Skip to content

Theory of Change

Design Method · All design methods
Evidence · no claims cited

Description

A Theory of Change works backward from a long-term outcome, asking at each step what else would have to be true for the next step to follow. The answers divide into two kinds: preconditions the design can produce, which become its own chain of intended effects, and assumptions it depends on and does not control — a supportive manager, a policy that permits the new practice, a labour market that rewards the credential. Both are written down, and the second kind is what distinguishes the method from a Logic Model.

Originating in the evaluation of comprehensive community initiatives, where no single intervention could plausibly be credited with the outcome, it was built for exactly the situation in which a programme cannot get there alone and needs to say so.

Design Implications

The method's contribution is not the diagram but the assumptions register. A learning design that teaches a capability nobody at work is permitted to use has a dependency it does not hold; stated, that is a finding addressed to the sponsor, and unstated the course looks like the thing that failed. Weiss's original argument for theory-based evaluation was precisely this — that initiatives fail for two distinguishable reasons, a wrong theory and a poor implementation, and that without a stated theory an evaluation cannot tell which happened.

Its second contribution is that it can hold a disagreement. Working backward with several parties in the room surfaces the cases where an employer's desired outcome and a learner's are not the same outcome, and the honest artifact records both rather than averaging them into one. Resolving that tension is rarely the design's to do; recording it is.

The costs are real and frequently unjustified. Where one party wants one thing and the chain is short, a Logic Model produces the same links for a fraction of the effort, and a theory of change run for a single-stakeholder course produces ceremony.

Context

Requirements

  • A long-term outcome specific enough to reason backward from — "improved outcomes" supports no backward step
  • The parties in the room, or at least their positions honestly represented; the method's distinctive yield comes from disagreement surfacing
  • Somebody with standing to record a disagreement as a disagreement rather than resolving it in the write-up
  • Tolerance for an artifact that states what the design cannot do

Constraints

  • The term is used for two different things — a participatory reflective process and a document produced for a donor — and the second has largely displaced the first in practice, which strips out the reflection that gave the method its value [-M]
  • Assumptions get logged and never revisited. A precondition recorded once and never checked is worse than an unrecorded risk, because it reads as handled [-M]
  • Backward reasoning invites over-specification: chains elaborated to a level of detail nobody can observe, which converts a hypothesis into a plan and makes revision feel like failure [~M]
  • Like any programme theory, it is a framework for organising evidence and not evidence; the systematic-review finding that stated theory often goes unused in the evaluation that follows applies here at least as strongly [~M]
  • Complex, emergent effects resist the form. Where the outcome depends on other actors' adaptive responses, a precondition tree understates how much is genuinely unknown [~M]

Implementation Variability

  • Workshop form: the parties build the backward chain together, and the process is as much the point as the artifact
  • Analyst form: a designer drafts and circulates; cheaper, and it loses the disagreement-surfacing the method exists for
  • Assumptions-first: start by listing what has to be true and only then draw the chain, which resists the tidy-diagram failure mode
  • Contribution-claim framing: the theory used to structure evidence for a contribution claim rather than an attribution claim, which is the honest posture where the outcome has many causes
  • Nested: an organisational theory of change into which individual programmes attach, so a course's distal outcome is somebody else's intermediate one

Target Learners

  • Not a learner-facing method
  • Its indirect beneficiaries are learners on programmes whose stated purpose is remote from the classroom — employability, health behaviour, civic participation — where the unexamined chain is long and the design's actual reach is routinely overstated
  • Where the method surfaces a precondition the learner's environment does not supply, it argues for changing the environment rather than teaching harder, which is a conclusion no learner-side analysis reaches on its own

Target Learning Goals

  • None directly. The method takes the goals as given and interrogates what surrounds them
  • It bears most on designs whose goals are instrumental — taught because of what they are supposed to enable, rather than for their own sake

Instructions

  1. State the long-term outcome, concretely enough that somebody could disagree with it, and name the parties who want it. Where two parties want different things, write both.
  2. Ask what has to be true immediately before it. Do not skip to what the programme will do; the question is about the world, not the intervention.
  3. Sort each answer into two piles — preconditions the design produces, and assumptions it depends on and does not control. The sorting is the method.
  4. Repeat backward until the chain reaches something the design already claims, usually a capability from Backward Design's desired results.
  5. Write the assumptions as first-class entries, each with what would falsify it and who could check.
  6. Mark the warrant on each link — published, held position, or the design's own reasoning.
  7. Decide what comes back to it, and when. A theory of change with no revisit point is a document; with one it is a design instrument.
  • Logic Model — the lighter sibling. Same chain, without the preconditions-and-assumptions split, and the right choice when the design controls most of the path
  • Backward Design — the same backward move inside a course, from results to evidence to instruction
  • Make the Invisible Visible (Power Dynamics) — the companion where the disagreement surfaced is about authority rather than about outcomes
  • Stakeholder Mapping — supplies the parties whose competing outcomes this method has to hold apart

Examples

  • Comprehensive community initiatives — the setting the method was developed for, in which many organisations act on one outcome and none can claim it, and the reason the assumptions register exists at all
  • International development programming — the DFID-commissioned review of the term's use found it covering both a reflective process and a donor deliverable, and is the standard reference for the gap between the two
  • The Center for Theory of Change (theoryofchange.org) — ActKnowledge's practitioner materials, which carry the backward-mapping and preconditions vocabulary most widely used

Key Sources

  • Weiss, C. H. (1995). Nothing as practical as good theory: Exploring theory-based evaluation for comprehensive community initiatives for children and families. In J. P. Connell, A. C. Kubisch, L. B. Schorr, & C. H. Weiss (Eds.), New Approaches to Evaluating Community Initiatives: Concepts, Methods, and Contexts (pp. 65–92). Aspen Institute.
  • Funnell, S. C., & Rogers, P. J. (2011). Purposeful Program Theory: Effective Use of Theories of Change and Logic Models. Jossey-Bass.
  • Vogel, I. (2012). Review of the Use of "Theory of Change" in International Development. UK Department for International Development.
  • Stein, D., & Valters, C. (2012). Understanding Theory of Change in International Development (JSRP Paper 1). Justice and Security Research Programme, London School of Economics.
  • Rogers, P. J. (2008). Using programme theory to evaluate complicated and complex aspects of interventions. Evaluation, 14(1), 29–48. doi:10.1177/1356389007084674
  • Coryn, C. L. S., Noakes, L. A., Westine, C. D., & Schröter, D. C. (2011). A systematic review of theory-driven evaluation practice from 1990 to 2009. American Journal of Evaluation, 32(2), 199–226. doi:10.1177/1098214010389321
  • Mayne, J. (2012). Contribution analysis: Coming of age? Evaluation, 18(3), 270–280. doi:10.1177/1356389012451663
Page metadata
Field Value
type method
id theory-of-change
title Theory of Change
description A design method that reasons backward from a desired long-term outcome to the preconditions it requires, making explicit both the links the design controls and the assumptions it depends on but does not.
status review
generated by claude/unspecified on 2026-09-07
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.