Test Anxiety Scale

Abstract

The Test Anxiety Scale is a type of personality inventory: a self-report instrument on which a respondent rates statements about how they think and feel in examinations, yielding a score for the disposition to react to evaluation with anxiety. Test anxiety is situation-specific, aroused by the threat of being judged rather than felt as a constant trait, and the scales measuring it were built to predict the performance decrement it produces. This article treats the scale as a case study in measuring an evaluative emotion: how the construct was operationalized from Mandler and Sarason's original questionnaire onward, why the reaction resolves into a cognitive worry component and a physiological emotionality component of which worry does most of the work in depressing performance, and how the cognitive-attentional account explains that decrement as task-irrelevant thinking that steals attention from the task.

Keywords: test anxiety, evaluative anxiety, worry and emotionality

The Test Anxiety Scale is the best-known member of a family of self-report instruments built to measure a single, sharply bounded emotion: the anxiety a person feels when their performance is being evaluated. Unlike a general anxiety inventory, which asks how a respondent feels much of the time, a test-anxiety scale asks specifically how they think and feel in examinations and other testing situations, and it does so because test anxiety is a situation-specific reaction — it appears when evaluation is threatened and recedes when it is not. Reading such a scale well means understanding that its target is not a free-floating trait but a disposition to respond to a particular class of situation, and that the reason to measure it at all is that the reaction predicts, and helps explain, why able people sometimes perform below their competence under examination.

That measurement problem was George Mandler and Seymour Sarason's. Working at Yale around 1950, they built the Test Anxiety Questionnaire to index individual differences in the anxiety learners bring to evaluative tasks, and they showed that this anxiety interacts with the situation: it impaired performance most on difficult tasks and under instructions that stressed evaluation (Mandler & Sarason, 1952). The instrument that Irwin Sarason later refined into the Test Anxiety Scale, and the several scales that followed it, all inherit that founding idea — that test anxiety is a reaction worth measuring because it is tied to a situation and to an outcome, the performance it disrupts (Sarason, 1978).

What the Test Anxiety Scale Is

The Test Anxiety Scale is a self-report inventory in which a respondent indicates, item by item, how characteristically they experience worry, tension, and bodily arousal in testing situations, and those responses are summed into a score representing the person's standing on the disposition (Sarason, 1978). The scale is domain-specific by design: its items refer to examinations and evaluative settings, not to life in general, so a high score identifies someone for whom the testing situation in particular is a reliable trigger of anxiety. This is what separates a test-anxiety measure from a broad anxiety inventory such as the Manifest Anxiety Scale, whose items sample a pervasive anxious tendency without tying it to any one class of situation.

Two features fix the construct the scale is built to measure. The first is that test anxiety is best understood as a trait-like disposition to enter an anxious state under evaluation — a standing readiness that is only realized as felt anxiety when a testing situation actually arises. A person high on the scale is not continuously anxious; they are reliably made anxious by evaluation. The second is that the reaction is defined partly by its consequence: the scales earn their keep because the disposition they index predicts a decrement in the very performance being evaluated, so the measurement is never merely descriptive of a feeling but is tied from the outset to an outcome in behaviour (Mandler & Sarason, 1952).

Because the target is situation-specific rather than global, a test-anxiety scale can be more sharply predictive of examination behaviour than a general anxiety inventory, while saying nothing about how the same person feels outside evaluative settings. That trade — narrow scope for sharper prediction within the scope — is the design logic of the whole family of instruments, and it is why several distinct scales, each carving the construct slightly differently, have coexisted rather than collapsing into one (Zeidner, 1998).

Worry and Emotionality: The Two Components

The single most consequential discovery about test anxiety is that it is not one thing but two. Richard Liebert and Larry Morris showed that the global reaction separates cleanly into a worry component and an emotionality component: worry is the cognitive concern about the consequences of failure — the stream of thought about doing badly, disappointing others, and what a poor result will mean — while emotionality is the perceived physiological arousal, the racing heart and tense stomach that accompany the situation (Liebert & Morris, 1967). The two components are correlated, because the same situation provokes both, but they are separable, and they behave differently.

The behavioural difference is what makes the distinction matter. Worry is consistently and more strongly related to depressed performance than emotionality is; a student's cognitive preoccupation with failing predicts their examination decrement better than the intensity of their felt arousal does (Deffenbacher, 1980). Emotionality tends to rise and fall with the immediate stress of the situation, peaking around the examination and subsiding after, whereas worry can precede the test by days and persist through it, occupying cognitive resources the whole time. A measurement built to predict performance therefore has reason to weight the worry items especially heavily, and modern instruments explicitly score the two components apart (Spielberger, 1980).

That worry dominates the association with performance is not merely a psychometric footnote; it is a claim about mechanism, and it points directly at the cognitive account of how test anxiety does its damage. A physiological reaction alone, however unpleasant, does not obviously consume the cognitive machinery a hard test requires. A stream of worry does. The empirical primacy of the worry component thus set up the attentional theory that became the dominant explanation of the test-anxiety–performance link, and it has been repeatedly reaffirmed: contemporary meta-analysis finds the cognitive side of test anxiety more tightly bound to achievement than the physiological side (Roos et al., 2021).

How Test Anxiety Impairs Performance: Cognitive Interference

Why should worrying about a test make one do worse on it? The dominant answer is the cognitive-attentional, or interference, account, whose clearest early statement was Jeri Wine's: the highly test-anxious person divides attention between the task and themselves, so that during an examination they are partly attending to the questions and partly to a running commentary of self-evaluative worry, and the fraction spent on the worry is subtracted from the task (Wine, 1971). On this view the deficit is not caused by arousal disorganizing behaviour but by attention being misallocated — the anxious examinee has less of a limited resource available for the work because some of it is consumed by task-irrelevant thinking.

Irwin Sarason developed this into a general account of cognitive interference, in which the intrusive, self-preoccupied thoughts of the anxious test-taker are the proximate cause of the decrement: thoughts about failing, about time running out, about how others are doing, all compete for the same processing capacity the task demands (Sarason, 1984). The account has a clear signature. Test anxiety should hurt most on tasks that place heavy demands on attention and working memory, where there is little spare capacity to lose, and least on easy or well-learned tasks where ample capacity remains — which is precisely the pattern Mandler and Sarason first observed with task difficulty (Mandler & Sarason, 1952). It also predicts that the damage is done through the working-memory resources the worry occupies, and that interventions redirecting attention back to the task should help.

The interference idea was later formalized as attentional control theory, which locates the damage precisely in working memory's executive control. On this account anxiety preferentially impairs the central executive — in particular its inhibition function, which normally suppresses task-irrelevant information, and its shifting function — so worry is not merely a competitor for capacity but a disruptor of the very system that keeps attention on the task (Eysenck et al., 2007). A key refinement is the distinction between processing efficiency and performance effectiveness: an anxious person can often maintain accuracy by recruiting extra effort and compensatory strategies, so the reaction shows up first as a loss of efficiency — more resources spent for the same result — and only degrades the result itself when the demand outstrips the compensation. This explains why test anxiety bites hardest exactly where the cognitive-interference account says it should: on demanding tasks that leave no spare executive capacity to compensate with.

The interference account also reframes what a test-anxiety scale is measuring when it predicts performance. A high worry score is, on this reading, an index of how much task-irrelevant cognition a person is prone to generate under evaluation, and the scale predicts the decrement because it predicts the interference. This is why the cognitive dimension of the reaction became the focus of both explanation and measurement: the cognitive test-anxiety construct, defined as the characteristic tendency toward interfering thoughts during evaluation, is the single facet most strongly and consistently tied to academic outcomes (Cassady & Johnson, 2002).

The Family of Test Anxiety Measures

There is no single canonical test-anxiety instrument but a lineage of them, each refining the construct and its scoring, and reading the literature requires knowing that the name Test Anxiety Scale denotes both a specific instrument and a category the specific instruments populate. The line begins with the Test Anxiety Questionnaire, a long true–false instrument, and continues through Irwin Sarason's shorter Test Anxiety Scale, which became the widely used general-purpose measure (Sarason, 1978). As the worry–emotionality distinction took hold, instruments were built to score the components separately, most influentially Charles Spielberger's Test Anxiety Inventory with its dedicated Worry and Emotionality subscales (Spielberger, 1980).

Later measures pushed the componential analysis further. Sarason's Reactions to Tests scale split the reaction into four factors — worry, tension, test-irrelevant thinking, and bodily symptoms — separating the pure cognitive-interference facet (test-irrelevant thinking) from the worry about consequences it had previously been folded into (Sarason, 1984). Instruments focused specifically on the cognitive facet followed, on the argument that if worry and interfering thought carry the predictive weight then a scale concentrating on them will predict achievement best (Cassady & Johnson, 2002). Each refinement is a bet about which cut of the construct is worth isolating, and the field has generally bet on the cognitive side.

Across all these instruments the empirical bottom line has been stable and is now very well established. Ray Hembree's early meta-analysis pooled hundreds of studies to fix the basic facts: test anxiety is reliably, negatively, and modestly correlated with performance and ability measures; it is higher in some groups than others; and it can be reduced by treatments that lower worry, with performance improving as it falls (Hembree, 1988). A comprehensive synthesis a decade later consolidated the construct's structure, correlates, and treatment into a single account (Zeidner, 1998), and a thirty-year meta-analytic review has since updated the estimates and mapped the predictors and correlates on modern data (von der Embse et al., 2018).

The Reaction in Motion

The three demonstrations below make manipulable the parts of test anxiety that prose can only describe. The first decomposes a test-anxiety score into its worry and emotionality components and shows that the two contribute unequally to the predicted performance decrement. The second runs the cognitive-interference account, dividing a fixed attentional capacity between the task and task-irrelevant worry. The third exposes the situation-specificity of the reaction, letting evaluative pressure and task difficulty combine to produce a decrement that a neutral setting does not.

Total test anxiety = 30 + 20 = 50. Predicted decrement = 0.6 × 30 + 0.2 × 20 = 22 points.

The same total anxiety predicts a larger decrement when more of it is worry, because worry is weighted more heavily than emotionality. Hold the total fixed and shift it toward worry: the predicted decrement rises. That asymmetry is the empirical signature of the two-component structure.

The worry–emotionality demonstration makes the two-component structure concrete. Setting the worry score and the emotionality score changes the overall level of test anxiety, but the predicted decrement in performance tracks the worry component far more than the emotionality component, because the two are weighted by how strongly each relates to performance. Holding the total fixed while shifting it from emotionality toward worry raises the predicted decrement, which is the empirical signature the demonstration is built around.

On-task attention = 100 − 40 = 60 units, so performance is 60% of what the task requires.

The task never changed; only the allocation did. As worry claims more of the pool, less capacity remains for the work, and performance falls with it. This is why the interference account predicts the largest deficits on demanding tasks, where there was little spare capacity to surrender.

The cognitive-interference demonstration exposes the attentional mechanism. A fixed pool of attentional capacity is divided between task-focused processing and task-irrelevant worry, and raising the intensity of worry shifts capacity away from the task, lowering the performance the remaining on-task attention can support. Watching performance fall as worry claims more of the pool shows why the account predicts the largest deficits on demanding tasks, where there was little spare capacity to surrender.

Evaluative decrement = (70 ÷ 100) × 0.80 × 40 = 22.4 points, so performance = 57.6%.

Switch the instructions to neutral and the decrement vanishes regardless of how test-anxious or how hard the task is: the disposition is realized as a loss only under evaluation. That interaction is the definitional core of a situation-specific anxiety.

The evaluative-stress demonstration shows that test anxiety is a state a situation must trigger. Setting a respondent's dispositional test anxiety, the difficulty of the task, and whether the instructions frame it as an evaluation determines the performance decrement — which stays near zero under neutral, non-evaluative instructions no matter how test-anxious the person is, and grows with anxiety and difficulty only once the situation is made evaluative. The interaction is the definitional core: the disposition is realized as a decrement only under evaluation.

Worked Example

Consider first the decomposition of a score into components. Suppose a respondent scores 30 on the worry component and 20 on the emotionality component, for a total test-anxiety level of 50. If worry and emotionality relate to performance with illustrative weights of 0.6 and 0.2, the predicted decrement is 0.6 × 30 + 0.2 × 20 = 18 + 4 = 22 points. Now hold the total fixed at 50 but redistribute it as 20 worry and 30 emotionality: the predicted decrement becomes 0.6 × 20 + 0.2 × 30 = 12 + 6 = 18 points. The same overall anxiety produces a larger decrement when more of it is worry, which is exactly the empirical asymmetry the two-component structure records.

Now run the interference account. Suppose attentional capacity is a fixed pool of 100 units and a respondent's worry claims 40% of it as task-irrelevant processing. On-task attention is then 100 − 40 = 60 units, and if performance is proportional to on-task attention the respondent works at 60% of the capacity the task requires. Raise worry so it claims 65% of the pool and on-task attention falls to 35 units, a performance of 35% — the same task, a worse result, entirely because attention was reallocated rather than because the task changed.

Finally, read the situational interaction. Suppose baseline performance is 80%, and under evaluative instructions the decrement is (anxiety ÷ 100) × difficulty × 40. For a highly test-anxious respondent (anxiety 70) on a hard task (difficulty 0.8), the decrement is (70 ÷ 100) × 0.8 × 40 = 0.7 × 0.8 × 40 = 22.4 points, so performance falls to about 58%. Under neutral instructions the evaluative decrement is zero and performance stays at 80%, however test-anxious the respondent is — the disposition costs nothing until the situation calls it out.

Discussion

The Test Anxiety Scale earns its place in cognitive psychology as a worked example of measuring a situation-specific emotion and tying that measurement to a cognitive mechanism. Mandler and Sarason did not set out to index a general anxious trait; they built an instrument for the anxiety of being evaluated, and they demonstrated from the start that its predictive value lay in an interaction with the situation — the disposition mattered most where evaluation was stressed and the task was hard (Mandler & Sarason, 1952). The subsequent history of the construct is the history of getting its internal structure right: the worry–emotionality split showed that a single score conceals two reactions that behave differently (Liebert & Morris, 1967), and the primacy of worry pointed measurement and theory alike toward the cognitive side of the emotion.

The strengths and limits of the construct follow from its situation-specificity. Because the scales target evaluation in particular, they can predict examination behaviour more sharply than a global anxiety measure, and they connect cleanly to an attentional account of why the anxiety impairs performance — worry as task-irrelevant thinking that consumes capacity a hard test needs (Wine, 1971). But the same narrowness means a test-anxiety score speaks only to the evaluative domain, and it raises a persistent question about the construct's boundaries: how far test anxiety is a distinct entity as opposed to a domain-restricted expression of a more general anxiety proneness. The relationship between test anxiety, trait anxiety, and the clinical anxiety disorders is genuinely unsettled, and recent work asks directly whether these are manifestations of the same underlying construct or different ones (Putwain et al., 2021).

The construct's durability is legible from its measurement history. The basic facts Hembree fixed by meta-analysis — a modest, reliable, negative relationship with performance, reducible by treatments that lower worry — have held up across decades and instruments (Hembree, 1988), even as the scales were repeatedly refined to isolate the cognitive facet that carries the predictive weight. Each refinement was a claim about which cut of the reaction is worth measuring, and the field's convergence on the worry and cognitive-interference facets is a substantive result about the emotion, not merely a psychometric preference. A measure is best understood by seeing what its construct was built to predict and how the measurement was reshaped as the mechanism came into focus.

Current Directions

Test anxiety remains an active research construct, and the questions now asked of it cluster around two themes: its physiological basis and its relationship to clinical anxiety. The first theme revisits the emotionality component with better methods, asking how tightly perceived and actual physiological arousal are coupled and how much either relates to performance. A systematic review and meta-analysis of test anxiety and physiological arousal found the association between the two to be weaker and more variable than the self-report emotionality literature had assumed, reinforcing the conclusion that the cognitive worry component, not the physiological one, carries the burden of the performance relationship (Roos et al., 2021). The emotionality that respondents report is only loosely tied to what their bodies are doing.

The second theme concerns the construct's boundaries. As test anxiety is increasingly studied in school and clinical settings together, the question of whether high test anxiety is continuous with the anxiety disorders — or a distinct, situation-bounded phenomenon — has become pressing for both classification and treatment. Recent work directly compares test anxiety, diagnosable anxiety disorders, and school-related wellbeing to ask whether they are manifestations of the same construct or different ones, with implications for whether test anxiety should be treated as an educational problem, a clinical one, or both (Putwain et al., 2021). The thirty-year meta-analytic review that maps the construct's modern predictors and correlates provides the empirical backdrop against which these boundary questions are now being settled (von der Embse et al., 2018).

Common Misconceptions

Test anxiety is just general anxiety showing up at exam time.

It is a situation-specific disposition. A person can be high in test anxiety without being generally anxious, and the scales predict examination behaviour precisely because they target the evaluative situation rather than a global trait (Zeidner, 1998).

The racing heart and nerves are what wreck the performance.

The physiological emotionality component is more weakly related to performance than the cognitive worry component; it is the worry and interfering thought, not the felt arousal, that carries most of the decrement (Liebert & Morris, 1967).

Test anxiety lowers scores by disorganizing the anxious person's behaviour.

The dominant account is attentional, not one of arousal-driven disorganization: worry consumes limited attention as task-irrelevant thinking, leaving less capacity for the task itself (Wine, 1971).

A single test-anxiety score captures the whole reaction.

The reaction is multi-component. Modern instruments deliberately score worry and emotionality — and sometimes tension and test-irrelevant thinking — apart, because a single total conceals facets that relate to performance very differently (Sarason, 1984).

Glossary

Cognitive interference.

The intrusion of task-irrelevant, self-preoccupied thoughts during a task, held to be the proximate cause of the test-anxiety performance decrement by diverting attention from the task.

Cognitive test anxiety.

The characteristic tendency to experience interfering, worry-laden thoughts during evaluation; the facet of test anxiety most strongly tied to academic performance.

Emotionality.

The physiological component of test anxiety: the perceived autonomic arousal, the racing heart and tense stomach, felt in an evaluative situation.

Evaluative anxiety.

Anxiety aroused specifically by the prospect of one's performance being judged; the broader category of which test anxiety is the schoolroom and examination instance.

Reactions to Tests (RTT).

Sarason's scale scoring test anxiety as four factors, worry, tension, test-irrelevant thinking, and bodily symptoms, separating the cognitive-interference facet from worry about consequences.

Self-report inventory.

A questionnaire scored from a respondent's own answers about their behaviour, feelings, or attitudes, as opposed to observer ratings or performance tasks.

Situation-specific anxiety.

Anxiety tied to a particular class of situation rather than felt pervasively; test anxiety is the situation-specific anxiety of being evaluated.

State anxiety.

A transient, situation-bound episode of anxiety, the felt reaction that a test-anxious disposition produces when an evaluation actually occurs.

Task-irrelevant thinking.

Thought during a task that is unrelated to performing it, worry about failing, about time, about others, that consumes attentional capacity the task requires.

Test Anxiety Inventory (TAI).

Spielberger's self-report scale that scores overall test anxiety together with separate Worry and Emotionality subscales.

Test Anxiety Questionnaire (TAQ).

Mandler and Sarason's original 1952 instrument, the progenitor of the test-anxiety scales, indexing individual differences in the anxiety brought to evaluative tasks.

Test anxiety.

The disposition to react with worry and physiological arousal in evaluative or testing situations, and the resulting state; measured by the test-anxiety scales.

Trait anxiety.

A stable, general disposition to experience anxiety across many situations, distinct from the situation-specific disposition that a test-anxiety scale measures.

Worry.

The cognitive component of test anxiety: concern about the consequences of failure, which carries most of the association with impaired performance.

Key Researchers

Nathaniel P. von der Embse. School psychologist at the University of South Florida; lead author of the thirty-year meta-analytic review that mapped the modern predictors and correlates of test anxiety. ORCID - Google Scholar - Faculty Page

Irwin G. Sarason (1929-2025). University of Washington psychologist who refined the Test Anxiety Scale and developed the general account of cognitive interference, framing intrusive self-preoccupied thought as the proximate cause of the test-anxiety performance decrement. Wikidata - In Memoriam

Seymour B. Sarason (1919-2010). Yale psychologist who, with George Mandler, built the Test Anxiety Questionnaire and established that evaluative anxiety interacts with task difficulty and instructions to impair performance. Wikipedia

Charles D. Spielberger (1927-2013). University of South Florida psychologist who developed the Test Anxiety Inventory with its separate Worry and Emotionality subscales, embedding the two-component structure directly into measurement. Wikipedia

Moshe Zeidner. Professor emeritus of educational psychology at the University of Haifa; author of the comprehensive synthesis of test-anxiety theory, structure, correlates, and treatment. Faculty Page

Frequently Asked Questions

What does the Test Anxiety Scale measure?
It measures test anxiety, a situation-specific disposition to react with worry and physiological arousal when one's performance is being evaluated, from a respondent's self-report about how they think and feel in testing situations (Sarason, 1978).

How is test anxiety different from general anxiety?
Test anxiety is tied to a particular class of situation, evaluation, rather than felt pervasively; a person can be high in test anxiety without being generally anxious, which is why a test-anxiety scale predicts examination behaviour more sharply than a global anxiety inventory (Zeidner, 1998).

What are the worry and emotionality components?
Worry is the cognitive concern about the consequences of failure, and emotionality is the perceived physiological arousal of the situation; they are correlated but separable, and worry is more strongly related to impaired performance (Liebert & Morris, 1967).

Why does test anxiety lower performance?
The dominant explanation is attentional: worry generates task-irrelevant thinking that consumes limited attention, leaving less capacity for the task, so the decrement is largest on demanding tasks where little spare capacity remains (Wine, 1971).

Which component matters more for performance?
The cognitive worry component carries most of the association with impaired performance, while the physiological emotionality component is more weakly related, a finding that pointed measurement toward the cognitive facet (Deffenbacher, 1980).

Is there one Test Anxiety Scale or several?
Several. The term names both Sarason's specific instrument and a family of scales, running from the Test Anxiety Questionnaire through the Test Anxiety Inventory and the Reactions to Tests scale, each carving the construct slightly differently (Sarason, 1984).

Does test anxiety really affect achievement?
Yes, reliably but modestly. Meta-analysis pooling hundreds of studies finds test anxiety negatively correlated with performance and ability, and reducible by treatments that lower worry (Hembree, 1988).

Is test anxiety the same as an anxiety disorder?
This is unsettled. Recent work asks directly whether test anxiety, the clinical anxiety disorders, and school wellbeing are manifestations of the same construct or different ones, with consequences for how test anxiety should be classified and treated (Putwain et al., 2021).

References