A quantitative research article can feel manageable until the Results section introduces statistical jargon. The challenge is deciding which numbers actually help answer the research question.
Scanning for p < .05 and moving on misses much of what clinicians need to know. Statistical interpretation also requires understanding what was measured, whether the analysis fits the study design and data, how large the observed effect was, how much uncertainty surrounds the estimate, and whether the finding is meaningful enough to influence practice (Grove & Cipher, 2025; Wasserstein & Lazar, 2016).
Start With the Research Question and Study Design
Before interpreting a statistical result, identify what the researchers were trying to determine. Look at who was studied, what intervention or exposure was examined, which outcomes were measured, and whether the study was primarily describing a population, comparing groups, examining relationships, or predicting an outcome.
The study design and structure of the data help determine which analyses are appropriate. Independent samples contain different participants in each group, while paired samples involve related observations, such as measurements taken from the same participants before and after an intervention. Level of measurement, distributional assumptions, sample size, and other features of the data can also influence the choice of statistical method (Grove & Cipher, 2025).
The table below shows common examples. These are starting points rather than fixed rules, and the assumptions of the statistical method still need to be appropriate.
| Research question | Common analysis when appropriate |
|---|---|
| Compare a continuous outcome between two independent groups | Independent-samples t test |
| Compare a continuous outcome in the same participants at two time points | Paired-samples t test |
| Compare means across three or more independent groups | One-way ANOVA |
| Compare categorical frequencies or proportions | Chi-square or Fisher’s exact test |
| Examine a linear relationship between two continuous variables | Pearson correlation |
| Examine an ordinal or ranked relationship | Spearman correlation |
| Predict a continuous outcome using one or more predictors | Linear regression |
| Analyze time until an event occurs | Survival-analysis methods |
Real studies may involve repeated observations, multiple predictors, missing data, confounding variables, clustering, or more complex designs. The practical question is whether the analysis is appropriate for the research question, study design, and structure of the data.
Read the Descriptive Statistics First
Descriptive statistics show who was studied and what the observed data looked like before the researchers make inferences beyond the sample. Start with the sample size and participant characteristics, including age, relevant clinical characteristics, baseline measurements, and differences between comparison groups.
Continuous variables are commonly summarized with means and standard deviations when those summaries appropriately represent the distribution, while medians and interquartile ranges can be more informative for skewed data. Categorical variables are typically summarized using counts and percentages (Grove & Cipher, 2025).
Suppose an intervention study reports that mean systolic blood pressure decreased from 148 mm Hg to 134 mm Hg. That descriptive result already tells the reader what occurred in the sample. Inferential statistics then help evaluate the uncertainty surrounding the observed difference and what conclusions the study can reasonably support.
Make Sure the Analysis Fits the Data
Statistical tests are not interchangeable. Researchers need to account for the study design, whether observations are independent or related, the type and distribution of the data, sample size, and the assumptions required by the selected analysis (Grove & Cipher, 2025).
An independent-samples t test, for example, is designed to compare two unrelated groups when its assumptions are reasonably satisfied. A paired-samples t test addresses related observations. Other data structures or violations of assumptions may require different approaches.
Readers do not need to reproduce every calculation to appraise a quantitative study. They do need enough statistical literacy to recognize when the analysis appears inconsistent with the design or the type of data collected.
Stop Treating p < .05 as the Answer
The p value is one of the most familiar statistics in healthcare research, but it is also frequently overinterpreted. The American Statistical Association cautions that a p value does not measure the probability that the null hypothesis is true, does not measure the probability that the results occurred “by chance,” and does not tell the reader how large or important an effect is (Wasserstein & Lazar, 2016).
A p value reflects how incompatible the observed data are with a specified statistical model under the assumptions used in the analysis. Many studies adopt an alpha level of .05 and describe results below that threshold as statistically significant, but the ASA specifically cautions against basing scientific conclusions solely on whether a p value crosses a particular threshold (Wasserstein & Lazar, 2016).
Results of p = .049 and p = .051 should not automatically be treated as fundamentally different evidence. A nonsignificant result also does not establish that no effect exists. Interpretation should consider the effect estimate, precision, study design, sample size, assumptions, and the broader evidence rather than relying on a binary significant versus nonsignificant label (Grove & Cipher, 2025; Wasserstein & Lazar, 2016).
Find the Effect Estimate and Confidence Interval
After identifying what was compared, look for the estimate that describes the direction and magnitude of the result. Depending on the study, this may be a mean difference, risk ratio, odds ratio, hazard ratio, correlation coefficient, regression coefficient, or another effect estimate.
Consider two hypothetical studies evaluating an intervention intended to reduce systolic blood pressure. One reports an average reduction of 1 mm Hg with p < .001. Another estimates an 8 mm Hg reduction with p = .06. Statistical significance alone would emphasize the first study, even though the second produced a much larger estimated effect. The uncertainty surrounding those estimates is essential to interpretation.
Consider the following result:
Mean systolic blood pressure difference: -8 mm Hg, 95% CI [-12, -4].
The point estimate indicates an average reduction of 8 mm Hg. The interval is relatively narrow and contains values that all indicate a reduction. Compare that with:
Mean difference: -8 mm Hg, 95% CI [-18, 2].
The point estimate is unchanged, but the second interval is substantially wider and includes values ranging from a larger reduction to a small increase. The estimate is therefore much less precise.
For difference measures such as a mean difference or risk difference, the null value is generally 0. For ratio measures such as relative risk, odds ratio, and hazard ratio, the null value is generally 1. Confidence intervals should be interpreted alongside the effect estimate and the clinical context rather than reduced to another binary test of whether the interval crosses the null value. Statistical reporting guidance recommends presenting effect estimates with measures of precision rather than relying exclusively on significance testing (Lang & Altman, 2015).
Statistical Significance and Clinical Importance Answer Different Questions
A statistically significant result may still be too small to matter in practice. Conversely, a potentially important effect may be estimated imprecisely in a study that lacks enough information to establish the effect confidently. Statistical significance alone therefore cannot determine whether a finding should influence clinical care (Wasserstein & Lazar, 2016).
Clinical interpretation requires attention to the magnitude of benefit, potential harms, patient burden, feasibility, and the outcome being measured. An intervention producing a very small change may have limited value even when the p value is small, while a modest effect may be more compelling when the outcome is important and the intervention is safe and feasible.
Statistical findings provide evidence for clinical judgment. They do not replace it.
Statistics Cannot Rescue a Weak Study
Statistical analysis should always be interpreted within the larger study design. Sophisticated modeling cannot convert an observational study into a randomized trial, and adjustment can account only for variables that were measured and appropriately included in the analysis. A large sample does not eliminate selection bias, and a very small p value cannot correct unreliable measurement or a poorly designed study (Grove & Cipher, 2025).
Missing data and loss to follow-up can also affect interpretation, particularly when the participants with missing information differ systematically from those who remain. Readers should consider how much information is missing, whether losses differ across groups, and how the researchers addressed missing observations (Grove & Cipher, 2025).
The central question remains whether the study design, measurement, data quality, and statistical analysis collectively support the conclusion being made.
Use a Consistent Reading Sequence
Quantitative Results sections become easier to interpret when they are approached in a consistent order. Begin with the research question and study design, then review the sample and descriptive statistics so the population and observed data are clear before focusing on inferential tests.
Next, determine whether the statistical analysis fits the design and data. Find the effect estimate and confidence interval to evaluate direction, magnitude, and precision, and then interpret the p value as one piece of the evidence rather than the final answer.
The last step is returning to the clinical question. Consider whether the estimated effect is meaningful, whether important bias or confounding could influence the result, whether the sample resembles the population in which the findings might be applied, and whether the conclusion is proportionate to the design and data.
DNP graduates may not routinely calculate every statistical test by hand, but they will continue deciding whether quantitative evidence is strong enough to influence patient care and practice change. The lasting skill is being able to interpret what the numbers support, how certain that conclusion is, and where the evidence still leaves room for doubt.
Related YourDNP resources:
- Research
- Critical Appraisal Worksheet
- Evidence Table & Synthesis Worksheet
- Research Pitfalls That Can Make Evidence Look Stronger Than It Is
References
Grove, S. K., & Cipher, D. J. (2025). Statistics for nursing research: A workbook for evidence-based practice (4th ed.). Elsevier.
Lang, T. A., & Altman, D. G. (2015). Basic statistical reporting for articles published in biomedical journals: The “Statistical Analyses and Methods in the Published Literature” or the SAMPL guidelines. International Journal of Nursing Studies, 52(1), 5–9. https://doi.org/10.1016/j.ijnurstu.2014.09.006
Wasserstein, R. L., & Lazar, N. A. (2016). The ASA’s statement on p-values: Context, process, and purpose. The American Statistician, 70(2), 129–133. https://doi.org/10.1080/00031305.2016.1154108