How to Write the Results Chapter of a Dissertation

The results chapter reports what you found, in the order of your research questions, and stops there. For each question you say what you tested, give the numbers a reader needs to check your conclusion, and state in one plain sentence what the result is. What the result means, why it happened, and what should be done about it all belong in the discussion.

In a five-chapter dissertation this is Chapter 4. Check your program's handbook or chapter template first. Many EdD, DNP, and PsyD programs require specific headings, and your committee will check for them.

Quantitative results do not speak for themselves. A table of coefficients with no sentence telling the reader what it shows is a common reason this chapter comes back.

Results or discussion?

The results chapter says what the data show. The discussion says what that means. Most of the confusion in this chapter is about where that line sits.

Belongs in resultsBelongs in discussion
Who was in the sample and how much data was missingWhy the response rate was low
Descriptive statistics for every variable in the analysisWhether those levels are high or low compared with other studies
Whether assumptions were met, and what you did if they were notWhether the design could have been stronger
The test, the statistics, and a plain statement of the resultHow the result compares with prior literature
Whether each hypothesis was supportedWhy a hypothesis was not supported
Themes and the quotes that support themWhat the themes mean for theory or practice

A quick test for any sentence: if it contains "because," "suggests," "may be due to," or a citation, it probably belongs in the discussion.

How to organize the chapter

If your program gives you required headings, use them. If it doesn't, this order works for a quantitative study.

  1. A short introduction. Restate the purpose and the research questions, word for word from Chapter 1, and say how the chapter is organized. One or two paragraphs.
  2. The sample. How many people were invited, how many responded, how many were kept, and why any were dropped. Then the demographics, in a table.
  3. Data preparation. How you handled missing data, any recoding or scale construction, and the reliability of each scale in your sample.
  4. Descriptive statistics. Means and standard deviations for continuous variables, counts and percentages for categorical ones. Every variable that appears in a later test should appear here first.
  5. Assumptions. For each test, say which assumptions you checked, how, and what you found. If one was violated, say what you did about it.
  6. Results by research question. One section per question, in the same order as Chapters 1 and 3. Restate the question, name the test, report the statistics, and state the result in plain language.
  7. Additional analyses, if any. Anything you ran that was not in your proposal goes here, labeled as exploratory.
  8. Summary. A paragraph or a table listing each research question and its answer. No interpretation.

The order of step 6 matters most. A committee member will have your research questions open next to this chapter. If question 2 is answered before question 1, or a question is never answered, they will notice.

How to report common tests

Each result needs four things: the descriptive statistics behind it, the test statistic with its degrees of freedom, the exact p value, and an effect size. Add a confidence interval where your software gives one. These formats follow APA style (7th edition). Use them unless your program requires something else.

TestReportFormat
Independent-samples t-testM and SD for each group, t, df, p, Cohen's d, CI of the differencet(126) = 2.90, p = .004, d = 0.51, 95% CI [0.13, 0.69]
Paired-samples t-testM and SD at each time point, t, df, p, dt(df) = x.xx, p = .xxx, d = x.xx
One-way ANOVAM and SD for each group, F, both df, p, eta squared, then post hoc comparisonsF(df1, df2) = x.xx, p = .xxx, η² = .xx
Correlationr, df (N minus 2), pr(df) = .xx, p = .xxx
Chi-squareCounts and percentages, χ², df, N, p, Cramér's Vχ²(df, N = xxx) = x.xx, p = .xxx, V = .xx
Multiple regressionModel first: R², F, both df, p. Then each predictor in a table: b, SE, β, t, p, CIR² = .xx, F(df1, df2) = x.xx, p = .xxx
Logistic regressionModel fit, then each predictor in a table: b, SE, odds ratio, CI, pOR = x.xx, 95% CI [x.xx, x.xx], p = .xxx

The formatting rules committees check

  • Report exact p values to two or three decimals (p = .032). Write p < .001 when it is smaller than that. Never write p = .000, which is how some software prints it.
  • No zero before the decimal for statistics that cannot exceed 1, such as p, r, and alpha. Write .032, not 0.032.
  • Italicize statistical symbols (M, SD, t, F, p, r, N, n). Greek letters are not italicized.
  • Two decimals for most statistics. Be consistent across the chapter.
  • Report non-significant results the same way as significant ones, with the full statistics. Do not write "approached significance" or "trended toward." The result met your alpha level or it did not.
  • Do not write "proved." A hypothesis is supported or not supported.

Worked example

These are models to adapt to your own study, not text to copy. The study and the numbers are made up for illustration.

Research question 1: Do burnout scores differ between nurses on night shifts and nurses on day shifts?

Weak: "An independent samples t-test was run in SPSS. The results were significant (p = .004). See Table 3."

Stronger: "Research question 1 asked whether burnout differed by shift. An independent-samples t-test was used. Levene's test indicated equal variances, F(1, 126) = 0.12, p = .73. Night-shift nurses (n = 64, M = 3.42, SD = 0.81) reported higher burnout than day-shift nurses (n = 64, M = 3.01, SD = 0.79), t(126) = 2.90, p = .004, d = 0.51, 95% CI of the difference [0.13, 0.69]. The hypothesis that burnout would differ by shift was supported."

The weak version names the software and a p value. The reader can't tell which group was higher, by how much, or whether the test was appropriate. The stronger version restates the question, names the test, reports the assumption check, gives the group statistics, and ends with a plain statement of the result. It also stops there. It does not say why night-shift nurses are more burned out. That is for Chapter 5.

When the result is not significant

"Research question 2 asked whether burnout was related to years of experience. The correlation was small and not statistically significant, r(126) = -.09, p = .31. The hypothesis was not supported."

Same structure, same level of detail, no apology. A non-significant result is a result. It means you didn't find the effect in this study. It does not mean the effect doesn't exist, and it does not need to be explained away.

Tables and figures

A table holds the numbers. The text tells the reader what to see in them. Don't do both jobs in both places.

Put it in a tableKeep it in the text
DemographicsOne or two sentences on who the sample was
Descriptive statistics for all variablesAny value the reader needs to follow your point
A correlation matrixThe correlations that answer a research question
Regression coefficients for every predictorThe overall model result and the predictors that mattered
Post hoc comparisons across several groupsWhich groups differed

A few rules that hold for APA-style tables:

  • Mention every table in the text before it appears. "Table 3 shows the means and standard deviations by shift."
  • Number tables in the order you mention them. The number goes above the table in bold, with the title in italics on the next line.
  • Don't paste software output. SPSS and Stata output is not formatted for a dissertation. Rebuild the table with only the columns a reader needs.
  • No vertical lines, and horizontal lines only at the top, under the headings, and at the bottom.
  • Explain abbreviations and significance markers in a note under the table.
  • Use a figure only when it shows something a table can't, such as an interaction or a trend over time.

If a paragraph repeats six numbers that are already in a table, cut the paragraph down to the one sentence that says what the table shows.

Qualitative and mixed methods studies

Qualitative

The chapter is usually called Findings, and it is organized by theme, not by interview question or by participant. A workable order:

  1. A short introduction restating the purpose and research questions.
  2. The participants, in a table, with pseudonyms and the characteristics that matter for your study.
  3. A brief account of the analysis, enough to show how you got from transcripts to themes. The full procedure belongs in Chapter 3.
  4. An overview of the themes, in a table that maps each theme to the research question it answers.
  5. One section per theme. State the theme as a claim, give two or three quotes that support it, and say what the quotes show. Note how widely the theme was shared and include any participant who did not fit.
  6. A summary that answers each research question in a sentence or two.

The common problem is a string of quotes with nothing between them. Every quote needs a sentence before it that sets it up and a sentence after it that says what it shows. Describing what participants said is expected here. Explaining it against the literature waits for the discussion.

Mixed methods

Report each strand by its own rules, then show where they meet. That usually means a joint display: a table that lines up the quantitative results against the qualitative themes so the reader can see where they agree, where they differ, and where one explains the other. Two sets of findings that sit side by side and never meet are one of the most common gaps committees flag in a mixed methods dissertation.

What committees push back on

The two I see most: results that claim more than the analysis supports, and students trying to explain away a non-significant result.

  • Claims the analysis doesn't support. "Mentoring reduced burnout" from a cross-sectional survey. A small difference described as a large one. A finding from one sample stated as if it holds everywhere.
  • Explaining away non-significance. "The result would have been significant with a larger sample." Non-significant means you didn't find it. It doesn't mean it isn't there, and it doesn't mean it is. Report it and move on.
  • Results that don't line up with the research questions. A question with no answer, an analysis with no question, or a different order from Chapter 1.
  • A test that is not the one in Chapter 3. If the analysis changed after the proposal, say so and say why.
  • Tables with no interpretation. Coefficients presented as if they explain themselves.
  • Software output pasted into the chapter, or long descriptions of which menus were clicked. Committees want the reasoning and the result, not the procedure.
  • No word on assumptions or missing data. Silence here raises more concern than an honest report of a violation.
  • A p value and nothing else. No descriptive statistics, no effect size, no direction.
  • Interpretation in the results chapter. Explanations, citations, and recommendations that belong in the discussion.
  • Numbers in the text that don't match the tables.

Checklist before it goes to your chair

  • The chapter follows your program's required headings, if it has them.
  • Every research question is answered, in the same order as Chapters 1 and 3.
  • Every test matches the analysis plan in Chapter 3, or the change is explained.
  • The sample, response rate, and missing data are reported.
  • Every variable in a test appears in the descriptive statistics first.
  • Assumptions are reported for each test.
  • Each result has descriptive statistics, the test statistic with degrees of freedom, an exact p value, and an effect size.
  • Non-significant results are reported in full and not explained away.
  • No claim goes beyond what the analysis supports.
  • Every table is mentioned in the text before it appears, and the numbers match.
  • No sentence explains why a result occurred or cites the literature.
  • The chapter ends with a summary that answers each question and does not interpret.

Get the results chapter worksheet

A free two-page Word file: a table that maps each research question to its test, the statistics to report, and the table it goes in, plus fill-in sentences for the common tests and the checklist.

I'll email you once or twice about dissertation help. No newsletter. Reply "stop" any time.

Related guides

Work with Matt

If your data is in hand and you want a methodologist alongside you for this chapter, that's Results Chapter Support ($1,095): I run your analyses or coach you through running them yourself, build APA-formatted tables, and walk you through what every result means. You write the chapter and defend it. If the chapter is already drafted, the Chapter Edit ($595) covers one chapter, back in five business days.

Previous
Previous

How to Write a Data Analysis Plan for Your Dissertation

Next
Next

How to Turn Your Dissertation into Journal Articles