SPSS Nursing Capstone: Complete Nursing Guide

For most nursing capstone students, SPSS shows up at exactly one point in the project: after data collection is complete, when pre-intervention and post-intervention numbers need to be compared and turned into a results section. The anxiety many students feel about this step is usually less about SPSS itself — the software is largely menu-driven — and more about not knowing which statistical test fits their data, and not knowing how to translate SPSS output (which is dense and full of unfamiliar abbreviations) into the kind of plain-language statement a results section needs. This guide focuses on exactly that: the small number of tests that cover the vast majority of nursing capstone analyses, how to choose between them based on your data, and how to read and report the output without getting lost in statistical jargon you don't need for a QI capstone.

What Most Nursing Capstone Data Actually Looks Like

Despite the breadth of statistical tests SPSS can run, the vast majority of nursing capstone analyses fall into a small number of patterns, because the vast majority of nursing capstone designs follow a small number of patterns: a single group, measured before and after an intervention (pre/post, same individuals); two groups compared at one point in time (e.g., a unit that received the intervention vs. a unit that didn't); or a relationship between two variables (e.g., does length of time since training correlate with compliance rate).

The most common of these by far is the pre/post, same-individuals design — you measure something (a knowledge score, a symptom severity score, a compliance rate) before your intervention, implement the intervention, then measure the same thing again afterward in the same individuals or the same unit's aggregate data. This design corresponds to a small number of statistical tests depending on what kind of data you collected: a paired-samples t-test if your outcome is a continuous score (like a 0-10 pain scale average, or a percentage), or a Wilcoxon signed-rank test if your outcome is ordinal and your sample is small (common in pilot-scale capstones), or a McNemar's test if your outcome is categorical/binary (e.g., percentage of patients who received an intervention as planned, yes/no).

If your design instead compares two different groups (rather than the same group before and after), the equivalent tests are an independent-samples t-test (continuous data, two groups), a Mann-Whitney U test (ordinal data or small samples, two groups), or a Chi-square test (categorical data, two groups). Knowing which of these patterns matches your design is the single highest-value step in the entire SPSS process — it determines everything else.

Choosing a Statistical Test for Common Capstone Designs

Entering Data and Setting Up SPSS for a Capstone Dataset

SPSS organizes data in two views: Variable View, where you define each variable (its name, type, and for categorical variables, value labels — e.g., 1 = "yes," 2 = "no"), and Data View, where each row is one case (one patient, one survey response, one chart audit) and each column is one variable. For a pre/post design measuring the same individuals, each person is one row, with separate columns for their pre-score and their post-score — not separate rows for "before" and "after," which is a common setup mistake that makes a paired test impossible to run correctly.

Defining value labels for categorical variables (in Variable View) makes your output much easier to read later — instead of seeing "1" and "2" in your results, you'll see "Received intervention" and "Did not receive intervention" (or whatever labels you've defined), which matters when you're translating output into a results section months after you entered the data and may not remember which number meant what.

For small capstone datasets — and most are small, often 10-40 cases — double-checking data entry by reviewing the Data View for obvious errors (a "9" in a column that should only contain 1s and 2s, a pre-score higher than any post-score when you expected the opposite, a missing value coded as 0 when it should be blank or a specific missing-value code) before running any analysis catches the errors that are easiest to introduce and hardest to notice once you're looking at statistical output instead of raw numbers.

Running and Reading a Paired-Samples T-Test (Most Common Capstone Analysis)

• In SPSS: Analyze > Compare Means > Paired-Samples T Test, then select your pre-score variable and your post-score variable as the pair
• In the output, find the "Paired Samples Statistics" table first — this shows the mean (average) for pre and post, which is often the single most important number for your results section ("the mean score increased from X to Y")
• Next, find the "Paired Samples Test" table — the column labeled "Sig. (2-tailed)" is your p-value. A p-value less than .05 is conventionally described as "statistically significant"
• If your p-value is below .05: you can report that the change was statistically significant, in addition to reporting the actual mean change
• If your p-value is above .05 (common in small pilot samples): report the mean change descriptively ("scores improved from X to Y, though this difference did not reach statistical significance, p = ...") — this is a legitimate and honest way to report a result, especially in a pilot framing
• Report both the means and the p-value together — neither alone tells the full story. A large mean change with a non-significant p-value (common with small samples) and a tiny mean change with a significant p-value (possible with large samples) mean very different things clinically
• Translate the result into plain language for your results section: "The mean [outcome] score increased from [X] at baseline to [Y] post-intervention (p = [value])" — save the clinical interpretation of what this means for the discussion section, not the results section

Reporting Results in Plain Language Without Losing Accuracy

One of the more common gaps between what SPSS produces and what a capstone results section needs is translation — SPSS output is dense, uses abbreviations (M for mean, SD for standard deviation, df for degrees of freedom, Sig. for significance/p-value), and presents far more numbers than a results section typically needs to report. Knowing which numbers from the output actually belong in your write-up — and which can stay in an appendix or simply inform your understanding without being quoted directly — keeps your results section readable.

For most nursing capstone results sections, the numbers worth reporting in the main text are: the sample size (n), the mean and standard deviation for pre- and post-scores (or percentages for categorical outcomes), and the p-value from your test. A sentence like "Mean patient-reported confidence scores increased from 2.4 (SD = 0.8) at baseline to 3.6 (SD = 0.6) post-intervention (n = 14, p = .003)" contains everything a reader needs to evaluate your finding, without requiring them to interpret an SPSS table.

If your p-value doesn't reach statistical significance — common with small samples — resist the temptation to either hide this result or overstate its meaning. "Did not reach statistical significance" is a factual, neutral statement; pair it with the actual mean change (which may still be clinically meaningful even without statistical significance) and let your discussion section address what this means for your project's conclusions, especially if you've framed the project as a pilot. If translating your SPSS output into a results section feels like the hardest part of finishing your capstone, get help with this paper from a writer who can help present your statistical results clearly and accurately.

Common Mistakes to Avoid

• Setting up pre/post data as separate rows instead of separate columns for the same case. A paired-samples t-test requires each individual's pre- and post-scores in the same row, in different columns. Separate rows for "before" and "after" make a paired test impossible to run correctly.
• Choosing a test based on what's familiar rather than what fits the data. An independent-samples t-test answers a different question than a paired-samples t-test. Match the test to your actual design (same group before/after vs. two different groups) — not to whichever test you've heard of.
• Treating a non-significant p-value as "the project failed." A p-value above .05, especially with a small sample, is common and doesn't mean nothing happened — report the actual mean change alongside the p-value and let the discussion address clinical significance.
• Reporting only the p-value without the means. A p-value alone doesn't tell a reader the size or direction of the change. Always report means (or percentages) alongside the p-value.
• Not defining value labels for categorical variables. Output showing "1" and "2" instead of meaningful labels is harder to interpret correctly when you're writing up results, especially after time has passed since data entry.
• Skipping a data entry double-check before running analysis. Small datasets are easy to introduce errors into — an out-of-range value, a miscoded missing value — and these errors are much easier to catch by reviewing the raw data than by puzzling over odd-looking statistical output.
• Using a t-test on ordinal data from a very small sample. If your outcome is an ordinal scale (e.g., a 5-point Likert-type item) and your sample is small, a non-parametric test (Wilcoxon signed-rank for paired data, Mann-Whitney U for two groups) is generally more appropriate than a t-test.
• Including raw SPSS output screenshots in the results section without translation. Most programs expect results reported in narrative/table form using plain-language statistics (means, percentages, p-values in sentence form), not pasted SPSS tables — check your program's expectations, but default to narrative reporting.

Related Guides