Difficulty: Hard
When outcome probabilities are credible, which measure averages financial results?
A Maximum observed payoff
B Unweighted score total
C Lowest possible loss
D Expected monetary value
Answer: D. Expected monetary value. Expected monetary value weights each payoff by its probability.
Which measure accounts for a decision maker’s risk preference?
A Expected utility
B Median cycle time
C Simple payoff maximum
D Raw data count
Answer: A. Expected utility. Expected utility represents outcomes using a utility scale that can reflect risk preferences.
If one option is no worse in every state and better in one, what condition is shown?
A Deadlock
B Dominance
C Sensitivity
D Data drift
Answer: B. Dominance. Dominance allows a clearly inferior alternative to be excluded from consideration.
Which analysis identifies the point where benefits cover costs?
A Scenario planning
B Data integration
C Break-even analysis
D Role-based access
Answer: C. Break-even analysis. Break-even analysis finds the value at which total benefits meet total costs.
How does robust optimization address uncertain inputs?
A Removes every constraint
B Maximizes a single uncertain input
C Assumes one forecast is exact
D Finds solutions resilient across cases
Answer: D. Finds solutions resilient across cases. Robust optimization seeks solutions that perform acceptably across uncertainty.
When a constraint’s shadow price is zero, what may this imply?
A Extra capacity adds no value
B The model has no variables
C The objective has no value
D All solutions are infeasible
Answer: A. Extra capacity adds no value. A zero shadow price can indicate that relaxing a nonbinding constraint adds no objective value.
Why use integer programming for indivisible decisions?
A Model only continuous percentages
B Require whole-number choices
C Ignore resource constraints
D Remove the objective function
Answer: B. Require whole-number choices. Integer programming restricts decision variables to whole-number values where needed.
What does Monte Carlo output help estimate?
A Only the largest input
B A fixed process sequence
C Outcome distribution under uncertainty
D One exact guaranteed answer
Answer: C. Outcome distribution under uncertainty. Repeated simulation estimates a range and distribution of possible results.
How does Bayesian updating use observed evidence?
A Replace probabilities with rankings
B Discard all prior information
C Guarantee a causal link
D Revise posterior beliefs
Answer: D. Revise posterior beliefs. Bayesian updating combines prior beliefs with evidence to form posterior beliefs.
Demand shifts and forecast errors grow over time. What model risk may be present?
A Explores plausible conditions
B Data compression
C Index fragmentation
D Role conflict
Answer: A. Explores plausible conditions. Model drift occurs when underlying relationships change and performance deteriorates.
When a recommendation changes customer behavior and its own input data, what risk emerges?
A Data normalization
B Feedback effects
C Linear feasibility
D Perfect calibration
Answer: B. Feedback effects. A feedback loop can change the system being modeled and alter later inputs.
Why can automation bias undermine DSS review?
A Data becomes more complete
B Models always show uncertainty
C Users over-trust automated advice
D Reviewers see every source
Answer: C. Users over-trust automated advice. Automation bias can lead users to accept recommendations without adequate scrutiny.
What does human-in-the-loop review add?
A Automatic approval of every result
B Removal of accountability
C Guaranteed unbiased choices
D Contextual judgment and oversight
Answer: D. Contextual judgment and oversight. Human review can add context and challenge model limitations.
When rankings shift after small weight changes, what should analysts test?
A Weight sensitivity
B Data masking
C Password strength
D Index coverage
Answer: A. Weight sensitivity. Weight sensitivity shows whether rankings depend heavily on chosen criterion weights.
Which feature separates a scenario from a probability forecast?
A It requires exact likelihoods
B Explores plausible conditions
C It always predicts one outcome
D It ranks options by utility
Answer: B. Explores plausible conditions. Scenario analysis explores plausible conditions without necessarily assigning probabilities.
Which record supports tracing transformed data through a DSS?
A Model screen theme
B User profile
C Data lineage
D Decision deadline
Answer: C. Data lineage. Data lineage shows how data moved and changed from source to result.
What is a key tradeoff of anonymous GDSS input?
A Latency versus encryption
B Access versus normalization
C Accuracy versus storage
D Candor versus accountability
Answer: D. Candor versus accountability. Anonymity may encourage candor but can make follow-up ownership harder.
How can explainable AI help reviewers assess advice?
A Show influential factors
B Force a single action
C Hide model inputs
D Remove counterexamples
Answer: A. Show influential factors. Explanations can reveal factors that influenced a recommendation.
Which question does a counterfactual explanation answer?
A How keys identify database rows
B Effect of changing one input
C A database key definition
D When a backup was created
Answer: B. Effect of changing one input. Counterfactuals describe how a different input might change a result.
Why evaluate a model under several data conditions?
A Eliminate all bias
B Increase the number of users
C Reveal model uncertainty
D Guarantee a stable future
Answer: C. Reveal model uncertainty. Testing varied conditions helps expose uncertainty and fragile assumptions.
Which practice prevents test data leaking into training?
A Copy target values into inputs
B Reuse all labels
C Tune on final test cases
D Separate evaluation data
Answer: D. Separate evaluation data. Data leakage can make evaluation seem more accurate than real deployment performance.
What should happen when critical decision data is incomplete?
A Assess sensitivity to missingness
B Treat blanks as zero always
C Hide the missing fields
D Assume all records are random
Answer: A. Assess sensitivity to missingness. Missing-data patterns can affect estimates and should be assessed explicitly.
Why preserve the option to delay an irreversible investment?
A Remove all uncertainty
B Retain real-option value
C Lock in every forecast
D Maximize sunk costs
Answer: B. Retain real-option value. Waiting can have value when new information may improve an irreversible decision.
When should decision accountability be documented?
A Only for discarded alternatives
B Only after a failure
C Before implementing the chosen action
D Never for model advice
Answer: C. Before implementing the chosen action. Documenting ownership and rationale makes decisions reviewable and accountable.
How should a DSS recommendation be treated when assumptions fail?
A Proceed without review
B Delete the source data
C Ignore changed conditions
D Reassess the model and choice
Answer: D. Reassess the model and choice. Recommendations depend on assumptions and should be revisited when those assumptions no longer hold.