Statistics Professional Standards & Competencies 2 — Questions and Answers
Question 1: A statistician discovers that a client has cherry-picked favorable results from multiple analyses to present to regulators. What is the most appropriate professional response?
- Rewrite the report to include all analyses
- Report the issue to regulators directly without informing the client
- Refuse to sign off on the report and explain the ethical concern to the client (Correct answer)
- Accept the client's decision as it is their data
Correct answer: Refuse to sign off on the report and explain the ethical concern to the client
A statistician must refuse to endorse misleading presentations and communicate the ethical issue to the client before taking further action.
Question 2: Which ASA (American Statistical Association) ethical principle prohibits using the same dataset to both generate and confirm a hypothesis without disclosure?
- Confidentiality
- Avoiding conflicts of interest
- Transparency in analysis and reporting (Correct answer)
- Responsibility to society
Correct answer: Transparency in analysis and reporting
Transparency requires disclosing when exploratory and confirmatory analyses are performed on the same data, to prevent inflated false-positive rates.
Question 3: A client requests that a statistician use a one-tailed test instead of a two-tailed test after seeing the data, solely to achieve a p-value below 0.05. This practice is best described as:
- Appropriate sensitivity analysis
- HARKing (Hypothesizing After Results are Known)
- Legitimate exploratory analysis
- P-hacking (Correct answer)
Correct answer: P-hacking
Changing the test direction after observing results to achieve significance is p-hacking, a form of research misconduct.
Question 4: Under the ASA's Ethical Guidelines, when a statistician is listed as a co-author, what is their primary responsibility?
- Verify only the statistical sections of the paper
- Take responsibility for the statistical accuracy of the entire work (Correct answer)
- Ensure only the data collection methods are sound
- Defer all methodological decisions to the principal investigator
Correct answer: Take responsibility for the statistical accuracy of the entire work
Co-authorship implies responsibility for the statistical integrity of the complete work, not just isolated sections.
Question 5: A government statistician is pressured by officials to delay releasing unemployment data until after an election. The most appropriate action is to:
- Comply if the delay is less than two weeks
- Release the data on the original schedule regardless of political pressure (Correct answer)
- Seek legal counsel before releasing any data
- Modify the report to make results less politically sensitive
Correct answer: Release the data on the original schedule regardless of political pressure
Official statistics must be released on pre-announced schedules free from political interference, per statistical agency independence principles.
Question 6: Which of the following best defines 'selective reporting' in statistical practice?
- Focusing a report on the most policy-relevant findings
- Publishing only statistically significant outcomes while suppressing null results (Correct answer)
- Using stratified sampling to target specific subgroups
- Reporting results separately for different demographic groups
Correct answer: Publishing only statistically significant outcomes while suppressing null results
Selective reporting occurs when null or unfavorable results are suppressed, distorting the evidence base.
Question 7: A statistician working for a pharmaceutical company is asked to analyze clinical trial data. A conflict of interest most clearly exists when:
- The statistician holds stock in a competitor company
- The statistician's bonus is tied to achieving a statistically significant result (Correct answer)
- The statistician previously worked in academia
- The statistician uses proprietary software owned by the company
Correct answer: The statistician's bonus is tied to achieving a statistically significant result
A financial incentive directly linked to the analytical outcome creates a conflict of interest that can bias analysis.
A statistician discovers that a client has cherry-picked favorable results from multiple analyses to present to regulators.
What is the most appropriate professional response?