Dissertation data analysis and results support turns the final evidence produced or selected through the approved methodology into transparent answers to the research questions without changing the analytical logic to obtain a preferred finding.
WritersArch treats analysis and results as a traceable evidence stage. The work can include defining the final analytic dataset or qualitative corpus, documenting exclusions and cleaning decisions, checking whether the planned analysis still fits the evidence, executing quantitative or qualitative procedures, integrating mixed-method findings, verifying software output, organizing tables and figures, and reporting expected, unexpected, null, or contradictory findings accurately. Interpretation beyond the immediate result belongs primarily to the later discussion stage.
What should dissertation data analysis and results establish?
Dissertation data analysis and results should establish what evidence entered the analysis, how that evidence was analysed, what the procedure produced, and which findings answer each research question.
- Final evidence: Which observations, interviews, documents, cases, records, measurements, or other evidence were actually analysed.
- Preparation decisions: Which cleaning, coding, exclusion, transformation, transcription, or data-management decisions affected the analytic material.
- Method fit: Whether the approved analytical procedure remains appropriate for the evidence that was actually obtained.
- Analysis execution: How quantitative, qualitative, mixed-method, or other approved procedures were applied.
- Traceable findings: Which output, pattern, estimate, theme, comparison, model, or other result answers each research question.
- Transparent reporting: How tables, figures, quotations, statistical output, and prose communicate results without manufacturing interpretation.
Lock the analysis to the final research questions and methodology
The analysis should begin from the final approved research questions and the methodology that produced the evidence rather than from whichever analytical procedure appears most likely to produce an interesting result.
| Alignment Element | What to Confirm Before Analysis | Why It Matters |
|---|---|---|
| Research question | The exact final wording and the type of answer each question requires. | Prevents analyses that are statistically or technically valid but irrelevant to the dissertation question. |
| Research design | The design actually implemented rather than only the design originally proposed. | Controls the kinds of claims the results can later support. |
| Sample or cases | Who or what actually entered the final evidence base. | Shows whether recruitment, access, attrition, or eligibility changed the planned evidence. |
| Variables or constructs | Which measures, categories, prompts, concepts, or extracted attributes are genuinely available. | Prevents the analysis from assuming evidence that was never collected or recorded. |
| Analysis plan | The procedure approved or justified for each research question. | Distinguishes planned analysis from later exploratory work. |
If implementation changed the sample, variables, instrument, data source, case boundary, or available evidence, that change should be documented. A substantial departure may also require the methodology chapter to describe what was actually done.
Establish the final analytic evidence before calculating or coding
The final analytic evidence should be defined before formal analysis so the researcher can identify exactly what was included, excluded, corrected, transformed, or left incomplete.
- Preserve the original evidence or source files separately from analysis-ready versions where the project permits it.
- Identify every participant, observation, case, transcript, document, record, measurement, or other unit eligible for analysis.
- Apply approved inclusion and exclusion rules consistently.
- Document missing, withdrawn, duplicate, corrupted, incomplete, or unusable material.
- Record transformations, recoding, derived variables, transcript corrections, anonymisation, or other preparation decisions.
- Freeze or version the analytic evidence so later changes remain traceable.
- Confirm that the final evidence still supports the planned analytical procedure.
Prepare quantitative data without erasing the audit trail
Quantitative data preparation should correct or document genuine data problems while preserving a reproducible record of how the original dataset became the analysis-ready dataset.
| Data Check | Analytical Question | Defensible Response |
|---|---|---|
| Variable names and types | Are numeric, categorical, ordinal, date, text, and identifier fields represented correctly? | Correct verified type or coding errors and retain a record of the change. |
| Ranges and categories | Do values fall within possible or defined ranges and category sets? | Investigate impossible or unexpected values rather than automatically deleting them. |
| Duplicate records | Does more than one row represent the same underlying observation when it should not? | Resolve duplicates through a documented rule based on the source data and design. |
| Missing values | Which variables and observations are incomplete, and could missingness affect the analysis? | Use a method-appropriate missing-data decision instead of treating deletion or replacement as neutral. |
| Derived variables | Are composite scores, transformations, categories, or calculated variables constructed according to the approved rule? | Document the exact calculation and verify it before analysis. |
| Unusual observations | Is an extreme value an error, a legitimate case, a subgroup signal, or a characteristic of the phenomenon? | Investigate first and base retention, transformation, or exclusion on a defensible analytical reason. |
How should missing data and outliers be handled?
Missing data and outliers should be handled according to their cause, structure, analytical consequence, research design, and relevant methodological guidance rather than through one universal deletion or replacement rule.
Missingness can affect different analyses in different ways. An outlier can reflect a recording error, a valid rare case, a different subgroup, or the phenomenon itself. Decisions about complete-case analysis, imputation, transformation, sensitivity checks, retention, or exclusion should therefore follow the actual method and be documented before their effect on the preferred conclusion becomes the reason for choosing them.
Prepare qualitative material for a transparent analytical process
Qualitative preparation should preserve the connection between the original material and the later codes, categories, themes, cases, narratives, or other analytical findings.
- Transcripts: Check accuracy against recordings where the approved process permits it and record consequential corrections.
- Identifiers: Use consistent participant, case, document, or event identifiers without exposing protected identities.
- Context: Preserve enough contextual information to understand what an extract or observation means.
- Source separation: Keep original evidence distinguishable from codes, annotations, memos, categories, and interpretations.
- Analytical notes: Record important coding and categorisation decisions so later findings can be reconstructed.
- Corpus boundary: Define which interviews, documents, field notes, records, or cases entered the analysis.
Distinguish raw evidence from codes and findings
Raw qualitative evidence, analytical codes, categories, themes, and final findings should remain conceptually distinct so the reader can see how interpretation developed from the source material.
A quotation does not become a theme merely because it is vivid, and a code does not become a major finding merely because it appears frequently. The analytical method should determine how recurrence, salience, contradiction, context, case differences, and conceptual importance contribute to the result.
Test whether the planned analysis still fits the actual evidence
The planned analytical method should be rechecked against the final evidence because an approach that was defensible at proposal stage can become unsuitable after the real sample, variables, records, or qualitative material are known.
| Evidence Condition | Fit Question | Required Decision |
|---|---|---|
| Variable or measurement structure differs from the proposal | Does the planned quantitative procedure still match the actual measurement level and available variables? | Retain the planned procedure only if its requirements remain defensible; otherwise justify the change. |
| Sample structure changed | Do group sizes, clustering, repeated observations, attrition, or case availability alter the analysis? | Reassess the procedure and the scope of inference. |
| Assumptions or model conditions are problematic | Can the intended procedure still produce a defensible result? | Use appropriate diagnostics, alternatives, transformations, sensitivity checks, or limitations where justified. |
| Qualitative evidence lacks expected depth or coverage | Can the selected analytical approach still answer the question from the material available? | Reassess the analytic claim rather than forcing the evidence into the planned structure. |
| Mixed-method strands do not align as expected | Can the planned integration still be executed meaningfully? | Preserve convergence and divergence rather than manufacturing agreement. |
An analytical change should be made because the evidence and method require it, not because another procedure produces a stronger-looking finding.
Use quantitative analysis to answer the research question
Quantitative analysis should use only the descriptive, comparative, associational, predictive, modelling, or other procedures needed to answer the research questions under the approved design.
| Analytical Purpose | Typical Evidence Structure | Possible Analysis Family | Reporting Focus |
|---|---|---|---|
| Describe a sample or variable | Numeric or categorical observations | Frequencies, proportions, measures of central tendency, variability, or distributions | What the observed data look like without implying a relationship that was not analysed. |
| Compare groups or conditions | Comparable measurements across defined groups, conditions, or periods | Method-appropriate group comparison procedures | Magnitude and direction of the comparison together with relevant uncertainty or supporting statistics. |
| Examine association | Two or more measured variables or constructs | Correlation, contingency analysis, regression, or another method suited to the data and design | Strength, direction, estimate, uncertainty, and the non-causal boundary where applicable. |
| Model or predict an outcome | Outcome information plus relevant predictors | Regression, classification, time-series, survival, multilevel, or another justified model family | Model estimates and performance at the level required by the design without overstating prediction as explanation. |
| Evaluate change over time | Repeated or longitudinal observations | Method-appropriate repeated-measure, longitudinal, trend, or time-based analysis | Observed change and uncertainty while respecting the actual temporal design. |
These are analysis families rather than automatic test prescriptions. The exact procedure depends on the final research question, measurement structure, design, assumptions, sample characteristics, methodological guidance, and approved analysis plan.
Report estimates, uncertainty, and magnitude where the method requires them
Quantitative findings should report the information needed to understand the result rather than reducing every conclusion to a significant or non-significant label.
Depending on the design and analysis, relevant reporting may include estimates, confidence intervals or other uncertainty measures, effect or association information, descriptive context, model diagnostics, test statistics, degrees of freedom, probability values, fit measures, or other method-specific output. The required combination is not universal and should follow the analytical method, discipline, programme guidance, and actual software output.
Make qualitative analysis show how findings were developed
Qualitative analysis should make the route from source material to the final analytical structure visible enough that themes, categories, cases, narratives, patterns, or concepts are not presented as unsupported labels.
| Analytical Approach | Possible Analytical Structure | What the Results Should Make Visible |
|---|---|---|
| Thematic analysis | Codes, patterns, candidate themes, reviewed themes, and final thematic structure according to the selected approach | How the reported themes represent meaningful patterns in the analysed material. |
| Qualitative content analysis | Codes, categories, concepts, frequencies where methodologically relevant, and interpretive relationships | How the analytical categories were developed and what evidence they represent. |
| Framework analysis | Cases, categories, matrix summaries, comparisons, and framework dimensions | How evidence varies across cases and analytical categories. |
| Grounded-theory-oriented analysis | Codes, categories, relationships, comparisons, memos, and developing conceptual explanation where appropriate | How conceptual relationships emerge through the method actually used. |
| Narrative, discourse, case, or other qualitative analysis | Method-specific analytical units and relationships | How the selected method transforms the source material into defensible findings. |
The selected approach should match the methodology rather than being chosen after analysis solely because one label appears to fit the emerging results.
Does coding frequency determine analytical importance?
Coding frequency does not automatically determine analytical importance because prevalence, salience, contradiction, context, negative cases, and conceptual significance can carry different weight across qualitative methodologies.
A less frequent pattern may matter because it exposes an important exception, boundary condition, subgroup experience, or alternative explanation. The results should follow the logic of the chosen qualitative method rather than treating every code count as equivalent to evidential importance.
Integrate mixed-method results according to the approved design
Mixed-method results should combine qualitative and quantitative evidence according to the integration logic established in the methodology rather than reporting two independent analyses and calling them integrated.
| Integration Relationship | What the Analysis Does | Possible Result |
|---|---|---|
| Convergence | Examines whether the strands support a compatible account of the research problem. | A combined finding showing where evidence agrees. |
| Divergence | Preserves meaningful disagreement between strands. | A result identifying where evidence types produce different answers or conditions. |
| Explanation | Uses one strand to clarify or explain a result from another. | A connected result in which qualitative evidence, for example, explains a quantitative pattern. |
| Expansion | Uses different evidence types to answer complementary dimensions of the problem. | A broader answer that remains explicit about what each strand contributes. |
| Joint comparison | Places evidence types into a common analytical display or relationship. | A joint display, case comparison, integrated matrix, or other design-appropriate synthesis. |
Disagreement should not be corrected into artificial consistency. Divergence can itself be an analytically important result.
Verify software output before treating it as a finding
Software output should be verified against the intended data, variables, filters, coding rules, model specifications, and analytical settings before any table, coefficient, theme report, or visualization is moved into the dissertation.
Quantitative work may use tools such as R, Python, SPSS, Stata, SAS, spreadsheets, or other approved analytical software. Qualitative projects may use applications such as NVivo, ATLAS.ti, MAXQDA, spreadsheets, coding scripts, or manual systems. These tools can implement or organize analysis, but the software name does not establish that the analytical decision is correct.
- Input verification: Confirm that the intended dataset, cases, variables, documents, or transcripts were used.
- Filter verification: Confirm that exclusions, subgroup filters, missing-data rules, or case selections match the analysis plan.
- Model verification: Confirm reference categories, estimators, transformations, interactions, contrasts, or other settings where relevant.
- Coding verification: Confirm code definitions, query settings, coding scope, case classifications, and source coverage where relevant.
- Output verification: Reconcile tables, figures, quotations, statistics, or themes with the underlying analysis before reporting them.
Distinguish planned analysis from exploratory analysis
Planned analysis should remain identifiable from later exploratory analysis so the dissertation does not present post-hoc discoveries as though they were specified before the researcher saw the results.
| Analysis Type | Purpose | Reporting Requirement |
|---|---|---|
| Planned analysis | Answer the predefined research question using the approved or prospectively justified procedure. | Report as the primary analysis where it remains methodologically valid. |
| Justified methodological adjustment | Repair a genuine mismatch between the planned procedure and the final evidence. | Explain what changed, why the adjustment was necessary, and whether the methodology also requires revision. |
| Exploratory analysis | Investigate an unplanned pattern, subgroup, relationship, theme, or anomaly. | Label it transparently as exploratory rather than rewriting it as a pre-specified test. |
Report unexpected, null, weak, and contradictory findings accurately
Unexpected, null, weak, mixed, or contradictory findings should be reported as evidence rather than hidden, repaired, or replaced because they do not support the hypothesis or expected argument.
| Result Pattern | Results-Stage Responsibility | What Belongs Later |
|---|---|---|
| Expected finding | Report the actual estimate, pattern, theme, comparison, or evidence supporting the result. | Explain why it aligns with theory or literature in the discussion. |
| Null or non-supporting finding | Report the result accurately with the method-appropriate supporting information. | Consider explanations and implications in the discussion. |
| Unexpected finding | Report what occurred without rewriting the original hypothesis. | Evaluate plausible explanations later. |
| Contradictory qualitative evidence | Preserve the contradiction, negative case, or variation where analytically relevant. | Discuss what the variation may mean. |
| Mixed-method divergence | Show where the strands disagree. | Interpret the reason and significance of the disagreement in the discussion. |
Organize the results around the research logic
The results chapter should organize findings so the reader can connect each major analytical output to the research question, hypothesis, analytical stage, theme, case, or other structure defined by the methodology.
A research-question structure works well when each question has a distinct analysis. A thematic structure can be appropriate for qualitative findings. A model, hypothesis, case, time-period, or analytical-stage structure may be more useful in other projects. The chapter should use whichever sequence makes the research logic easiest to follow.
The same finding should not be reproduced unnecessarily in prose, a table, and a figure. Prose should direct attention to the result that matters, while the table or figure carries the detailed information it is best suited to display.
Choose text, tables, figures, and quotations by information type
Results should be presented in the format that makes the evidence easiest to inspect, with prose, tables, figures, quotations, and other displays performing different informational roles.
| Format | Use It When | Avoid It When |
|---|---|---|
| Prose | The reader needs the central result, direction, relationship, or analytical transition explained succinctly. | The paragraph would merely reproduce many values already visible in a table. |
| Table | Readers need to inspect multiple related values, categories, model estimates, cases, or comparisons. | The information consists of one simple value or relationship that prose communicates more clearly. |
| Figure | A pattern, trend, distribution, comparison, interaction, process, or relationship becomes clearer visually. | The visual adds decoration but no analytical information. |
| Quotation or qualitative extract | A source excerpt helps illustrate an analytical finding while preserving relevant context. | The extract is treated as though one quotation alone proves the entire theme or pattern. |
| Joint display | A mixed-method design requires connected comparison of qualitative and quantitative evidence. | The two evidence types have no defined integration relationship. |
Formatting, numbering, captions, notes, statistical notation, and placement should follow the student's actual programme and citation-style requirements where those requirements apply.
Keep an analysis trail that supports verification and revision
An analysis trail should preserve the substantive decisions needed to reconstruct how the final evidence became the reported results.
- Quantitative records: Cleaning scripts, syntax, notebooks, transformations, variable definitions, model specifications, diagnostic output, or versioned datasets where appropriate.
- Qualitative records: Codebooks, coding frameworks, memos, case summaries, analytic matrices, decision logs, or versioned coding records where appropriate.
- Mixed-method records: Integration matrices, joint displays, case-linking records, or strand-comparison notes where relevant.
- Output records: Verified tables, figures, model output, theme reports, extracts, and other evidence used in the chapter.
- Decision records: Exclusions, methodological adjustments, sensitivity decisions, recoding choices, and other consequential analytical changes.
The exact documentation depends on the methodology, confidentiality constraints, software environment, and programme expectations. The objective is traceability, not publication of every working file.
Keep results reporting separate from broader interpretation
The results stage should report what the approved analysis found, while broader explanations, comparison with prior literature, theoretical meaning, implications, study-wide limitations, recommendations, and contribution claims belong primarily to the discussion stage.
| Data Analysis & Results Owns | Discussion & Conclusion Owns |
|---|---|
| What evidence entered the analysis | What the finding means in relation to the research problem |
| What analytical procedure was executed | Why the finding may have occurred |
| What estimates, patterns, themes, comparisons, or models were produced | How the finding compares with prior literature |
| Immediate factual context required to understand the result | Theoretical, practical, educational, managerial, clinical, or policy implications where justified |
| Tables, figures, quotations, and other result displays | Study-wide limitations, recommendations, contribution, and final conclusion |
Some programmes combine results and discussion. Where they do, the reporting and interpretation can appear together, but the reader should still be able to distinguish what came from the analysis from what the researcher subsequently interprets.
Use supervisor feedback to locate the analytical defect
Supervisor or committee feedback should be mapped to the part of the analysis workflow it challenges so revision corrects the evidence, procedure, output, or reporting problem rather than merely rewriting the marked sentence.
| Feedback | Likely Problem | Defensible Revision |
|---|---|---|
| “The analysis does not answer the question.” | The analytical procedure and research question are misaligned. | Remap the question to the final evidence and the procedure capable of producing the required answer. |
| “There are too many tests.” | Analyses were accumulated without a clear role in the research logic. | Retain procedures that answer research questions or provide necessary analytical support. |
| “Assumptions are not addressed.” | The selected quantitative procedure lacks required diagnostic or methodological justification. | Evaluate the conditions relevant to that procedure and report the appropriate response. |
| “Themes are descriptive.” | Codes or topic labels are being presented without an analytical relationship. | Revisit how the qualitative method moves from source material to categories, themes, patterns, cases, or concepts. |
| “The table is unclear.” | The display does not reveal what question, variables, categories, cases, or result it represents. | Redesign the table around the information the reader needs to inspect. |
| “The numbers do not match.” | Prose, tables, figures, datasets, or software output may come from inconsistent analytical versions. | Trace every reported value back to the verified analysis output and repair all dependent locations. |
| “Results and discussion are mixed.” | Interpretation, literature comparison, or implications have entered a separately structured results chapter. | Retain immediate result context and move broader interpretation to the appropriate discussion section. |
Run a data analysis and results coherence check
Data analysis and results are coherent when the final evidence, preparation decisions, analytical procedure, software implementation, reported outputs, and research-question answers can be traced through one consistent analytical chain.
- Research questions: Which final question does each major analysis answer?
- Evidence boundary: What exact observations, participants, cases, documents, records, or transcripts entered the analysis?
- Preparation: Which cleaning, coding, transformation, exclusion, missing-data, or transcription decisions could affect the result?
- Method fit: Does the analytical approach still fit the final evidence and implemented design?
- Quantitative checks: Were the assumptions, diagnostics, uncertainty, magnitude, or model conditions relevant to the selected method addressed?
- Qualitative checks: Can themes, categories, cases, narratives, or other findings be traced to the source material and analytical process?
- Mixed-method checks: Is the relationship between strands actually implemented and is divergence preserved where it occurs?
- Software verification: Were the correct data, filters, variables, codes, models, and settings used?
- Exploration: Is planned analysis distinguishable from justified adjustments and exploratory work?
- Reporting: Are null, weak, unexpected, contradictory, and mixed findings reported rather than hidden?
- Displays: Do tables, figures, quotations, and joint displays add information without unnecessary duplication?
- Traceability: Can important results be reconstructed from the analytical records?
- Boundary: Does the page stop before broader literature comparison, implications, recommendations, contribution, and conclusion unless the programme intentionally combines stages?
What should you send for dissertation data analysis and results support?
Dissertation data analysis and results support works best when WritersArch receives the final research questions, approved methodology, final evidence, analysis plan, variable or coding documentation, software output or scripts, current results draft, programme requirements, and supervisor feedback.
- Final research questions or hypotheses: The exact questions the analysis must answer.
- Approved methodology: The implemented research design and planned analysis.
- Final evidence: An anonymised dataset, transcripts, documents, cases, records, or other permitted research material where appropriate.
- Codebook or variable definitions: Labels, values, scoring rules, derived variables, category definitions, or measurement documentation.
- Data-preparation records: Cleaning notes, exclusion rules, missing-data decisions, transcript preparation, or transformation records.
- Analysis plan: The intended quantitative, qualitative, mixed-method, or other procedure.
- Software materials: Syntax, scripts, notebooks, model output, coding reports, query settings, or other analytical records where available.
- Qualitative analytical materials: Codebooks, memos, coding frameworks, case matrices, themes, or relevant extracts where available.
- Current results draft: Existing prose, tables, figures, findings, or chapter sections requiring review.
- Programme guidance: Requirements governing statistical reporting, qualitative presentation, tables, figures, chapter structure, or combined results/discussion formats.
- Supervisor or committee feedback: Comments identifying analytical, reporting, consistency, or chapter-boundary problems.
Start your data analysis and results support
WritersArch can review the full analytical chain and identify whether the main weakness lies in evidence preparation, method fit, quantitative execution, qualitative analysis, mixed-method integration, software verification, results organization, tables and figures, or reporting consistency.
Send the dissertation materials through the project order page so the analysis can be scoped around the actual research questions, methodology, evidence, and programme requirements. Existing clients can continue through client login.
Move from data analysis and results to the discussion and conclusion
Once the approved analysis has produced transparent findings and those findings have been reported accurately, the next stage is to explain what they mean, compare them with prior research, and develop evidence-proportional implications and conclusions through dissertation discussion and conclusion support.