The TRAMS: The Team-Referent Attributions Measure in Sport

Objectives: To provide initial evidence for the construct, concurrent, and predictive validity of the TeamReferent Attributions Measure in Sport (the TRAMS). Design: Cross-sectional in Studies 1 and 2, and multiple time points in Study 3. Method: Study 1 required participants (N 1⁄4 500) to complete the TRAMS for their “least successful” and “most successful” performances in the preceding three months. In Study 2, after performance, participants (N 1⁄4 515) completed the TRAMS and the Causal Dimension Scale for Teams (CDS-T; Greenlees et al., 2005). Study 3 required participants (N 1⁄4 165) to complete a measure of pre-competition collectiveefficacy prior to performance (Day 1, Time 1), the TRAMS following performance (Day 1, Time 2), and a measure of subsequent collective-efficacy prior to subsequent performance (Day 7e9, Time 3). Results: Study 1 supported the factor structure of the TRAMS across least successful and most successful conditions. Study 2 provided further support for the factor structure of the TRAMS, together with evidence of concurrent validity with subscales of the CDS-T. Study 3 revealed, following team defeat, interactions between controllability and generalisability dimensions: Controllability had a significant effect upon subsequent collective-efficacy when causes of team defeat were also perceived to generalise across situations and/or across teams. Following team victory, stable attributions were positively associated with subsequent collective-efficacy. Conclusions: This article provides initial evidence for the validity of the TRAMS and demonstrates for team-referent attributions the theoretical advantages of examining a broader conceptualisation of generalisability attributions and interactive effects of attributions. © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). Team-referent attributions refer to team members' individual explanations for collective outcomes. The attributions made for group outcomes are proposed to have an important role in the affective, cognitive, physiological and behavioural responses of group members (Allen, Coffee, & Greenlees, 2012). For example, the explanations we form for group outcomes influence emotions (e.g., pride or shame), efficacy for future performances (i.e., confidence to perform in the future), hormones (e.g., elevation or reduction in cortisol and testosterone levels), and subsequent behaviour (e.g., increased or reduced involvement). A central premise within attribution research is that there is a dimensional structure underpinning the reasons people give for their failures and successes. The Attribution Theory of Achievement Motivation (ATAM; e.g., Weiner, 1985) considers three primary fax: þ44 (0)1786 466477. offee), [email protected] Ltd. This is an open access article u attribution dimensions: locus of causality, stability, and controllability. Locus of causality refers to the extent to which causes are seen as either residing within or outside an individual. Stability refers to the extent to which causes are seen as either stable or variable. Controllability refers to the extent to which causes are seen as regulated by individuals/teams or something over which control cannot be exerted. For example, an attribution to “poor genetics” is usually (but not always) categorised as internal, stable, and uncontrollable. An important advancement in the measurement of teamreferent attributions, based upon the ATAM model, was the development of the Causal Dimension Scale for Teams (CDS-T; Greenlees, Lane, Thelwell, Holder, & Hobson, 2005). This self-assessment, situation-specific questionnaire, derived from the Revised Causal Dimension Scale (CDS-II; McAuley, Duncan, & Russell, 1992), requires teammembers to identify the main cause of a team outcome and then to rate that cause along a series of bipolar items that correspond to four attribution dimensions: locus of causality, stability, team control (control by the team), and external control nder the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/). P. Coffee et al. / Psychology of Sport and Exercise 16 (2015) 150e159 151 (control by others). To develop the CDS-T, Greenlees and colleagues made two major amendments to the CDS-II: First, the authors reworded each item to reflect a team rather than a self-referent attribution; the word “you” was replaced with “your team” throughout the revised questionnaire. Second, the authors generated four new items to increase the item pool for each attribution dimension from three items to four items. Greenlees et al. reported good overall fit for the four-factor CDS-T (c2(98) 1⁄4 210.21; RMSEA 1⁄4 .05, p > .05; NNFI 1⁄4 .95; CFI 1⁄4 .96) with coefficient alpha reliabilities ranging from .74 to .82. There were, however, some concerns about low factor loadings for items with six factor loadings below .70 and one factor loading below .50. The CDS-T has subsequently been used to explore the correlates of team-referent attributions in sport settings (Allen, Jones, & Sheffield, 2009; Chow & Feltz, 2008; Dithurbide, Sullivan, & Chow, 2009; Greenlees et al., 2007; Shapcott, Carron, Greenlees, & El Hakim, 2010). For example, Greenlees et al. demonstrated that attributions following success weremore likely to be perceived as internal, stable, and controllable, and Dithurbide et al. reported evidence that collective efficacy was higher when causes of prior performance were considered less stable. Allen and colleagues explored interactive effects of attributions on collective-efficacy and reported, following team defeat, an interaction for external control and stability, and, following team victory, an interaction for team control and stability. Across conditions, the nature of the interactions was the same: If causes were perceived as stable, higher levels of control (team controldfollowing team defeatdand external controldfollowing team victory) were positively associated with subsequent collective-efficacy. Despite the progress in the team-referent attribution literaturedafforded through the development of the CDS-Tdtheoretical advancements are limited by the lack of an alternative validated measure to the CDS-T for assessing situation-specific team-referent attributions in sport. This reasoning is underpinned by the following two key points: First, the CDS-T suffers from the same conceptual andmeasurement issues inherent in its parentmeasure, the CDS-II. In reference to the CDS-II, it has been noted that the assessment of personal (team, for the CDS-T) and external control is not congruent with the ATAM framework and that respondents have considerable problems interpreting scale anchors (Biddle & Hanrahan, 1998; Biddle, Hanrahan, & Sellars, 2001). Typically, with research using the CDS-II, the conceptual modification of controllability results in very high correlations between locus of causality and personal control (e.g., Crocker, Eklund, & Graham, 2002). Similarly, very high and significant correlations between locus of causality and team control have been noted in research using the CDS-T (rs 1⁄4 .69e.79, ps < .01, Allen et al., 2009; r 1⁄4 .84, p < .01, Dithurbide et al., 2009), suggesting cause for concern regarding the discriminant validity of the subscales. Second, empirical evidence from research into self-referent attributions (Coffee & Rees, 2008a, 2008b, 2009; Crocker et al., 2014) provides support for a broader conceptual approach to assessing attributions in sport (Allen et al., 2012; Rees, Ingledew, & Hardy, 2005). Rees et al. (2005) proposed that attribution research in sport should focus upon the main effects of controllability, together with the interactive effects of controllability and generalisability dimensions (stability, globality, and universality). This proposal is underpinned by three key points. First, that controllability is a key dimension upon which attention should be focused. Second, that attribution research would benefit from examining a broader conceptualisation of generalisability dimensions; that is, in addition to stability, examining the globality and universality of causes. As noted before, stability refers to the extent to which causes are seen as either stable or variable. Globality refers to the extent to which causes are seen to affect a wide range of situations or a narrow range of situations, and universality refers to the extent to which causes are seen as common to all people/teams or unique to individuals/teams. Third, that to model generalisability implies the need to move beyond main effects and consider interactive effects of attribution dimensions. (The reader is referred to Coffee & Rees, 2008a, and Rees et al., 2005, for a more elaborate discussion of these proposals.) To permit investigation of these proposals, Coffee and Rees (2008a) developed a measure of Controllability, Stability, Globality, and Universality attributions (the CSGU). The CSGU is a 16-item self-referent, situation-specific self-report questionnaire that assesses controllability and the three generalisability dimensions of stability, globality, and universality. The authors reported a good fit for the four-factor structure (least successful condition: RMSEA 1⁄4 .04, SRMR 1⁄4 .04, NNFI 1⁄4 .98; CFI 1⁄4 .98; most successful condition: RMSEA 1⁄4 .04, SRMR 1⁄4 .05, NNFI 1⁄4 .97; CFI 1⁄4 .98), together with coefficient alpha reliabilities ranging from .81 to .91. In line with proposals by Rees et al. (2005), the CSGU has been used to explore the main and interactive effects of attribution dimensions upon emotional and cognitive consequences of attributions (Coffee & Rees, 2008a, 2008b, 2009; Crocker et al., 2014). For example, Crocker and colleagues reported significant relationships between guilt and attributions of controllability, stability and globality, and between physical self-concept and shame and attributions of controllability and globality. Coffee and Rees have reported, following less successful performances, interactive effects for controllability and generalisability dimensions, and, following more successful performances, main effects for generalisability dimensions upon self-efficacy. The interactive effects, following less successful performances, demonstrated that controllability was positively associated with subsequent self-efficacy when causes were perceived to generalise across time (stability; Coffee & Rees, 2008a, 2009) or situations (globality; Coffee & Rees, 2008b). The purpose of the current article is to provide initial evidence for the construct, concurrent, and predictive validity of a new, fourfactor (controllability, stability, globality, and universality) measure of team-referent attributions: the Team-Referent Attributions Measure in Sport (the TRAMS). The TRAMS assesses individual group member perceptions of the causes of group outcomes. In Study 1, we examined the construct validity of the TRAMS across least successful and most successful conditions using confirmatory factor analysis (CFA). In Study 2, following team defeat and team victory, we again tested the factor structure of the TRAMS through CFA, together with examining evidence for concurrent validity by exploring correlations between the TRAMS dimensions and dimensions of the CDS-T. We hypothesized that high correlations would emerge between the TRAMS controllability subscale and the CDS-T team control subscale, and between the TRAMS stability subscale and the CDS-T stability subscale. In Study 3 we explored the predictive validity of the TRAMS and examined main and interactive effects of team-referent attributions on collectiveefficacy following team defeat and team victory.