Most nursing capstones use SPSS for one job: comparing outcome data before and after an intervention and turning it into a results section. The hard part is rarely the software; it is choosing the right test and reporting it accurately.
This guide covers the designs most capstones use, a test-selection table, data setup, running and reading a paired-samples t-test, non-parametric options and how to report results in plain language.
The Three Questions That Pick Your Test
- Same people or different groups? Pre/post in the same people needs a paired test; two separate groups need an independent test.
- What type of data? Continuous scores, ordinal ratings or categories such as yes/no.
- How big is the sample? Small samples and skewed data often call for a non-parametric test.
Answer those three and the table below gives you the test.
What Most Capstone Data Looks Like
SPSS can run hundreds of tests, but most capstones fit a few designs. Recognizing yours is the biggest step.
- One group, before and after: the same people or unit measured pre- and post-intervention. This is the most common design.
- Two groups at one point: for example, a unit that received the intervention compared with one that did not.
- A relationship: whether two measures move together, such as training hours and compliance.
- Description only: means, frequencies and percentages, often enough for a small pilot.
If you measured aggregate unit data, such as a monthly compliance rate, rather than individual people, you may have too few data points for a formal test. Descriptive statistics and a run chart are often the honest choice.
Choosing a Test for Common Capstone Designs
| Design | Data type | Test |
|---|---|---|
| Same group, pre/post | Continuous, roughly normal differences | Paired-samples t-test |
| Same group, pre/post | Ordinal, skewed or small sample | Wilcoxon signed-rank test |
| Same group, pre/post | Binary, such as yes/no | McNemar's test |
| Two groups, one time point | Continuous | Independent-samples t-test |
| Two groups, one time point | Ordinal, skewed or small sample | Mann-Whitney U test |
| Two groups, one time point | Categorical | Chi-square test (Fisher's exact test if expected counts are small) |
| Relationship between two measures | Continuous | Pearson correlation (Spearman for ordinal data) |
| No comparison | Any | Descriptive statistics |
Check your program's guidance too. Some faculty prefer descriptive results for pilots, while others expect an inferential test.
Setting up Your Data in SPSS
SPSS has two views. In Variable View you define each variable: its name, type, measure level and, for categories, value labels such as 1 = "Yes" and 2 = "No". In Data View, each row is one case and each column is one variable.
For a paired pre/post design, put each person's pre-score and post-score in the same row, in two columns. Separate rows for "before" and "after" make a paired test impossible.
Example layout for a pre/post knowledge test.
ID Unit Pre_score Post_score Role 01 A 12 17 1 02 A 9 15 2 03 B 14 16 1
Role uses value labels: 1 = "Registered nurse", 2 = "Healthcare assistant".
Use short variable names without spaces, and use the Label column for the full description that appears in output.
Checking Your Data Before Analysis
Capstone datasets are often small, sometimes 10 to 40 cases, so one error can change a result. Check the data before you run anything.
- Run Frequencies on every variable to spot impossible values, such as a 9 where only 1 or 2 is allowed.
- Code missing values consistently, and never as 0 if 0 is a real score.
- Compare a few rows against your original forms.
- Check that every paired case has both a pre- and a post-score.
Running and Reading a Paired-Samples T-Test
- Choose Analyze, then Compare Means (called Compare Means and Proportions in some recent versions), then Paired-Samples T Test.
- Select the pre-score and post-score variables as one pair and run the test.
- In the Paired Samples Statistics table, note the mean, standard deviation and n for pre and post.
- In the Paired Samples Test table, find the significance value. The two-sided p-value is the one usually reported.
- Report the means, the change and the p-value together.
A p-value below .05 is conventionally called statistically significant. A p-value above .05 does not mean nothing happened; with a small sample, a real change may not reach significance.
When to Use Non-Parametric Tests
Paired t-tests assume the differences between pre and post scores are roughly normally distributed. With small samples, single Likert-type items or skewed data, that assumption is often doubtful.
| Situation | Use instead | Report |
|---|---|---|
| Paired ordinal or skewed data | Wilcoxon signed-rank | Medians, Z and p |
| Two independent groups, ordinal or skewed | Mann-Whitney U | Medians, U and p |
| Paired yes/no outcomes | McNemar's test | Percentages and p |
You can check normality with a histogram of the differences and the Shapiro-Wilk test under Explore, but with very small samples, visual checks and your program's guidance matter more than one test.
Reporting Results in Plain Language
SPSS output contains far more numbers than a results section needs. Report the sample size, the means and standard deviations (or medians, or percentages) and the p-value.
Example results sentence (hypothetical figures). Twenty-two nurses completed both tests. Mean knowledge scores rose from 11.4 (SD = 2.6) before the education session to 15.9 (SD = 2.1) afterwards, a statistically significant increase, t(21) = 7.12, p < .001.
Non-significant example. Mean confidence scores rose from 3.1 (SD = 0.9) to 3.5 (SD = 0.8), but the difference did not reach statistical significance, t(14) = 1.62, p = .13.
Keep interpretation for the discussion. Results state what happened; the discussion explains what it means clinically and why.
Statistical and Clinical Significance
A small change can be statistically significant in a large sample, and a meaningful change can miss significance in a small one. Your discussion should address both.
- Report the size of the change, not only whether p is below .05.
- Explain whether the change matters for patients or practice.
- For pilots, frame results as preliminary and say what a larger sample could test.
Presenting Results in an APA Table
When you compare several outcomes, a table is clearer than a paragraph of numbers. Number it, give it a short italic title in your word processor and explain it in the text.
Example table layout (hypothetical figures).
Table 1 Pre- and Post-Intervention Scores (n = 22) Measure Pre M (SD) Post M (SD) p Knowledge 11.4 (2.6) 15.9 (2.1) < .001 Confidence 3.1 (0.9) 3.6 (0.7) .02
Do not repeat every number from the table in your text. Highlight the main findings and refer the reader to the table for the rest.
Common Mistakes to Avoid
| Mistake | Fix |
|---|---|
| Pre and post scores on separate rows | One row per case, two columns |
| A familiar test that does not fit the design | Match the test to paired or independent data |
| Calling a non-significant result a failure | Report the change and discuss clinical meaning |
| Reporting p without means | Always give means, medians or percentages too |
| No value labels | Label categories in Variable View |
| Skipping data checks | Run frequencies before analysis |
| A t-test on a small set of ordinal ratings | Use a non-parametric test |
| Pasting raw SPSS tables | Report in sentences or a formatted APA table |
How EssayDonkey Helps with SPSS and Capstone Statistics
Send your PICOT, a description of your design, your de-identified dataset or SPSS output, your program template and rubric. Say which test your faculty expects, if any.
You get a custom results section, discussion or full capstone paper written from scratch by a writer with nursing research experience, with statistics reported accurately in APA 7 style. We never invent or alter data; everything is based on what you provide. Free revisions within the scope of your original request are included.
Need Help Turning SPSS Output into a Results Section?
Send your data or output, PICOT and rubric. Your writer builds a custom results and discussion section with accurate, plain-language reporting.
Order Your Results SectionFree revisions · Full refund if late · Written from scratch for your order
Frequently Asked Questions
Usually not. Many universities provide SPSS through campus computers or a student license. Check with your library or IT service.
A paired-samples t-test for continuous scores, a Wilcoxon signed-rank test for ordinal or skewed data, and McNemar's test for yes/no outcomes.
Report the actual change and the p-value honestly. Small samples often miss significance, and the discussion can address clinical meaning.
One row per person, with the pre-score and post-score in separate columns.
Excel can produce descriptive statistics and some tests, but many programs expect SPSS or similar. Check your program's requirements.
Usually not in the main text. Report results in sentences or an APA-style table, and add output to an appendix only if your program asks.
There is no fixed minimum, but very small samples have little power to detect change. Many capstones with fewer than about 20 paired cases report descriptive results or use a non-parametric test, as their faculty advise.
A run chart plots a measure over time, such as weekly compliance. It suits QI projects with aggregate data where formal tests are not appropriate.