Ingest: The Nature of Theory in Information Systems

Type: kb/sources/types/ingest-report.md

Source: the-nature-of-theory-in-information-systems-gregor-2006.md Captured: 2026-07-14 From: https://openresearch-repository.anu.edu.au/bitstreams/77077d0f-f636-40a8-89fb-d2d2365ad7ad/download

Classification

Genre: scientific-paper -- a peer-reviewed MIS Quarterly research essay that builds and tests a structural taxonomy of theory types against a sample of information-systems articles. Domains: theory-taxonomy, design-theory, prescription, information-systems Author: Shirley Gregor, Australian National University; this is the mature primary statement of her five-type theory taxonomy and explicitly supersedes the shorter taxonomy presentation as the general account.

Summary

Gregor classifies theories by four possible primary goals—analysis/description, explanation, prediction, and prescription—whose combinations yield five interrelated types. Type V, theory for design and action, “says how to do something” by giving explicit prescriptions such as methods, techniques, or principles of form and function for constructing an artifact. All communicable theory requires representation, constructs, relationship statements, and scope; causal explanations, testable propositions, and prescriptive statements are contingent on purpose. Prescription is framed as a special case of prediction: if a recipe is acted upon, it is expected to cause an artifact of a certain type to come into being. This mature formulation therefore treats design-and-action as a purpose-defined subtype, while the availability and capability of whoever acts on the prescription remain outside the taxonomy.

Connections Found

The load-bearing target is Actionable methodology: this paper supplies the mature Type V definition while preventing its collapse into the theory–operator relation. A knowledge base holds theories, descriptions, and prescriptions with asymmetric linking is a secondary connection because Gregor's purpose taxonomy allows theory to be prescriptive, whereas Commonplace's theoretical and prescriptive profiles are local text contracts rather than ontological partitions. Relative to Gregor 2002, the 2006 paper is the broader and more structurally explicit account, not merely independent corroboration.

Extractable Value

  1. Primary goal is the classification key. A theory belongs to Type V because prescription is its primary goal, although classification can require judgment and comprehensive bodies of theory can contain multiple types. This favors keeping theory for design and action as the inherited subtype name. [quick-win]
  2. Actionability is not exhausted by Type V membership. Gregor's prescription describes what follows if a recipe is acted upon; it does not guarantee a suitable actor, authority, resources, or context. Adding those conditions changes the question from theory structure to a relation between a theory and an operating situation. [deep-dive]
  3. The mature structural minimum is smaller than the 2002 paper's strongest formulation. Representation, constructs, relationships, and scope are common to all theory; causal explanation, testable propositions, and prescription vary with purpose. A later note should not require every actionable design theory to be a complete Type IV explanatory-predictive theory. [quick-win]
  4. Prescription encodes a causal expectation. Acting on a specified method or structure should cause an artifact of a certain type to exist. This is more precise than treating actionable theory as prose that merely sounds practical. [just-a-reference]
  5. Human-machine interaction is inside the IS domain. The relevant systems include technological and social phenomena and what emerges when they interact; people can therefore be constructs or components of the target system, not only external readers of a theory. [just-a-reference]
  6. Theory types are interrelated but not ordered prerequisites. Type V may arise from other theory, practice, observation, or inventive construction ahead of good supporting theory; its origin does not decide its type. [just-a-reference]

Inherited Vocabulary

Exact terminology and definitions

  • “Theory for design and action” / “Type V.” Table 2 defines it by the phrase “says how to do something” and by explicit prescriptions—methods, techniques, and principles of form and function—for constructing an artifact (printed p. 620; PDF p. 11). The dedicated section says it concerns principles of form and function, methods, and justificatory theoretical knowledge used in IS development (printed p. 628; PDF p. 19).
  • “Prescription.” One of four primary theory goals. Gregor calls it a special case of prediction in which theory describes a method, a structure, or both for constructing an artifact; if acted upon, the recipe is expected to cause an artifact of a certain type to come into being (printed p. 619; PDF p. 10).
  • “Prescriptive statements.” Statements specifying how people can accomplish something in practice, including constructing an artifact or developing a strategy (printed p. 620; PDF p. 11).
  • “Primary goals.” What a theory is for: analysis/description, explanation, prediction, or prescription. Combinations yield the five types, and ambiguous cases require judgment about which goal is primary (printed p. 619; PDF p. 10).
  • “Scope.” The degree of generality of relationship statements, expressed with modal qualifiers, together with boundary statements limiting those generalizations (printed p. 620; PDF p. 11).
  • “Causal explanations.” Relationship statements among phenomena that show causal reasoning, distinct from covering-law or probabilistic reasoning alone (printed p. 620; PDF p. 11).

Necessary conditions versus illustrative features

  • Necessary for all communicable theory in the 2006 taxonomy: a physical means of representation, constructs, statements of relationship, and a specified scope (printed pp. 620–621; PDF pp. 11–12; recapitulated printed p. 634, PDF p. 25).
  • Necessary to distinguish Type V: explicit prescriptive statements directed at constructing an artifact—its “how to” primary purpose (printed pp. 619–620, 628; PDF pp. 10–11, 19).
  • Contingent on purpose, not universal minima: causal explanations, empirically testable propositions, and prescriptive statements are separate contingent components in Table 3 (printed p. 620; PDF p. 11). A Type V theory may also contain them in combination, but the taxonomy does not require every Type V theory to have Type IV's causal explanation and testable propositions.
  • Illustrative rather than defining: methods, techniques, principles of form/function, methodologies, decision-support-system prescriptions, relational-database design, and the Markus et al. system features/development principles are examples of the Type V form (printed pp. 628–630; PDF pp. 19–21).
  • Contribution criteria rather than membership conditions: utility, novelty, persuasive effectiveness, completeness, simplicity, consistency, ease of use, and result quality concern the value of design-science work, not the structural definition of Type V (printed p. 629; PDF p. 20).
  • Not required: actual execution, a currently present operator, authorization, success, self-application, reflection, or reflexivity. “If acted upon” makes application a condition in the predicted causal sequence, not a condition for classification as Type V (printed p. 619; PDF p. 10).

Assumed system boundary and people

The disciplinary boundary is explicitly human-machine systems. IS studies more than the technological or social system separately: it includes phenomena emerging from their interaction and therefore needs theory linking natural, social, and artificial worlds (printed p. 613; PDF p. 4). People may be inside the target-system boundary as users, human actors, work context, or organizational participants; Gregor's Type V example lists users and work context among its primary constructs (printed p. 630; PDF p. 21). Separately, a person may stand outside a particular artifact boundary as designer or practitioner acting on the theory. The taxonomy does not force one universal system boundary; each theory must state its own scope and boundaries (printed p. 620; PDF p. 11).

Causal connections

  • Prescription → action → artifact: when an actor follows the described method or structure, the recipe is expected to cause an artifact of the specified type to come into being (printed p. 619; PDF p. 10). The actor is presupposed in the causal application but not modeled as a mandatory theory component.
  • Design principle → expected outcome: Table 1 says practice prescriptions carry the implicit expectation that the prescribed method is better in some sense than alternatives (printed p. 613; PDF p. 4).
  • People/technology knowledge → artifact design → observed impacts: knowledge of people and technology informs new artifacts, which can then be studied for workplace and societal effects; design theory and explanatory-predictive theory are strongly interrelated (printed p. 629; PDF p. 20).
  • Relationships are not all causal: theory statements may be associative, compositional, directional, conditional, or causal, with the required relation type controlled by the theory's purpose (printed p. 620; PDF p. 11). Type I explicitly specifies no causal relationships; Type III predicts without developed causal explanation; Type IV includes both testable propositions and causal explanations (printed p. 620; PDF p. 11).

Distinctions that should constrain later Commonplace notes

  • Use theory for design and action for Gregor's Type V subtype; do not attribute actionable theory to Gregor as an exact synonym.
  • Keep primary purpose separate from operational state. A prescription can be Type V before anyone executes it; execution is a later event.
  • Keep theory content separate from theory–operator fit. Capability, authority, access to a target, and contextual applicability are relational conditions added by the proposed Commonplace concept.
  • Keep the actor who applies a prescription separate from people modeled inside the target system, unless a stated system boundary deliberately includes both.
  • Do not import the 2002 paper's strongest “special case of Type IV” language as an unqualified mature requirement. In 2006, explanatory and testable components are contingent, while explicit prescription distinguishes Type V.
  • Do not treat the five types as mutually exclusive containers for whole bodies of work: the categories are interrelated, and a comprehensive body can combine types (printed pp. 614, 629–630, 633–634; PDF pp. 5, 20–21, 24–25).
  • On subtype/state/relation: Type V is unambiguously a subtype in Gregor's taxonomy. If actionable additionally means that a particular actor can apply the prescription within a particular boundary, the evidence favors a relational property. The transition in which it is applied is an operational event/state. Neither classification decides anything about Commonplace, reflection, or reflexivity.

Limitations (our opinion)

Editorial opinion. The taxonomy is analytic and only lightly tested through classification of 50 journal articles; the author acknowledges that alternatives and subcategories may exist. Its broad theory vocabulary reduces exclusion but also makes boundaries contestable, and “prescription as a special case of prediction” does not specify what actor competence, authority, fidelity, or environmental conditions are needed for the predicted construction to occur. The paper's IS boundary is sociotechnical, yet it provides no general account of when a designer/operator belongs inside the system being analyzed. It therefore grounds the vocabulary of design-and-action theory but cannot decide whether Commonplace is reflective or reflexive, nor whether any particular Commonplace artifact is in fact actionable.

Use the 2006 taxonomy to constrain Actionable methodology: define Type V within Gregor's taxonomy and present operator-dependent actionability as a separately argued relation, not a sixth theory type.