PSY343 Psychometrics: Test Construction Study Notes (Stellenbosch University)

Psychometrics is the scientific study of psychological measurement, and test construction is one of its most practical and examinable applications. For Stellenbosch University PSY343 students, this topic usually combines theory, statistics, and applied decision-making: defining what to measure, writing good items, piloting a test, analysing item performance, and evaluating reliability and validity. Strong exam answers show not only what the concepts mean, but also how each step in test construction affects score quality, fairness, and interpretation.

1. The logic of psychometrics and the purpose of test construction

Psychometrics is built on a simple but demanding idea: if a psychological attribute is not directly observable, it can still be measured indirectly through behaviour. Test construction is the process of creating a tool that captures this hidden attribute as accurately, consistently, and fairly as possible. In PSY343, this means understanding that a test is never just a set of questions. It is an instrument designed to produce scores that can support decisions, inferences, or research conclusions.

What psychometrics measures

Psychometric tests are used to measure constructs such as intelligence, personality, anxiety, depression, attitudes, motivation, aptitude, and specific abilities. These constructs are not physical objects; they are theoretical dimensions inferred from responses. The central challenge is that different people may respond similarly for different reasons, and the same person may respond differently across contexts. Good test construction reduces ambiguity by defining the construct carefully and by linking items to observable indicators.

A useful exam distinction is between:

Observed score: the score a person actually obtains on a test.
True score: the stable component of the score associated with the construct.
Error score: random or systematic influences that distort the observed score.

This distinction is central to classical test theory, where:

Observed score = True score + Error

This formula is not just mathematical decoration. It explains why tests need reliability evidence, why item quality matters, and why repeated measurement can vary slightly even when the underlying trait has not changed. A test with a large error component is less useful because it cannot separate stable individual differences from noise.

Why test construction matters

Test construction matters because poor tests produce misleading results. A badly written achievement test may confuse students about what is being assessed. A flawed personality scale may measure social desirability instead of the intended trait. A biased aptitude test may unfairly disadvantage groups because of language or cultural loading. In practice, test construction affects:

selection and placement decisions,
diagnosis and screening,
educational assessment,
clinical evaluation,
organisational recruitment,
research measurement,
programme evaluation.

When a test is used for high-stakes decisions, such as admission or diagnosis, the consequences of weak construction become especially serious. A small measurement error can misclassify a candidate, and a validity problem can create systematic unfairness. This is why test construction is not merely a technical activity; it is a scientific and ethical responsibility.

Core principles of a good test

A sound psychometric test usually has the following characteristics:

Clear construct definition
The test must measure a well-specified attribute. Vague constructs produce vague items.
Adequate content coverage
Items should represent the full domain of the construct, not only one narrow aspect.
Reliability
Scores should be stable and internally coherent.
Validity
Scores should support the intended interpretation and use.
Fairness
The test should minimise irrelevant disadvantage and bias.
Practicality
The test should be usable within realistic constraints of time, cost, and administration.
Standardisation
Administration, scoring, and interpretation must be consistent.

These principles interact. For example, a test can be very reliable but still invalid if it consistently measures the wrong thing. A test can be practical but weak if it sacrifices content breadth for speed. Exam answers become stronger when they show that psychometric quality is multidimensional rather than one-dimensional.

Test construction as a process

Test construction typically follows a sequence:

define the construct,
identify the purpose of the test,
specify the test blueprint,
write items,
review items for content and bias,
pilot the test,
analyse item statistics,
revise or discard weak items,
establish reliability and validity evidence,
standardise scoring and interpretation.

This process is iterative, not linear. A pilot study often reveals problems that force a return to the blueprint or item-writing stage. A strong test constructor therefore thinks in cycles of refinement rather than in a single pass from idea to finished instrument.

Example: measuring test anxiety

Suppose a lecturer wants to build a test anxiety scale for university students. The construct must first be defined carefully. Test anxiety might include cognitive worry, emotional tension, physiological symptoms, and avoidance behaviour. If the constructor writes only items about nervous feelings before exams, the scale may miss avoidance and physical symptoms. The resulting instrument might be reliable but incomplete. A good blueprint ensures balanced coverage of all important subdimensions.

This example shows why psychometrics is both conceptual and statistical. The best statistics cannot rescue a poor construct definition, and a good construct cannot be measured properly without statistical evaluation.

2. Defining constructs, writing a blueprint, and planning the test

Test construction begins long before the first item is written. The most important early decision is what exactly the test is supposed to measure and how that construct will be represented in item form. In many exam questions, students lose marks because they jump straight to reliability or item analysis without showing how construct definition comes first. In psychometrics, clarity at the planning stage determines the quality of everything that follows.

Construct definition and operationalisation

A construct is an abstract psychological attribute inferred from patterns of behaviour, responses, or performance. Examples include self-esteem, reasoning ability, impulsivity, or academic motivation. To test a construct, it must be operationalised, meaning translated into observable indicators or items.

Good operationalisation requires:

a theoretical definition,
a behavioural expression of the construct,
an explanation of what should count as evidence,
a statement of what should not be included.

For example, if the construct is conscientiousness, the test should include behaviours related to organisation, responsibility, and persistence, but not simply general intelligence or compliance with authority. If the construct is verbal reasoning, items should focus on language-based inference, not arithmetic speed.

A common exam point is the difference between:

theoretical construct: the abstract idea;
operational definition: the measurable manifestation in the test.

The stronger the operational definition, the easier it is to create items that are relevant and interpretable.

Purpose of the test

The purpose of the test determines item format, scoring, and interpretation. Different purposes include:

screening: identifying possible cases requiring follow-up;
selection: distinguishing among candidates for entry or employment;
diagnosis: assessing whether a person meets criteria for a condition;
placement: allocating people to appropriate levels or groups;
research: measuring variables in a study;
evaluation: assessing program impact or treatment effect.

A screening instrument may prioritise sensitivity, while a selection test may prioritise prediction of future performance. A clinical measure may require strong validity and careful norms. Because purpose affects design, a test cannot be evaluated properly unless its intended use is explicit.

Test blueprint or table of specifications

The test blueprint is one of the most important planning tools in test construction. It specifies the content domains, the relative weight of each domain, the cognitive level required, and the number of items per category. In education, this is often presented as a table of specifications.

A blueprint prevents two common problems:

underrepresentation of important content;
overrepresentation of trivial content.

For example, if a PSY343 lecturer is constructing a short test on item analysis, the blueprint might include:

item difficulty,
item discrimination,
distractor analysis,
reliability,
validity.

If 50% of the course time was spent on reliability and validity, but the test includes only one item each on these topics, the test would not adequately represent the curriculum. A blueprint aligns the test with learning outcomes and content emphasis.

Simple blueprint example

Content domain	Weight	Number of items	Item type
Construct definition	20%	8	MCQ and short answer
Reliability	25%	10	MCQ
Validity	25%	10	MCQ and applied questions
Item analysis	20%	8	Calculation and interpretation
Ethics and fairness	10%	4	Short answer

This kind of table helps ensure that the final test reflects both the conceptual structure of the module and the relative importance of each area.

Choosing a response format

Response format affects measurement quality. Common formats include:

multiple-choice questions (MCQs)
true/false items
Likert-type scales
ranking items
short-answer items
essay questions
performance tasks

Each format has strengths and weaknesses. MCQs allow objective scoring and broad sampling of content, but good distractor construction is difficult. Likert scales are useful for attitudes and self-report traits, but they may be vulnerable to acquiescence, central tendency, and social desirability biases. Essay items can capture complex reasoning, but scoring reliability may be lower unless clear rubrics are used.

The format must match the construct. A personality trait is usually better measured with self-report scales than with right-or-wrong items. An achievement domain is often better measured with performance items than with agreement statements. A poor match between construct and format creates construct-irrelevant variance, which weakens validity.

Writing item objectives before writing items

A smart way to plan the test is to write item objectives first. Each objective specifies what the item should assess. For example:

distinguish between reliability and validity,
identify the correct interpretation of a correlation coefficient,
apply the formula for item difficulty,
evaluate whether a test item is biased.

By writing objectives first, the test constructor avoids drifting into items that are interesting but irrelevant. This step also helps create parallel forms and balanced coverage.

Planning for fairness and accessibility

Modern test construction must consider fairness from the beginning. This includes:

language simplicity appropriate to the target group,
avoiding culturally loaded examples unless culturally relevant content is intended,
avoiding double negatives and ambiguous phrasing,
ensuring that test directions are clear,
considering accessibility for candidates with disabilities.

A test may be statistically strong yet unfair if the wording disadvantages people because of language proficiency rather than the construct being measured. In South African higher education, where students may come from diverse linguistic backgrounds, this issue is especially significant. If a test on psychometrics uses unnecessarily complex English, it may measure reading burden more than subject knowledge.

Planning checklist

Before item writing begins, a robust plan should answer:

What construct is being measured?
What is the purpose of the test?
Who is the target population?
What content domains must be included?
What weighting should each domain receive?
What item format fits the construct?
What time limit is appropriate?
What level of reading ability is assumed?
What scoring procedure will be used?
What evidence will later be needed for reliability and validity?

These questions are foundational because every later decision depends on them. A carefully designed blueprint is one of the strongest predictors of a useful final test.

3. Item writing, response formats, and common item-writing errors

Item construction is where theoretical planning becomes a practical instrument. A test can fail not because the construct was wrong, but because items were poorly written. In psychometrics, item quality is crucial because each item contributes to the overall score and to the meaning of that score. Weak items increase error, distort difficulty levels, and reduce reliability and validity.

General principles of good item writing

Good test items are:

clear,
concise,
relevant to the construct,
free of unnecessary complexity,
free of ambiguity,
matched to the target population,
consistent with the blueprint,
balanced in difficulty.

The item should measure the intended construct and little else. This is sometimes summarised as: each item should have a single clear task. If the respondent must solve two problems at once, ambiguity increases and interpretation becomes weaker.

Multiple-choice item construction

Multiple-choice items are common in educational and aptitude testing because they can sample many content areas efficiently and are easy to score objectively. A standard MCQ contains:

a stem,
one correct answer or keyed response,
several distractors.

A good stem poses a clear problem or question. The options should be plausible, similar in length and grammatical structure, and free of clues that reveal the answer.

Example of a strong MCQ

Which of the following best describes reliability?

A. The extent to which a test measures what it claims to measure
B. The consistency of scores across repeated measurement
C. The degree to which a test predicts future performance only
D. The fairness of a test across cultural groups

Correct answer: B

This item works because the distractors reflect common confusions with validity, predictive validity, and fairness. It checks understanding rather than rote memorisation alone.

Common MCQ errors

Ambiguous stem
The question does not clearly indicate what is being asked.
More than one plausible correct answer
This destroys score interpretability.
Grammatical clues
One option fits the stem grammatically while others do not.
Unequal option length
The correct answer is often the longest, which can cue test-wise students.
Negative wording without necessity
Questions using “except” or “not” increase confusion.
Tricky distractors that are implausible
If distractors are obviously wrong, the item becomes too easy and less informative.

True/false items

True/false items are simple and quick, but they have serious limitations. Because there are only two options, guessing probability is 50%, which reduces discrimination. They are best used for preliminary checks, practice activities, or when many items are needed and broad coverage is more important than precision. If used, statements must be unambiguous and not excessively broad.

An item such as “All valid tests are reliable” is problematic because it may be true in a technical sense, but students can argue about wording. A better item would clearly express a single factual claim.

Likert-type items and scales

Likert-type items are common for measuring attitudes, opinions, and self-reported traits. Respondents indicate degree of agreement, frequency, or intensity on ordered categories such as:

strongly disagree,
disagree,
neutral,
agree,
strongly agree.

Several principles matter here:

Each item should express only one idea.
Each response category should be clearly defined.
The direction of scoring must be consistent or carefully reversed.
Items should avoid double-barrelled wording.

Example of a good Likert item

“I feel nervous before important tests.”

This item is clear, singular, and directly relevant to test anxiety.

Example of a bad Likert item

“I feel nervous and unprepared before important tests and assignments.”

This item is double-barrelled because it mixes nervousness, preparedness, tests, and assignments. A respondent may agree with one part but not another, making the response difficult to interpret.

Open-ended and performance items

Open-ended items allow deeper responses and may assess reasoning, synthesis, or explanation. However, they require scoring rubrics to improve objectivity. Performance tasks are especially useful when the test seeks to measure real-world application, such as interpreting psychometric statistics, analysing case material, or constructing a test blueprint.

The main challenges are:

more time to answer,
more time to score,
potential scorer bias,
lower inter-rater reliability if rubrics are weak.

To reduce these problems, rubrics should define criteria at different quality levels. In exam settings, it is useful to mention that open-ended tasks can increase content validity and reduce cueing, but may decrease scoring efficiency.

Writing for the target population

The quality of an item depends on the population for whom it is intended. A test for first-year psychology students should not assume advanced statistical knowledge unless that knowledge is part of the construct. Similarly, an item for a multilingual context should avoid idioms, slang, and highly local references unless they are essential to the content.

Consider these two versions:

“The item showed poor discrimination.”
“The item did not separate high-performing from low-performing students well.”

The second version is easier for some audiences to understand. Clarity should not be sacrificed in order to appear sophisticated.

Item-writing rules that improve quality

A practical set of rules includes:

Write one clear idea per item.
Avoid unnecessary complexity.
Avoid clues in grammar or length.
Match wording to the reading level of the target group.
Avoid sarcasm, humour, and cultural assumptions unless intended.
Keep the stem and options parallel.
Make distractors plausible.
Avoid absolute words such as “always” or “never” unless the content demands them.
Avoid double negatives.
Review each item for bias and ambiguity.

The role of item difficulty in item writing

Items should vary in difficulty. A test made up only of very easy items or only of very hard items provides limited information about individuals. A balanced test usually includes a range of items so that it can distinguish between low, medium, and high levels of the trait or ability. During item writing, the test constructor should aim for some easy, some moderate, and some difficult items, depending on the purpose of the test.

For example, in an achievement test, some items can assess foundational knowledge, while others assess application and analysis. This creates a score distribution with meaningful spread rather than a pile-up at the top or bottom.

Case example: a poor item and its repair

Suppose an item asks:

“Validity means the same as reliability, right?”

This is poor because it is informal, leading, and imprecise. A repaired version would be:

“Which statement best distinguishes validity from reliability?”

This revised item is clearer, academically appropriate, and directly linked to conceptual understanding. In psychometric item construction, refinement often makes the difference between a weak and an examinable item.

4. Pilot testing, item analysis, reliability, and validity evidence

Once items have been written and reviewed, they must be tested empirically. This is where theory meets data. Pilot testing shows whether items function as intended, and item analysis identifies which items should be kept, revised, or discarded. Reliability and validity evidence are then built from the results. In exam answers, this section is often where students can gain substantial marks by demonstrating an understanding of statistics in context.

Why pilot testing is necessary

A pilot test is a small-scale administration of the test to a sample similar to the target population. The purpose is to identify problems before full use. Pilot testing can reveal:

unclear instructions,
confusing items,
inappropriate difficulty,
poor distractors,
timing problems,
unexpected response patterns,
technical or administration errors.

No matter how carefully a test is designed, actual respondents may interpret items differently from what the constructor intended. Pilot testing is therefore a reality check.

Item difficulty

In classical test theory, item difficulty is often represented by the proportion of respondents who answer correctly. For an MCQ, if 80 out of 100 students answer correctly, the difficulty index is 0.80. Despite the name, a higher value means an easier item. An item difficulty of 0.20 means only 20% answered correctly, so the item is difficult.

A balanced test generally avoids items that are too easy or too difficult unless the purpose calls for them. If almost everyone gets an item right, it contributes little to distinguishing among individuals. If almost everyone gets it wrong, it also contributes little because it cannot separate respondents well.

Item discrimination

Item discrimination refers to how well an item distinguishes between high-scoring and low-scoring respondents. A good item is more likely to be answered correctly by people with higher total scores than by people with lower total scores. Discrimination can be examined through item-total correlations or other indices.

If a high-performing group answers an item correctly much more often than a low-performing group, the item discriminates well. If low-performing students do as well as high-performing students, the item is weak. In some cases, negative discrimination indicates a serious problem, such as a miskeyed answer, ambiguous wording, or content that measures something unrelated to the test construct.

Distractor analysis

For MCQs, distractor analysis examines whether incorrect options function properly. A good distractor should attract some respondents, especially those who have not mastered the content. If a distractor is never chosen, it is probably implausible and should be revised or replaced. If a distractor is chosen more often by high-performing students than low-performing students, the item may be misleading or flawed.

A simple distractor analysis table can show the proportion choosing each option:

Option	Proportion chosen
A	0.12
B	0.63
C	0.15
D	0.10

If B is the keyed answer, this item may be functioning well, provided the high-performing students are disproportionately selecting B. If option C is chosen heavily by strong students, that may suggest ambiguity or a partially correct distractor.

Reliability

Reliability is the consistency or stability of scores. It does not tell us whether the test measures the right construct; it tells us whether the measurement is dependable. Major forms of reliability include:

test-retest reliability: consistency over time,
internal consistency: coherence among items,
split-half reliability: consistency between two halves of a test,
inter-rater reliability: agreement between scorers,
parallel-forms reliability: consistency across equivalent versions.

Test-retest reliability

This is useful when the construct should remain relatively stable across the interval between tests. If scores change dramatically without a real change in the trait, the test may be unstable. However, the time interval matters: too short and memory inflates the correlation; too long and real change may reduce it.

Internal consistency

This assesses whether items on the same test are measuring the same underlying construct. If items are too heterogeneous, the internal consistency may be low. Common indicators include split-half approaches and coefficients such as Cronbach’s alpha. A high alpha often suggests items hang together, but an alpha that is too high can sometimes indicate item redundancy rather than ideal breadth.

Inter-rater reliability

This matters for essays, performance tasks, and observational ratings. If two raters score the same response differently, score meaning becomes unstable. A clear rubric and rater training improve agreement.

Validity

Validity concerns the degree to which evidence and theory support the intended interpretation of test scores. It is not a property of the test in the abstract alone; it is a property of the score interpretation for a specific purpose. A test may be valid for one use and not another.

Major sources of validity evidence include:

content validity: the items represent the domain adequately,
criterion-related validity: scores relate to relevant outcomes,
construct validity: scores behave as expected within a theory,
face validity: the test appears appropriate, though this is not strong technical evidence.

Content validity

Content validity is especially important in achievement tests. If the test should cover five topics but only tests two, it lacks content validity. This is why the blueprint is essential. Content experts often review whether the items match the intended domain.

Criterion-related validity

This involves the relationship between test scores and an external criterion. If a selection test predicts later job performance, it has predictive validity. If it correlates with an established measure administered at the same time, it may show concurrent validity.

Construct validity

Construct validity asks whether the test behaves as expected within a theoretical framework. For example, a test of anxiety should correlate positively with related constructs like stress, but not so strongly with unrelated constructs that it becomes indistinguishable from them. It should also show sensible patterns across groups or contexts if theory predicts them.

Reliability and validity are related but different

This is one of the most important exam distinctions. Reliability is necessary but not sufficient for validity. A test cannot be valid if it is wildly inconsistent, because unstable scores cannot support meaningful interpretation. However, a reliable test may still be invalid if it consistently measures the wrong thing.

For example, a mathematics exam that mainly measures reading comprehension may produce consistent scores, but those scores would not validly represent mathematical ability. This distinction should be stated clearly in exam answers.

Common interpretation errors

Students often make these mistakes:

treating reliability and validity as synonyms,
assuming high reliability automatically means high validity,
assuming face validity is enough,
confusing item difficulty with test difficulty,
assuming a correlation alone proves validity.

A strong PSY343 response explains that psychometric evidence is cumulative. Multiple forms of evidence are needed before a test can be confidently used.

5. Scoring, standardisation, ethics, and exam-focused synthesis

The final stage of test construction is not simply scoring the test. It is standardising the entire process so that scores can be interpreted responsibly. In applied psychometrics, this is where ethics, fairness, and interpretation become central. A well-constructed test must be administered consistently, scored accurately, normed appropriately, and used in a way that respects the rights of test takers.

Scoring systems

The scoring method depends on the test format and purpose. Common approaches include:

dichotomous scoring: 1 for correct, 0 for incorrect;
polytomous scoring: partial credit for graded responses;
summated rating scales: adding Likert responses across items;
rubric-based scoring: applying criteria to open-ended answers;
weighted scoring: some items count more than others.

Scoring should reflect the construct. In a knowledge test, correct answers may be scored dichotomously. In a performance task, partial credit can recognise partially correct reasoning. In a personality scale, the focus is usually on pattern of endorsement rather than right or wrong answers.

Standardisation of administration

Standardisation means that all test takers receive the same instructions, conditions, time limits, and scoring rules as much as possible. Without standardisation, differences in score may reflect differences in administration rather than real differences in the construct.

Standardisation usually involves:

fixed instructions,
consistent timing,
controlled environment,
clear handling of queries,
same materials and response sheets,
standard scoring procedures.

If one group gets additional explanation and another does not, comparability is weakened. Standardisation is especially important when scores are used for selection or diagnosis.

Norms and interpretation

Raw scores alone often do not mean much. A score becomes interpretable when it is compared with a relevant reference group, or norm. Norms may be presented as:

percentile ranks,
standard scores,
stanines,
z-scores,
T-scores,
age norms or grade norms.

Norms help answer questions like: Is this score high or low relative to peers? However, norms must be relevant to the population being tested. Using norms from a very different group can create misleading conclusions. A test normed on one population may not be appropriate for another if the groups differ in language, education, or context.

Ethical principles in test construction

Ethics is inseparable from psychometrics. A test constructor must consider:

informed use of the test,
confidentiality,
competence of the administrator,
fairness and non-discrimination,
appropriate interpretation of results,
avoidance of harm.

Testing can affect people’s educational opportunities, employment, and self-concept. Poorly designed tests can reinforce inequality. For this reason, fairness is not an optional add-on; it is a core psychometric concern.

Bias and fairness

A test item is biased when it disadvantages a group for reasons unrelated to the construct. Bias may arise from language complexity, cultural assumptions, unfamiliar contexts, or stereotype threat. To reduce bias, constructors should use diverse review panels, pilot the test on representative samples, and examine differential performance patterns.

It is important to distinguish difference from bias. Two groups may score differently for many reasons, including true differences in the construct, unequal opportunity to learn, or measurement bias. A score difference by itself does not prove bias, but it does warrant investigation.

Revision and test refinement

Test construction is iterative. After scoring and analysis, weak items are revised or removed. Revision decisions may be based on:

low discrimination,
extreme difficulty,
poor distractor functioning,
ambiguity reported by respondents,
evidence of bias,
weak contribution to reliability.

Sometimes an item is not discarded outright but rewritten for a second pilot. This is efficient when the conceptual content is valuable but the wording is weak. Other times, the item must be removed because the problem is conceptual rather than editorial.

An integrated construction model

A high-quality test emerges from the integration of several components:

clear construct definition ensures relevance;
test blueprint ensures coverage;
good item writing ensures clarity;
pilot testing reveals real-world problems;
item analysis improves functioning;
reliability evidence supports consistency;
validity evidence supports interpretation;
standardisation supports comparability;
ethical safeguards support fair use.

If any one of these is missing, the test becomes weaker. A beautiful item set without norms is hard to interpret. Good reliability without validity gives false confidence. Careful validity evidence without standardisation leaves room for procedural noise. Psychometrics is therefore best understood as a system.

Exam-style summary of key contrasts

Concept	Main question	What it tells us	What it does not tell us
Reliability	Are scores consistent?	Stability and coherence	Whether the test measures the right construct
Validity	Does the score interpretation make sense?	Accuracy of interpretation	Whether scores are perfectly error-free
Difficulty	How many got it right?	Ease or hardness of an item	Whether the item is useful by itself
Discrimination	Does the item separate high and low performers?	Item quality	Whether the whole test is valid
Standardisation	Was administration consistent?	Comparability of scores	Whether the test measures the intended trait
Fairness	Is the test unbiased?	Equitable measurement	Whether all group differences disappear

A compact exam answer framework

For a question on test construction, a strong response usually follows this order:

define psychometrics and test construction,
explain construct definition and operationalisation,
describe blueprinting and content coverage,
discuss item writing and response formats,
explain pilot testing and item analysis,
distinguish reliability from validity,
mention standardisation, norms, and ethics,
conclude with the importance of fairness and iterative revision.

This order mirrors the actual process of building a test and gives the answer a logical flow. It also makes it easier to include examples and statistics without losing structure.

Final synthesis

Psychometrics is about making invisible psychological traits measurable in a responsible way. Test construction turns theory into a practical instrument by defining the construct, writing items, piloting the test, analysing item behaviour, and gathering evidence for reliability and validity. The best tests are not merely technically clever; they are conceptually coherent, statistically sound, and ethically defensible. For Stellenbosch PSY343 students, mastering test construction means being able to move fluently between theory, numbers, and application. That ability is exactly what examiners look for: not isolated facts, but an integrated understanding of how psychological measurement works from first principle to final score interpretation.