Overclaiming relative to sample
The claim that re-purposing was most visible in the post-merger setting needs site-by-site counts before it can carry argumentative weight.
Read the rule behind this10 rooms, grouped by stage from choosing a topic to answering your committee. Every comment points to a passage of your thesis or a published source.
Free to start, no card needed. Then $9 a month billed yearly, or $12 month to month.
Plays from YouTube only after you press play.
A walkthrough of Thesisroom, from choosing a topic to practicing your defense.
Thesisroom has 10 rooms, each built for one part of a thesis.
Shows what published research already holds near your topic.
What it does not do
It does not tell you what to research.
leadership and team performance in remote work
The record holds nine papers on health workforce staffing beside thirty-seven on team dynamics. Is the smaller group thin because it has been answered, or because nobody has asked?
You type what you are thinking about, in your own words. Thesisroom searches published research databases and shows you the shape of what is already there: the groups your interest sits between, how many papers each one holds, when each was busiest, who publishes in them, and the papers to start from.
Every paper is a real record with an identifier that resolves. Nothing is written for you and nothing is summarised. The counts are the record's own arithmetic, so where the field is crowded and where it is sparse is something you see rather than something you are told.
Explore a topic when you are choosing your subject. Check a claim when you know what you want to argue.
Thesisroom never says which part of the field to work in. It shows you what is there and asks you questions about it, because the person who has to live with the answer for three years is you.
Reads published research, not your writing.
Your words go straight to the academic databases as a search. There is no document to upload and nothing of yours is read, because at this point in a degree there is usually nothing written to read.
Shows the shape.
Papers are grouped by the topics the databases already assign them, so the labels are the field's own words and you can take them to a supervisor. Papers the record has not classified yet appear in their own group rather than being dropped.
Turns the results into questions.
Up to three questions, each one naming a specific paper from your results that prompted it. One of them presses on the thinnest part of what came back. A question that cannot name a paper from your results is dropped rather than reworded, so a search sometimes carries two.
Thesisroom does not tell you what to research, and it does not say a quiet corner of the record is an opportunity. A thin group of papers can mean a question nobody has asked or one that turned out to be a dead end, and the databases cannot tell those apart. Naming which is which is for you and your supervisor.
Asks where your proposal's question and its plan do not obviously meet.
What it does not do
It does not say whether a project can be done.
This study asks whether cooperative membership raises smallholder resilience to price shocks across the cereal, horticulture and dairy sectors in Kenya, Uganda and Tanzania, and whether any effect differs by sector and by country.
In Research question
I will conduct semi-structured interviews with members of one maize marketing cooperative in Nakuru County, Kenya.
In Research design and method
Which of the three sectors does a single maize marketing cooperative speak to?
Thesisroom reads your research proposal and finds the pairs of passages where what you ask for and what you plan sit apart. A question naming three sectors beside a method describing one site. A claim resting on records the access section says survive for part of the period. An analysis needing more schools than the recruitment section describes.
Each pair comes back as one question, with both passages quoted from your own proposal and the section each sits in. Nothing is added. The two passages are the whole of the reason the question is there, and you can see both without leaving the page.
Every other room hands you comments. This one hands you a question, because in the first weeks of a degree the person who knows the answer is you. Thesisroom reports what your proposal states and asks how the two parts meet. What to do about it stays yours.
Reads what you stated.
Thesisroom reads the proposal for what it asks, what its method describes, and what its resources section says is in hand. Only what the document states counts. A proposal naming no sample size is a proposal with no sample size to hold anything against.
Quotes both halves.
Every question carries two passages, copied word for word from your proposal and found in it before you see them. A passage Thesisroom cannot find in your text is dropped, and the question with it.
Asks about the join.
One question per pair, about how the two passages meet. You answer it, or you change the proposal, and either way the decision was yours to make.
Thesisroom does not say whether a project can be done. It knows nothing about your programme, your supervisor, your funding, or the year ahead of you, and a tool that guessed at those would be guessing at the things that matter most. It reports what your proposal states and asks about the join.
Looks up every reference in published research databases such as Crossref and OpenAlex.
What it does not do
It does not rewrite your citations or generate replacement references.
Arowosegbe, O. B., Olutimehin, D. O., Odunaiya, O. G., Soyombo, O. T.
Risk management in global supply chains: Addressing vulnerabilities in shipping and logistics
10.51594/ijmer.v6i3.962
Verified against published research databases such as Crossref and OpenAlex
Thesisroom extracts every entry in your bibliography. For each one it checks whether the work is in published research databases such as Crossref and OpenAlex, whether the DOI resolves and whether the paper has been retracted. On a paid run it also reads the abstract of a first set of verified papers against the claim you cite each one for. The result is a traffic-light list. Green is verified. Orange needs a second look. Blue is found in a repository and not verified. Magenta is unverifiable, which is the ordinary result for books, working papers and reports that no registry indexes.
Most of this is not done by reasoning. Thesisroom queries the registries directly, and reads retraction status as a marker those records carry. Which registries a reference reaches depends on where it turns up: once one confirms the work, the rest are not searched. Judgement is reserved for the final step, whether a specific claim in your text is supported by the cited paper's abstract. Every verdict states its reason in plain words, with a link to the paper where one exists, so you can check the verdict yourself.
What you see
Source response
Crossref returned a record for this DOI with the same title and first author. OpenAlex holds the same work.
Nine named sources, and what each one settles.
Crossref resolves DOIs. OpenAlex covers author disambiguation and non-journal works. PubMed confirms biomedical work curated by the National Library of Medicine. DataCite confirms a DOI those three do not know, as it registers DOIs for datasets, software and theses held in repositories. Those four decide whether a reference is verified. Retraction Watch data reaches the check through Crossref's record of the work, and PubMed marks a retracted publication in its own article type, so a retraction outranks every other verdict. Semantic Scholar supplies the abstract the claim step reads. OpenAIRE and HAL find repository and preprint copies and Unpaywall locates open-access copies, and a work found only there, or only as a repository copy in OpenAlex, is marked found and not verified rather than verified. DOAJ is asked by the journal's ISSN, and a listed journal is named on the result. An unlisted one gets no line, because absence from DOAJ says nothing against a journal.
Deterministic where possible.
Most broken citations are caught by direct database lookups. Whether a paper exists is answered by the registry, not by a model's memory.
Never invents citations.
Thesisroom cannot suggest a citation you did not include. There is no code path for that. The bibliography you upload is the universe.
Checked once, then checked again every week.
Every source a run returns a verdict on joins your Library: every source you've checked, kept up to date. The weekly pass re-checks it against the same registries for as long as it stays there. A paper can be retracted years after you cite it, and the verdict on your screen is the verdict as of the last check.
Thesisroom does not rewrite your citations. It does not generate replacement references. The audit names what is wrong. The fix stays yours. Absence from the Directory of Open Access Journals is recorded and shown, and it is not read as a judgement about the venue.
Re-checks your sources every week and tells you when one changes.
What it does not do
It does not assess or rank the papers it finds.
Hernandez & Park (2019). Humble leadership in remote contexts: a dyadic process model.
The work has been retracted.
Source: Retraction Watch
Thesisroom re-checks your bibliography against published research databases such as Crossref and OpenAlex on a weekly cycle. A paper can be retracted, a DOI can stop resolving, and a venue can be de-listed long after you cited the work. When a source's status changes, verified to retracted, or back to verified, the alert appears here, with the registry that produced it.
It also watches the record around the claims Blind Spots has read. Once a week it searches for each of those claims and names work the record ranks highly that your bibliography does not contain. Those appear in their own section, never mixed with the first kind: a source you cited being retracted is urgent, and a paper you have not met is not.
Alerts arrive as one weekly digest email, which you can turn off, and they wait on the Reference Watch page until you dismiss them. A retracted source with a stored claim passage offers a direct route into Find Sources, pre-filled with the passage that needs new support.
Your bibliography is the watchlist.
Every source you have verified through Citation Check joins your Library. Reference Watch monitors that list. Nothing is added on its own.
Weekly re-check, nine named sources.
Every source in your bibliography is re-verified each week against the same registries Citation Check uses. The check is deterministic. No model decides whether a paper is retracted.
One search per claim, per week.
For the claims Blind Spots covered least, Thesisroom searches published research databases once a week and keeps what the registry returns. Nothing is generated: every paper is a registry record with a resolvable identifier, carried through as it came back.
Alert only on change.
A source that stays verified stays silent. An alert appears only when the status actually moved, and it names the registry that moved it.
Reference Watch does not assess the papers it finds, does not rank them by quality, and does not decide what is worth reading. It searches for claims Blind Spots has already read, never for a topic it chose. The writing stays yours.
Margin comments on your chapter, the way a careful supervisor marks it.
What it does not do
It does not ghostwrite or produce alternative drafts.
The studies reviewed above show that Green Finance is associated with improved environmental performance and operational efficiency. But association and mechanism are different things.
Acknowledgement of a gap without full resolution
The text correctly identifies a critical gap between association and mechanism but does not fully address how the proposed model will establish causation versus correlation
Rule · Argument structure
Thesisroom reads each chapter of your thesis against the standards of your field. It returns marginal comments anchored to specific sentences, classified by severity: must address, revise, or consider. Each comment quotes the sentence it rests on, says what to change, and links to the rule behind it.
The comments come most serious first, the first five on the page and the rest one press away. You can mark a comment resolved as you work through them.
What you see
Chapter 3 · §3.2
The interviews revealed three recurring orientations toward institutional voice, which I label endorsement, withdrawal, and re-purposing. Each orientation appeared across all four sites, though re-purposing was most visible in the post-merger setting. The next section reads these patterns against the typology proposed by Hirschman.
Taken together, these results establish that procedural justice is the dominant mechanism through which humility shapes voice, a result that holds across the full sample and resolves the tension left open in the literature.
The claim that re-purposing was most visible in the post-merger setting needs site-by-site counts before it can carry argumentative weight.
Read the rule behind thisHirschman's typology is named here but never used to interpret the three orientations. Either apply it or drop the reference.
Read the rule behind thisAnchored to your own sentences.
Every comment points at a passage in your chapter and carries the reason behind it. A comment that cannot be traced back to your own text is never shown to you. You get margin notes, each marked with how much it matters, not new paragraphs.
Field-aware.
A chemistry chapter and a history chapter get audited against different standards. Your field selection on signup decides which rubric applies.
Comments, not rewrites.
Every output is a margin comment. Thesisroom does not produce replacement paragraphs.
Thesisroom does not ghostwrite. It does not produce alternative drafts. The audit names where work is needed. The writing stays yours.
Counts the most-cited papers on each claim's topic that you have not cited.
What it does not do
It does not tell you what to cite.
Your claim (p. 42): Procedural justice mediates the effect of leader humility on team voice.
You cited 3 of 28 most-cited papers on this topic.
Owens & Hekman (2016) How does leader humility influence team performance? Exploring the mechanisms of contagion and collective promotion focus.
For each claim in your thesis, Thesisroom counts how many highly-cited papers exist on that topic in OpenAlex, a published research database, then compares that count against your own bibliography. The result is a ratio: you cited three of twenty-eight. Both numbers come from the registries. Nothing decides whether three of twenty-eight is good enough for your argument, because that is a judgment about your thesis, not a fact about the record.
For each claim, two lists follow, whatever its coverage: the most-cited papers on that topic and the recent papers cited above their field average, each absent from your bibliography and verified through the same registries as Citation Check so you know the paper is real and not retracted. Each shows a sentence from its own abstract and whether that language supports or questions your claim. Intellectual neighbours are researchers citing three or more of the same sources you cite: people asking your question with your evidence, found by following the citation graph rather than by asking a model.
Your claims set the topic.
Claims that share a topic are searched together, once. Each topic's keywords come from its own claims, their surroundings, your thesis title and your field: a word that is not in that material is not searched on.
The count is arithmetic.
Papers above a citation threshold form the field; your verified sources are matched against them. A retracted source counts toward nothing.
The order is the registry's.
The most-cited list is sorted by citation count and the recent list by how many times its field's average each paper is cited, per OpenAlex. Never by opinion, and every paper is checked for retraction before it is shown to you.
Blind Spots does not tell you what to cite. It counts what exists and shows you what you missed. Whether a gap matters for your argument is your decision.
Finds real papers for a claim you need to support.
What it does not do
It does not invent sources or pick one for you.
The passage you need a citation for
Rising sovereign debt deepens the economic damage of climate shocks, creating a feedback loop that constrains adaptation spending in the most exposed economies.
Paste the sentence where you need a citation. Thesisroom reads the verifiable claim in it, searches the academic databases, and returns real papers in the order the databases rank them. Each result carries an abstract, a citation count, open-access status, and a one-click link to read it.
It is a research aid for your decision, not a recommendation engine. You enter the claim you have already decided to argue, the search returns real work, and you choose which paper, if any, supports it. Nothing is auto-inserted into your thesis.
Claim, not topic.
You paste the sentence you need to support. Thesisroom extracts the verifiable claim and the search keywords from it, never a citation.
The database ranks, not a model.
Results come back in the academic databases' own relevance order. Thesisroom does not sort them by a model's opinion or mark one as the answer.
Every result is verified.
This is Citation Check run in reverse. Retracted papers and papers without an abstract are filtered out, so you see only work that would pass the same checks.
Thesisroom does not invent sources and does not pick one for you. It does not write your in-text citation and does not auto-replace anything. The choice stays yours.
Finds contradictions between chapters, drifting definitions and unanswered questions.
What it does not do
It does not rewrite your thesis or resolve a contradiction for you.
Chapter 3, p. 67
The mediation held across both co-located and remote teams, with no moderating effect of team distribution.
Chapter 5, p. 134
Remote teams showed a markedly weaker pathway, suggesting distribution constrains the mechanism.
Thesisroom reads your full thesis and extracts every claim, every defined term, and every research question, each as a verbatim quote, with its page when you upload a PDF. It then reads those layers against each other: which definitions drift when a term is used again three chapters later, which research questions are stated but never answered, and which claims sit in tension across chapters.
The map is the one part of Thesisroom you keep open while you write. Type a term and see everywhere you used it, with those page numbers. That search runs against the map already built, so it involves no analysis of any kind.
Chapter by chapter.
Every claim is pulled as an exact quote and checked against your thesis text before it is stored. A passage that cannot be found is rejected, not shown.
Then chapter against chapter.
Definitions, questions, and claims are compared across chapters, which is where drift, unanswered questions, and contradictions become visible.
The map persists.
Once built, the map stays searchable between runs. Searching costs nothing and consults nothing beyond your own stored words.
Argument Map does not rewrite your thesis, suggest what to argue, or tell you how to resolve a contradiction. It shows you your own words and where they conflict. The argument is yours.
Practice the questions an examiner or committee asks, in three styles.
The three: Practice, Rapid round and Mock panel. Every question is tied to a passage in your thesis.
What it does not do
It does not write your answers or hand you a script.
On page 82, you discuss the unsupported moderation hypothesis H4, where the interaction term was near zero. Yet, your NCA analysis on page 73 finds IS/IT is a necessary condition for high resilience. How do you reconcile these two seemingly contradictory findings?
Thesisroom reads your full thesis once. It identifies the lines of attack an examiner or committee could take: methodology choices, generalisability, theoretical alignment, contribution claims. A full session builds up to 40 questions as you practice, grouped by difficulty, easiest first, each tied to a specific passage in your thesis.
Inside a session you receive one question at a time. You answer in writing or by voice. Thesisroom gives written feedback on clarity, depth and accuracy: what worked, and what to strengthen. At the end of a session, a report names your three weakest areas and recommends what to revisit.
Every question presses on a specific passage. One you can answer is a passage that held. One that stops you is a passage an examiner or committee would press on too, and that is the same fact whether you read it in month four or the week before you defend.
What you see
Your fourth chapter argues that procedural justice mediates the effect of leader humility on team voice. How do you defend that mediation against the reverse-causality reading the data also supports?
From your thesis · p. 134
The dyadic models reported in Table 4.3 indicate that variance in team voice is most strongly accounted for by the perceived procedural justice composite, and only weakly by leader humility absent that mediator.
Sample answer and feedback
I anchor the mediation in the two-wave subset reported in chapter four, where leader humility at T1 predicts procedural justice at T2 but not the reverse. I acknowledge the cross-sectional design cannot rule out reverse causality outright, and I present the mediation as the more parsimonious reading rather than a settled claim.
Clear thesis and an honest concession. Tighten the link between the two-wave evidence and the specific causal claim before the examiner or committee presses on the sample size.
Reads your thesis.
Thesisroom parses your thesis and identifies the claims, methodology choices, and theoretical commitments that carry the most weight.
Anchors every question.
Each generated question references a specific passage. The interface lets you open that passage in one click, with the relevant sentence highlighted.
Gives honest feedback.
Feedback follows rubrics tuned to your level and field. Thesisroom tells you when an answer is weak, not only when it is fine.
Thesisroom does not write your answers. It does not feed you talking points to memorise. The questions are designed to make you think, not to give you a script.
Shows whether a paper supports, contradicts or complicates your argument.
What it does not do
It does not tell you what to argue, cite or write.
Your claim (p. 87): Procedural justice mediates the effect of leader humility on team voice.
What the paper found: Perceived fairness of process fully accounted for the humility-to-voice pathway in both samples.
Longitudinal three-wave design
The paper tests the mediation across three waves. The thesis relies on cross-sectional data in the chapter where this claim lives.
Most relevant to: Methodology
Upload a published paper. Thesisroom reads it against your stored claims, your bibliography, and your chapter structure, and returns three outputs. A claim map shows which of your claims the paper supports, contradicts, or complicates, each match tied to the exact passage in your thesis. Gap detection names what the paper does that your thesis has not addressed, with the chapter it is most relevant to.
The third output is a cite-or-skip verdict: the paper is checked against the same nine named sources as Citation Check, the claim map decides relevance, and the two combine into a plain classification. Whether to cite is your decision. Thesisroom states what it found, never what to do.
Your claims are the reference frame.
The paper is read against claims already anchored in your thesis. Every match quotes your own passage back to you, verbatim, with its page.
Classification, not opinion.
Each relationship is classified as supports, contradicts, complicates, or neutral, with a confidence score. Weak matches are dropped rather than shown.
The registries settle the record.
Whether the paper is retracted or verifiable is answered by the same registries Citation Check queries, not by a model.
Paper Match does not tell you what to argue, what to cite, or what to write. It maps the relationship between a paper and your thesis. The decision stays yours.
Errors you fix, not comments you weigh.
The ten rooms above read your argument and hand you comments and questions to weigh: a question an examiner or committee may ask, a structural comment worth acting on, a source worth a second look. Judgement is required to act on any of them, which is why those rooms cost something to run and are metered accordingly.
A check is the other kind of work. It recomputes something you already reported and tells you where the arithmetic does not hold. There is nothing to weigh and no judgement to make, so no model runs, nothing is metered, and checks are unlimited on every plan, free included.
Thesisroom takes every significance test you reported in a sentence, recalculates the p-value from the test statistic and degrees of freedom you gave, and tells you where the reported figure cannot be right. It reads results written in your text, not results in tables, across five families: t, F, r, z, and chi-square.
t(28) = 2.20, p = .60
Candidates in the humility condition reported significantly greater voice, t(28) = 2.20, p = .60, supporting the second hypothesis.
The reported p-value is not achievable from t(28) = 2.20, and the significance decision at .05 changes.
The recomputation approach is published by Nuijten, Hartgerink, van Assen, Epskamp and Wicherts (2016) in Behavior Research Methods, and implemented in the statcheck package by Epskamp and Nuijten. Thesisroom applies a stricter rounding rule than the published method.
When Thesisroom can find your reference list and read your citation style, it matches every citation in your text against your bibliography and every entry against your text, then names the citations with no entry and the entries nothing cites. It reads your text only, never the list itself, because a reference list holds every entry beside its own year and reading it would make every entry look cited.
Harman, 1967
Common Method Bias (CMB) is assessed using two approaches: Harman's Single Factor Test (Harman, 1967) and the Full-Collinearity VIF Technique.
This citation appears in your text and Thesisroom found no entry for it in your reference list.
References matches against the bibliography entries Citation Check has already parsed, rather than parsing the list a second time. Two parsers would eventually disagree, and two different entry counts on two pages of one product would mean neither could be trusted.
Thesisroom reads the chapter list it detected in your document and every numbered figure and table reference written in your text, and reports where a sequence breaks: a chapter numbering gap, a figure number that appears nowhere, a contents entry that matches no detected heading, and the word count against a limit you set. It counts what is written in your text, so a figure numbered only inside an image is outside its reach.
Chapter 4 · Chapter 6
Chapter numbering skips 5: chapter 4 is followed by chapter 6.
Structure reads the chapter list the parser already stored and the same parsed text every other surface reads. It never comments on argument, evidence or quality; those belong to Chapter Audit. Every comment is counting or matching, so you can verify any of them by looking at your own document.
A check never tells you what to write. Thesisroom reports that a number does not hold and shows you the arithmetic. You decide what your manuscript should say.
A passage you can check is evidence.
The studies reviewed above show that Green Finance is associated with improved environmental performance and operational efficiency. But association and mechanism are different things.
Acknowledgement of a gap without full resolution
The text correctly identifies a critical gap between association and mechanism but does not fully address how the proposed model will establish causation versus correlation
Rule · Argument structure
You upload your thesis once. Thesisroom parses it once. The rooms that read it then run against the same parsed text, the same bibliography, the same field rubric. Scope Check reads your research proposal, and Citation Check a reference list when you add one. A comment in Citation Check cross-references with a comment in Chapter Audit, a citation marked in Citation Check opens Find Sources pre-filled with the passage it came from, and a weak answer in Defense Practice can route you back to the relevant chapter.
Some of it keeps running after you upload. Reference Watch re-checks your bibliography weekly and alerts you when a source's status changes, and Paper Match reads any new paper against your claims on demand. Uploading a new version keeps the old one, every previous run stays available, and you can compare two runs side by side before your defense.
Thesisroom has ten rooms. Research Compass shows what published research already holds near a topic you type. Scope Check reads a research proposal and asks about the places where what it asks for and what it plans do not obviously meet. Defense Practice writes the questions an examiner or committee could ask, each tied to a passage in your thesis. Citation Check looks every source up in published research databases such as Crossref and OpenAlex, drawing on nine registries. Crossref and OpenAlex find the work, then each registry is asked the question it can answer, and the result names every one that answered. Chapter Audit reads each chapter against the standards of your field. Find Sources returns verified papers that support a specific claim. Reference Watch re-checks your bibliography weekly, alerts you when a source is retracted, and names highly cited work that your bibliography does not contain. Paper Match shows whether an uploaded paper supports, contradicts, or complicates your argument. Blind Spots counts the most-cited papers and the recent papers cited above their field average on each claim's topic that you have not cited. Argument Map tracks every claim, definition, and research question across chapters and marks contradictions between chapters.
Thesisroom is a defense preparation tool, not a writing service. It does not rewrite text to bypass plagiarism detectors. It does not fabricate citations. It does not ghostwrite chapters. It holds you to the same standard your examiner or committee does.
Read the full integrity policy