How to write the results chapter: present what your data show, in the order your research questions were asked, using tables for the numbers and text for the pattern — and say nothing about what it means. Chapter 4 reports; Chapter 5 interprets. Nearly every results chapter that comes back with corrections has either mixed those two jobs, buried the findings in raw software output, or answered fewer questions than Chapter 1 asked. This guide fixes all three.
Step 1: Draw the Line Between Chapter 4 and Chapter 5
Before writing a sentence, learn the test that decides where any given sentence belongs. If a sentence could be true only after looking at your data, it is a result. If it requires an argument, another author, or a judgement about why, it is discussion.
Result: “Perceived workload was negatively correlated with job satisfaction, r(148) = −.42, p < .001.” Discussion: “This is consistent with Rahman (2024), and may reflect the shift patterns common in the sector.” The first belongs in Chapter 4; the second does not. A useful discipline while drafting: if you find yourself typing “because”, “this suggests”, “in line with” or any citation, stop — you have crossed into Chapter 5.
Some faculties do require a combined “Results and Discussion” chapter, and a few qualitative programmes expect interpretation woven in. Your handbook decides; the rest of this guide assumes the separated structure that most Malaysian faculties use.
Step 2: Follow the Standard Presentation Order
Examiners read Chapter 4 looking for specific things in a predictable sequence. Give them that sequence:
- Response rate and sample profile. How many were distributed, how many returned, how many usable, and the demographic breakdown. “Of 250 questionnaires distributed, 212 were returned and 198 were usable, giving an effective response rate of 79.2%.”
- Data screening. Missing data and how you handled it, outliers, and the assumption checks your chosen tests require.
- Reliability of the instrument. The reliability coefficient for each construct, reported for the data you actually collected — not the value from the original study you adapted the scale from.
- Descriptive statistics. Means and standard deviations for each variable, usually in one table.
- Inferential results, one section per research question or hypothesis, in exactly the order they appeared in Chapter 1.
- A summary of findings — a short table or list mapping each hypothesis to supported or not supported.
The traceability requirement in point 5 is where chapters most often fail silently: three research questions in Chapter 1 and two answered in Chapter 4 is an instant correction, and it is entirely visible before submission.
Step 3: Build the Table First, Then Write the Text
Write each results section by making its table or figure first and its prose second. A table that has to be built around existing prose usually ends up duplicating it.
The APA table conventions your examiner will check: a table number above the table, then a brief italicised title on its own line; horizontal rules only — no vertical lines, no shading, no coloured headers; every column headed and every unit stated; and general notes below the table. Numbers are aligned on the decimal point, and values that cannot exceed one — correlations and p values — are written without a leading zero, as .42 rather than 0.42. Full formatting rules sit alongside the citation rules in our guide to APA 7th edition format, and your faculty template overrides the manual where the two differ.
One rule saves more marks than any other: never paste raw software output. An SPSS window screenshot, with its grey borders and its columns you did not ask for, tells the examiner you did not decide what mattered. Retype the numbers you need into a clean table. If your faculty wants the full output as evidence, it goes in an appendix.
The prose beside a table should not read the table aloud. Its job is to state the pattern and point at the number that proves it: “Job satisfaction recorded the highest mean among the four constructs (M = 4.12, SD = 0.58), while perceived workload recorded the lowest (M = 2.87, SD = 0.91), as shown in Table 4.3.”
Step 4: Use the Standard Sentence Template for Each Test
Every statistical test has a conventional reporting form, and using it makes your chapter instantly readable to an examiner. Statistical symbols are italicised; report exact p values to two or three decimals, and write p < .001 when the value is smaller than that. These templates cover the tests most Malaysian FYPs use — choosing between them is a separate decision, covered in our guide to which statistical test to use.
- Independent-samples t-test: “Male respondents (M = 3.84, SD = 0.62) scored significantly higher than female respondents (M = 3.51, SD = 0.70), t(196) = 3.42, p = .001, d = 0.49.”
- One-way ANOVA: “A one-way ANOVA showed a significant difference in satisfaction across the three departments, F(2, 195) = 5.62, p = .004.” Follow with the post-hoc test and state which pairs differed.
- Pearson correlation: “Training frequency was positively correlated with performance, r(196) = .38, p < .001.”
- Multiple regression: “The model was significant and explained 41% of the variance in retention, R² = .41, F(3, 194) = 44.87, p < .001. Perceived workload was the strongest predictor (β = −.34, p < .001), while training frequency was not significant (β = .07, p = .284).”
- Chi-square: “The association between department and preferred training mode was significant, χ²(2, N = 198) = 8.94, p = .011.”
Two habits complete the pattern. Report an effect size wherever your test provides one — significance tells the reader the effect probably exists, effect size tells them whether it matters, and examiners increasingly ask for both. And state the decision explicitly: after each test, write one sentence saying whether the hypothesis was supported, so your summary table has something to summarise.
Step 5: Report Non-Significant Results Properly
Every hypothesis you stated gets an answer, and “no significant relationship was found” is an answer. Report it in the same format as a significant result, with the same statistics. Two things to avoid: dressing it up (“approaching significance”, “a marginal trend”) — a result above your alpha is not significant and hedging reads as reluctance; and quietly dropping the hypothesis from the chapter, which is an integrity problem rather than a presentation choice, and one an examiner detects by counting Chapter 1’s questions.
Resist explaining why the result was not significant here. That explanation — small sample, restricted range, a measurement issue — is genuinely valuable, and it belongs in Chapter 5.
Step 6: Handle Qualitative Findings the Same Way
A qualitative Chapter 4 follows the same discipline in different clothing. Open with the participant profile in a table, using pseudonyms or codes rather than names. Organise the chapter by theme, not by participant and not by interview question, and introduce each theme with a short statement of what it captures and how widely it appeared. Support each theme with one or two verbatim quotations, each attributed to its code, indented and reproduced exactly — including the participant’s own phrasing and any code-switching between Malay and English, which is normal in Malaysian data and should not be silently translated away. If you translate a quotation, present the original and mark the translation as yours.
The line holds here too: the quotation is the finding; what the theme reveals about the phenomenon is discussion. Quantify carefully as well — “most participants” is vague, “seven of the twelve participants” is a result.
Step 7: Run the Consistency Audit
Before submitting, check four things in order, all of which take under an hour and each of which is a standard examiner catch.
- Question coverage. Put Chapter 1’s research questions beside Chapter 4’s sections. Every question has a section; every section answers a question.
- Method match. Every analysis reported here was named in your methodology chapter. An unannounced test raises the question of how many others you ran.
- Number match. Each figure in the text matches its table, and the sample size is the same everywhere — the number after data cleaning, not the number distributed.
- Cross-reference match. Every table is numbered in the order it is mentioned, and every table is actually mentioned in the text.
Then update your abstract, because the figures you finally settled on are the ones it should quote.
Common Mistakes That Send Chapter 4 Back
- Interpreting. Citations and explanations in the results chapter, which then leave Chapter 5 with nothing to say.
- Pasted output. Software screenshots instead of tables you designed.
- Narrating the table. A paragraph that lists every cell rather than naming the pattern.
- Missing assumption checks. Reporting a parametric test without evidence that its assumptions held.
- Shifting sample sizes. 250 in one paragraph, 198 in the next, with no explanation of the difference.
- Unanswered hypotheses. Fewer results sections than research questions.
- Chart clutter. Three-dimensional pie charts and decorative colours where a plain table would be clearer.
Chapter 4 is where careless formatting costs marks you already earned with your data. Tesify keeps your chapters, tables and citations in one structured place, so your figures stay consistent from the results section to the abstract. Your thesis, 100% written by you. Used by 9,000+ students. Write Chapter 4 in Tesify.
Frequently Asked Questions
What is the difference between Chapter 4 and Chapter 5?
Chapter 4 reports what the data show; Chapter 5 explains what it means, compares it with earlier studies and draws implications. If a sentence needs a citation or an argument, it belongs in Chapter 5.
How long should the results chapter be?
However long the findings require — it is the one chapter whose length is set by your data rather than a convention. Tables carry most of the content, so a thorough Chapter 4 is often shorter in words than the literature review.
Should I include SPSS output in Chapter 4?
No. Retype the values you need into clean tables you designed. If your faculty wants the full output as evidence, it belongs in an appendix, referenced from the chapter.
Do I report descriptive statistics for demographic variables?
Yes, as frequencies and percentages in the sample profile table. Means are for your measured variables; counts and percentages are for categories such as gender, department or year of study.
Where do reliability results go?
In Chapter 4, before the inferential analyses, computed on your own collected data. Reporting only the original developer’s reliability value is a common shortcut examiners notice, because it says nothing about your sample.
Should I report results that were not significant?
Always, in the same format as significant ones. Every hypothesis stated in Chapter 1 needs an answer, and omitting an unsupported one is an integrity problem rather than an editing decision.
How do I report an exact p value?
Give it to two or three decimals without a leading zero — p = .032 — and write p < .001 when it is smaller than that. Avoid reporting p = .000, which your software prints but which is not a possible value.
Do I need effect sizes?
Report them wherever the test provides one. Significance indicates that an effect is unlikely to be chance; effect size indicates whether it is large enough to matter, and postgraduate examiners increasingly expect both.
Should each table have a paragraph explaining it?
Each table needs text that names the pattern and points to the key value, not a paragraph reciting the cells. If your prose repeats the table, delete the prose and keep the table.
How do I present findings from mixed methods?
Usually quantitative results first, then qualitative themes, then a short integration section showing where they converge or conflict — with the integration kept descriptive here and interpreted in Chapter 5.
Can I use charts instead of tables?
Use a figure when shape or trend is the point, and a table when the precise values are. Never present the same data as both, and avoid decorative three-dimensional effects, which distort the very comparison the figure exists to make.
My results contradict my hypotheses. Is that a problem?
No — unsupported hypotheses are a normal, publishable outcome and examiners are used to them. Report them accurately here and use Chapter 5 to explain plausibly why. What damages a thesis is a chapter whose results have been quietly tidied to agree with its predictions.