Theory Development (isr-theory-development)
When to trigger
- Your predictions are descriptive ("IT use relates to performance") with no mechanism
- You have an analytical model but the propositions are unmotivated or the comparative statics are not theorized
- A reviewer says "where is the IS theory?" or "this could be any social science"
- You must connect a behavioral and an economic/design logic (intradisciplinary work)
Two theorizing modes — match yours to your genre
ISR houses behavioral/empirical and analytical/economic & design-science research as co-equal genres. The shape of "theory" differs:
- Behavioral mode. Build an explicit causal mechanism centered on the IT artifact: how a feature of the system, platform, or algorithm changes cognition, affect, behavior, interaction, or organizing — then derive hypotheses a priori. State the boundary conditions and the level(s) (individual → group → firm → network → society). Theorize mediation (the process) and moderation (when it strengthens/weakens), not just main effects.
- Analytical mode. Specify agents, payoffs, information structure, and the IT-enabled decision (e.g., platform pricing, security investment, contracting, recommendation, data sharing). The assumptions are the theory: justify each, derive equilibrium results, and read the comparative statics as the propositions. The contribution is the qualitative insight ("when X, the platform optimally does Y"), not the algebra.
Make it sociotechnical and intradisciplinary
ISR's distinctive identity is bridging IS silos. The strongest manuscripts let the social and the technical co-determine the outcome and connect perspectives that usually stay separate — e.g., an economic model whose parameters are disciplined by behavioral evidence, or a behavioral study whose design is informed by a design-science artifact. Single-paradigm theory is publishable but the intradisciplinary, silo-bridging move is what ISR prizes. If you are combining methods, the 2025 multimethod-research framework editorial (ISR 36(2)) expects you to state how each method's theory informs the others.
Worked micro-rewrite: generic prediction → IT-anchored mechanism
The reviewer complaint "this could be any social science" is fatal at ISR. The fix is to make a feature of the IT artifact the causal pivot, not the setting.
- Before (context-only): "H1: Users who receive more feedback exhibit higher engagement." — True of any feedback in any medium; the technology is a backdrop, so an SE reads it as organizational behavior, not IS.
- After (IT-anchored, behavioral): "Because the platform's algorithmic feedback is quantified and socially comparative (a design property absent in offline feedback), it triggers upward social comparison that raises effort for below-median users but induces disengagement among bottom-decile users — a non-monotonic effect the feature's granularity uniquely produces." — Now the artifact's design does the theoretical work and the prediction changes if the feature changes.
- After (analytical variant): model a platform choosing feedback granularity g; users hold comparison-sensitivity θ; derive that the engagement-maximizing g is interior and falls as the θ-distribution's left tail thickens. The comparative static ∂g*/∂(tail mass) < 0 is the proposition — the IS insight ("platforms should coarsen feedback when discouraged users are numerous"), not the algebra.
Both variants pass the swap test in the next section: change the feature, and the prediction moves.
Anchor every construct to the IT artifact
The IT artifact must do theoretical work. For each construct, ask: would the prediction change if the technology were different? If not, the theory is not yet about IS.
Checklist
Anti-patterns
- Decorated correlations: "H1: A is positively related to B" with no mechanism.
- Algebra-as-theory: deriving an equilibrium without articulating the IS insight it yields.
- Technology as backdrop: a mechanism that would hold for any tool.
- Forced multimethod: stapling a model to an experiment without a shared theoretical logic.
Output format
【Genre】behavioral / analytical / design-science
【Mechanism or model】[...]
【Predictions】hypotheses (a priori) / propositions (comparative statics)
【IT-artifact role】load-bearing? [...]
【Sociotechnical + intradisciplinary】[...]
【Boundary conditions】[...]
【Next step】isr-literature-positioning or isr-methods
1---2name: isr-theory-development3description: Use when building or refining the mechanism for an Information Systems Research (ISR) manuscript — deriving behavioral hypotheses for empirical work or analytical propositions/equilibrium predictions for modeling work, always centered on the IT artifact and the sociotechnical interplay across levels of analysis. Builds the theory; it does not test it (isr-methods, isr-data-analysis).4---56# Theory Development (isr-theory-development)78## When to trigger910- Your predictions are descriptive ("IT use relates to performance") with no mechanism11- You have an analytical model but the propositions are unmotivated or the comparative statics are not theorized12- A reviewer says "where is the IS theory?" or "this could be any social science"13- You must connect a behavioral and an economic/design logic (intradisciplinary work)1415## Two theorizing modes — match yours to your genre1617ISR houses **behavioral/empirical** and **analytical/economic & design-science** research as co-equal genres. The shape of "theory" differs:1819- **Behavioral mode.** Build an explicit causal **mechanism** centered on the IT artifact: how a feature of the system, platform, or algorithm changes cognition, affect, behavior, interaction, or organizing — then derive hypotheses *a priori*. State the boundary conditions and the level(s) (individual → group → firm → network → society). Theorize mediation (the process) and moderation (when it strengthens/weakens), not just main effects.20- **Analytical mode.** Specify agents, payoffs, information structure, and the IT-enabled decision (e.g., platform pricing, security investment, contracting, recommendation, data sharing). The **assumptions are the theory**: justify each, derive equilibrium results, and read the **comparative statics** as the propositions. The contribution is the qualitative insight ("when X, the platform optimally does Y"), not the algebra.2122## Make it sociotechnical and intradisciplinary2324ISR's distinctive identity is **bridging IS silos**. The strongest manuscripts let the social and the technical co-determine the outcome and connect perspectives that usually stay separate — e.g., an economic model whose parameters are disciplined by behavioral evidence, or a behavioral study whose design is informed by a design-science artifact. Single-paradigm theory is publishable but the intradisciplinary, silo-bridging move is what ISR prizes. If you are combining methods, the 2025 multimethod-research framework editorial (ISR 36(2)) expects you to state how each method's theory informs the others.2526## Worked micro-rewrite: generic prediction → IT-anchored mechanism2728The reviewer complaint "this could be any social science" is fatal at ISR. The fix is to make a feature of the IT artifact the causal pivot, not the setting.2930- **Before (context-only):** "H1: Users who receive more feedback exhibit higher engagement." — True of any feedback in any medium; the technology is a backdrop, so an SE reads it as organizational behavior, not IS.31- **After (IT-anchored, behavioral):** "Because the platform's algorithmic feedback is *quantified and socially comparative* (a design property absent in offline feedback), it triggers upward social comparison that raises effort for below-median users but induces disengagement among bottom-decile users — a non-monotonic effect the feature's granularity uniquely produces." — Now the artifact's design does the theoretical work and the prediction changes if the feature changes.32- **After (analytical variant):** model a platform choosing feedback granularity g; users hold comparison-sensitivity θ; derive that the engagement-maximizing g is interior and falls as the θ-distribution's left tail thickens. The comparative static ∂g*/∂(tail mass) < 0 *is* the proposition — the IS insight ("platforms should coarsen feedback when discouraged users are numerous"), not the algebra.3334Both variants pass the swap test in the next section: change the feature, and the prediction moves.3536## Anchor every construct to the IT artifact3738The IT artifact must do theoretical work. For each construct, ask: would the prediction change if the technology were different? If not, the theory is not yet about IS.3940## Checklist4142- [ ] Genre named (behavioral vs analytical vs design-science) and theory shaped accordingly43- [ ] Behavioral: explicit mechanism + a priori hypotheses + mediation/moderation + level(s)44- [ ] Analytical: agents/payoffs/information stated; assumptions justified; propositions = comparative statics45- [ ] IT artifact does load-bearing theoretical work46- [ ] Sociotechnical co-determination is visible; an intradisciplinary bridge is attempted47- [ ] Boundary conditions and scope stated4849## Anti-patterns5051- **Decorated correlations**: "H1: A is positively related to B" with no mechanism.52- **Algebra-as-theory**: deriving an equilibrium without articulating the IS insight it yields.53- **Technology as backdrop**: a mechanism that would hold for any tool.54- **Forced multimethod**: stapling a model to an experiment without a shared theoretical logic.5556## Output format5758```59【Genre】behavioral / analytical / design-science60【Mechanism or model】[...]61【Predictions】hypotheses (a priori) / propositions (comparative statics)62【IT-artifact role】load-bearing? [...]63【Sociotechnical + intradisciplinary】[...]64【Boundary conditions】[...]65【Next step】isr-literature-positioning or isr-methods66```
brycewang-stanford