ISM Hierarchical Model
Foundational driving factors appear at the bottom; more dependent factors appear toward the top.
Interpretive Structural Modeling
Structure expert judgements about how factors influence one another, then build a transparent hierarchy of drivers and dependent outcomes.
Browser-based analysis
Expert-defined relationships
Transparent matrices and hierarchy
Why use this tool?
Decision Supported
Use ISM when a complex problem contains interrelated barriers, risks, enablers, or variables that need a structured expert view.
What It Tells You
It identifies hierarchy levels, foundational drivers, dependent factors, linkage factors, and the direct or transitive paths connecting them.
Example Input
Define a custom problem or load a supply-chain template, adapt the factors, and confirm each V, A, X, or O relationship.
Sample Output Interpretation
Factors at the base with high driving power usually warrant early management attention because they influence several factors above them.
- 1DesignDefine the problem and factors
- 2Expert SurveyRecord pairwise relationships
- 3AnalysisBuild matrices and hierarchy
1. Design
Define the problem and factors before sending the questionnaire to an expert or facilitation group.
All factors, judgements, calculations, and exported files remain in your browser. Save the project JSON before refreshing if you want to continue later.
Suggested starting factors — review and adapt them to your organisation or research context.
Supply-chain templates are optional examples. Their factors are not universal, exhaustive, or validated for every organisation.
Template scope
Review and Edit Factors
Use 2–12 clearly defined factors at a similar level of detail. If you load a sample template, review, rename, add, remove, or reorder its factors to match your organisation or research context before starting the Expert Survey. Reordering factors after relationships have been entered will clear those judgements.
Important: Complete the factor review before collecting expert judgements.
2. Expert Survey
Import an ISM questionnaire or use the current design, then confirm one relationship for every unique pair.
Need more context about a factor?
Review its name and description in the Design section before making pairwise judgements.
Questionnaire
Not loaded
VFactor i influences factor j
AFactor j influences factor i
XBoth factors influence each other
ONo direct relationship
Expert judgement required
The tool does not infer relationships. Select the symbol that best reflects the agreed contextual relationship for each pair. Use X only when each factor directly influences the other; overusing reciprocal relationships can collapse the hierarchy into one level.
3. Analysis
Import a completed expert response, review the matrices, and generate the ISM hierarchy.
Sample response: Loads an illustrative electronics manufacturing supply-continuity scenario with completed pairwise judgements and a multi-level hierarchy. Review the matrices before generating results; the relationships are synthetic teaching judgements, not validated evidence.
4. Build SSIM
5. Reachability matrix
6. Apply transitivity
7. Partition levels
8. Build hierarchy
9. Review power
The Structural Self-Interaction Matrix records one V, A, X, or O judgement for each unique factor pair.
The initial reachability matrix converts direct judgements to binary links and includes each factor’s relationship with itself.
Analysis Results and Interpretation
Review hierarchy levels, transitive relationships, and driving/dependence power.
Final Reachability Matrix
Direct links are shown as 1. Relationships added through transitivity are shown as 1*.
Level Partitioning
At each iteration, factors whose reachability set equals the intersection set are assigned to the current level and removed from the next iteration.
Driving and Dependence Power
MICMAC-Style Classification
About Interpretive Structural Modeling
Interpretive Structural Modeling structures expert knowledge about contextual relationships among factors. It converts pairwise judgements into a directed reachability model, applies transitivity, and partitions factors into hierarchy levels.
ISM is useful when variables are interdependent and a team needs a transparent way to discuss which factors may act as foundational drivers. It structures judgement; it does not establish statistical or causal proof.
Guidance
- In Design, define one specific problem, its boundary, and a focused set of distinct factors at a comparable level of abstraction.
- Export the questionnaire JSON and send it to a knowledgeable expert or use it in a facilitated group workshop.
- In Expert Survey, import the questionnaire, judge every unique factor pair, and export the completed survey JSON. Use X only when influence is genuinely reciprocal.
- In Analysis, import the completed response and review the SSIM and initial reachability matrix before applying transitivity.
- Check whether base, dependent, and linkage factors make practical sense. Revisit unclear definitions or judgements when the model collapses into one level.
- Use the hierarchy and MICMAC-style classification to support discussion and prioritisation, then validate proposed actions with evidence and stakeholder review.
JSON file workflow
- Questionnaire JSON: Created with Export Questionnaire JSON in Design and opened in Expert Survey.
- Completed survey JSON: Created with Export Completed Survey JSON after the expert confirms every relationship and opened in Analysis.
Multiple experts: V, A, X, and O are categorical judgements, so this version does not automatically average independent responses. Use one agreed group response or reconcile individual responses before analysis.
Methodology, Assumptions, and Limitations
Algorithm Logic
SSIM conversion
V: rij = 1, rji = 0; A: 0, 1; X: 1, 1; O: 0, 0. Every diagonal value rii = 1.
Transitive closure
If rij = 1 and rjk = 1, the final matrix sets rik = 1. Added links are displayed as 1*.
Reachability set
R(i) = {j | rij = 1}: factors that factor i can reach, including itself
Antecedent set
A(i) = {j | rji = 1}: factors that can reach factor i, including itself
Intersection set
I(i) = R(i) ∩ A(i)
Level partitioning
Factors where R(i) = I(i) form the current top level, are removed, and the test repeats.
Driving and dependence
Driving power = final-matrix row total; dependence power = final-matrix column total
MICMAC-style classes
Driving and dependence are marked high or low against their model averages, producing Autonomous, Dependent, Linkage, and Independent/Driving groups.
The hierarchy is produced from confirmed expert relationships. Transitive links are logical implications of those judgements; they are not additional observed evidence.
Assumptions
Experts understand the problem context, factor definitions are distinct, and the selected directional relationships are meaningful for that context.
Limitations
Results are sensitive to factor selection and expert judgement. ISM does not estimate effect size, probability, time lag, or statistical causality.
FAQ
Are the suggested factors validated for my organisation?
No. They are illustrative starting points only. Review, rename, add, or remove factors to match your organisation or research context.
Does the tool decide relationships automatically?
No. Every V, A, X, or O relationship must be selected or confirmed by the user or expert group.
Can several experts complete the questionnaire?
Yes, but this version analyses one completed response at a time. Ask the group to agree a consensus response or reconcile separate expert responses before analysis.
What does a transitive relationship mean?
It is an inferred pathway added when one factor reaches another through an intermediate factor. It is not a separately entered direct judgement.
Why does my model show only one hierarchy level?
A single level means every factor can reach every other factor through the current direct and transitive paths. Review reciprocal X judgements and circular directional chains. Separate levels should only appear when the confirmed relationship structure supports them.
Does high driving power prove causality?
No. Driving power describes reachability within the expert-defined model. It does not demonstrate statistical causality or quantify the strength of an effect.
Is my model uploaded?
No. The tool processes factors, judgements, matrices, and exports locally in the browser without a backend or external API.