DASS · STATISTICAL ANALYSIS
Analysis-Planning Worksheet
A fictional example
This invented workshop study illustrates planning decisions. It contains no client results, fitted model, or sample-size recommendation. Proposed models require review against the actual design and data.
Question: What is the mean score difference between randomized workshop and comparison groups at week 12?
Outcome and population: A continuous skill-assessment score for enrolled participants. Document its scoring and measurement evidence before interpreting a difference as improvement.
Assignment and observations: Individual randomization; baseline and weeks 4, 8, and 12. One row per participant per visit, with participant ID, assignment, scheduled visit, actual assessment date, and score. Record any additional clustering.
Target comparison: Week-12 group difference under the prespecified analysis strategy. A difference in change or an overall interaction is a distinct comparison.
Candidate approach: A longitudinal model with categorical visit and a justified within-person covariance structure. Decide whether and how baseline is included, which covariates are used, and how the week-12 contrast is obtained. Specify assumptions before evaluating results.
Missing information: Track availability by group and visit; record reasons where feasible. Assess the assumptions supporting the primary missing-data approach and plan sensitivity analyses for plausible departures.
Checks and reporting: Review model convergence, residuals, dependence, influential observations, and precision. Report the contrast with a confidence interval, observation counts, specification, limitations, and reproducible code.
Keep distinct: In an unconstrained model, a group coefficient of 1 and group × week-12 coefficient of 4 imply a final-visit group difference of 5, while the difference in change is 4. These are illustrative coefficients without uncertainty estimates.
Prepared October 10, 2026 · Planning aid, not a complete statistical analysis plan.
DASS · BLANK WORKSHEET
1. Question, design, and data
Research question and intended interpretation: descriptive, predictive, or causal?
Target comparison: groups, time point, population, and effect scale
Outcome, units, scoring, measurement evidence, and distribution
Assignment / sampling design, eligibility, and comparison conditions
Observation schedule and actual timing; repeated measurements and identifiers
Clustering: people within schools / clinics / sites, and assignment unit
Available sample, number of independent clusters, and precision / power planning
DASS · BLANK WORKSHEET
2. Analysis and reporting
Candidate model, time specification, baseline handling, and covariance structure
Adjustment variables and rationale; avoid unplanned post-assignment adjustment
Missing outcomes / predictors, reasons, assumptions, and primary approach
Diagnostics, convergence, sensitivity analyses, and criteria for revisions
Primary contrast, additional comparisons, and multiplicity plan
Reporting: estimates, uncertainty, counts, limitations, tables, and reproducible code
Analysis-plan version, review date, reviewer, and documented deviations
Guides: Mixed models vs. repeated-measures ANOVA; Pre/post analysis with a comparison group. Research support: statisticalsolutions.org/researchers.