Test of a Social Cognitive Model of Proactive Career Behavior

Defining and Measuring Proactive Career Behavior

In a comprehensive review of the literature on proactive behavior in organizations over 20 years ago, Crant (2000) observed that, though the topic has received much attention from scholars, it has not produced “an integrated research stream … There is no single definition, theory, or measure driving this body of work; rather, researchers have adopted a number of different approaches toward identifying the antecedents and consequences of proactive behavior, and they have examined them in a number of seemingly disconnected literatures” (p. 435). Despite a great deal of valuable scholarship on work proactivity over the intervening years, Crant’s (2000) observation still appears to have a good deal of currency. For example, the behaviors of interest in the current study have been variously termed, among other things, proactive career behaviors (Claes & Ruiz-Quintanilla, 1998), enacted aspirations (Tharenou & Terry, 1998), career initiative (Parker & Collins, 2010), career engagement (Hirschi et al., 2013), and career competencies (Blokker et al., 2019). While different researchers have operationalized proactive career-enabling behaviors somewhat differently, the content overlap among their measures is often notable.

For the purposes of this study, we will adopt Parker and Collins’ (2010) definition of proactive career initiative behavior as the “individual’s active attempts to promote [their] career rather than a passive response to the job situation.” We will operationalize this concept with the proactive career behavior scale used by Claes and Ruiz-Quintanilla (1998), which borrows from earlier measures and has been adopted by a number of subsequent researchers (e.g., Strauss et al., 2012). The scale includes career planning, proactive skill development, career consultation, and network-building subscales. Though they do not constitute an all-inclusive set of career management strategies (many workers undoubtedly call on additional strategies, such as job-crafting, ongoing career exploration, preemptive job searching, and career change), they could be considered a fairly common and representative group of behaviors that can be intentionally used to help preserve and advance one’s career prospects.

A Social Cognitive Perspective on Proactive Career Behavior

Though the diversity of approaches to the study of proactive career behavior has clearly produced a fruitful stream of inquiry, we are mindful of Crant’s (2000) depiction of it as a literature lacking integration. In addition to Crant’s own effort at organizing this literature, other researchers have made notable efforts to classify and integrate a number of disparate approaches to the study of proactive workplace behaviors (e.g., Parker & Collins, 2010). In this article, we will use the CSM model (Lent & Brown, 2013) as a theoretical framework for conceptualizing proactive career initiative behaviors, their predictors, and their outcomes. We think the CSM model offers a viable approach for organizing at least a portion of the inquiry on proactive career behaviors, including efforts to understand how these behaviors can be acquired and modified and toward what ends, complementing studies that have explored their correlates more empirically than theoretically.

As the most recent of the five models of social cognitive career theory (SCCT; Lent et al., 1994; Lent & Brown, 2008 ), the CSM model was designed to help explain how people help to guide their own educational and career behavior throughout the lifespan, including processes involved in career preparation, entry, adjustment, change, withdrawal, and renewal. Brown and Lent (2019) recently reviewed the growing base of research on the CSM model, which has included applications to career exploration and decision-making (Lent et al., 2016), multiple role management (Roche et al., 2017), job search behaviors (Lim et al., 2016), management of sexual identity in the workplace (Tatum et al., 2017), and retirement planning (Penn & Lent, 2021).

Building on Bandura’s (1986) general social cognitive theory, SCCT highlights three central social cognitive mechanisms: self-efficacy, outcome expectations, and goals. Self-efficacy is defined as judgments of one’s ability to organize and execute behaviors required to perform particular tasks. Outcome expectations are beliefs about the consequences of performing certain behaviors or courses of action. Goals refer to the intention to engage in a particular activity or to perform it at a specified level of competence. As shown in Figure 1, self-efficacy and outcome expectations are hypothesized to predict career-related goals and, along with goals, to prompt corresponding adaptive behavior, such as engaging in career exploration activities which, in turn, enable more ultimate outcomes (e.g., higher levels of career decidedness, lower levels of decisional discomfort).OPEN IN VIEWER

As also shown in Figure 1, the CSM model features additional influences on career development, such as environmental, personality, and learning mechanisms. For example, environmental variables, often operationalized as supports and barriers (Lent et al., 2000), are posited to predict self-management outcomes both directly and indirectly via their linkages to the social cognitive variables. Personality factors also play important roles in the prediction of career outcomes within the model, with their specific pathways depending on the personality factor and the relevant career process under consideration. For example, persons high in conscientiousness are more likely to set specific goals for career exploration and decision-making and to transform these goals into actions (Lent & Brown, 2013).

The Present Study

This study was designed to extend prior research by assessing the tenability of a CSM-derived model of the relationships among a theory-based set of predictors and outcomes of proactive career behavior. We should note that SCCT’s models are designed to be tested with social cognitive variables that match one another along relevant dimensions, such as content and domain-specificity (Lent & Brown, 2006). Because we could not locate existing measures of self-efficacy and outcome expectations that were adequately tailored to the measure of proactive career behavior used in this study, we preceded model testing with a measure development phase involving initial validation of self-efficacy and outcome expectation scales linked to proactive career behaviors. This phase included additional measures (e.g., occupational role self-efficacy, negative career outlook) for validity estimation purposes.

We subsequently used the domain-specific social cognitive measures to test the CSM model. In addition to providing an omnibus test of the model, we examined invariance of the model across gender. We did not have specific hypotheses about model fit differences by gender but were interested in a preliminary exploration of the model’s range of generalizability, treating possible gender differences as an empirical question. The sample contained roughly equal and adequate numbers of men and women for invariance testing (relative sub-sample sizes for other possible demographic group comparisons were less adequate). Given the complexity of the larger CSM model presented in Figure 1 and the early stage of research applying it to proactive career behaviors, we streamlined the model test in a few ways, in particular, by limiting the number of theory-relevant predictors (e.g., goals were not included) and by examining the tenability of the relationships among the variables with a cross-sectional design. Such model streamlining has been common in other initial applications of the CSM model to different aspects of career development (e.g., Tatum et al., 2017).

Participants

The sample consisted of 511 adult workers (248 women, 263 men) who completed an online survey. Their mean age was 34.91 years (SD = 5.49). All participants reported being currently employed full-time (i.e., at least 40 h per week) within an organizational setting (i.e., not self-employed). We selected workers in the age range of 25–44 years, which conforms to Super’s Establishment stage, a period characterized by active efforts at career stabilization, consolidation, and advancement for many workers (Hartung, 2021). In terms of race/ethnicity, they identified as 66% White or European American, 12% Black or African American, 11% Latinx, and 9% Asian/Pacific Islander American, with 1% or fewer identifying as multiracial, Native American, or other ethnic groups. They were distributed throughout the U.S., with most living in the Mid-Atlantic, Southeast, and Midwest (22%, 21%, and 20%, respectively). In terms of education, 40% had a 4-year college degree, 27% a graduate or professional degree, 9% a 2-year college degree, and 24% a high school degree. They represented a wide range of (mostly white collar) occupations, with the largest groupings including information technology (16%), health (10%), education and training (9%), finance (7%), and manufacturing (7%). Annual salaries ranged from less than $15,000 (1%) to $200,000 or more (4%), with about half of the sample earning between $25,000 and $74,999 per year. Data were gathered within the first year of the COVID-19 pandemic.

Procedure and Instruments

Participants were individuals who had signed up to complete online surveys through Qualtrics Research Services (QRS). They received a modest incentive, such as credit toward gift cards, for their participation. Samples recruited through QRS have been shown to be reasonably representative of U.S. population demographics (Boas et al., 2018). To participate, respondents completed a university approved informed consent form and indicated that they were at least 25 years old and currently working at least 40 h per week. They were then provided access to the survey, which included two validity items used to screen out inattentive respondents. QRS also used bot detection to prevent non-human responses and IP address checks to prevent multiple responses from the same participant. We retained data only from participants who had responded appropriately to the screening mechanisms (e.g., passing both validity items) and who completed all demographic items, allowing us to confirm their fit to the sampling parameters (e.g., full-time workers). Total scale scores were calculated by summing item responses and dividing by the number of items on each scale.

Data Analysis

We first assessed the psychometric properties of the new measures of self-efficacy and outcome expectations prior to using them in model testing. Data from the first 200 participants were analyzed in the measurement development phase, providing initial estimates of reliability and validity for the novel scales. This phase included a factor analysis to test the factor structure of proactive behavior and its companion scales (self-efficacy and outcome expectations), followed by calculation of internal consistency reliability coefficients, descriptive statistics, and intercorrelations among the new and more established scales used to represent the CSM model. As part of the validity assessment, we also examined the utility of the new social cognitive measures in explaining unique variance in proactive career behavior while controlling for more global traits (proactive personality, negative outcome beliefs), supervisory support, and self-efficacy linked to general occupational role performance.

In the second, model testing phase of the study, data from the remaining 311 participants were used to assess the tenability of the posited measurement and structural models. The measurement model tested a correlated 8-factor representation of the CSM model variables, while the structural model examined hypothesized path relationships among the variables, as portrayed by Figure 2. Model testing employed the MLM estimation procedures of Mplus 8.4 (Muthén & Muthén, 1998-2019). Hu and Bentler’s (1999) 2-index method was used to assess adequacy of model-data fit. Using this method, fit may be considered adequate if (a) SRMR values ≤ .08 and (b) CFI values ≥ .95 or RMSEA values ≤ .06.

Initial Measurement Validation

We first examined the factor structures of the proactive behavior, self-efficacy, and outcome expectation scales. Strauss et al. (2012) had found support for a higher order structure of the proactive behavior items, with four first order factors (career planning, career consultation, skill development, and network-building), each of which loaded highly on a single second order factor. Using confirmatory factor analysis (CFA), we tested this structure by setting (a) each item to load on its corresponding subscale (or facet) and (b) each subscale to load on a common, second order factor. This model produced good fit to the data, SRMR = .07, RMSEA = .05 (90% CI = .02, .07), CFI = 0.97, Satorra-Bentler (S-B) χ² (50) = 71.11, p < .05. All items were found to load substantially on their subscale (range = 0.61–0.84), and the four subscales each loaded highly on the second order proactive behavior factor (range = 0.52–0.93). These results thus replicated the factor structure obtained by Strauss et al.

Because the self-efficacy and outcome expectation scales had been designed to parallel the four-facet content of the proactive behavior scale, we subjected each of them to an identical higher order CFA. Both analyses yielded support for the higher order structure: for self-efficacy, SRMR = 0.04, RMSEA = 0.06 (90% CI = 0.03, 0.08), CFI = .97, S-B χ² (50) = 80.40, p < 0.01; for outcome expectations, SRMR = 0.05, RMSEA = 0.06 (90% CI = 0.03, 0.08), CFI = 0.95, S-B χ² (50) = 81.94, p < .01. Item-subscale loadings for self-efficacy ranged between 0.57 and 0.78; for outcome expectations, the range was 0.54–.83. Loadings of the subscales on the second order factor ranged between 0.89 and 0.95 for self-efficacy and between 0.68 and 0.90 for outcome expectations. Thus, the findings suggest that the proactive behavior, self-efficacy, and outcome expectations scales may each be represented by four first order factors and one second order factor.

Given the CFA findings and our interest in the higher order constructs for present path modeling purposes, we calculated total scale scores representing each of the three constructs and examined their correlations with one another and the other scales. Table 1 contains summary statistics for the proactive behavior and social cognitive scales, proactive personality, occupational self-efficacy, negative career outlook, and the three criterion variables. Scale scores on all measures were distributed normally, with most skew and kurtosis values below 1; one kurtosis value (for proactive career behavior) was slightly above 1. Internal consistency reliability estimates ranged between 0.71 (internal and external job marketability) and 0.92 (supervisory support); values for the new self-efficacy and outcome expectation scales were, respectively, 0.91 and 0.88.

OPEN IN VIEWER

Table 1.

Correlations, Means, Standard Deviations, and Internal Consistency Estimates for Measure Development Phase.

Variables	1	2	3	4	5	6	7	8	9	10	M	SD	α
1. PC self-efficacy	—										3.75	0.60	0.91
2. PC outcome exp	0.56	—									4.06	0.60	0.88
3. Org support	0.51	0.53	—								3.61	0.91	0.92
4. PC behavior	0.68	0.46	0.44	—							3.88	0.56	0.86
5. Proactive psnlty	0.59	0.47	0.37	0.60	—						5.55	0.86	0.91
6. Car satisfaction	0.52	0.37	0.52	0.39	0.31	—					3.52	0.91	0.90
7. Org rewards	0.48	0.30	0.59	0.42	0.39	0.53	—				3.21	0.89	0.90
8. Int marketability	0.57	0.53	0.53	0.53	0.52	0.51	0.61	—			3.89	0.71	0.71
9. Ext marketability	0.46	0.20	0.12	0.37	0.31	0.25	0.24	0.41	—		3.98	0.66	0.71
10. Occ self-efficacy	0.56	0.56	0.36	0.41	0.58	0.38	0.28	0.47	0.39		4.91	0.76	0.87
11. Negative outlook	−0.28	−0.35	−0.17	−0.27	−0.40	−0.15	−0.06	−0.27	−0.21	−0.45	2.35	1.03	0.88

As expected, the proactive career self-efficacy and outcome expectation scales were substantially interrelated and both correlated moderately to strongly with the measures of proactive career behavior, supervisory support, proactive personality, and occupational role self-efficacy; they also produced small to moderate negative correlations with negative career outlook, suggesting that those possessing higher levels of proactive career self-efficacy and outcome expectations were less likely to hold pessimistic appraisals of their career futures. Both new social cognitive measures also produced mostly moderate to strong correlations with the criterion variables: career satisfaction, organizational rewards, and internal and external job marketability.

As an initial examination of the incremental utility of the new self-efficacy and outcome expectation scales, we predicted proactive behavior using a two-step hierarchical regression. Proactive personality, negative career outlook, occupational role self-efficacy, and supervisory support were entered at the first step, and proactive self-efficacy and outcome expectations were entered at the second step. The equation accounted for 42% of the variance in behavior at the first step and an additional 12% of the variance at the second step; however, only proactive self-efficacy (0.47) and proactive personality (0.33) produced significant beta weights at the final step of the equation. Collectively, these findings offer initial psychometric support for the new scales, especially proactive career self-efficacy.

Measurement and Structural Model Tests

To reduce the number of parameter estimates in relation to sample size in testing measurement and path models at the latent variable level, we represented each theoretical construct with subscales or item parcels. Specifically, proactive personality traits, supervisory support, career satisfaction, and organizational rewards—each of which had been shown to yield single-factor structures in prior research—were modeled with three item parcels apiece, with parcels created using the balancing method of Little et al. (2013). Job marketability was represented by its two subscales, internal and external marketability. Consistent with the higher order CFA findings, the proactive behavior, self-efficacy, and outcome expectations constructs were each indexed by their four parallel facets (career planning, skill development, career consultation, and network-building); we allowed the matching facets to covary across the three constructs to take their content similarities into account.

Limitations and Implications for Practice and Future Research

The study’s limitations should be considered in the context of efforts to interpret or generalize its findings; they also have important implications for future research. First, though the sample was geographically diverse and reasonably representative of U.S. working adults in terms of race/ethnicity, gender, and income level, the participants were mainly employed in white collar jobs and the method of recruitment (via an online research panel) required internet access and was prone to self-selection bias.

We also restricted the sample to workers at early to mid-career stages. While career proactivity and sustainability are likely to be relevant across the career lifespan, we reasoned that Establishment stage workers might be likely to be especially active at positioning themselves for career growth and advancement. Consistent with this assumption, the recent “Great Resignation” phenomenon appears to have been largely driven by job turnover among mid-career employees between 2020 and 2021 (Cook, 2021). Despite the relevance of proactivity to mid-career workers, future research should test the model with younger and older workers as well. Also, although we found invariance of model fit across gender in our sample, it would be valuable to test the CSM model’s generalizability in relation to a wider range of worker characteristics and contextual affordances (e.g., race/ethnicity, occupational type and level, educational level, and socio-economic status). It is likely that certain contexts, such as low-skilled employment, afford much less opportunity to engage in the sort of proactive behaviors we studied than do other, more resource-rich contexts.

Second, while we preceded model testing with a measure development and validation phase by splitting our sample, our measures of self-efficacy and outcome expectations had not been employed in prior research. We designed them to be consistent with SCCT measurement guidelines (Lent & Brown, 2006), including tailoring their content to parallel that of the previously established career initiative measure, and the new scales were found to yield promising reliability and validity estimates. However, the reference to common behaviors across the behavior, self-efficacy, and outcome expectation measures likely elevated their correlations, though not to the extent that they could not be differentiated from one another (see Lent & Brown’s, 2006, discussion of linked measurement issues). Still, there is need for future research to confirm the adequacy and predictive utility of the novel measures in independent samples.

Third, the measure of supervisory support we used may have captured only a portion of the career development support that workers receive. It did not, for example, assess support from fellow workers, mentors, or friends and family members. While supervisory support has produced stronger relationships to subjective career success/satisfaction than have measures of support from mentors and from more general sources of workplace support (Ng & Feldman, 2014), it would be valuable to examine additional aspects of environmental support in future research applying the CSM model to proactive career behavior engagement.

Fourth, to facilitate this study of proactive career behaviors from a social cognitive position, we streamlined the CSM model test in several ways, for example, by including only one aspect of personality and by omitting goals as a theorized antecedent of behavior. We also did not include other model elements, such as experiential/learning sources of self-efficacy and outcome expectations (cf. Ireland & Lent, 2018). These source variables (e.g., personal performance accomplishments, modeling) are particularly important to study because of their relevance to the design of interventions to promote career self-management.

Parenthetically, we should note that the proactive behavior measure we employed contains only a sampling of the behaviors that workers may use to protect and advance their career options. Future research might examine the extent to which this measure overlaps with or is distinct from other measures of career proactivity, exploration, engagement, and related constructs. Such research might address concerns about construct proliferation in the career literature (Brown, 2015) and also identify areas of convergence between vocational and organizational psychology inquiry on proactivity and career self-management. In this regard, the CSM model could be used as an integrative framework for studying additional ways in which workers may seek strategically to advance or sustain their career options (e.g., engagement in organizational citizenship, job-crafting, self-advocacy, and proactive job searching).

Fifth, the findings are limited by the cross-sectional nature of the study’s design. Paired with the measure development aspect of this study, the cross-sectional design was intended to build a foundation for further CSM-based study of proactive career behavior. Though the network of hypothesized paths we observed was largely consistent with the data, this does not constitute evidence of causation or temporal precedence. Longitudinal studies, especially those containing at least three time points, will provide more rigorous tests of mediation and temporal precedence; and experimental and theory-based intervention studies could also provide targeted tests of the model’s causal assumptions and yield a stronger foundation for its extension to practice. Sixth, the study was conducted within the first year of the COVID-19 pandemic. It is unclear what impact the pandemic may have had on the observed model relationships, but the findings do need to be interpreted in the context of the ongoing pandemic and the uncertainties and stresses it adds to career development.

Finally, all of the study’s measures involved self-report and tapped only the worker’s perspective. While predictors like self-efficacy and outcome expectations necessitate subjective appraisals because individuals are the ultimate arbiters of their own beliefs, it would be valuable to include study of variables that can be assessed from external or inter-subjective perspectives. For example, though we asked participants to report on their progress at receiving organizational rewards, such as promotions and pay raises, it would be useful to gather data on objective indicators of career success as well in future studies applying the CSM model to proactive career behaviors and their outcomes. Existing research suggests that proactivity indices are likely to yield stronger relationships to measures of subjective than objective career success (Fuller & Marler, 2009). This should not be too surprising because individuals are able to define the terms and standards of their success when it is assessed subjectively. In addition, no matter how well performed, workers often have little control over the linkage of their proactive behaviors to objective outcomes. For example, a worker who does “all the right things” is still subject to the appraisals of supervisors, organizational policies and biases, and the availability of resources for pay raises, promotions, and other rewards.

The present findings need to be replicated and extended in order to strengthen their implications for practice. However, they tentatively suggest that use of proactive behaviors—and the organizational supports and person factors (e.g., self-efficacy, proactive traits) that enable their use—may promote a variety of positive outcomes, at least from the worker’s perspective, such as an enhanced sense of job marketability. However, related to our above observation about the tenuous link between proactive career behaviors and indicators of objective success, an important caveat is that proactive behavior engagement is ultimately a matter of doing what one can to manage one’s career prospects, albeit with no certainty that it will protect one’s job or advance one’s career. In other words, career sustainability may be a critical aspiration for most workers but it is, unfortunately, not a given.

In sum, the current study extended the social cognitive CSM model to the prediction of proactive career initiative behavior and to beneficial outcomes associated with engagement in these types of behavior. We found support for most paths in the model, which may offer a promising approach to studying the predictors and outcomes of proactive career behavior. Further study of the model in the context of proactive career behavior may help to integrate an extremely important yet largely fragmented area of inquiry (Crant, 2000)—one that extends across several psychological specialties and other fields, such as education—and enable the design of theory-based strategies that career counselors and organizational practitioners can employ to support workers’ efforts at career persistence and advancement.