Nursing Capstone Data Analysis

By the time students reach the data analysis chapter, two very different situations tend to show up. Either the project ran on schedule and there is real before-and-after data to present, or — just as commonly — implementation slipped, the timeline ran short, and what exists is partial data, process measures, or staff feedback rather than a clean outcome comparison. Both situations are completely normal, and both can produce a strong chapter. This guide covers how to present data with simple statistics (percentages, pre/post comparisons, maybe a paired t-test or chi-square if your program expects it), how to present it without formal statistics (qualitative themes, process counts), how to build simple tables and charts that communicate clearly, and — most importantly — how to discuss results honestly when they are mixed, incomplete, or did not turn out as hoped, while still tying everything back to the PICOT outcome.

Start by re-reading your PICOT outcome and evaluation plan

Before writing a single sentence of this chapter, go back to your PICOT question and your methodology chapter's evaluation plan. The data analysis chapter's entire job is to answer the "O" — the outcome — that you committed to back in chapter one, using the measures you said you would collect back in chapter three. If your PICOT question asked whether hourly rounding would reduce call-light frequency, this chapter needs to report call-light frequency data, however that data turned out. If your evaluation plan said you would track flowsheet documentation completion as a process measure, this chapter needs to report that number too, even if it is the headline finding rather than a footnote.

This sounds obvious, but it is the single most common organizational problem in this chapter: students present data that is easy to get (a satisfaction survey that happened to be lying around) instead of data that answers the PICOT question (the specific outcome stated in chapter one). If the data you actually have does not perfectly match what you originally proposed to measure — which happens often — say so explicitly and explain the substitution, rather than silently presenting different data and hoping no one notices the mismatch.

Presenting results with simple statistics

Many capstone projects lend themselves to a straightforward pre/post comparison: a metric before the intervention, the same metric after, and the difference. Fall rate per 1,000 patient days before implementation versus during/after the implementation period. Hand-hygiene compliance percentage from a baseline audit versus a post-implementation audit. Average pain reassessment time before versus after a new documentation prompt. These comparisons can be reported as simple percentages and differences — "hand-hygiene compliance increased from 68% (baseline, n=50 observed opportunities) to 84% (post-implementation, n=52 observed opportunities), a 16 percentage-point increase" — and for many BSN/MSN capstones, this level of statistical sophistication is entirely sufficient and expected.

Some programs, particularly at the MSN or DNP level, expect a formal statistical test to determine whether an observed difference is statistically significant rather than just descriptively different. The two most common tests for capstone-level pre/post data are the paired t-test, used when comparing the means of a continuous variable measured on the same subjects before and after (for example, comparing each patient's pre- and post-intervention HbA1c readings in a diabetes self-management education project), and the chi-square test, used when comparing proportions or categorical outcomes between two groups or two time points (for example, comparing the proportion of patients who fell in the baseline period versus the implementation period). If your program expects this level of analysis, most schools provide access to SPSS or similar software, and many capstone committees are realistic about the fact that a single nursing student is not expected to be a biostatistician — focus on correctly setting up the comparison and correctly interpreting the p-value (typically, p < .05 is considered statistically significant) rather than on advanced statistical theory.

What if the sample is too small for statistical significance?

This is extremely common with short-timeline, single-unit capstones, and it is not a failure — it is an expected limitation that belongs in your discussion. A change from 2 falls to 0 falls over an 8-week period on one unit is a meaningful descriptive finding even if it is not statistically significant in a formal test (and with such small numbers, a chi-square test might not even be appropriate — note this rather than running a test that does not fit the data). Report the raw numbers and percentages honestly, and address statistical significance (or its absence) directly in your discussion rather than letting a reader wonder whether you understood the limitation.

Presenting results without formal statistics

Not every capstone project produces numeric outcome data, and that is fine — many produce qualitative or process-based findings that are just as valid for a practice-change project. If your project centered on staff education, a common and legitimate set of findings includes: the number and percentage of eligible staff who completed training, themes from staff feedback collected via a brief survey or informal interviews after implementation (for example, "staff reported the new rounding script felt natural after the first week, though several noted documentation took slightly longer initially"), and process counts like the number of audits completed, the number of educational sessions held, or the number of times a new tool was used.

For qualitative feedback, even a small amount of staff input can be presented as simple themes with illustrative quotes (de-identified): "Three recurring themes emerged from post-implementation staff feedback (n=8 respondents): increased confidence in pain reassessment documentation, a perceived modest increase in time per round, and a desire for the rounding log to be integrated directly into the EHR rather than a separate paper form." This is legitimate data, presented honestly, and it gives your recommendations chapter something concrete to build on (e.g., recommending EHR integration as a next step based directly on staff feedback).

Combining both approaches

Many of the strongest capstone results chapters use both: a simple quantitative process measure (90% of staff completed training; documentation compliance rose from 60% to 88%) alongside qualitative or anecdotal context (staff feedback themes) and, where available, an early look at the outcome measure itself, clearly labeled as preliminary. This combination tells a fuller story than numbers alone and gives you more to discuss in the next chapter.

Presenting incomplete or "anticipated" results honestly

Here is the scenario that probably describes more capstone papers than any other: the implementation timeline in the methodology chapter said weeks 5-7 for rollout and weeks 7-10 for data collection, but by the time the paper is due, the project is in week 6, training just finished, and there is no post-implementation outcome data yet — maybe not even a full cycle of process data. This is not a crisis, and it does not mean your results chapter is empty. It means your results chapter needs to be framed around what has happened so far and what is anticipated.

The key move is precise language. Do not write "the intervention reduced fall rates by 20%" if that has not actually been measured yet — that is a misrepresentation, however unintentional. Instead: "As of the writing of this paper, the intervention has been in place for three weeks. Process data indicate that 22 of 24 staff (92%) completed training, and documentation audits show an 85% compliance rate with the new rounding protocol in week three, up from a 0% baseline (since the protocol did not previously exist). Based on the evidence reviewed in Chapter 2, where similar interventions produced fall-rate reductions of 15-30% over 8-12 week periods, a meaningful reduction in this unit's fall rate is anticipated with continued implementation, though outcome data collection will continue beyond the timeframe of this paper." This framing is honest, uses the process data you do have, cites the literature's expected effect sizes to set realistic expectations, and does not claim an outcome that has not been observed. Faculty see this situation constantly and generally respond far better to honest framing of partial data than to results that quietly overstate what was actually measured.

Tables and simple charts that communicate clearly

A single well-labeled table comparing baseline and post-implementation values does more work than several paragraphs of prose, and most capstone results chapters benefit from at least one. The simplest effective format is a two- or three-column table: the metric name, the baseline value, and the post-implementation (or most recent available) value, sometimes with a third column for percentage change. If you have data across multiple time points (weekly audits, for example), a simple bar chart or line graph showing the trend over time can be more informative than a single before/after snapshot, especially if the trend shows steady improvement even if the final number has not yet reached a target.

Keep visuals simple. A bar chart comparing two or three values, or a line graph showing a trend across 4-8 weekly data points, communicates far better than a complex multi-panel figure. Label axes clearly, include units (percentage, rate per 1,000 patient days, count), and reference every figure and table in the surrounding text — a chart that is never discussed in the prose is a missed opportunity to interpret it for the reader. If you are using a spreadsheet tool to generate charts, simple bar and line charts in default styling are perfectly appropriate; capstone papers are not graded on data-visualization design sophistication.

Example results table: hand-hygiene compliance project

What belongs here vs. what belongs in discussion

A clean way to think about the boundary between this chapter and the next is: results presents WHAT happened, discussion explains WHY and SO WHAT. This chapter should describe the data — numbers, percentages, themes, comparisons — with minimal interpretation. Save the comparison to the literature reviewed in chapter two, the explanation of unexpected findings, the discussion of limitations, and the implications for practice for the discussion and recommendations chapter. Some programs combine results and discussion into a single chapter, in which case this distinction becomes a matter of paragraph order rather than separate chapters, but the logic still applies: present the data cleanly first, then interpret it.

Common Mistakes to Avoid

• Presenting data that does not match the outcome measure stated in the PICOT question, without explaining why a different measure is being reported.
• Claiming an outcome was achieved ("the intervention reduced falls by 20%") when the data only covers process measures or a partial implementation period.
• Running a formal statistical test (like chi-square) on a sample too small for the test to be meaningful, without acknowledging the limitation.
• Presenting only raw numbers without percentages or context, making it hard for a reader to judge the size or importance of a change.
• Including charts or tables that are never referenced or explained in the surrounding text.
• Mixing interpretation and discussion into the results chapter so heavily that there is nothing left to say in the discussion chapter.
• Treating incomplete implementation as something to hide or gloss over rather than framing honestly as preliminary or anticipated results.
• Failing to explicitly connect the results back to the PICOT outcome anywhere in the chapter, leaving the reader to make the connection themselves.

Related Guides