EPPP: PART 2, SKILLS • DOMAIN 2: ASSESSMENT AND INTERVENTION

Data Reconciliation — Reconcile discrepancies across assessment sources

Integrating conflicting clinical data to form coherent, accurate psychological formulations.

Historical Context & Motivation

The challenge of reconciling inconsistent information across assessment sources has been a fundamental concern in clinical psychology since the field's earliest attempts at systematic evaluation. Throughout much of the nineteenth century, psychological assessment relied almost exclusively on a single clinician's interview impressions, which meant that discrepancies between data sources were rarely encountered—because only one source existed. As the discipline matured, however, practitioners began supplementing clinical interviews with standardized tests, behavioral observations, collateral reports, and physiological measures, each of which introduced its own perspective on the client's functioning and, inevitably, its own potential for disagreement with other sources.

The concept of data reconciliation emerged from this growing awareness that no single assessment instrument or informant captures the full picture of a person's psychological experience. When a self-report measure suggests moderate depression but a behavioral observation reveals minimal psychomotor retardation, or when a parent describes severe conduct problems that a teacher does not observe, the clinician must determine how to integrate—and sometimes prioritize—these competing narratives. This section traces the historical milestones that shaped our modern approach to managing discrepant assessment data.

1905
Binet-Simon Scale Introduced
Alfred Binet and Théodore Simon develop the first standardized intelligence test, establishing the principle that psychological attributes can be measured objectively—separate from clinician impression alone.
1943
MMPI Published
The Minnesota Multiphasic Personality Inventory introduces validity scales that detect response distortion, highlighting for the first time that self-report data may systematically diverge from clinical observation.
1959
Campbell & Fiske's Multi-Trait Multi-Method Matrix
Donald Campbell and Donald Fiske propose the MTMM framework, formalizing the expectation that the same construct measured by different methods will show convergent validity—and that discrepancies signal method variance.
1991
Achenbach's Cross-Informant Syndromes
Thomas Achenbach systematically documents low-to-moderate cross-informant agreement in child and adolescent assessment, demonstrating that informant discrepancies are the rule rather than the exception.
2013
DSM-5 and Multi-Source Assessment
The DSM-5 cross-cutting symptom measures formalize multi-informant assessment, embedding data reconciliation as a core clinical competency in contemporary diagnostic practice.

The central question this history raises is both practical and epistemological: when multiple sources of clinical data disagree, how should a practitioner determine which data most accurately reflect the client's true psychological state? Addressing this question requires an understanding of the sources of discrepancy, systematic frameworks for evaluating data quality, and clinical judgment grounded in psychometric principles.

Core Principles of Data Reconciliation

Effective data reconciliation rests on several foundational principles that guide clinicians in making sense of conflicting information. These principles draw from psychometric theory, systems thinking, and evidence-based clinical reasoning. Understanding them transforms the experience of encountering discrepant data from a confusing obstacle into a clinically informative process that deepens the overall assessment.

1

Multi-Method Assessment

No single method (interview, test, observation, collateral report) is inherently superior. Each method samples different aspects of functioning and introduces unique sources of error. Using multiple methods provides converging evidence and reveals clinically meaningful divergences.
2

Informant Variance as Signal

Discrepancies between informants are not merely noise—they often carry diagnostic meaning. A child who behaves differently at home versus school may exhibit context-dependent symptoms, which itself is a clinically relevant finding.
3

Psychometric Contextualization

Each data source must be evaluated in terms of its reliability, validity, base rates, and known biases. A measure with strong test-retest reliability and validated norms may receive greater weight than an unstructured clinical impression.
4

Hierarchical Integration

Clinicians apply a systematic hierarchy: convergent data are accepted with higher confidence, divergent data are explored for explanations, and contradictory data trigger hypothesis testing to determine which source best captures the construct of interest.
5

Cultural & Contextual Sensitivity

Discrepancies may arise from cultural differences in symptom expression, communication styles, or differing norms. Reconciliation must account for the sociocultural context in which each data source was obtained.
KEY TAKEAWAY
Think of data reconciliation like being a detective reviewing witness statements after an event. Each witness saw the event from a different angle, through their own perceptual lens, and under different conditions. A skilled detective does not simply discard conflicting accounts—rather, they ask why the accounts differ and uses that understanding to reconstruct a more complete and accurate picture of what actually happened. Similarly, the clinician's task is not to eliminate disagreement but to interpret it meaningfully.

Visual Model of Data Reconciliation

The following diagram illustrates the multi-source assessment convergence model, showing how different data streams feed into a reconciliation process. At the center, the clinician evaluates areas of convergence and divergence, weighing each source according to its psychometric properties and contextual relevance before arriving at an integrated clinical formulation.

The diagram shows four assessment sources (left) feeding into a central reconciliation engine, which outputs three categories: convergent (sources agree, high confidence), divergent (context-specific variation), and contradictory (requiring further investigation).

As the diagram illustrates, the reconciliation process is not merely additive. The clinician does not simply average across sources. Instead, each finding is categorized based on whether it converges with, diverges from, or directly contradicts other sources. Convergent findings across multiple methods represent the most reliable clinical conclusions. Divergent findings often reveal important context-dependent variation—for instance, a child who exhibits oppositional behavior only at home but not at school. Contradictory findings demand careful hypothesis testing: the clinician must consider response bias, measure validity, situational factors, and diagnostic alternatives before resolving the inconsistency.

How Discrepancies Arise: Sources of Variance

To reconcile discrepancies effectively, one must first understand the mechanisms that produce them. In classical test theory, any observed score on a psychological measure is a composite of the true score and error. When we extend this framework across multiple methods and informants, we recognize that different sources introduce systematically different types of error, which accounts for much of the discrepancy clinicians encounter in practice.

CLASSICAL TEST THEORY
X = T + E
Where X = observed score, T = true score (the construct of interest), E = measurement error. Different assessment methods produce different E values, leading to discrepant X values even when T remains constant.

Sources of Discrepancy Variance

The method variance component, first articulated in the multi-trait multi-method matrix framework, refers to systematic error attributable to the assessment method itself rather than the construct being measured. For example, self-report inventories are subject to social desirability bias, denial, and limited insight, while projective techniques may introduce examiner interpretation bias. Structured interviews improve inter-rater reliability but may miss nuances captured in free-form clinical dialogue.

Informant variance represents another major source of discrepancy. Achenbach, McConaughy, and Howell (1987) conducted a landmark meta-analysis demonstrating that the average correlation between different informants rating the same child's behavior was only r = .28. This low agreement does not necessarily indicate that one informant is wrong; rather, it reflects the reality that different informants observe the individual in different settings, at different times, and through different relational lenses. Parents observe the child at home during evenings and weekends, teachers observe in structured academic environments, and the child's own self-report captures internal experiences inaccessible to external observers.

ACHENBACH META-ANALYTIC FINDINGS
r̄ (same informant type) ≈ .60 | r̄ (different informant type) ≈ .28
Mean correlations from Achenbach et al. (1987). Same-type informants (e.g., mother–father) agree moderately, while cross-type informants (e.g., parent–teacher) show only modest agreement. This pattern is consistent across behavioral, emotional, and social domains.

Additional sources of discrepancy include temporal variance (symptoms fluctuate over time, so assessments administered days or weeks apart may capture different clinical states), setting variance (behavior is context-dependent), and construct underrepresentation (a measure may tap only a narrow facet of a multidimensional construct). Understanding these variance sources is essential for determining whether a discrepancy reflects genuine clinical complexity or artifactual measurement noise.

A Decision Framework for Reconciliation

When faced with discrepant assessment data, clinicians benefit from a structured decision framework that guides the reconciliation process. The following diagram presents a decision tree that practitioners can use to systematically evaluate and resolve inconsistencies. This framework integrates psychometric considerations, contextual factors, and clinical reasoning into a stepwise process that moves from detection of the discrepancy through to its resolution.

A five-step decision framework for resolving discrepancies: evaluate psychometric quality, check validity indicators, assess context, consider cultural factors, and generate hypotheses—culminating in an integrated formulation.

At Step 1, the clinician evaluates the psychometric properties of each discrepant source—internal consistency, test-retest reliability, criterion validity, and normative adequacy. A source with poor psychometric properties receives less weight in the reconciliation. At Step 2, the clinician examines validity indicators embedded within specific measures—for instance, the MMPI-3 validity scales (F, L, K, VRIN, TRIN) may reveal overreporting, underreporting, or random responding that explains the discrepancy. At Step 3, contextual and setting differences are assessed: the same client may genuinely function differently across environments, making the discrepancy itself a valid clinical finding. Step 4 integrates cultural and developmental considerations, recognizing that cultural norms around emotional expression or developmental stage may systematically affect certain assessment modalities more than others. Finally, at Step 5, the clinician generates specific clinical hypotheses to explain the discrepancy and seeks additional data to test them, ultimately arriving at an integrated clinical formulation.

Worked Example: Reconciling Multi-Informant Data

Consider the following clinical scenario, which demonstrates the reconciliation framework in practice. This example involves a 10-year-old boy, Marcus, referred for evaluation due to academic underperformance and behavioral concerns.

📋 CLINICAL SCENARIO
Marcus, age 10, is referred by his school for a comprehensive psychological evaluation. His teacher reports significant inattention, disorganization, and difficulty completing assignments. His mother describes him as "moody and withdrawn" at home, with frequent tearfulness and low energy. Marcus's self-report (CDI-2) yields a T-score of 45 (within normal limits for depression), and his CBCL (completed by mother) yields clinically elevated Internalizing scores. The TRF (teacher report) yields clinically elevated Attention Problems and Externalizing scores, with Internalizing within normal limits.
Reconciling Marcus's Discrepant Assessment Data
1
Step 1 — Evaluate Psychometric QualityAll instruments used (CDI-2, CBCL, TRF) have strong psychometric properties with well-established reliability and validity. The CBCL and TRF were normed on large representative samples, and the CDI-2 has adequate internal consistency (α ≈ .86) for the age group. No instrument is dismissed on psychometric grounds.
All three sources pass psychometric scrutiny—the discrepancy is not attributable to poor measurement.
2
Step 2 — Check Validity IndicatorsThe CDI-2 includes an inconsistency index; Marcus's responses show no random or contradictory patterns. However, at age 10, children's capacity for introspective self-report on internalizing symptoms is still developing. Marcus may lack the vocabulary or insight to endorse items capturing his emotional experience accurately. The mother's and teacher's reports show no evidence of exaggeration or minimization patterns.
Marcus's self-report may underrepresent internalizing symptoms due to developmental limitations in self-awareness, not due to deliberate minimization.
3
Step 3 — Assess Contextual DifferencesThe mother observes Marcus primarily during evenings and weekends—times when fatigue and emotional vulnerability may be heightened. The teacher observes during structured academic tasks, where inattention and behavioral disruption are most salient. This context difference explains why the mother reports internalizing symptoms (moodiness, tearfulness at home) while the teacher reports externalizing and attentional problems (disorganization, task avoidance at school).
The discrepancy is partly explained by setting variance: Marcus's presenting problems manifest differently across home and school contexts.
4
Step 4 — Consider Cultural and Developmental FactorsMarcus's family cultural background emphasizes emotional stoicism for boys, which may lead him to underreport sadness and withdrawal on the self-report measure. Developmentally, 10-year-olds often externalize emotional distress through behavioral disruption rather than verbalizing it, which aligns with the teacher's observation of conduct problems that may, at their root, be emotionally driven.
Cultural norms around male emotional expression and developmental stage support the hypothesis that Marcus's externalizing behaviors at school may represent expressed internalizing distress.
5
Step 5 — Generate and Test Hypotheses → Integrated FormulationThe primary hypothesis is that Marcus is experiencing a depressive episode that manifests as withdrawal and tearfulness at home (observed by mother) and as inattention, irritability, and behavioral disruption at school (observed by teacher). His self-report underestimates severity due to limited insight and cultural influences. This hypothesis is consistent with DSM-5 criteria for Major Depressive Disorder in children, which recognizes irritability as a mood criterion and concentration difficulties as a cognitive symptom. The clinician conducts a follow-up semi-structured interview (K-SADS) with both Marcus and his mother to test this hypothesis, which confirms threshold depressive symptoms across domains.
Integrated Formulation: Marcus presents with a depressive disorder that manifests across settings with context-appropriate symptom expression—internalizing at home, externalizing at school. The apparent discrepancy is diagnostically informative rather than contradictory.

Reconciliation Strategies: Strengths and Limitations

Clinicians employ several distinct strategies when reconciling discrepant data, and each carries its own advantages and potential pitfalls. The table below summarizes the major reconciliation strategies, their appropriate applications, and their limitations. Awareness of these trade-offs helps practitioners select the most appropriate approach for each clinical situation.

Comparison of five common strategies for reconciling multi-informant assessment data
StrategyDescription & When to UseLimitations
"OR" Rule (Inclusive)A symptom is considered present if endorsed by any informant. Maximizes sensitivity; appropriate when missing a true condition is more costly than a false positive (e.g., suicidality screening).Inflates false positive rates; may lead to over-diagnosis. Not appropriate when specificity is prioritized.
"AND" Rule (Restrictive)A symptom is considered present only if endorsed by all (or most) informants. Maximizes specificity; appropriate when false positives carry high costs (e.g., involuntary commitment).Inflates false negative rates; may miss genuine conditions observable in only one context. Underestimates prevalence.
Optimal InformantWeight is given to the informant best positioned to observe the construct. Parents are prioritized for externalizing behaviors in young children; self-report is prioritized for internalizing symptoms in adolescents and adults.Requires prior knowledge of which informant is "optimal" for each construct. Evidence base is incomplete for many clinical populations and diagnoses.
Clinical ConsensusThe clinician uses professional judgment to integrate all data qualitatively, weighting sources based on the specific case context. Most flexible and individualized approach.Subject to cognitive biases (anchoring, confirmatory bias). Reliability depends heavily on clinician expertise. May be difficult to replicate or justify.
Statistical AggregationScores from multiple sources are combined using algorithms (e.g., averaging T-scores, factor analysis across informants). Used in research settings and large-scale screening programs.May obscure clinically meaningful discrepancies by averaging them away. Assumes commensurability across measurement scales. Rarely used in individual clinical practice.
KEY TAKEAWAY
No single reconciliation strategy is universally correct. The appropriate strategy depends on the clinical question, the stakes of the decision, and the specific informants involved. Think of it like choosing a camera lens: a wide-angle lens (the OR rule) captures everything in view but with less precision, while a telephoto lens (the AND rule) zooms in with high specificity but may miss important peripheral information. The skilled clinician selects the "lens" that best serves the clinical purpose at hand.

Connection to Advanced Clinical Integration

Data reconciliation as practiced in routine assessment is the foundation for more advanced integrative practices in clinical psychology. As clinicians develop expertise, they move from the structured, stepwise reconciliation described above toward a more fluid, pattern-recognition-based approach that draws on extensive clinical experience and deep knowledge of psychopathology. The table below contrasts the foundational reconciliation skills covered in this lesson with the advanced integrative competencies that emerge with clinical maturation.

Progression from foundational reconciliation skills to advanced clinical integration
Foundational ReconciliationAdvanced Clinical Integration
Follows a structured, stepwise decision frameworkIntegrates data fluidly using pattern recognition and expert heuristics
Resolves discrepancies between 2–4 sourcesSynthesizes data across dozens of variables, including longitudinal patterns
Focuses on whether data converge, diverge, or contradictFocuses on generating and testing complex biopsychosocial formulations
Uses known psychometric properties to weight sourcesIntegrates psychometric data with theoretical models (e.g., attachment theory, cognitive schemas) to generate individualized case conceptualizations
May apply a single reconciliation rule (OR, AND, optimal informant)Dynamically shifts reconciliation strategy based on evolving clinical hypotheses across the assessment process

Looking ahead, advances in computational modeling and machine learning are beginning to offer algorithmic approaches to data reconciliation that complement clinical judgment. Latent variable models, for instance, can estimate a latent trait (the "true" construct) from multiple fallible indicators, statistically modeling the measurement error associated with each source. Bayesian approaches allow clinicians to update their diagnostic confidence as each new data source is incorporated, providing a formal framework for the intuitive process of hypothesis revision. While these methods are currently more common in research than in clinical practice, they represent the direction in which evidence-based assessment is moving and underscore the importance of the foundational reconciliation competencies covered in this lesson.

Practice Problems

PROBLEM 1CONCEPTUAL
A clinician administers the BDI-II to a 35-year-old client who endorses minimal depressive symptoms (total score = 8). However, during the clinical interview, the client displays psychomotor retardation, flat affect, and tearfulness. Explain why these two data sources might produce discrepant results and identify which source of variance is most likely operating.
PROBLEM 2BASIC APPLICATION
A school psychologist receives a CBCL from a mother (Internalizing T = 72, clinically significant) and a TRF from a teacher (Internalizing T = 48, within normal limits) for a 7-year-old girl. Based on Achenbach's meta-analytic findings, is this degree of discrepancy expected or unusual? Justify your answer using the cross-informant correlation data.
PROBLEM 3INTERMEDIATE
A psychologist is evaluating a 16-year-old for ADHD. The following data are obtained: (a) Conners-3 Parent Rating Scale: clinically elevated Inattention; (b) Conners-3 Teacher Rating Scale: within normal limits; (c) CPT-3 (continuous performance test): clinically impaired sustained attention; (d) Self-report: endorses significant concentration difficulties. Apply the five-step reconciliation framework to integrate these data and develop a formulation.
PROBLEM 4APPLIED
You are conducting a forensic evaluation of a 42-year-old man involved in a personal injury lawsuit following a motor vehicle accident. He reports severe PTSD symptoms on the PCL-5 (total score = 68). His MMPI-3 profile shows elevated F and Fp scales, suggesting symptom overreporting. Behavioral observations during the evaluation reveal appropriate emotional reactivity and no avoidance behavior when discussing the accident. His treating therapist's notes document consistent reports of nightmares and hypervigilance over the past six months. How do you reconcile these data? Which reconciliation strategy would you apply and why?
PROBLEM 5CRITICAL THINKING
Critically evaluate the following claim: "Because cross-informant agreement is generally low (r ≈ .28), clinicians should rely primarily on self-report data, as individuals are the best experts on their own psychological experiences." Under what conditions is this claim defensible, and under what conditions does it fail? In your analysis, address at least three specific clinical scenarios where privileging self-report would lead to suboptimal assessment outcomes.

Summary: Data Reconciliation Across Assessment Sources

Data reconciliation is the clinical competency of systematically integrating information from multiple assessment sources—including self-report measures, clinical interviews, collateral reports, and behavioral observations—when those sources produce discrepant findings. Rooted in classical test theory and the multi-trait multi-method framework, effective reconciliation requires clinicians to understand that discrepancies arise from method variance, informant variance, temporal fluctuations, setting differences, and cultural factors—and that these discrepancies are often clinically informative rather than mere noise.

The five-step decision framework guides practitioners through evaluating psychometric quality, checking validity indicators, assessing contextual differences, considering cultural and developmental factors, and generating testable clinical hypotheses. Multiple reconciliation strategies exist—including the OR rule, AND rule, optimal informant, and clinical consensus—each suited to different clinical contexts and decision stakes. Mastering data reconciliation is essential for accurate diagnosis, effective treatment planning, and competent practice across all domains of behavioral health.

Varsity Tutors • EPPP: Part 2, Skills • Data Reconciliation — Reconcile discrepancies across assessment sources