versioned; the program it describes will change, and a model that does not change with it becomes funder fiction
• Evaluation findings loop back: arrows that data supports get strengthened claims; arrows that data breaks get redesigned activities or revised claims — in the model, in writing
• The model is a communication tool at three altitudes: one-page visual for funders and boards, detailed version for staff and evaluators, plain-language version for participants and community — same chain, three translations

Common Failure Modes

Non-Negotiables

1. Every arrow states its mechanism or gets marked as the weak link.
2. Assumptions and external factors appear on the model itself.
3. Every outcome has measure, source, and cadence at design time.
4. Baseline before service — wherever possible.
5. Outcomes report starters and completers.
6. The model versions with the program.

YourVisionYourCreation LLC — yourvisionyourcreation.com Licensed under CC BY 4.0