For a quantitative or mixed-methods dissertation, the data analysis report is Chapter 4 — the chapter that turns raw data into evidence for your research questions. This guide covers every stage, from data cleaning through statistical testing to results presentation, for dissertation-level work.
In a dissertation, the data analysis report documents the full analytical process applied to the data collected for your study — from the research questions and dataset through preparation, statistical testing, and visualisation, to a results presentation that answers each research question or hypothesis directly. It typically sits as Chapter 4, between Methodology (Chapter 3) and Discussion (Chapter 5).
The chapter's job is narrowly defined: present what the data show, organized around your research questions, without yet interpreting what it means for the field. Interpretation belongs in the discussion chapter that follows.
| Section | Content |
|---|---|
| Introduction | Restates the research questions/hypotheses; previews chapter structure |
| Sample Description | Final sample size, response rate, demographic/descriptive table |
| Data Cleaning Summary | Missing values, outliers, transformations — what was done and why |
| Preliminary/Descriptive Analysis | Descriptive statistics, distributions, assumption checks |
| Results by Research Question | One subsection per research question/hypothesis, with the test, statistic, and result |
| Summary of Findings | A brief, interpretation-free recap tying results back to each question |
| Appendices | Full output tables, syntax/code, additional plots |
Every dissertation Chapter 4 should open by restating each research question or hypothesis from Chapter 1/3 exactly as previously stated — committees check for consistency. For each one, identify:
Before any analysis, describe who or what is actually in your final dataset:
Data cleaning is the most time-consuming phase of any real data analysis. Document every decision — your report must be reproducible. Key issues to address and report:
Always report what you started with and what you ended with. "The original dataset contained 12,450 records. After removing 340 duplicates and 128 records with missing outcome data, the analysis dataset contained 11,982 records." This transparency is a mark of professional-quality reporting.
Our dissertation statistics specialists run the analysis and write the full chapter — descriptives, hypothesis testing, visualisations, and APA-formatted output tables.
EDA is the phase where you understand the data before applying formal statistical tests. Report:
EDA findings should inform your choice of statistical test. If the data is heavily skewed, you may need non-parametric tests. If groups are very unequal in size, this may affect power calculations.
Report each statistical test with: the test name, the null hypothesis, the test result, the p-value, effect size, and confidence interval. Always check and report whether test assumptions were met.
| Question type | Data type | Appropriate test |
|---|---|---|
| Compare two groups | Continuous, normal | Independent t-test |
| Compare two groups | Continuous, non-normal | Mann-Whitney U |
| Compare 3+ groups | Continuous, normal | One-way ANOVA |
| Compare 3+ groups | Non-normal | Kruskal-Wallis |
| Association between two categorical variables | Categorical | Chi-squared test |
| Correlation between two continuous variables | Continuous, normal | Pearson's r |
| Correlation between two variables | Ordinal or non-normal | Spearman's ρ |
| Predict a continuous outcome | Mixed | Linear regression |
| Predict a binary outcome | Mixed | Logistic regression |
Effective data visualisation communicates findings that text and tables cannot. Rules for dissertation data analysis chapters:
Most committees want APA-formatted tables in the body (not raw software output) with narrative text introducing and interpreting each one. Full/raw statistical output usually belongs in an appendix, referenced from the body.
Report it exactly as you would a supported one — a non-significant result is a legitimate finding, not a failure. State the result plainly in Chapter 4; save the "why might this be" discussion for Chapter 5. Never reshape the analysis after the fact to force significance.
Keep Chapter 4 to "what the data show" — save "what it means" for Chapter 5 (Discussion). Mixing interpretation into the results chapter is one of the most common reasons committees send chapters back for revision.