Fixing tests, assumption checks, sample size and missing-data rules ahead of analysis, the PU525 Unit 9 plan example sets out how a composite home-visiting evaluation will be judged. Searches like "pu 525 unit 9 assignment example", "pu525 unit 9 sample" and "pu525 unit 9 example" land here.
What a finished PU525 Unit 9 statistical analysis plan looks like
The plan runs about six pages under numbered headings modeled on published analysis plans. It opens with the evaluation question and a single primary hypothesis: the proportion exclusively breastfeeding at six months differs between enrolled and comparison mothers. A variable table lists the outcome, the exposure and six covariates with measurement level and coding. The primary analysis is a comparison of two proportions reported as a risk difference with a 95 percent interval, followed by logistic regression adjusting for maternal age, parity, WIC participation and prior breastfeeding. A sample size section shows that 272 mothers per group give 80 percent power to detect 45 versus 57 percent, inflated to 332 for attrition. The missing-data section specifies multiple imputation with 20 datasets and a complete-case sensitivity analysis. Shell tables with empty cells close the plan.
How a PU525 Unit 9 example is structured
The plan is ordered so that every decision is recorded before the data could influence it. The question and primary hypothesis come first and are singular on purpose: one primary outcome protects against searching many outcomes for a significant one. Variables are defined next, with coding fixed, since recoding after seeing results is a quieter form of the same problem. The primary analysis precedes secondary ones, and each names its fallback: Fisher's exact test if expected cell counts fall below five, and a simpler model if the adjusted one fails to converge. Sample size follows because it depends on the chosen test. Missing data receive a full section rather than a sentence, stating the assumed mechanism and how a sensitivity analysis would reveal whether conclusions depend on it. Shell tables end the plan, showing exactly what will be reported.
One primary hypothesis
Exclusive breastfeeding at six months, enrolled versus comparison. Secondary outcomes are listed separately and labeled exploratory, so no one can later promote a secondary result to headline status.
Variables coded in advance
Outcome, exposure and covariates with levels, codes and sources. Parity is fixed as first birth versus later births, a choice recorded here instead of made after the results arrive.
Tests and their fallbacks
A two-proportion comparison with a risk difference, then adjusted logistic regression. Each analysis names the assumption it depends on and the alternative used if that assumption fails.
Power and attrition
The calculation behind 272 per group, the effect size it targets and why, and the inflation to 332 to cover expected losses. Inputs are cited to the program's composite pilot data.
Missing outcomes
Imputation under a missing-at-random assumption, the variables in the imputation model, and a complete-case analysis whose disagreement with the main result would be reported, not hidden.
Shell tables
Empty tables for participant characteristics, the primary comparison and the adjusted model, formatted exactly as the final report will present them.
Where marks go in PU525 Unit 9
Analysis plans in PU525 tend to be graded on alignment between questions and tests, justification of methods, treatment of assumptions and missing data, and completeness. Alignment marks require each hypothesis to have one named analysis, with the outcome's measurement level driving the choice. Justification marks go to reasons stated for each test and for the effect size in the power calculation. The missing-data row is where plans most often fall short: a sentence saying incomplete records will be removed earns little, while a stated mechanism, a method and a sensitivity check earn the row. Deductions also follow plans with several primary outcomes, tests chosen without checking assumptions, sample sizes given without the inputs that produced them, and shell tables missing entirely. A plan written in the past tense, as though results already exist, draws comment in many sections.
Get a PU525 Unit 9 example written to your instructions
Describe the study your PU525 Unit 9 plan covers, its outcome, groups and data source, then attach the prompt and rubric. The plan that comes back fixes every test, assumption and missing-data rule in advance, with a power calculation whose inputs are shown. Your first custom sample is free and arrives in 24-48 hours.
PU525 Unit 9 questions, answered
Why commit to the analysis before seeing the data?
Because choices made after seeing results tend to favor significant findings, even without intent. Fixing the primary outcome, the tests and the treatment of incomplete records in advance means the reported result is the one planned, not the most striking of many tried. Funders and journals increasingly expect this, and the course treats it as basic practice.
What is multiple imputation, and is it always required?
It replaces each missing value with several plausible values drawn from a model of the observed data, analyzes each completed dataset and combines the results. It is not always required. With very little missing data, a complete-case analysis may be defensible if justified. The example uses imputation because 18 percent attrition is too large to ignore.
Where do the numbers in a power calculation come from?
From prior evidence: a pilot, a published study in a similar population, or a national estimate. The expected comparison-group proportion and the smallest difference worth detecting drive the result. The example cites its composite pilot for the 45 percent baseline and explains why a 12-point increase is the smallest the program would consider meaningful.