Theory of Change: mapping how a programme is supposed to work

A pathway of outcomes, not activities. The chain is written in changes — in behaviour, condition, capacity — with activities attached as what triggers each change. “Train teachers” is an activity; “teachers use formative assessment weekly” is the outcome the training must cause.

Assumptions, stated at the links. Every arrow hides beliefs: that trained teachers have time to apply training, that attendance responds to school quality, that local authorities will keep funding. The theory’s value concentrates in these — they are what evidence can support or break.

The ceiling of accountability: the point on the pathway beyond which the programme claims contribution, not control. Declaring it prevents both overclaiming at the top and blame games at the bottom.

Preconditions and context: what must already hold for the pathway to operate at all — the difference between a theory of the programme and a theory of the programme in this system.

The approach grew from theory-based evaluation in the 1990s — Carol Weiss’s work on surfacing the assumptions of complex community initiatives (Aspen Institute) is the usual origin citation — and matured into today’s workshop-built pathway maps.

For a rigorous application protocol in trial and programme contexts, De Silva and colleagues’ Theory of Change: a theory-driven approach to enhance the Medical Research Council’s framework for complex interventions is the practical standard, including how to run development workshops and integrate the theory into evaluation design.

Breuer and colleagues’ systematic review, Using theory of change to design and evaluate public health interventions, documents both the approach’s spread and its recurring weakness in practice: theories built once, shelved, and never revised against evidence.

• At design time, for any intervention with a chain longer than one link. If impact depends on intermediate behaviour change, the theory is where that dependence becomes visible and plannable.
• At evaluation time, even retrospectively. An evaluation without a programme theory measures whatever is measurable; with one, it tests the links that matter — including the assumption most likely to fail.
• For portfolios and strategies, where a shared theory exposes which grants actually address which links.
• Not as decoration. If no one intends to test assumptions or revise the theory, the workshop cost buys a wall chart; be honest about that trade before booking the facilitator.

A six-month employability programme (training, mentoring, employer placements) built its theory backwards from “sustained employment at 12 months”. The workshop — implementers, two employers, four programme alumni — produced a pathway whose contested link surfaced immediately: staff believed skills gaps drove youth unemployment; alumni argued the binding constraint was employers’ screening practices. Both mechanisms went into the theory, tagged contested.

The evaluation was designed against that disagreement: skills assessments and employer-side interviews sat on the rival links. Midline evidence broke the tie — completion and skills gains were high and flat across employment outcomes, while placement conversion varied fourfold by employer, tracking whether the employer had adjusted its screening. The assumption log recorded the skills-deficit link as unsupported for this population.

The revision was structural: resources shifted from extending training to employer engagement and placement design, and the theory’s version two carried the evidence trail. At endline, the programme could show not only outcomes but a defensible account of why it had changed shape — which is what a theory of change is for.

• Activities in boxes, renamed outcomes. “Workshops delivered” with an arrow to “awareness raised” is a work plan wearing theory’s clothes.
• Assumption-free arrows. A pathway with no stated assumptions has hidden them, not resolved them.
• The everything diagram. Forty boxes and all plausible arrows — unfalsifiable and unusable; a theory is a commitment to the pathway you believe most.
• One theory, never revised. The Breuer review’s central finding: built for the proposal, untouched by the evidence.
• Consensus theatre. Smoothing genuine disagreement out of the workshop; contested links are the theory’s most valuable content.
• No ceiling of accountability — leading to impact claims the design cannot support and blame the programme does not deserve.

A theory of change is only as good as the causal knowledge behind it; in weak-evidence fields the pathway encodes plausible belief, and its value is disciplining that belief, not certifying it.

Linear box-and-arrow form struggles with genuinely complex systems — feedback loops, emergence, adaptive opposition. The remedy is honesty about grain: theories for complex programmes should model fewer links with more attention to context and revision, or pair with methods built for complexity.

And the approach cannot by itself establish causation: it tells the evaluation where to look. Attribution needs the evaluation designs — experimental, quasi-experimental, or theory-based causal methods — that the theory makes targeted.

The logic model is the compact, linear cousin — a results summary rather than an argued causal theory.

The logical framework approach turns a pathway into a management matrix with indicators and risks.

Realist evaluation deepens single links into context–mechanism–outcome configurations.

Contribution analysis uses the theory of change as the spine of a causal claim assessed against evidence and rivals.

A used theory of change generates qualitative evidence continuously — workshop transcripts, stakeholder interviews on assumptions, implementation notes — and the analytic task is mapping that material onto specific links. Evidano codes interview and document corpora against the theory’s assumptions as a codebook, so each link accumulates its supporting and contradicting evidence with quotes attached, and revision meetings start from an evidence table rather than recollection.

The theory itself — the backwards mapping, the contested links, the revision decisions — is a negotiation among people who know the programme. The tool keeps score; the stakeholders own the argument.