Statistical analysis in a nursing dissertation connects a clinical or professional question with evidence that can be examined systematically. It helps students describe participants, identify patterns, estimate uncertainty and decide whether the findings support the claims made in the discussion.
Statistics are not separate from nursing. They provide one way of examining questions about patient outcomes, workforce experiences, service quality, education and clinical practice. However, statistical output has value only when it is aligned with the research question, study design, variables and nursing context.
This guide explains that relationship conceptually. It does not repeat the software procedures covered in our step-by-step SPSS data-analysis guide.
The relationship between nursing questions and numerical evidence
Nursing research often begins with an uncertainty arising from practice. A student may want to know whether an education intervention improves medication knowledge, whether burnout differs between staff groups or whether staffing levels are associated with patient outcomes. Statistical analysis helps translate these questions into evidence that can be examined.
The relationship works in both directions. The nursing question determines which data are relevant, while the available data limit which conclusions can be justified. A dissertation should therefore establish alignment before analysis begins. Collecting many variables and testing them later without a clear rationale can produce results that are difficult to interpret and increase the likelihood of misleading conclusions.
From a clinical concept to a measurable variable
Concepts such as pain, anxiety, adherence, clinical confidence and quality of care are meaningful in nursing but cannot be analysed until they are defined operationally. Operationalisation specifies how each concept will be observed or measured.
For example, “anxiety” might be represented by a score from a validated instrument. “Medication adherence” might be measured through a questionnaire, pharmacy record or documented behaviour. Each choice affects the meaning, reliability and limitations of the eventual findings.
The variables must also be classified correctly. A nominal variable places observations into categories without an inherent order. An ordinal variable has ordered categories. Continuous variables can represent measured quantities across a numerical range. These properties influence which summaries and statistical tests are appropriate.
Statistical analysis therefore begins before software is opened. It begins when the student defines the concepts, selects measures and explains how the variables answer the research question.
How research design determines the role of statistics
The same statistical result can mean different things in different study designs. A randomised study, observational survey, retrospective record review and pre–post service evaluation provide different levels and types of evidence.
For instance, an association between workload and burnout in a cross-sectional survey does not establish that workload caused burnout. Other variables may influence both. Similarly, improvement after an intervention does not automatically prove that the intervention caused the change if no suitable comparison or control for alternative explanations exists.
A strong dissertation explains what the design permits the student to claim. Statistical significance cannot repair weaknesses in sampling, measurement or design. Instead, analysis should operate within the boundaries established by the methodology.
Descriptive statistics establish the research context
Descriptive statistics summarise what the dataset contains. Frequencies and percentages can describe categories such as nursing role, ward type or completion status. Means, medians, standard deviations and ranges can summarise numerical variables such as age, length of stay or questionnaire scores.
These summaries are not merely introductory. They help the reader judge whether the sample resembles the population of interest, whether groups are balanced and whether unusual values may affect interpretation. For example, a mean can be misleading when a small number of extremely long hospital stays create a highly skewed distribution. The median may then provide a more representative summary.
Tables should present relevant information without reproducing every value from the software output. The narrative should direct attention to patterns that matter to the research question.
Inferential statistics address differences and relationships
Inferential statistics help students use sample data to evaluate a question about a wider population. A dissertation might examine whether two groups differ, whether scores change over time or whether two variables are related.
The appropriate test depends on the question, variable types, number of groups, independence of observations and relevant assumptions. Our statistical-test selection guide for nursing dissertations explains these decisions in detail.
The test must follow the research question rather than the student’s preferred software technique. Running several tests until one produces a small p-value weakens the credibility of the analysis. Planned analysis, transparent reporting and acknowledgement of exploratory work are more defensible.
Statistical significance is not the same as clinical importance
A p-value contributes information about compatibility between the observed data and a specified statistical model. It does not show the size, importance or clinical value of an effect. A very small difference may be statistically significant in a large sample but irrelevant to patients or services. Conversely, a potentially important effect may remain statistically uncertain in a small dissertation sample.
Students should therefore report and interpret effect sizes and confidence intervals where appropriate. Effect size describes the magnitude of a difference or relationship. A confidence interval communicates the range of values that remain reasonably compatible with the estimate under the model.
Nursing interpretation should then ask whether the effect is meaningful in practice. Does it represent a noticeable improvement for patients? Would it justify changing a service? Is the difference large enough to matter when costs, burden and safety are considered? These questions connect statistical findings back to nursing.
Data quality directly affects statistical conclusions
Analysis cannot correct inaccurate coding, duplicate cases, impossible values or misunderstood missing data. Before testing a hypothesis, students should inspect the dataset, verify labels and coding, identify missingness and confirm that each row represents the intended unit of analysis.
For example, recording repeated measurements from one participant as though they came from different independent participants can produce misleading precision. Treating an undocumented value as zero can also change the result and its interpretation.
Our guide to cleaning nursing research data before SPSS analysis covers the practical checking process. The dissertation methodology should document important cleaning and coding decisions rather than presenting the final dataset as though it appeared automatically.
Assumptions connect the data with the selected test
Statistical tests rely on assumptions about the data and model. These may concern independence, distribution, variance, linearity or expected cell counts. Assumptions are not administrative boxes to tick; they influence whether the analysis provides a defensible answer.
However, students should avoid using one assumption check mechanically. For example, normality should be considered alongside graphical evidence, sample size, the variable being analysed and the robustness of the proposed test. Our normality-testing guide for nursing research explains this decision more fully.
If an assumption is not met, the response might involve a different test, a transformation, a robust method or a more cautious interpretation. The choice should be justified rather than hidden.
Statistical analysis must remain aligned across dissertation chapters
The research question, objectives, methodology, results and discussion should describe the same analytical logic. If an objective asks whether two groups differ, the results should report the relevant comparison. If an objective asks about association, the analysis should examine that relationship.
Misalignment occurs when students introduce tests that were not justified in the methods, report variables unrelated to the objectives or make causal claims from descriptive results. A useful audit is to create a table linking each research question or objective with its variables, statistical method, result and interpretation.
| Dissertation element | Statistical connection |
|---|---|
| Research question | Defines the difference, relationship or outcome being examined |
| Variables | Represent the concepts in measurable form |
| Research design | Determines what can be inferred |
| Statistical method | Provides an analysis appropriate to the question and data |
| Results | Report estimates, uncertainty and relevant test information |
| Discussion | Explains clinical meaning, limitations and implications |
How statistics support evidence-based nursing
Evidence-based nursing integrates research evidence with clinical expertise, patient preferences and the circumstances of care. Statistical analysis contributes to the research-evidence component by showing the size, direction and uncertainty of quantitative findings.
Nevertheless, statistics do not make a decision independently. Evidence may be statistically convincing but difficult to implement, unacceptable to patients or based on a population unlike the local service. Nursing judgement is required to assess applicability, feasibility, equity and potential harm.
The NMC Code requires nurses to practise effectively and use the best available evidence. A dissertation demonstrates this principle when it treats statistics as evidence to interpret critically rather than numbers to report mechanically.
Presenting statistical findings clearly
A results chapter should report findings without hiding them inside copied software output. Each table requires a clear title, understandable labels and appropriate precision. The accompanying text should identify the relevant result rather than repeat every table entry.
Where applicable, reporting should include the test statistic, degrees of freedom, p-value, effect size and confidence interval. The exact requirements depend on the analysis. Our guides to the nursing SPSS results chapter and discussing SPSS results address reporting and interpretation separately.
The discussion should not simply announce whether a result was significant. It should compare the finding with previous evidence, consider plausible explanations, evaluate limitations and explain what the result may mean for nursing practice.
Common misunderstandings about nursing statistics
- “The software chose the test.” Software performs an instruction; the researcher must justify the method.
- “A significant result proves the hypothesis.” A result supports a statistical inference within assumptions and design limitations; it does not provide absolute proof.
- “A non-significant result means no relationship exists.” The study may have insufficient precision, an unsuitable measure or a genuinely small effect.
- “More tests produce a stronger dissertation.” Unnecessary analyses create confusion and increase the chance of misleading findings.
- “Statistics are separate from clinical nursing.” Interpretation requires clinical knowledge, awareness of patient priorities and understanding of practice.
A practical alignment checklist
- The statistical analysis answers the stated research question.
- Every analysed variable has a clear operational definition.
- The method matches the variable types and study design.
- Data cleaning and missing-data decisions are documented.
- Relevant assumptions have been assessed appropriately.
- Results include magnitude and uncertainty where possible.
- Statistical significance is not confused with clinical importance.
- Claims remain within the limits of the design and sample.
- Tables and narrative reporting are consistent.
- The discussion translates findings into cautious nursing implications.
Conclusion
Statistical analysis matters in a nursing dissertation because it connects a research question with transparent evidence. Its purpose is not to decorate the results chapter with numbers or demonstrate software skills. Good analysis begins with clinically meaningful concepts, uses methods appropriate to the design and ends with a cautious interpretation of what the findings mean for nursing.
When those connections remain aligned, statistics strengthen the dissertation’s credibility and help the student move from data to evidence-informed conclusions without overstating what the study can prove.
References
- American Statistical Association (2016) ‘The ASA statement on p-values: Context, process, and purpose’, The American Statistician, 70(2), pp. 129–133. doi: 10.1080/00031305.2016.1154108.
- Grove, S.K. and Cipher, D.J. (2020) Statistics for Nursing Research: A Workbook for Evidence-Based Practice. 3rd edn. St Louis, MO: Elsevier.
- Nursing and Midwifery Council (2018) The Code: Professional Standards of Practice and Behaviour for Nurses, Midwives and Nursing Associates. London: NMC.
- Polit, D.F. and Beck, C.T. (2021) Nursing Research: Generating and Assessing Evidence for Nursing Practice. 11th edn. Philadelphia, PA: Wolters Kluwer.
