
Action research is a research approach that combines practical change with the development of scientific knowledge. Researchers work with practitioners to design and implement an intervention. They study its implementation, analyze the resulting data, and reflect on the findings. This process informs subsequent action and generates research insights (Davison et al., 2004).
The approach is appropriate when you want to help shape a real-world change and investigate it scientifically. Describing a problem, completing a consulting assignment, or recommending improvements is not sufficient on its own.
In information systems research, action research examines the relationships among people, organizations, and information technology. Common settings include system implementation, digital workflows, and digital transformation in public administration.
A transparent study requires four connected elements:
- A justified choice of method and access to a suitable research setting.
- A research question that connects the intervention to a scientific knowledge gap.
- A documented process of data collection, analysis, and reflection.
- A clear distinction among practical contributions, research contributions, and limitations.
This guide explains each element. A fictional public administration example illustrates their application throughout the article. A checklist, a research cycle template, and a reflection log support your own study planning.
The materials are also available in a fillable action research workbook (XLSX). It includes blank templates and a separately labeled fictional example.
1. Action research is suited to the scientific investigation of practical change
Action research is appropriate when a practical intervention is integral to your research objectives and can actually be investigated. Its defining feature is not a particular data collection technique. It is the connection between collaborative action and scientific inquiry (Davison et al., 2004).
How does action research fit within information systems research?
Information systems researchers use a range of research approaches and methods. Wilde and Hess examine this methodological landscape and its development. Their work helps position action research within the discipline and supports comparisons among methodological approaches (Wilde & Hess, 2007).
A methodological overview does not replace detailed guidance on conducting a study. Study planning requires original methodological sources. This guide draws particularly on canonical action research as developed by Davison, Martinsons, and Kock (Davison et al., 2004).
Action research encompasses a family of related research traditions. These traditions differ in their use of theory, approaches to participation, and process models (Baskerville, 1999).
The principles of canonical action research address the researcher–client agreement, the cyclical process, theory, change through action, and learning through reflection. This framework does not define every form of action research (Davison et al., 2004).
Practical contributions and research contributions serve different purposes
An action research study must distinguish practical outcomes from research insights. A successful project is not automatically a contribution to knowledge.
- The practical contribution concerns the specific setting. Examples include a tested handoff procedure, clearer responsibilities, or a systematically documented system implementation.
- The research contribution advances scientific knowledge. It may explain relationships, refine theoretical assumptions, or develop evidence-informed design principles.
An unsuccessful intervention can still generate valuable research insights. Its conditions, implementation, and outcomes must be investigated systematically. Practical success must not be claimed when the evidence does not support it (Davison et al., 2004; Casey & Coghlan, 2026).
Why is action research relevant to information systems?
Action research investigates information systems within their organizational settings. It reveals how technical features, work procedures, and human activities interact.
A new software feature does not change a workflow independently of responsibilities and skills. Organizational procedures also influence whether employees use the feature as intended or work around it.
Researchers can investigate these relationships while helping practitioners shape a change. The focus extends beyond the technical system to decisions, everyday use, and participants’ different perspectives (Davison et al., 2004).
Lindgren, Henfridsson, and Schultze provide a real-world example. Their action research study developed and tested design principles for competence management systems. It connected individual and organizational competence with the role of information technology (Lindgren et al., 2004).
The study illustrates how an information systems project can generate research-based design knowledge. It does not establish a required study duration, number of organizations, or number of cycles.
When is action research appropriate for an academic study?
The approach is suitable when your research objectives, field access, and ability to intervene are aligned. Assess these conditions before making a final methodological commitment (Staron, 2025).
- A specific problem has both practical and scientific relevance.
- Practice partners are willing to participate in inquiry and change.
- A justified intervention can actually be implemented.
- Its implementation and possible consequences can be observed.
- Participants support transparent research documentation.
- Findings can be reported under agreed confidentiality and data protection arrangements.
When is another research approach a better choice?
Another design is often more appropriate when the proposed change cannot be implemented. Additional interviews do not compensate for a lack of access to intervene.
- You can conduct interviews but cannot implement an intervention.
- The organization expects consulting services without scientific analysis.
- A technical solution will be developed independently of the organizational setting.
- The available time does not allow implementation and meaningful evaluation.
- Data protection constraints, dependencies, or publication restrictions prevent responsible research.
Diagnosis can be an important part of action research. On its own, however, it does not constitute a complete intervention study. The same applies to a workshop whose recommendations are not subsequently investigated (Staron, 2025).
How does action research differ from related approaches?
These approaches differ primarily in their research objectives and underlying logic. They may use some of the same data collection methods.
| Approach |
Focus and distinction |
| Case study research |
Investigates phenomena within their context. A collaboratively designed change intervention is not required. |
| Design science research |
Develops and evaluates artifacts and the associated design knowledge. Sustained collaborative organizational change is not always required. |
| Action design research |
Closely integrates artifact development, organizational intervention, and evaluation. It is not interchangeable with every form of action research. |
| Participatory action research |
Emphasizes shared decision-making power, collaborative knowledge production, and often emancipatory objectives. Workshop participation alone does not justify this designation. |
Action research can include artifact development. A design-oriented question therefore does not automatically require design science research. Explain which approach best represents your actual objectives and research process (Hevner et al., 2004; Sein et al., 2011; Baum et al., 2006).
2. The research question must connect practical change with a knowledge gap
An appropriate research question goes beyond asking whether an intervention works. It examines the change process, the conditions shaping it, and what can be learned scientifically. Your research design must show how the intervention and data will address that question (Davison et al., 2004).
The article on research questions and research gaps in information systems explains how to move from a topic to a justified question. This section adds a method-specific assessment: whether that question is suited to action research.
The research design matters more than the question word
A question beginning with “how” or “why” is not automatically an action research question. The same wording may support different research designs.
For example, “How does collaboration change after a system implementation?” could be investigated through an observational case study. Action research would additionally involve researchers and practitioners collaboratively shaping a change and reflecting on it scientifically.
The question does not need to spell out every methodological step. Its connection to the intervention, data, and research objectives must nevertheless be clear.
What kinds of questions are suited to action research?
Process-oriented, explanatory, and learning-oriented questions are often suitable. The examples below are original teaching aids, not a definitive classification.
| Research objective |
Possible research question |
| Understand a change process |
How does collaboration between business and IT teams develop during the jointly designed implementation of a digital workflow? |
| Explain relevant conditions |
Under what organizational conditions is a collaboratively developed handoff procedure used or bypassed? |
| Connect technology and organization |
How do a jointly introduced status field and agreed responsibilities change the coordination of digital application handoffs? |
| Learn from revisions |
What insights into responsibility for exceptions emerge from iteratively testing and revising a digital handoff procedure? |
Each example connects an observable change with a research objective. An actual study must also justify its research gap and theoretical foundation.
Fictional teaching example: From a practical problem to a research question
The following public administration example runs throughout this guide. It is entirely fictional. Neither the proposed research gap nor the observations presented later have been empirically established.
A municipal team processes digital applications. After an initial completeness check, applications move between processing staff and specialist reviewers. Responsibilities and exception handling are not consistently defined.
- Topic: Coordination in digital application processing.
- Practical problem: Responsibilities and exception handling remain unclear during handoffs.
- Unit of analysis: Handoff practices between the participating roles.
- Assumed research gap: The interaction between technical status visibility and organizational responsibility for exceptions is insufficiently understood in the selected context.
- Research objective: Understand when a collaboratively developed handoff procedure is used and what limits its use.
- Planned intervention: Jointly develop and test a status field, an assigned responsible role, and a handoff checklist.
The research question is: “How does a collaboratively introduced handoff procedure change the coordination of digital applications, and what conditions explain its use?”
An actual literature review would need to establish the proposed research gap. The fact that your organization has not been studied before is not sufficient. Explain why its contextual features might extend or challenge existing explanations.
The main question and subquestions should address the same study
The main question provides a coherent research objective. Subquestions specify aspects of diagnosis, intervention, and interpretation. They should not create unrelated projects.
The fictional example could use the following subquestions:
- Which technical and organizational conditions make existing handoffs difficult?
- How is the collaboratively developed procedure actually implemented and used?
- What changes and unintended consequences emerge during the pilot?
- Which contextual conditions and alternative explanations account for the observations?
- What insights justify revisions and refine the existing understanding?
Not every operational project question needs to become a research subquestion. Questions about deadlines or task assignments often belong in project planning.
Which questions require revision or a different design?
Some questions are too broad, assume a particular outcome, or require another research design. They are not necessarily unscientific. They need a justified revision or methodological choice.
| Initial question |
Issue and possible revision |
| “How can we digitize our local government?” |
Too broad. Specify the process, participants, change, and research objective. |
| “What problems do employees experience with the application processing system?” |
Initially a diagnostic question. Without an actual intervention, it does not support a complete action research study. |
| “How satisfied are employees with the new system?” |
A survey may be sufficient. Action research requires a justified connection to collaboratively designed change. |
| “How can a digital handoff tool be designed?” |
May fit design science research. Determine whether artifact design or the collaborative change process is central. |
| “By what percentage does the procedure reduce processing time?” |
Requires a defensible quantitative estimate of an effect. Labeling the study action research does not, by itself, provide that justification. |
| “Why does our new procedure improve collaboration?” |
Assumes success. Ask openly about changes, absent improvements, and the conditions shaping them. |
A suitability check for your research question
The question fits the method when your research objectives align with actual opportunities to intervene. Address the following questions before committing to the approach:
- What scientific knowledge gap will the study address?
- Why is an actual change necessary or particularly useful for answering the question?
- What intervention can researchers and practice partners develop collaboratively?
- Which people, organizational procedures, and technical functions will be investigated?
- What data will document the baseline, implementation, and possible changes?
- How will you examine contrary evidence and alternative explanations?
- What knowledge contribution will extend beyond solving the local problem?
- Are access, implementation authority, and time sufficient for action and reflection?
These questions are a planning aid, not a scoring system. If you can only investigate the problem, define the study accordingly. Do not present an intervention study that was never implemented as completed action research.
Theory informs the intervention and its interpretation
Theory helps explain why an intervention is appropriate for the problem. Here, a mechanism means an assumed relationship among action, context, and outcome.
The teaching example takes a sociotechnical perspective. Its working assumption is that visible responsibility supports coordination only when work procedures are also agreed upon.
This assumption must be investigated rather than taken for granted. New observations may refine or challenge it. Theory is an explicit principle of canonical action research (Davison et al., 2004).
The research question may evolve when revisions are justified
New diagnoses and unexpected observations may justify refining the question. However, it must not be silently adjusted to fit preferred results.
Document the original wording, reason for revision, revised question, and implications for the study. Assess whether previously collected data still address the question. Readers must be able to trace this development in the research report (Davison et al., 2004; Casey & Coghlan, 2026).
3. Research cycles connect intervention, data, and reflection
A research cycle is more than a sequence of project tasks. It connects a justified action with systematic analysis and scientific learning.
This guide uses diagnosis, planning, action, evaluation, and reflection as an organizing framework. The framework draws on canonical action research. It is not a universally required model for every action research tradition (Davison et al., 2004).
Establish collaboration before the intervention
A research agreement supports practical collaboration and research independence. It clarifies objectives, roles, resources, and boundaries.
- Who decides on interventions and technical changes?
- Who collects, analyzes, and interprets each type of data?
- Who participates in defining the problem and evaluating the intervention?
- What information may be published?
- How will conflicts, revisions, and termination be handled?
- Who assumes responsibility after the research project ends?
In the teaching example, researchers are responsible for the study design and analysis. Processing staff and specialist reviewers help develop the procedure. Management provides resources. IT staff assess implementation and access permissions.
Organizational approval does not automatically replace employees’ voluntary participation in research. These decisions serve different purposes (Davison et al., 2004; Davison & Chatterjee, 2024).
Diagnosis: Establish a defensible understanding of the problem and context
Diagnosis must do more than confirm the initial problem description. Compare different perspectives, documents, and actual practices.
The outcome is a justified problem definition, a bounded unit of analysis, and initial explanatory assumptions.
In the teaching example, the team examines typical handoffs and exceptions. Interviews, observations, and work instructions may reveal different understandings of responsibility.
Planning: Design the intervention and evaluation together
The intervention needs a clear rationale grounded in the diagnosis. Decide what will change and how it will be investigated before implementation.
- Describe the intended change and the assumed mechanism.
- Specify responsibilities and technical prerequisites.
- Define evaluation criteria and appropriate data sources.
- Plan data collection points and analytical procedures.
- Document risks, alternatives, and stopping criteria.
In the teaching example, the intervention combines a status field, an assigned responsible role, and a handoff checklist. Participants agree on what constitutes a complete handoff.
The intervention should have identifiable boundaries in time and scope. More extensive changes require particularly careful documentation of concurrent developments (Staron, 2025).
Action: Document what was actually implemented
The study must capture what happened, not just what was planned. Record timing, participants, revisions, and deviations.
Demonstrating a prototype outside the field does not establish a change in everyday work. Explain precisely what intervention took place and what use could be observed (Staron, 2025).
In the teaching example, the handoff procedure is piloted within a limited scope. Revisions receive version identifiers and a documented rationale.
Evaluation: Investigate change, unintended consequences, and absent effects
Evaluation relates the data to the baseline and agreed criteria. It examines intended changes, unintended consequences, and expected changes that did not occur.
Keep observations separate from explanations. A before-and-after difference does not establish causality on its own. A different mix of cases, new employees, or concurrent training may also influence outcomes.
The fictional example assumes that the status field is used but responsibility during staff absences remains unclear. This statement is invented. It is not an empirical finding.
Reflection: Justify learning and subsequent action
Reflection connects findings to practical decisions and research interpretation. Examine your assumptions, conflicting perspectives, and influence as a researcher.
In the example, unclear responsibility for exceptions might limit use. Alternative explanations could include a different mix of cases or additional attention during the pilot.
In the hypothetical next cycle, the team adds an escalation procedure. It focuses on staff absences and exceptions. Whether the revision helps remains an open question.
Justify whether to continue, revise the approach, or conclude the study. There is no universal minimum number of cycles. Explain the actual number and its implications for the study’s limitations (Davison et al., 2004).
Data work continues throughout the research process
Data collection, analysis, and reflection do not belong exclusively at the end of a cycle. New observations may justify revisions during implementation. Document the reason and resulting decision (Casey & Coghlan, 2026).
- Diagnosis produces an evidence-informed problem definition.
- Planning connects the intervention, theory, and evaluation criteria.
- Action creates an actual change setting that can be investigated.
- Evaluation examines developments using appropriate data.
- Reflection supports research insights and decisions about continuation or closure.
Practical and research contributions are developed through this documented process. They do not arise automatically from completing a sequence of phases.
What data does the study need?
The research question determines the choice of data. Action research is not restricted to qualitative methods. It may use qualitative data, quantitative data, or a deliberately integrated combination (Casey & Coghlan, 2026).
| Data source |
Potential contribution |
| Interviews and reflection discussions |
Understand experiences, expectations, conflicts, and different interpretations. |
| Observations |
Capture actual practices, workarounds, and interactions. |
| Documents and artifact versions |
Reconstruct changes to procedures, decisions, and development steps. |
| Approved process or usage data |
Describe patterns and changes where definitions and data quality are comparable. |
| Research and reflection logs |
Make the researcher’s decisions, assumptions, and influence transparent. |
Justify the selection and amount of data collected. There is no universally required number of interviews or participants. What matters is adequate coverage of the research question and relevant perspectives.
Analyze qualitative data systematically
Select an analytical approach suited to the research question. For example, structured qualitative content analysis can organize accounts of responsibility, information gaps, and exceptions.
Document categories, coding rules, and revisions. Connect interpretations to specific evidence. The overview of content analysis provides an introduction.
Use quantitative data without overstating effects
Process measures can describe changes. They require clear definitions and comparable observation conditions.
For example, define what counts as a clarification request or processing interruption. Examine missing values, case mix, and changes in recording practices.
Report denominators and observation periods. Statistical analysis does not replace contextual analysis or reflection. Quantitative causal claims require additional, appropriate methodological support.
Triangulation tests and extends interpretations
Triangulation compares different data sources, methods, or perspectives. Contradictions are themselves a subject of investigation.
A positive interview account may coexist with frequent workarounds. Examine whether different situations, roles, or time points explain the discrepancy.
An integration matrix can connect a claim, data source, contrary evidence, and implication. This makes the path from data to interpretation and decision visible (Casey & Coghlan, 2026).
Limited time requires a clearly bounded study
For a thesis, a manageable problem is advisable. A single handoff process is generally easier to plan than an organization-wide transformation. This is planning guidance, not a methodological minimum requirement.
Check field access, approvals, and technical dependencies before selecting the method. Reserve time for analysis and reflection.
A single cycle may be defensible. However, it limits the investigation of subsequent revisions and longer-term developments. Report this limitation explicitly.
4. Research quality depends on transparent reasoning and evidence
Research quality requires a traceable connection among the question, intervention, data, and conclusions. A chronological project narrative is not sufficient.
The five principles of canonical action research provide guidance. Recent methodological work adds perspectives on collaboration, ongoing data work, and transparent reporting (Davison et al., 2004; Staron, 2025; Casey & Coghlan, 2026).
How does a study generate a research contribution?
A research contribution explains what the study adds to existing knowledge. It goes beyond reporting that a local initiative was implemented.
- What change was actually observed?
- Which explanation is supported by the data?
- What contrary evidence and alternative explanations remain?
- Which theoretical assumption is supported, refined, or challenged?
- Which contextual conditions limit the claim?
One possible contribution is a context-specific proposition. It states a justified relationship and can guide further research. It must not be presented as a universal law.
In the fictional teaching example, a proposition might be: “Visible processing statuses support coordination when responsibility for exceptions has been jointly clarified.”
This proposition is only a teaching illustration. An actual study would need to develop it from data and examine alternative explanations.
Separate the practical and research contributions in the example
- Intended practical contribution: A tested handoff procedure and clearer responsibilities.
- Possible research contribution: An evidence-based explanation of the relationship between technical visibility and organizational responsibility.
- Open question: Whether, and under what conditions, the explanation applies in other settings.
This distinction prevents practical satisfaction from being treated as scientific proof. It also prevents a theoretical interpretation from being presented as a demonstrated practical improvement.
Contextual detail supports judgments about transferability
Findings are initially tied to the setting studied. Assessing transferability requires a description of relevant conditions, not just a generic limitations statement.
In the example, relevant features would include the application processing system, division of work, frequency of exceptions, and team structure. Different legal or organizational conditions might produce different relationships.
Explain which features matter for the interpretation. This allows other researchers to assess how the findings relate to their own settings (Casey & Coghlan, 2026).
Reflexivity makes the researcher’s influence visible
Researchers participate in shaping the change. Their influence is therefore part of the study and requires reflection.
Document your prior assumptions, decisions, and role conflicts. Consider whether project responsibilities or personal expectations influence your interpretation.
Retain divergent perspectives. Joint reflection does not require artificial consensus. Disagreement may reveal important differences in interests and experiences (Casey & Coghlan, 2026).
Address ethics and data protection before collecting data
Close relationships with the field can facilitate access while creating dependencies. Employees must not confuse organizational expectations with voluntary participation in research.
Explain the objectives, risks, and intended uses of data clearly. Include critical and dissenting perspectives (Davison & Chatterjee, 2024).
- Determine the required institutional ethics review.
- Consult the responsible offices about the legal basis and information requirements.
- Limit personal data to what is justified by the research objectives.
- Define access permissions, secure storage, retention, and deletion.
- Assess reidentification risks in small teams and role descriptions.
- Keep research data separate from individual performance evaluation.
- Explain withdrawal options and their limits before participation.
- Agree on confidentiality reviews without suppressing inconvenient findings.
These considerations do not replace legal advice or an institution-specific review.
The methods chapter must explain what actually happened
A citation and a cycle diagram are not enough. Explain how the specific study was planned, conducted, and revised.
- Justify the method in relation to alternatives.
- Identify the research tradition, process model, and philosophical position.
- Describe the context, unit of analysis, and field access.
- Explain research roles, participation, and agreements.
- Connect the intervention and evaluation criteria to the problem and theory.
- Describe data selection, collection, and analysis.
- Reconstruct the cycles actually completed.
- Justify revisions, interruptions, and decisions to conclude the study.
- Explain reflexivity, ethics, and quality assurance.
- Identify limitations in access and the claims the evidence can support.
These considerations support transparent reporting. Their implementation depends on the research question and chosen tradition (Davison et al., 2004; Staron, 2025).
Sample methods paragraph for the fictional study
“The study is designed as canonical action research. It investigates changes in digital application handoffs within a municipal team. Researchers and practice partners collaboratively develop a handoff procedure. The process follows diagnosis, planning, action, evaluation, and reflection, drawing on Davison et al. (2004). Interviews, observations, and approved process data are related to the timing of the intervention. Revisions and alternative explanations are documented in a reflection log. Practical and research contributions are reported separately.”
This paragraph is a planning aid. A completed study needs specific information about the actual process, data, and deviations.
Common mistakes undermine the research logic
- A consulting project is labeled action research without an additional research rationale.
- The intervention remains a recommendation and is never implemented.
- The research question is retrospectively adjusted to fit preferred results.
- Practical success is treated as equivalent to a research contribution.
- Only supportive perspectives are reported.
- Planned procedures and actual implementation are conflated.
- Data analysis is presented only as a final step.
- Before-and-after differences are interpreted as causal effects without adequate examination.
- The researcher’s role remains unexamined.
- A local explanation is generalized without assessing contextual conditions.
Action research depends on access, relationships, and organizational decisions. Change cannot be fully controlled. This makes systematic analysis and transparent reporting particularly important.
5. The checklist, templates, and Excel workbook support study planning
The following materials connect the research question, intervention, and contribution to knowledge. They are original teaching aids, not validated measurement instruments.
Their methodological foundations are the cited works on canonical action research, practical implementation, and ongoing data analysis (Davison et al., 2004; Staron, 2025; Casey & Coghlan, 2026).
Download and complete the Excel workbook
We have compiled the planning tools into a fillable Excel workbook. It includes blank input fields, drop-down lists, and a separate fictional teaching example.
Download the action research Excel workbook: research question planner, checklist, cycle plan, and reflection log (XLSX)
| Worksheet |
Purpose |
| Start |
Instructions, suggested workflow, and usage guidance. |
| Project |
Record the research context, roles, contributions, and study conditions. |
| Research Question |
Align the research gap, question, intervention, and data access. |
| Quality Checklist |
Assess checkpoints and document evidence, follow-up actions, and responsibilities. |
| Cycle Plan |
Plan individual research cycles and document their actual implementation. |
| Reflection Log |
Record observations, interpretations, contrary evidence, and decisions separately. |
| Teaching Example |
Explore the templates through an entirely fictional public administration example. |
| References |
Review the methodological foundations and their role in the templates. |
Save your own copy first. Then complete the light-yellow input cells. The worksheets are unprotected and can be extended. Adjust row heights for longer answers.
The file contains no macros and does not calculate a scientific quality score. Because of its width, the cycle plan is primarily intended for on-screen use.
Enter confidential research data only in an approved environment. Use evidence IDs rather than identifiable raw data where possible. Do not publish a completed workbook containing confidential information.
Material A: Planning and quality checklist
We have created a fillable Excel template for this checklist. Use the “Quality Checklist” worksheet to document status, supporting evidence, and follow-up tasks. Download the planning and quality checklist template (XLSX).
Assess each item as “Met,” “Not met,” “Open,” or “Not applicable.” Add supporting evidence or a follow-up action.
Use “Not met” when an assessed criterion is currently not satisfied. Use “Open” when the assessment or clarification is still pending. Explain any “Not applicable” rating.
- The research question is derived from a scientific knowledge gap.
- The methodological rationale explains the role of the actual intervention.
- The question, intervention, data, and intended research contribution are aligned.
- People, organization, and information technology are clearly connected in the study.
- Practical and research contributions are stated separately.
- The unit of analysis and contextual boundaries are documented.
- Field access, implementation authority, and resources are confirmed.
- Participation rights and justified exclusions are documented.
- Roles, dependencies, and conflicts of interest are described.
- The intervention, assumed mechanism, and alternatives are explained.
- The process model and evaluation criteria are identified.
- The selection of data sources and participants is justified.
- Data collection points are aligned with the intervention and research question.
- Analytical procedures and the treatment of conflicting evidence are specified.
- Ethics review, data protection, and secure storage arrangements are addressed.
- Voluntary participation and limits on withdrawal are explained clearly.
- Changes and decisions have version identifiers and evidence IDs.
- Divergent findings and unsuccessful actions are reported.
- Reflection memos examine the researcher’s influence and alternative explanations.
- Continuation or closure is justified through reflection.
- Transferability is discussed in relation to specific contextual features.
- The report distinguishes plans, actual implementation, and limitations.
Material B: Research cycle template
We have created a fillable Excel template for planning and documenting research cycles. Use the “Cycle Plan” worksheet. The separate “Teaching Example” worksheet illustrates a fictional application. Download the research cycle template (XLSX).
You can also copy the blank template below into your research records. Complete observations and research insights only after conducting the actual investigation.
| Field |
Your study |
| Cycle ID, version, date, and responsible researchers |
________________________________ |
| Objective, research question, and relevant subquestion |
________________________________ |
| Baseline and unit of analysis |
________________________________ |
| Participants and roles |
________________________________ |
| Theoretical rationale for the intervention |
________________________________ |
| Planned action and stopping criteria |
________________________________ |
| Data sources and selection rationale |
________________________________ |
| Data collection points and access |
________________________________ |
| Analytical procedures and integration |
________________________________ |
| Observations with evidence IDs |
________________________________ |
| Reflection and alternative explanations |
________________________________ |
| Practical implications |
________________________________ |
| Research insights and limitations |
________________________________ |
| Implications for the research question |
________________________________ |
| Decision about the next cycle and rationale |
________________________________ |
Material B: Completed fictional example
The following example is for teaching purposes only. Observations and conclusions are hypothetical. No empirical evidence of effectiveness is available.
| Field |
Fictional entry |
| Cycle ID and status |
C1. Fictional plan with assumed observations. No actual data collection. |
| Objective and research question |
Investigate how a collaboratively introduced handoff procedure changes coordination and what conditions explain its use. |
| Baseline and unit of analysis |
Unclear responsibilities for digital application handoffs in a fictional municipal team. |
| Participants and roles |
Researchers: design and reflection. Staff: collaborative development. Management: resources. IT: technical implementation. |
| Intervention rationale |
Sociotechnical working assumption: Technical visibility supports coordination in conjunction with agreed organizational procedures. |
| Planned action |
Pilot a status field, an assigned responsible role, and a handoff checklist. Do not start while data protection risks remain unresolved. |
| Data and collection points |
Interviews and documents before the pilot. Observations during use. A reflection meeting afterward. Approved process data as a supplement. |
| Analytical procedures |
Code accounts of responsibility, information gaps, and exceptions. Compare sources and time points in a process matrix. |
| Assumed observations |
The status field is used. Responsibility during staff absences remains unclear. These statements are fictional. |
| Reflection and alternative explanations |
Unclear responsibility for exceptions might limit use. Case mix and additional attention during the pilot might also influence outcomes. |
| Hypothetical practical implication |
Add an escalation procedure. No actual improvement has been demonstrated. |
| Possible research insight |
Proposition to investigate: Visibility supports coordination when responsibility for exceptions is clearly assigned. |
| Implications for the research question |
The main question remains unchanged. A subquestion focuses more closely on staff absences and exceptions. |
| Next cycle |
C2 investigates exceptions after the escalation procedure is added. Document the decision and revised data collection plan. |
Material C: Decision and reflection log
We have also created a fillable Excel template for recording decisions and reflections. Use the “Reflection Log” worksheet to keep observations, interpretations, and decisions separate. Download the decision and reflection log template (XLSX).
Complete the log after important events and decisions. Distinguish observations, interpretations, and actions.
- Date, cycle, and version:
- Observation and corresponding evidence ID:
- Connection to the research question or subquestion:
- Your interpretation:
- Alternative explanation or contrary evidence:
- Your influence and potential role conflict:
- Participant perspectives and unresolved disagreement:
- Decision, responsible person, and approval:
- Implications for the intervention and research question:
- Next check and deadline:
The log shows how data and reflection inform subsequent action. This connection matters more than a polished retrospective success story (Casey & Coghlan, 2026).
Frequently asked questions about action research
Is action research always qualitative?
No. The research question determines which methods are appropriate. Qualitative and quantitative data may both be used. Their combination requires a clear rationale. Using multiple types of data does not, by itself, establish a coherent mixed methods design (Casey & Coghlan, 2026).
What is the best research question for action research?
There is no universal model question. Appropriate questions connect an actual intervention with a scientific knowledge gap. They often address change processes, contextual conditions, and learning from revisions.
How many research cycles are required?
There is no universal minimum. Justify the number in relation to the question, emerging insights, and actual opportunities. Report which revisions were investigated and which remain unexplored (Davison et al., 2004).
Can I study my own organization?
Yes. Being an insider can provide contextual knowledge and access. It also increases the need for reflexivity and safeguards against dependency. Document your roles, assumptions, and conflicts of interest (Casey & Coghlan, 2026).
Is an improvement workshop sufficient for action research?
Not automatically. The research question and actual process are decisive. If the workshop only produces recommendations, the study may lack an investigated change. A workshop can, however, form part of a justified intervention.
What if the intervention fails?
Report the failure and the conditions surrounding it. A transparent analysis may still generate valuable research insights. Do not claim practical benefits that the data cannot support.
References
- Baskerville, R. L. (1999). Investigating information systems with action research. Communications of the Association for Information Systems, 2, Article 19. https://doi.org/10.17705/1CAIS.00219
- Baum, F., MacDougall, C., & Smith, D. (2006). Participatory action research. Journal of Epidemiology and Community Health, 60(10), 854–857. https://doi.org/10.1136/jech.2004.028662
- Casey, M., & Coghlan, D. (2026). Beyond the cycles: Rethinking data collection, analysis and interpretation in action research in organisational setting: Implications for QuARC. Systemic Practice and Action Research, 39, Article 18. https://doi.org/10.1007/s11213-026-09766-1
- Davison, R. M., Martinsons, M. G., & Kock, N. (2004). Principles of canonical action research. Information Systems Journal, 14(1), 65–86. https://doi.org/10.1111/j.1365-2575.2004.00162.x
- Davison, R. M., & Chatterjee, S. (2024). Ethics I: Authors and their research. Information Systems Journal, 34(1), 1–5. https://doi.org/10.1111/isj.12480
- Hevner, A. R., March, S. T., Park, J., & Ram, S. (2004). Design science in information systems research. MIS Quarterly, 28(1), 75–105. https://doi.org/10.2307/25148625
- Lindgren, R., Henfridsson, O., & Schultze, U. (2004). Design principles for competence management systems: A synthesis of an action research study. MIS Quarterly, 28(3), 435–472. https://doi.org/10.2307/25148646
- Sein, M. K., Henfridsson, O., Purao, S., Rossi, M., & Lindgren, R. (2011). Action design research. MIS Quarterly, 35(1), 37–56. https://doi.org/10.2307/23043488
- Staron, M. (2025). Guidelines for conducting action research studies in software engineering. e-Informatica Software Engineering Journal, 19(1), Article 250105. https://doi.org/10.37190/e-Inf250105
- Wilde, T., & Hess, T. (2007). Forschungsmethoden der Wirtschaftsinformatik: Eine empirische Untersuchung [Research methods in information systems: An empirical study]. Wirtschaftsinformatik, 49(4), 280–287. https://doi.org/10.1007/s11576-007-0064-z