Temperament is measured by systematically recording individual differences in emotional reactivity, activity, attention, approach, inhibition, and self-regulation. Researchers use age-appropriate caregiver questionnaires, structured laboratory tasks, observations in everyday environments, and, for some research questions, physiological indicators and self-reports. Each method samples a different part of a person's behavior. A sound assessment therefore begins with an explicit definition of the temperament dimension, the child's developmental stage, the setting in which behavior is observed, and the inference that a score is intended to support. A questionnaire describes patterns reported across occasions; a laboratory task records responses to a particular standardized challenge; a home observation documents behavior in a familiar social ecology. These sources can complement one another without producing identical rankings. Gartstein and Ramirez Gonzalez, 2025; Stifter, Willoughby, and Towe-Goodman, 2008.
Temperament measurement is especially consequential in infancy and childhood because children have a rapidly changing behavioral repertoire and ordinarily cannot explain the internal states that adult personality questionnaires ask respondents to describe. The familiar words easy, difficult, shy, sensitive, and active are too broad to serve as measurements by themselves. A research instrument must specify which actions count, when they occur, who reports them, how often they are sampled, and how scores are compared. A high score describes an observed or reported tendency under the instrument's assumptions; it is not a psychiatric diagnosis, a fixed identity, or a direct reading of neural circuitry. Gartstein and Rothbart, 200300169-8); Planalp et al., 2017.
What Does a Temperament Assessment Actually Measure?
Reactivity, regulation, and developmental context
In a major contemporary framework, temperament consists of constitutionally based individual differences in reactivity and self-regulation, influenced by heredity, maturation, and experience. Reactivity includes how quickly, intensely, and for how long someone responds emotionally, motorically, or attentively to stimulation. Self-regulation includes processes that modulate such responses, including attention deployment and the development of effortful control. Researchers operationalize these constructs through observable frequencies, intensities, durations, latencies, and ratings rather than relying on the label temperament as though it were one measurable substance. Rothbart, 2007; Gartstein and Ramirez Gonzalez, 2025.
For example, fear may be studied through hesitancy, facial expression, vocalization, gaze, body movement, proximity seeking, and latency to approach an unfamiliar person or object. Activity level may be coded as the frequency and vigor of movements within age-appropriate circumstances. Attentional orienting may be indexed by the time an infant maintains attention toward a stimulus. These are indicators from which a dimension is inferred, and individual behaviors can have multiple explanations. A quiet infant might be alert, tired, ill, absorbed, or inhibited. The researcher's task is to distinguish a recurrent disposition from the immediate conditions producing one act. Planalp et al., 2017.
A temperament dimension is typically continuous. Children can differ by degree along the same dimension, and their relative standing can change as they develop. The three broad domains often examined in the Rothbart tradition—negative affectivity, surgency or positive emotionality, and regulatory capacity or effortful control—are organizing constructs assembled from narrower scales. They are not universal diagnostic categories, and the precise factor solution can vary with instrument, age group, and population. Gartstein and Rothbart, 200300169-8); Bosquet Enlow et al., 2016.
Temperament, momentary state, and personality
A child crying during one unfamiliar procedure displays a state or episode. A child's recurring tendency to respond intensely across multiple relevant situations may support an inference about a temperament characteristic. The distinction is statistical and developmental: a pattern must be sampled repeatedly enough to separate recurring individual differences from fatigue, hunger, novelty, caregiving, and chance. A temperament score is therefore a summary of observations within a specified window, not a prediction of every future response. Planalp et al., 2017.
Personality assessment overlaps with temperament research when studying emotional stability, sociability, and self-control across development. Yet adult personality traits, early temperament dimensions, and psychiatric constructs are measured with different item content and validation evidence. Translating one score into another requires direct empirical justification. An infant questionnaire should not be interpreted using adult personality norms, and an adult inventory should not be presented as a test of infant temperament. Evans and Rothbart, 2007; Rothbart, 2007.
The Main Ways Researchers Measure Temperament
Caregiver-report questionnaires
Caregivers complete sets of behaviorally anchored items concerning the child's reactions during a defined recent period. Well-constructed questions describe familiar events, such as the child's response when an activity is interrupted or when an unfamiliar adult approaches, and ask how often a specified behavior occurred. The respondent may also mark an event as inapplicable when the situation did not arise. Such items make it possible to collect information about repeated behavior in many settings with relatively modest research costs. They are especially practical for large longitudinal samples and for narrow dispositions that may rarely appear during one laboratory visit. Gartstein and Rothbart, 200300169-8).
The caregiver simultaneously serves as an observer and a source of interpretation. Knowledge accumulated over months can improve coverage, but the report can reflect the caregiver's expectations, mood, standards, memory, relationship with the child, and opportunities to witness a behavior. Parents may also observe different behaviors at different times. Consequently, caregiver report has substantive value while requiring independent evidence on reliability, agreement, factor structure, and associations with external outcomes. Parade and Leerkes, 2008; Stifter, Willoughby, and Towe-Goodman, 2008.
Structured laboratory observation
Laboratory assessment presents comparable stimuli to participants and records their responses. A fear episode might introduce an unfamiliar object; an anger or frustration episode might temporarily interrupt an engaging activity; a positive-affect episode might involve playful interaction. Trained coders can independently rate latency, intensity, duration, facial expressions, vocalizations, and motor behavior. The Laboratory Temperament Assessment Battery, or Lab-TAB, is a prominent example of this family of methods. Its use allows specified situations to be reproduced and its behavioral definitions to be examined systematically. Planalp et al., 2017.
Standardization improves comparability within an episode, but a laboratory response remains a response to that episode. unfamiliarity with the room, separation from a caregiver, recent sleep, developmental ability, and the order of tasks can affect performance. A single task has limited coverage of a disposition expected to appear across situations. Researchers therefore strengthen inference by using several episodes, repeated assessments, documented coding rules, and analysts who are masked to relevant hypotheses when feasible. Planalp et al., 2017.
Naturalistic observation at home or in childcare
Naturalistic observation examines behavior during ordinary feeding, play, caregiving, transitions, peer contact, and other everyday activities. An observer may record behavior continuously, sample it at predetermined intervals, or apply global ratings after a documented observation period. Home and childcare observations supply ecological context that a fixed experimental task cannot fully reproduce. They can reveal whether a child's responses differ around familiar adults, peers, routines, or unfamiliar demands. Stifter, Willoughby, and Towe-Goodman, 2008.
Natural settings introduce unequal opportunities for behavior. A child cannot demonstrate a response to unfamiliar peers on a day with no peer interaction; a child may appear calm because the caregiver proactively prevents frustration. Observer presence can modify family behavior, and a short visit samples only a small portion of ordinary life. Comparisons require a careful record of what happened, whose behavior was available for coding, and whether participants encountered comparable situations. A low frequency of an event is not automatically evidence of low trait intensity. Stifter, Willoughby, and Towe-Goodman, 2008.
Self-report, teacher report, and physiological measures
With increasing age, children and adolescents can report some internal reactions and self-regulatory experiences, while teachers and other adults can assess behavior in settings that parents rarely observe. Adult temperament questionnaires, including the Adult Temperament Questionnaire, describe a related but developmentally different approach. These instruments complement child caregiver measures; their dimensions, interpretation, and validity must be evaluated on their own terms. Evans and Rothbart, 2007.
Physiological recordings such as heart rate, respiratory sinus arrhythmia, electrodermal activity, or measures of stress-related systems may provide evidence about arousal or regulatory processes in a particular protocol. They measure physiological responses, not an entire temperament trait. The same physiological signal can be associated with multiple experiences and is affected by age, physical activity, respiration, measurement conditions, and other factors. Associations with behavioral temperament dimensions require convergent validation rather than an assumption of one-to-one correspondence. Contemporary reviews describe physiological and biological indicators as complementary evidence within a broader multi-method research program. Gartstein and Ramirez Gonzalez, 2025.
Major Temperament Questionnaires by Developmental Stage
The Infant Behavior Questionnaire–Revised (IBQ-R)
The IBQ-R was designed for infants approximately 3 to 12 months old. Its original full form contains 191 items organized into 14 narrower scales, including dimensions such as activity level, distress to limitations, fear, sadness, smiling and laughter, approach, soothability, and duration of orienting. Instead of asking whether the infant is generally difficult, items refer to specific actions in recent everyday situations. In the original development study, parents of 360 infants contributed data across three age bands. The authors reported satisfactory internal consistency for 14 retained scales and moderate agreement across caregivers on some ratings. Gartstein and Rothbart, 200300169-8).
Researchers sometimes use short or very short IBQ-R versions to reduce respondent burden. A shorter form sacrifices some item-level detail, and the very short form emphasizes broad factors rather than every narrow dimension. Choosing the version is a design decision: an inquiry about specific soothing processes may call for a fuller scale than a population survey examining broad temperament structure. Form length alone cannot establish measurement precision or validity in a particular sample. A recent historical review describes the original, short, and very short versions and places them within a larger developmental measurement program. Gartstein and Ramirez Gonzalez, 2025.
A study of six-month-old infants found adequate internal consistency for the IBQ-R subscales and mother–father agreement for 11 of 14 subscales, alongside limited convergence with laboratory observations. The extent of concordance was related to caregiver depressive symptoms. These findings support the questionnaire's usefulness for some dimensions while showing why internal reliability alone cannot establish that a parent rating and an experimental observation are interchangeable. Parade and Leerkes, 2008.
The Early Childhood Behavior Questionnaire (ECBQ)
The ECBQ is intended for toddlerhood, approximately 18 to 36 months. Its development study specified 18 fine-grained caregiver-report dimensions and examined samples with cross-sectional and longitudinal information. The proposed higher-order structure included Surgency/Extraversion, Negative Affectivity, and Effortful Control. Across the study's scales, the authors described adequate internal consistency, generally moderate caregiver agreement, and evidence of longitudinal stability. This instrument addresses behaviors that become observable as mobility, language, social expectations, and voluntary regulation develop. Putnam, Gartstein, and Rothbart, 2006.
The boundary between infant and toddler instruments reflects changes in what children can do. The ability to wait, redirect attention, inhibit a prepared response, or manage competing instructions develops over time. Researchers should therefore select items that are possible and meaningful at the child's age. Apparent score differences caused by switching instruments cannot by themselves establish that the underlying disposition changed. A well-designed longitudinal study specifies its construct correspondence and may administer overlapping measures at transitional ages. Putnam, Gartstein, and Rothbart, 2006.
The Children's Behavior Questionnaire (CBQ)
The CBQ evaluates fine-grained temperament dimensions in early and middle childhood, commonly across ages 3 through 7 or 8 depending on form and research protocol. Its items address emerging attentional control, inhibitory capacities, emotions, activity, and approach or withdrawal within familiar situations. Rothbart and colleagues analyzed the instrument's structure and measurement properties in a foundational study, while subsequent work produced shorter versions for research contexts in which the full questionnaire is impractical. Rothbart et al., 2001; Putnam and Rothbart, 2006.
The CBQ short form contains 94 items organized into 15 scales, and its very short form contains 36 items corresponding to three broad dimensions. The short-form development and evaluation study reported that psychometric properties varied across subgroups, with lower internal consistency on some scales among African American and lower-income samples. This is a concrete reminder that a scale's published reliability is a property of scores observed in specified populations, not a certificate guaranteeing equivalent performance everywhere. Putnam and Rothbart, 2006.
Adult temperament questionnaires
Adult temperament research relies more extensively on self-reports about habitual attention, emotion, sensitivity, motivation, and self-control. Evans and Rothbart examined the Adult Temperament Questionnaire in an undergraduate sample and a community sample. Their factor analyses yielded somewhat different higher-order solutions and showed overlap with personality dimensions. A questionnaire written for adults can therefore be informative about adult temperament constructs while requiring separate validation from childhood scales and from measures of personality disorders. Evans and Rothbart, 2007.
Other temperament traditions emphasize different dimensional systems. Their measures should be selected according to the construct definition and the empirical question, not because all questionnaires using the word temperament are assumed to evaluate the same thing. Cross-theory comparisons need documented overlap in dimensions, item content, and criterion associations. This is particularly important when an instrument combines temperament with character, behavior problems, sensory traits, or clinical screening indicators. Evans and Rothbart, 2007; Gartstein and Ramirez Gonzalez, 2025.
How Laboratory Temperament Assessment Works
Standardized episodes and behavioral coding
A structured protocol defines the eliciting situation before the participant enters the laboratory. It specifies the stimulus, duration, caregiver position, task sequence, allowable prompting, stopping conditions, and the behavior to record. A recording may then be divided into time intervals. Coders evaluate the occurrence, intensity, or latency of behaviors such as vocal distress, avoidance, approach, smiling, struggling, or attentional persistence. Multiple behaviors may be combined into an episode score, and multiple episodes may contribute to a broader composite when supported by a stated measurement model. Planalp et al., 2017.
The Lab-TAB literature illustrates the distinction between episode-level behavior and an inferred underlying disposition. In a study of 1,076 infants assessed at six and twelve months, researchers evaluated the measurement properties of laboratory episodes and compared scores with caregiver-reported temperament. Laboratory measures showed some temporal stability while their association with parental ratings was generally modest. This supports the practical utility of structured observation without making any one task a definitive behavioral standard. Planalp et al., 2017.
Observer training and reproducible scoring
Direct observation depends on human classification unless automated procedures have themselves been validated. A rigorous protocol defines a facial expression or movement precisely enough that trained observers can apply it consistently. Researchers can double-code a subset of recordings, quantify agreement, monitor drift over time, and document disagreements. Reliability is evaluated at the score level and in relation to the statistical method used; simple percentage agreement is insufficient for many continuous or low-base-rate behaviors. The goal is a reproducible measurement procedure whose limits are explicit. Planalp et al., 2017.
Blinding can reduce expectation effects. A coder who knows the child's questionnaire score or developmental outcome may, even unintentionally, search for supporting behaviors. Masking those details where practicable helps separate observation from interpretation. Protocols must also document inter-episode timing and the possibility that an earlier frustration task influences a later task. A later fear response can reflect accumulated distress as well as the intended stimulus. Planalp et al., 2017.
Why one dramatic response is weak evidence
A vivid episode is psychologically compelling and scientifically underdetermined. A child may react strongly to one mask, toy, separation, sound, or unfamiliar adult because of a transient state, previous experience, or sensory context. Temperament concerns relative tendencies across appropriate observations. Several observations, within-person variation, and replication across settings yield a more defensible estimate than a single encounter. Researchers can also distinguish the probability of a response from its magnitude when it occurs. Stifter, Willoughby, and Towe-Goodman, 2008; Planalp et al., 2017.
Parent Reports and Direct Observation: Why Do Scores Disagree?
They sample different evidence
A questionnaire aggregates the caregiver's impressions of weeks of behavior, while a structured task samples minutes in a limited context. Parents may witness behavior during feeding, bedtime, sibling interactions, illness, and play. The laboratory may deliberately elicit novelty or frustration that occurs infrequently at home. Agreement should be strongest when the two procedures capture sufficiently similar dimensions under comparable circumstances. Even then, overlapping method-specific variance can remain. Divergence does not automatically invalidate either method; it identifies a question about what each method captures. Stifter, Willoughby, and Towe-Goodman, 2008.
Stifter and colleagues investigated a sample of 955 infants using standard caregiver report, ratings made during extended home visits, and coding of reactions in specific tasks. A model incorporating both trait and method factors fit their data better than models attributing all variation exclusively to one source. Agreement was more evident for infant positivity than for negativity. The conclusion is specific: different methods can contain common construct information while preserving distinctive perspectives and biases. It does not follow that every disagreement is harmless or that every procedure is equally valid for every use. Stifter, Willoughby, and Towe-Goodman, 2008.
Informant agreement is dimension-specific
Agreement between parents, between caregivers and teachers, or between parents and observers can vary substantially across activity, fear, approach, negative affect, and regulation. Behavior with high visibility and frequent opportunities may be easier to rate consistently than internal or rarely observed processes. A caregiver who sees the child when tired and another who sees the child during structured group activities may describe different but genuine patterns. Informant disagreements should be analyzed in relation to context and item coverage rather than averaged away without explanation. Parade and Leerkes, 2008; Stifter, Willoughby, and Towe-Goodman, 2008.
Parent mood, expectations, and the relationship
Caregiver depressive symptoms, stress, beliefs about acceptable behavior, and comparisons with siblings or peers can shape responses. In a six-month IBQ-R study, the association between caregiver report and observation varied with caregiver depressive symptoms. Such findings establish that reporter characteristics may influence scores; they do not imply that caregivers experiencing distress are intrinsically unreliable or that their reports lack substantive information. A sound design measures relevant reporter characteristics where the hypothesis calls for them and considers more than one reporting source. Parade and Leerkes, 2008.
Relationship effects are part of the phenomenon. A child may exhibit different regulation when comforted by a familiar adult, placed in a group, or approached by a stranger. A parent who routinely adapts the environment can reduce the frequency of frustration while caring effectively for a reactive child. A simple behavioral frequency therefore reflects both the child's disposition and the opportunities created by the environment. Interpreting a score sensibly requires knowledge of the caregiving ecology in which its observations were produced. Stifter, Willoughby, and Towe-Goodman, 2008.
Reliability: Are Temperament Scores Reproducible?
Internal consistency and item coverage
Internal consistency concerns whether items intended to represent a scale tend to vary together in the assessed sample. It cannot demonstrate that all items validly measure the proposed disposition. A narrowly repetitive set of questions may correlate highly while covering little of the construct. Conversely, a broad multidimensional definition may produce more heterogeneous items. Reliability statistics must be interpreted along with the instrument's content, score structure, item distribution, and the intended conclusion. The foundational IBQ-R, ECBQ, and CBQ papers provide empirical information about the reliability of scores obtained with their respective forms. Gartstein and Rothbart, 200300169-8); Putnam, Gartstein, and Rothbart, 2006.
Test–retest stability and developmental change
Repeated testing asks whether individuals retain some relative position over time. In children, this test must distinguish measurement error from genuine maturation and changing opportunities for behavior. A toddler's capacity for voluntary inhibition is different from an infant's regulatory repertoire, and both mean-level development and rank-order change can occur. A correlation between two waves indicates relative consistency within the sampled group; it does not mean the group's average score remained constant or that every child maintained their rank. Longitudinal analyses need age-appropriate measures and careful evidence of construct continuity. Putnam, Gartstein, and Rothbart, 2006; Gartstein and Ramirez Gonzalez, 2025.
Inter-rater reliability and observer agreement
Inter-rater reliability for coded observations addresses whether independent observers apply the same rules to the same behavior. Mother–father or parent–teacher agreement answers a somewhat different question because informants may witness different behavior. It should not be treated as a pure estimate of coder skill. A study can report reliable coders and nevertheless find low caregiver–observer correlations, reflecting distinct situations and measurement formats. Distinguishing those comparisons prevents misleading claims that an instrument is either proved or disproved by one coefficient. Planalp et al., 2017; Parade and Leerkes, 2008.
Validity: Does a Score Support the Intended Interpretation?
Content and construct validity
Content validity asks whether items and situations adequately represent the defined dimension for the participant's developmental level and relevant contexts. A scale described as emotional reactivity should cover the relevant reactions without conflating them indiscriminately with disease, parenting difficulties, or social desirability. Construct validity is accumulated through theoretical analysis and empirical findings involving structure, relations with other variables, developmental patterns, and alternative explanations. No single validity coefficient establishes every future application of a measure. Gartstein and Rothbart, 200300169-8); Bosquet Enlow et al., 2016.
Convergent, discriminant, and predictive evidence
Convergent validity examines whether an instrument relates to measures expected to assess a shared dimension. Discriminant evidence asks whether it can distinguish theoretically different characteristics. Predictive associations test whether a score is related to later outcomes under specified conditions. A statistically significant association with later adjustment does not by itself establish causal influence or justify a deterministic prognosis for an individual child. Where outcome measurement relies on the same caregiver, shared informant variance can inflate associations. Strong designs diversify data sources, specify temporal ordering, and control plausible alternative explanations. Stifter, Willoughby, and Towe-Goodman, 2008; Gartstein and Ramirez Gonzalez, 2025.
Factor structure and comparability of scales
Statistical factors are patterns of covariance among observed items or scores. A three-factor description can be useful, but it is a model whose fit may depend on how questions are phrased and who participates. A diverse U.S. study of the IBQ-R concluded that the widely used factor arrangement required adjustments in its sample. The finding argues for local validation and careful invariance testing before interpreting group score differences as substantive temperament differences. Bosquet Enlow et al., 2016.
A German study of the IBQ-R likewise found a factor pattern that did not fully reproduce the proposed original higher-order configuration. This does not render all cross-cultural comparisons impossible; it changes the empirical burden. Translation quality, item meaning, response distributions, and equivalence of factor loadings and thresholds may need evaluation before averages from different populations are placed on one interpretive scale. Vonderlin, Ropeter, and Pauen, 2012; Bosquet Enlow et al., 2016.
The Risks of Short Forms and Broad Scores
Very short questionnaires reduce cost and fatigue. They also make a narrow component less visible and can change the precision of individual scores. A survey designed to estimate broad group differences may work adequately with three composite domains yet provide too little information to support claims about an individual child's fear, frustration, or attentional shifting. Shortening a scale should be guided by its validation study, the intended population, and the statistical resolution required for the research question. Putnam and Rothbart, 2006.
A second risk is interpreting a composite as a homogeneous psychological process. Broad negative affectivity can include distinguishable reactions such as sadness, frustration, and fear. The child who avoids novelty and the child who protests interruption may obtain similar broad scores through different configurations. Whenever a research claim concerns a narrow behavior or mechanism, the relevant lower-order evidence should be examined rather than inferred automatically from the broad score. Gartstein and Rothbart, 200300169-8).
Age, Development, and Longitudinal Assessment
An infant's attentional control, a toddler's ability to follow rules, and a school-age child's self-regulation are developmentally related but operationally different. A measurement system needs to acknowledge these transitions. The changing capacity to move independently, talk about feelings, tolerate frustration, and respond to adult instructions transforms what researchers can observe. Age-appropriate items protect against mistaking normal developmental constraints for unusually low regulation or high emotional reactivity. Gartstein and Ramirez Gonzalez, 2025.
Longitudinal programs can use overlapping assessments, harmonized narrow constructs, repeated observational episodes, and statistical models of measurement invariance. Researchers should report exactly when questionnaires changed and whether the correspondence among scores was established empirically. A simple numerical comparison of a six-month IBQ-R score, a three-year ECBQ score, and an adult questionnaire score lacks meaning when scales have different items and norms. An apparent developmental trajectory can partly reflect changing measurement. Putnam, Gartstein, and Rothbart, 2006; Evans and Rothbart, 2007.
Measurement occasions also matter. A caregiver report during an episode of acute illness or family disruption may represent a real but temporarily altered pattern. Recording health, caregiving, sleep, environment, and major contextual shifts makes subsequent interpretation stronger. The fact that children develop does not erase stable individual differences; it means developmental stability and change must be measured as distinct quantities. Gartstein and Ramirez Gonzalez, 2025.
Culture, Language, and Population Representation
Temperament behaviors are embedded in cultural routines and expectations. The frequency of encountering unfamiliar adults, the meaning of assertive behavior, norms surrounding emotional expression, and opportunities for independent exploration may differ across social contexts. A translated item can be grammatically faithful while sampling a behavior that occurs at a different rate because of local practices. Validity therefore includes whether the situations described in the instrument are recognizable, representative, and comparably interpreted by respondents. Bosquet Enlow et al., 2016.
Socioeconomic background, respondent education, home language, disability access, and recruitment channels can influence both participation and instrument functioning. A scale developed in a comparatively homogeneous sample warrants reexamination when applied to a substantially different population. The CBQ short-form study found differences in internal consistency across demographic groups; the IBQ-R factor study likewise raised questions about structural comparability in a diverse U.S. sample. Responsible group comparisons require documented measurement equivalence and careful interpretation of any remaining differences. Putnam and Rothbart, 2006; Bosquet Enlow et al., 2016.
A group mean says little about a particular child without information about within-group variability, measurement error, and overlapping distributions. Researchers should avoid describing cultural or demographic groups as essentially endowed with a single temperament. Cross-cultural findings become informative when sampling, translation, invariance, family context, and alternative explanations are reported transparently. Gartstein and Ramirez Gonzalez, 2025.
Common Sources of Measurement Error and Bias
Recall, response tendencies, and missing opportunities
Caregiver questionnaires usually depend on retrospective judgments across a reference period. Salient events may be remembered more vividly than mundane behavior. Some respondents favor extreme ratings; others avoid them. Many instruments allow a not applicable response because a child may not have encountered a particular event. Treating missing opportunities as evidence that the target behavior did not occur can bias the resulting score. Researchers should document missingness, item response options, how scales handle inapplicable items, and whether the scoring algorithm was followed. Gartstein and Rothbart, 200300169-8).
Observer expectancy and context effects
A coder may be influenced by knowledge of parental reports, experimental condition, or the desired outcome. Task order, experimenter behavior, camera position, ambient noise, and the caregiver's visibility can affect observable responses. Predefined coding manuals, standardized stimuli, reliable double coding, and masking where feasible strengthen observation. Yet standardized laboratory settings do not remove all situational effects; they make many of them easier to identify and analyze. Planalp et al., 2017.
Shared-method variance and selection bias
When the same adult rates both temperament and later behavioral difficulties, a correlation can partly reflect shared response style or interpretation. When research samples disproportionately include caregivers who have the time and resources to complete long surveys, findings may not generalize to families with different circumstances. Selective attrition across longitudinal waves introduces further risks. Study quality depends on recruitment transparency, missing-data handling, method independence, and replication—not merely the apparent precision of statistical estimates. Stifter, Willoughby, and Towe-Goodman, 2008; Bosquet Enlow et al., 2016.
How to Design a Strong Multi-Method Temperament Study
A well-designed study starts with a specific research question. To evaluate broad temperamental negative affectivity in infancy, investigators may combine age-appropriate caregiver ratings with multiple structured emotional episodes. To study behavior in childcare, they may add teacher observations or repeated naturalistic recordings. The choice of informants and settings follows the construct and the intended inference. A narrowly targeted experiment requires a different protocol from a population cohort seeking representative developmental trajectories. Planalp et al., 2017; Gartstein and Ramirez Gonzalez, 2025.
The sampling plan should describe the participant age range, recruitment frame, caregiver arrangements, observation contexts, and timing. Protocols should specify the number and selection of episodes, instrument versions and licenses, observer training, handling of missing or inapplicable ratings, and preregistered analyses when appropriate. Multiple measurement methods permit investigators to model method-specific and shared variance. The Stifter study demonstrates that representing both sources explicitly can reveal why seemingly contradictory reports coexist. Stifter, Willoughby, and Towe-Goodman, 2008.
Researchers should examine score distributions, internal consistency where meaningful, coder agreement, test–retest stability, factor structure, convergent evidence, and relationships with independent outcomes. They should state the population and occasion to which these analyses apply. When predicting later behavior, external validation samples and calibration matter: a relationship in a group does not guarantee accurate identification of an individual outcome. A multi-method design improves triangulation only when the methods are interpreted in light of their distinct strengths, biases, and sampled contexts. Parade and Leerkes, 2008; Gartstein and Ramirez Gonzalez, 2025.
What Temperament Scores Can and Cannot Tell Parents and Practitioners
A temperament score can help describe a child's characteristic way of responding to novelty, frustration, activity demands, or attentional challenges. It may help adults notice regularities and consider how routines, expectations, or contexts relate to a child's experience. The scientific interpretation remains conditional on the quality of the measure and the setting. A questionnaire completed by one caregiver can be a valuable account of behavior without constituting a comprehensive individual evaluation. Parade and Leerkes, 2008.
Temperament is part of normative psychological variation. High emotional reactivity, behavioral inhibition, strong activity, or low effortful control on a questionnaire does not itself establish a disorder or a need for treatment. Clinical assessment, when indicated by persistent impairment or other concerns, requires its own validated procedures and developmental history. Instruments designed for atypical behavior screening or developmental risk have distinct purposes and should not be substituted casually for dimensional temperament scales. For example, the Temperament and Atypical Behavior Scale explicitly addresses indicators of atypical self-regulatory and developmental functioning, which differs from a general characterization of ordinary temperamental variation. Assessment of behavioral and emotional problems in infancy, 2015.
Parents also need an appropriately scaled interpretation of prediction. A research association between childhood inhibition and later difficulty describes a pattern across a population. The probability for any one child depends on multiple interacting developmental and environmental influences and on how risk was estimated and validated. Scores should therefore be communicated as measured tendencies with uncertainty, not as personality verdicts or irreversible futures. Gartstein and Ramirez Gonzalez, 2025.
Scientific Disagreements and the Current Evidence
Established findings
Several widely used questionnaires have developmental validation studies showing interpretable dimensions and useful consistency under specified conditions. Structured observations can produce reproducible measurements of emotional responses to controlled eliciting events. Parent report and observational measures frequently capture overlapping but nonidentical variance. Developmental appropriateness, multiple informants, and careful specification of constructs remain central principles supported by converging evidence. Gartstein and Rothbart, 200300169-8); Planalp et al., 2017; Gartstein and Ramirez Gonzalez, 2025.
Findings with important limits
The exact higher-order factor structure of temperament instruments varies across some samples and translations. Parent–observer agreement differs by dimension, population, and protocol. Short forms can preserve broad information while losing fine-grained coverage, and their precision is not uniform across groups. These are empirical limitations with direct implications for instrument selection and score interpretation. They do not justify discarding all questionnaires or treating laboratory behavior as an unambiguous representation of the person. Bosquet Enlow et al., 2016; Putnam and Rothbart, 2006; Stifter, Willoughby, and Towe-Goodman, 2008.
Continuing questions
Disagreements persist about the precise boundaries among temperament, early personality, regulatory skills, and situational effects; about the comparability of factor models across ages and cultures; and about how to combine physiological indicators with behavior and report data. Machine-coded movement, voice, or facial information can become scientifically useful only after demonstrations of validity, fairness, and robustness in appropriate developmental samples. The weight of evidence should follow the specific instrument and inference, rather than claims that any emerging technology directly detects temperament. Gartstein and Ramirez Gonzalez, 2025.
Frequently Asked Questions
Can a parent accurately assess an infant's temperament?
Caregivers can provide informative reports because they witness repeated everyday behavior. The agreement with laboratory observation varies across temperament dimensions and can depend on caregiver characteristics and measurement context. A caregiver questionnaire is best understood as one valuable source of data whose quality must be evaluated, not as an infallible judgment or a mere personal opinion. Parade and Leerkes, 2008; Stifter, Willoughby, and Towe-Goodman, 2008.
What is the most widely used infant temperament questionnaire?
The IBQ-R is among the prominent research measures for infancy, with age-specific subscales and several available lengths. Whether it is the most appropriate instrument depends on the infant's age, research purpose, target dimensions, language, and available validation evidence. Its name should not be used as a generic substitute for every temperament measure. Gartstein and Rothbart, 200300169-8); Gartstein and Ramirez Gonzalez, 2025.
Is a laboratory observation more objective than a questionnaire?
A laboratory protocol can standardize stimulation and document visible behavior more directly. It still relies on operational definitions, coder decisions, and a narrow set of eliciting contexts. A questionnaire is susceptible to reporter effects but covers behavior across many occasions. Method choice is determined by the research question, and independent sources may improve interpretation. Planalp et al., 2017.
Do different temperament tests measure the same traits?
They may share theoretically related dimensions, but questionnaires developed for infants, toddlers, older children, and adults use different behavioral indicators. Even instruments within a common research tradition can vary in factor structure across samples. Comparability requires evidence of overlapping content and measurement equivalence. Putnam, Gartstein, and Rothbart, 2006; Bosquet Enlow et al., 2016.
Can temperament be measured in one visit?
A single visit can measure responses to tasks administered during that visit. Inferring a stable temperament dimension requires additional evidence connecting those responses to characteristic behavior across occasions, reporters, or later assessments. Several episodes and multiple methods strengthen that inference. Planalp et al., 2017.
Does a high temperament score diagnose a mental disorder?
No. Temperament questionnaires generally characterize dimensional psychological differences. A diagnosis involves separate clinical criteria, impairment, developmental context, and appropriate professional assessment. A screening tool for atypical behavior has a different function from a research questionnaire describing normative temperament. Assessment of behavioral and emotional problems in infancy, 2015.
How can scientists tell whether temperament changes with age?
Repeated assessments can reveal mean-level change and rank-order stability, but only after researchers consider age-related changes in items, informants, and the child's behavioral capacities. Longitudinal inference is stronger when instruments overlap or are linked using empirical evidence rather than raw score resemblance. Gartstein and Ramirez Gonzalez, 2025.
Related Articles
Cluster Personality Traits, Temperament and Individual Differences — The broad scientific chapter connecting temperament, personality traits, development, and individual-differences research.
Rothbart's Temperament Model: Reactivity and Effortful Control — The theoretical model underlying several age-specific temperament questionnaires.
Infant Temperament: Reactivity, Self-Regulation and Early Development — Early individual differences and the developmental context of infant measurement.
References
Bosquet Enlow, M., White, M. T., Hails, K., Cabrera, I., and Wright, R. J. (2016). The Infant Behavior Questionnaire–Revised: Factor structure in a culturally and sociodemographically diverse sample in the United States. Infant Behavior and Development, 43, 24–35. https://doi.org/10.1016/j.infbeh.2016.04.001
Evans, D. E., and Rothbart, M. K. (2007). Developing a model for adult temperament. Journal of Research in Personality, 41(4), 868–888. https://doi.org/10.1016/j.jrp.2006.11.002
Gartstein, M. A., and Ramirez Gonzalez, M. A. (2025). Temperament development in infancy: What we have learned about the origins of individual differences in the past 25 years. Infant Behavior and Development, 80, 102101. https://doi.org/10.1016/j.infbeh.2025.102101
Gartstein, M. A., and Rothbart, M. K. (2003). Studying infant temperament via the Revised Infant Behavior Questionnaire. Infant Behavior and Development, 26(1), 64–86. https://doi.org/10.1016/S0163-6383(02)00169-800169-8)
Parade, S. H., and Leerkes, E. M. (2008). The reliability and validity of the Infant Behavior Questionnaire–Revised. Infant Behavior and Development, 31(4), 637–646. https://doi.org/10.1016/j.infbeh.2008.07.009
Planalp, E. M., Van Hulle, C., Gagne, J. R., and Goldsmith, H. H. (2017). The infant version of the Laboratory Temperament Assessment Battery (Lab-TAB): Measurement properties and implications for concepts of temperament. Frontiers in Psychology, 8, 846. https://doi.org/10.3389/fpsyg.2017.00846
Putnam, S. P., Gartstein, M. A., and Rothbart, M. K. (2006). Measurement of fine-grained aspects of toddler temperament: The Early Childhood Behavior Questionnaire. Infant Behavior and Development, 29(3), 386–401. https://doi.org/10.1016/j.infbeh.2006.01.004
Putnam, S. P., and Rothbart, M. K. (2006). Development of short and very short forms of the Children's Behavior Questionnaire. Journal of Personality Assessment, 87(1), 102–112. https://doi.org/10.1207/s15327752jpa8701_09
Rothbart, M. K. (2007). Temperament, development, and personality. Current Directions in Psychological Science, 16(4), 207–212. https://doi.org/10.1111/j.1467-8721.2007.00505.x
Rothbart, M. K., Ahadi, S. A., Hershey, K. L., and Fisher, P. (2001). Investigations of temperament at three to seven years: The Children's Behavior Questionnaire. Child Development, 72(5), 1394–1408. https://doi.org/10.1111/1467-8624.00355
Stifter, C. A., Willoughby, M. T., and Towe-Goodman, N. (2008). Agree or agree to disagree? Assessing the convergence between parents and observers on infant temperament. Infant and Child Development, 17(4), 407–426. https://doi.org/10.1002/icd.584
Vonderlin, E., Ropeter, A., and Pauen, S. (2012). Assessment of temperament with the Infant Behavior Questionnaire Revised (IBQ-R): The psychometric properties of a German version. https://doi.org/10.1024/1422-4917/a000187