Statistical Analysis Workflow: 10 practice questions

Question 1 of 10

A retailer asks whether a new layout increases weekly spend. Which definition states the analysis target before any calculation?

1. AA statistical question only needs a dataset and a preferred test; cases, variables, population, and estimand can all be chosen after inspecting the output.
2. BA statistical question should identify the cases, the measured variables, the population of interest, and the population quantity being estimated.
3. CIt is enough to name the observed sample and response column, because the target population and population quantity are implied by whichever statistic is computed.
4. DAn analysis plan is a record of the significant results selected after all models have been compared.

Question 2 of 10

Why is the population estimand different from the value calculated in one observed sample?

1. AA statistical question only needs a dataset and a preferred test; cases, variables, population, and estimand can all be chosen after inspecting the output.
2. BIt is the software script that runs the final test, regardless of the study question or design.
3. CIt is enough to name the observed sample and response column, because the target population and population quantity are implied by whichever statistic is computed.
4. DAn estimand is the population quantity the analysis is designed to estimate, not the value computed from one sample.

Question 3 of 10

What is the main protection gained by writing the analysis plan before inspecting the result?

1. AAn analysis plan links the question and study design to summaries, models, diagnostics, and reporting decisions before the result is known.
2. BA numerical summary always contains every feature visible in a graph.
3. CIt is the software script that runs the final test, regardless of the study question or design.
4. DAn analysis plan is a record of the significant results selected after all models have been compared.

Question 4 of 10

Before selecting a formal model, what can an exploratory display contribute to the workflow?

1. AThe graph should replace numerical summaries because a display preserves all inferential information, uncertainty, and every design qualification for the target population.
2. BAn analysis plan is a record of the significant results selected after all models have been compared.
3. CGraphical exploration can reveal structure and assumption problems before a formal method is selected.
4. DIt is the software script that runs the final test, regardless of the study question or design.

Question 5 of 10

A mean looks stable but the histogram shows two separated groups. How should the analyst use both pieces of evidence?

1. AA numerical summary always contains every feature visible in a graph.
2. BThe graph should replace numerical summaries because a display preserves all inferential information, uncertainty, and every design qualification for the target population.
3. CA graph establishes the population relationship whenever the plotted sample pattern is strong.
4. DQuantitative summaries and graphical displays complement one another because each can reveal features the other compresses or obscures.

Question 6 of 10

A sample mean exceeds last year's value. What can be concluded about the target population at this stage?

1. AA graph is only decoration after fitting and cannot reveal violations of model assumptions or data-quality problems.
2. BA numerical summary always contains every feature visible in a graph.
3. CA sample provides evidence about a population, but a sample result is not automatically a fact about that population.
4. DThe graph should replace numerical summaries because a display preserves all inferential information, uncertainty, and every design qualification for the target population.

Question 7 of 10

A scatterplot shows a strong pattern in a convenience sample. Why is the graph alone insufficient for a population claim?

1. AA graph describes the observed data and may challenge a model, but it does not by itself justify inference beyond the design.
2. BThe analyst should run every available procedure, retain the one with the smallest p-value, and redefine the research question to match that output.
3. CA graph establishes the population relationship whenever the plotted sample pattern is strong.
4. DA graph is only decoration after fitting and cannot reveal violations of model assumptions or data-quality problems.

Question 8 of 10

A prespecified linear model gives a positive estimate, but the scatterplot shows curvature and two clusters, and the sample is convenience-based. What should the analyst do before treating the estimate as confirmatory?

1. AA graph establishes the population relationship whenever the plotted sample pattern is strong.
2. BThe most familiar software command should determine the model even when its assumptions conflict with the sampling and measurement process.
3. CA graph is only decoration after fitting and cannot reveal violations of model assumptions or data-quality problems.
4. DExploratory graphics and classical confirmatory methods can be used together, and disagreements between them should be investigated.

Question 9 of 10

The target is median customer spend and the file contains repeated rows per customer, but an analyst selects the first software procedure for independent means. Which ordering principle was violated?

1. AThe most familiar software command should determine the model even when its assumptions conflict with the sampling and measurement process.
2. BThe research question and data-generating process should determine the method, rather than the list of procedures available in software.
3. CThe analyst should run every available procedure, retain the one with the smallest p-value, and redefine the research question to match that output.
4. DA qualified conclusion reports only whether a conventional significance threshold was crossed.

Question 10 of 10

A convenience sample yields a small positive estimate with a narrow interval, while diagnostics show model mismatch. Which elements must the final conclusion include?

1. AA qualified conclusion reports the estimated magnitude, its uncertainty, the population and conditions covered, and the assumptions that matter.
2. BIt reports the sample estimate as a universal fact.
3. CA qualified conclusion reports only whether a conventional significance threshold was crossed.
4. DIt is enough to name the observed sample and response column, because the target population and population quantity are implied by whichever statistic is computed.