Academic writing: claims, evidence and substantial revision
A polished sentence can carry a weak argument. Start by deciding what a reader needs to understand and what your evidence actually supports. This route ends with a short argument, its source ledger and a revision note explaining a change in reasoning rather than merely correcting punctuation.
Course topic: Writing, Composition and Argument
Study route and exercise: University Reading List editorial project. How we curate · Sources checked:
Scope and responsibility
The books do different jobs: OpenStax supplies a free genre-based writing course and handbook; They Say / I Say makes conversational moves visible; The Craft of Research connects questions, evidence and warrants. None guarantees a disciplinary style or replaces an assignment’s requirements. The worked data and source labels below are invented teaching material, never a real study or citation.
Choose your starting point
I need a free first writing course
Start with OpenStax. Choose the argument and research sections rather than attempting every genre at once. Use the handbook when a particular sentence or documentation problem arises.
I can draft but struggle to respond to others
Choose They Say / I Say for summary, response and objections. Keep OpenStax available for genre and documentation. A template helps reveal a move; it does not supply the evidence or thinking.
My topic has not become a research argument
Choose The Craft of Research if you need to connect a question to a problem for readers, then justify a claim. Add They Say / I Say only if the resulting relationships remain hard to express.
Compare the three books
Writing Guide with Handbook
Michelle Bachelor Robinson; Maria Jerskey; Toby Fulwiler · 2021
Free course and reference
- Choose it when
- You want genre models, staged writing work and a handbook in one freely readable resource.
- Use another route when
- You only need a compact guide to responding to another author this week.
- Selected reading
- 2021 original publication: sections 10.3, 12.5, 13.1 and 14.1, plus the relevant handbook entry.
- Limits and trade-offs
- The breadth requires selection. Its sample assignments are examples, not the requirements for your course; do not import their page or source counts automatically.
Try before choosing
After section 10.3, mark one claim, one evidence statement and the linking warrant in an original three-sentence practice paragraph. Use a supplied fictional packet or a genuine located source, and label which.
Check your work: The warrant explains why the evidence matters instead of repeating it. Source status remains visible, and the claim does not exceed what that source can establish.
They Say / I Say: The Moves That Matter in Academic Writing
Gerald Graff; Cathy Birkenstein · Edition: 6 · 2024
Conversation and revision companion
- Choose it when
- You can identify another writer’s claim but need to represent it fairly and make your own response explicit.
- Use another route when
- You expect sentence frames to generate a research question or make unsupported claims reliable.
- Selected reading
- Sixth edition, 2024, basic text: chapters 2, 4, 6, 8 and 12. The new chapter 7 concerns personal experience.
- Limits and trade-offs
- The selected ISBN is the basic paperback, not the separate with-readings product or a standalone digital licence. Personal experience can contribute a perspective without proving prevalence.
Try before choosing
Use sixth-edition chapters 2 and 4 to write a fair two-sentence account of an invented opposing position, followed by a response that agrees in one respect and qualifies another.
Check your work: The response addresses the strongest stated reason, not a caricature. If the voice is invented, it is not attributed to a real author or entered as a scholarly citation.
The Craft of Research
Wayne C. Booth; Gregory G. Colomb; Joseph M. Williams; Joseph Bizup; William T. FitzGerald · Edition: 5 · 2024
Research reasoning spine
- Choose it when
- You have or can develop a question and want to test reasons, evidence, warrants and objections.
- Use another route when
- You mainly need introductory grammar practice or examples of every writing genre.
- Selected reading
- Fifth edition, 2024: chapters 1–2, 5–9, then 11–12.
- Limits and trade-offs
- It asks you to make choices about audience and argument; reading it will not replace gathering relevant sources. Fifth-edition chapter numbers differ from earlier editions.
Try before choosing
Use fifth-edition chapters 5–9 to distinguish a topic, a question and why a reader should care about the answer. Add one reason, its warrant and a serious objection.
Check your work: A reader can identify the question and its consequence. The warrant can be challenged separately from the evidence; a broad topic label is not presented as a research problem.
Check your preparation
How much writing experience is necessary?
A short school essay is enough for the free route. If drafting a paragraph is difficult, first write one specific claim and explain it to an imagined reader. The Craft of Research becomes easier with an actual question and a small set of sources to test.
What belongs in a source ledger?
The source’s real author, title, date, location or stable identifier, the exact relevant passage or data, your paraphrase, and how it bears on the claim. Keep quotation and paraphrase separate. Record an unverified lead as a lead; do not create a bibliographic entry for a document you have not located.
How can I tell a warrant from evidence?
Evidence is what you ask the reader to accept as support: for example, a documented count. A warrant is the connecting principle that makes the evidence relevant to the claim. Ask whether that principle applies to this case; calling it “obvious” does not test it.
Six steps with a concrete output
Outputs marked ready: 0 / 6
These are your own study marks, not an assessment. They are not sent to this site.
-
1. Define reader, question and stakes
Selected reading: The Craft of Research, fifth edition, chapters 1–2; OpenStax 10.3.
Make this: Write a question about one proposed campus service change, name the reader who can act, and explain what that reader cannot yet decide. Produce a 100-word problem statement.
Check your work: The statement identifies a specific uncertainty and consequence rather than announcing a broad topic such as “technology”.
-
2. Summarise before responding
Selected reading: They Say / I Say, sixth edition, chapter 2, The Art of Summarizing; chapter 4, Three Ways to Respond.
Make this: Choose one real, accessible source. Write a 60-word fair summary and a separate 60-word response. Point to the source passage that warrants your summary.
Check your work: The original author could recognise the summary; the response is clearly your own and does not silently change the author’s claim.
-
3. Build the evidence and warrant map
Selected reading: The Craft of Research, fifth edition, chapters 5–8; OpenStax 10.3.
Make this: For the invented counts below, label claim, reason, evidence, warrant and qualifier. Draw an arrow for each inferential connection and identify the weakest one.
Check your work: The map does not equate seat occupancy with rejected visitors, learning outcomes or causation.
-
4. Integrate sources without inventing them
Selected reading: OpenStax 12.5 and 14.1; The Craft of Research, fifth edition, chapter 12.
Make this: Make a two-entry ledger using genuinely located sources for your own topic. Explain each source’s role, quote one short passage accurately and paraphrase another with attribution. Keep the fictional exercise outside this real bibliography.
Check your work: Every bibliographic entry resolves to an actual source, and every quotation can be checked against its original.
-
5. Give the objection its best form
Selected reading: They Say / I Say, sixth edition, chapter 6 on a naysayer and chapter 8 on why it matters; The Craft of Research, chapter 9.
Make this: Write the strongest 80-word objection to your proposal, then answer it by conceding a limit, narrowing the claim or identifying a needed check. Avoid an obviously foolish opponent.
Check your work: The objection changes at least one claim or research need; the answer does more than repeat the original position.
-
6. Revise the reasoning visibly
Selected reading: They Say / I Say, sixth edition, chapter 12 on substantial revision; The Craft of Research, chapter 11.
Make this: Draft 500 words for your named reader. Mark one removed overstatement, one rearranged paragraph and one new qualification. Add a 100-word note explaining how each change improves the argument.
Check your work: A reader can reconstruct the conclusion, evidence, warrant, objection and unresolved question without guessing what a citation is meant to do.
Worked exercise
Worked revision: what do four seat counts justify?
Original fictional draft: “Our university must build a new study hall because every student is unable to find a seat. A survey proves the library is always full.” Supplied invented data: one observer counted occupied seats in the same 100-seat room at 15:00 on four weekdays: 92, 95, 91 and 94. No survey was conducted; no record of rejected visitors, morning use or other rooms exists. Revise this into an honest claim useful to a campus facilities reader.
Assumptions
The numbers, room and observation log are invented for practice, with no real university or published source behind them. Treat each count as accurate for that snapshot only. “Practice log P” below is a fictional source label, not a bibliographic citation or evidence for a real policy.
- Check the descriptive evidence: total occupied seats are 92+95+91+94=372 across four observations; mean occupancy is 372/4=93 of 100 seats, or 93%. The observed range is 91–95%. The observations are not a random sample of all opening hours.
- Remove invented authority: replace “a survey proves” with the actual source status, “four invented 15:00 counts in practice log P”. Delete “every student” and “always”; neither has been measured.
- Expose the warrant: consistently high occupancy at one selected time can justify investigating capacity at that time. It does not imply that anyone was turned away, that all rooms were full or that new construction is the best response.
- Test the objection: empty seats remain in all four snapshots. That weakens an immediate building claim but does not settle whether seats are suitable, available to groups or occupied at other times. Those are research questions, not findings to add.
- Revise the action: request a broader, specified occupancy study before comparing extended hours, redistribution of rooms and construction. Keep the recommendation proportionate to the narrow evidence.
Answer and interpretation
Revised practice paragraph: In the fictional room, four weekday counts at 15:00 recorded 91–95 occupied seats out of 100 (practice log P; invented data). These snapshots suggest that afternoon capacity merits investigation, but do not establish unmet demand or justify a new building. A facilities review should compare multiple rooms and times and record whether users actually cannot find suitable seats before evaluating alternatives. The warrant is that repeated high use can justify further inquiry, not that high use alone proves a shortage.
Test the result
Recompute 372/400=0.93 using all observed seat positions; this equals the mean percentage only because each snapshot has the same denominator. Remove one snapshot and “always” remains unsupported. Search the paragraph for a nonexistent survey, a causal effect or a fabricated author/year citation: none should remain.
Change an assumption
When two correct averages answer different questions
Add a separate, invented one-time observation: room U has 93 occupied seats out of 100; room V has 10 out of 20, observed at the same instant. A draft says “average occupancy is 71.5%, so 28.5% of all seats are empty and capacity is sufficient”. Preserve the snapshot-only scope. Which parts need revision?
- The mean of the two room percentages is (93%+50%)/2=71.5%. This is a legitimate equal-room-weight average; do not call it arithmetically wrong without saying what is being estimated.
- For the share of all observed seats occupied, pool the counts: (93+10)/(100+20)=103/120≈85.8%. Empty seats are 17/120≈14.2%, not 28.5%. Equal weighting of unequal rooms answers a different question.
- Neither quantity establishes that everyone could find a suitable seat or that capacity is sufficient over time. Availability by room, group needs and other times remain unmeasured.
Answer and interpretation
Revise to “At this fictional instant, 103 of 120 seats across two rooms were occupied (about 85.8%); the unweighted mean room occupancy was 71.5%.” Explain which denominator matters to the reader. Remove the conclusion about sufficient capacity unless additional evidence supports that inference.
Test the result
If both rooms had equal capacity, the pooled and equal-room averages would coincide. Here 93+10+17=120. A correct descriptive average can still be used to support an unjustified conclusion; check both arithmetic and warrant.
Common mistakes
A citation as a decoration
State what a source contributes and show the inferential connection. A parenthesis at the end of a paragraph does not make every sentence supported.
A template as someone else’s thought
Use conversational frames to see the move, then make the content specific. Do not invent a consensus simply because a frame begins “Many scholars argue”.
Copyediting before testing the claim
Fix unsupported scope, missing evidence and a bad warrant before polishing every sentence. A fluent overstatement is still an overstatement.
Before moving on
- A real-source ledger distinguishes located sources, quotations, paraphrases and unverified leads.
- A claim map makes evidence, warrant, qualifier and objection inspectable.
- A revision note explains a substantive change and keeps invented exercise material out of the real bibliography.
Sources and edition checks
Sources checked:
- OpenStax: Writing Guide with Handbook, official book
The official free book is Writing Guide with Handbook, original publication 2021. The catalogue’s print ISBN does not make the paid print copy a prerequisite for the free web route.
- OpenStax: section 10.3, thesis, reasoning and evidence
Section 10.3 was checked for argument, reasoning and evidence. The original seat-count example here is not its example or an OpenStax research finding.
- OpenStax: section 12.5, integrating research
Section 12.5 was checked for integrating research. Its particular assignment requirements are not silently adopted as universal writing rules.
- Norton: They Say / I Say, sixth-edition paperback
Norton verifies sixth edition, 2024, Graff and Birkenstein, basic paperback ISBN 9781324070030. The product description’s reference to another version’s contributors does not change the authors of this selected basic text.
- Norton: sixth-edition contents for ISBN 9781324070030
The ISBN-specific official contents verify the numbered chapter assignments, including new chapter 7 and substantial revision at chapter 12. Do not apply older-edition numbering.
- University of Chicago Press: The Craft of Research, fifth edition
The publisher’s fifth-edition contents verify question/problem at chapters 1–2, argument at 5–9 and revision/incorporating sources at 11–12.
- University of Chicago Press: Spring 2024 catalogue
The publisher’s Spring 2024 catalogue identifies ISBN 9780226826677 as paper. Author list, 2024 year and fifth edition are retained; no price or availability claim is made.
These are proposed study activities, not official course assignments or a claim that every book has been read cover to cover.