Difficulty: Medium
When resignations rise after a policy change, what evidence is needed before claiming cause?
A Compare other factors and evidence
B Publish the policy immediately
C Assume timing proves causality
D Ignore the workforce context
Answer: A. Compare other factors and evidence. Timing alone does not show the policy caused the increase.
Two departments report different absence rates. What should analysts check first?
A Metric definitions and workforce context
B The managers’ personalities
C Only employee ages
D Whether charts use colors
Answer: A. Metric definitions and workforce context. Different definitions or working conditions may explain the difference.
What should HR examine when a model flags one group as high-risk?
A Names of flagged employees publicly
B How to discipline the group
C Inputs, errors, and fairness
D Whether managers like the model
Answer: C. Inputs, errors, and fairness. Potential bias and data issues should be reviewed before acting on predictions.
Which checks help explain a sharp recruiting conversion drop?
A Change the hiring standard automatically
B Blame the recruiter immediately
C Remove older records
D Check stages, dates, and data
Answer: D. Check stages, dates, and data. Stage definitions and data quality should be checked before interpreting a shift.
How can HR fairly compare teams of different sizes?
A Show rates and denominators
B Display counts without context
C Rank supervisors by totals
D Hide small-team results without note
Answer: A. Show rates and denominators. Rates plus denominators make size differences easier to interpret.
Survey response is low in one shift. What should analysts consider?
A Access, timing, and trust
B Assume low engagement
C Publish respondent names
D Exclude the shift silently
Answer: A. Access, timing, and trust. Low participation may reflect survey access or trust rather than employee attitudes.
What should HR review when course completion rises but skills do not?
A Attendance incentives only
B Course title formatting
C Employee job grades
D Learning design and transfer
Answer: D. Learning design and transfer. Completion alone does not establish learning or on-the-job transfer.
What limitation arises when a model predicts turnover from salary alone?
A Salary always causes exits
B Important context may be missing
C Prediction becomes causal
D No fairness review is needed
Answer: B. Important context may be missing. Salary alone may omit other relevant factors and can mislead decisions.
How should analysts respond when a staffing forecast misses actual needs?
A Hide the forecast history
B Review assumptions and recalibrate
C Blame line managers
D Discard all workforce data
Answer: B. Review assumptions and recalibrate. Forecast errors should lead to assumption review and model adjustment.
Which safeguard helps when a small-team dashboard exposes one person’s absence?
A Publish the exact value
B Add employee names
C Share the dashboard widely
D Suppress or combine small cells
Answer: D. Suppress or combine small cells. Suppressing small groups can reduce re-identification risk.
HR wants to join survey data with health records. What should happen first?
A Assess purpose, necessity, and safeguards
B Merge every field immediately
C Share names with managers
D Assume access means permission
Answer: A. Assess purpose, necessity, and safeguards. Sensitive linkage needs a clear purpose, necessity assessment, and safeguards.
How should analysts address duplicate employee IDs across systems?
A Average all records together
B Delete one system
C Continue with totals
D Resolve identity matching errors
Answer: D. Resolve identity matching errors. Incorrect record matching can distort person-level and aggregate results.
What evidence helps evaluate a retention program after launch?
A A promotional announcement
B Baseline and suitable comparison
C One manager’s opinion
D A larger chart only
Answer: B. Baseline and suitable comparison. Baseline and comparison evidence help separate program effects from broader trends.
What should a report disclose when a metric definition changes?
A Only the newest value
B Definition change and comparability limits
C An assumption of real improvement
D No note about the change
Answer: B. Definition change and comparability limits. A definition change can break trend comparability and should be disclosed.
How should HR respond when a leader requests names behind engagement scores?
A Send raw responses immediately
B Use aggregated need-to-know reporting
C Share comments with identifiers
D Open all employee records
Answer: B. Use aggregated need-to-know reporting. Reports should use aggregated information unless identifiable detail is necessary and authorized.
One location reports strong engagement but high turnover. What should HR explore?
A Discard the engagement data
B Assume all staff are content
C Different measures and employee groups
D Assume survey fraud
Answer: C. Different measures and employee groups. Different measures and populations can produce distinct patterns that need explanation.
How can HR compare time-to-hire after a recruiting redesign?
A Only the final annual number
B Candidate names and ages
C Unrelated training completion
D Comparable stages and periods
Answer: D. Comparable stages and periods. Comparable start and end points are needed for fair before-and-after analysis.
Which details should accompany an unexplained dashboard risk score?
A Employee rankings
B A guarantee of accuracy
C Unreviewed model code
D Drivers, limits, and context
Answer: D. Drivers, limits, and context. Decision-makers need enough context to understand a score’s drivers and limits.
Analysts find a relationship between overtime and exits. What is a sound interpretation?
A Overtime definitely caused exits
B No relationship can exist
C Association needing further study
D The model proves individual intent
Answer: C. Association needing further study. The observed association suggests a question but does not prove causation.
How should HR compare workforce measures across countries?
A Assume identical labor systems
B Use one threshold universally
C Check local definitions and context
D Ignore law and culture
Answer: C. Check local definitions and context. Local rules and labor-market context can affect what a metric means.
Managers want to act on a prediction about one employee. What is a key concern?
A Avoid unsupported individual decisions
B Use the score as fact
C Notify the team publicly
D Skip human review
Answer: A. Avoid unsupported individual decisions. Predictions are uncertain and need careful, fair, and context-aware human review.
What protects survey respondents when HR reports quotations?
A Include department and job title
B Remove identifying details
C Share original files widely
D Keep no review process
Answer: B. Remove identifying details. Quotations should be de-identified to reduce the chance of identification.
Which population detail should HR clarify when contractors are included?
A The dashboard font
B Who approved the chart
C Population included in the measure
D Unrelated employee benefits
Answer: C. Population included in the measure. The metric must state whether contractors are included in its population.
How should analysts respond when a model performs poorly after reorganization?
A Revalidate on current data
B Keep using the old score
C Hide the performance drop
D Expand its use immediately
Answer: A. Revalidate on current data. Changed workforce conditions can make an older model less reliable.
What makes an HR analytics recommendation responsible?
A A confident tone alone
B A high model score
C Evidence, limits, and review
D A large sample by itself
Answer: C. Evidence, limits, and review. Responsible recommendations communicate evidence, uncertainty, and the need for review.