Opening Remarks
What Theory Is Not
References, data, lists of variables, diagrams, and hypotheses are not theory (Sutton & Staw, 1995).
Sutton & Staw (1995) list five elements that manuscripts present in place of theory: references to prior work, data, lists of variables or constructs, diagrams, and hypotheses. Each is part of a good paper, and none of them explains why something happens. Many manuscripts contain all five and still lack an explanation; reviewers notice this most often in the results and the discussion, where the paper has to show what the findings mean.
This unit covers the second half of a manuscript (mechanisms, results, and discussion) and the revision of the whole text.
Learning Outcomes
After completing this unit, you will be able to:
- Name the mechanisms behind the effects or patterns your manuscript reports.
- Present results as evidence for or against your theoretical argument.
- Structure a discussion that answers the research question and states its scope.
- Turn limitations into boundary conditions.
- Revise a manuscript systematically for structure, theory, evidence, and style.
Mechanisms
Mechanisms
Explain why effects occur, not only that they occur (Thatcher, 2025).
- Weak: “Users with strong IT identity explored more features.”
- Strong: “The effect of IT identity on feature exploration is likely driven by internalization: when users see IT as part of who they are, they seek out new capabilities to maintain self-coherence.”
Diagnostic: every reported effect gets a “because” clause.
Rule 7 of Thatcher (2025) asks authors to link every finding back to the mechanism that produces it. A paper that reports inputs and outputs without the process in between is a “black box” paper; reviewers penalize it because the paper does not explain anything. The diagnostic in Thatcher (2025, Appendix E) is simple: after each effect, add “because [mechanism].” If no mechanism comes to mind, the theory section has not developed one.
In qualitative research, mechanisms appear as the processes and relationships in the emerging model; the requirement is the same: the findings section explains how and why the observed patterns arise.
Results as Theory Tests
- Weak: “We ran a regression to test H3 and found a significant effect (p < .05).”
- Strong: “Shielding moderated the effect of stress on performance (β = −.21, p < .01), supporting the idea that real-time capacity dampens strain-related impairment.”
Lead with the finding and its meaning, not with the procedure (Thatcher, 2025).
Rule 10 of Thatcher (2025): results paragraphs that open with “We ran …” or “We conducted …” lead with process. A stronger opening states what the data show and what that means for the theory, then reports the estimate. Report direction, magnitude, and uncertainty (e.g., effect sizes with confidence intervals), not only significance.
Further hints for the results section:
- Report what you found and hold the interpretation for the discussion; the results section links findings to hypotheses or propositions, the discussion explains their significance.
- Do not describe every step of the analysis; put details in captions or an appendix for reviewers.
- In qualitative papers, the findings section presents the emerging concepts and their relationships, supported by evidence from the data (e.g., quotes, data tables); the claim leads, the evidence follows.
Visuals
- Every important element of the argument appears in a figure or table.
- Each figure and table speaks for itself: the caption explains what is tested and how to read it.
- Every figure and table is referenced in the text.
- Labels in visuals match the labels in the text.
Thatcher (2025) calls this the “Gordon Davis Rule” (Rule 11), after the founder of MIS Quarterly: reviewers scan figures and tables before reading the text, so the model figure and the main results table should let them verify the argument at a glance. Rule 12 extends this to robustness checks and alternative specifications: if a test defends the contribution, show it in a table instead of mentioning it in passing.
Conventions: number tables and figures separately; use a figure only when the data lend themselves to visual representation; avoid figures with too many variables at once; describe in the text how a table is structured before discussing its content.
Discussion
Discussion Strategy
- Open with a short restatement of the most important results.
- Restate the research question and answer it explicitly.
- Interpret the results in light of the literature from the introduction.
- Limit conclusions to what the data support.
- Derive implications for theory and practice.
The discussion is where the paper explains what its results mean and why they differ from or confirm what others have found. Begin with a restatement of the key findings and use it to set up the ideas you want to interpret. Do not open with problems of the method or with the results you feel least confident about; place these in the limitations.
Relate the discussion back to the introduction: repeat the research question and give a brief, direct answer. Interpret the findings with reference to the sources cited in the introduction and, where needed, additional sources. Separate what the data support from what you infer beyond them.
Boundary Conditions
Theorize, don’t speculate (Thatcher, 2025).
- Weak: “Future research should examine how IT identity interacts with other individual or organizational factors.”
- Strong: “IT identity amplifies the effect of digital self-efficacy on innovation only when organizational structures allow autonomy. Future research could test whether identity-based motivation weakens in highly centralized settings.”
Rule 8 of Thatcher (2025): generic statements about future research and limitations (“our sample is limited,” “more research is needed”) signal that the authors have reached the edge of their argument. A stronger approach uses the mechanism of the study to specify where and why the findings hold, and turns limitations into boundary conditions that future research can test. The strong example stays within the theoretical system of the paper: it names the mechanism (identity internalization) and the condition under which it operates (discretionary space).
Scope and Resonance
- Close the discussion with a scope sentence: what the field now knows, and where it applies.
- Mirror the introduction: return to the opening claim or puzzle.
- End with the contribution, not with an apology.
- Weak: “Although this study has limitations, more research is needed to understand the long-term impact of CIOs.”
- Strong: “By showing that CIO presence influences AI orientation, this paper identifies a concrete lever firms can adjust as they confront strategic uncertainty around AI.”
Thatcher (2025) (Rule 16) recommends ending with resonance: the conclusion echoes the opening and reaffirms the contribution. The scope sentence diagnostic (Thatcher, 2025, Appendix E) consists of two sentences: the knowledge claim and its boundary conditions. A paper that closes with “more research is needed” fails the test.
Rule 17 adds the teachability test: a colleague should be able to explain the model, the central concept, or the mechanism on a whiteboard in three minutes.
Style
Compression
Compress without hollowing: every sentence advances the argument (Thatcher, 2025).
- Say each idea once, then move on.
- Give each sentence a distinct job (define, justify, apply).
- Prefer precise verbs: “influences” instead of “has the potential to influence.”
Rule 13 of Thatcher (2025): concise writing is dense writing. Many drafts restate the same idea two or three times in one paragraph; compression removes the repetition, not the theory. Ask of each sentence what new job it does compared with the previous one; if it only rewords the same point, cut or revise it.
Clutter
| Clutter | Examples | Alternative |
|---|---|---|
| Hedges | might, may, could, appears to, potentially, somewhat | State the claim with the confidence the evidence supports |
| Nominalizations | utilization, implementation, the investigation of | use, implement, investigate |
| Metadiscourse | “In this section, we will discuss …” | Lead with the claim |
| Shudder quotes | the so-called “digital” turn | Use the term or define it |
Pinker (2014) argues that academic prose is often hard to read because writers hedge, turn verbs into abstract nouns, and comment on their own text instead of guiding the reader’s attention to the subject. Thatcher (2025) (Rule 14) turns these observations into a revision checklist. Hedges are appropriate where the evidence is uncertain; the problem is hedging by default.
Use active voice: write “we measured” instead of “it was measured.” Active voice is usually shorter and makes clear who did what.
Hemingway Pass
Heuristics for a style pass during revision (Thatcher, 2025):
- Search for hedges and remove those the evidence does not require.
- Convert “-tion” nouns back into verbs.
- Keep passive voice below about 10% of sentences.
- Aim for an average of 14 to 18 words per sentence.
The thresholds come from Thatcher (2025) and are heuristics, not rules of the journals. They help to locate problems: a section with long average sentences or frequent passive voice usually contains other forms of clutter as well. Run the pass after the content is settled; polishing sentences that will later be cut wastes time.
Revision
Diagnostics Box
Four passes after drafting, not during drafting (Thatcher, 2025):
| Pass | Focus | Diagnostics |
|---|---|---|
| A | Structure | Topic-sentence ladder; Control-F label audit |
| B | Theory | Contribution clarity; mechanism naming; scope sentence |
| C | Evidence | Visuals as evidence; measurement and definition alignment; robustness at a glance; magnitude and uncertainty |
| D | Style | Hemingway pass |
The Common Diagnostics Box in Thatcher (2025, Appendix E) turns the rules into mechanical checks with pass and fail criteria. The passes take between 10 and 40 minutes each. Writing and diagnosing are separate activities: finish a full draft of a section first, then run the diagnostics. The self-study phase before the third day applies the passes to your manuscript.
Writing Routines
- Show up and write: daily production, even one page or one paragraph.
- Start with revision: edit the previous day’s work before writing new text.
- Meet prepared: send an agenda and a one-page input before meetings with supervisors.
Thatcher (2025) credits two principles to his mentors. The Pascoe Principle (“You don’t wait for inspiration. You show up. You write.”) treats writing as a daily practice: block time, set a minimum output, and track it. The George Law (“Don’t waste your advisor’s time. Come prepared. Be clear. Move the work forward.”) applies to supervision: structure every meeting, bring a draft or a one-page sketch, and follow up with a three-bullet summary of next steps.
AI in Writing
- AI tools cannot be authors; authors remain responsible for every claim (Committee on Publication Ethics, 2023).
- Disclose the use of AI tools as your target journal requires.
- Verify every reference and every factual claim an AI tool produces.
- Never upload confidential manuscripts (e.g., as a reviewer) to AI tools.
The Committee on Publication Ethics states that AI tools cannot meet the requirements for authorship because they cannot take responsibility for the submitted work, and that authors who use AI tools must disclose how they used them (Committee on Publication Ethics, 2023). Publishers and journals translate this position into their own policies, which differ in detail: some require a statement in the methods section, others in the acknowledgments; some allow AI-assisted language editing without disclosure. Check the policy of your target journal before submission.
Legitimate uses include language editing and checking the consistency of labels. Problematic uses include generating arguments or literature summaries that you do not verify against the sources; fabricated or misattributed references are a known failure of language models and a reason for desk rejection.
Mock Review
The Mock Review
The self-study phase simulates a review round:
- By 22.12.2026: upload your revised manuscript, after running the four diagnostic passes.
- By 04.01.2027: write two reviews as reviewer for the manuscripts of two peers.
- By 07.01.2027: write one decision letter as editor for the manuscript of a third peer.
- Third day: draft a response matrix to your reviews and your decision letter.
Every participant takes each role: author, reviewer, and editor.
With four participants, each manuscript has exactly three other readers: two reviewers and one editor (see the role table in Administrivia). Write each review for the target journal of the manuscript: a short summary of the paper, the main concerns (positioning, contribution, theory, method, presentation), minor points, and a recommendation (reject, major revision, minor revision, accept). Number your points so that the author can respond to them one by one.
The decision letter synthesizes the two reviews, as in a real review process: it states the decision, names the three to five issues the revision has to address, and resolves disagreements between the reviewers. Writing it trains the perspective of the desk-reject simulation from the first day at the next stage of the process. On the third day, the lecturer comments on the decision letters as editor-in-chief, and the response matrix from Thatcher (2025, Appendix G) turns reviews and decision into a revision plan.
Homework
During the self-study phase:
- Revise your manuscript, running the four diagnostic passes, and upload it by 22.12.2026.
- Write two reviews by 04.01.2027 and one decision letter by 07.01.2027.
- Prepare the third day:
- Look up deductive, inductive, and abductive reasoning and classify your manuscript.
- Note keywords on positivist, interpretive, and critical research.
- Add the dominant methods of your conversation to your conversation map.