VALIDITY

The Concept of Validity

The concept of validity occupies a central and foundational position in the entire discipline of
psychological measurement and educational evaluation. It embodies the fundamental
question that underlies all forms of testing and assessment: Does the test actually measure
what it purports to measure? The notion of validity goes beyond mere accuracy or precision
of measurement—it involves the appropriateness, meaningfulness, and usefulness of the
specific inferences drawn from test scores in relation to their intended purpose and
population.
Cohen (2021) defines validity as a judgment or estimate concerning how well a test
measures what it claims to measure in a particular context. This definition highlights two key
ideas: first, that validity is a matter of degree rather than an all-or-none property; and second,
that it is a judgment—a conclusion derived from accumulating evidence rather than an
inherent attribute of a test. Thus, validity does not belong to a test itself but rather to the
interpretations and uses of test scores. For instance, an intelligence test may be valid for
predicting academic performance but invalid for assessing emotional adjustment.
Historically, the evolution of the concept of validity reflects the scientific maturation of
psychological assessment. Early test developers often equated validity with simple
correlation between test scores and observable performance. Over time, however,
psychologists came to understand that validity must incorporate both empirical evidence
and theoretical justification. Modern test theory views validation as a scientific inquiry
into test score meaning (Messick, 1989)—a process that integrates logical reasoning,
statistical evidence, and the practical consequences of test use.
Importantly, validity is not universal or eternal. No test is valid for all purposes,
populations, or circumstances. A test may be valid for assessing cognitive ability in one
cultural or linguistic group but invalid in another where language, educational experience, or
cultural expectations differ. For example, an aptitude test standardized on English-speaking
university students may not maintain the same validity when administered to non-native
English speakers or to adults with limited formal education. Consequently, psychologists and
educators often undertake local validation studies, examining whether the test’s structure,
item content, and predictive relationships hold true within their specific context.
Validation is, therefore, an ongoing, cumulative process—not a single event that can be
finalized once and for all. As societies evolve, technologies change, and populations
diversify, the assumptions underlying test interpretation must be revisited. A test that was
valid twenty years ago may no longer maintain its validity today if the skills, knowledge
bases, or cultural references it measures have shifted.
In essence, validity concerns the truthfulness and defensibility of the conclusions drawn
from test scores. To claim that a test is valid is to assert that it measures the intended
construct accurately and meaningfully, that its scores correlate logically with relevant
external criteria, and that its use produces fair and just outcomes for all individuals assessed.
Face Validity
Among the various forms of validity, face validity is perhaps the most intuitive and easily
understood—yet also the most superficial. It refers to the degree to which a test appears, on
the surface, to measure what it claims to measure, as judged by the test-taker or a casual
observer. Face validity is a matter of perception rather than of empirical verification; it
reflects whether the test “looks right” to those who take or administer it.
Consider, for instance, a test titled “The Assertiveness Inventory” that includes items such as
“I am comfortable expressing disagreement with others” or “I find it easy to say no when
necessary.” To most examinees, these questions clearly relate to assertive behavior, giving the
test a high degree of face validity. On the other hand, a projective test like the Rorschach
Inkblot Test—where individuals describe what they see in ambiguous shapes—may seem to
have little to do with personality assessment, and hence appears low in face validity, even
though it might yield valuable clinical insights when interpreted correctly.
While face validity has no direct psychometric foundation, it holds practical and ethical
importance. Tests with low face validity can provoke anxiety, resistance, or even legal and
ethical challenges. In educational or employment contexts, examinees who perceive a test as
irrelevant or arbitrary may become demotivated or distrustful, leading to reduced
performance or noncompliance. In organizational settings, employees and applicants may
question the fairness of assessment tools that lack apparent relevance to the job.
For example, imagine an employment screening test for a customer service position that
consists largely of abstract reasoning puzzles. Even if the test predicts job performance
statistically, candidates might view it as unrelated to customer service skills, thereby
perceiving it as unfair. Conversely, a situational judgment test that presents realistic customer
interaction scenarios would be seen as face valid, increasing both acceptance and perceived
legitimacy.
However, a test can lack face validity yet still be empirically valid. Many psychological
instruments—particularly projective tests or implicit measures—may appear unrelated to
their intended constructs but have strong empirical validity supported by research.
Conversely, a test may look perfectly relevant but fail to correlate with actual behavior or
outcomes. Therefore, while face validity contributes to test acceptance and cooperation, it
does not provide scientific evidence of measurement accuracy.
In summary, face validity functions as the social dimension of testing—it helps ensure that
the test is understandable, acceptable, and motivating for participants and stakeholders, even
though it does not guarantee psychometric soundness.
Content Validity
Whereas face validity depends on subjective perception, content validity relies on systematic
expert evaluation. Content validity refers to the degree to which the items, questions, or
tasks of a test adequately and representatively sample the entire domain of behavior or
knowledge that the test is intended to measure. It addresses whether the test content covers
the full scope of the construct and whether the sampling of items is balanced and appropriate.
A test with strong content validity demonstrates that the items collectively reflect all
important facets of the construct, neither omitting essential elements nor including irrelevant
material. For example, an end-of-term examination in psychology should include questions
covering all major topics—research methods, biological bases of behavior, cognitive
processes, and social influences—rather than focusing exclusively on one area such as
learning theory. Similarly, an aptitude test for secretarial work should include sections on
typing accuracy, proofreading, and communication skills, not merely spelling or grammar.
Defining and Structuring the Content Domain
The first step in establishing content validity involves defining the content domain, which is
the total universe of behaviors, tasks, or knowledge statements relevant to the construct. The
domain must have clearly defined boundaries (what is included and excluded) and a
structure (how the content areas are organized and weighted).
Murphy and Davidshofer (2005) illustrate this process using an example of a “world history”
test domain. The domain can be structured according to geography (Europe, the Americas,
Asia and Africa) and historical era (18th and 19th centuries), with proportional weights
assigned to each sub-area based on curricular emphasis. A test that samples these areas
proportionately demonstrates stronger content validity than one that disproportionately
focuses on a narrow section of the domain.
In applied psychology, the same principle guides job analysis and test blueprinting. A
content-valid employment test for an air traffic controller might derive from detailed
observation and analysis of job tasks—such as monitoring radar screens, interpreting flight
data, and responding to emergencies. The test items would then sample these functions
systematically to reflect real-world job demands.
Assessing and Quantifying Content Validity
Evaluating content validity typically involves expert judgment rather than statistical analysis.
Panels of subject-matter experts review each test item to determine whether it represents an
essential component of the domain. They may classify items as “essential,” “useful but not
essential,” or “irrelevant.” Lawshe’s (1975) Content Validity Ratio (CVR) provides a
quantitative index of the proportion of experts who consider each item essential. The average
CVR across items can serve as a general indicator of the test’s content validity.
Another widely used approach is the creation of a test blueprint, analogous to an architect’s
plan. The blueprint specifies the topics to be covered, the relative weighting of each topic,
and the number and format of items for each area. In educational settings, curriculum
designers use blueprints to ensure that exams representatively sample the learning objectives
of a course. In industrial-organizational psychology, test blueprints ensure that selection
instruments reflect the critical competencies identified in job analyses.
Content Validity and Reliability
Although content validity and reliability are conceptually distinct, they are closely
intertwined. Reliability refers to the consistency of measurement, whereas content validity
refers to the appropriateness of the material measured. A test may be highly reliable yet
invalid if it consistently measures the wrong content. Conversely, a test that samples the right
content inconsistently may have high content validity but low reliability. Both are essential
for defensible measurement. As Cronbach and colleagues (1972) observed, content validity
studies and reliability studies both rest on the assumption that a test samples a domain of
possible items; they differ only in the emphasis placed on specifying that domain.
In summary, content validity ensures that a test reflects the conceptual and practical
boundaries of the construct it is designed to measure. It provides the foundation upon which
more complex forms of validity—such as criterion-related and construct validity—are built.
Criterion-Related Validity
While content validity concerns the representativeness of test items, criterion-related
validity addresses the empirical relationship between test performance and an external
standard or criterion. It evaluates how well a test predicts or corresponds to an outcome of
practical significance. In simple terms, it asks: Do test scores correlate with actual
performance, behavior, or achievement?
A criterion serves as an external measure of the construct being tested. Examples include
academic grades, supervisor ratings, performance productivity, or clinical improvement
scores. For criterion-related validity to be meaningful, the criterion itself must be reliable,
relevant, and uncontaminated by extraneous factors.
Criterion-related validity is particularly crucial in applied settings such as employment
selection, academic admissions, and clinical diagnosis, where the goal of testing is decision-
making or prediction rather than mere description. Two major subtypes of criterion-related
validity are concurrent validity and predictive validity.
Concurrent Validity
Concurrent validity examines the extent to which test scores are correlated with criterion
measures obtained at approximately the same time. It assesses whether the test reflects
current performance or status. For instance, if a newly developed depression inventory
correlates highly with clinicians’ current diagnostic ratings of depression severity, it exhibits
strong concurrent validity.
In employment contexts, concurrent validity may involve administering a new test to current
employees and correlating their scores with existing performance appraisals. High
correlations suggest that the test measures relevant job-related traits or skills. In clinical
psychology, concurrent validity might be assessed by comparing a new anxiety scale to
established anxiety inventories or behavioral observations collected simultaneously.
The advantage of concurrent validity is its practical efficiency—results can be obtained
quickly without waiting for future outcomes. However, concurrent studies may be limited by
restriction of range (for example, if all employees tested are already competent and
performing well) or by situational variables affecting both measures simultaneously.
Therefore, concurrent validity provides an important but sometimes limited snapshot of test
utility.
Predictive Validity
Predictive validity evaluates how well test scores forecast future performance or outcomes.
It requires a temporal separation between test administration and criterion measurement.
For example, a college entrance examination demonstrates predictive validity if applicants’
scores correlate strongly with their subsequent academic grades or graduation rates.
Predictive validity studies are especially valuable in selection and placement contexts, where
organizations seek to predict future success. Aptitude tests, admission tests (such as the GRE
or SAT), and employment assessments are all validated primarily through predictive studies.
However, predictive validation demands longitudinal data collection, which can be time-
consuming and resource-intensive. Participants must be followed over time, and future
performance data must be accurately recorded. Attrition, changes in circumstances, or
uncontrolled variables can all affect predictive correlations.
For instance, a test of managerial potential might predict performance in leadership roles five
years later. If the correlation remains strong despite time and situational variability, the test
demonstrates robust predictive validity. On the other hand, if situational factors—like
changes in organizational culture—interfere, predictive validity may decline even if the test
itself is technically sound.
In practice, both concurrent and predictive validity studies contribute complementary
evidence. Concurrent validity establishes the test’s relevance to present behavior, while
predictive validity establishes its utility for future outcomes.
Construct Validity
Construct validity is the most comprehensive and theoretically fundamental form of validity.
It addresses whether a test truly measures the abstract construct or psychological attribute it
purports to assess. Constructs are theoretical entities—such as intelligence, motivation,
anxiety, or creativity—that cannot be directly observed but are inferred from behavior,
performance, or self-report patterns.
Construct validity encompasses all other forms of validity, serving as the unifying concept
that integrates content, criterion-related, and other evidential approaches. Establishing
construct validity involves demonstrating that the test behaves in accordance with theoretical
expectations about the construct’s nature, structure, and correlates.
Sources of Evidence for Construct Validity
Construct validation relies on the accumulation of diverse types of evidence:
1. Convergent Evidence: The test correlates positively with other measures of the same
or similar constructs. For instance, a new measure of social anxiety should correlate
strongly with established social anxiety inventories or behavioral avoidance
observations.
2. Discriminant Evidence: The test does not correlate significantly with measures of
unrelated constructs. A social anxiety scale should not correlate with measures of
physical fitness or unrelated personality traits.
3. Factor-Analytic Evidence: Statistical analysis of test items reveals patterns
consistent with theoretical expectations. For example, within an intelligence test,
verbal and quantitative items should load on distinct but related factors, mirroring the
theoretical model of intelligence.
4. Developmental and Group-Differences Evidence: Scores vary across groups or life
stages in ways predicted by theory. For example, self-control scores should increase
 from childhood to adulthood, and anxiety scores should differ between clinical and
nonclinical populations.
5. Experimental Manipulation Evidence: When the construct is theoretically linked to
certain conditions, test scores change accordingly. For example, inducing stress
should elevate anxiety scores, whereas relaxation training should reduce them.
Construct Validity as a Cumulative Process
Construct validation is not a single study but an ongoing program of research. Each new
piece of evidence strengthens or challenges the existing interpretation of the test. Over time, a
network of empirical findings builds what Cronbach and Meehl (1955) called a nomological
net—a system of theoretical relationships among constructs and observable indicators.
Messick (1995) expanded this concept by proposing a unitary view of validity, arguing that
all evidence—content, criterion, and construct—contributes to understanding score meaning.
In this integrated framework, construct validity also encompasses the consequences of test
use, including its social and ethical implications.
For example, the construct validity of an intelligence test depends not only on its correlations
with academic achievement but also on whether it unfairly disadvantages certain cultural
groups or reinforces biased social assumptions. Thus, construct validity merges scientific
accuracy with ethical responsibility.
Validity, Bias, and Fairness
In contemporary testing practice, discussions of validity are inseparable from considerations
of bias and fairness. A test cannot be considered valid if it yields systematically unfair
outcomes for particular groups, even if it is statistically reliable and theoretically well-
constructed.
Test Bias
Test bias occurs when a test systematically underestimates or overestimates the true ability or
trait level of specific groups. Bias can stem from cultural, linguistic, or socioeconomic
differences that influence how test items are interpreted. For example, vocabulary items that
include culturally specific idioms may disadvantage examinees from different linguistic
backgrounds. Similarly, gender-biased scenarios in aptitude tests can influence performance
independent of actual ability.
Psychometricians detect bias through differential item functioning (DIF) analysis or
regression techniques. If test scores predict the criterion differently across groups—say, if the
same score predicts higher job performance for one group than another—then bias exists.
Bias undermines both validity and fairness, as the test no longer measures the intended
construct equivalently across populations.
Test Fairness
Fairness extends beyond the technical detection of bias to encompass the ethical, social, and
procedural equity of testing. A fair test treats all examinees equally, ensuring that differences
in scores reflect genuine differences in the construct, not extraneous influences such as
language proficiency, test anxiety, or socioeconomic status.
Fairness also concerns the interpretation and use of test results. Even a technically valid test
can be unfairly applied if its scores are misused to justify discriminatory decisions or if
accommodations for disabilities are not provided. The Standards for Educational and
Psychological Testing (APA, 2014) emphasize that fairness and validity are inseparable: a test
that is not fair cannot be valid for its intended purpose.
In practice, fairness involves ongoing evaluation of testing procedures, norming samples, and
interpretation frameworks. Test developers must ensure that items are culturally neutral, that
norms are representative of diverse populations, and that interpretations respect contextual
factors. In high-stakes testing, such as college admissions or employment selection, fairness
is not only a psychometric issue but also a social and legal imperative.
Conclusion
Validity remains the cornerstone of psychological and educational assessment, embodying
the scientific, theoretical, and ethical foundations of measurement. A valid test provides more
than consistent scores—it provides meaningful, interpretable, and justifiable information
about human behavior and attributes.
Through face validity, we ensure that tests are accepted and perceived as relevant; through
content validity, that they adequately represent the intended domain; through criterion-related
validity, that they predict or reflect real-world performance; and through construct validity,
that they align with theoretical understanding and empirical evidence.
Yet validity does not end with statistics or theory. It extends into the realms of fairness,
cultural sensitivity, and responsible use. The validity of a test is therefore not a static quality
but a living, evolving judgment, continually refined through empirical research, theoretical
reflection, and ethical scrutiny.
As Cronbach wisely stated, “Validity is not a property of the test or of the scores alone; it is a
property of the meaning of those scores.” Thus, the pursuit of validity is not merely a
technical enterprise—it is an enduring scientific and moral commitment to measure human
capacities truthfully, interpret them fairly, and apply them wisely.