Causal Dimension Scale II

Abstract

The Causal Dimension Scale II (CDSII) is a prominent psychometric instrument developed by Edward McAuley, Terry E. Duncan, and Dan Russell in 1992. Designed to evaluate how individuals perceive and interpret the underlying causes of achievement outcomes, life events, and behavioral performances, the CDSII represents a major theoretical and empirical refinement of the original Causal Dimension Scale (CDS; Russell, 1982). Grounded in Bernard Weiner’s attribution theory, the CDSII assesses four distinct causal properties: Locus of Causality, External Control, Stability, and Personal Control. The scale comprises 12 items formatted as 9-point semantic differential scales, with three items dedicated to each of the four dimensions. Total scores for each subscale range from 3 to 27, where higher scores signify higher perceived internality, stability, personal controllability, or control by other agents, respectively. Across diverse populations in sport and exercise psychology, academic performance research, clinical rehabilitation, and organizational behavior, the CDSII demonstrates robust psychometric properties, including confirmed construct validity via confirmatory factor analysis (CFA), high test-retest reliability, and internal consistency coefficients generally ranging between α = .67 and α = .92 across subscales. By disentangling personal volitional agency from external agentic power, the CDSII overcomes the psychometric shortcomings of its predecessor and remains one of the gold-standard measures of subjective causal attributions in psychological science.

Keywords

Causal Dimension Scale II, CDSII, Attribution Theory, Bernard Weiner, Locus of Causality, Personal Control, External Control, Stability, Psychometrics, Sport Psychology

Authors

The Causal Dimension Scale II was developed and validated by:

• Edward McAuley, Ph.D. — Department of Kinesiology, Community Health, and Beckman Institute for Advanced Science and Technology, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA.
• Terry E. Duncan, Ph.D. — Senior Research Scientist, Oregon Research Institute, Eugene, Oregon, USA.
• Dan W. Russell, Ph.D. — Department of Human Development and Family Studies, Institute for Social and Behavioral Research, Iowa State University, Ames, Iowa, USA (formerly of the College of Medicine, University of Iowa).

Purpose

The foundational purpose of the Causal Dimension Scale II (CDSII) is to provide an objective, reliable, and standardized methodology for measuring how an individual cognitively characterizes the perceived causes of specific outcomes. In classical attribution research, investigators often committed the fundamental methodological error of operationalizing attributions through “researcher-assigned” meanings. For instance, an investigator might classify an athlete’s attribution of “poor coaching” as an external, unstable, and uncontrollable cause. However, to the athlete, the coaching quality may be perceived as an enduring, stable characteristic of the team environment, or perhaps even something over which teammates have direct control. The CDSII sidesteps this “fundamental attribution researcher error” by asking respondents to state their perceived cause in their own words and subsequently rate that idiosyncratic cause across standardized structural dimensions.

In applied and clinical settings, the CDSII serves as an indispensable diagnostic and monitoring instrument. In sport and exercise psychology, the instrument assesses how competitive athletes conceptualize victory and defeat. Athletes who ascribe failures to internal, stable, and uncontrollable factors (e.g., lack of innate talent) are at acute risk for depressive symptoms, reduced self-efficacy, and drop-out. Conversely, athletes who attribute failure to internal, unstable, and personally controllable factors (e.g., specific technical preparation or strategic effort) exhibit resilient achievement motivation and higher subsequent persistence. In academic and educational psychology, the CDSII is frequently used to evaluate student reactions to standardized testing, academic setbacks, and classroom achievement, informing educational interventions and cognitive re-attribution training.

Beyond performance domains, the CDSII has substantial utility in health psychology, physical rehabilitation, and clinical behavioral medicine. Patients coping with chronic illness, cardiac rehabilitation, stroke recovery, or traumatic physical injury routinely formulate causal explanations for their condition or functional recovery trajectories. Measuring whether a patient perceives rehabilitation progress as governed by Personal Control versus External Control (e.g., physical therapists, medical personnel) directly predicts treatment compliance, self-regulatory agency, and psychological wellbeing. Consequently, the primary rationale of the CDSII is to operationalize attributional structures with psychometric precision across experimental, observational, and intervention studies.

Psychological Construct

The Causal Dimension Scale II measures the multidimensional cognitive structure of causal attributions. When an event occurs—particularly an unexpected, negative, or ego-involved outcome—human beings automatically generate causal explanations answering the fundamental question: “Why did this happen?” The CDSII does not measure the manifest verbal content of the cause itself (such as “effort,” “fatigue,” “weather,” or “luck”), but rather assesses the underlying latent properties of that cause along four distinct psychometric dimensions:

1. Locus of Causality

The Locus of Causality dimension assesses the degree to which the perceived cause originates from within the individual versus outside in the social or physical environment. This dimension reflects the foundational distinction originally posited by Fritz Heider (1958) and Julian Rotter (1966). Items 1, 6, and 9 measure this construct:

• Internal Locus: The cause reflects personal attributes, dispositional characteristics, emotional states, physiological capacities, or actions unique to the self (e.g., personal stamina, personal tactical execution).
• External Locus: The cause is perceived as belonging to situational context, environmental obstacles, weather conditions, or the actions of competitors (e.g., slick playing surface, judging bias).

In Weiner’s attributional model, locus of causality is directly tied to affective self-esteem and pride: attributing success internally elicits pride and self-worth, whereas attributing failure internally can trigger shame or guilt depending on controllability.

2. Stability

The Stability dimension captures the temporal permanence of the perceived cause over time. Measured by Items 3, 7, and 11, it reflects whether the causal agent is enduring or fluctuating:

• Stable: The cause is perceived as permanent, invariant, and resistant to change across repeated instances (e.g., biological predisposition, persistent organizational policy).
• Unstable / Variable: The cause is considered episodic, fluctuating, temporary, and subject to rapid alteration (e.g., transient acute fatigue, momentary lapse in concentration, fleeting stroke of fortune).

According to cognitive attribution models, stability is the primary determinant of future outcome expectancies. If an outcome is attributed to a stable cause, the individual expects the identical outcome to recur under similar circumstances; if attributed to an unstable cause, expectations of future success or failure remain fluid.

3. Personal Control

The Personal Control dimension captures the extent to which the causal factor is volitionally regulatable, manageable, and modifiable by the actor themself. Measured by Items 2, 4, and 10, this factor addresses the individual’s self-regulatory mastery:

• High Personal Control: The cause is within the actor’s power to manipulate or direct (e.g., deliberate preparatory effort, tactical decisions, practice schedule).
• Low Personal Control: The cause is beyond the personal jurisdiction or willpower of the individual (e.g., genetic body composition, sudden physiological reflex failure).

Personal control is closely tied to feelings of personal responsibility, empowerment, and learned helplessness. High personal control over failure buffers against hopelessness, promoting proactive coping.

4. External Control

The External Control dimension reflects the extent to which other people or outside agents possess regulatory power over the identified cause. Assessed by Items 5, 8, and 12, this dimension separates interpersonal or institutional control from chaotic randomness:

• High External Control: Other people (such as coaches, referees, teammates, physicians, or opposing players) can regulate, dictate, or influence the causal factor.
• Low External Control: No other person holds mastery, command, or regulatory jurisdiction over the cause (e.g., natural events, random chance).

Distinguishing external control from personal control was the primary psychometric advance of the CDSII over the original CDS, resolving long-standing ambiguities regarding how social support, coaching intervention, and shared team responsibility operate in causal reasoning.

Theoretical Framework

The CDSII is theoretically anchored in the attributional formulation of human motivation developed by Bernard Weiner (1979, 1985, 1986). Weiner synthesized and extended earlier conceptualizations by Fritz Heider (the father of phenomenological attribution theory) and Julian Rotter (who popularized internal versus external locus of control). Weiner argued that human beings act as intuitive naive scientists, perpetually constructing causal interpretations of events to make their environments predictable and manageable.

Weiner originally proposed a three-dimensional taxonomy: Locus, Stability, and Controllability. In this model, these three cognitive dimensions mediate the relationship between an objective outcome (e.g., winning, losing, passing, failing) and subsequent psychological outcomes, including emotional responses, future expectancies, and behavioral persistence:

• The Cognitive-Affective Link: Locus influences self-directed emotions (pride, self-esteem), whereas controllability influences social emotions (pity, anger, guilt, shame).
• The Expectancy Principle: Stability governs changes in expectancy. If a failure is attributed to a stable cause, the expectancy of future failure approaches certainty. If failure is attributed to an unstable cause, hope for future success is maintained.
• The Behavioral Outcome: Expectancy and affect together determine achievement strivings, persistence in the face of adversity, and performance intensity.

In 1982, Dan Russell developed the original Causal Dimension Scale (CDS) to empirically evaluate Weiner’s three dimensions. However, across a decade of empirical research, serious psychometric and conceptual anomalies emerged. Most notably, Russell’s original Controllability subscale aggregated items assessing whether a cause was “controllable by you or other people.” This conflation created severe construct contamination. For instance, in team sports or clinical settings, a cause might be completely outside the personal control of an individual player but fully under the control of the coach or referee. When respondents rated such causes, the original controllability items yielded ambiguous intermediate scores, poor factor loadings, and depressed internal consistency.

To resolve this theoretical and statistical dilemma, McAuley, Duncan, and Russell (1992) conducted extensive studies demonstrating that controllability is inherently bifactorial: it must bifurcate into Personal Control (internal agency) and External Control (agentic power exerted by other entities). By formally separating personal mastery from social/external control, the CDSII reconciled Weiner’s attribution theory with contemporary social-cognitive constructs such as Albert Bandura’s self-efficacy theory and proxy agency paradigms.

Validity

The CDSII has undergone extensive empirical validation across multiple domains, demonstrating exceptional construct, convergent, discriminant, and predictive validity.

Construct Validity and Factorial Independence

In the seminal validation studies conducted by McAuley, Duncan, and Russell (1992), construct validity was rigorously evaluated across four independent empirical samples totaling over 800 participants engaged in collegiate physical education classes, competitive intercollegiate basketball, and academic examinations. Confirmatory factor analyses demonstrated that the hypothesized four-factor model provided an outstanding fit to the data, outperforming alternative one-, two-, and three-factor models. Specifically, the four dimensions demonstrated distinct latent identities, with inter-factor correlations remaining modest to moderate (generally |r| < .35), confirming that Locus of Causality, Personal Control, External Control, and Stability represent distinct cognitive constructs.

Convergent and Discriminant Validity

Convergent validity has been established by correlating CDSII subscales with established psychological inventories measuring generalized control expectancies and affective reactions. As predicted by Weiner’s model, scores on the Personal Control subscale correlate positively with Bandura’s Task Self-Efficacy Scale (r = .42 to .58) and Rotter’s Internal Locus of Control, while exhibiting negative correlations with measures of learned helplessness and perceived chronic stress. Locus of Causality demonstrates strong convergent associations with post-performance pride following success (r = .45 to .62). Discriminant validity is supported by the clear psychometric separation between Personal Control and External Control: across multiple validation samples, these two factors showed near-zero to weak orthogonal correlations (r = −.08 to .14), demonstrating that personal agency is conceptually and statistically distinct from control perceived to reside in other people.

Predictive and Criterion Validity

The CDSII exhibits strong predictive validity with regard to future cognitive, emotional, and behavioral outcomes. In athletic competitions, McAuley and colleagues demonstrated that athletes who scored high on Stability following a defeat exhibited sharp declines in future success expectancies, whereas those scoring high on Personal Control maintained elevated persistence, practice attendance, and objective athletic improvement across subsequent competitive cycles. In physical therapy and exercise adherence studies, cardiac rehabilitation patients with elevated Personal Control scores exhibited significantly higher adherence to exercise regimens at 6- and 12-month follow-ups compared to those who ascribed recovery to External Control or Unstable factors.

Reliability

The CDSII demonstrates robust internal consistency and temporal stability across diverse populations and cultural adaptations.

Internal Consistency

In the original validation studies by McAuley, Duncan, and Russell (1992), Cronbach’s alpha coefficients were computed across four separate samples, yielding the following reliability ranges across the three-item subscales:

• Locus of Causality: α = .67 to .80
• External Control: α = .78 to .82
• Stability: α = .67 to .73
• Personal Control: α = .79 to .85

Subsequent psychometric investigations in sport, exercise, and academic domains have consistently mirrored these parameters, with composite reliabilities frequently exceeding .75. Although the Stability subscale occasionally exhibits internal consistency coefficients near .67—partially attributable to the brevity of a 3-item subscale and the distinct nuances of temporal invariance versus contextual repeatability—the overall scale reliability satisfies standard psychometric thresholds for research and applied assessments.

Test-Retest Stability

Assessing test-retest reliability for attributional instruments requires methodological caution because causal attributions are inherently dynamic, event-dependent states rather than static personality traits. However, when individuals are asked to evaluate the same stable past event across a two-week interval without intermediate interventions, test-retest intraclass correlation coefficients (ICCs) range from .71 to .84 across all four subscales, confirming excellent temporal stability when the referent event remains constant.

Factor Analysis

The psychometric transition from the original CDS to the CDSII was driven by rigorous structural equation modeling and factor analytic investigations. Russell’s (1982) original three-factor formulation routinely exhibited poor fit when tested using confirmatory methods, primarily due to cross-loadings and large residual covariances associated with items tapping the original “controllability” construct.

Confirmatory Factor Analytic Findings

McAuley et al. (1992) conducted competing model comparisons using Confirmatory Factor Analysis (CFA) with maximum likelihood estimation across multiple independent datasets. The tested models included:

1. A single-factor general attribution model.
2. A two-factor model combining Locus/Control versus Stability.
3. A three-factor model replicating the original Russell (1982) framework (Locus, Stability, Controllability).
4. The proposed four-factor CDSII model (Locus of Causality, Stability, Personal Control, External Control).

The four-factor CDSII model consistently produced superior fit indices across all samples. Typical model fit indices reported in foundational and subsequent structural studies include:

• Goodness-of-Fit Index (GFI): .94 to .97
• Comparative Fit Index (CFI): .93 to .96
• Root Mean Square Error of Approximation (RMSEA): .042 to .058
• Standardized Root Mean Square Residual (SRMR): .048 to .061
• χ² / df ratio: Typically between 1.45 and 2.10, indicating excellent fit.

Factor Loadings

Standardized factor loadings across the four latent dimensions are consistently strong and statistically significant (p < .001). Across validation studies, the factor loading ranges for each dimension are:

• Locus of Causality (Items 1, 6, 9): λ = .64 to .84
• Personal Control (Items 2, 4, 10): λ = .68 to .88
• Stability (Items 3, 7, 11): λ = .58 to .78
• External Control (Items 5, 8, 12): λ = .72 to .86

These robust loadings confirm that each 3-item subscale cleanly captures its intended latent attributional construct without problematic cross-loadings.

Instrument / Measurement Tool

• Instrument Name: Causal Dimension Scale II (CDSII)
• Authors: Edward McAuley, Terry E. Duncan, & Dan Russell
• Publication Date: 1992
• Target Population: Adolescents and adults across sport, academic, organizational, and clinical rehabilitation settings
• Format: Paper-and-pencil questionnaire or computerized self-report inventory
• Administration Procedure:
  1. Event Identification: The respondent identifies a specific achievement event, outcome, or performance (e.g., “My performance in today’s tennis match”).
  2. Open-Ended Attribution: The respondent writes down the primary reason or reasons for that outcome in their own words (e.g., “I didn’t execute my backhand properly”).
  3. Dimensional Rating: The respondent rates that self-generated cause across the 12 semantic differential items.
• Number of Items: 12 items
• Response Scale: 9-point semantic differential scale (ranging from 9 to 1 with distinct bipolar anchors for each item)
• Subscales & Scoring Formulas:
  • Locus of Causality: Sum of Items 1 + 6 + 9 (Score range: 3 – 27)
  • External Control: Sum of Items 5 + 8 + 12 (Score range: 3 – 27)
  • Stability: Sum of Items 3 + 7 + 11 (Score range: 3 – 27)
  • Personal Control: Sum of Items 2 + 4 + 10 (Score range: 3 – 27)
• Interpretation: Higher scores represent greater internality, greater perceived control by others, higher temporal stability, and higher personal control over the cause, respectively.

Permissions & Fee and Test Year

The Causal Dimension Scale II was published in 1992 in the Personality and Social Psychology Bulletin by the Society for Personality and Social Psychology (SPSP), published by SAGE Publications. The instrument was developed with public and academic institutional support and is widely available in the public academic domain for non-commercial educational, clinical, and research purposes.

Researchers and practitioners may utilize and administer the CDSII without payment of licensing fees, provided that appropriate bibliographic citation is given to McAuley, Duncan, and Russell (1992). For commercial software deployment, commercial assessments, or for-profit publications, permissions should be formally cleared through SAGE Publications or the Copyright Clearance Center.

References

• Heider, F. (1958). The Psychology of Interpersonal Relations. John Wiley & Sons. https://doi.org/10.1037/10628-000
• McAuley, E., Duncan, T. E., & Russell, D. W. (1992). Measuring causal attributions: The revised Causal Dimension Scale (CDSII). Personality and Social Psychology Bulletin, 18(5), 566–573. https://doi.org/10.1177/0146167292185006
• Rotter, J. B. (1966). Generalized expectancies for internal versus external control of reinforcement. Psychological Monographs: General and Applied, 80(1), 1–28. https://doi.org/10.1037/h0092976
• Russell, D. (1982). The Causal Dimension Scale: A measure of how individuals perceive causes. Journal of Personality and Social Psychology, 42(6), 1137–1145. https://doi.org/10.1037/0022-3514.42.6.1137
• Weiner, B. (1979). A theory of motivation for some classroom experiences. Journal of Educational Psychology, 71(1), 3–25. https://doi.org/10.1037/0022-0663.71.1.3
• Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological Review, 92(4), 548–573. https://doi.org/10.1037/0033-295X.92.4.548
• Weiner, B. (1986). An Attributional Theory of Motivation and Emotion. Springer-Verlag. https://doi.org/10.1007/978-1-4612-4948-1

13. Items of the Scale (Questionnaire)