The statistical analysis plan in a DNP capstone manuscript is not optional decoration in Chapter 3, it is the commitment the student makes to the committee about how the project's success will be measured. A statistical analysis plan that names "appropriate statistical methods will be used" signals that the student does not have a specified plan, which will generate a revision request before the methodology chapter is approved. Choosing the correct statistical test requires knowing the data type (continuous vs categorical vs count), the design structure (one group pre-post vs two groups concurrent), and the distribution assumption (parametric vs non-parametric). This page covers the tests used most commonly in DNP capstone projects and how to select and report them correctly.
Statistical Test Selection Guide for DNP Capstone Projects
Paired t-test: Used when comparing means of a continuous variable before and after an intervention in the same group (one-group pre-post design). Requires: one group, two time points, continuous outcome, approximately normally distributed data. Examples: pre-post comparison of PHQ-9 scores, pre-post comparison of pain scores, pre-post comparison of mean hand hygiene compliance percentage. Report: mean ± SD at pre and post, t-statistic, degrees of freedom (n−1), p-value, 95% confidence interval for the mean difference, Cohen's d effect size. Example reporting: "PHQ-9 scores decreased from a pre-implementation mean of 12.4 (SD = 3.1) to a post-implementation mean of 8.7 (SD = 2.8), a statistically significant improvement (t(42) = 5.23, p < 0.001, 95% CI [2.13, 5.27], d = 0.79)."
Wilcoxon Signed-Rank Test: The non-parametric equivalent of the paired t-test. Used when the continuous outcome data are not normally distributed (e.g., small sample, skewed data, ordinal scale outcomes). Examples: pre-post comparison of Likert-scale knowledge scores, pre-post comparison of pain scores with a ceiling effect. Report: median and IQR at pre and post, Z-statistic, p-value, effect size r (= Z / √N). Example reporting: "Staff knowledge scores improved from a pre-implementation median of 14 (IQR: 11–17) to a post-implementation median of 22 (IQR: 19–24), a statistically significant improvement (Z = −4.12, p < 0.001, r = 0.63)."
McNemar Test: Used when comparing proportions of a categorical (binary) outcome before and after an intervention in the same group. Examples: pre-post comparison of CAUTI occurrence (yes/no), pre-post comparison of screening completion (completed/not completed), pre-post comparison of protocol compliance (compliant/non-compliant) where individual-level data are available. Report: proportion compliant at pre and post, McNemar chi-square statistic, degrees of freedom (1), p-value. Note: the McNemar test requires paired categorical data, the same individuals observed before and after. If aggregate proportions are compared (e.g., CAUTI rate per 1,000 catheter days before vs after) without individual-level paired data, the McNemar test is not appropriate, use a rate comparison with chi-square or a proportion z-test instead.
Chi-Square Test (Pearson χ²): Used when comparing proportions between two independent groups, or when comparing a pre-implementation and post-implementation proportion where aggregate (not paired individual-level) data are available. Examples: comparison of complication rates in control vs intervention units, comparison of screening completion rates before and after implementation using aggregate monthly percentages. Report: χ² statistic, degrees of freedom, p-value, effect size Cramér's V. Chi-square requires a minimum expected frequency of 5 in each cell, if cell counts are small (n < 5 in any cell), use Fisher's Exact Test instead.
Independent Samples t-test: Used when comparing means between two independent groups (intervention vs control, or two different time periods in two different units). Requires: two independent groups, continuous outcome, approximately normal distribution. Examples: comparing post-intervention pain scores between patients who received the intervention vs those who did not (concurrent control group). Report: mean ± SD for each group, t-statistic, degrees of freedom, p-value, 95% CI for the difference, Cohen's d.
Mann-Whitney U Test: The non-parametric equivalent of the independent samples t-test. Used when continuous data from two independent groups are not normally distributed. Report: median and IQR for each group, U statistic, p-value, effect size r.
Statistical Process Control (SPC) Charts and Run Charts
SPC charts and run charts are used to monitor quality improvement outcomes over time, they are not inferential tests but graphical tools for distinguishing common cause variation (expected random fluctuation in the process) from special cause variation (signals that the process has genuinely changed). In DNP QI projects, SPC charts are used alongside or instead of pre-post t-tests when weekly or monthly data are collected over the implementation period.
Run Chart: A plot of the outcome measure (CAUTI rate, compliance percentage, PHQ-9 score) over time, with the median as the centreline. Run chart analysis applies four rules for detecting non-random patterns: (1) a shift (6 or more consecutive data points above or below the median; (2) a trend) 5 or more consecutive data points all going up or all going down; (3) too many or too few runs (assessed against a probability table; (4) an astronomical data point) a value clearly outside the expected range. Run charts are appropriate when the implementation period is short (8 to 16 weeks) or the data are counts rather than rates.
SPC Chart Types: P-chart, for proportion data (percentage of patients screened, percentage of catheters with daily necessity assessment documented). U-chart (for count data per unit of opportunity (CAUTI events per 1,000 catheter days). I-MR chart (Individuals and Moving Range)) for continuous data measured one at a time (weekly mean PHQ-9 score, monthly mean length of stay). The correct chart type depends on the data type: using a P-chart for count data or an I-MR chart for proportion data is a statistical error that committees will catch. SPC charts include upper and lower control limits set at ±3 standard deviations from the process mean, data points beyond these limits signal special cause variation.
Effect Size and Clinical Significance: Why p-value Alone Is Not Enough
A statistically significant p-value (p < 0.05) indicates that the observed difference is unlikely to be due to chance, it does not indicate that the difference is clinically meaningful. DNP capstone manuscripts are required to report both statistical significance (p-value) and clinical significance (effect size and MCID comparison). Effect size quantifies the magnitude of the change, independent of sample size.
Cohen's d (for t-tests): Small = 0.2, Medium = 0.5, Large = 0.8. A paired t-test result of p = 0.03 with d = 0.18 is statistically significant but practically negligible. A result of p = 0.08 with d = 0.72 is clinically meaningful but did not reach statistical significance, possibly due to a small sample.
MCID (Minimally Clinically Important Difference): The smallest change in an outcome that patients or clinicians consider meaningful. For the PHQ-9, the MCID is a 5-point reduction. For the GAD-7, it is a 4-point reduction. For CAUTI rate, the MCID is typically comparison to the NHSN benchmark. Report the actual change observed and state whether it meets the MCID threshold: "The mean PHQ-9 score decreased by 3.7 points, which did not meet the MCID threshold of 5 points, suggesting that while the change was statistically significant, it may not represent a clinically meaningful improvement for the majority of patients."
What data type is your primary outcome (continuous, categorical, or count per 1,000 units) and what is your study design?
Statistical analysis plan support identifies the correct test for your data type and design, writes the analysis plan for Chapter 3, and supports results reporting for Chapter 4 including table formatting, output interpretation, and effect size calculation. Share your PICOT outcome measure, the data source, the design (one-group pre-post, two-group, time-series), and the sample size, and the correct test, reporting format, and interpretation will be specified.
Descriptive Statistics in DNP Capstone Results
Every DNP capstone Results chapter begins with descriptive statistics before any inferential test is reported. Descriptive statistics characterise the sample and the outcome data: for continuous variables, report mean and standard deviation (or median and IQR for non-normal data); for categorical variables, report frequencies and percentages; for time-series data, report the baseline process average (centreline) and control limits. The descriptive statistics table (Table 1 in most DNP manuscripts) shows participant or unit characteristics (age, sex, years of experience, unit census, catheter days) that help the reader understand the sample to which the findings apply. Inferential test results (Tables 2 and onward) report the pre-post comparisons, between-group comparisons, or SPC chart signals as appropriate to the project design.
See also: DNP data analysis help · DNP capstone manuscript help · DNP IRB proposal help
Statistical Methods for DNP: Frequently Asked Questions
Do I need a statistician for my DNP capstone data analysis?
Most DNP capstone projects use straightforward pre-post designs with standard inferential tests (paired t-test, Wilcoxon, McNemar) that a methodologically prepared DNP student can conduct independently using SPSS, Excel, or R. A statistician is valuable (and sometimes required by the committee) when the design is more complex: when covariates need to be controlled (requiring ANCOVA or logistic regression), when a time-series analysis with multiple data points requires ARIMA modelling, or when the sample is small and power analysis is needed to interpret non-significant results. If your committee chair has not specified a statistician, begin with the standard test for your design and data type. If the committee requests more complex analysis at the proposal defence, consult a statistician before committing to a method you cannot execute correctly.
What statistical software is acceptable for DNP capstone data analysis?
IBM SPSS Statistics is the most commonly used and most supported statistical software in DNP programs. Microsoft Excel with the Data Analysis ToolPak is acceptable for basic descriptive statistics and simple inferential tests (t-tests, chi-square) and is available without additional cost. R is free, powerful, and increasingly used in DNP programs with quantitatively oriented faculty, but requires more technical preparation than SPSS or Excel. JMP, SAS, and Minitab are accepted where available. The software choice is rarely mandated by the programme, what matters is that the output is correctly interpreted and reported in APA 7th edition table format, not as a screenshot of raw software output.
What if my results are not statistically significant?
Non-significant results do not disqualify a DNP capstone project. The committee evaluates whether the project was designed and implemented correctly, not whether the intervention produced a p < 0.05 result. A non-significant result that is honestly reported, correctly interpreted (was the sample too small to detect a real effect? was the implementation period too short?), and connected to a meaningful discussion of clinical significance and limitations is a stronger manuscript than a significant result that is overclaimed or whose limitations are minimised. Report the non-significant result with the test statistic, confidence interval, and effect size, then discuss whether the sample size was sufficient to detect the MCID, and what the confidence interval range implies about the plausible effect.
Need expert help with your DNP capstone project?
Get Help NowCommon Questions
What is a DNP capstone project and how is it different from a PhD dissertation?
A DNP capstone project is a practice-focused doctoral scholarly project that applies evidence-based practice, quality improvement, or program evaluation methods to address a clinical problem. Unlike a PhD dissertation, which generates new knowledge through primary research, a DNP capstone translates existing evidence into practice change. It does not require original data collection in most cases and is evaluated on practice impact rather than research contribution.
Which DNP specialisation tracks do you support?
We support all 13 major DNP specialisation tracks: Family Nurse Practitioner (FNP), Adult-Gerontology Acute Care NP (AGACNP), Adult-Gerontology Primary Care NP (AGPCNP), Psychiatric-Mental Health NP (PMHNP), Pediatric NP (PNP), Neonatal NP (NNP), Women's Health NP (WHNP), Certified Nurse Midwife (CNM), Certified Registered Nurse Anesthetist (CRNA), Clinical Nurse Leader (CNL), Nurse Executive/Healthcare Leadership, Population Health, and Nursing Informatics.
Can you help with just one chapter of my DNP proposal or do I need the full project?
You can order help with any individual component: a single proposal chapter, just the PICOT question, just the IRB protocol, or just the data analysis section. You do not need to order the full project. Many students come to us mid-project needing targeted help with one specific deliverable.
Does my DNP capstone project need IRB approval?
Most DNP capstone projects are classified as quality improvement (QI) or program evaluation and do NOT require full IRB review under 45 CFR 46; they qualify for a QI determination or exempt status. However, the determination must be documented. We help you complete the QI determination checklist and, where needed, write the full IRB protocol for exempt or expedited review.