What Is Decision-Making Competence?
Decision-making competence is the measurable ability to apply defensible principles of judgment and choice across different problems. In psychological research, it commonly refers to performance on tasks that test whether people evaluate information consistently, recognize uncertainty, compare alternatives using stated rules, resist irrelevant presentation effects, and calibrate confidence to accuracy. The central scientific question is whether these skills cohere sufficiently to be measured, and whether scores predict consequential behavior beyond conventional measures of intelligence. Bruine de Bruin, Parker, and Fischhoff (2007) established a widely used adult battery and examined its construct and external validity.
A decision-making competence score describes performance on a particular assessment under specified conditions. It is not a global ranking of personal wisdom, a certification that someone will choose correctly in every situation, or a clinical diagnosis. Decisions depend on goals, values, available information, time, incentives, institutional constraints, and chance. An excellent process can produce an unfavorable outcome, while a defective process can sometimes produce a fortunate one. Meaningful assessment therefore requires an explicit standard for the quality of the process, appropriate measures of performance, and evidence that the result generalizes beyond the test.
Researchers commonly distinguish belief assessment, value assessment, integration of beliefs and values, and metacognition. Belief assessment asks how accurately and coherently a person represents uncertain events. Value assessment concerns the significance assigned to possible outcomes. Integration concerns the application of decision rules to competing alternatives. Metacognition concerns understanding one's own knowledge and uncertainty. These are partly related capabilities, and their empirical correlations are a research question rather than an assumption. Parker and Fischhoff (2005) and Bruine de Bruin et al. (2020) provide the foundational individual-differences framework.
Scientific Status and Conceptual Boundaries
Competence, performance, and outcomes
Competence is a dispositional capacity inferred from observed behavior across tasks. Performance is the behavior actually recorded on a specified task, with its particular instructions and constraints. Decision quality may mean the quality of the reasoning process, the fit between choice and the person's priorities, or the realized outcome. These criteria must be separated before the researcher assigns a score. A person who uses the available information consistently may still experience loss because an adverse event occurs; another person may take a reckless gamble that happens to pay off. Outcome alone is consequently an incomplete test of decision competence.
Judgment is the evaluation of propositions, probabilities, evidence, or attributes. Choice is the selection of an action from available alternatives. Reasoning describes procedures for deriving or evaluating conclusions, including deductive and probabilistic procedures. Cognitive biases are task-dependent systematic patterns of deviation from a declared benchmark. Heuristics are strategies that may be efficient or adaptive in particular environments. A logical fallacy is an error in argument structure, while a statistical error concerns a specific inference or procedure. Neither ordinary bias-task performance nor a low score on a competence battery independently identifies a mental disorder.
Decision-making competence must also be distinguished from decision-making capacity in medical ethics and law. Clinical capacity concerns whether a person can understand relevant information, appreciate consequences, reason about options, and communicate a choice in a particular context. Such capacity is assessed through clinical and legal standards, and it may vary across decisions and time. Psychological decision-making competence batteries are research instruments, not substitutes for capacity evaluations or legal determinations.
Descriptive behavior and normative evaluation
A descriptive model predicts what people actually do. A normative model stipulates the principle against which a response is assessed, such as basic probability coherence, an explicitly announced decision rule, or irrelevance of irrecoverable costs. These are different statements. A surprising response is not automatically an error: the assessment must show which features are equivalent, which information is relevant, and which assumptions are in force.
When framing tasks use mathematically equivalent outcomes, consistency is typically the benchmark. When temporal probability tasks compare nested events, probability monotonicity supplies the benchmark: an event that is a subset of another cannot have higher probability than the containing event. When people are instructed to follow a lexicographic rule, correct execution of that rule is the criterion. Some value-sensitive judgments admit several rational answers, and observers cannot identify a uniquely optimal preference without knowing the person's goals and constraints.
How the Field Developed
From classic behavioral tasks to individual differences
Much early behavioral decision research compared average responses under different experimental conditions. Studies of framing, overconfidence, probability judgment, and sunk costs documented systematic patterns within particular tasks. Parker and Fischhoff's 2005 research brought several such measures into a common assessment and asked whether performance differences between individuals were coherent and related to life outside the laboratory. They found generally small positive associations between tasks and meaningful associations between aggregated competence and external characteristics. This established the individual-differences problem: whether the same people tend to perform relatively well across diverse judgment tasks. Parker and Fischhoff (2005).
Bruine de Bruin, Parker, and Fischhoff extended and adapted the tasks into Adult Decision-Making Competence, or A-DMC. Their 2007 research evaluated seven task types in the development and validation program, examined reliability, and related the resulting assessments to intelligence, socioeconomic characteristics, styles of decision making, and a separate inventory of negative life outcomes. The commonly used standard A-DMC subsequently consists of six components, with path independence often omitted because of measurement concerns. It is important to distinguish the original seven-task development from the six-component battery typically described in later applications. Bruine de Bruin et al. (2007); Bruine de Bruin et al. (2020).
Youth and life-span extensions
Youth Decision-Making Competence, or Y-DMC, adapts the multi-task strategy to late adolescents. It was evaluated against extensive external indicators in the 2005 longitudinal-study population. Later work extended task-based competence assessments to preadolescent children, while also examining how performance changes with age and cognitive experience. The scoring principles and item characteristics need age-appropriate validation; a measure for adults cannot simply be treated as interchangeable with a measure for children. Weller et al. (2012); Parker and Fischhoff (2005).
The Adult Decision-Making Competence Battery
Applying decision rules
Applying Decision Rules asks participants to choose among alternatives described by several attributes while following specified procedures. Some items require selecting an option with the highest value on a priority attribute; others require combining attributes in the way the instructions define. Performance is scored against the prescribed rule, not against a researcher's private opinion about which product or treatment the participant ought to prefer.
The task measures whether the person can keep relevant information available, understand a verbal rule, ignore distracting characteristics, and execute a comparison accurately. Errors may reflect working-memory demands, reading comprehension, numeracy, difficulty mapping instructions to attributes, or lapses of attention. A low score does not by itself identify which component caused the error. The ability is particularly relevant to structured decisions where a rule or constraint is agreed upon before alternatives are considered. Bruine de Bruin et al. (2020).
Resistance to framing
Resistance to Framing presents substantively equivalent options in different verbal forms, for example describing a probability as survival rather than mortality. A consistency score measures the extent to which choices or evaluations remain stable when only the presentation changes. The benchmark is invariance to information that has been shown to be normatively irrelevant under the task assumptions.
A framing difference can arise from attention, affect, the accessibility of losses or gains, or a different interpretation of the wording. Careful assessments hold objective content equivalent, check comprehension, and distinguish framing of a logically equivalent outcome from framing that truly changes the implied meaning. Responses need to be scored across multiple problems rather than treating one reaction as the definitive measure of a person's judgment. Bruine de Bruin et al. (2007).
Underconfidence and overconfidence
Under/Overconfidence measures the correspondence between the confidence assigned to knowledge answers and the proportion answered correctly. One common implementation asks for true-or-false judgments and then elicits confidence from chance-level to certainty. The difference between average confidence and observed accuracy characterizes group-level overconfidence or underconfidence in the test's item set.
Consider a participant who reports 80% confidence across a sufficiently large set of comparable answers but is correct on 65%. The 15-percentage-point gap indicates overconfidence for that set. It does not identify the calibration of every single answer and does not show that the same person will be overconfident in unrelated subjects. Item difficulty, selection of questions, response format, and the range of confidence permitted can alter the result. This component intersects with the neighboring scientific subject of Probability Forecasting and Calibration, which addresses probabilistic prediction more extensively.
Consistency in risk perception
Consistency in Risk Perception measures whether probability judgments about logically related events obey basic constraints. If event A necessarily implies event B, P(A) cannot exceed P(B). A probability for an event occurring within the next year should not exceed the probability of its occurring at least once within the next five years when the event definitions and reference point are held fixed. Two mutually exclusive, collectively exhaustive alternatives must sum to one.
The task does not require researchers to know the true probability of a future personal event. It can evaluate internal coherence without observing the event's eventual occurrence. Consequently, logical consistency and empirical predictive accuracy are distinct. A person can produce internally consistent probability statements that are poorly calibrated to actual event frequencies, or accurate forecasts that contain isolated coherence violations. Bruine de Bruin et al. (2020).
Recognizing social norms
Recognizing Social Norms asks participants to estimate how common particular judgments are among a defined reference group. Their estimates are compared with independently elicited responses from that group. The scientific target is social prediction accuracy, not agreement with the group's moral judgments and not conformity to a social norm.
A score depends on the social reference population, the behaviors selected, the sample used to estimate the actual norm, and the measurement context. If the benchmark group is drawn from a different community or era, the resulting apparent error may reflect a faulty comparison rather than a generalized weakness of social judgment. This component also depends on exposure to relevant social environments and must be interpreted cautiously in culturally diverse populations. Parker and Fischhoff (2005).
Resistance to sunk costs
Resistance to Sunk Costs presents choices involving prior investments of money, effort, or time that cannot be recovered. The normative rule is that irrecoverable expenditures, when they do not affect future benefits or costs, should not determine the present choice. Participants perform better when they compare future consequences and avoid continuing solely because they have already invested resources.
This principle requires a genuine sunk cost. A previous investment that produces future obligations, information, reputation effects, or valuable options may remain relevant through those forward-looking consequences. Test scenarios simplify these complications to isolate the targeted tendency. In actual organizations, escalation of commitment can involve strategic bargaining, accountability, and multiple stakeholders that cannot be reduced to one vignette score. Bruine de Bruin et al. (2007).
The historical path-independence component
Some original seven-task versions also tested path independence: whether the final choice depends on irrelevant intermediate steps when the available options and preferences are substantively unchanged. The component was subsequently excluded from widely used six-task forms because its psychometric behavior raised concerns. Studies must specify precisely which version and component count they used. Mixing six- and seven-task composites without harmonized scoring can create misleading comparisons. Bruine de Bruin et al. (2007); Bavolar's validation study.
How Scores Are Computed and Interpreted
Component scores and aggregate indices
A-DMC reports task-specific scores and may combine them into an overall index. A component based on correct rule applications uses different raw units from a component based on confidence calibration or resistance to framing. Aggregation therefore commonly requires standardization or another stated weighting method. The choice of reference sample for standardization matters: a standardized score communicates standing relative to that sample, not an absolute natural unit of rationality.
A composite can be useful when researchers expect shared variation across tasks and need a broad measure. It is less informative when the purpose is to explain a particular decision failure. A strong composite can conceal weakness in one domain, and averaging does not establish that all six tasks reflect a single underlying psychological mechanism. Reporting both aggregate and component results improves interpretation.
What makes a score reliable?
Reliability concerns the consistency of measurement rather than its truth. Internal consistency asks whether relevant items or components behave sufficiently coherently. Test-retest reliability asks whether people maintain a similar relative standing after an interval under conditions where their ability is expected to remain reasonably stable. Reliability is affected by the number of items, task heterogeneity, sampling, score distribution, changes in instructions, and the length of the retest interval.
Some components intentionally test different skills, so demanding a high inter-item correlation across every element can be conceptually inappropriate. Yet unreliable component scores also attenuate correlations with external criteria and make individual interpretation hazardous. Validation studies must identify which reliability coefficient is being calculated for which score and population. Reports of acceptable aggregate reliability cannot be generalized automatically to every subscale, language, or age group. Bruine de Bruin et al. (2007); Bavolar.
Factor structure and construct validity
Construct validity requires a network of evidence showing that the intended construct explains performance patterns better than plausible alternatives. Positive associations between components support commonality, but they do not prove unidimensionality. Different studies have evaluated one-factor and more differentiated solutions. Their conclusions depend on task selection, sample composition, measurement errors, and analytic decisions.
A well-specified assessment should consider convergent validity with other decision-related measures, discriminant validity against partly distinct traits, sensitivity to education and numeracy, and invariance across demographic groups. The theoretical issue is whether decision-making competence is best understood as one broad ability, several correlated competencies, or a useful composite of distinct skills. Current evidence supports meaningful common variance while leaving the exact latent structure open to refinement. Appelt et al. (2011); Parker et al. (2018).
Is Decision-Making Competence Different From Intelligence?
Reasoning ability, executive functioning, reading comprehension, and numerical literacy help people interpret alternatives and apply formal rules. Accordingly, decision-making competence is positively associated with measures of cognitive ability. Some components, particularly Applying Decision Rules and framing resistance, can impose substantial fluid-reasoning demands. Other components may rely more strongly on accumulated experience, insight into personal knowledge, and regulation of impulses.
This overlap creates a legitimate validity challenge. If a task requires specialized vocabulary or heavy mental arithmetic, a low score could partly reflect schooling or language proficiency instead of the target decision principle. Conversely, an association between a competence composite and outcomes that remains after statistical adjustment for intelligence suggests incremental predictive information, not absolute independence from intelligence. The initial A-DMC evidence and later reviews support this more nuanced interpretation. Bruine de Bruin et al. (2007); Bruine de Bruin et al. (2020).
Numeracy is especially important because probabilities, percentages, ratios, and expected values appear in several judgment tasks. Objective numeracy measures mathematical competence, while subjective numeracy reflects perceived numerical ability and preference for numerical information. Both can shape the interpretation of risk communication. A score change after simplifying numerical presentation may show that the interface removed an unnecessary comprehension obstacle, rather than that a deep decision trait changed. Reyna and Brainerd (2023).
Cognitive control and neuroscience
Research using cognitive and brain measures provides additional hypotheses about mechanisms, including logical reasoning, monitoring of conflict, and integrating socially relevant information. A 2019 neuroimaging investigation associated A-DMC performance with reasoning measures and structural variation in regions implicated in cognitive control and social decision making. These are correlational findings. Regional associations do not supply an individual diagnostic marker or show that changing a brain feature would cause an improvement in decision competence. Román et al. (2019).
Motivation, Emotion, Experience, and Decision Strategy
A person's effort can vary with the stakes, familiarity, fatigue, incentives, and perceived legitimacy of a test. Deliberation may be useful when alternatives are complex and time is available. Experienced decision-makers may use well-learned recognition processes that are effective in their domain. Emotional responses can prioritize information, signal personally important consequences, or bias interpretation in context. No simple rule states that faster intuitive responses are always inferior to slower analytical responses.
The dual-process family of theories describes distinctions among more automatic and more deliberate processes, but the mapping of every laboratory task to precisely two mental systems is contested. An incorrect answer to a framing item does not prove that one specific system was active, just as a correct answer does not demonstrate that reasoning was entirely deliberate. Explanations should be tied to experimentally distinguishable mechanisms and process measures. Bruine de Bruin et al. (2020).
Metacognitive awareness can support requests for second opinions, appropriate confidence, and adaptation of strategy. Yet self-ratings of competence are not interchangeable with measured performance. Some people underestimate themselves while performing well, and some report strong confidence despite errors. Assessment should retain direct behavioral tasks and treat self-report as a complementary measure, subject to its own reliability and response biases.
Does a Laboratory Score Predict Everyday Decisions?
Evidence from real-world outcomes
The Decision Outcomes Inventory, or DOI, asks about concrete negative outcomes a person might have experienced, while recognizing that not every individual had equal exposure to every opportunity for such an outcome. Early validation studies found that better A-DMC performance was associated with fewer reported negative life outcomes, including after adjustment for selected demographic and cognitive variables. A later item-level analysis found that many individual adverse outcomes were more likely in participants with lower competence scores even after accounting for age and socioeconomic status. Bruine de Bruin et al. (2007); Parker et al. (2015).
These findings are evidence of association and predictive validity within the studied samples, not a causal demonstration that low test scores generated each adverse event. The outcome inventory involves retrospective self-report, unequal risk exposure, socioeconomic constraints, and events with multiple causes. Avoiding an adverse event can reflect prudent choice, but it can also reflect better resources, fewer opportunities for exposure, or good luck. Valid analyses should address these complications explicitly.
Evidence over eleven years
An influential longitudinal analysis compared Y-DMC around age 19 with A-DMC around age 30, eleven years later. The authors reported positive associations across the two instruments even after adjustment for cognitive ability. The analysis included 146 people who completed both assessments, while broader study samples were used for additional comparisons. This supports relative stability of something captured by the batteries across a substantial interval. The two instruments were not identical, attrition and sample characteristics matter, and stability of relative rank should not be confused with an absence of skill development. Parker, Bruine de Bruin, Fischhoff, and Weller (2018).
Prediction is conditional, not deterministic
To claim predictive usefulness, researchers should report how much outcome prediction improves beyond a plausible baseline and whether the effect replicates in different populations. Out-of-sample prediction, preregistered models, and confidence intervals are particularly valuable when tests are used to make decisions about people. A statistically significant association can be practically modest; a large correlation in a selected sample may shrink in a more heterogeneous field setting.
The relevant predictive target also matters. Accuracy in recognizing social norms does not entail that someone will manage money well; correct application of a simple decision rule does not guarantee skill in navigating conflicting values or incomplete information. Broad competence measures should be supplemented by domain-specific observation for consequential applications. Appelt et al. (2011).
Development, Aging, and Cultural Generalizability
Childhood and adolescence
The acquisition of decision skills involves numerical learning, probability concepts, inhibition, social experience, and the capacity to hold multiple attributes in mind. Studies with children have explored whether components of competence already covary during preadolescence. Weller and colleagues assessed 101 children aged 10 to 11 and examined the structure of decision performance, extending the scientific question beyond adult samples. Developmental differences cannot be read as universal deficits without checking whether instructions and tasks are suitable for the child's experience and abilities. Weller et al. (2012).
Adult aging
Research on adult age differences shows heterogeneous patterns. Some tasks emphasizing fluid reasoning, such as rule application or resistance to framing, tend to be more difficult with age in the studied samples. Other tasks, including resistance to sunk costs, may be preserved or improve, potentially reflecting experience or changed priorities. Self-perceptions of competence can move differently from observed task scores. Thus a single average age effect obscures meaningful task differences. Bruine de Bruin, Parker, and Fischhoff (2012).
Translation and measurement invariance
Cross-cultural validation is more than translating words. It requires checking whether item content, reference norms, comprehension demands, and task meaning remain comparable. A Slovak student-sample validation reported broadly similar findings to the American adult research while identifying some group differences across subscales. Such replication is useful evidence but does not establish identical norms for every language or social setting. Bavolar.
Social-norm questions are particularly sensitive to reference groups, and consumer-choice items may presuppose familiarity with certain products. Researchers should test factor structure, score distributions, item functioning, and criterion validity within the intended population. International comparisons require evidence of measurement equivalence before attributing score gaps to enduring psychological differences.
How Assessment Studies Can Go Wrong
Task contamination and method variance
A competence battery can inadvertently assess education, English reading fluency, attentional endurance, familiarity with multiple-choice tests, or exposure to particular experimental paradigms. Correlations among tasks may also emerge from shared testing methods. Therefore the existence of positive task correlations is a starting point for construct validation, not its endpoint. Alternative item formats and independent outcome measures can help separate general decision skills from method-specific advantages.
Researchers should describe whether participants have previously encountered famous examples. Repeated exposure to classic framing or confidence questions can alter performance, particularly when answers are learned. Item leakage makes a test less informative about flexible application to novel problems.
Normative ambiguity and preference-sensitive choices
Some tasks have unambiguous mathematical standards; others hinge on assumed goals or values. Probability monotonicity is a clear coherence requirement for nested events. By contrast, what counts as a good choice between uncertain career alternatives depends on a person's priorities and constraints. Assessments should avoid scoring disagreement with the investigator's preferences as an error. In sunk-cost problems, a prior investment matters only through future consequences if these genuinely arise; scenario writers must rule them out to justify a simple criterion.
Sample selection, statistical power, and replication
University samples often have restricted variation in educational attainment, numerical skills, and socioeconomic circumstances. Such range restriction may distort observed relationships. Small samples can create unstable factor solutions, particularly when each component contains relatively few items. Multiple outcomes and unreported analyses increase the danger of selective findings. Replication requires transparent scoring, codebooks, sample characteristics, missing-data rules, preregistration when appropriate, and reporting of null results. Appelt et al. (2011).
A systematic interpretation separates established evidence for measurable individual differences from more preliminary claims that any single test can forecast behavior reliably across diverse decisions. The 2005, 2007, and 2018 validation programs provide meaningful convergent evidence, but practical deployment should be justified with validation in the actual target population rather than citation alone.
Measuring Improvement and Debiasing
Practice, instruction, and genuine transfer
An intervention may improve a task score because the learner memorized its structure, became familiar with the wording, or developed a more general strategy. These mechanisms have different implications. A strong evaluation includes novel items, comparison conditions, delayed follow-up, and ideally real-world outcomes. Gains on the exact exercises used in training establish near transfer. Applying a principle to unfamiliar decisions in a different domain establishes a more demanding form of transfer.
Experimental work has reported improvements following brief debiasing training in selected judgment tasks, including longitudinal follow-up in particular interventions. These findings support the possibility of teachable skills under defined conditions. They do not demonstrate that every debiasing course yields durable changes across real-life decisions. Morewedge et al. (2015). A deeper review of the intervention literature belongs to the separately published article Cognitive Debiasing Strategies: Which Interventions Work?.
Scaffolds for better decisions
A practical scaffold can make essential information easier to compare: specify the decision, list feasible alternatives, distinguish known facts from forecasts, state the relevant values, identify hard constraints, and record the decision rule. For probabilistic judgments, use explicit reference classes, calibrated numerical expressions, and feedback from resolved cases. For sunk-cost questions, compare future consequences while documenting any genuine forward-looking effects of past commitments.
Such procedures reduce unnecessary cognitive load and make disagreements auditable. Their benefit depends on user comprehension, implementation, and the surrounding institution. A complex checklist can add burden without improving decisions when task demands are low; a good aid should be tested against an appropriate comparator. Bruine de Bruin et al. (2020).
Can decision-making competence itself be trained?
Research on component interventions and transferable judgment habits provides a plausible foundation, but evidence for durable improvement in an entire competence composite is more limited than evidence for improvement in specific taught tasks. A-DMC is primarily an assessment framework; it should not be presented as an established clinical treatment or guaranteed personal-improvement program. Randomized studies with blinded or objective outcomes, active comparison groups, and long follow-up are required before broad causal claims can be made.
Applications in Education, Organizations, and the Artificial Era
In education, competence assessment can reveal which aspects of judgment need explicit teaching: understanding probabilities, applying conditional rules, distinguishing evidence from presentation, or recognizing the limits of knowledge. Its use should be formative and accompanied by suitable validation. Low subscale scores may indicate insufficient familiarity with task materials rather than an immutable deficiency.
In organizations, task-specific exercises may complement experience-based evaluations when selecting and training for complex roles. High-stakes personnel decisions should not rely on a research battery without strong predictive, fairness, and contextual evidence. Good governance records the decision criterion, evaluates the quality of available information, and separates the decision process from outcome luck.
Human interaction with automated recommendations adds another assessment problem: can people determine when to rely on a tool, verify an uncertain claim, and override recommendations that conflict with independently justified evidence? These are empirical questions about human judgment and decision support. A-DMC can inform hypotheses, but a general competence score does not directly measure calibration of trust in a particular AI system. Evaluation should include task-specific accuracy, reliability of recommendations, user expertise, transparency, and consequences of errors. The published research on AI as Authority: Trust, Expertise, Automation Bias, and Human Decision-Making addresses the related technology-centered intent.
The Artificial Era expands the contexts in which judgment is exercised, including settings where humans evaluate machine-produced explanations, forecasts, and recommendations. The relevant psychological commitment is operational: define what judgment skill is being measured, specify a defensible benchmark, test performance on independent cases, and track errors over time. Philosophical accounts of artificial rationality and human-machine history address a different explanatory level from the empirical validation of human competence instruments.
A Responsible Assessment Protocol
A sound research or applied protocol begins by identifying the target construct and decision domain. Investigators then specify the normative standards for each task, the population for which the instrument was validated, the version and language of the battery, the scoring method, and the comparison sample. They distinguish process measures from outcomes, test the reliability of component scores, and check the degree to which the measure adds useful information beyond simpler predictors.
The next stage investigates generalization. Use independent tasks and, where feasible, behavioral outcomes that are collected prospectively. Report uncertainty intervals, missingness, subgroup performance, floor and ceiling effects, and evidence for measurement invariance. If a training intervention is involved, compare it with meaningful alternatives and evaluate whether participants apply the learned principles after delays and in novel circumstances.
Finally, interpret each result in proportion to what the evidence warrants. A strong A-DMC score supports an inference about performance across the included research tasks in an appropriate reference population. It does not authorize a sweeping inference about moral character, legal capacity, competence at every job, or future life outcomes. Decision quality is an interaction among skills, opportunities, values, information, constraints, and stochastic outcomes.
Frequently Asked Questions
What does a decision-making competence test actually measure?
It measures how people perform on selected behavioral judgment and choice tasks against declared standards, such as probability consistency, following a rule, calibration of confidence, or resistance to irrelevant framing. The exact content depends on the instrument and version. A composite summarizes correlated skills but does not replace the individual subscale results.
Is the Adult Decision-Making Competence test an IQ test?
No. Intelligence, executive functioning, literacy, and numeracy can influence performance, and competence scores are correlated with some cognitive measures. The A-DMC battery also tests particular judgment principles, sensitivity to framing, calibration, and value-sensitive decision processes. Incremental associations with external outcomes have been reported after accounting for selected ability measures, although this does not establish complete independence. Bruine de Bruin et al. (2020).
Is the battery composed of six tasks or seven?
The original research and development examined seven task types, including path independence. The standard adult battery commonly described in subsequent applications uses six components: applying decision rules, resistance to framing, under/overconfidence, consistency in risk perception, recognizing social norms, and resistance to sunk costs. Researchers must report the version actually administered.
Does a poor score prove a person makes bad decisions in real life?
No. Associations with reported adverse outcomes support limited external validity at the group level. Outcomes depend on exposure, opportunity, material resources, circumstance, and luck. A score should never be converted into a categorical prediction of one person's life or into a clinical diagnosis. Parker et al. (2015).
Can people improve their scores?
Instruction, experience, practice, feedback, and decision aids can improve performance on particular tasks. Demonstrating lasting improvement in broad competence requires independent tasks and valid follow-up, because practicing familiar questions is not the same as transferring judgment skills. Debiasing studies provide promising specific results alongside unresolved questions about generalization. Morewedge et al. (2015).
Why do older adults sometimes outperform younger adults?
Different tasks depend on different resources. Fluid reasoning demands can make some problems harder with age, while experience and familiarity with relevant trade-offs can help with other tasks. Empirical age patterns vary by component and sample. Bruine de Bruin et al. (2012).
How does decision competence differ from cognitive reflection?
Cognitive reflection tests focus on whether respondents detect and reconsider compelling initial answers on a small set of problems. A-DMC covers several distinct judgment and choice principles. The measures may correlate, but neither is interchangeable with the other, and their reliability and validity must be assessed separately.
Can this score be used to decide whether someone is medically capable of consent?
No. Clinical decision-making capacity is a decision-specific ethical and legal question assessed by appropriate professionals and established standards. Research batteries measuring cognitive biases and judgment principles do not replace such evaluations.
Does competence mean always choosing the option with the highest expected value?
Expected-value maximization is one normative criterion when probabilities and utilities are specified and the model's assumptions fit the problem. Many real decisions involve ambiguous evidence, personal goals, risk preferences, fairness obligations, or constraints that make simple monetary expected value inadequate. Competence requires applying an appropriate declared criterion, not assuming every person's values are identical.
Related Articles
Cognitive Debiasing Strategies: Which Interventions Work? — The evidence on training, feedback, and procedures intended to reduce systematic decision errors.
Probability Forecasting and Calibration: How to Judge Uncertainty — Measuring the correspondence between numerical confidence and realized events.
Overconfidence Bias: Confidence, Accuracy, and Calibration — Specific patterns of excessive confidence and the standards used to assess them.
Judgment and Decision-Making in Psychology: Models and Processes — The conceptual foundations of judgment, preferences, and choice.
Base Rate Neglect: Why Prior Probabilities Matter — A specialized discussion of prior probabilities in uncertain judgments.
AI as Authority: Trust, Expertise, Automation Bias, and Human Decision-Making — Human evaluation of recommendations from artificial systems.
References
Appelt, K. C., Milch, K. F., Handgraaf, M. J. J., and Weber, E. U. (2011). The Decision Making Individual Differences Inventory and guidelines for the study of individual differences in judgment and decision-making research. Judgment and Decision Making, 6(3), 252–262. https://doi.org/10.1017/S1930297500001455
Bavolar, J. (2013). Validation of the Adult Decision-Making Competence in Slovak students. Judgment and Decision Making, 8(3), 386–392. https://doi.org/10.1017/S1930297500006057
Bruine de Bruin, W., Parker, A. M., and Fischhoff, B. (2007). Individual differences in adult decision-making competence. Journal of Personality and Social Psychology, 92(5), 938–956. https://doi.org/10.1037/0022-3514.92.5.938
Bruine de Bruin, W., Parker, A. M., and Fischhoff, B. (2012). Explaining adult age differences in decision-making competence. Journal of Behavioral Decision Making, 25, 352–360. https://doi.org/10.1002/bdm.712
Bruine de Bruin, W., Parker, A. M., and Fischhoff, B. (2020). Decision-making competence: More than intelligence? Current Directions in Psychological Science, 29(2), 186–192. https://doi.org/10.1177/0963721420901592
Morewedge, C. K., Yoon, H., Scopelliti, I., Symborski, C. W., Korris, J. H., and Kassam, K. S. (2015). Debiasing decisions: Improved decision making with a single training intervention. Policy Insights from the Behavioral and Brain Sciences, 2(1), 129–140. https://doi.org/10.1177/2372732215600886
Parker, A. M., Bruine de Bruin, W., and Fischhoff, B. (2015). Negative decision outcomes are more common among people with lower decision-making competence: An item-level analysis of the Decision Outcome Inventory (DOI). Frontiers in Psychology, 6, 363. https://doi.org/10.3389/fpsyg.2015.00363
Parker, A. M., Bruine de Bruin, W., Fischhoff, B., and Weller, J. (2018). Robustness of decision-making competence: Evidence from two measures and an 11-year longitudinal study. Journal of Behavioral Decision Making, 31(3), 380–391. https://doi.org/10.1002/bdm.2059
Parker, A. M., and Fischhoff, B. (2005). Decision-making competence: External validation through an individual-differences approach. Journal of Behavioral Decision Making, 18(1), 1–27. https://doi.org/10.1002/bdm.481
Reyna, V. F., and Brainerd, C. J. (2023). Numeracy, gist, literal thinking and the value of nothing in decision making. Nature Reviews Psychology, 2, 421–439. https://doi.org/10.1038/s44159-023-00188-7
Román, F. J., Colom, R., Hillman, C. H., Kramer, A. F., Cohen, N. J., and Barbey, A. K. (2019). Cognitive and neural architecture of decision making competence. NeuroImage, 199, 172–183. https://doi.org/10.1016/j.neuroimage.2019.05.076
Weller, J. A., Levin, I. P., Rose, J. P., and Bossard, E. (2012). Assessment of decision-making competence in preadolescence. Journal of Behavioral Decision Making, 25, 414–426. https://doi.org/10.1002/bdm.744