Skip to content

Guide change efforts with systemic understanding, focusing interventions on one framework component at a time as needed

Principle · All principles
Evidence · 25 claims (18 for, 7 mixed) · 20 studies (7 theoretical, 5 review, 2 causal, 2 associational, 2 qualitative, 2 design), q1–q2 · 1 of 20 report an effect size · 25 claims rest on one study

Conditional relationship

This page owns the read-the-system-first relationship in introducing a change. When a person or team introduces an innovation (a course model, a platform, a teaching practice, a programme) into a school, district, college department or organisation where other people must adopt it and keep using it, the expectation is this. Lasting, faithful use at a horizon of a term or more (term-plus) is more likely if the team first maps the system the innovation enters, then plans the other changes it needs as one coordinated bundle, and then directs effort at whichever part is blocking now. That system includes its purpose, the people and groups in it, its subsystems (assessment, timetable, incentives, roles), the adopters' concerns, the causes of any resistance, and the conditions around it such as leadership commitment, time and training. The comparison is introducing the innovation on its own, or choosing it because it worked on average elsewhere and then simply rolling it out. The outcome is adoption and continued use, observed as behavioural use and sustained practice, not whether the innovation works where it is used. That second question belongs to the innovation's own evidence.

The relationship is conditional on four things. Someone must have time and access to look at the system before and during the change. There must be more than one lever to pull (the innovation's attributes, adopters' support, the environment, the surrounding rules). The horizon must be long enough for use to stabilise. And the reading of the system must be revisited as implementation proceeds, rather than written once. The old page's second clause holds here: the parts need not be worked on in a fixed order, and at a given moment it may be most effective to work on one part of the system, such as the adopters, the innovation's attributes or the environment, while keeping the whole in view.

No claim in the wiki tests this relationship. None compares a change effort planned from a systemic reading with one that was not. What the wiki holds is mostly authors' arguments: a review digest of the classic change models, practitioner guides and a handbook chapter, all q1–q2 with no effect size. Alongside them are four findings closer to observation. In one quasi-experiment, achievement effects of personalized learning were larger where implementation was fuller and longer. In a set of district pilots, teachers who helped choose a tool used it more and with more fidelity. In an evaluation, follow-up contact by change agents related most strongly to adoption success. In a group-randomised trial, schools that adopted fastest had healthier organisations at the start, while later adopters improved most. The default design below keeps the old page's guidance as a labelled proposal.

How this page sits beside its neighbours: Match interventions to Stages of Concern owns one part of the system in detail. It covers the adopters: what each person is concerned about and doing with the innovation, and which support fits. This page owns the wider decision about which part of the system to work on, and it hands the adopter part to that page. The Professional Development pattern and Ongoing study teams and peer coaching own how one group of teachers is supported to change practice. Field-test before full implementation and the Continuous Improvement of Learning Materials process own trialling and revising the innovation itself. The methods that do the mapping are Stakeholder Mapping, Needs Analysis, Theory of Change and Logic Model.

Default design, while the relationship is untested

The old page's guidance and the framework it comes from, made concrete. Every step is a design proposal; its evidence status is stated beside it.

  1. Start from the problem and the local need, not the solution. Before choosing or launching anything, write down the problem the change is meant to solve, for whom, and how it shows locally (data, observations, the people affected). Then check that the innovation addresses that problem in this setting. Average results elsewhere do not settle that. Evidence: overfocusing on intervention selection risks a 'magic bullet' view because average positive outcomes do not guarantee local success [+W] (a state guidance document's argument). The method is Needs Analysis. Allow two to four weeks for this (an untested proposal).
  2. Map the system the innovation enters. In one or two working sessions of about two hours each, with people from each group that will be affected, list five things. First, the system's purpose and its members. Second, its subsystems the innovation touches: assessment, timetable, staffing, materials, data, incentives, reporting. Third, the stakeholders and what each stands to gain or lose. Fourth, the neighbouring systems: district or head office, parents, regulators, suppliers. Fifth, what each would have to do differently. Evidence: an innovation foreign to the rest of its system tends to be rejected [+M] (the Ellsworth digest's own synthesis, in two editions of one source). A tool's effectiveness results from the whole configuration of events, activities and contexts in which it is used [+W]. Include the capacity of the people who must carry the change, individually and as a staff, since reform progress depends on teachers' individual and collective capacity and its link with school-wide capacity [+M] (a literature review's opening synthesis; no study counts printed). Session lengths are untested proposals. The method is Stakeholder Mapping.
  3. Plan the bundle, not the single item. From the map, list the other changes the innovation needs in order to fit. Examples are an assessment that rewards what the innovation teaches, time in the timetable, a changed role for the teacher or tutor, a data routine someone will act on, and a leadership statement. Plan them together, with a named owner for each. Write the chain from these changes to the intended outcome as a Theory of Change or Logic Model. Evidence: the same digest claim ([+M] as above). Effects of a bundled programme fell where its parts were implemented less fully: personalized learning effects vary with fidelity [+M]. That was a quasi-experiment in about 40 schools, and its authors call the implementation link suggestive.
  4. Involve the adopters in choosing and shaping it. Let the people who will use the innovation review options, recommend one, and say what would need to change for them to use it. Evidence: educator involvement in choosing a pilot tool is associated with higher engagement and fidelity, while lack of autonomy contributed to low usage [+M] (district math-tool pilots, interviews and surveys; associational). An article on narrative inquiry asserts more broadly that teachers' internal drive, rather than top-down prescription, leads to the most successful change [+W], but prints no evidence for it.
  5. Work on the adaptive side as well as the technical side. Before or alongside training in the technical routine, give time to the values, beliefs and relationships the routine depends on. Hold conversations about why it matters, build trust between the people who must do it together, and share examples from colleagues. In the restorative-practices guide this "ramp-up" runs a year or more. Evidence: technical implementation falls short when the adaptive components have not shifted [+W]; relational elements are preconditions for restorative practices to work [+W]. Both come from one practitioner guide. The year is that guide's figure for restorative practices and is untested elsewhere.
  6. Diagnose the blockage, then work on that part. When progress stalls, ask which part of the system is blocking, and act on that part while keeping the map in view:
  7. Adopters: concerns or skills. Read where each person stands and match support, following Match interventions to Stages of Concern.
  8. Resistance: find its cause first. It may be threatened status, missing knowledge or skills, values the innovation conflicts with, or low confidence in the system. Then change the innovation's attributes or how they are perceived (diagnose resistance, then modify innovation attributes; perceived innovation attributes relate to intention to use and actual adoption [+M], a survey of 300 university staff about AI systems).
  9. Environment: leadership commitment, time, training, resources. Address the missing condition (Ely's conditions for change; untested here).
  10. The innovation itself: if it is adapted, keep a list of its critical components so that none is removed by accident (the framework's own advice; untested).

Evidence status: diagnosing before acting is the framework's recommendation and is untested as a design choice. 7. Keep contact going after launch. Plan follow-up contact with each site or team through the first term of use. That means a visit, call or short meeting every two to four weeks, mostly encouragement and help with what has come up. Evidence: follow-up or post-implementation contact related most strongly to adoption success [+M]. This is an evaluation of a national dissemination network, reported second-hand, with no effect size. The interval is an untested proposal. 8. Do not make the systemic reading a checklist. Revisit the map with the people in it at each review point, and ask what has changed and whether the bundle still fits. A filled-in template is not the aim. Evidence: continuous improvement can fall short when it becomes a compliance-based checklist activity [~M] (interviews with leaders, about improvement methods generally). 9. Set the horizon and the review points to match the size of the change. Evidence: compact collaboration activities did not yield systemic change within the grant period [~M]; restorative practices outcomes follow a slow trajectory [~W] (a practitioner's estimate). A default arc, every quantity an untested proposal: - Months 0–2: the problem statement, the map and the bundle plan. - Months 2–4: a small trial with volunteers, with the adopters involved in shaping it. - Term 1 of wider use: follow-up contact every two to four weeks, and a review of the map at the end of the term. - Terms 2–3: support shifts from the change team to colleagues and local leads, with contact about monthly. - Year 2 onward: review each term, and judge the change against the outcomes it was meant to produce.

Move from trial to wider use when the trial group is using the innovation as intended and the blocking parts named in the map have owners and plans. Move from close support to light support when use is routine at a review point. Expect the early gains to be in relationships and practice, and gains in learner outcomes and culture to take longer.

When a condition is not met: - If there is no time to map the whole system, at least ask each group that must change what they would need to do differently, and what would stop them. Record the answers as the map. - If the innovation cannot be changed (a mandated product), work on how it is perceived, on the environment and on the adopters.

Fitting the design to a situation

The three facts that most change the decision:

  • Who must change what, and who decided. An innovation the adopters chose differs from one handed to them. One that changes only the user's own practice differs from one that needs assessment, timetable or roles to change too. If unstated, ask: who will have to do something differently, did they have a say in choosing it, and what else in the system has to change for it to fit?
  • Which part is blocking now. Adopters' concerns, a cause of resistance, a missing condition in the environment, or the innovation's own attributes each call for a different move. If unstated, ask: where has it stalled, and what do the people involved say is stopping them?
  • The horizon and the outcome that counts. First use, faithful routine use, and effects on learners come at different times. If unstated, ask: by when must what be visible, and is the target use of the innovation or its effect on learners?
If the brief says… Then change… Basis
The innovation is already chosen and mandated from above Skip the selection step. Map the system and plan the bundle around the mandate. Put early effort into what adopters can still shape (how, when, with what support) and into how the innovation is perceived Educator involvement in choosing a pilot tool is associated with higher engagement and fidelity [~M] (shows the cost of no voice in district pilots; untested for mandates generally)
The change is a single tool or course for one teacher or one class, touching nothing else Keep the map to one page: the assessment, the time and the one or two people around the user. Do not run the full process untested proposal
A district-, college- or organisation-wide change across many sites Map each site as well as the whole. Expect sites with healthier organisations to adopt fastest, and do not leave the others out: plan more support and longer for them Later-implementing PBIS schools showed the greatest organizational health improvements, while fast adopters started higher [~M] (a group-randomised trial of one school-wide programme in 37 schools; hedged "tended")
Adopters are adult professionals in adult-workplace settings (staff, clinicians, trainers) Map roles, workload, incentives and supervision as the subsystems. Use the adopters' own work tasks as the trial material. Put follow-up contact into the working week Follow-up contact related most strongly to adoption success [+M] (school dissemination; carried to workplaces untested)
The change concerns pre-k-12 schools where families and community shape what teachers can do Include families and community organisations in the map, and plan a part of the bundle for them Teachers' ability to deliver strong instruction depends on school context, family resources and the larger community [+W] (a book chapter's framing)
The change is a technical routine resting on values or relationships (restorative practices, inclusion, a feedback culture) Before formal training, give a ramp-up period to the reasons and to trust between the people involved. Judge early progress on relationships and practice, not outcomes Technical implementation falls short when adaptive components have not shifted [+W]; restorative outcomes follow a slow trajectory [~W] (one practitioner guide)
An ambitious, inquiry-type innovation that changes teacher and learner roles Plan role changes, participation and coherence across the system as part of the bundle. Do not plan to scale by distributing materials Inquiry environments are hard to scale because they require changes in roles, participation and system coherence [+W] (a handbook chapter's conjecture)
The innovation is technology Map who supports it, the devices, the data, and what users do with what it produces, as well as the tool itself. Plan its use inside the lesson or workflow, not only its installation A tool's effectiveness results from the whole configuration in which it is used [+W]; digital education innovations are an interplay of technological, pedagogical, organisational and governance dimensions [+W] (a book introduction's conceptual argument)
The innovation is meant to narrow gaps between better-off and lower-resourced learners Put access and the conditions of lower-resourced learners in the map. Check use and effects separately for them at each review point Without attention, educational innovations tend to exacerbate learning gaps [~W] (an article citing Reich 2020)
Short funding or a fixed grant period (one year or less) Name the small, nearer gains the period can show (use, practice, relationships), and do not promise systemic change within it. Plan who carries the work on after the grant Compact collaboration did not yield systemic change within the grant period [~M] (interviews and site visits; perceived small gains)
Scarce facilitator time (one coordinator for many sites, a busy shift, volunteers) Keep the map and the bundle plan but make contact short: a 10–15 minute call per site every two to four weeks. Train one local lead per site to carry the rest Follow-up contact related most strongly to adoption success [+M] for keeping contact; the short format is an untested proposal
A single planning meeting only, of a fixed length Use the time for the problem statement (about a quarter), the map of who must change what (about half), and named owners for the two or three parts most likely to block (the rest). For example, in 90 minutes: about 20, 45 and 25 minutes untested proposal
The brief names its own planning routine or framework (its own improvement cycle, logic model template or implementation checklist) Keep it. Add only what it lacks: a list of the other parts of the system that must change, and a review of that list with the people in it at each cycle untested proposal; the risk of a routine done as compliance: continuous improvement can fall short as a checklist [~M]
The brief gives its own categories for the system (its own stakeholder groups, its own framework components) Map in those categories, not Ellsworth's. Check only that adopters, environment and the innovation's own attributes are covered somewhere untested proposal
Some of what the map would record is sensitive (staff performance, conflicts, a family's circumstances) Record roles and conditions, not named individuals' failings. Let people choose what they say in a group and what they tell the change team privately untested proposal
Some adopters join remotely, or across sites, and others are in the room Map and review in a shared document everyone can edit. Hold follow-up by call or video at the same interval. Ask remote sites separately what their own conditions are, since they differ from the main site's untested proposal
Adopters have low confidence in the change, or in the system that is making it Treat low confidence as a cause of resistance to diagnose. Ask what has failed before. Start with a small, visible early success and keep promises about support untested proposal; from the framework (resistance cause: low confidence in the system)
The change ends in a product or service for a real audience (a new course launching to students, a public programme) Include the audience in the map and the trial. Judge the trial on whether the audience can use what is produced, as well as on adopters' use untested proposal

Twelve of the eighteen rows cite a claim. All are q1–q2 except the trial row (q3), and none tests a systemic plan against a non-systemic one. Each row carries a neighbouring finding to the setting named and says so where the claim's own setting differs. Six rows are untested proposals.

Observation, state and explanation

What can be observed is use under recorded conditions: who is using the innovation, how much, which of its components are present, what else in the system has changed, what people say about it, and what is blocking. "The school is not ready", "staff are resistant" and "the culture has not shifted" are inferred states. A launch event, a signed plan or training attendance is not use.

Response Plausible explanations Observation that could change the interpretation
Use starts after launch and fades within a term No follow-up support; something else in the system works against it (assessment, timetable, incentives); the innovation does not solve a problem the users have Ask users what stopped them. Check whether the assessment and timetable changed with it. Compare sites that kept contact with those that did not.
Use is reported, but what happens differs from what was planned Critical components dropped in adapting it; the routine is run without its reasons (technical without adaptive); reports are compliance Observe use against a list of critical components. Ask users why each step matters.
One site or group adopts quickly; others do not Healthier starting conditions at the fast site; different leadership; different fit with local needs Compare baseline conditions across sites. Ask the slow sites what they would need.
Open objection from one group Threat to status; missing skill; a values conflict; low confidence after earlier failed changes Ask the group what they expect to lose. Offer the skill and see whether the objection persists. Ask about the last change they were asked to make.
Use is high, but learner outcomes do not move The innovation does not work here; outcomes come later than use; outcomes are measured on something the innovation does not target Check the logic model's chain step by step. Compare with the innovation's own evidence. Set a later review point.

These rows are proposals, untested as diagnostics. Asking people why they do not use something is itself an intervention and can change their use, so record each probe. Keep the explanations that remain open rather than picking the one that fits the change team's hopes.

Evidence and qualifications

No claim in the wiki compares change efforts guided by a systemic reading with efforts that are not, or compares working on one part of the system at a time with working on all at once. The claims below are the framework's argument and some findings about parts of the relationship.

Settings, innovations and outcomes differ across these claims: dissemination of school programmes in the 1970s, personalized learning, district technology pilots, a school-wide behaviour programme, restorative practices. Do not rank them by label. None measures whether planning from a systemic reading changes adoption.

The evidence by question

Further evidence, not yet read against this model

Claims bearing on the topic that have been read but are not part of the model above, each with how far it bears on the page.

Objective and learner-valued goal

The change team's objective is usually lasting, faithful use of the innovation, and then the outcome it was chosen for. Elicit separately what the adopters want from the change: less workload, better results for their own learners, recognition, not to be blamed when it fails, or to keep a practice they value. Agreement and divergence are both observations. A teacher who values her current practice may adopt the visible parts of an innovation and keep the substance of the old practice. A coordinator who values meeting a deadline may report adoption that has not happened.

For example, a college may intend an adaptive courseware rollout to raise pass rates in a gateway course, while the instructors want to stop spending evenings marking homework. If the rollout gives them a dashboard to act on each week, it adds work to the very thing they hoped to reduce, and use may fade. Record both aims, and plan the bundle so that one part of it serves the adopters' own goal.

What would revise this model?

The expectation should weaken under any of these findings:

  • Change efforts that mapped the system and planned a bundle show no more lasting use, at a term or more, than comparable efforts that introduced the innovation alone with the same support.
  • Simple, well-supported single innovations (a tool with good follow-up contact) are routinely kept without changes elsewhere in the system. Then the bundle step is needed only for innovations that conflict with the system, and this page should say how to tell which those are.
  • Programmes routinely improve the context they enter, as the PBIS trial suggests. Then waiting for context to be "ready" is the wrong default.
  • Systemic planning tools are mostly run as compliance. Then the method fails in practice whatever its logic.

Do not protect the model by calling every failure a part of the system nobody mapped, after the fact.

Adoption, fidelity of use, persistence of use and effects on learners are separate outcomes. The claims here speak mostly to the first two, a little to the third, and not at all to whether a systemic plan changes the fourth.

Examples

Key Sources

  • Ellsworth, James B. (2000). A Survey of Educational Change Models. ERIC Digest. https://eric.ed.gov/?id=ED444597
  • Gilbert, L. S. (1999). Where Is My Brain? Distributed Cognition, Activity Theory, and Cognitive Tools. Proceedings of the AECT National Convention. https://eric.ed.gov/?id=ED436153
  • Willie, Charles V.; Alves, Michael J. (1996). Controlled Choice: A New Approach to School Desegregated Education and School Improvement. Education Alliance Press. https://eric.ed.gov/?id=ED430265
Page metadata
Field Value
type principle
id systemic-understanding-guides-change-efforts
aliases analyze-computer-use-as-activity-system, simultaneous-desegregation-choice-improvement
title Guide change efforts with systemic understanding, focusing interventions on one framework component at a time as needed
description For a team introducing an innovation that others must adopt, mapping the system it enters, planning the other changes it needs as one coordinated bundle and working on whichever part blocks now is expected to produce more lasting use than introducing the innovation alone; no claim in the wiki tests this, and the evidence is mostly the change literature's own argument.
status review
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content
sources 3 — ellsworth-2000, gilbert-1999, willie-1996

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