Operationalize Interest Congruence: A Comparative Examination of Four Approaches

Abstract

Although interest congruence is a cornerstone of career counseling, little is known about the relative importance of different operationalization approaches to interest congruence (i.e., how to calculate interest congruence). Using a sample of U.S. employees (n = 303), the current study comparatively examined four profile-based conceptual congruence approaches, namely Euclidean distance, angular agreement, profile deviance, and profile correlation, in terms of their predictions for job and life satisfaction, turnover intention, and perceived person–job fit. The results found that profile correlation demonstrated complete dominance (i.e., ubiquitously stronger predictive utility) over the other three congruence indices in predicting all four career outcomes. Therefore, the current study portrays profile correlation as a preferred operationalization approach to interest congruence and offers rich implications for congruence research and practice.

Keywords

interest congruence Holland theory operationalization profile correlation

Interest congruence, which denotes congruence between one’s interest profile and occupational environment, is a key foundation of vocational psychology and career counseling (Su et al., 2015; Wilkins & Tracey, 2014). Following Holland’s seminal model of vocational interests, extensive research has examined and supported the role of interest congruence in career development (Nye et al., 2017; Su et al., 2015); however, relatively little research has explored the optimal operationalization of interest congruence, which concerns how to determine interest congruence. More importantly, existing research on the operationalization of interest congruence (e.g.,(Brown & Gore, 1994; Camp & Chartrand, 1992; Young et al., 1998) predominantly examined congruence indices that capitalize on top interest letter(s) without taking into account important uninteresting or disliked activities in the whole interest profile (Dik et al., 2010; Phan & Rounds, 2018); therefore, the optimal operationalization of interest congruence based on the whole interest profile remains unclear (Su et al., 2015).

The profile-based operationalization of interest congruence is critical for career counseling because how to match a client’s vocational interest to occupational environment could influence the occupational results of interest assessment and consequently direct the process and outcomes of career counseling (Brown & Gore, 1994; Su et al., 2015). Therefore, to search for a clinically optimal operationalization of interest congruence based on the whole interest profile, the focus of the present study was to comparatively examine four profile-based operationalization approaches to interest congruence (briefly referred to as congruence in the following sections) using their relative predictions for four important career outcomes as the criteria.

Operationalization of Interest Congruence

Holland’s (1997) interest model has been the most influential for calculating congruence in the field likely because it explicitly formulates congruence based on a commensurate taxonomy of individual and environmental characteristics (Su et al., 2015; Wilkins & Tracey, 2014). This commensurate model describes both vocational interests and environment using orientations toward Realistic, Investigative, Artistic, Social, Enterprising, and Conventional (RIASEC) activities, whose relative similarity can be represented by a hexagon structure. Prediger and Vansickle (1992) further found through factor analysis that the RIASEC scores can be represented by two orthogonal dimensions: people/things (P/T) and ideas/data (I/D). Although the typological and dimensional representations of the interest profile might appear different, they are conceptually compatible and mathematically related because the six interest types represent six different combinations of the P/T and I/D dimensional scores, whereas the P/T and I/D dimensional system could capture the unique characteristics underlying the six interest types.

While Holland’s (1997) theory proposes congruence as a primary predictor of career success, initial research on congruence generated equivocal results, leading to a debate regarding the scientific utility and validity of Holland’s congruence (Tinsley, 2000). However, using more comprehensive study pools and profile-based information of RIASEC interests, recent meta-analytic and primary research has consistently revealed positive relationships between congruence and important career outcomes such as job satisfaction and academic performance (Nye et al., 2017; Tracey & Robbins, 2006). Although the associations between congruence and satisfaction (.16–.17; Tsabari et al., 2005) and between congruence and task performance (.27; Nye et al., 2017) appeared small to moderate in meta-analytical research, Meyer et al. (2001) noted that .20–.30 is a reasonable association range in social sciences. Therefore, Su et al. (2015) concluded that the evidence has robustly supported the relation of congruence with career development but called for research to address other issues of congruence, including the appreciable variances or moderators of the congruence–criterion link.

Among potential psychological and methodological moderators of the congruence–criterion link (Tracey, 2003; Tracey et al., 2012; Tsabari et al., 2005), the operationalization of congruence could be an important methodological moderator that has direct clinical implications. Different approaches to congruence operationalization, which depict how to calculate congruence based on what information, can be categorized into three major domains: top-letter(s)-oriented approaches, profile-based empirical approaches, and profile-based conceptual approaches (see Figure 1 for a summary). Top-letter(s)-oriented approaches, such as Holland’s (1997) congruence index and Brown and Gore’s (1994) C index, calculate congruence based on an individual’s top interest type(s) and have been the dominant operationalization of congruence in congruence research before the 21st century (Brown & Gore, 1994; Young et al., 1998). By contrast, profile-based empirical approaches, such as polynomial regression, operationalize congruence using the estimated score of congruence criteria (e.g., satisfaction) in a regression model involving individual and environmental RIASEC profiles (Edwards, 1994; Nye et al., 2018). Likewise, profile-based conceptual approaches, such as Euclidean distance (based on Prediger’s dimensional model), angular agreement (based on Prediger’s dimensional model), profile correlation (based on Holland’s typological model), and profile deviance (based on Holland’s typological model), use individual and environmental RIASEC profiles but specify the algorithm of congruence based on a priori considerations (e.g., adhere to the hexagon structure or not; Tracey et al., 2012).

Figure 1.
The taxonomic summary of congruence operationalizations.

Although these approaches are often correlated with one another (Young et al., 1998), they prioritize different RIASEC information in the matching process and could lead to different occupational recommendations with the same interest profile (Su et al., 2015). For example, in comparison to Holland’s (1997) congruence index, Brown and Gore’s (1994) C index could help an individual, who likes both R and I activities more than the other activities, obtain more reasonable occupational recommendations because the congruence index quantifies only the agreement of the first letter and could inappropriately ignore one of the R and I interests, while the C index quantifies the agreement of the top three letters and retains information about interests in both R and I activities. In fact, the congruence index and the C index could lead to different occupational recommendations in the ONET database because ONET lists anesthesiologist assistants, athletic trainers, computer and information research scientists, dentists, environmental engineers as the top five occupations for R and lists acupuncturists, anesthesiologist assistants, architectural and engineering managers, athletic trainers, and fuel cell engineers as the top five occupations for R and I. Therefore, it is reasonable to argue that the choice of congruence operationalization could directly affect the career counseling process and outcomes.

Given the clinical salience of the operationalization of congruence, there has been research exploring the optimal operationalization based on top-letter(s)-oriented approaches. Brown and Gore (1994) examined the ability of top-letter(s)-oriented approaches to differentiate individuals with like but out of order three-letter codes and recommended the C index. However, their study did not examine the criterion validity of competing approaches, which is a key aspect of an optimal operationalization. Therefore, Camp and Chartrand (1992) and Young et al. (1998) examined the predictions of top-letter(s) oriented indices for satisfaction but found nonsignificant relationships across competing congruence indices. In fact, meta-analytic research that used top-letter(s)-oriented approaches tended to find a fluctuating and often nonsignificant estimate of the congruence–satisfaction relationship (Nye et al., 2017), which has contributed to a previous debate about the validity of Holland’s congruence hypothesis (Tinsley, 2000). Nye et al. (2017) recently found in their meta-analysis that including more interest letters can improve the predictive power of congruence. Therefore, the accumulated evidence appears clear, suggesting that top-letter(s)-oriented approaches can hardly capture all meaningful information related to congruence and do not provide the best ground for the optimal congruence operationalization.

In comparison to top-letter(s)-oriented approaches, profile-based approaches have been widely conceptualized as a more accurate strategy of congruence operationalization because profile-based approaches more comprehensively capture RIASEC information and avoid arbitrary processing of tied scores (Tracey et al., 2012; Wille et al., 2014). Including less endorsed interests of the RIASEC profile in the calculation of congruence not only incorporates more information in congruence calculation but also embraces meaningful information about uninteresting or disliked activities, whose importance in predicting career criteria has been supported by research on incongruence (Dik et al., 2010; Phan & Rounds, 2018). Additionally, profile-based approaches have consistently revealed significant congruence–criterion relationships (e.g., Nye et al., 2018; Tracey et al., 2012) and thus represent a better direction than top-letter(s)-oriented approaches when searching for the optimal approach to congruence operationalization.

Between profile-based empirical and profile-based conceptual approaches, some scholars (Edwards, 1994; Nye et al., 2018) advocate regression-based polynomial congruence because it imposes less arbitrary constraints (e.g., weights of different letters and definitions of discrepancy) than conceptual approaches. However, we argue that conceptual approaches are more relevant for career counseling than polynomial congruence because like regression, which relies on sample information to maximize the prediction of predictors, polynomial congruence essentially capitalizes on sample data to estimate the upper boundary of the predictive utility of congruence without delineating how to operationalize congruence for an individual client. Practically speaking, the regression coefficients in polynomial congruence can be calculated only based on group information rather than individual information. Therefore, the relative utility of profile-based conceptual approaches appears to be a paramount issue for the clinical operationalization of congruence. However, the field has conducted limited conceptual and empirical exploration to clarify this issue, rendering the choice of the congruence operationalization largely inconsistent, if not unintentional, in the literature.

Profile-Based Conceptual Operationalizations of Congruence

To facilitate career counseling based on interest congruence, we thus comparatively examined four common and important profile-based conceptual approaches to congruence operationalization (Su et al., 2015; Tracey & Robbins, 2006): Euclidean distance (based on Prediger’s dimensional model), angular agreement (based on Prediger’s dimensional model), profile deviance (based on Holland’s typological model), and profile correlation (based on Holland’s typological model). Table 1 summarizes their differences and presents top five occupational recommendations associated with each approach for an illustrative real case (using occupational RIASEC scores of the ONET database). Notably, different operationalization approaches generate different occupational recommendations for the same illustrative case, demonstrating the clinical importance of choosing an appropriate congruence operationalization approach. Figure 2 provides visual demonstrations of the four approaches.

Table 1.
Application of Four Profile-Based Conceptual Approaches in an Illustrative Case.

Index Calculation Structural Assumption Focal Profile Difference Top Five ONET Occupations

Euclidean distance The square root of the aggregated squared deviance of the P/T and I/D scores between individual and environmental profiles Adherence to the normative hexagon and P/T–I/D structure Elevation difference on the P/T–I/D plane Merchandise displayers and window trimmers, dietetic technicians, informatics nurse specialists, physical therapist assistants, and cooks, private household

Angular agreement 1−D/90, where D represents the angular discrepancy between individual and environmental vectors on the P/T–I/D plane Adherence to the normative hexagon and P/T–I/D structure Angular (pattern) difference on the P/T–I/D plane Education teachers, postsecondary, recreational therapists, preschool teachers, except special education, singers, and fashion designers

Profile deviance The square root of the aggregated squared deviance of the RIASEC scores between individual and environmental profiles None Elevation difference of the RIASEC scores Instructional designers and technologists, architects, except landscape and naval, landscape architects, reporters and correspondents, and curators

Profile correlation Correlation between individual and occupational RIASEC scores None Pattern difference of the RIASEC scores Reporters and correspondents, film and video editors, poets, lyricists and creative writers, video game designers, and technical writers

Note. Individual and occupational P/T and I/D scores: P/T = (S + .50 × (A + E))/2 – (R + .50 × (C + I))/2 and I/D = .50 × (A + I) − .50 × (E + C). Individual RIASEC scores = 4.25, 5.50, 7.00, 4.25, 5.58, and 4.58, respectively. P/T = people/things; I/D = ideas/data; RIASEC = realistic, investigative, artistic, social, enterprising, and conventional activities.

Figure 2.
(Top) A visual demonstration of Euclidean distance and angular agreement. (Bottom) A visual demonstration of profile correlation and profile deviance.

Among the four approaches, Euclidean distance quantifies incongruence using the distance between interest and occupational profile locations on a plane defined by the P/T and I/D dimensions (Tracey & Robbins, 2006). Based on the structural correspondence between the typological and dimensional systems, the P/T scores = (S + .50 × (A + E))/2 – (R + .50 × (C + I))/2, while the I/D scores = .50 * (A + I) − .50 × (E + C). Angular agreement quantifies congruence using the angle between interest and occupational vectors on the P/T–I/D plane (Tracey & Robbins, 2006). Therefore, both Euclidean distance and angular agreement derive the P/T–I/D scores from the RIASEC scores based on the normative RIASEC and P/T–I/D structure. Research has examined the predictive validity of congruence using both Euclidean distance and angular agreement and found that Euclidean distance negatively predicted college success and angular agreement positively predicted college success (Tracey et al., 2012; Tracey & Robbins, 2006). In practice, the Personal Global Inventory (PGI) uses Euclidean distance to derive educational and vocational recommendations (Tracey, 2002).

In contrast to Euclidean distance and angular agreement, profile deviance quantifies incongruence using the square root of the aggregated squared difference between interest and occupational RIASEC scores (Su et al., 2015). Profile correlation quantifies congruence using the Pearson correlation between interest and occupational RIASEC scores (Su et al., 2015; Tracey et al., 2012). Notably, both profile deviance and correlation employ the RIASEC scores without relying on the P/T–I/D structure underlying the RIASEC scores; additionally, profile deviance treats the RIASEC scores separately and indicates the aggregated elevation differences between interest and occupational RIASEC scores, whereas profile correlation focuses on the relative strength of the RIASEC scores (i.e., the rank order) within each profile and indicates the deviance of the two RIASEC patterns. Research has supported the predictive validity of profile correlation in its positive prediction for college success (Allen & Robbins, 2010; Tracey et al., 2012). In practice, the ONET Interest Profiler uses profile correlation to derive educational and vocational recommendations (McCloy et al., 1999).

Optimal Operationalization: Typological Versus Dimensional and Pattern Versus Elevation

We argue that profile correlation likely outperforms Euclidean distance and angular agreement in predicting career outcomes because it is independent of the normative P/T–I/D and RIASEC structure. It is critical to note that although extensive research has supported the P/T–I/D and RIASEC structure (Rounds & Tracey, 1996; Xu & Tracey, 2016), such a structural arrangement is derived from RIASEC correlations in sample data and thus serve as a normative model describing the population-specific (instead of individual-specific) pattern of the RIASEC scores (e.g., Realistic and Social tend to be inversely associated; Tracey et al., 2006). However, research has shown that people’s individual profiles of the RIASEC scores exhibit varying adherence to the normative model (e.g., Realistic and Social could be similarly endorsed; Tracey, 2008; Tracey et al., 2006); thus, applying the normative model in the individual operationalization of congruence could inappropriately enforce population structural constraints in individual profiles (i.e., reduce six types to two orthogonal dimensions) and consequently generates inaccurate congruence results. For example, while an individual scores 5 on both Realistic and Social and three on the other types, Euclidean distance and angular agreement (based on the normative dimensional model) would conclude that this individual has no preference on and between the P/T and I/D dimensions and any well-rounded occupations could fit this person. However, this individual clearly leans toward Realistic and Social activities, and only profile correlation can detect this profile gravitation and point to R and S heavy occupations (e.g., dentists).

Additionally, we argue that profile correlation likely outperforms profile deviance in predicting career outcomes because it focuses on the RIASEC pattern (i.e., relative strength of the RIASEC scores). The importance of pattern congruence in fact is a core idea of Holland’s (1997) model, which can be seen in Holland’s prototypical congruence index. Although Holland’s congruence index focuses on the agreement of only the top interest letter, it advocates the match between interest and occupational profile patterns because the top letter in fact is a simplistic representation of the RIASEC pattern (Holland, 1997). Additionally, Holland’s (1997) theory proposes pattern-based congruence as a primary predictor of career success and conceptualizes differentiation of interest elevations as a contributing factor of career indecision, suggesting that pattern-based congruence is plausibly a better index than elevation-based congruence when facilitating career success is the primary goal. Therefore, while profile correlation expands the information source from the top letter to the whole RIASEC profile, it maintains Holland’s theoretical focus on the RIASEC pattern. By contrast, profile deviance directly builds on the individual RIASEC scores instead of their relative strength; thus, it heavily capitalizes on the separate RIASEC elevations and could be only remotely associated with the theoretically important RIASEC pattern (Su et al., 2015). Given the important theoretical differences of Euclidean distance, angular agreement, profile deviance, and profile correlation, it is plausible to argue that the current comparative examination could not only help identify an optimal operationalization but also shed light on two important but often ignored issues in congruence research and theory: using the normative structure in individual congruence and the relative importance of pattern match.

In this study, we used four important career outcomes in employees as evaluative criteria: job satisfaction, life satisfaction, turnover intention, and perceived person–job fit. Both job and life satisfaction are important well-being-related career outcomes (Lent, 2004) with job satisfaction focusing on work-specific fulfillment and life satisfaction focusing on general work–life fulfillment and indicating a distal outcome of career development. Turnover intention is also an important career criterion because it indicates employee’s persistence in their job. In fact, satisfaction and persistence are two major outcomes of interest congruence in Holland’s (1997) theory. Perceived person–job fit is another important criterion because it represents a subjective evaluation of general person–environment correspondence and should result from objective interest congruence (Wessel et al., 2008).

Summary of the Present Study

In summary, the tension between the clinical importance of operationalizing interest congruence and the knowledge gap regarding the optimal operationalization constitutes a compelling rationale for our comparative examination of four profile-based conceptual congruence indices, namely Euclidean distance, angular agreement, profile deviance, and profile correlation. Because profile correlation capitalizes on the individual RIASEC pattern without relying on the normative P/T–I/D and RIASEC structure, we hypothesized that profile correlation outperforms the other three indices in predicting job satisfaction, life satisfaction, turnover intention, and perceived person–job fit.

Method

Sample

The current sample consisted of 303 employees from the United States ranging in age from 19 to 50 years (M = 39.20, SD = 11.64). Of the sample, 39.3% (n = 119) identified as men, 60.4% (n = 183) identified as women, and 0.3% (n = 1) identified as gender nonbinary. In terms of race/ethnicity, 74.9% (n = 227) identified as White/European American, 7.9% (n = 24) identified as African/African American/Black, 5.6% (n = 17) identified as Hispanic/Latinx American, 7.3% (n = 22) identified as Asian/Asian American, 0.3% (n = 1) identified as American Indian/Native American, 0.7% (n = 2) identified as Arab American/Middle Eastern/North African, and 3.3% (n = 10) identified as multiracial. In terms of social economic status, 4.0% (n = 12) identified as lower class, 35.6% (n = 108) identified as working class, 49.5% (n = 150) identified as middle class, 10.6% (n = 32) identified as upper-middle class, and 0.3% (n = 1) identified as upper class. In terms of employment status, 85.8% (n = 260) identified as full-time employees and 14.2% (n = 43) identified as part-time employees. They reported a variety of occupations including sales, manager, teacher, engineer, casher, nurse, chemist, journalist, and accountant.

Procedure

We recruited participants through an online crowdsourcing platform Amazon Mechanical Turk (MTurk). MTurk has been a popular data collection platform in social science allowing researchers to access targeted populations in a relatively affordable and valid way (Buhrmester et al., 2011; Casler et al., 2013; Paolacci & Chandler, 2014; Ramsey et al., 2016). Research has also shown that with an appropriate design MTurk can serve as a valuable data collection resource for counseling psychology research (e.g., Dahling et al., 2013; Xu, 2020). Following a procedure used previously in similar studies, we invited voluntary participants to fill out a demographic form and the research questionnaires for monetary compensation (US$1, which is appropriate for the length and complexity of the survey; Hara et al., 2018). All participants’ responses remained anonymous and confidential throughout the study. To enhance data validity, we only invited participants with an above 95% history approval rate and used four validity screening items (e.g., “please choose ‘3’”) throughout the survey to retain valid responses (303 of 378 total responses), which were defined as those following the instruction of the validity items. After excluding invalid responses, we did not detect missing values in the final data set. Institutional review board approval for this study was obtained.

Measures

Vocational interest

We used the short version of PGI (PGI-Short; Tracey, 2010) to measure vocational interest. The PGI-Short has psychometric advantages over traditional interest measures such as the Strong Interest Inventory and the Self-Directed Search because while it resembles the traditional interest measures in terms of developing items based on the circular model of vocational interests, it additionally used item response theory to improve item effectiveness and reduce differential item functioning across gender. The inventory contains 40 activities that derive eight basic interest scales (i.e., Social Facilitating, Managing, Business Detail, Data Processing, Mechanical, Natural/Outdoors, Artistic, and Helping) around the circular structure on the plane defined by the P/T and I/D dimensions. While the eight interest scales provide a more fine-grained assessment of interest, the inventory also generates the RIASEC scores based on the eight interest scales. Two sample items were “sculpture a statue” and “drive a bus.” Participated rated each item on a 7-point Likert-type scale ranging from 1 (strongly dislike) to 7 (strongly like). Higher scores indicate a higher level of interests. Tracey (2010) reported α coefficients of .73–.83 across interest types and found strong support for the structural and convergent validity of the PGI-Short. The current study revealed α coefficients of .60–.88 for the interest scales of the PGI-Short.

Job satisfaction

We used the Michigan Organizational Assessment Questionnaire Job Satisfaction subscale (JSS; Cammann et al., 1979) to measure job satisfaction. The 3-item JSS is one of the scales most commonly used to measure an individual’s general sense of satisfaction with the current job (Bowling & Hammond, 2008). A sample item was “In general, I like working here.” Participants rated each JSS item on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicated a higher level of job satisfaction. Bowling and Hammond (2008) found meta-analytic evidence that supported the internal consistency reliability of the JSS (.84) and the construct validity of the JSS (in associations between JSS and a variety of antecedents, consequences, and correlates of job satisfaction). The current study found an α coefficient of .94 for the JSS.

Life satisfaction

We used the Satisfaction With Life Scale (SWLS; Diener et al., 1985) to measure life satisfaction. The 5-item SWLS is perhaps the most widely adopted measure of life satisfaction in the literature (Diener et al., 2013). A sample item was, “The conditions of my life are excellent.” Participants were invited to rate the SWLS items on a 7-point Likert-type scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicated a higher level of life satisfaction. Diener et al. (2013) summarized the psychometric evidence of the SWLS and found strong support for its internal consistency reliability (above .80) and construct validity (in associations between the SWLS and non-self-report measures of satisfaction and societal difference). The current study found an α coefficient of .91 for the SWLS.

Turnover intention

We measured turnover intention using the Turnover Intention Scale (TIS; Bothma & Roodt, 2013). A sample item was, “How often have you considered leaving your job?” Participants were invited to rate the TIS items on a 5-point Likert-type scale ranging from 1 (never) to 5 (always). Higher scores indicated a higher level of turnover intention. Bothma and Roodt (2013) found support for the internal consistency reliability of the TIS (.80) and support for the criterion validity of the TIS in its ability in differentiating actual leavers and stayers. The current study found an α coefficient of .78 for the TIS.

Perceived person–job fit

We measured perceived person–job fit using the Person-Job Fit Scale (PJFS; Chuang et al., 2016). Building on person–environment correspondence theories, the 4-item PJFS measures an individual’s perception of fit between the self and the current job in terms of interests, personality, skills, and style and thus offers a brief but comprehensive assessment of correspondence perception. A sample item was, “How would you describe the match between your interests (e.g., social vs. unsocial, artistic vs. inartistic, and conventional vs. unconventional) and those you desire for a job?” Participants rated the PJFS items on a 7-point Likert-type scale ranging from 1 (no match) to 7 (complete match). Higher scores indicated a stronger level of perceived fit. Chuang et al. (2016) reported an α coefficient of .84 for the PJFS and supported the construct validity of the PJFS in its associations with other fit measures and in its predictions for important work-related criteria (e.g., job satisfaction and organizational citizenship behavior). The current study found an α coefficient of .83 for the PJFS.

Analysis

Based on the participants’ reported occupational title and description of occupational activities and duties, we identified their Standard Occupational Classification (SOC) occupational title and occupational RIASEC profile in the ONET database. We first used their self-reported occupational title to search for a matched SOC occupation in the ONET database. If such an attempt was unsuccessful, we alternatively searched by the participants’ description of occupational activities and duties. Specifically, three counseling psychology doctoral students, who had received training on vocational psychology and interest assessment, each conducted the same searching procedure (i.e., used the participants’ self-reported occupation title and description to find their SOC title in the ONET database) for a third of the participants. Meanwhile, a vocational psychology scholar independently conducted the same searching procedure for all the participants. We later compared the separate coding results and reviewed potential discrepancy to reach consensus regarding the participants’ SOC occupations. Because each occupation is linked to an RIASEC profile in the ONET database, we identified each participant’s occupational RIASEC scores based on their SOC occupational title.

Then, we calculated individual and occupational P/T scores using P/T = (S + .50 × (A + E))/2 – (R + .50 × (C + I))/2 and calculated the I/D scores using I/D = .50 × (A + I) − .50 × (E + C). Last, we calculated Euclidean distance (= the square root of the aggregated squared deviance of the P/T and I/D scores between individual and environmental profiles), angular agreement (= 1−D/90, where D represents the angular discrepancy between individual and environmental vectors on the P/T–I/D plane; Fisher et al., 1985), profile deviance (= the square root of the aggregated squared deviance of the RIASEC scores between individual and environmental profiles), and profile correlation (= the correlation between individual and occupational RIASEC scores).

We used dominance analysis (Azen & Budescu, 2003) to examine the relative importance of Euclidean distance, angular agreement, profile deviance, and profile correlation in predicting job satisfaction, life satisfaction, turnover intention, and perceived person–job fit. Dominance analysis is appropriate for this study because by comparing the additive predictions of two predictors in every possible baseline model (defined as every possible subset of remaining predictors), dominance analysis could indicate the proportion of models in which one predictor (we focused on profile correlation in the present study) outperforms the other one. In other words, for each career criterion, dominance analysis compares the additive predictions of profile correlation and another congruence approach to indicate which congruence approach could uniquely explain more variance of that criterion. Dominance analysis replicates this procedure across all possible baseline models, which consist of congruence approaches other than profile correlation and the compared congruence approach; therefore, the final proportion of profile correlation being a superior predictor took account of covariance among predictors. We were particularly interested in the complete dominance of profile correlation, which occurs when profile correlation is more important than the other predictor in all possible baseline models (Azen & Budescu, 2003). Complete dominance indicates the highest level of relative importance of a predictor. We examined the four career criteria separately because they captured related but distinct aspects of career development.

Results

Table 2 shows the means, standard deviations, and correlations of the four congruence indices and four career criteria, which generally indicate that the analysis variables were correlated with one another. We first examined outliers and normality of the data and found that the standardized scores of all the variables were below 3 and the absolute skewness and kurtosis values of all the variables were below 1, suggesting that no outliers exited in the data set, and all the variables were normally distributed (Weston & Gore, 2006). Additionally, we calculated the correlation of gender and employment status with each index of congruence and did not find significant correlations, indicating that the two demographic variables were accounted for in later analyses. Therefore, we continued to conduct dominance analysis.

Table 2.
Means, Standard Deviations, and Correlations of Variables.

Variable Mean SD Euclidean Distance Angular Agreement Profile Deviance Profile Correlation Job Satisfaction Life Satisfaction Turnover Intention

Euclidean distance 3.42 1.54 1.00

Angular agreement 0.21 0.56 −0.72 1.00

Profile deviance 5.36 1.47 0.77 −0.71 1.00

Profile correlation 0.19 0.48 −0.71 0.85 −0.84 1.00

Job satisfaction 5.12 1.50 −0.16 0.17 −0.14 0.18** 1.00

Life satisfaction 4.41 1.43 −0.10 0.10 −0.14* 0.15 0.45 1.00

Turnover intention 2.87 0.84 0.19 −0.18 0.18 −0.21 −0.71 −0.33 1.00

Perceived person–job fit 5.10 1.13 −0.16 0.16 −0.14* 0.20 0.58 0.29 −0.48

Note. N = 303.

*p < .05. **p < .01.

Table 3 summarizes the dominance analysis results. First, regarding job satisfaction, the results showed that profile correlation provided greater ΔR² in pairwise comparison to any of the three competing congruence indices (i.e., Euclidean distance, angular agreement, and profile deviance) in any prediction models (proportion = 100%), suggesting that profile correlation is constantly a better predictor of job satisfaction than any of the three competing congruence indices. Second, the results showed that profile correlation contributed greater ΔR² of life satisfaction in pairwise comparison to any of the three competing congruence indices in any prediction models (proportion = 100%), suggesting that profile correlation is constantly a better predictor of life satisfaction than any of the three competing congruence indices. Third, the results showed that profile correlation contributed greater ΔR² of turnover intention in pairwise comparison to any of the three competing congruence indices in any prediction models (proportion = 100%), suggesting that profile correlation is constantly a better predictor of turnover intention than any of the three competing congruence indices. Last, the results showed that profile correlation contributed greater ΔR² of perceived person–job fit in pairwise comparison to any of the three competing congruence indices in any prediction models (proportion = 100%), suggesting that profile correlation is constantly a better predictor of perceived person–job fit than any of the three competing congruence indices. Therefore, the results of dominance analysis collectively supported the complete dominance of profile correlation over Euclidean distance, angular agreement, and profile deviance in predicting all four career outcomes.

Table 3.
Summary of Dominance Analysis Results.

Predictor Variable Euclidean Distance Angular Agreement Profile Deviance Profile Correlation

Job satisfaction

Euclidean distance 0.00 0.50 1.00 .00

Angular agreement 0.50 0.00 0.50 .00

Profile deviance 0.00 0.50 0.00 .00

Profile correlation 1.00 1.00 1.00 .00

Life satisfaction

Euclidean distance 0.00 0.25 0.00 .00

Angular agreement 0.75 0.00 0.50 .00

Profile deviance 1.00 0.50 0.00 .00

Profile correlation 1.00 1.00 1.00 .00

Turnover intention

Euclidean distance 0.00 1.00 1.00 .00

Angular agreement 0.00 0.00 0.25 .00

Profile deviance 0.00 0.75 0.00 .00

Profile correlation 1.00 1.00 1.00 .00

Perceived person–job fit

Euclidean distance 0.00 0.50 0.50 .00

Angular agreement 0.50 0.00 0.50 .00

Profile deviance 0.50 0.50 0.00 .00

Profile correlation 1.00 1.00 1.00 .00

Note. N = 303. The table displays the proportion of models in which inclusion of the row variable results in a larger ΔR² than inclusion of the column variable. Boldface type indicates the dominance frequency of profile correlation.

Discussion

Although interest congruence serves as the cornerstone of career counseling, its optimal operationalization remains unclear. Thus, the current study comparatively examined four profile-based conceptual congruence approaches, Euclidean distance, angular agreement, profile deviance, and profile correlation, in terms of their predictions for job and life satisfaction, turnover intention, and perceived person–job fit. The results found that profile correlation demonstrated complete dominance (i.e., ubiquitously stronger predictive utility) over the other three congruence indices in predicting all four career outcomes. In general, the findings support the superior importance of profile correlation in comparison to the other three operationalization approaches to congruence and offer important theoretical and practical implications for vocational psychology on interest congruence.

Idiosyncratic Structure Versus Normative Model in Interest Congruence

Perhaps one of the most important implications of the advantage of profile correlation over Euclidean distance and angular agreement is that it provides insights into a long-standing but rarely explored issue in congruence research, which concerns the appropriateness of using the normative structure in estimating individual congruence. While extensive research has supported the structural validity of the normative circular model and the P/T–I/D system of the RIASEC interests (Rounds & Tracey, 1996; Xu & Tracey, 2016), research has also shown that people’s individual RIASEC structure does not always adhere to this normative model (Tracey, 2008; Tracey et al., 2006). Although the normative model helpfully depicts the population RIASEC structure and consequently provides useful guidance for interest measure development, calculating congruence is essentially an individual-centered process that builds on the idiosyncratic RIASEC structure. Therefore, enforcing the normative RIASEC model in congruence practice and research ignores individual differences (and potential occupational difference) in terms of adherence to the normative model and could result in inaccurate and even distorted congruence evaluations. Interestingly, although research has acknowledged such individual differences and explored the benefit of using the normative RIASEC structure to organize information (Tracey, 2008), there is a clear conceptual oversight regarding the implication of assuming the normative RIASEC structure in individual congruence. This issue can also be seen in research (e.g., Tracey et al., 2012) that jointly used Euclidean distance and profile correlation as the congruence indicators but did not compare their predictions for career outcomes. Thus, by comparing the predictive performance of normative model-independent and normative model-dependent indices, the current study served as the first study examining and supporting the importance of adopting an idiosyncratic approach to interest congruence. Notably, at the surface level, the results showed that more data points generated stronger congruence–criterion correlations; however, it is the theoretical implication regarding the use of a normative model in congruence calculation that marks one of the major contributions of the present study.

Complete Match Versus Pattern Congruence in Interest Congruence

As interesting as the relative importance of profile correlation, Euclidean distance, and angular agreement is, the advantage of profile correlation over profile deviance provides insights into the relative importance of complete match and pattern congruence. Cronbach and Gleser (1953) in their seminal article outlined three parameters of multivariate profiles: elevation (i.e., level), dispersion (i.e., variance), and pattern (i.e., rank order). Because profile deviance builds on the quantitative discrepancy across all six interest types, it takes account of differences in all three parameters of elevation, dispersion, and pattern (Cronbach & Gleser, 1953; Tracey, 2003) and essentially reflects a deviation from a complete match; by contrast, profile correlation by definition capitalizes on pattern congruence. The relative importance of complete match and pattern congruence taps into a substantive question regarding the dominant criterion of profile similarity. Although it might make intuitive sense that a complete match could lead to optimal outcomes of person–environment interaction, Holland’s (1997) model places a theoretical emphasis on pattern congruence and implicitly discounts other two congruence criteria (i.e., elevation and dispersion congruence). The current study served as the first research examining Holland’s implicit hypothesis regarding the dominant role of pattern congruence and suggested that pattern congruence is a key congruence criterion and elevation and dispersion congruence contribute less to career outcomes. The current findings might speak to a phenomenon that people are more sensitive to whether their major occupational activities align with their dominant interests than to the extent to which they are fulfilling each interest (or they tend to be intolerant of incongruence of interest priority but adaptive to mismatched interest intensity). Certainly, these speculations require future research and theory development.

Research and Practical Implication: Operationalize Interest Congruence

Revealing the superiority of profile correlation in predicting career outcomes, the current study also provides implication for operationalizing congruence in research. Notably, because conceptual approaches impose potentially arbitrary constraints on the operationalization of congruence, the current examination of the four profile-based conceptual congruence indices might have less direct implications for evaluating the optimal predictive utility of congruence than data-driven approaches. However, the current findings still meaningfully suggested that profile correlation could be a sound index to congruence because it captures the key element of interest congruence that relates to career outcomes. Xu (2020) noted that while extensive research has acknowledged congruence as a valid predictor of career outcomes, how to achieve congruence remains less clear. Therefore, we encourage future research to examine the predictors of profile correlation to enhance the literature on the development mechanism of interest congruence.

Furthermore, while the field has diverse operationalizations of congruence, the current study sheds light on the potential optimal operationalization of congruence in interest assessment and interpretations. Given its superior importance for all four career outcomes, counselors can more heavily rely on profile correlation (i.e., pattern congruence) than on the other three approaches to derive occupational options. Interestingly, the ONET Interest Profiler is probably the only free interest assessment that offers occupational recommendations based on profile correlation; therefore, it could be a good starting point of career assessment. However, the ONET platform still emphasizes the top three interest letters in searching for occupations and does not offer easy access to occupational information based on the full RIASEC pattern. Because the current results portrayed profile pattern as a critical aspect of congruence, we encourage existing and prospective interest assessments and occupational databases to design the occupation searching system based on the full RIASEC pattern and use profile correlation to derive occupational recommendations. That said, using the whole RIASEC pattern requires careful comparison of the RIASEC scores and demands more information particularly on the occupational end; thus, it could be unrealistic for extremely complex situations (e.g., multiple tied scores) and might not be as practical for career counseling use as a top-letter(s) match. Therefore, we encourage practitioners to tailor their congruence approach to the counseling situation.

It should be noted that we used the environmental scores from the O*NET database, which largely uses job analysis to generate occupational RIASEC scores. Therefore, the current results might not be generalizable to congruence-related research and practice that uses others in the occupation to generate occupational RIASEC scores such as the Strong Interest Inventory (Dik et al., 2007). However, research on the moderation of the type of environment measure has not decisively suggested a superior system of occupational scores (Tsabari et al., 2005), and more importantly, it remains unclear whether different systems of occupational scores could change the pattern of relative importance of the four congruence approaches. Therefore, it is necessary for future research to replicate the current study using the environmental RIASEC scores from the Strong Interest Inventory.

Notably, while profile correlation outperformed the other three congruence approaches, an eyeball examination of the predictive magnitude of all four approaches revealed two issues that are worth discussion. First, the advantage of profile correlation in terms of the numeric predictive strength did not appear substantial. We argue that the small difference in predictive strength is a more severe threat to the differential use of the four approaches in congruence research than to the differential use of congruence indices in counseling because a small difference in predictive magnitude is unlikely to generate vastly different results regarding the congruence–criterion link; however, different operationalization approaches based on a small quantitative difference could lead to qualitatively different occupational choices that likely have consequential influences on educational choices, career paths, and even life directions (see different occupations showing different primary interest types in Table 1). Therefore, even though the numeric difference appeared small, the predictive superiority of profile correlation still has meaningful implications for career counseling that uses congruence to direct the individual assessment and interpretation process.

Relatedly, the current study found that the predictions of all four approaches for the criteria were small to moderate, which is consistent with previous research (Nye et al., 2017; Tsabari et al., 2005). We propose that one way to address this challenge to Holland’s theory is to continue investigating psychological and methodological moderators. While research has supported the moderation of interest flexibility, environmental constraints, and cultural norms (Tracey, 2003; Tracey et al., 2012; Tsabari et al., 2005), it would be interesting for future research to examine the moderation of personality, work centrality, values, and interest measures. Additionally, given Meyer et al.’s (2001) notion on the reasonable association range in social sciences, it might not be surprising that a single construct of interest congruence cannot explain a substantial proportion of the outcomes of complex person–environment interaction. However, it may be important for future research to integrate interest congruence with value and ability congruence when examining the broader idea of person–environment fit.

Limitations and Suggestions for Future Research

Several limitations should be noted when understanding the current results. First, the current study focused on subjective career outcomes (e.g., job satisfaction) because of their direct relation to well-being (Lent, 2004). However, the results might not be generalizable to objective career outcomes (e.g., performance), which are also important for people’s career decision making. Therefore, it would be interesting for future research to examine the relative prediction of the four competing operationalization approaches for objective career outcomes. Second, the current study adopted a cross-sectional approach in comparing the predictions of the four congruence indices. However, a cross-sectional examination might ignore a potential confounding variable of person–environment mutual selection (Wilkins & Tracey, 2014). Although the mutual selection effect likely dampens the predictions of all four operationalization approaches, it could affect them differently (e.g., people might more quickly select themselves out based on elevations than based on patterns) to make the current results possible. Therefore, we encourage future research to use a longitudinal design to replicate the current study. Last, we examined the relative importance of the four competing congruence approaches in a Western context, which might capitalize on people’s intolerance of pattern incongruence (e.g., engage in relatively boring activities) in Western cultures. However, other cultural contexts (e.g., Eastern cultures) might emphasize following organizational structure (e.g., organizationally prescribed activities) in person–environment interaction (Leung et al., 2011) and consequently elevate the importance of level and dispersion congruence. To examine the cross-cultural validity of the current conclusion, it is necessary for future research to compare the four operationalizations of interest congruence in contexts (e.g., East Asian countries) that are culturally different from North America.

Conclusions

Although interest congruence is a cornerstone of career counseling, little is known about the relative importance of different profile-based conceptual operationalization approaches to congruence. The current study comparatively examined four operationalization approaches Euclidean distance, angular agreement, profile deviance, and profile correlation and found support for the superiority of profile correlation over the other three approaches. The findings not only portray profile correlation as the preferred approach in operationalizing congruence in practice and research but also speak to the issue of using the normative RIASEC structure in individual congruence and the centrality of pattern similarity in congruence calculation. To conclude, the current study further refines Holland’s approach to interest congruence by providing important guidance on its operationalization.

Index	Calculation	Structural Assumption	Focal Profile Difference	Top Five O*NET Occupations
Euclidean distance	The square root of the aggregated squared deviance of the P/T and I/D scores between individual and environmental profiles	Adherence to the normative hexagon and P/T–I/D structure	Elevation difference on the P/T–I/D plane	Merchandise displayers and window trimmers, dietetic technicians, informatics nurse specialists, physical therapist assistants, and cooks, private household
Angular agreement	1−D/90, where D represents the angular discrepancy between individual and environmental vectors on the P/T–I/D plane	Adherence to the normative hexagon and P/T–I/D structure	Angular (pattern) difference on the P/T–I/D plane	Education teachers, postsecondary, recreational therapists, preschool teachers, except special education, singers, and fashion designers
Profile deviance	The square root of the aggregated squared deviance of the RIASEC scores between individual and environmental profiles	None	Elevation difference of the RIASEC scores	Instructional designers and technologists, architects, except landscape and naval, landscape architects, reporters and correspondents, and curators
Profile correlation	Correlation between individual and occupational RIASEC scores	None	Pattern difference of the RIASEC scores	Reporters and correspondents, film and video editors, poets, lyricists and creative writers, video game designers, and technical writers

Variable	Mean	SD	Euclidean Distance	Angular Agreement	Profile Deviance	Profile Correlation	Job Satisfaction	Life Satisfaction	Turnover Intention
Euclidean distance	3.42	1.54	1.00
Angular agreement	0.21	0.56	−0.72**	1.00
Profile deviance	5.36	1.47	0.77**	−0.71**	1.00
Profile correlation	0.19	0.48	−0.71**	0.85**	−0.84**	1.00
Job satisfaction	5.12	1.50	−0.16**	0.17**	−0.14*	0.18**	1.00
Life satisfaction	4.41	1.43	−0.10	0.10	−0.14*	0.15**	0.45**	1.00
Turnover intention	2.87	0.84	0.19**	−0.18**	0.18**	−0.21**	−0.71**	−0.33**	1.00
Perceived person–job fit	5.10	1.13	−0.16**	0.16**	−0.14*	0.20**	0.58**	0.29**	−0.48**

Predictor Variable	Euclidean Distance	Angular Agreement	Profile Deviance	Profile Correlation
Job satisfaction
Euclidean distance	0.00	0.50	1.00	.00
Angular agreement	0.50	0.00	0.50	.00
Profile deviance	0.00	0.50	0.00	.00
Profile correlation	1.00	1.00	1.00	.00
Life satisfaction
Euclidean distance	0.00	0.25	0.00	.00
Angular agreement	0.75	0.00	0.50	.00
Profile deviance	1.00	0.50	0.00	.00
Profile correlation	1.00	1.00	1.00	.00
Turnover intention
Euclidean distance	0.00	1.00	1.00	.00
Angular agreement	0.00	0.00	0.25	.00
Profile deviance	0.00	0.75	0.00	.00
Profile correlation	1.00	1.00	1.00	.00
Perceived person–job fit
Euclidean distance	0.00	0.50	0.50	.00
Angular agreement	0.50	0.00	0.50	.00
Profile deviance	0.50	0.50	0.00	.00
Profile correlation	1.00	1.00	1.00	.00

Footnotes

Acknowledgments

We thank Papa Adams and Daewon Kim for their assistance in coding.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Hui Xu

References

Allen

Robbins

(2010). Effects of interest-major congruence, motivation, and academic performance on timely degree attainment. Journal of Counseling Psychology, 57, 23–35. https://doi.org/10.1037/a0017267

Azen

Budescu

D. V.

(2003). The dominance analysis approach for comparing predictors in multiple regression. Psychological Methods, 8, 129–148. https://doi.org/10.1037/1082-989X.8.2.129

Bothma

C. F.

Roodt

(2013). The validation of the turnover intention scale. SA Journal of Human Resource Management, 11, 1–12. https://doi.org/10.4102/sajhrm.v11i1.507

Bowling

N. A.

Hammond

G. D.

(2008). A meta-analytic examination of the construct validity of the Michigan Organizational Assessment Questionnaire Job Satisfaction Subscale. Journal of Vocational Behavior, 73, 63–77. https://doi.org/10.1016/j.jvb.2008.01.004

Brown

S. D.

Gore

P. A.

(1994). An evaluation of interest congruence indices: Distribution characteristics and measurement properties. Journal of Vocational Behavior, 45, 310–327. https://doi.org/10.1006/jvbe.1994.1038

Buhrmester

Kwang

Gosling

S. D.

(2011). Amazon’s Mechanical Turk: A new source of inexpensive, yet high-quality, data? Perspectives on Psychological Science, 6, 3–5. https://doi.org/10.1037/e527772014-223

Cammann

Fichman

Jenkins

Klesh

(1979). The Michigan organizational assessment questionnaire [Unpublished manuscript]. University of Michigan.

Camp

C. C.

Chartrand

J. M.

(1992). A comparison and evaluation of interest congruence indices. Journal of Vocational Behavior, 41, 162–182. https://doi.org/10.1016/0001-8791(92)90018-u

Casler

Bickel

Hackett

(2013). Separate but equal? A comparison of participants and data gathered via Amazon’s MTurk, social media, and face-to-face behavioral testing. Computers in Human Behavior, 29, 2156–2160. https://doi.org/10.1016/j.chb.2013.05.009

10.

Chuang

Shen

C. T.

Judge

T. A.

(2016). Development of a multidimensional instrument of person–environment fit: The perceived person–environment fit scale (PPEFS). Applied Psychology, 65, 66–98. https://doi.org/10.1111/apps.12036

11.

Cronbach

L. J.

Gleser

G. C.

(1953). Assessing similarity between profiles. Psychological Bulletin, 50, 456–473. https://dx-doi-org-s.web.bisu.edu.cn/10.1037/h0057173

12.

Dahling

J. J.

Melloy

Thompson

M. N.

(2013). Financial strain and regional unemployment as barriers to job search self-efficacy: A test of social cognitive career theory. Journal of Counseling Psychology, 60, 210–218. https://doi.org/10.1037/a0031492

13.

Diener

Emmons

R. A.

Larsen

R. J.

Griffin

(1985). The satisfaction with life scale. Journal of Personality Assessment, 49, 71–75. https://doi.org/10.1207/s15327752jpa4901_13

14.

Diener

Inglehart

Tay

(2013). Theory and validity of life satisfaction scales. Social Indicators Research, 112, 497–527. https://doi.org/10.1007/s11205-012-0076-y

15.

Dik

B. J.

R. S.

Hansen

J.-I. C.

(2007). An empirical test of the Modified C Index and SII, O*NET, and DHOC occupational code classifications. Journal of Career Assessment, 15, 279–300. https://doi.org/10.1177/1069072707301203

16.

Dik

B. J.

Strife

S. R.

Hansen

J. I. C.

(2010). The flip side of Holland type congruence: Incongruence and job satisfaction. The Career Development Quarterly, 58, 352–358. https://doi.org/10.1002/j.2161-0045.2010.tb00183.x

17.

Edwards

J. R.

(1994). The study of congruence in organizational behavior research: Critique and a proposed alternative. Organizational Behavior and Human Decision Processes, 58, 51–100. https://doi.org/10.1006/obhd.1994.1029

18.

Fisher

G. A.

Heise

D. R.

Bohrnstedt

G. W.

Lucke

J. F.

(1985). Evidence for extending the circumplex model of personality trait language to self-reported moods. Journal of Personality and Social Psychology, 49, 233–242. https://doi.org/10.1037/0022-3514.49.1.233

19.

Hara

Adams

Milland

Savage

Callison-Burch

Bigham

J. P.

(2018). A data-driven analysis of workers’ earnings on Amazon Mechanical Turk. Proceedings of the 2018 CHI conference on human factors in computing systems. https://doi.org/10.1145/3173574.3174023

20.

Holland

J. L.

(1997). Making vocational choices: A theory of vocational personalities and work environments (3rd ed.). Psychological Assessment Resources.

21.

Lent

(2004). Toward a unifying theoretical and practical perspective on well-being and psychosocial adjustment. Journal of Counseling Psychology, 51, 482–509. https://doi.org/10.1037/0022-0167.51.4.482

22.

Leung

Hou

Gati

(2011). Effects of parental expectations and cultural-values orientation on career decision-making difficulties of Chinese University students. Journal of Vocational Behavior, 78, 11–20. https://doi.org/10.1016/j.jvb.2010.08.004

23.

McCloy

Campbell

Oswald

F. L.

Lewis

Rivkin

(1999). Linking client assessment profiles to O*NET occupational profiles. National Center for O*NET Development.

24.

Meyer

G. J.

Finn

S. E.

Eyde

L. D.

Kay

G. G.

Moreland

K. L.

Dies

R. R.

Eisman

E. J.

Kubiszyn

T. W

Reed

G. M.

(2001). Psychological testing and psychological assessment: A review of evidence and issues. American Psychologist, 56, 128–165. https://doi.org/10.1037/0003-066X.56.2.128

25.

Nye

C. D.

Prasad

Bradburn

Elizondo

(2018). Improving the operationalization of interest congruence using polynomial regression. Journal of Vocational Behavior, 104, 154–169. https://doi.org/10.1016/j.jvb.2017.10.012

26.

Nye

C. D.

Rounds

Drasgow

(2017). Interest congruence and performance: Revisiting recent meta-analytic findings. Journal of Vocational Behavior, 98, 138–151. https://doi.org/10.1016/j.jvb.2016.11.002

27.

Paolacci

Chandler

(2014). Inside the Turk: Understanding Mechanical Turk as a participant pool. Current Directions in Psychological Science, 23, 184–188. https://doi.org/10.1177/0963721414531598

28.

Phan

W. M. J.

Rounds

(2018). Examining the duality of Holland’s RIASEC types: Implications for measurement and congruence. Journal of Vocational Behavior, 106, 22–36. https://doi.org/10.1016/j.jvb.2017.11.011

29.

Prediger

D. J.

Vansickle

T. R.

(1992). Locating occupations on Holland’s hexagon: Beyond RIASEC. Journal of Vocational Behavior, 40, 111–128. https://doi.org/10.1016/0001-8791(92)90060-d

30.

Ramsey

S. R.

Thompson

K. L.

McKenzie

Rosenbaum

(2016). Psychological research in the internet age: The quality of web-based data. Computers in Human Behavior, 58, 354–360. https://doi.org/10.1016/j.chb.2015.12.049

31.

Rounds

Tracey

T. J.

(1996). Cross-cultural structural equivalence of RIASEC models and measures. Journal of Counseling Psychology, 43, 310. https://doi.org/10.1037/0022-0167.43.3.310

32.

Murdock

Rounds

(2015). Person-environment fit. In Savickas

M. L.

Hartung

P. J.

Walsh

W. B.

(Eds.), APA handbook of career intervention (pp. 81–98). American Psychological Association.

33.

Tinsley

(2000). The congruence myth: An analysis of the efficacy of the person–environment fit model. Journal of Vocational Behavior, 56, 147–179. https://doi.org/10.1006/jvbe.1999.1727

34.

Tracey

T. J.

(2002). Personal Globe Inventory: Measurement of the spherical model of interests and competence beliefs. Journal of Vocational Behavior, 60, 113–172. https://doi.org/10.1006/jvbe.2001.1817

35.

Tracey

T. J.

(2003). Interest traitedness as a moderator of interest–occupation congruence. Journal of Vocational Behavior, 62, 1–10. https://doi.org/10.1016/s0001-8791(02)00011-8

36.

Tracey

T. J.

(2008). Adherence to RIASEC structure as a key career decision construct. Journal of Counseling Psychology, 55, 146–157. https://doi.org/10.1037/0022-0167.55.2.146

37.

Tracey

T. J.

(2010). Development of an abbreviated Personal Globe Inventory using item response theory: The PGI-Short. Journal of Vocational Behavior, 76, 1–15. https://doi.org/10.1016/j.jvb.2009.06.007

38.

Tracey

T. J.

Allen

Robbins

S. B.

(2012). Moderation of the relation between person–environment congruence and academic success: Environmental constraint, personal flexibility and method. Journal of Vocational Behavior, 80, 38–49. https://doi.org/10.1016/j.jvb.2011.03.005

39.

Tracey

T. J.

Lent

R. W.

Brown

S. D.

Soresi

Nota

(2006). Adherence to RIASEC structure in relation to career exploration and parenting style: Longitudinal and idiothetic considerations. Journal of Vocational Behavior, 69, 248–261. https://doi.org/10.1016/j.jvb.2006.02.001

40.

Tracey

T. J.

Robbins

S. B.

(2006). The interest–major congruence and college success relation: A longitudinal study. Journal of Vocational Behavior, 69, 64–89. https://doi.org/10.1016/j.jvb.2005.11.003

41.

Tsabari

Tziner

Meir

E. I.

(2005). Updated meta-analysis on the relationship between congruence and satisfaction. Journal of Career Assessment, 13, 216–232. https://doi.org/10.1177/1069072704273165

42.

Wessel

J. L.

Ryan

A. M.

Oswald

F. L.

(2008). The relationship between objective and perceived fit with academic major, adaptability, and major-related outcomes. Journal of Vocational Behavior, 72, 363–376. https://doi.org/10.1016/j.jvb.2007.11.003

43.

Weston

Gore

J. P. A.

(2006). A brief guide to structural equation modeling. The Counseling Psychologist, 34, 719–751. https://doi.org/10.1177/0011000006286345

44.

Wilkins

K. G.

Tracey

T. J.

(2014). Person environment fit and vocational outcomes. In Coetzee

(Ed.), Psycho-social career meta-capacities (pp. 123–138). Springer.

45.

Wille

Tracey

T. J.

Feys

De Fruyt

(2014). A longitudinal and multi-method examination of interest–occupation congruence within and across time. Journal of Vocational Behavior, 84, 59–73. https://doi.org/10.1016/j.jvb.2013.12.001

46.

(2020). The ripple prediction of ambiguity aversion in career decision-making. Journal of Counseling Psychology, 67, 123–131. https://doi.org/10.1037/cou0000354

47.

Tracey

T. J. G.

(2016). Stability and change in interests: A longitudinal examination of grades 7 through college. Journal of Vocational Behavior, 93, 129–138. https://dx-doi-org-s.web.bisu.edu.cn/10.1016/j.jvb.2016.02.002

48.

Young

Tokar

D. M.

Subich

L. M.

(1998). Congruence revisited: Do 11 indices differentially predict job satisfaction and is the relation moderated by person and situation variables? Journal of Vocational Behavior, 52, 208–223. https://doi.org/10.1006/jvbe.1997.1587