Assessing the status of locus of control as an indicator of core self-evaluations

https://doi.org/10.1016/j.paid.2015.11.002Get rights and content

Highlights

  • We investigate whether locus of control is a core evaluation of the self or environment.

  • Locus of control does not load on CSE after removing common method variance.

  • Perceptions of the environment impact locus of control but not the other CSE traits.

Abstract

Although core self-evaluation (CSE) is a significant personality predictor of work outcomes, additional research that assesses the structural validity of CSE is needed. It has been suggested that the inclusion of locus of control may have biased past CSE findings because this trait is better suited as a core evaluation of the environment. We investigated this issue in the current study. In Studies 1 and 2 we assessed the factor structure of the CSE factor after controlling for common method variance, which is a salient threat to the validity of higher-order constructs. We found that locus of control no longer loaded on CSE when we controlled for social desirability (Study 1) and when the traits were measured at different times (Study 2). To directly test whether locus of control is better suited as a core evaluation of the environment, in Study 3 we primed 110 employees to think about the predictability of their work environment. Results of this experiment revealed that ratings of locus of control changed as a function of the environment-based manipulation, whereas ratings of the other traits did not. Our research suggests that locus of control may not be an appropriate indicator of CSE.

Introduction

Several personality traits share a common self-evaluative theme that colors how people appraise themselves and their abilities, which in turn spill over to color appraisals of their experiences and environment. Judge, Locke, and Durham (1997) proposed the construct of core self-evaluation (CSE) to account for the dispositional source of these appraisals. CSE is a higher-order multidimensional construct that is believed to underlie the shared variance among traits that are fundamental, broad, and self-evaluative in nature. The specific traits are self-esteem, generalized self-efficacy, emotional stability, and locus of control. Although CSE was originally proposed to predict satisfaction with one's life and job (Judge, Locke, Durham, & Kluger, 1998), it has since been found to predict a variety of outcomes, including approach/avoidance motives (Ferris, D. L., et al., 2011, Ferris, D. L., et al., 2013), decisiveness (Di Fabio & Palazzeschi, 2012), and safety behavior (Yuan, Li, & Lin, 2014), among others (see Chang, Ferris, Johnson, Rosen, & Tan, 2012, for a review).

Although empirical evidence for the criterion-related validity of CSE has been encouraging, several issues have been raised with respect to the structural validity of this higher-order personality construct (e.g., Chen, G., 2012, Dormann, C., et al., 2006, Johnson, R. E., et al., 2011b, Johnson, R. E., et al., 2008, Schmitt, N., 2004). By “structural validity,” we mean the degree to which the set of personality traits belong together – both theoretically and empirically – as indicators of the higher-order construct (see Johnson, Rosen, Chang, Djurdjevic, & Taing, 2012). For example, some scholars have suggested that CSE may simply be a broader representation of emotional stability (Bono & Judge, 2003) or self-esteem (Johnson et al., 2008). Indeed, these two traits, along with generalized self-efficacy, tend to exhibit the highest factor loadings (>.70) on the CSE construct. Less clear, however, is the status of locus of control vis-à-vis the higher-order construct because it has markedly lower loadings relative to the other traits (e.g., Bono, J. E. and Judge, T. A., 2003, Johnson, R. E., et al., 2011b, Judge, T. A., et al., 2002). Findings like these have culminated in calls for further research on the structural validity of the CSE construct in general and the role of locus of control in particular (Chang, C.-H., et al., 2012, Chen, G., 2012).

We take up this call in the present study by investigating the suitability of locus of control as an indicator of the CSE construct. Addressing this issue is important because the structural validity of higher-order personality constructs ought to first be established prior to examining criterion-related validity (Johnson, R. E., et al., 2011a, Johnson, R. E., et al., 2012). Testing relationships of the higher-order construct with other variables before sufficient housecleaning has been performed on the CSE construct places the cart in front of the horse. Before doing so, it ought to be first established that the lower-order traits all load equivalently high onto the higher-order personality construct. Moreover, establishing the structural validity of CSE also avoids problems of construct contamination when irrelevant variables are erroneously included within its construct space, thereby biasing estimates of relationships and effect sizes. In light of these concerns, we examined the structural validity of CSE by benchmarking locus of control against the theoretical criteria of its trait indicators and by quantitatively investigating its fit across three empirical studies.

Judge et al. (1997) argued that traits must be fundamental, broad in scope, and self-evaluative to be counted as indicators of CSE. Traits that are fundamental are central to a person's self-concept. Generalized self-efficacy is fundamental, for example, because it reflects one's overall capability to perform effectively and cope with environmental demands. Traits that are broad in scope transcend specific times and situations. Self-esteem is broad, for example, because it captures a person's global sense of self-worth as opposed to their contextualized self-worth at work or school or home or elsewhere. Lastly, traits that are self-evaluative involve value judgments of the self as good/bad or effective/ineffective as opposed to being merely descriptive. Generalized self-efficacy is evaluative, for example, because it is a direct appraisal of the value of one's skills and abilities, as opposed to non-evaluative traits (e.g., agreeableness) that merely describe people's usual patterns of thinking and behaving (e.g., being cooperative and compassionate with others). The question we address in the remainder of this section is “Does locus of control satisfy these three criteria?”

Locus of control represents people's beliefs about how controllable and responsive the environment is (Rotter, 1966). People with internal loci of control believe that the environment is responsive to personal agency and that outcomes (e.g., incentives and punishment) can be predictably obtained. Those with external loci of control, in contrast, view the environment as mostly unresponsive and outcomes as relatively uncontrollable. With respect to the CSE criteria, locus of control appears to be fundamental and broad in scope. That is, it is fundamental and broad because locus of control involves generalized beliefs about the controllability of the environment that extend across multiple contexts (e.g., at home and work) and times, and it spills over to color people's perceptions, attitudes, and behaviors (Spector, P. E., 1982, Spector, P. E., 1986). With respect to these qualities, locus of control resembles self-esteem, generalized self-efficacy, and emotional stability.

Locus of control does not, however, fit cleanly with the criterion of being self-evaluative (Johnson, Rosen, Chang, & Lin, 2015). The other traits are evaluative in that they judge oneself and one's capabilities as being good/bad or effective/ineffective. Rather than referencing the self, locus of control primarily references the environment (i.e., how responsive and controllable it is; Rotter, J. B., 1966, Spector, P. E., 1982). Also, the consequences of locus of control differ regarding how people feel about themselves as compared to the other traits. It is typically desirable for people to have high levels of self-esteem, generalized self-efficacy, and emotional stability, all of which enhance feelings of self-worth and well-being. Whether or not a high level of locus of control (i.e., an internal locus) is desirable varies. When detrimental outcomes are experienced and people view the environment as responsive and controllable, it may actually harm their feelings of self-worth and well-being because such beliefs indicate that people could have acted differently to avoid the unfavorable result (Kelley, H. H., 1973, Weiner, B., 1985). In cases where detrimental outcomes are experienced, it is better for people to view the environment as uncontrollable (i.e., an external locus of control), thus enabling them to attribute such outcomes to external factors. Conceptually, it does not appear that locus of control is interchangeable with the other CSE traits.

The lack of conceptual overlap between locus of control with self-esteem, generalized self-efficacy, and emotional stability is also hinted at by empirical evidence. For example, meta-analytic estimates of the relationships among the four CSE traits reveal that the average inter-correlation for locus of control is noticeably lower than the inter-correlations among the other three traits (Judge, T. A., et al., 2002, Judge, T. A., et al., 1998). More recent primary studies have also noted the small inter-correlations of locus of control with self-esteem, generalized self-efficacy, and emotional stability (Johnson, Rosen, & Djurdjevic, 2011). It should be pointed out, however, that although locus of control has smaller inter-correlations with the other traits and weaker loadings on the higher-order CSE construct, these inter-correlations and loadings still tend to be statistically significant. Thus, to the extent that statistical significance is one benchmark for evaluating factor loadings, then these findings suggest that locus of control may be an indicator of CSE.

Nevertheless, there is one salient shortcoming with the CSE data that have been collected to date, namely all of the trait indicators are typically measured from a single source all at the same time. Doing so creates conditions that are especially ripe for common method variance (CMV; Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). CMV refers to variation in observed data that is due to the method used rather than the constructs of interest (Campbell & Fiske, 1959). CMV is of special concern for higher-order personality constructs because it can bias the emergence of such constructs when shared method variance combines with shared variance due to conceptual overlap (Johnson, R. E., et al., 2011a, Johnson, R. E., et al., 2012). With respect to CSE, for example, it is unclear how much of the small albeit statistically significant overlap of locus of control with the other traits is due to method artifacts. Given that locus of control appears to involve beliefs about the environment rather than the self, we suspect that much of its overlap with the other traits is due to method artifacts and not conceptual overlap. Thus, we predict that locus of control will no longer load on the CSE construct after CMV is controlled for in the higher-order model (Hypothesis 1).

To test this prediction, we examined the fit of the higher-order CSE construct after systematically applying statistical and procedural remedies for CMV across two studies. Statistical remedies, such as controlling for measured and unmeasured sources of CMV, account for the potential biasing effects of CMV during the data analysis stage (see Podsakoff et al., 2003). The statistical remedy in Study 1 involved measuring social desirability, which is a common source of CMV for self-evaluative traits and attitudes (e.g., Crampton & Wagner, 1994), and then partialing it out from the manifest indicators for self-esteem, generalized self-efficacy, emotional stability, and locus of control. While statistical remedies are commonly used, they are limited in that they deal with the symptoms of CMV rather than its sources. Procedural remedies target the sources of CMV by minimizing the influence of method artifacts via study design. The procedural remedy we used in Study 2 was temporal separation, which involved measuring each indicator at a different time. Temporal separation is particularly effective when examining personality traits and other constructs that are relatively stable (Johnson, Rosen, & Djurdjevic, 2011). We conducted a third and final study in which we primed employees to think and write about the extent to which their work environment is predictable and controllable (vs. unpredictable and uncontrollable). Employees rated their self-esteem, generalized self-efficacy, emotional stability, and locus of control following this manipulation, and we examined whether this environmental prime influenced only locus of control or the other CSE traits as well.

Section snippets

Participants and procedure

Two hundred and one students enrolled in undergraduate psychology and business courses participated in exchange for extra credit. Two hundred and twenty five participants were initially recruited, for a response rate of 89%. Participants completed an online survey that included measures of the CSE traits and social desirability. Participants' average age was 22.7 years (sd = 3.2), about half were female (52%), and they were mostly Caucasian (63%), Asian (19%), or African American (12%). We

Study 2

The results of Study 1 suggest that CMV may be in part responsible for the loading of locus of control on CSE. We demonstrated this by utilizing a statistical remedy (i.e., controlling for social desirability). However, a limitation of statistical remedies is that they do not prevent CMV from infiltrating data in the first place (Podsakoff et al., 2003). More effective remedies are procedural ones that target the source of CMV by altering how the data are collected. In the case of relatively

Study 3

Our findings from the previous two studies indicate that the higher-order CSE model has better fit when locus of control is excluded (vs. included) as an indicator (Hypothesis 1). We predicted as much because locus of control is not self-evaluative in the same way that the other CSE traits are. Locus of control is imbued in part by beliefs about the external environment and how responsive and controllable it is (Rotter, J. B., 1966, Spector, P. E., 1982). Although this logic provides the

General discussion

CSE represents appraisals of the self that influence how individuals perceive and experience the world around them, yet prior studies have operationalized CSE as a higher-order personality construct with indicators that reflect not only evaluations of the self (i.e., generalized self-efficacy, self-esteem, emotional stability) but possibly the environment as well (i.e., locus of control). Moreover, a growing body of research has provided evidence that locus of control has weaker loadings on the

References (48)

  • G. Chen

    Evaluating the core: Critical assessment of core self-evaluations theory

    Journal of Organizational Behavior

    (2012)
  • G. Chen et al.

    Validation of a new generalized self-efficacy scale

    Organizational Research Methods

    (2001)
  • P.T. Costa et al.

    Gender differences in personality traits across cultures: Robust and surprising findings

    Journal of Personality and Social Psychology

    (2001)
  • S.M. Crampton et al.

    Percept–percept inflation in microorganizational research: An investigation of prevalence and effect

    The Journal of Applied Psychology

    (1994)
  • D.P. Crowne et al.

    A new scale of social desirability independent of psychopathology

    Journal of Consulting Psychology

    (1960)
  • C. Dormann et al.

    A state-trait analysis of job satisfaction: On the effect of core self-evaluations

    Applied Psychology: An International Review

    (2006)
  • D.L. Ferris et al.

    When is success not satisfying? A moderated mediation model of the relation between core self-evaluation and job satisfaction

    The Journal of Applied Psychology

    (2013)
  • D.L. Ferris et al.

    Approach or avoidance (or both?): Integrating core self-evaluations within an approach/avoidance framework

    Personnel Psychology

    (2011)
  • L.R. Goldberg

    A broad-bandwidth, public domain, personality inventory measuring the lower-level facets of several five-factor models

  • J.F. Hair et al.

    Multivariate data analysis with readings

    (1998)
  • R.J. Hall et al.

    Item parceling strategies in SEM: Investigating the subtle effects of unmodeled secondary constructs

    Organizational Research Methods

    (1999)
  • K.J. Harris et al.

    Do social stressors impact everyone equally? An examination of the moderating impact of core self-evaluations

    Journal of Business and Psychology

    (2009)
  • H.S. Hodgins et al.

    Autonomy and control motivation and self-esteem

    Self and Identity

    (2007)
  • H.S. Hodgins et al.

    The cost of self-protection: Threat response and performance as a function of autonomous and controlled motivations

    Personality and Social Psychology Bulletin

    (2010)
  • Cited by (24)

    • Eat to avoid negative self-awareness: Locus of control and core self-evaluation as serial mediators in the effect of stress on food intake

      2019, Appetite
      Citation Excerpt :

      Whereas the other three CSE indicators references the self in that they judge oneself and one's capabilities as being good/bad or effective/ineffective, LOC primarily references the environment and represents people's beliefs about how controllable and responsive the environment is (Rotter, 1966; Spector, 1982). Thus, Johnson et al. (2015, 2016) concluded that LOC may be a fundamental evaluation of the environment rather than the self. The lack of conceptual overlap between LOC with the three indicators of CSE is also hinted at by empirical evidence.

    • Psychometric properties of the Spanish version of the Core Self-Evaluations Scale (CSES-SP)

      2018, Personality and Individual Differences
      Citation Excerpt :

      Finally, the correlations between the CSES-SP and optimism were stronger than the correlations with locus of control. This result confirms misgivings about including the locus of control as an indicator of the core self-evaluations (Johnson, Rosen, Chang, & Lin, 2016). Other indicators, such as dispositional optimism, should be reconsidered as substitutes (Judge et al., 1997).

    View all citing articles on Scopus
    View full text