Measures and analysis planning
Capstone and DNP Data Analysis Planning Guide
Analysis begins when the measure is defined, not when the spreadsheet is full. The project question, design, data type, sampling process and decision determine which summaries, displays and tests are defensible.
Quick answer
Build the analysis plan before opening the final dataset
Write the decision question, unit of analysis, comparison, period and measure definitions. For each variable, record name, meaning, type, allowed values, source, timing, missing code and transformation. This prevents a label such as compliance from hiding multiple definitions.
Start with data quality and descriptive analysis. Check duplicates, impossible values, denominators, missingness, time order and subgroup sizes. A sophisticated test cannot repair an unreliable measure or a comparison that the design never created.
For improvement projects, time-order displays can reveal shifts, trends and instability that a single before-and-after average hides. If inferential statistics are proposed, check assumptions, independence, sample size and whether the design supports the intended interpretation.
A decision-first method
Align question, measure, data and analysis
Document every choice so another reviewer can reproduce the result from the authorized data.
Restate the approved question
Identify population, activity, comparison, outcome and period.
Create a measure dictionary
Define numerator, denominator, exclusions, direction and source.
Map the data structure
Identify unit, repeated observations, groups, time and variable types.
Run a quality plan
Predefine checks for missing, duplicate, impossible and inconsistent records.
Select descriptive outputs
Choose counts, proportions, center, spread and time displays that answer the question.
Justify any inferential method
Match assumptions and design before calculating a p value or interval.
Predefine interpretation
State which claims the design can and cannot support.
Working decisions
Analysis selection starts with the decision
| Question | Possible descriptive output | Boundary |
|---|---|---|
| How often did the process occur? | Count and proportion with defined denominator | Depends on complete eligible-case identification |
| How did performance change over time? | Run chart or time-ordered proportions | Concurrent changes may affect the pattern |
| How do two independent groups differ? | Group summaries and justified comparison | Assignment and confounding control the claim |
| Are two measures associated? | Scatterplot, cross-tabulation or correlation as appropriate | Association does not establish causation |
| What themes appear in text? | Transparent coding and theme evidence | Themes depend on sampling and analytic process |
Use the current program, course or project requirements as the controlling source. The table is a planning aid, not an institutional ruling.
Original worked example
Defining a process measure before analysis
Measure: eligible discharge records with documented follow-up within the approved interval divided by all eligible discharge records reviewed each week. The dictionary defines eligibility, the interval, duplicate handling and missing status. A weekly run chart and period summaries describe performance. The report does not call the change causal because the project lacks a controlled comparison.
A complete operational definition makes the numerator, denominator, timing and interpretation auditable.
Common problems and repairs
Capstone data-analysis errors and repairs
Choosing a test from a list
Repair: start with the question, design, variable type and data structure.
Changing the denominator after seeing results
Repair: predefine eligibility and document authorized deviations.
Treating missing values as zero
Repair: code missingness separately and assess its pattern.
Claiming causation from before-and-after data
Repair: describe the observed change and relevant alternative explanations.
Sending protected row-level data for help
Repair: use approved systems, de-identification and simulated examples.
Final control
Capstone analysis audit
- Approved question and design fixed.
- Unit of analysis identified.
- Measure dictionary complete.
- Eligibility and missing-data rules documented.
- Quality checks completed.
- Outputs match variable types and time structure.
- Assumptions checked for inferential methods.
- Claims stay within the design.
- Files are stored and shared under policy.
Keep the record authentic
Never fabricate, delete inconvenient observations, change definitions to improve results or share protected data outside approved systems. Report the authentic analysis and its limitations.
Sources and version check
Sources used for this guide
- AHRQ Plan-Do-Study-Act Worksheet, for test planning, measures and learning cycles
- AHRQ Quality Improvement Models, for disciplined change and evaluation
- AHRQ Collect and Use Data for Quality Improvement, for measure selection and data-use planning
- NIST Engineering Statistics Handbook, for statistical method and diagnostic references
- CDC Epi Info, for public-health data and analysis context
Verify the active requirements
Requirements vary by institution, instructor, site and project version. Use the current instructions, rubric, handbook and approvals as the controlling sources.
Frequently asked questions
Capstone and DNP Data Analysis Planning Guide questions
What should I verify first for capstone project data analysis?
Verify the current institution, program, course or project requirements and identify the exact decision you need to make. Public examples do not override your active requirements.
Can Bright Writers help me plan capstone project data analysis?
Yes. Help can include requirement breakdown, planning, concept explanation, evidence organization, rubric review, feedback on your own draft and a prioritized revision plan.
Can I use the example on this page in my submission?
Use it to understand the reasoning moves, then create your own work from your assignment, approved evidence and authentic project record. Do not submit the sample as your answer.
What should I send when requesting help?
Send the current instructions, rubric or policy, your own attempt, relevant feedback, the exact point of difficulty and the deadline. Remove passwords and protected personal information.
Will Bright Writers complete or submit the work for me?
No. Support is educational and review-based. You remain the learner, author and person responsible for every submission and institutional decision.
Does every DNP project need inferential statistics?
No. The required analysis depends on the question, design, data and program. Descriptive and time-ordered methods may be more appropriate for some improvement projects.
Can I email my dataset for help?
Do not send identifiable or protected data. Follow institutional and site rules. Use de-identified, aggregated or simulated data only when permitted.