1. Purpose and scope of psychological assessment
Psychological assessment is the systematic collection, integration and interpretation of information about a person's psychological functioning, abilities, behaviour, experiences and context. It is undertaken to answer a defined question, not simply to administer a test. Questions may concern learning needs, cognitive strengths, emotional distress, personality patterns, career interests, rehabilitation, treatment planning or research. Assessment is a process that combines professional judgement with evidence from multiple sources. It should identify strengths and resources as well as difficulties, and it should lead to decisions that are understandable and useful to the person concerned.
Assessment differs from testing. A psychological test is one instrument or procedure within a wider assessment. A complete assessment may include interviews, observation, records, informant reports, behavioural tasks, rating scales and standardised tests. The choice depends on the referral question, age, language, disability access, culture and practical setting. A test score without context can be misleading. For example, a low score may reflect unfamiliar language, fatigue, anxiety, poor vision, interrupted schooling or an unsuitable norm group rather than the ability the test intends to measure.
Assessment should be purposeful, proportionate and collaborative. Before beginning, the practitioner clarifies who requested the assessment, who will receive the report, what decisions may follow and what limits apply to confidentiality. The person should understand the procedures and have opportunities to ask questions. When a person cannot provide full legal consent, applicable law and professional standards guide consent from an authorised representative while seeking the person's assent and participation as far as possible. The assessor must remain within professional competence and refer when the question requires expertise they do not possess.
2. The assessment cycle and referral questions
A well-designed assessment follows a sequence. First, clarify the referral question and the decision that the findings are expected to inform. Second, identify the constructs that must be examined and the evidence needed. Third, select methods that are valid for the intended purpose and appropriate for the individual. Fourth, collect information under conditions that are as fair and consistent as possible. Fifth, interpret findings in relation to history, context and measurement limitations. Finally, communicate conclusions, recommendations and uncertainty in accessible language and review whether the assessment achieved its purpose.
Referral questions should be specific and neutral. 'Why is this learner lazy?' embeds an unsupported judgement. A better question is: 'What factors may explain the learner's persistent difficulty completing reading tasks, and what supports may improve access and progress?' The revised question invites investigation of reading skills, attention, language exposure, instruction, sensory access, emotional wellbeing and environmental conditions. It also avoids treating the learner as the problem. Similar care is needed in workplace, clinical and forensic settings, where labels can influence opportunities and legal decisions.
The assessor should distinguish the question asked by the referrer from the questions important to the person being assessed. A school may ask about classroom performance while a learner may be concerned about stigma or privacy. An employer may request a fitness assessment while the employee may need reasonable adjustments. Ethical practice makes the purpose transparent, explores competing interests and explains what can and cannot be concluded. If the referral question is outside the scope of the methods available, the practitioner should narrow it, obtain further information or decline the assessment.
3. Psychological constructs and operational definitions
A construct is a theoretical attribute that cannot usually be observed directly, such as intelligence, anxiety, working memory, motivation, resilience or self-esteem. Researchers and practitioners infer constructs from patterns of responses and behaviour. Because constructs are abstract, they must be operationalised: the assessor specifies the observable indicators and procedures used to represent them. Anxiety, for example, might be examined through self-reported worry, physiological arousal, avoidance and observed behaviour in relevant situations. Each indicator captures part of the construct rather than the whole of it.
Operational definitions influence what a test measures and what conclusions can be drawn. A timed vocabulary test may measure aspects of acquired verbal knowledge and speed under particular conditions; it cannot represent every dimension of intelligence. A self-report stress scale measures the person's responses to selected statements during a defined period; it may not capture all contextual sources of stress. Construct underrepresentation occurs when important parts of a construct are omitted. Construct-irrelevant variance occurs when scores are influenced by factors unrelated to the intended attribute, such as confusing wording or inaccessible response formats.
Good assessment therefore asks whether the construct is clearly defined, whether the indicators represent it adequately and whether alternative explanations have been considered. Constructs may overlap. A person who reports poor concentration may be experiencing sleep loss, pain, anxiety, depression, medication effects or an attention-related developmental condition. A single measure rarely separates these possibilities. Multiple methods, developmental history and context help determine what a score means and prevent an instrument from being treated as a direct window into a person's mind.
4. Principles of psychometrics
Psychometrics is the science of psychological measurement. It studies how tests are constructed, scored, evaluated and interpreted. Core concepts include reliability, validity, standardisation, norms, measurement error, fairness and utility. Psychometric quality is not a permanent property that applies equally to every use. Evidence must support the interpretation of scores for a particular purpose and population. A test suitable for screening may not be sufficiently precise for diagnosis, placement or high-stakes selection.
Measurement is imperfect. A person's observed score reflects the underlying attribute, temporary influences and measurement error. Fatigue, distraction, illness, motivation, anxiety, examiner behaviour and environmental conditions can influence performance. The standard error of measurement expresses the expected uncertainty around an observed score under a measurement model. Consequently, scores should usually be interpreted as estimates with uncertainty rather than exact quantities. Small score differences may not be meaningful, especially when confidence intervals overlap.
Psychometric evidence should be evaluated rather than accepted because a test is popular or professionally branded. The assessor examines the quality of the development sample, the intended population, administration instructions, scoring procedures, reliability evidence, validity evidence, norm dates and accessibility. A measure may perform differently across languages, educational systems and cultural settings. Local validation, careful adaptation and transparent reporting are especially important when a test developed elsewhere is used in a new context.
5. Reliability: consistency and precision
Reliability concerns the consistency or precision of measurement. Test–retest reliability examines whether scores are reasonably stable when the same people complete a measure on two occasions, provided the construct is expected to remain stable. Internal consistency examines how closely items within a scale relate to one another. Inter-rater reliability concerns agreement between observers or scorers. Parallel-forms reliability compares scores from different versions designed to measure the same construct. Each method addresses a different source of inconsistency.
Internal consistency is often summarised using coefficients such as Cronbach's alpha or omega. A high coefficient does not prove that a scale is unidimensional or valid. It may be increased by adding many similar items, and it may be inappropriate for a measure intentionally covering distinct domains. Inter-rater agreement can be examined through percentage agreement or statistics such as kappa, selected according to the data and study design. Correlation alone may be insufficient because two raters can correlate strongly while systematically assigning different scores.
Reliability is affected by the range of ability in the sample, test length, item quality, administration conditions and scoring rules. A measure may be reliable in one population and less reliable in another. Practitioners should report the relevant reliability evidence and avoid overinterpreting tiny differences. Reliability is necessary for meaningful interpretation but is not sufficient: a bathroom scale that consistently adds five kilograms is reliable in a narrow sense but inaccurate. Validity evidence is also required.
6. Validity and evidence for score interpretation
Validity concerns the degree to which evidence and theory support the interpretation and use of test scores for a proposed purpose. Modern validity evaluation draws on several complementary sources. Evidence based on test content asks whether items adequately represent the construct. Evidence based on response processes examines whether respondents understand and answer items in the intended way. Evidence based on internal structure investigates whether the relationships among items fit the proposed dimensions. Relations with other variables examine expected associations and distinctions. Consequences of testing consider the effects of score use, including unintended harm.
Content validity is especially important for achievement and competency tests. A mathematics assessment should sample the taught domain in proportion to its learning objectives rather than overemphasising a narrow topic. Criterion-related evidence examines how scores relate to relevant outcomes, either concurrently or predictively. Predictive evidence may support selection or screening, but correlation does not guarantee that a test causes later performance. Construct validity requires a coherent pattern of evidence that the measure behaves as theory predicts and differs from constructs it should not simply duplicate.
Validity is not a single coefficient printed in a manual. It is an argument supported by evidence, theory and the intended use. A scale validated for research screening may not be appropriate for making an individual diagnosis. A test normed on adults may not be valid for children. A translated instrument may preserve literal wording while changing the meaning of items. Reports should specify the purpose, population and limitations for which the evidence applies, and should avoid claims that exceed the available support.
7. Standardisation, norms and score interpretation
Standardisation means that administration, instructions, timing, scoring and interpretation procedures are specified so that results can be compared meaningfully. Deviations may be necessary to provide access for a person with a disability, but the assessor must record them and consider how they affect comparability. A standardised procedure does not automatically make a test culturally fair; it makes the procedure more consistent. Appropriate interpretation still requires evidence that the construct and items function as intended for the person assessed.
Norm-referenced interpretation compares an individual's score with scores from a defined reference group. Criterion-referenced interpretation compares performance with a specified standard or skill domain. A percentile rank indicates the percentage of the norm group scoring at or below a particular point; it is not the percentage of questions answered correctly. Standard scores transform raw scores to a common scale. For example, a z score expresses distance from the norm-group mean in standard deviation units. A score of zero on a z scale is the group mean, not an absence of ability.
Norms depend on who was sampled, when data were collected and whether the sample represents the population of interest. Outdated norms may misrepresent current performance because education, technology, language and social conditions change. Local norms may be needed when the reference population differs substantially. Reports should name the comparison group and explain that a percentile is a relative position, not a judgement of personal worth. Norms describe distributions; they do not establish what an individual can achieve with support.
8. Measurement scales and data types
Measurement theory commonly distinguishes nominal, ordinal, interval and ratio scales. Nominal data classify observations into categories without an inherent order, such as preferred learning mode. Ordinal data have an order but unequal distances cannot be assumed, as with ranked preferences or many rating categories. Interval scales have equal units but no meaningful absolute zero under the measurement model. Ratio scales have equal units and a meaningful zero, allowing ratio comparisons. The distinctions influence which summaries and statistical analyses are appropriate.
Psychological variables are often measured using rating scales. A single Likert-type item is ordinal in a strict interpretation, while a multi-item scale may sometimes be treated as approximately continuous when evidence supports that approach. The decision should consider item structure, distribution, scale properties and analytical purpose. A numerical code assigned to a category does not turn it into a quantitative measurement: coding '1' for one group and '2' for another does not imply that the second group has twice the attribute.
Researchers should define variables and scoring rules before analysis. Reverse-scored items must be recoded correctly, missing data must be handled transparently and composite scores should follow the test manual or validated protocol. Data transformations should not be selected merely to obtain a preferred result. The scale of measurement guides summaries, visualisation, statistical tests and interpretation. Clear documentation makes analyses reproducible and reduces avoidable scoring errors.
9. Interview methods and clinical history
Interviews allow the assessor to explore experiences in the person's own words. Structured interviews use predetermined questions and rules, improving consistency. Semi-structured interviews combine core questions with follow-up prompts, allowing depth while retaining coverage. Unstructured interviews are more flexible but can vary substantially between assessors. The appropriate format depends on the purpose, setting, risk, time and need for comparability. Interviewers should explain the purpose, use accessible language and avoid leading questions.
An effective interview includes open questions, attentive listening, clarification and respectful summaries. Open questions invite narrative; focused questions establish details such as frequency, duration and context. Silence can give the person time to think. Interviewers should distinguish what the person reports from the assessor's interpretation. Sensitive questions are introduced with explanation and permission where possible. If distress arises, the assessor pauses, checks safety and follows appropriate procedures rather than pursuing information at any cost.
A clinical or developmental history may include presenting concerns, onset and course, health, sleep, medication, education, work, relationships, developmental milestones, family context, culture, language, strengths and previous support. Not every assessment requires every domain. Information should be proportionate to the referral question and collected with respect for privacy. Collateral information from relatives, teachers or colleagues can add context, but consent, relevance and possible conflicts of interest must be considered. Disagreement between sources is information to examine, not a reason to assume one person is dishonest.
10. Behavioural observation and functional analysis
Behavioural observation records actions in context using defined procedures. It may take place in natural settings, structured tasks or simulations. The observer specifies what behaviour is being counted or rated, when observation occurs and how records are made. Frequency, duration, latency, intensity and antecedent–behaviour–consequence sequences are common dimensions. Operational definitions improve agreement: 'leaves the seat without permission during a ten-minute task' is more observable than 'is disruptive'.
Functional assessment examines the relationship between antecedents, behaviour and consequences. An antecedent occurs before the behaviour; a consequence follows it and may influence whether it happens again. The same behaviour can serve different functions in different settings. A learner leaving a task may be escaping work that is too difficult, seeking social interaction, responding to sensory overload or communicating distress. Observation should therefore test hypotheses rather than assume motive from appearance.
Observation has limitations. People may change behaviour when they know they are being observed, observers may hold expectations, and brief samples may not represent typical functioning. Multiple observations across times and contexts can improve confidence. Inter-rater checks help identify inconsistent coding. Ethical observation requires appropriate consent or lawful authority, privacy protection and attention to power relationships. Observers should record context and avoid intrusive surveillance that is not justified by the assessment purpose.
11. Self-report inventories and rating scales
Self-report measures ask people to describe their own experiences, beliefs, symptoms, preferences or behaviour. They can efficiently collect information that may not be observable, such as worry, mood or perceived quality of life. Rating scales may use frequency, intensity or agreement categories. Clear instructions specify the reference period, response format and whether there are right or wrong answers. The assessor should check that the person can read and understand the items or provide an appropriate accessible format.
Self-report is influenced by insight, memory, literacy, language, social desirability, response style and the perceived consequences of disclosure. Some people minimise difficulties because of stigma or fear; others may overreport because they are distressed or interpret items differently. These possibilities should not be treated as evidence of deception without support. Confidentiality and the intended use of results should be explained, and the assessor should consider whether the measure is suitable for the person's culture and circumstances.
Screening instruments identify people who may benefit from further evaluation; they do not independently establish a diagnosis. A cut-off score balances sensitivity and specificity and will produce false positives and false negatives. Positive predictive value also depends on how common the condition is in the population being screened. A high score should prompt appropriate follow-up, not a label delivered without context. Repeated measures can monitor change, but the same limitations and measurement error remain relevant.
12. Cognitive and intelligence assessment
Cognitive assessment examines processes such as reasoning, memory, attention, language, processing speed, visuospatial ability and executive functioning. Intelligence tests sample selected cognitive tasks and compare performance with a norm group. Modern models recognise that cognitive ability is multidimensional and influenced by development, education, health, opportunity and context. A composite score can summarise performance, but it may conceal meaningful differences between domains. Interpretation should consider the pattern of scores and the person's everyday functioning.
An intelligence quotient or standard score is a norm-referenced estimate, not a complete measure of potential, creativity, wisdom, practical skill or human value. Scores are affected by test content, language, familiarity with formal testing, sensory or motor access and the quality of norms. A person who has had limited access to schooling may be disadvantaged by tasks that assume particular educational experiences. Fair assessment may require culturally and linguistically appropriate instruments, interpreters, accessible formats or alternative methods, with limitations clearly documented.
Neuropsychological assessment may examine memory, attention, language, perception, motor skills and executive functioning following neurological illness, injury or developmental concerns. It can contribute to diagnosis, rehabilitation planning and monitoring, but findings must be integrated with medical history, imaging where available and functional information. Practice effects can influence repeated testing, and fatigue or pain can affect performance. The assessor should avoid making strong conclusions from isolated low scores without considering validity, context and the broader profile.
13. Achievement, aptitude and educational assessment
Achievement tests assess knowledge or skills acquired through instruction, such as reading, mathematics, science or language proficiency. Aptitude tests estimate readiness or performance potential in a defined domain, although aptitude is shaped by opportunity and learning. Diagnostic educational assessment investigates the processes underlying a learning difficulty and helps identify suitable instruction. These purposes are distinct: a low achievement score may reflect inadequate teaching opportunity, language difference, absence, disability or a specific skill difficulty.
Curriculum alignment is essential. An achievement test should represent the content and competencies that learners were expected to learn. Item difficulty, discrimination, coverage and accessibility should be reviewed. Classroom assessment may combine tests, projects, oral responses, practical demonstrations and teacher observations. Multiple forms of evidence can give a fuller picture than one examination. Formative assessment is used to guide teaching while learning is occurring; summative assessment evaluates achievement at a defined point.
Educational decisions should be supportive and proportionate. Assessment can identify where a learner needs explicit instruction, additional time, assistive technology, language support or changes in task presentation. It should not be used to deny opportunity or assume a fixed ceiling. Feedback should describe skills demonstrated, areas for development and practical next steps. Learners and families should understand the findings and have a route to ask questions or request review.
14. Personality assessment and trait models
Personality assessment examines relatively enduring patterns of emotion, thought and behaviour. Trait models describe dimensions such as openness, conscientiousness, extraversion, agreeableness and neuroticism, while other frameworks examine narrower characteristics. Trait measures are generally most useful for describing patterns and generating hypotheses rather than making definitive claims about character. Personality varies across situations and can change with development, experience and major life events.
Self-report inventories are common, but results can be influenced by impression management, self-knowledge, item interpretation and cultural expectations. Informant reports may provide another perspective but can reflect relationship history and observer bias. Projective techniques use ambiguous stimuli to elicit responses; their scoring and interpretation require particular caution, and evidence varies substantially by method. No personality instrument should be treated as a device that reveals hidden truth without uncertainty.
Reports should describe measured dimensions and their likely relevance to the referral question. Avoid pejorative language and deterministic statements such as 'this person will always behave this way'. A trait score is not a moral judgement or a diagnosis. Where personality assessment informs recruitment, promotion or clinical decisions, fairness, job or treatment relevance, privacy and the consequences of error must be considered. Feedback should include strengths and contextual qualifications.
15. Neuropsychological assessment
Neuropsychological assessment studies the relationship between brain functioning and cognitive, emotional and behavioural performance. Referral reasons may include acquired brain injury, stroke, epilepsy, neurodegenerative conditions, developmental concerns or unexplained changes in memory and executive functioning. The assessment may sample attention, learning, memory, language, visuospatial processing, motor coordination, processing speed and executive control. Selection depends on the clinical question, medical history and the person's stamina and access needs.
Interpretation considers premorbid functioning, education, language, culture, sensory and motor abilities, mood, pain, sleep, medication and the timing of assessment relative to illness or injury. A low score may indicate a cognitive difficulty, but alternative explanations must be explored. Patterns across tasks can be more informative than a single result. Functional history helps determine whether test findings correspond to everyday difficulties such as managing medication, planning journeys or completing work tasks.
Neuropsychological reports can guide rehabilitation, environmental adjustments, safety planning and communication with families or services when consent permits. They should distinguish established findings from hypotheses and explain the limits of prediction. Cognitive performance can change with recovery, treatment, fatigue and environmental support. Reassessment should be timed to answer a meaningful question and account for practice effects. Collaboration with medical and rehabilitation teams is often important.
16. Forensic and occupational assessment
Forensic assessment addresses questions arising in legal contexts, such as psychological functioning relevant to a legal issue, risk formulation, capacity or the impact of an event. The assessor's role is to provide an independent, evidence-based opinion within the limits of expertise. The person being assessed may not be the person who commissioned the report, and confidentiality may be limited. The purpose, audience, procedures and potential use of findings must be explained before information is collected.
Forensic conclusions require careful reasoning, alternative hypotheses and transparent limitations. Risk assessment should not present a score as a certain prediction of future behaviour. Structured professional judgement may organise relevant historical, clinical and contextual factors, but it remains uncertain and must be linked to proportionate risk management. Cultural context, access to resources, protective factors and the possibility of bias must be considered. The assessor should avoid advocacy disguised as assessment and should not exceed the referral question.
Occupational assessment may examine skills, interests, work-related functioning, leadership or support needs. Tests used in recruitment must be relevant to the role, reliable and fair, and should not unnecessarily collect sensitive information. Reasonable adjustments may be required so that a person can demonstrate relevant capability. Employers should receive only information necessary for the stated decision, subject to law and consent. High-stakes decisions require documented procedures, review mechanisms and safeguards against discrimination.
17. Assessment of children and adolescents
Assessment of children and adolescents requires developmental knowledge and attention to family, school, peer and community settings. Behaviour that is expected at one age may be unusual at another, and skills develop at different rates. The assessor considers developmental history, language exposure, educational opportunity, health, sleep, sensory functioning and relationships. Information from caregivers and teachers can be valuable, but it should be gathered with appropriate consent and interpreted in light of each observer's setting and expectations.
Children should receive explanations that match their developmental level and be invited to express their views. Assent is important even when a caregiver provides legal consent. Confidentiality and its safeguarding limits should be explained to both the young person and caregiver. Adolescents may need some private discussion, depending on local law and professional guidance. The assessor must respond appropriately to disclosures of abuse, neglect, exploitation or immediate danger, following safeguarding procedures.
Assessment should lead to practical support rather than simply attach a label. Recommendations may involve classroom adjustments, targeted instruction, family support, health referral, communication tools or changes to the environment. Strengths and interests are essential to planning. Reports should distinguish developmental variation from clinically significant impairment and avoid treating cultural or linguistic differences as deficits. Follow-up is important because developmental needs and environmental demands change over time.
18. Culture, language and assessment fairness
Culture influences how people understand questions, express emotion, interpret tasks, respond to authority and view psychological services. Language proficiency affects performance on verbally loaded tests, while schooling and familiarity with formal examinations influence test-taking strategies. Cultural fairness does not mean assuming all members of a group are alike. It means examining whether procedures provide a valid opportunity for each person to demonstrate the construct being assessed and whether interpretations are supported for that context.
Test translation requires more than substituting words. Developers should examine conceptual equivalence, item relevance, response options, reading level and the meaning of examples. Back-translation can identify some discrepancies but cannot guarantee cultural validity. Adaptation may require expert review, cognitive interviews, pilot testing, norm development and analysis of measurement invariance. Where evidence is limited, the assessor should state the uncertainty and use multiple methods rather than presenting imported norms as universally applicable.
Fairness also concerns consequences. A measure may be statistically predictive while producing unjust barriers for a group because of unequal educational opportunity or irrelevant item content. Differential item functioning and measurement invariance analyses can help investigate whether items operate similarly across groups. Practitioners should not infer ability or pathology from accent, cultural practice, religion or socioeconomic status. Cultural humility includes recognising the assessor's own assumptions and inviting the person to explain what matters in their context.
19. Disability inclusion and accessible testing
Accessible assessment removes avoidable barriers while preserving the construct being measured. Adjustments may include large print, screen readers, sign-language interpretation, additional breaks, accessible rooms, alternative response methods or adapted instructions. The correct adjustment depends on the person's needs and the construct. If a timed motor response is central to the task, changing the timing may alter what is measured; if speed is irrelevant, additional time may improve access without undermining the intended interpretation.
Disability should not be equated with inability. Assessment distinguishes the person's skills from barriers created by inaccessible environments. A learner who cannot demonstrate knowledge through handwriting may be able to do so orally or with assistive technology. A person with hearing loss may require visual instructions. The assessor records adjustments, considers comparability and explains any limits. Consultation with the person and relevant specialists can help identify appropriate procedures.
Universal design can improve accessibility for many people by offering clear instructions, flexible response formats and predictable layouts. However, universal design does not remove the need for individual adjustments. Reports should describe functional needs and effective supports without disclosing unnecessary sensitive details. Decisions should comply with applicable disability rights and education or employment law. The objective is valid, respectful assessment and fair participation, not forcing every person to use an identical procedure.
20. Test construction and item analysis
Test construction begins with a clear construct definition and a test blueprint that maps content domains to item numbers, formats and intended cognitive demands. Developers write items, review them for clarity and bias, pilot them with a relevant sample and analyse responses. Item difficulty describes how often an item is answered correctly in a defined sample for a scored test. Item discrimination examines whether an item differentiates between people with higher and lower performance on the intended construct. These statistics depend on the sample and should not be interpreted mechanically.
Classical test theory represents an observed score as a true-score component plus measurement error. Item response theory models the probability of a response as a function of person ability and item characteristics, such as difficulty and discrimination; some models also include guessing. These frameworks offer different tools for test development and equating. They require appropriate data, model fit and expertise. A complex statistical model cannot rescue a poorly defined construct or biased content.
Good item writing avoids ambiguity, unnecessary reading difficulty, double negatives, clues to the answer and irrelevant cultural references. Distractors in multiple-choice items should be plausible and reflect common misconceptions. Performance tasks need clear rubrics and scorer training. Test developers examine reliability, validity evidence, fairness, accessibility and consequences before operational use. A test should be revised or withdrawn when evidence indicates that its scores no longer support the intended decisions.
21. Screening, diagnosis and decision thresholds
Screening identifies people who may need further evaluation. Diagnostic assessment determines whether a defined pattern is present under applicable criteria and professional standards. Monitoring measures track change over time. These purposes require different levels of precision and different consequences for error. Screening tools are often designed to be brief and sensitive, accepting that some people who screen positive will not meet diagnostic criteria after fuller evaluation.
Sensitivity is the proportion of people who truly have a condition who are correctly identified by a positive screen. Specificity is the proportion of people without the condition who are correctly identified by a negative screen. False positives occur when a screen indicates possible difficulty that is not confirmed; false negatives occur when a screen misses a person who does have the condition. Positive and negative predictive values depend partly on prevalence in the population tested. A cut-off should therefore be selected for the purpose and setting, not copied uncritically.
Decision thresholds involve trade-offs. In a low-cost early support programme, a sensitive screen may be useful because missing a person in need could be costly. In a high-stakes decision, false positives may have serious consequences and confirmation may be essential. Scores should be combined with clinical interview, history and other evidence. People should be told what a screening result means, what it does not mean and what follow-up options are available.
22. Ethical principles and informed consent
Ethical assessment respects autonomy, dignity, beneficence, non-maleficence, justice, competence and professional integrity. Informed consent requires understandable information about the purpose, procedures, expected duration, possible discomfort, benefits, risks, confidentiality, recipients of results and voluntary nature of participation. Consent is not merely a signature; it is an ongoing process. The person should have an opportunity to ask questions and withdraw where applicable, while being informed of any limits imposed by law or the assessment context.
Confidentiality is essential but not absolute. The assessor explains circumstances in which disclosure may be required or permitted, such as safeguarding obligations or serious and imminent safety concerns, according to applicable law and professional standards. Information should be shared only with appropriate people and only to the extent necessary. Reports should avoid irrelevant intimate details. Digital records and test data require secure storage, controlled access and appropriate retention practices.
Ethical practice also includes avoiding conflicts of interest, using tests only within competence, respecting copyright and test security, and preventing misuse of results. Practitioners should not release raw test materials to unauthorised recipients or allow unqualified people to interpret restricted instruments. They should document decisions, seek supervision and correct errors. When an assessment is commissioned by an organisation, the assessor must explain the dual obligations to the individual and the commissioning body.
23. Bias, response styles and validity concerns
Response bias refers to systematic patterns that can distort answers. Social desirability involves presenting oneself in a favourable way; acquiescence is a tendency to agree; extreme or midpoint responding concerns preferences for particular response categories. Impression management may be conscious or unconscious. Such patterns are not proof that a person is lying. They may reflect culture, uncertainty, literacy, anxiety, misunderstanding or a genuine tendency in how the person uses rating scales.
Performance validity concerns whether test performance provides an interpretable estimate of the ability being measured. Symptom validity concerns the interpretability of symptom reports. These are complex professional concepts and should not be reduced to a binary judgement of honesty. Poor performance may result from fatigue, pain, severe distress, cognitive impairment, language barriers or misunderstanding. Validity indicators require appropriate training, validated procedures and cautious interpretation in the context of the whole assessment.
Bias can enter through item content, sampling, administration, scoring, interpretation and decision-making. Stereotype threat, examiner expectations and unequal access to preparation may influence performance. Standardised instructions, blind scoring where feasible, diverse development samples and structured interpretation can reduce some risks. They cannot eliminate the need for critical reflection. Reports should state when validity concerns limit conclusions and should avoid accusing a person without adequate evidence.
24. Feedback, reporting and recommendations
A psychological report communicates the referral question, methods, relevant background, findings, interpretation, limitations and recommendations. It should be accurate, proportionate and written for its intended audience. Technical terms should be explained. Scores are reported with appropriate norms and uncertainty, and conclusions should distinguish direct observations, self-report, collateral information and professional inference. The report should not include irrelevant personal information or claim certainty beyond the evidence.
Feedback is a conversation as well as a document. The assessor explains what the results mean, checks the person's understanding, invites questions and discusses emotional reactions. Strengths should be included alongside needs. Recommendations should be specific, feasible and prioritised. 'Improve attention' is not an actionable recommendation; 'provide written task steps, reduce competing noise during independent work and check understanding before starting' is more useful when supported by findings.
Reports can have lasting effects in schools, workplaces, healthcare and legal systems. The assessor considers who may read them, how decisions may be made and whether the language could stigmatise or be misunderstood. Where appropriate, the person receives a copy and knows how to request clarification or correction. Follow-up can determine whether recommendations were implemented and whether they improved functioning. Assessment is successful when it supports better decisions, not merely when a report is completed.
25. Assessment in clinical, school and organisational settings
In clinical settings, assessment may clarify presenting concerns, establish baseline functioning, support diagnosis, guide treatment and monitor outcomes. Measures should be selected according to the person's symptoms, language, medical context and goals. Repeated symptom scales can help identify improvement or deterioration, but a score should not replace conversation about functioning, adverse effects or preferences. Risk assessment and safeguarding procedures must be integrated when relevant.
In schools, assessment supports learning, inclusion, educational planning and evaluation of interventions. It should distinguish achievement from aptitude and consider instruction quality, attendance, language, disability access and classroom environment. Teachers contribute valuable observations but should not be expected to diagnose conditions outside their professional role. Families and learners should understand the purpose and how information will be used. Recommendations should be practical for the resources and realities of the school.
In organisations, assessment may inform recruitment, development, leadership coaching, team design and occupational support. Selection measures must relate to essential job requirements and be evaluated for fairness and predictive value. Personality tests should not be used as simplistic pass-or-fail devices. Candidates should receive clear information about procedures and data use. Employers should consider reasonable adjustments, privacy and the consequences of automated scoring. The validity of a tool must be evaluated for the actual role and population.
26. Digital assessment and artificial intelligence
Digital assessment includes computer-delivered tests, remote interviews, mobile questionnaires, adaptive testing and sensor-based measures. Benefits may include efficient scoring, flexible access and timely feedback. Risks include unequal internet access, device differences, privacy breaches, distraction, identity verification problems and reduced control over testing conditions. Remote assessment requires a plan for technical failure, accessibility, emergency response and confirmation that the environment is suitable for the task.
Computer-adaptive testing selects items based on previous responses to estimate ability efficiently. Its accuracy depends on calibrated item banks, model assumptions and appropriate content coverage. Automated scoring can improve consistency for some tasks but may reproduce bias in training data or fail to recognise valid forms of expression. Artificial intelligence systems should be transparent enough for accountable use, independently evaluated and subject to human review. An algorithmic score should not become an unchallengeable decision about a person.
Digital data can be highly sensitive. Practitioners should know where information is stored, who can access it, whether it is used to train other systems and how long it is retained. Consent must explain relevant data flows in plain language. Remote tests should not be assumed equivalent to in-person versions without evidence. Professional accountability remains with the practitioner or organisation using the tool; technology does not remove duties of competence, fairness, confidentiality and care.
27. Monitoring change and evaluating interventions
Outcome monitoring uses repeated measures and other evidence to evaluate whether an intervention is helping. Baseline measurement provides a reference point, while follow-up assesses change. Reliable change indices can help determine whether a difference is larger than expected measurement error, and clinically significant change considers whether the person has moved into a more functional range. These methods depend on appropriate reliability estimates and reference data and should not be treated as universal thresholds.
Progress is multidimensional. Symptom reduction may occur without improvement in relationships, education, work or quality of life; conversely, meaningful functional gains may occur before symptoms fully resolve. Client-defined goals, attendance, adverse effects, satisfaction and contextual changes should be considered. If progress is limited, the team reviews the formulation, treatment fit, barriers, dosage, therapeutic relationship and possible alternative explanations rather than blaming the person.
Evaluation should be routine but humane. Repeated questionnaires can become burdensome if they do not inform decisions. The assessor explains why information is collected and how it will be used. Data should be interpreted alongside the person's narrative. At service level, aggregate outcomes can identify inequities or quality problems, but small groups require privacy safeguards. Evaluation supports learning and improvement when findings lead to thoughtful action.
28. Integrating findings and professional judgement
Integration means weighing evidence across methods and sources rather than mechanically averaging scores. Converging evidence increases confidence when independent methods support a similar conclusion. Divergent findings may reveal context-specific functioning, response bias, measurement limitations or a construct with several components. For example, a person may perform well in a quiet testing room but struggle in a crowded classroom. The difference is meaningful and should guide investigation rather than be dismissed as inconsistency.
Professional judgement should be structured, evidence-informed and open to challenge. The assessor states the reasoning that links data to conclusions, identifies alternative explanations and describes uncertainty. Confidence should be proportional to evidence quality. When information is missing, the report says so. When a question cannot be answered, the assessor explains why and identifies what additional evidence might help. Avoiding false certainty is a mark of competence, not weakness.
An integrated assessment ends with recommendations linked to findings and priorities agreed with the person. The plan may include referral, further evaluation, treatment, teaching adjustments, workplace accommodation or no additional intervention if evidence does not indicate a need. Decisions should be reviewed when circumstances change. The assessor remains alert to unintended consequences, including stigma, exclusion or misuse of scores. The purpose of measurement is to support fairer and more effective human decisions.
Review outcomes and revise the plan when new evidence emerges.
↓ Integrated interpretation, with uncertainty and limitations stated ↓
Chapter summary and key terms
Psychological assessment is a purposeful integration of evidence; testing is only one possible component. Measurement depends on clear constructs, suitable methods and appropriate interpretation. Reliability concerns consistency, while validity concerns support for score interpretations and uses. Norms are meaningful only when the reference group is appropriate. Interviews, observation, self-report, cognitive measures and records each offer distinct evidence and limitations. Fair assessment accounts for culture, language, disability and context. Ethical practice requires consent, confidentiality, competence, transparent reporting and recommendations that improve decisions.
Key terms: construct; operational definition; psychometrics; reliability; validity; standardisation; norm-referenced score; criterion-referenced score; percentile rank; standard error of measurement; sensitivity; specificity; predictive validity; response bias; item analysis; functional assessment; informed consent; measurement invariance.
Structured university examination questions with model answers
Attempt each question before reviewing its model answer. In formal examinations, support answers with examples and relevant terminology.
1. Distinguish psychological testing from psychological assessment.
Model answer: Testing is the use of a particular instrument or procedure. Assessment is the broader process of integrating tests, interviews, observation, history and contextual evidence to answer a referral question.
2. Outline the main stages of an assessment cycle.
Model answer: Clarify the referral question; define constructs and information needs; select appropriate methods; collect evidence; interpret it in context; communicate findings and recommendations; and review whether the assessment achieved its purpose.
3. Explain construct underrepresentation and construct-irrelevant variance.
Model answer: Underrepresentation occurs when a measure omits important parts of the intended construct. Irrelevant variance occurs when scores are influenced by factors unrelated to it, such as inaccessible instructions or unfamiliar language.
4. Why is reliability necessary but insufficient for validity?
Model answer: Reliability concerns consistency; validity concerns whether interpretations and uses are supported. A measure can consistently measure the wrong attribute, so consistency alone cannot justify conclusions.
5. Compare test–retest, internal consistency and inter-rater reliability.
Model answer: Test–retest examines stability over time; internal consistency examines relationships among items; inter-rater reliability examines agreement between observers or scorers. They address different error sources.
6. Explain why validity is an argument rather than a single coefficient.
Model answer: Validity requires converging evidence about content, response processes, internal structure, relationships with other variables and consequences for the intended use. One statistic cannot establish all these claims.
7. Distinguish norm-referenced and criterion-referenced interpretation.
Model answer: Norm-referenced scores compare a person with a defined group. Criterion-referenced scores compare performance with a specified standard or skill domain.
8. Interpret a percentile rank of 75.
Model answer: The score is at or above approximately 75 percent of the defined norm group. It does not mean 75 percent of items were answered correctly or that ability is 75 percent.
9. Explain why norms must be appropriate and current.
Model answer: Norms reflect a particular sample, time and context. If the person differs from the reference group or norms are outdated, comparisons may be misleading.
10. Differentiate nominal, ordinal, interval and ratio measurement.
Model answer: Nominal categories have no order; ordinal categories have order but unequal or unknown intervals; interval scales have equal units without a meaningful absolute zero; ratio scales have equal units and a meaningful zero.
11. Give two advantages and two limitations of self-report scales.
Model answer: They efficiently capture private experiences and standardise questions. Limitations include memory error, response styles, literacy or language barriers and social desirability.
12. Why is a screening result not a diagnosis?
Model answer: Screening estimates whether further assessment may be needed. False positives and negatives occur, and diagnosis requires fuller evaluation of criteria, context, impairment and alternative explanations.
13. Define sensitivity and specificity.
Model answer: Sensitivity is the proportion of people with a condition correctly identified as positive. Specificity is the proportion without the condition correctly identified as negative.
14. What is positive predictive value and why does prevalence matter?
Model answer: It is the proportion of positive results that are true positives. It changes with prevalence; in low-prevalence populations, a larger share of positive screens may be false positives.
15. Describe the role of behavioural observation.
Model answer: It records defined behaviour in context, helping identify patterns, antecedents, consequences and setting effects. Multiple observations and clear coding improve interpretability.
16. Explain antecedent–behaviour–consequence analysis.
Model answer: It records what happens before a behaviour, the observable behaviour itself and what follows. The sequence helps generate hypotheses about behavioural function and maintenance.
17. Compare structured and semi-structured interviews.
Model answer: Structured interviews use fixed questions and rules for consistency. Semi-structured interviews retain core questions while allowing relevant follow-up and clarification.
18. Why should an assessor distinguish report from inference?
Model answer: It makes the evidence trail transparent, reduces bias and allows readers to see which statements come from the person's account, observation, records or professional interpretation.
19. Discuss fairness when using tests across languages and cultures.
Model answer: Assessors examine conceptual and linguistic equivalence, norm suitability, item relevance, schooling and response processes. Literal translation alone does not establish validity.
20. Explain the purpose of a test blueprint.
Model answer: It maps content domains and cognitive demands to items, helping ensure representative coverage and alignment with the construct or curriculum.
21. What is item discrimination?
Model answer: It indicates how well an item differentiates people with different levels of the measured attribute in the studied sample. It should be interpreted alongside item content and test purpose.
22. Explain why intelligence scores should not be treated as fixed potential.
Model answer: Scores are estimates from selected tasks and are influenced by opportunity, language, health, education and access. They do not capture every form of ability or future development.
23. Identify ethical requirements for informed consent.
Model answer: Explain purpose, procedures, risks, benefits, confidentiality and limits, recipients of results and voluntary participation in accessible language; allow questions and treat consent as ongoing.
24. How should an assessor handle conflicting informant reports?
Model answer: Explore setting, timing, expectations and opportunities for observation. Differences may reflect context-specific behaviour; seek additional evidence rather than automatically privileging one source.
25. What makes a recommendation useful?
Model answer: It is specific, feasible, prioritised, linked to assessment evidence and expressed in language the person or organisation can act on.
26. Discuss risks of automated psychological assessment.
Model answer: Risks include biased training data, opaque scoring, privacy loss, technical inequity and overreliance on algorithms. Independent evaluation, human oversight and clear data governance are required.
27. Explain converging and divergent evidence.
Model answer: Converging evidence from different methods supports a shared interpretation. Divergence may reveal contextual differences, measurement limitations or distinct dimensions requiring further investigation.
28. What is the standard error of measurement?
Model answer: It estimates the expected uncertainty around an observed score due to measurement error, reminding users that scores are not exact values.
29. Differentiate formative and summative educational assessment.
Model answer: Formative assessment guides teaching and learning while it occurs. Summative assessment evaluates achievement at a defined endpoint or reporting period.
30. Describe how assessment findings should be communicated.
Model answer: Explain the referral question, methods, results, uncertainty and limitations; use accessible language, include strengths, link practical recommendations to evidence and invite questions.
Further study and responsible use
Consult current psychometric texts, test manuals, peer-reviewed validation studies, professional assessment standards and applicable national laws. Check whether evidence supports the specific test, population, language and decision. Restricted instruments should be administered and interpreted only by appropriately qualified professionals. This chapter is an academic learning resource, not permission to administer restricted tests or make clinical, educational, employment or legal decisions without appropriate competence and oversight.