Data Chapters With Unexpected Results: How to Write the Chapter Your Study Isn't Giving You

A doctoral candidate I worked with last spring had spent nine months on her data chapter. The proposal had predicted a clear directional finding — interview participants would report a specific shift in practice after the new protocol was introduced. The dissertation plan was built around that finding. The IRB was approved around that finding. The committee had signed off on the methodology on the strength of the finding it was designed to surface. The data, when it finally came in, showed something else. Not the opposite — something adjacent. The pattern the interviews produced was real, it was substantial, and it was not the pattern the proposal said would appear. She emailed me a single sentence: I don't think I have a dissertation anymore.

This is the most common catastrophic sentence in mid-dissertation work, and almost no one writes about it. The data chapter is the part of the dissertation where the proposal meets reality, and reality doesn't always agree with the proposal. When the meeting goes well, the chapter is mechanical — you report the findings you predicted, discuss them against the literature, link to the conclusion. When the meeting goes badly, the chapter becomes the thing the candidate can't write. The committee is waiting for the finding the proposal promised. The candidate knows the data doesn't show it. The candidate assumes the chapter is now unwriteable, the proposal is now wrong, the dissertation is now impossible, and the only honest move is to delay until the situation resolves itself. The situation does not resolve itself. The chapter does not get easier by waiting.

This article is for the candidate sitting in front of a data set that didn't go the way the proposal said. Below are four sections covering what "unexpected" actually means in a data chapter, why the standard chapter structure fails you when the results diverge, the four-move protocol for presenting an unexpected finding without surrendering the dissertation, what the committee will actually push back on (and what they won't), and — when it applies — the harder question of whether the real problem is the data or the proposal. The goal is to give you a way to write the chapter the data is actually giving you, not pretend it is a different chapter than the one in front of you.

What "Unexpected" Actually Means in a Data Chapter

"Unexpected" in a data chapter is doing more work than it looks like. It can mean four very different things, and each one writes differently. The first is a result that is directionally opposite to the predicted one — you expected an increase, the data shows a decrease. The second is a result that is directionally consistent but the wrong size — you expected a small effect, the data shows a large one, or vice versa. The third is a result that is the right kind of finding on a different question — your predictors didn't behave the way you hypothesized, but a different relationship inside the data did. The fourth is no clear result at all — the data is messy, null, mixed, or underpowered, and there isn't one finding to point at.

The reason the distinction matters is that each one has a different writeable chapter structure. The opposite-direction chapter needs to defend the new direction against the predicted one and reframe the contribution. The wrong-size chapter needs a discussion section explaining why the magnitude shifted and what that means. The different-question chapter needs a re-statement of the research question — quietly, with permission from the committee — and a re-anchoring of the contribution. The no-clear-result chapter is the hardest. It needs a methodologically honest data presentation, a discussion that theorizes the null, and a conclusion that names what was learned from the absence. None of these are failures. All of them are writeable chapters that have defended and passed.

The thing to know is that the proposal never actually promised the result. The proposal promised a question, a method, and a framework. It said: if the data shows X, we will interpret it against Y. The chapter is allowed to say: the data showed Z, and we will interpret it against Y. Most candidates, when the actual results diverge from the prediction, hear the proposal as a contract for a specific answer. It isn't. It is a contract for a specific kind of inquiry. As long as the inquiry was rigorous and the answer is reported honestly, the dissertation is intact. If blank-page paralysis is the part that's actually unmapped — the inability to open the chapter at all because the predicted content isn't there — Procrastination: The Three Types walks through why this stalls most candidates and which of the three types usually drives the data-chapter freeze.

This week's tactic: before you write a single word of the chapter, classify the unexpected result into one of the four types above. Resist the urge to write your way into the classification. Write the classification first — one paragraph, no more — and pin it above your drafting scratch file. Every subsequent decision about chapter structure is downstream of that paragraph. If you find yourself resisting the classification, that resistance is the signal: it is the thing the chapter will need to argue, and it is the thing the committee will eventually ask you to defend.

Why the Usual Chapter Structure Fails You

The standard chapter-four template — restate the research questions, present the results, discuss, link to the next chapter — is built for the predicted case. Each section has an implicit contract: the research questions will come back with yes-or-no answers; the results section will report magnitudes; the discussion will interpret those magnitudes against the literature; the link will say what comes next. When the actual findings match the prediction, the template works almost without effort. When the findings diverge, every section of the template becomes a place where the candidate has to lie. Restating the research questions as written becomes dishonest if the data doesn't answer them. Presenting the results becomes painful if the magnitudes don't match. The discussion becomes unwriteable if the interpretation would have to pretend the data said something it didn't.

The candidate's instinct, under that pressure, is one of two moves. The first is to write the chapter as if the data matched the prediction. This produces a clean structure but eventually collapses at the defense, where the committee reads the methodology, the data tables, the discussion, and notices the alignment isn't there. The second is to write the chapter as a confession — to open with "the data did not show what was expected," spend the chapter explaining why, and leave the reader unsure whether the dissertation is finished or whether more data collection is required. Both are wrong moves. The first is dishonest. The second is excessive. Neither is the right chapter.

The right chapter is one where the structure bends to fit the data without abandoning the proposal's purpose. The proposal's purpose was a research question, not a specific answer. A data chapter that presents what the data actually showed, names the type of divergence from the prediction, defends the analytic choices that produced the unexpected finding, and reframes the contribution in light of it — that is a chapter. It is a more honest chapter than the predicted-result version would have been. It is also a chapter the committee can pass, because the committee's job at the defense is to evaluate whether the candidate conducted rigorous inquiry and reported it accurately. The committee's job is not to enforce that the data matched the proposal. The committee knows the data doesn't always match the proposal. They have read dissertations where the data didn't match the proposal. They have supervised candidates through the rewrite. They have signed off on the revised chapter. The candidate is the one who doesn't know this yet.

This week's tactic: take the standard chapter-four template and mark, in pencil, the sentences that assume the data will confirm the prediction. There are usually three to five of them — a sentence in the research-questions restatement ("as hypothesized"), a sentence in the results framing ("consistent with the prediction"), a sentence in the discussion ("supports the framework of…"), a sentence in the link ("these findings suggest the next chapter will address…"). Cross each one out physically. Now read the chapter skeleton without them. What remains is the actual structure of the data chapter you have. The structure is shorter than the textbook version. That is fine. Honest chapters are usually shorter than performative ones.

The Four-Move Protocol: Present, Explain, Defend, Reframe

When the data chapter has to absorb an unexpected finding, four moves carry almost the entire load. They are: present what the data actually showed, explain why the divergence is interpretable rather than a methodology failure, defend the choices that produced the finding, and reframe the contribution of the study in light of it. Each move does a specific job, and the chapter fails when one of them is skipped.

Present. The first move is to describe the finding without flinching. This is harder than it sounds. Candidates under surprise pressure soften the language, hedge the magnitudes, organize the tables so the eye slides past the unexpected finding. Don't. The first paragraph of the results section should name what the data showed, in plain language, without framing it as a problem to be solved in the next paragraph. The data showed what it showed. State it. The committee will read this paragraph four times — once now, once at the proposal-defense review, once with the chair, once at the defense. Every softening in this paragraph becomes a question later. The discipline is to write it the way you would describe a finding you had predicted.

Explain. The second move is to explain why the finding is interpretable — that is, why it is a result rather than an artifact. This requires the candidate to do a specific kind of methodological work in the discussion section: examining the data collection process for bias, the analytic choices for sensitivity, the sample for representativeness, and the operationalization of key variables for whether they measured what the proposal said they would. The goal here is not to convince the committee that the finding matches the prediction. The goal is to convince the committee that the finding is real, that the methodology supported it, and that the divergence from the prediction is a feature of the inquiry rather than a flaw in the research. This move is where most candidates underperform. They offer the explanation in a single paragraph and move on. The committee reads it as insufficient. The explanation needs to be a section — pages, not paragraphs — and it needs to address the most obvious alternative interpretations directly.

Defend. The third move is to defend the choices that produced the finding. The committee will, at the defense, ask versions of "why this method" and "why this analysis" and "why this sample." The chapter needs to answer those questions on the page, not at the defense table. The defense is a check on whether the chapter can defend itself. A chapter that cannot defend its analytic choices on the page cannot be defended at the defense. This is the move that closes most candidates down — they think they need to defend the finding, when they actually need to defend the methodology. The methodology is defendable. The finding is what the methodology produced. The candidate's job is to defend the rigor of the inquiry; the inquiry's job is to produce the finding.

Reframe. The fourth move is to reframe the contribution of the study. Every dissertation makes a contribution. When the data matches the prediction, the contribution is straightforward — the study confirmed (or refined) what was expected. When the data diverges, the contribution is still real, but it has shifted. The reframe is to name where the contribution now sits, in light of the unexpected finding. This is not spin. It is a precise intellectual move: the contribution is no longer "we showed X," and it is also not "we tried to show X and failed." It is "the inquiry surfaced Y, which the proposal's framework did not predict and which is now a contribution to the literature in its own right." The reframe is the move that converts the unexpected chapter from a liability into a contribution. Without it, the chapter reads as a failed prediction. With it, the chapter reads as a research surprise that produced new knowledge.

This week's tactic: draft the chapter as four discrete sections, one per move. Don't try to integrate them into a flowing narrative on the first pass. Just write them in sequence: a present section that describes the finding; an explain section that addresses the most likely alternative interpretations; a defend section that walks through the methodology choices; a reframe section that names the new contribution. The integration comes in the second pass. The first pass is to get the four moves on the page. Most candidates who have been stuck for months find that the present section is three paragraphs they could write today. The block is downstream of those three paragraphs, not upstream.

What the Committee Will Actually Ask

The committee does not, in practice, ask the questions the candidate fears. The candidate fears the committee will say "your data doesn't match your proposal" or "this finding contradicts your hypothesis" or "what do we do now." The committee almost never opens with any of those. The committee opens with methodology questions — how the sample was recruited, how the data was coded, how the analysis was run, what was done to control for the variables most likely to confound. The committee opens with methodology because methodology is what they can evaluate. The candidate is usually braced for the methodology questions to be attacks. They are not. They are the committee doing their job. The candidate's task is to answer them cleanly, with the data underlying each answer visible, and to use the methodology answers to establish that the finding is what the methodology produced.

There are three specific questions that come up in almost every defense of an unexpected-result chapter. The first is a version of "what would have to be true for this finding to be an artifact rather than a result?" — the committee asks this because they want to see the candidate demonstrate they have considered the alternative interpretations. The second is a version of "if you could re-run the study with one change, what would it be and why?" — the committee asks this because it tests whether the candidate understands the methodological levers well enough to name them. The third is a version of "what does this finding now allow us to claim that the proposal didn't anticipate?" — the committee asks this because they want to see the candidate articulate the reframed contribution. None of these questions is a trap. All three of them are direct invitations to use the chapter's present/explain/defend/reframe structure out loud.

The candidate's best move at the defense is to anticipate the three questions above, draft answers to each one before the defense, and bring those drafts into the room as a written reference. The discipline of writing the answers in advance does two things: it surfaces the methodological gaps in the chapter while there is still time to revise them, and it converts the defense from a Q&A with a hostile committee into a structured conversation with the chair. The chair is, almost always, on the candidate's side at this stage. The chair has read the chapter. The chair has flagged the unexpected finding internally. The chair wants the defense to go well. The candidate's job is to give the chair something to point at when the committee asks their questions. The four-move chapter structure is that something.

This week's tactic: write out answers to the three committee questions above — the artifact question, the re-run question, the reframed-contribution question — in advance of any defense scheduling. Each answer should be one page, single-spaced, plain language, with the relevant table or figure from the chapter referenced by page number. Read each answer out loud to yourself. If any sentence sounds like a hedge, rewrite it. The hedges are where the committee will pull on the thread. If the advisor relationship is the part that has gone cold during this stretch — your chair has stopped responding or stopped pushing — the move is named explicitly in Advisor Conflict Protocol, and the three-outcome decision rubric applies whether the divergence is a methodology problem or a personal one.

When the Real Problem Is the Proposal, Not the Data

There is a smaller, harder case where the unexpected result is not really unexpected — the proposal was built on a prediction the methodology could not have produced, and the data showed what the methodology was always going to show. In this case, the four-move chapter structure still works, but the reframe move is bigger. The contribution is no longer "the inquiry surfaced a new finding." The contribution is "the inquiry revealed that the original framework was insufficient to explain the phenomenon." This is a more significant intellectual move, and it requires explicit permission from the committee — usually in the form of a revised research question, a revised framework, or a revised methodology section in the chapter before the data chapter. The candidate cannot make this move unilaterally. The candidate has to bring it to the chair and the committee. The chair has to agree to it. The committee has to bless it before the defense.

This case is rarer than candidates fear. Most unexpected findings are interpretable findings that need a four-move chapter. The proposal-revision case is reserved for situations where the methodology as designed could not have produced the predicted finding — where the data didn't just diverge, it diverged because the original framework was missing a variable the data was always going to surface. The candidate usually senses this one before they can name it. The chapter feels unwriteable because it is unwriteable as currently framed. The proposal needs a quiet revision. The chair needs to agree. The committee needs a memo. The defense is still passable, but it requires more setup than the standard four-move chapter does. If you are sensing this case in your own work, the move is to schedule a conversation with the chair — not an email, a real conversation, forty-five minutes — and to walk through what the methodology could and could not have produced. If you can't get the chair into a conversation, you have a chair problem, not a data problem. The protocol for that is different.

The discipline in either case is the same. Write the chapter that the data gave you, not the one the proposal imagined. The proposal was a contract for rigorous inquiry. It was not a contract for a specific answer. A dissertation that conducts rigorous inquiry and reports it honestly — even when the honest report diverges from the prediction — is a dissertation that defends and passes. The candidate sitting in front of a data set that didn't match the prediction is the candidate sitting in front of the actual dissertation. The dissertation they're afraid they lost is the dissertation the proposal sketched. That one was never going to be the dissertation they wrote.

This week's tactic: if you can't tell which case you are in, do one diagnostic move. Re-read the methodology chapter as if you were a methodologist who had not seen the proposal. Ask yourself: was the methodology capable of producing the prediction? If yes, the unexpected finding is interpretable and the four-move chapter works. If no, the methodology was always going to produce something different, and you are in the proposal-revision case, which requires a chair conversation before the chapter can move forward. The diagnostic takes an hour. The chapter, after the diagnostic, takes weeks. The chapter, without the diagnostic, takes months. The methodology chapter itself — how it is structured, what it commits to, how it constrains the data chapter — is covered in How to Write Your Dissertation Methodology Chapter, and if the methodology section is also frozen, that's the upstream chapter to unblock first.