Abstract
This article examines within-person consistency in personality expression across situations as an individual difference variable that is distinct from the typically studied personality trait level. The focus of the study is the manner in which personality consistency (a conceptualization of personality strength) influences the choice and interpretation of situations and, ultimately, the enactment of organizational citizenship behavior. We conducted an experience sampling study of 167 employees over 10 workdays. At each survey, participants reported their conscientiousness, agreeableness, situation perceptions, and organizational citizenship behavior. Results demonstrated that even after controlling for the linear and quadratic effects of personality trait level (and several other variables): (1) personality consistency increased within-person consistency in organizational citizenship behavior across situations and (2) this relationship was partially mediated by perceived consistency of situational strength and trait-relevant situational content. More broadly, the findings show that individual differences in personality are not restricted solely to the personality trait level. Rather, within-person consistency in personality expression across situations is itself an important individual difference: one that possesses appreciable behavioral consequences in the workplace and one that, consequently, is deserving of considerable future research.
Keywords
Introduction
Employees do not behave in an identical manner across situations. For instance, recent research has shown that there is considerable within-person variation in the three primary aspects of job performance: task performance, organizational citizenship behavior (OCB), and counterproductive work behavior (CWB; Dalal, Lam, Weiss, Welch, & Hulin, 2009). Specifically, Dalal, Bhave, and Fiset’s (2014) review found that on average across studies, 62%, 43%, and 49% of the variability in task performance, OCB, and CWB, respectively, was attributable to within-person factors. Through compilation and composition processes (Chan, 1998), variation in job performance over time within employees is expected to be an important source of variation in the performance of the organization as a whole—as evidenced by changes over time in measures such as the return on investment and the price-to-earnings ratio (Richard, Devinney, Yip, & Johnson, 2009). Moreover, within-person variability in job performance across situations is often desirable. For example, both newer contextual and older contingency theories suggest that leaders are more effective if they can systematically alter their behavior across situations (e.g., Vroom, 2000).
It is therefore important to understand within-person variability in job performance. In addition, though, it is likely that some employees may habitually exhibit more within-person performance variability than others. Indeed, individual differences in behavioral variability across situations or time have long been known to exist (see Fiske & Rice, 1955, for a review). Recent research on job performance (Sonnentag & Frese, 2012) has identified within-person performance variability (as contrasted with the mean level of performance) as an understudied area—and has called for research that specifically studies such variability as an outcome.
If within-person performance variability (e.g., OCB variability) is to be studied as an outcome, what are its likely predictors? Traditionally, personality traits have been viewed as a force for within-person stability. Therefore, personality traits cannot account for within-person variability in behavior. Conversely, situations—and the cognitive-affective responses they engender—have been viewed as forces for within-person behavioral variability (Mischel & Shoda, 1995; Weiss & Cropanzano, 1996). Recent conceptualizations of personality, however, have moved beyond the conventional trait (or mean across occasions) level of personality and have embraced the study of within-person consistency (i.e., absence of variability across situations or occasions) in state levels of personality (Dalal, Meyer, Bradshaw, Green, Kelly, & Zhu, 2015). Whereas individual differences in personality traits are a force for stability, individual differences in within-person personality consistency are a force for stability in variability (Fleeson, 2004, 2007).
Our emphasis on consistency, above and beyond the mean (trait) level, parallels research in several other areas. For instance, research on affect has demonstrated that above and beyond individual differences in trait (mean) affect levels, individual differences in affect variability (Beal, Trougakos, Weiss, & Dalal, 2013; Eid & Diener, 1999) are important predictors of behavior. At a more macrolevel, research on organizational climate has demonstrated the importance of studying climate strength above and beyond climate level (e.g., S. Lee & Dalal, 2016).
In the current paper, we further discuss the existence of, and the relationship between, these two individual difference conceptualizations of personality. We moreover answer the aforementioned calls for research predicting within-person consistency (vs. variability) in job performance—specifically, OCB. We do so by showing that individual differences in personality consistency predict within-person consistency in OCB consistency—and that they do so incrementally, beyond individual differences in personality trait levels. In addition, we examine the interplay between personality and situations in predicting OCB consistency. Specifically, we examine two perspectives. One perspective is the traditional statistical moderation perspective. According to this perspective, the perceived consistency of experienced situations should interact with personality consistency in predicting OCB consistency, with each attenuating the impact of the other. The second perspective, however, views personality as influencing the choice, shaping, and interpretation of situations. According to this perspective, the perceived consistency of experienced situations should result in part from personality consistency and should mediate the impact of personality consistency on OCB consistency. Overall, then, this paper emphasizes an understudied aspect of job performance (viz., within-person consistency), demonstrates how it can result from an understudied aspect of personality (viz., individual differences in within-person personality consistency), and examines two routes for situational influence (viz., moderation and mediation).
Personality Consistency
Although personality traits are traditionally considered relatively stable across situations and over time (Allport, 1937), equating “personality” merely with trait level (i.e., a mean score on a trait measurement scale) oversimplifies the phenomenon of personality (Dalal et al., 2015). In an influential critique, Mischel (1968) asserted that the expression of personality is more situation specific than personality trait theorists had believed. Several responses to Mischel’s work by personality psychologists acknowledged the need to assess, rather than assume, the consistency of personality across situations. For example, Bem and Allen (1974) discussed individual differences in the consistency of responses to personality scales, Lanning (1988) discussed individual differences in the extent of behavior changes related to equivalent situational changes, Paunonen (1988) discussed individual differences in the extent to which various personality traits are relevant to specific individuals, and Baumeister and Tice (1988) discussed individual differences in whether personality traits are possessed at all. Subsequently, Fleeson (2001) argued that a personality trait should be thought of as a density distribution of personality states. Stated differently, Fleeson argued that the personality trait should be thought of as the mean level of trait-relevant expressions across many situations or time (i.e., occasions).
Fleeson’s (2001) perspective, which has now become very influential in personality psychology (Dalal et al., 2015), has two notable implications. The first is that there exists considerable inconsistency (variability) in personality trait expressions across situations. For instance, meta-analytic findings suggest that from 49% to 78% of the variability in personality trait expression actually occurs within individuals across situations rather than between individuals (Fleeson & Gallagher, 2009). Even in workplace settings, where the range of situations encountered is presumably somewhat narrower than in life as a whole, personality trait expression exhibits considerable within-person variability across situations (Judge, Simon, Hurst, & Kelley, 2014).
The second implication of Fleeson’s (2001) work is that the extent of within-person consistency (vs. variability) in personality trait expression differs across people. For example, two people might have identical mean (trait) levels of conscientiousness across situations. However, one person may exhibit high conscientiousness consistency across situations (i.e., a narrower and higher-peaked distribution of conscientiousness), whereas the other person may exhibit low conscientiousness consistency (i.e., a broader and lower-peaked/flatter distribution; Fleeson, 2001).
Moreover, for a given person, the extent of consistency in personality trait expression across situations appears, like the mean level, to be relatively stable across situations or over time. In other words, there is stability in variability (Fleeson, 2004, 2007; Judge et al., 2014). This suggests that the mean (trait) level of personality expression and the consistency of personality expression across situations reflect two relatively distinct individual differences.
Personality consistency is a conceptualization under the “theoretical umbrella” of personality strength (Dalal et al., 2015: 270). Drawing from extant research literatures (e.g., self-monitoring, traitedness, self-concept clarity) relevant to personality psychology and organizational behavior, Dalal et al. (2015) provided a theoretical model of the impact of personality strength on within-person situational and behavioral consistency. Formally, these authors defined personality strength as “the forcefulness of implicit or explicit internal cues regarding the desirability of potential behaviors” (Dalal et al., 2015: 263). The strength of a particular trait reflects the relevance of the trait to the individual (Baumeister & Tice, 1988; Paunonen, 1988), such that traits are more relevant to, or reflective of, individuals high in rather than low in personality strength. Thus, individuals high in personality strength are guided by internal cues such as personality trait level to a greater extent than individuals low in personality strength. The forcefulness of these implicit cues for individuals high in personality strength results in a narrower distribution of, or more consistency in, personality states across situations.
In the current paper, we empirically test the personality strength propositions proposed by Dalal et al. (2015). The Dalal et al. model is notable because it emphasizes the prediction of within-person job performance consistency, an important but understudied outcome of interest. Moreover, in so doing, the model emphasizes the predictive power of within-person personality consistency above and beyond personality trait levels. Finally, the model emphasizes two routes—mediation and moderation—for the interplay of personality consistency with situations encountered on the job.
Although both personality consistency and personality trait level are individual differences, and are distinguishable from each other, they may be related quadratically: individuals who have moderate personality trait levels may exhibit high, moderate, or low personality consistency levels across situations, but individuals who have very high or very low trait levels may exhibit only high consistency levels (Paunonen, 1988). Thus, hypotheses involving personality consistency should emphasize effects incremental to those of the personality trait score and its square (Dalal et al., 2015).
Operationalization of Performance and Personality
Conceptual models of job performance have typically distinguished between three forms of employee performance: task performance, OCB, and CWB (Dalal et al., 2009; Rotundo & Sackett, 2002). In the current paper, we focus on OCB. OCB is defined as behavior that though less often recognized and rewarded by supervisors than task performance, contributes to organizational effectiveness by “[shaping] the organizational, social, and psychological context that serves as the catalyst for task activities and processes” (Borman & Motowidlo, 1997: 100). OCB includes behavior such as helping other employees, persisting beyond required levels on work tasks, volunteering to perform nonrequired activities, supporting organizational objectives, and complying with organizational rules and procedures beyond required levels (Borman & Motowidlo, 1997).
From a predictor-criterion conceptual matching perspective (Borman & Motowidlo, 1997; R. J. Schneider, Hough, & Dunnette, 1996), OCB is a form of employee performance that is particularly relevant to personality (and that does not require high levels of cognitive ability). In contrast, task performance, according to this perspective, is influenced less by personality than by cognitive ability. Moreover, meta-analytic results suggest that OCB is itself associated with important outcomes: OCB levels “account for at least as much variance in managerial evaluations of [overall employee] performance as task performance,” and they exert an appreciable influence on organization-level outcomes such as productivity, profitability, customer satisfaction, and unit turnover (N. P. Podsakoff, Whiting, Podsakoff, & Blume, 2009: 131). Moreover, like task performance, OCB exhibits considerable within-person variability across time (Dalal et al., 2014). Finally, compared to CWB—another “contextual” form of job performance—OCB has higher base rates of occurrence (Dalal et al., 2009) and therefore less likelihood of restricted range in individual differences in within-person performance consistency. In sum, therefore, this paper focuses on OCB because OCB is related to personality (arguably more strongly than task performance; Borman & Motowidlo, 1997), because OCB predicts important outcomes at the individual employee and organizational levels, and because OCB occurs more frequently within a workday than, say, CWB.
We turn now to the operationalization of personality. It is conceivable that an individual could exhibit high consistency associated with one personality trait but low consistency associated with another personality trait (e.g., Baumeister & Tice, 1988). Therefore, in the current study, we examine personality consistency associated with two traits: conscientiousness and agreeableness. Conscientious individuals are thorough and dedicated; as a result, they have a lower threshold for engaging in compliance-, persistence-, and organizational goal-oriented behavior such as OCB (P. M. Podsakoff, MacKenzie, Paine, & Bachrach, 2000). Agreeable individuals strive toward communion and are particularly willing to behave in a manner that serves the group; as a result, they have a lower threshold for engaging in prosocial behavior such as OCB (Ilies, Scott, & Judge, 2006). In other words, OCB is behavior that is conceptually relevant (Tett & Burnett, 2003) and empirically related (Dalal, 2005; Ilies et al., 2006; P. M. Podsakoff et al., 2000) to conscientiousness and agreeableness. We turn next to a discussion of why consistency in OCB should relate to consistency in conscientiousness and agreeableness.
Personality Consistency and OCB Consistency
Fleeson’s (2001) influential observation that personality is a density distribution of states suggests the primacy of within-person theorizing, with personality traits simply being one way of aggregating personality states—and personality consistency being another. In other words, personality-behavior relationships should be examined first at the within-person level and then, to study individual differences, aggregated up to the between-person level in various ways. Specifically, starting at the within-person level, OCB is more likely to be enacted in situations when the individual is high (vs. low) in conscientiousness and agreeableness. For instance, Judge et al. (2014) found that within-person models of agreeableness, conscientiousness, and extraversion that included autoregressive, concurrent, and cross-lagged effects explained approximately 50% of the within-person variance in daily OCB.
Next, to study individual differences in personality (and their impact on OCB), it is necessary to aggregate personality states to the between-person level. One way of doing so is to calculate the mean score. This is the traditional personality trait perspective. Accordingly, people who on average score high (vs. low) on conscientiousness and agreeableness across situations should also on average exhibit more OCB. However, a second way of aggregating personality states is to calculate the standard deviation. This is the personality consistency perspective. Accordingly, people who characteristically exhibit more (vs. less) variability in conscientiousness and agreeableness across situations should also characteristically exhibit more variability in OCB. The impact of personality consistency on OCB consistency, in other words, should occur even after accounting for the impact of personality traits on habitual OCB. To illustrate this, we examine the impact of personality consistency on OCB consistency after controlling for the personality and OCB mean scores (and their squared terms).
Hypothesis 1: Even after controlling for mean (trait) and squared mean levels of personality, and mean and squared mean levels of OCB, personality consistency is related positively to within-person consistency in OCB across situations.
Personality Consistency and Perceived Situational Consistency
The situation plays an important role in the relationship between personality consistency and OCB consistency (Dalal et al., 2015). In this regard, we note that there is a difference between objective (physico-biological) situations and subjective situations (Hattrup & Jackson, 1996). The subjective situation is the result of a person’s interpretation of the objective situation and is therefore a function not just of the objective situation but also of the person’s personality (Hattrup & Jackson, 1996). Depending on individuals’ personality-driven social cognitive processes, the same objective situation (e.g., criticism from a supervisor) could result in various subjective interpretations (e.g., a threat to job security vs. a learning opportunity). Because a person’s behavior in a situation is a function of the situation as he or she interprets it, researchers have typically concluded that “the most relevant level of situational analysis for theory building and research is the [subjective] situation” (Hattrup & Jackson, 1996: 512).
Moreover, it is necessary to distinguish between two fundamental aspects of subjective situations: namely, situational content (and therefore situational trait relevance) and situational strength (Tett & Burnett, 2003; see also Alaybek, Green, & Dalal, 2018). Situational content refers to the situational cues and features that “afford” motives, needs, goals, and traits (Rauthmann et al., 2014). The trait relevance (Tett & Burnett, 2003) of situational characteristics pertains to the probability of personality expression in the situation (e.g., Ten Berge & De Raad, 2001). For instance, situational characteristics of “sociality” (e.g., social interaction) call for or “afford” agreeableness (Rauthmann et al., 2014) and might result in agreeableness-related behavior (e.g., Ten Berge & De Raad, 2001). In contrast, situational strength refers to implicit or explicit situational cues regarding expected behavior (Meyer, Dalal, & Hermida, 2010; Mischel, 1968). Situational cues vary in the extent to which they are clear and consistent and indicate behavioral constraints and consequences, with higher levels reflecting more situational strength (Meyer et al., 2010). In sum, because we are examining the consistency of the personality traits of agreeableness and conscientiousness, we examine agreeableness-relevant and conscientiousness-relevant situation consistency; moreover, because personality has been shown to influence perceptions of situational strength beyond objective properties of the situation (Meyer et al., 2014, Study 3), we examine the consistency of perceived situational strength.
Personality strength theory (Dalal et al., 2015) suggests that individuals high in personality consistency rely more on internal cues, and less on external (situational) cues, to determine the desirability of potential behaviors. According to the theory, individuals high in personality consistency use several mechanisms to perceive situations in a way that aligns with their underlying personality trait level—and that the result is higher consistency, across situations, in both situational trait relevance and situational strength.
The first mechanism is situational choice (B. Schneider, 1987). Individuals high (vs. low) in personality consistency are likely to have strong preferences regarding the types of situations they are willing to enter: specifically, they are more likely to enter situations compatible with their personality trait profile, thereby resulting in more consistency in their perceptions of situations over time. For instance, an individual high in conscientiousness consistency and conscientiousness trait level would choose to frequently enter situations perceived as high in “duty” characteristics (Rauthmann et al., 2014). Moreover, when situations are compatible with personality trait profiles, the behavioral demands of situations are compatible with the behavioral demands of personality—and therefore the situations are likely to be perceived as weaker than would otherwise be the case. Because individuals high in personality consistency are disproportionately likely to enter situations consistent with their personality trait profiles, they are disproportionately likely to view situational strength as consistently low. For instance, an individual high in conscientiousness consistency and conscientiousness trait level would be disproportionately likely to view situational strength as consistently low (i.e., high situational strength consistency juxtaposed with low situational strength level).
The second mechanism is situational shaping. People high in personality consistency are likely to shape situations to be consistent with their own personality trait profiles. For instance, individuals high in conscientiousness consistency and conscientiousness trait level are likely to consistently shape situations (e.g., via modeling and incentives) to make them more relevant to high conscientiousness. Moreover, individuals high (vs. low) in personality consistency are disproportionately likely to view situations as consistently malleable. For instance, an individual high in conscientiousness consistency and conscientiousness trait level would be disproportionately likely to view situations as consistently malleable—and therefore situational strength as consistently low. In contrast, individuals low in personality consistency are unlikely to view situations as consistently malleable.
The third mechanism is selective attention (Ocasio, 2011) to features of situations. Attention—its initial allocation, sustenance, and control—is a central aspect of person-in-situation models (e.g., Beal, Weiss, Barros, & MacDermid, 2005; Ocasio, 2011). Of relevance to the current work, individuals are likely to differ in their attention to the features of the situations they enter. Among the psychological characteristics of situations (e.g., Parrigon, Woo, Tay, & Wang, 2016), some are relevant to the expression of personality traits (e.g., Rauthmann et al., 2014). We suggest that individuals high in personality consistency focus disproportionately on those trait-relevant features of situations that are consistent across situations and consistent with their personality trait levels, thus resulting in high within-person consistency in perceived situations. For example, individuals who are high in conscientiousness consistency and who have high conscientiousness trait scores will focus primarily on high-conscientiousness-relevant situational cues across situations. Conversely, individuals low in personality consistency will focus more equally on all features of situations—that is, features unique to individual situations, features common to multiple situations but inconsistent with these individuals’ personality trait levels, and features that are both common to multiple situations and consistent with these individuals’ personality trait levels. Moreover, because individuals high in personality consistency will focus to a greater extent on situational features consistent with their trait levels, they will disproportionately perceive situational strength as consistently low across situations.
The three mechanisms lead to the same contention: across a set of situations encountered over time, individuals high in personality consistency should exhibit more consistency in their perceptions of the encountered situational content and strength.
Hypothesis 2: Even after controlling for mean (trait) and squared mean levels of personality, and mean and squared mean levels of perceived situations, personality consistency is related positively to within-person consistency in perceived situations.
Perceived Situational Consistency and OCB
Perceptions of situations influence the behavioral requirements of the situation. According to the cognitive-affective processing system (Mischel & Shoda, 1995), behavior is dependent on how people process situational characteristics. Between individuals, different cognitions and affects (e.g., taking precautions, seizing opportunities) are activated by situational characteristics. Identical situational characteristics can therefore result in different behavior across individuals (Mischel & Shoda, 1995). An employee who perceives criticism as a threat, for instance, may in turn engage in CWB, whereas an employee who perceives criticism as helpful feedback may engage in OCB.
Although people differ from each other in their responses to a given situation, they also differ within themselves in their responses from one situation to another (Mischel & Shoda, 1995). In terms of trait-relevant situations, within-person differences in situation perception have predicted within-person differences in state behavioral expressions. For instance, R. A. Sherman, Rauthmann, Brown, Serfass, and Jones (2015) observed that when situations were perceived as high in sociality cues, participants tended to be more talkative and gregarious than usual.
Additionally, people differ from each other in their within-person patterns of situation-behavior relationships (Mischel & Shoda, 1995). For instance, one employee may behave in a more gregarious manner when meeting with coworkers than when meeting with the supervisor, whereas another employee may exhibit the opposite tendency. One such pattern, as demonstrated by previous studies, is that perceived similarity in situations relates to similarity in behavior enacted in those situations (e.g., Champagne & Pervin, 1987; R. A. Sherman, Nave, & Funder, 2010). Therefore, people who behave a particular way in one situation are likely to behave in a similar way in a similar situation. To the extent that an employee perceives situations to be similar (i.e., consistent in their psychological ingredients such as trait relevance or situational strength), he or she is likely to engage in consistent levels of OCB across situations. We therefore expect an impact of perceived situational consistency on OCB consistency.
Hypothesis 3: Even after controlling for mean and squared mean levels of perceived situations, and mean and squared mean levels of OCB, within-person consistency in perceived situations is related positively to within-person consistency in OCB across situations.
A Mediation Model
People high in personality consistency vary less across situations in their personality states than do people low in personality consistency. People high in personality consistency, moreover, are likely to focus not only on internal cues but also on features of situations that are consistent with their personality trait profiles. Thus, their situational perceptions further confirm their personality-based predilections. Given that situational consistency in turn predicts behavioral consistency, we predict that perceived situational consistency (specifically, trait-relevant situational consistency and situational strength consistency) will mediate the relationship between personality consistency (specifically, conscientiousness consistency and agreeableness consistency) and behavioral consistency (specifically, OCB consistency). In other words, we suggest that the relationship between personality consistency and OCB consistency is partially accounted for—that is, mediated—by the tendency of people scoring high on personality consistency to perceive higher consistency in the trait relevance and strength of situations. We predict partial, rather than full, mediation because of the possibility of additional mediators (e.g., consistency in goal level and goal orientation) not specified in the Dalal et al. (2015) model. Figure 1 presents the hypothesized model.
Hypothesized Model of Personality Consistency
Hypothesis 4: Even after controlling for the mean (trait) personality level (and its square), mean level of perceived situations (and its square), and mean level of OCB (and its square), within-person consistency in perceived situations partially mediates the positive relationship between personality consistency and within-person OCB consistency.
Ancillary Research Question: The Moderating Effect of Personality Consistency
Although perceived situational consistency predicts behavioral consistency (R. A. Sherman et al., 2010), this may be truer for some individuals than for others (Klirs & Revelle, 1986). For instance, Champagne and Pervin (1987) found that the size of the positive relationship between perceived situational similarity and predicted behavior similarity varied across participants. Recall that personality strength, of which personality consistency is a conceptualization, pertains to the impact of internal (to the person) behavior-relevant cues (Dalal et al., 2015). To the extent that internal (personality) cues are salient, they compete with external (situational) cues in determining behavior. In other words, situations may have less effect on the behavior for individuals high (vs. low) in personality consistency because, as discussed previously, such individuals are less reactive to situational cues.
However, we examine potential moderation effects as a research question rather than a formal hypothesis because we have already hypothesized that personality consistency will influence performance consistency indirectly through (i.e., mediated by) perceived situational consistency. Specifically, because people can often choose, interpret, and/or shape situations, the impact of perceived situational consistency on behavioral consistency may already be in harmony with the impact of personality consistency on behavioral consistency. Personality consistency, then, may not also moderate the relationship between situational consistency and behavioral consistency (Dalal et al., 2015). Moderation may instead be more prevalent in instances where individuals are in situations they have not chosen (either by happenstance or coercion)—and where it is difficult to shape the situation to be consistent with their personality trait profile (Dalal et al., 2015). For this reason, we examine potential moderation effects in an exploratory fashion, using a research question rather than a formal hypothesis statement.
Research Question: Even after controlling for personality mean (trait) level (and its square), mean level of perceived situations (and its square), and mean level of OCB (and its square), does personality consistency moderate—specifically, weaken—the relationship between within-person consistency in perceived situations and within-person consistency in OCB?
Overview of the Current Study
The current study extends organizational researchers’ view of personality beyond trait level by examining personality consistency as an individual difference variable in its own right. Specifically, the study builds on a recent conceptual review of the personality strength literature (Dalal et al., 2015) by using the experience sampling method (ESM; Beal, 2015) to examine the process by which personality consistency (a conceptualization of personality strength) associated with two personality traits (conscientiousness and agreeableness) relates to consistency in perceived trait-relevant situational content and situational strength as well as, ultimately, consistency in OCB. As a result of the focus on within-person consistency as an individual difference, the analysis is at the between-person rather than the within-person level.
Method
Participants
Participants were 185 employees: 109 undergraduate business and psychology students from a large mid-Atlantic university who worked at least 20 hours per week (mean number of hours worked per week = 25.74) and 76 full-time employees (mean number of hours worked per week = 40.41). Despite some empirical evidence to the contrary (e.g., Peterson, 2001), and despite arguments that samples do not matter as much as popularly believed (Highhouse & Gillespie, 2009), undergraduate samples can be criticized for their lack of generalizability to a working population. To address this concern, we supplemented the working student subsample with the full-time employee subsample.
The working student sample was recruited using an online research participation portal maintained by a large university in the mid-Atlantic United States, whereas the full-time employee sample was recruited using snowball sampling, a method that yields a study sample through referrals (Biernacki & Waldorf, 1981). Recruitment for the full-time employee sample began with researchers identifying known working professionals who might be interested in participating and would meet the requirements for the study (described in the following section) and by asking them to nominate others. No incentives were provided to individuals for recruiting sample participants. The study was described (truthfully yet vaguely) to participants as an examination of employees’ momentary reactions at work.
Working students were most frequently employed in office and clerical positions (30.56%), service positions (28.70%), and sales positions (21.30%). Full-time employees were typically professionals (38.16%) or employed in office or clerical positions (25.00%). The average age of the working students was 22.65 years (SD = 4.02) and that of the full-time employees was 27.66 years (SD = 8.98). Of the working students, 48.60% were women; of the full-time employees, 64.50% were women. Working students were 33.90% White, 27.50% Asian, 15.60% Black or African American, 14.70% Hispanic or Latino, 7.30% Other, and 1.00% Native American. Full-time employees were 71.10% White, 6.60% Asian, 13.10% Black or African American, 7.90% Hispanic or Latino, and 1.30% Other.
When subsample membership was controlled for statistically in analyses, it was never a statistically significant predictor, nor did it influence substantive conclusions associated with focal constructs. The absence of effects for subsample membership justifies an aggregation of the two subsamples into a single overall sample of 185 employees.
Procedure
To investigate consistency in personality, work situations, and behavior over time, we used the ESM (Beal, 2015). This sampling strategy is advantageous in that it enables better measurement of dynamic within-person processes by reducing recall bias and allows for the investigation of variability over time (Dalal et al., 2014; Fisher & To, 2012). Participants were first administered a screening survey to evaluate whether they qualified for the study. Participants were required to be at least 18 years old, work a minimum of 20 hours per week, work with coworkers and/or customers rather than alone (because some of the situations and behaviors of interest were interpersonal in nature), work on site at their job rather than teleworking (for the same reason), and be allowed to respond to surveys during work hours using their smartphones. If individuals passed the screening survey, we collected their contact information and their working hours for the next 10 workdays. Each qualifying participant was assigned an identification number to maintain confidentiality.
Qualifying participants were briefed on the procedure for the ESM study. Working students received an in-person introduction to the ESM study in the research lab, which typically occurred individually or in pairs, while full-time employees received the same introduction via e-mail. All participants received $1.67 for each valid ESM survey completed, for a possible total of $50.10 (if no surveys were missed).
We used an interval-contingent ESM approach, wherein participants received survey links via text message to their personal smartphones three times per day for 10 workdays. A qualifying workday included a work shift that was at least 4 hours long, so as to adequately assess a variety of situations and opportunities for OCB. Survey signal times were calculated based on the work schedules provided by participants and were sent at fixed intervals, with three evenly spaced signals sent per work shift such that, for reasons described in the next section, surveys were spaced at least an hour apart and the first survey of the shift occurred at least an hour after the start of the shift. Average time between surveys was 2.41 hours (SD = 1.03).
Each survey contained the same instructions and items, but the order of scales and of items within scales was randomized to prevent order effects and to reduce tedium. To create a “research alliance” between researchers and participants, which Csikszentmihalyi and Larson (1987) have argued is important in ESM studies due to the high demands on participants in these studies, we sent participants several e-mails during the ESM portion of the study. These e-mails contained not just reminders about the study requirements but also encouragement to continue participation in the study. Participants also had access to a website, created for this study, which provided answers to frequently asked questions related to study procedures (and compensation) and listed an e-mail address that could be used to contact the researchers with further questions.
To be included in analyses, participants needed to complete a minimum of 10 valid surveys. Our reasoning for this, based on prior research (Snijders & Bosker, 1993), was that 10 data points would permit a somewhat stable estimate of within-person variability (specifically, the standard deviation) needed for the operationalization of consistency across situations. To ensure relatively even distributions of survey responses within a workday, we required that surveys be completed within an hour of being received to be treated as valid. If participants missed any surveys, they had the option of receiving surveys on an additional, 11th workday, with three surveys sent using the same interval-contingent ESM approach described above. Of the 185 participants, 18 did not complete 10 or more valid surveys; thus, their data were excluded from all analyses. The resulting sample of 167 participants completed an average of 24.46 surveys (range = 10–32) for a total of 4,085 valid surveys (i.e., surveys completed within an hour of being received) that were used for analyses.
Measures
For each measure, we wanted participants to reflect on their most recent performance episode. A performance episode is “a behavioral segment that is thematically organized around an organizationally relevant goal or objective” (Beal et al., 2005: 1055) and that is time bound. As such, for each ESM survey, participants were instructed to focus on behavior, thoughts, and feelings associated with their main goal(s) or objective(s) with respect to their most recent major work-related task.
To test our hypotheses, it was necessary to aggregate our within-person (i.e., ESM) data to the between-person level so that standard deviations (and means) could be calculated for each person on the measures described subsequently. Because of this, the reliability of each measure was calculated in two ways. First, reliability coefficients were calculated across persons for each survey (reliability for the Day 1, Time 1 survey; reliability for the Day 1, Time 2 survey; etc.). Second, a reliability coefficient was calculated across persons using temporally aggregated data. For measures consisting of dichotomous items (see below), we used the KR-20 reliability coefficient rather than Cronbach’s alpha (Lord & Novick, 1968).
Personality (conscientiousness and agreeableness)
To capture momentary levels of conscientiousness and agreeableness, 7 items per factor (so, 14 in total) were taken from John, Donahue, and Kentle’s (1991) Big Five Inventory. Items from the original inventory were excluded if they explicitly indicated that they represented a tendency (e.g., “tends to find fault with others”) because these items could not readily be reflected as personality states. Because of the need to keep ESM surveys short (e.g., Dalal et al., 2009), items were also excluded if they very closely paralleled other items (consistent with Cattell’s, 1978, recommendations to avoid “bloated specific” items).
Participants were instructed to answer items using the stem “With respect to your most recent major work-related task since the last survey (or since the start of your workday if this is your first survey of the day), you were . . .” Participants responded using a 1 (strongly disagree) to 5 (strongly agree) response scale. Sample items include “Doing a thorough job” (for conscientiousness) and “Cooperating with others” (for agreeableness). Mean reliability (Cronbach’s alpha) across persons but within occasion was .86 for conscientiousness and .85 for agreeableness. Reliability across persons and across occasions was .93 for conscientiousness and .92 for agreeableness.
Perceived situational content
To assess momentary perceived situational characteristics that were relevant to the traits of conscientiousness and agreeableness, Sociality and Duty items were adapted from the Rauthmann et al. (2014) Situational Eight DIAMONDS scale. As per Rauthmann et al., the situational aspects relating to Sociality resemble the personality trait of agreeableness, whereas the Duty items are representative of factors such as working, goal pursuit, and achievement, which resemble the personality trait of conscientiousness. Moreover, empirical evidence suggests high correlations between a situational content scale and its personality affordance (i.e., the relevance of or need for personality to a situation): in Rauthmann et al., Duty and the affordance of conscientiousness correlated at .60, whereas Sociality and the affordance of agreeableness correlated at .71. Sample items include “There was an opportunity for social interaction” (Sociality situations) and “Someone depended on me to do something” (Duty situations). Participants were instructed to indicate whether each statement described their work environment during their most recent major work-related task, using dichotomous agree and disagree response options. Mean reliability (coefficient KR-20) across persons but within occasion was .72 for conscientiousness-relevant situations and .76 for agreeableness-relevant situations. Reliability across persons and across occasions was .87 for conscientiousness-relevant situations and .88 for agreeableness-relevant situations.
Perceived situational strength
To assess momentary perceived situational strength, items were adapted from Meyer et al. (2014). We used 12 items in total, 3 from each of the four facets of situational strength (clarity, consistency, constraints, and consequences). The 3 items from each facet were those items with the highest factor loadings per facet, as reported by Meyer et al. Consistent with Meyer et al., we operationalized overall situational strength as a unit-weighted composite of all the items. Participants were instructed to indicate whether each statement described their work environment during their most recent major work-related task, using dichotomous agree and disagree response options. A sample item is “Specific information about work-related responsibilities was provided.” Mean reliability (coefficient KR-20) across persons but within occasion was .79. Reliability across persons and across occasions was .89.
OCB
OCB was assessed using a combination of five items from K. Lee and Allen (2002) and three items from Motowidlo and Van Scotter (1994). Items tapping both interpersonally and organizationally directed behavior were used, given LePine, Erez, and Johnson’s (2002) finding of strong meta-analytic relationships between various forms of OCB. We excluded items from the original scales (i.e., K. Lee & Allen, 2002; Motowidlo & Van Scotter, 1994) if they were (1) strongly task related (to avoid conflating OCB with task performance), (2) redundant with more general items from the other scale, and/or (3) unlikely to vary within a given day. Participants were asked to respond with either Yes or No to indicate whether they had enacted each behavior during their most recent major work-related task. Methodological choices such as these are common—and necessary—for ESM measures of behavior/performance (see, e.g., the discussion in the Method section of Study 1 in Dalal et al., 2009). A sample item is “Supported and encouraged a coworker with a problem.” Mean reliability (coefficient KR-20) across persons but within occasion was .85. Reliability across persons and across occasions was .94.
Consistency Operationalizations
Personality consistency was operationalized via the standard deviation of momentary scores (i.e., personality states) of conscientiousness and agreeableness across surveys (Fleeson, 2004, 2007). We multiplied standard deviations by −1 so that higher numbers would represent personality consistency rather than variability (see S. Lee & Dalal, 2016). The three situational consistency constructs (i.e., situational strength consistency, conscientiousness-relevant situation consistency, and agreeableness-relevant situation consistency) and OCB consistency were operationalized in the same way.
Control Variables
Focal variable (conscientiousness, agreeableness, perceived situations, and OCB) responses were averaged across surveys to provide between-person measures of focal variable mean scores. These mean scores were used as controls in analyses so as to demonstrate the impact of the consistency score (e.g., conscientiousness consistency) after controlling for the impact of the mean score (e.g., conscientiousness trait level). Because the relationships between the mean scores and consistency scores may be quadratic rather than linear, we also included the square of the mean scores (e.g., conscientiousness trait level squared) as controls. Finally, because agreeableness and conscientiousness mean scores were quite highly correlated in our study, we controlled for agreeableness mean scores (and their square) in the conscientiousness analyses and vice versa.
We also included other controls pertaining to the study design. We controlled for the number of surveys completed by each participant over the course of the study. We did this because even though all participants retained for the final sample provided data from at least 10 surveys, variation in the number of surveys completed by participants (range = 10–32) may nonetheless be related to the precision of our operationalization of consistency (Snijders & Bosker, 1993). Additionally, we controlled for the average length of the time interval between surveys for each participant. We did this because although participants’ work shifts needed to be at least 4 hours long, their shifts could have been much longer (e.g., 9 hours). Because the ESM surveys were spaced evenly across the work shift, controlling for the time between surveys ensures the generalizability of findings across shift lengths.
Results
Variance and Stability of Within-Person Measures
To investigate the percentage of total variance attributable to within-person as compared to between-person sources, we estimated multilevel, unconditional means models using the lme4 package in R version 3.2.2 for each of the ESM measures. As expected, and consistent with previous findings (e.g., Judge et al., 2014), for all variables a substantial proportion of variance was attributable to within-person sources (see Table 1).
Variance Decomposition and Stability of Personality, Situation, and Behavioral Constructs
Note: n = 167 participants. The second and third columns in the table represent the percentages of overall variance in a construct attributable to within-person and between-person sources. Results indicate that all constructs exhibit appreciable within-person variance. The third column of numbers, “Between-Person Percentage,” is the intraclass correlation coefficient, or ICC(1), for each construct, albeit expressed as a percentage. The last two columns in the table examine temporal stability—in other words, test-retest reliability—by correlating scores from the first and second halves of people’s data over the course of the study. For each construct (e.g., conscientiousness), temporal stability was examined for both the within-person mean scores and the within-person consistency (i.e., standard deviation multiplied by −1) scores. Results indicate that the mean scores are highly stable over time and that the consistency scores are quite stable.
p < .01.
The last two columns of Table 1 indicate the temporal stability (test-retest reliability) of the variables. We computed stability by splitting each participant’s survey responses into first and second halves and then correlating the within-person mean scores and the within-person consistency scores (standard deviations multiplied by −1) between the two halves of study data (Fleeson, 2004, 2007; Judge et al., 2014). The idea of personality as a density distribution of states (Fleeson, 2004) suggests that (1) each participant will exhibit a distribution of personality state scores during the study; (2) the personality mean scores should remain stable, indicating support for the conventional understanding of personality traits; and (3) the personality consistency scores should also remain stable. The results show strong stability for the mean scores, with correlations (rs) ranging from .74 to .90 (all p < .01), and moderate stability for the consistency scores, with correlations ranging from .45 to .58 (all p < .01). The higher stability of the mean (compared to consistency) is in line with other ESM studies of personality consistency (Fleeson, 2007; Judge et al., 2014) and with ESM studies in general (Hektner, Schmidt, & Csikszentmihalyi, 2007).
Descriptive Statistics for Variables
To test the study’s hypotheses and research question, we aggregated focal construct scores (i.e., personality states, momentary situation perceptions, and momentary OCB levels) to the person level in the manner described previously. Table 2 provides descriptive statistics (of raw variables) and correlations (after mean-centering) among the between-person mean scores (and their squares) and the between-person consistency scores. With regard to trait-relevant situation scores, the current correlation between conscientiousness-relevant and agreeableness-relevant situations (r = .40) was comparable to the analogous correlation (r = .41) reported by R. A. Sherman and colleagues (2015).
Descriptive Statistics for and Correlations Between Focal Constructs and Control Variables
Note: n = 167 participants. Descriptive statistics are based on raw (observed) scores. Correlations are based on mean-centered scores.
p < .05.
p < .01.
Turning now to trait scores, we found a high correlation between conscientiousness and agreeableness (r = .77) when these traits were operationalized as mean state scores (averaged across ESM surveys). Conscientiousness and agreeableness have been found to load on the same higher-order “stability” (or “α,” versus “plasticity” or “β”) factor in broad-bandwidth two-factor models of personality (e.g., Digman, 1997), but the correlation obtained here is appreciably higher than the between-person conscientiousness-agreeableness correlation when the traits are measured using traditional, cross-sectional surveys (meta-analytic correlation, r = .31; van der Linden, te Nijenhuis, & Bakker, 2010). Importantly, however, previous ESM studies have also reported high correlations between these two variables. A “mega-analysis” combining 15 ESM studies (see Fleeson & Gallagher, 2009) produced a correlation of .62 between mean state conscientiousness and agreeableness (W. Fleeson, personal communication, March 7, 2017). Similar results were recently obtained by Finnigan and Vazire (in press). This suggests the strength of the relationship between conscientiousness and agreeableness might differ across levels of analysis (Dalal et al., 2014)—or, more technically, in this case, across levels of original measurement—such that the correlation is higher in aggregated ESM surveys than in traditional cross-sectional surveys. Although the correlation between conscientiousness and agreeableness mean (trait) scores is not of focal interest to the current paper, we responded to the higher-than-expected correlation by controlling for both personality traits (conscientiousness and agreeableness) and their squares in all analyses involving the effects of personality consistency.
Relationship Between Personality Consistency and Personality Trait Scores
Personality consistency scores are unlikely to be completely orthogonal from the corresponding personality trait scores (Paunonen, 1988), and as discussed previously, the true relationship between the two may be quadratic. To examine this, we regressed respondents’ personality consistency scores onto the corresponding personality mean scores (i.e., personality trait scores; Fleeson, 2004, 2007) and squared personality mean scores.
Results indeed suggested a quadratic relationship between personality consistency and the associated personality mean score. For both conscientiousness consistency and agreeableness consistency, adding a squared personality mean score at Step 2 in the regression equations resulted in statistically significant incremental variance beyond the personality mean score (see Table 3). Graphing the relationship between conscientiousness mean score and conscientiousness consistency revealed a quadratic relationship such that as expected, personality consistency is higher when conscientiousness mean (trait) scores are low and high than when they are at moderate levels. A similar relationship was found for agreeableness. As a result, and as previously discussed, all focal analyses involving personality consistency include statistical controls for both the personality mean (trait) score and its square.
Relationship Between Personality Trait and Personality Consistency: Regressing Personality Consistency on the Linear and Squared Personality Mean Scores
Note: n = 167 participants. All variables were mean-centered prior to the analysis. b = unstandardized regression coefficient; SE = standard error for unstandardized coefficient; β = standardized regression coefficient.
p < .01.
Hypothesis Test Results
We tested all hypotheses and the ancillary research question in a moderated mediation model (see Figure 1). We mean-centered all variables (except for interaction and nonlinear terms) at the between-person level and controlled statistically for several variables, as described previously. We assessed our moderated mediation model using the SPSS macro PROCESS (specifically, Model 74) with 1,000 bootstrap samples (Hayes, 2013). We regressed OCB consistency onto an operationalization of perceived situational consistency (i.e., situational strength consistency, conscientiousness-relevant situation consistency, or agreeableness-relevant situation consistency), an operationalization of personality consistency (i.e., conscientiousness consistency or agreeableness consistency), and the control variables. We tested four models, one for each of the personality consistency–perceived situational consistency combinations: (1) conscientiousness consistency and situational strength consistency, (2) agreeableness consistency and situational strength consistency, (3) conscientiousness consistency and conscientiousness-relevant consistency, and (4) agreeableness consistency and agreeableness-relevant consistency. Tables 4 through 7 provide the unstandardized coefficients, standard errors of unstandardized coefficients, confidence intervals (CIs) for unstandardized coefficients, and standardized coefficients for the paths within each of these models.
Test of Hypothetical Model: Conscientiousness Consistency and Situational Strength Consistency
Note: n = 167 participants. All variables were mean-centered prior to the analysis. b = unstandardized regression coefficient; SE = standard error for unstandardized coefficient; LLCI, ULCI = lower and upper limits of 95% confidence interval for unstandardized coefficient; β = standardized regression coefficient; OCB = organizational citizenship behavior.
p < .05.
p < .01.
Hypothesis 1 stipulated that personality consistency would be incrementally (beyond the control variables) positively related to OCB consistency. In all four models, conscientiousness consistency (unstandardized coefficients b = 0.06, p < .05; b = 0.09, p < .01; Tables 4 and 6) and agreeableness consistency (b = 0.11, p < .01; b = 0.12, p < .01; Tables 5 and 7) were positively associated with OCB consistency. Thus, Hypothesis 1 was supported.
Test of Hypothetical Model: Agreeableness Consistency and Situational Strength Consistency
Note: n = 167 participants. All variables were mean-centered prior to the analysis. b = unstandardized regression coefficient; SE = standard error for unstandardized coefficient; LLCI, ULCI = lower and upper limits of 95% confidence interval for unstandardized coefficient; β = standardized regression coefficient; OCB = organizational citizenship behavior.
p < .05.
p < .01.
Test of Hypothetical Model: Conscientiousness Consistency and Conscientiousness-Relevant Situation Consistency
Note: n = 167 participants. All variables were mean-centered prior to the analysis. b = unstandardized regression coefficient; SE = standard error for unstandardized coefficient; LLCI, ULCI = lower and upper limits of 95% confidence interval for unstandardized coefficient; β = standardized regression coefficient; OCB = organizational citizenship behavior.
p < .05.
p < .01.
Test of Hypothetical Model: Agreeableness Consistency and Agreeableness-Relevant Situation Consistency
Note: n = 167 participants. All variables were mean-centered prior to the analysis. b = unstandardized regression coefficient; SE = standard error for unstandardized coefficient; LLCI, ULCI = lower and upper limits of 95% confidence interval for unstandardized coefficient; β = standardized regression coefficient; OCB = organizational citizenship behavior.
p < .01.
Hypothesis 2 stipulated that personality consistency would be incrementally positively related to situational strength consistency and situational content consistency. Model results provide partial support. Conscientiousness consistency was significantly and positively associated with situational strength consistency (b = 0.07, p < .05; Table 4), but conscientiousness consistency was not significantly related to conscientiousness-relevant situation consistency (b = 0.02, p = .48; Table 6). Agreeableness consistency was significantly and positively associated with situational strength consistency (b = 0.09, p < .01; Table 5) and agreeableness-relevant situation consistency (b = 0.10, p < .01; Table 7). In sum, Hypothesis 2 was supported in three out of four comparisons.
Hypothesis 3 stipulated an incremental positive relationship between perceived situational consistency and OCB consistency. As shown in the models, situational strength consistency (b = 0.39, p < .01; b = 0.34, p < .01; Tables 4 and 5), conscientiousness-relevant situation consistency (b = 0.17, p < .05; Table 6), and agreeableness-relevant situation consistency (b = 0.30, p < .01; Table 7) were significantly related to OCB consistency over time. Thus, Hypothesis 3 was supported.
Hypothesis 4 stipulated that within-person consistency in perceived situations over time would partially mediate the relationship between personality consistency and OCB consistency. The moderated mediation analysis previously described, using macros by Hayes (2013), employs bootstrapping procedures to construct CIs to determine the significance of the indirect effect. Each model was run with 1,000 bootstrapped samples to construct 95% CIs. If the CI excludes zero, one can conclude that mediation has occurred and that the indirect effect is statistically significant. The indirect effects for each model are presented in Table 8. Of primary relevance for Hypothesis 4 are the “average level of moderator” indirect effect rows (the other rows will be discussed in the next section on moderation). When running the model with situational strength consistency as the mediator between conscientiousness consistency and OCB consistency, the CI revealed that as expected, the average indirect effect (0.026) was statistically significant at the 95% level of 0.005 to 0.051. This held true for agreeableness consistency as well (average indirect effect = 0.030, 95% CI = [0.009, 0.056]). Turning now to situational content consistency, the average indirect effect of agreeableness consistency (0.030) on OCB consistency through agreeableness-relevant situation consistency was statistically significant at the 95% level of 0.009 to 0.061; however, the average indirect effect for conscientiousness consistency (0.003) through conscientiousness-relevant situation consistency was not significant (95% CI = [–0.006, 0.019]). In sum, Hypothesis 4 was supported in three out of four comparisons.
Personality Consistency as Simultaneously an Antecedent and Second-Stage Moderator: Indirect Effects of Personality Consistency on Organizational Citizenship Behavior Consistency via Perceived Situational Consistency at Different Levels of Personality Consistency
Note: n = 167 participants. All variables were mean-centered prior to the analysis. Indirect effects are provided for 1 SD below the mean (“low level”), the mean (“average level”), and 1 SD above the mean (“high level”) of the moderator, personality consistency. Situational content consistency = trait-relevant situation consistency associated with the same personality trait as the personality consistency predictor; SE = standard error for effect; LLCI, ULCI = lower and upper limits of 95% confidence interval.
Though the magnitude of the effects is small for mediation and for the other hypothesized paths in the moderated mediation models, these small effects are still important. Just as a moderator variable is not orthogonal to its components (i.e., the two variables used to create the product term), personality consistency is not orthogonal to the personality trait score and its square. Obtained effect sizes for moderators in the literature are typically small (between 1% and 3% of the variance; Champoux & Peters, 1987) but still important (Evans, 1985) because they explain the residual variance of interactions after partialling out the component main effects (McClelland & Judd, 1993). Although our focus is mediation, a similar argument applies: we found small effect sizes for personality consistency after partialling out the component main effects for personality mean (trait) scores and its square, along with other control variables. Therefore, our effect sizes, though small, suggest that personality consistency explains unique variance in perceived situational consistency and OCB consistency.
Ancillary Research Question
In addition to testing the formal hypotheses, we examined the research question proposed in the Introduction section: namely, whether personality consistency moderates (specifically, attenuates) the relationship between perceived situational consistency and OCB consistency. To examine whether—as we speculated in the Introduction section of the paper—the moderation effects are already “baked into the cake” via the impact of personality consistency on perceived situational consistency, it is necessary to examine personality consistency simultaneously as an antecedent to perceived situational consistency and a moderator (attenuator) of the relationship between perceived situational consistency and OCB consistency.
Indirect effects at three levels of the moderator (mean plus and minus 1 SD) for all four models are provided in Table 8. Additionally, an overall index of moderated mediation is provided for each of the four moderated mediation models. Given earlier results supporting mediation, the overall test for moderated mediation essentially indicates whether moderation is obtained in the presence of mediation. In the model with conscientiousness consistency as the predictor (and moderator), results supported moderated mediation for situational strength consistency as a mediator (index = −0.040, 95% CI = [–0.111, –0.002]) but not for conscientiousness-relevant situation consistency as a mediator (index = 0.001, 95% CI = [–0.010, 0.026]). Conversely, in the model with agreeableness consistency as the predictor (and moderator), results supported moderated mediation for agreeableness-relevant situation consistency as a mediator (index = −0.042, 95% CI = [–0.113, –0.001]) but not for situational strength consistency as a mediator (index = −0.058, 95% CI = [–0.137, 0.004]). Moreover, in the two models for which moderated mediation was significant (i.e., the 95% CI excluded zero), the size of the indirect effect (mediation) decreased with increasing personality consistency (see Table 8 for conditional indirect effects), thereby supporting the contention that personality consistency attenuates the second-stage relationship (between perceived situational consistency and OCB consistency) in the mediation. Figure 2 is a plot of the interaction between conscientiousness consistency and situational strength consistency (Aiken & West, 1991).
Interaction Between Conscientiousness Consistency and Situational Strength Consistency in Predicting Organizational Citizenship Behavior Consistency (Test of Ancillary Research Question)
In sum, vis-à-vis the research question, personality consistency moderated (specifically, attenuated) the relationship between perceived situational consistency and behavioral consistency in two of the four comparisons. Overall, the results suggest some evidence of moderation via personality consistency despite its preexisting effects as the antecedent in a mediation relationship.
Discussion
A noteworthy finding from this study was that personality can be simultaneously a volatile and a stable characteristic of individuals. Consistent with previous research in both personality psychology (e.g., Fleeson & Gallagher, 2009) and organizational behavior (Judge et al., 2014), our results showed that considerable variance in both conscientiousness and agreeableness state scores was attributable to within-person sources. More importantly, however, there was evidence of stability in variability: the first half of participants’ survey responses correlated appreciably with the second half for not just the means but also the consistencies of the study’s main variables (personality, perceived situations, and OCB). These results provide support for the idea that personality involves two relatively distinct (though quadratically related) individual differences: personality trait and personality consistency.
The study’s focal hypotheses pertained to the process by which personality consistency relates to OCB consistency across situations. Personality traits (mean scores) are forces for stability and therefore are insufficient for the study of OCB consistency. In contrast, personality consistency is a force for stability in variability—and therefore is helpful in predicting OCB consistency. More specifically, we generally found support for the following hypotheses: (1) personality consistency was related positively to perceived situational consistency and OCB consistency, (2) perceived situational consistency was in turn related positively to OCB consistency, and (3) perceived situational consistency partially mediated the relationship between personality consistency and OCB consistency. Interestingly, despite this mediated effect, we found some support for the idea that personality consistency also moderates (specifically, attenuates) the effect of perceived situational consistency on OCB consistency.
Practical Implications
Despite the fact that the current research is the first empirical test of a theory (Dalal et al., 2015) and more generally represents an understudied perspective (personality consistency across situations), the findings have important practical implications. One important implication pertains to how personality consistency is viewed by others. People with highly consistent personalities may be viewed as reliable (vs. erratic). On the other hand, they may be viewed as dogmatic (vs. adaptable). We therefore suggest that unlike high versus low levels of personality traits, it is not the case that either high or low levels of personality consistency are inherently “good” or “bad.” Rather, the extent to which a consistent personality is desirable depends on the job setting. High personality consistency would be beneficial for jobs that require considerable perseverance and behavioral consistency but harmful for jobs that require considerable behavioral flexibility.
Traditionally, employee selection methods have measured trait levels of personality characteristics in order to choose employees, resulting in a validity ceiling of 0.3 when using personality to predict performance (Mischel, 1968).With the caveat that employee selection has typically been viewed in the context of maximizing task performance, and therefore that selection-related implications must necessarily be viewed as tentative until our results are replicated using task performance (vs. OCB) as an outcome of interest, we suggest that one way to overcome this validity ceiling is for practitioners to select for high or low levels of personality consistency with the aim of maximizing person-job fit. For instance, consistency in trait expression may be preferred when the job requires resistance to strong situations (e.g., the ability to resist influence tactics or conformity pressures). On the contrary, variability in trait expression may be preferred when the employee needs to display behavioral flexibility (or adaptive performance).
Personality consistency can also be the focus of training and/or work (re)design interventions for job incumbents (Dalal et al., 2015). Opportunities for job crafting (Wrzesniewski & Dutton, 2001) would increase person-job fit for employees high in personality consistency. Conversely, when the external imposition of structuring mechanisms (e.g., assigned goals) is necessary, employees low in personality consistency are likely to respond better than those high in personality consistency.
Finally, organizations could incorporate personality consistency into the feedback component of performance appraisal (Dalal et al., 2015). Employees low in personality consistency may benefit from insight into their unique “if . . . then . . .” situation–behavior profiles (Mischel & Shoda, 1995). In contrast, employees high in personality consistency may benefit from the opportunity to gain insight into their “one-size-fits-all” tendency.
Limitations and Future Research Directions
The ESM design, used in the current study, may be less susceptible to same-source bias than traditional one-shot survey designs because responses are collected on multiple occasions and the focus is on change from one occasion to the next. Additionally, meta-analytic evidence (Carpenter, Berry, & Houston, 2014) shows that self-rated and other-rated OCB have similar patterns of relationships with antecedents. Other-rated OCB, moreover, adds very little incremental validity beyond self-rated OCB for self-rated antecedents such as conscientiousness (Carpenter et al., 2014). This evidence supports our use of self-rated OCB in the current study. Future research could nonetheless attempt to replicate our findings using other-rated OCB.
Using ESM also meant that items included in the study had to be adapted from existing (non-ESM) sources and that measures had to be kept short to reduce participant burden (see Dalal et al., 2009, for a discussion of these issues). Nonetheless, all measures demonstrated acceptable reliability. In addition, although ESM studies are generally believed to provide a representative sampling of individuals’ work situations (Beal, 2015), we do not know the population of work situations encountered by our respondents. Nor do we yet have empirical evidence to support the idea that one reason personality consistency is related positively to perceived situational consistency is that people with consistent personalities focus primarily on similarities between situations that are also consistent with their own personality trait levels, whereas people with less consistent personalities focus to a greater extent on all aspects of situations (i.e., differences as well as similarities). A major direction for future personality consistency research, therefore, involves assessing the psychological ingredients of situations with the aim of testing this idea.
Several additional avenues exist for future research. Any given study, especially an ESM study, cannot examine the entire construct space of traits, situations, and work behavior or performance. Future research could extend the findings in this study and attempt to replicate our findings through different operationalizations of personality consistency, perceived situational characteristics, and behavior/performance characteristics. The current research operationalized personality strength statistically by calculating the standard deviation of personality states across situations. Although Dalal et al. (2015) specifically recommended this particular operationalization of personality strength highly, those authors also listed several other operationalizations. Personality strength could be measured using several other statistical approaches: for example, an “appropriateness” index from item response theory (Dalal et al., 2015). In addition, Dalal et al. proposed numerous nonstatistical operationalizations of personality strength. For instance, “content-independent” operationalizations refer to those extant literatures that examine the consistency in behavior associated with other constructs such as attitudes. Given the similarity between attitudes and personality and given their similar effects on behavior (see S. J. Sherman & Fazio, 1983), attitude strength measures could be adapted to measure personality strength. A third category of operationalizations involves “content-general” individual differences such as self-monitoring and interpersonal dependency. Considerable future research is needed to identify the empirical interrelationships between these categories of operationalization, as well as the best operationalizations within each category.
Because this study examined the personality consistency of conscientiousness and agreeableness, future research could extend our findings using other personality traits within the Big Five and beyond (e.g., Honesty-Humility; K. Lee & Ashton, 2008). Moreover, research that simultaneously includes a large number of personality traits can begin to assess whether personality consistency is a unitary construct or trait specific—an important question that is as yet unresolved (Dalal et al., 2015). Future research could also examine forms of job performance beyond OCB: for instance, task performance (of particular relevance to selection-related implications) and CWB. Similarly, although we focused on particularly work-relevant situational dimensions, future research could focus on other situational characteristics (e.g., deception; Rauthmann et al., 2014). Furthermore, future research could examine the possibility of additional (i.e., beyond those specified in the Dalal et al., 2015, model) mediators of the relationship between personality consistency and performance consistency: for instance, consistency in the level and orientation of self-set goals.
Finally, in closing, although the current paper emphasized the prediction of behavior/performance consistency (rather than the mean level) via personality consistency (rather than personality traits), we suggest that the current paper nonetheless has implications for research on situational strength. Specifically, research on situational strength has typically viewed situational strength as a moderator (specifically, an attenuator) of the relationship between personality traits and behavior/performance mean levels. There has been considerable empirical support for this contention (e.g., S. Lee & Dalal, 2016; Meyer, Dalal, & Bonaccio, 2009; Meyer et al., 2014; see Meyer et al., 2010, for a review), which essentially pits personality against situations in an “either/or” framework. The theorizing in the current paper (see also Dalal et al., 2015), however, suggests the possibility of a second role for situational strength: namely, as a mediator of the relationship between personality and behavior/performance. This “both/and” perspective notes that far from being in direct opposition to personality, the role of situations (including situational strength) is often in harmony with—and in fact attributable to—personality. Future research should simultaneously examine these two perspectives on situational strength with the intent of determining the circumstances under which empirical results are better characterized by each perspective. The mediation perspective also suggests that nascent attempts to examine the antecedents of situational strength (Alaybek, Dalal, Sheng, Morris, Tomassetti, & Holland 2017) should feature not just situational but also personality antecedents.
