The multi-factor structure of the Brief Self-Control Scale: Discriminant validity of restraint and impulsivity

Abstract

The Brief Self-Control Scale (BSCS; Tangney, Baumeister, & Boone, 2004) was developed to assess dispositional self-control as it is conceptualized by contemporary theoretical perspectives; however, concerns regarding its unidimensionality and validity remain. This article addresses these concerns using three samples to define, confirm and test a multi-factor version of the BSCS. Results of Study 1 (n = 909) identified two factors of the BSCS: restraint and impulsivity. Results of Study 2 (n = 364) confirmed that the 2-factor structure fit the data well and performed better than 1-factor structures. Finally, results from Study 3 (n = 175) demonstrated the utility of the multi-factor BSCS in predicting self-reported affective and behavioral outcomes. Implications for research and theoretical development are discussed.

Introduction

Building from control theory conceptualizations of the self-regulatory process (Carver & Scheier, 1982), most researchers consider self-control to represent the “operate phase”, wherein one initiates, alters, or maintains behavior in response to environmental demands. Thus, self-control can encompass a range of responses from overriding impulses to deliberative action (Carver, 2005). Individual differences in the capacity for self-control have been identified by numerous authors (cf., Carver, 2005) and add uniquely to our understanding of how people interact with their environment. Unfortunately, there is a dearth of well-validated measures of this construct. In part, this is due to conceptual issues regarding the operationalization of self-control (i.e., differentiating it from other aspects of the self-regulatory process such as standards and self-monitoring; Baumeister & Heatherton, 1996). In part, it is due to the fact that many scale developers design measures for use in specific populations or for specific purposes (e.g., clinical populations or eating behaviors; Tangney et al., 2004).

Tangney et al. (2004) noted that researchers have used several assessments to measure dispositional self-control; however, they questioned the content validity and generalizability of these assessments. Accordingly, they developed the 36-item Self-Control Scale (SCS) and a corresponding 13-item Brief Self-Control Scale (BSCS). To date, these scales have been used in over 60 studies, but the vast majority of researchers have used the BSCS. Several authors provided preliminary evidence for the reliability and construct validity of the BSCS (Carver et al., 2010, Gailliot et al., 2006, Schmeichel and Zell, 2007); however, questions regarding its unidimensionality and validity still remain.

Given the scarcity of well-validated and generalizable measures of self-control, the prolific use of the BSCS, and concerns regarding its validity; the aims of the current studies are twofold. First, we wanted to determine the appropriate factor structure of the BSCS. Tangney et al. (2004) provided no empirical justification for the unidimensionality of the BSCS. They noted that the BSCS represents the same content as the SCS and that the results of an exploratory factor analysis (EFA) suggested several distinct factors of the SCS. Unfortunately, they only provided a brief description of the factor analysis in a footnote of their validation paper. They listed the factor labels and the number of items on each factor, but did not indicate which items loaded onto each factor. To our knowledge, no existing articles have published results of exploratory or confirmatory factor analysis examining the structure of the BSCS. Clearly delineating the factors of such a widely-used measure can help to clarify conceptual ambiguities regarding the components of self-control, differentiate self-control from other self-regulatory constructs and identify areas where further refinement of these operationalizations are needed.

Second, we wanted to demonstrate the utility of examining each factor measured by the BSCS. Several authors have failed to find significant main effects of BSCS with expected outcomes (e.g., Barber et al., 2009, Haws and Poynor, 2008). If the BSCS represents conceptually-distinct factors related to self-regulatory outcomes, treating it as a unidimensional self-control scale can result in construct contamination that attenuates results. Thus, identifying these factors and their differential predictive validity can help to inform appropriate scoring, use and interpretation of the BSCS.

No additional refinement or development of the BSCS has been published in the peer-reviewed literature since the scale was originally created; however, Friese and Hofmann (2009) provided initial support for the multi-factor structure of the BSCS. They indicated that the effects of impulsivity on self-regulatory behavior disappeared when controlling for self-control as measured by the BSCS, but self-control evidenced incremental validity above impulsivity. Moreover, Fulford, Johnson, and Carver (2008) found different patterns of correlations between BSCS items and outcomes. After examining the correlations with scales measuring hypomania and narcissism, they indicated that items representative of self-discipline demonstrated positive relationships with narcissism, whereas items representative of impulse-control (i.e., reverse-coded impulsivity) correlated negatively with mania.

This 2-factor structure of BSCS as represented by self-discipline and impulse-control is similar to the distinction made between restraint and impulsivity, which has broad theoretical support from psychodynamic, trait, biological, cognitive, and developmental literatures (Carver, 2005). Carver indicated that while restraint represents the tendency to be deliberative or disciplined and engage in effortful control, impulsivity represents the tendency to be spontaneous and act on intuition or heuristics. Although related, these two components operate simultaneously, and they compete with one another to affect behavioral outcomes (Carver, 2005, Hofmann et al., 2009). As such, the empirical and theoretical literature provided a basis for suggesting that (a) the BSCS may measure multiple factors, and (b) these factors evidence differential relationships with important correlates.

The current paper incorporates three studies used to define, revise, and confirm the structure of the BSCS. In Study 1, we use EFA to identify the factors measured by the BSCS and confirmatory factor analysis (CFA) to refine these factors. In Study 2, we use CFA to confirm this structure, and in Study 3, we demonstrate the utility of each factor in predicting self-report affective and behavioral outcomes.