Interpreting Statistical Output for a Nursing Dissertation or DNP Project

Running a statistical test is only half the job; interpreting and reporting the output correctly is what a committee actually reads. This guide picks up after your analysis has been run in SPSS, R, or another package, and focuses on what the numbers mean, how to report them honestly, and the difference between statistical significance and practical importance.

P-ValuesEffect SizeConfidence IntervalsAPA ReportingResults Writing

Key Takeaways

Quick answer. A p-value tells you how compatible your data are with the null hypothesis, not whether a finding matters. Report effect size and confidence intervals alongside every p-value, describe results factually in the results chapter, and save interpretation of what a finding means for the discussion chapter.

  • This guide assumes a test has already been selected and run; see the biostatistics guide for that earlier step.
  • A p-value is not the probability that the null hypothesis is true, and it does not measure how large or important an effect is.
  • Statistical significance and practical or clinical significance are different questions and both deserve attention.
  • The specific numbers in your paper must come from your own actual analysis, never invented or estimated.

What This Guide Covers, and What It Does Not

This guide assumes you have already chosen and run an appropriate statistical test for your research question and data, whether in SPSS, R, or another package, and that you now have real output in front of you. It focuses on what that output means and how to write it up defensibly. If you have not yet reached that point, the foundational concepts of hypothesis testing and test selection are covered in our biostatistics for nursing students guide, which this guide builds directly on.

It also will not invent specific statistical values, such as a fabricated t-statistic or p-value, and present them as if they were example results, because doing so would misrepresent what real output looks like and could be mistaken for guidance to report numbers that were not actually produced by an analysis. Instead, this guide explains the reporting conventions in words, so you can apply them correctly to your own actual numbers.

What a P-Value Actually Tells You

A p-value is the probability of observing a result as extreme as, or more extreme than, the one obtained, if the null hypothesis were actually true. That is a precise, conditional statement, and it is worth holding onto exactly, because two very common misinterpretations cause real problems in a results or discussion chapter.

Common misinterpretationWhat is actually true
"The p-value is the probability the null hypothesis is true."A p-value is calculated assuming the null hypothesis is true; it does not tell you the probability that it is true.
"A smaller p-value means a bigger or more important effect."A p-value is influenced heavily by sample size and does not by itself measure the size or importance of an effect.
"A non-significant p-value proves there is no effect."A non-significant result may reflect an inconclusive test, often due to a small sample, rather than genuine proof of no effect.

Committees notice when a discussion chapter conflates statistical significance with practical importance, or treats a non-significant finding as a confident null result. Getting this distinction right, and stating it carefully, is one of the more reliable ways to demonstrate genuine statistical literacy rather than a mechanical run-and-report approach.

Why Effect Size and Confidence Intervals Matter Alongside Significance

A p-value alone tells you whether a result met a chosen significance threshold. It does not tell you how big, or how clinically meaningful, the effect actually is. A statistically significant result can come from a very small effect size, particularly in a large sample, and might have limited real-world or clinical relevance. Conversely, a non-significant result in a small sample might still show a meaningful effect size that is worth discussing cautiously as a signal for future, adequately powered research, rather than dismissed outright.

This is why reporting an effect size, and a confidence interval around your estimate, alongside every p-value gives a fuller and more honest picture than significance alone. A confidence interval also communicates the precision of your estimate: a wide interval signals more uncertainty than a narrow one, even when both surround the same point estimate. Reporting all three together, the test statistic, the effect size, and the confidence interval, is now standard expectation in nursing research reporting, not an optional extra.

How to Report Common Statistical Results in APA Style

APA 7 sets specific conventions for how statistical results are formatted in text, and following them consistently matters for both clarity and grading. Rather than inventing illustrative numbers here, this section describes the general shape of a complete report for a few common analyses; your actual output supplies the real values.

AnalysisWhat a complete report generally includes
Independent-samples t-testThe test statistic with degrees of freedom, the p-value, and a standardized effect size such as Cohen's d, often alongside group means
Chi-square test of independenceThe test statistic with degrees of freedom, sample size, the p-value, and an effect size such as Cramer's V
Correlation (e.g., Pearson's r)The correlation coefficient itself, the sample size, and the p-value; the coefficient functions as its own effect size
ANOVAThe F statistic with both degrees of freedom, the p-value, an effect size such as eta-squared, and follow-up post hoc results where relevant
RegressionThe relevant coefficient, its standard error or confidence interval, the p-value, and a model-level statistic such as R-squared

For the exact formatting details, italics, spacing, and rounding conventions, consult APA 7 directly and see our APA 7 guide and APA tables and figures guide, particularly if you are presenting results in a table rather than in narrative text.

How Sample Size and Power Affect Interpretation

Statistical power is the probability that a test will detect an effect if one genuinely exists in the population. Power is influenced heavily by sample size: a study with a small sample has lower power, meaning it is more likely to miss a real effect and produce a non-significant result even when a true effect is present. This matters directly for interpretation, because a non-significant finding in a small, likely underpowered sample should be discussed as inconclusive rather than as evidence that no effect exists.

If your proposal included an a priori power analysis, your discussion chapter should refer back to it honestly: state whether the achieved sample matched the planned sample, and if it fell short, acknowledge that shortfall as a limitation affecting the confidence you can place in a non-significant result specifically. This is a more defensible and more commonly expected approach than presenting a small-sample null finding without qualification.

What low power looks like in practice

SituationRisk it createsHow to discuss it
Small final sample relative to the planned power analysisReal effects may go undetected (a Type II error)Note the achieved versus planned sample size as a limitation
Non-significant result with a moderate effect sizeThe result may reflect low power rather than a true absence of effectReport the effect size and frame the finding as inconclusive, not as proof of no effect
Many subgroup comparisons on a modest sampleEach subgroup analysis has less power than the full-sample analysisInterpret subgroup findings cautiously and note the reduced power explicitly

Multiple Comparisons and Correction

When a study runs many statistical tests, such as comparing several outcome variables between groups or conducting numerous subgroup analyses, the probability of finding at least one statistically significant result purely by chance increases with every additional test. This is the multiple comparisons problem, and it is a real methodological concern that reviewers and committees are attentive to, particularly in studies with a large number of outcome measures.

Several accepted approaches exist to address this, including a Bonferroni correction, which adjusts the significance threshold based on the number of comparisons being made, and other less conservative correction methods used in some analyses. Whether a correction is required, and which method is appropriate, depends on your specific design and the number of comparisons involved, and this is a decision best confirmed with a statistician or your methodologist rather than assumed. What belongs in your write-up, regardless of the specific method, is transparency: state clearly how many comparisons were made and whether, and how, you accounted for that in your interpretation of significance.

How Missing Data Affects Interpretation

Very few real data sets are complete, and how missing data are handled changes what your statistical output actually represents. Methodologists commonly distinguish three mechanisms behind missing data: data missing completely at random, where missingness is unrelated to any variable in the study; data missing at random, where missingness relates to other observed variables but not to the missing value itself; and data missing not at random, where missingness relates to the value that is missing. This distinction is real and citable, and it matters because different mechanisms justify different handling approaches, and the wrong approach can bias your results in ways that a p-value alone will not reveal.

Common approaches to handling missing data include listwise deletion, which excludes any case with a missing value on a variable used in an analysis, and multiple imputation, which estimates plausible values based on patterns in the observed data and produces more statistically sound estimates under certain missingness assumptions. Which approach is defensible depends on the missingness mechanism, the amount of missing data, and your specific analysis, and this decision is worth confirming with a statistician rather than defaulting to whatever a software package does automatically.

What to report about missing data

ElementWhy it matters for interpretation
The amount of missing data per variableA small amount of missingness generally has less effect on interpretation than a large amount
The method used to handle itDifferent methods carry different assumptions about the missingness mechanism
Any meaningful difference between complete and incomplete casesA systematic difference suggests missingness may not be random, which affects how confidently a finding generalizes

Silently dropping incomplete cases without reporting how many were dropped, or why, is one of the more easily avoidable gaps a statistical reviewer will flag, since it leaves a reader unable to judge how much the final sample actually reflects the population you set out to study.

Writing the Results Narrative Around Your Output

A results section built around statistical output should do three things for each analysis: state what was found factually, connect that finding explicitly back to the specific research question or hypothesis it addresses, and stop there. Do not use the results section to speculate about mechanisms, compare your finding to other studies, or discuss implications; those belong in the discussion chapter. This mirrors the same report-versus-interpret principle described in our dissertation results chapter guide, and it applies whether your data are qualitative or quantitative.

A simple structure for reporting each analysis

  1. Restate briefly which hypothesis or question this analysis addresses.
  2. Report the test used and the complete statistical output: test statistic, degrees of freedom where applicable, p-value, effect size, and confidence interval.
  3. State in one sentence what the result shows, in plain but precise language, without interpreting its meaning or importance.
  4. Move to the next analysis or hypothesis in the same format.

Have your output and need help writing it up correctly?

Share your analysis output, hypotheses, and program handbook with your brief, and a specialist can help you interpret and write up your results chapter in proper APA format. The price is shown before you pay, and every delivered paper includes 14 days of free revisions.

Get my instant quote →

An Illustrative Story

Illustrative example, not a real client. This short story is invented to show the pattern, and it contains no real data or numbers.

The problem. A DNP student ran her post-intervention analysis and reported only a p-value below the significance threshold in her draft results chapter, describing the intervention as a clear success.

The tension. Her statistical reviewer asked what the effect size was, and she realized she had not calculated or reported one, and had not distinguished statistical significance from practical relevance anywhere in the chapter.

The turn. She went back to her actual output, pulled the effect size and confidence interval that had been sitting in the same table the whole time, and rewrote her results sentences to report all three together.

The proof. Her revised discussion chapter could now honestly address both how confident the finding was and how meaningful its size appeared to be in practice.

The payoff. Her committee's feedback shifted from "incomplete reporting" to specific, substantive questions about her intervention's implementation, which is the kind of feedback that moves a project toward defense.

Common Mistakes

Final Checklist

Frequently Asked Questions

What is the difference between statistical significance and clinical significance?

Statistical significance reflects whether a result met a chosen probability threshold, often influenced by sample size. Clinical or practical significance asks whether the size of the effect matters in real-world terms, which is why effect size should always be reported and considered alongside the p-value.

My result was not statistically significant. Should I still report the effect size?

Yes. Reporting the effect size and confidence interval for a non-significant result is honest practice, and it allows a reader to judge whether the finding might still be worth further, adequately powered investigation.

Which software should I use to run my analysis?

That depends on your program's requirements and your own comfort; SPSS and R are both common. This guide focuses on interpreting whichever program's output you have, not on operating a specific package.

Can you tell me what p-value or effect size to report in my paper?

No. The specific numbers in your paper must come from your own actual analysis of your own real data; they cannot be estimated, assumed, or supplied by anyone else.

How do I know which effect size measure to use?

It depends on the test: Cohen's d is common for comparing two means, Cramer's V or phi for chi-square tests, eta-squared for ANOVA, and the correlation coefficient itself for correlational analyses. Your statistics text or software's documentation for the specific test will confirm the appropriate measure.

Do qualitative results need this same kind of statistical reporting?

No. Qualitative findings are reported and interpreted differently; see our qualitative data analysis guide for that process.

Report Completely, Interpret Carefully

Good statistical reporting is complete, not just significant: a test statistic, a p-value, an effect size, and a confidence interval, tied back plainly to the question it answers. Keep interpretation for the discussion chapter, and let your own real output, never an invented number, be the only source of the figures in your paper.

Have output from SPSS or R that needs an accurate, properly formatted writeup? Get my instant quote. The price is shown before you pay, every delivered paper includes 14 days of free revisions, and refund terms are on the money-back guarantee page. Please use any model paper in line with your institution's academic-integrity policy.