Context and Validation of the Korean Subjective Underemployment Scale (K-SUS): 한국의 불완전 취업 실태 및 주관적인 불완전 취업 척도 타당화

Abstract

Underemployment is a multidimensional construct that captures various suboptimal work statuses. Although recent economic depressions and insecure job markets in Korea have increased underemployment, no appropriate scale exists to measure underemployment in a Korean context. Therefore, the aim of this study was to translate the Subjective Underemployment Scales (SUS) into Korean and validate the Korean SUS (K-SUS) with a sample of 427 Korean employees. We found that a bifactor model fit the data best, suggesting a different internal structure from the English version. Scores from the K-SUS were also invariant across gender, age, income, and employment status. In addition, we found evidence for construct validity by relating scores from the K-SUS to similar constructs and indicators of well-being. Findings from the current study help describe Korean employees’ experiences of subjective underemployment and suggest how psychologists, employers, and policy makers can address structural and psychological issues related to underemployment.

Keywords

Underemployment Scale Korea temporary work irregular work contingent work psychology of working 불완전 취업 척도 한국 비정규직 임시직 계약직 일의 심리학

In the era of radical economic, sociopolitical, and technological change, labor markets have become flexible and unstable, resulting in the increase of underemployment across the globe (International Labor Organization, 2017). In this context, vocational psychologists and other vocational and mental health professionals can help understand underemployment, particularly how it affects well-being and functioning. However, in measuring underemployment, scholars have overused single-item scales, categorical indicators, and objective variables and therefore not fully incorporated the diverse components of underemployment (e.g., Dooley et al., 2000; Friedland & Price, 2003; Maynard et al., 2006). To address this concern, scholars developed the Subjective Underemployment Scales (SUS; Allan et al., 2017). These scholars hoped to stimulate underemployment research across diverse national, social, and economic contexts. However, the SUS was developed with a sample from the United States, limiting the understanding of underemployment to only the U.S. context. In Korea, flexible labor markets, economic depressions, and underemployment are prevalent, but scholars have not developed multidimensional measures of subjective underemployment. Therefore, in the current study, we aimed to translate the SUS into Korean and validate the Korean version of the SUS (K-SUS).

Theoretical Framework

Despite the basic definition of underemployment as work below various standards of full employment (Feldman, 1996), multiple disciplines have defined underemployment differently. For example, economists, sociologists, and public health researchers have primarily based it on financial constraints and unstable work conditions such as low wages, job insecurity, and flexible work (Benach et al., 2014; Campbell, 1981; Zvonkovic, 1988). However, they do not necessarily capture psychological experiences of underemployed workers (Burris, 1983). To operationalize underemployment with a more holistic approach, Feldman (1996) proposed five domains of underemployment: (a) possessing more education or/and (b) higher skills and experience than is required (overqualification); (c) being involuntarily employed outside one’s field, training, or expertise (field underemployment); (d) being involuntarily engaged in part-time or temporary employment (involuntary temporary work and hours underemployment); and (e) earning less income than previous work or peers (underpayment). Scholars incorporated Feldman’s perspective and developed the SUS to assess multifaceted, subjective, and continuous dimensions of underemployment (Allan et al., 2017). On top of domains listed above, Allan and colleagues (2017) also included poverty wage employment, which is part of other definitions of underemployment. Taken together, the SUS included underpayment, status, hours, involuntary temporary work, field, and poverty wage employment.

Despite efforts to conceptually understand and empirically measure underemployment constructs, underemployment research has not been prolific in Korea. Most research has interchangeably used the concepts of nonstandard work, irregular work, or temporary work to understand underemployment (e.g., B. Lee, 2013; H. Lee & Lee, 2015; Shin, 2009). However, these conceptualizations have focused on working hours or temporariness, which neglects other dimensions of underemployment. Another study in Korea developed an Underemployment Scale by combining a single psychological indicator of overqualification and objective indicators of income and level of education (Roh & Kim, 2009). Despite this multifaceted measure of underemployment, it mostly measures objective underemployment, which proposes theoretical and statistical limitations. For instance, objective indicators do not capture different experiences of people with the same income or educational levels (McGuinness, 2006), and subjective measures of underemployment are more likely to predict psychological outcomes than objective ones (Maynard & Feldman, 2011; McGoldrick & Robst, 1996). In summary, no appropriate measures assessing subjective underemployment exist in Korea, making it critical to develop an appropriate scale to understand the subjective underemployment of Korean workers.

Domains of Underemployment in South Korea

As mentioned above, the SUS measures six dimensions of underemployment (hours, temporary work, status, field, underpayment, and poverty wage employment). The following section explains how the dimensions of the SUS would operate in Korean context and how underemployment is related to Korean workers’ well-being.

Hours

Hours-based underemployment refers to working fewer hours than desired due to work circumstances or contracts. The prevalence of part-time workers has increased in Korea (Organization for Economic Cooperation and Development [OECD], 2018a), and over half of contingent/temporary workers reported that they are involuntarily working part-time because they cannot find full-time work (Y. Kim, 2016; D. H. Lee & Lee, 2000). However, hours-based underemployment may be different in Korea because of the prevalence of overemployment—working more hours than desired (OECD, 2018a). This has led to new legislation attempting to reduce working hours for people in larger companies (Umeda, 2018), but this unique context may make hours-based underemployment less central to underemployment in Korea.

Temporary work

Temporary work is a dimension in which workers are involuntarily involved in time-limited work such as contract jobs (Feldman, 1996). Structural issues, such as globalization, have caused more flexible job markets and offshoring in Korea, which have in turn increased temporary work (H. Lee & Lee, 2015). In addition, policies to protect workers’ contracts have become weaker after the Asian financial crisis, leading to temporary workers to be more vulnerable (H. Lee & Lee, 2015). Moreover, the intersection between temporary work and underpayment has become more profound in Korea. Specifically, Korean temporary workers receive lower wages because of discrimination from management (Shin, 2009). This suggests that Korean subjective underemployment could be situated at the intersection of multiple underemployment domains.

Status and field

Status underemployment is when workers have a lower position at work than their previous position or compared to peers with similar knowledge, skills, and experiences, and field underemployment occurs when employees work outside of their area of education, training, or expertise (Allan et al., 2017; Feldman, 1996). Status and field underemployment seem to have worsened in Korea. For example, 21% of workers are overqualified for job requirements, meaning that skills and status mismatch exist, but overqualified job applicants tend to downgrade their occupational opportunities due to the unstable job market (OECD, 2015). Moreover, approximately 50% of Korean employees work in a different field than their training and education (Y. Lee, 2016). Furthermore, mismatched and overqualified workers are more likely to earn less and have limited contracts than well-matched workers in Korea (McGowan & Andrews, 2015; OECD, 2015). This suggests that status and field underemployed workers are also likely to have temporary contracts due to the unstable job market, showing the interconnection of underemployment experiences in Korea.

Korean workers might be overqualified and work outside of their education and training due to their traditional Confucian culture with an emphasis on education. People influenced by Confucianism are likely to be motivated to acquire higher education, and approximately 70% of people pursue a college education (Sorensen, 1994; Statistics Korea, 2018). Despite the positive influence of Confucianism culture on educational levels, it sometimes functions as oppressing system, referred to as “Hakbeol.” In the Hakbeol system, individuals are mindlessly motivated to enter prestigious colleges or companies to obtain higher social statuses and recognition and to avoid social stigma (Seth, 2002). Due to cultural systems of upward social mobility through higher education, job applicants are highly likely to be overqualified, and they tend to experience status or field underemployment given the constrained job market in Korea. Moreover, Korean people tend to conduct upward comparisons to evaluate themselves within their collectivistic culture (Jang, 2009; White & Lehman, 2005). Specifically, in the Korean cultural context, people may frequently compare themselves with others with similar qualifications but who have higher positions or a job related to their education.

Underpayment and poverty wage employment

Underpayment focuses on perceptions of people’s pay based on their skills, experiences, and knowledge, whereas poverty wage employment does not provide enough wages to meet basic needs (Allan et al., 2017). In Korea, increasing economic inequality may propagate the financial components of underemployment. For example, the top quantile of households earned 4.5 times more than the bottom quantile (Hwang & Lee, 2012; OECD, 2018b). This wage gap may exaggerate people’s inability to meet their basic needs, leading to greater perceptions of underpayment or poverty wage employment.

Underemployment and Well-Being

Several theories have proposed that underemployment frustrates people’s psychological and basic needs (e.g., psychology of working framework; Blustein, 2008, latent deprivation theory; Jahoda, 1981). Specifically, the psychology of working theory and latent deprivation theory argue that people not only obtain an income through work but also psychological benefits, including daily structure, social connection, self-determination, and meaningfulness, which leads to increased well-being (Blustein, 2008; Jahoda, 1981). These theories propose that underemployment may lead to poorer well-being by harming people’s ability to meet these needs.

Consistent with these theories, previous research supports the connection between underemployment and poorer well-being in and outside of work. For example, longitudinal and cross-sectional data showed that underemployment is associated with poorer mental health and well-being, such as depression and psychological distress (Cassidy & Wright, 2008; Dooley et al., 2000), as well as job-related well-being constructs, including meaningful work and job satisfaction (Allan et al., 2017; Friedland & Price, 2003; T. Kim & Allan, 2019). Because underemployment is relatively new concept in Korea, there are few studies exploring the relations among underemployment and work-related and general well-being. Rather, scholars have mostly focused on the financial or political consequences of underemployment (e.g., I. Kim et al., 2019; Shin, 2009). However, scholars have recently begun to study psychological consequences by revealing that temporary workers are likely to experience greater depression and poorer life and job satisfaction (e.g., W. Kim et al., 2016; B. Lee, 2013). This suggests that a more psychological approach to examining the relations among underemployment and well-being indicators is needed in Korea.

The Present Study

The primary goal of this study was to translate and validate the Korean version of the SUS and explore its relations to other relevant variables in a Korean context. We used confirmatory factor analysis (CFA) to examine the factor structure and aimed to demonstrate measurement invariance of the K-SUS across gender, age, income, and employment status. Moreover, we evaluated construct validity to identify how the K-SUS measures the construct as intended by relating scores from the K-SUS to other work-related constructs. For the construct validity, we included perceived overqualification, job insecurity, pay satisfaction, and person–job (P-J) fit, which are similar but different constructs than subjective underemployment. Specifically, we included perceived overqualification and P-J fit to understand status and field underemployment. We used job insecurity to explore involuntary temporary work and hours underemployment. We also included pay satisfaction to identify the relations among pay satisfaction, underpay, and poverty wage employment. Moreover, previous studies have found that underemployment predicts poorer well-being. Thus, we included indicators of well-being: job satisfaction, meaningful work, recognition, and life satisfaction for predictive validity to explore the relations between K-SUS and relevant measures available at a future time. We especially included recognition because it is a component of meaningfulness for Korean workers (Tak et al., 2015; Tak et al., 2017).

Method

Participants

The sample consisted of 427 Korean employees. Regarding gender, 65.6% (n = 280) of the sample self-identified as women, 33% (n = 141) as men, and 1.4% (n = 6) did not provide a gender. In total, 57.6% of participants were employed full-time by employer (n = 246), 19.7% were temporary employees (n = 84), 15% were part-time employees (n = 64), 3.3% were employed part-time self-employed (n = 14), 2.1% were employed full-time self-employed (n = 9), and 2.3% had other statuses such as outsourced and unlimited-term contract workers (n = 10). Regarding highest degree of education achieved, 63.9% (n = 273) had a bachelor’s degree, 17.8% (n = 76) had a professional degree, 10.5% (n = 45) had a trade/vocational school diploma, and 7.7% (n = 33) had some high school or less. Participants reported various job titles such as company staff, teacher, nurse, part-time worker, and civil servant. Participants reported their subjective social class as lower class (2.1%, n = 9), working class (29.8%, n = 127), middle class (40%, n = 171), upper-middle class (25.8%, n = 110), and upper class (2.3%, n = 10).

Instruments

Subjective underemployment

Scholars developed the SUS to assess components of subjective underemployment (Allan et al., 2017). The SUS consists of six subscales: underpayment, status discrepancy, hours discrepancy, involuntary temporary work, field, and poverty wage employment. The SUS contains 37 items with a 7-point Likert-type scale ranging from 1 = strongly disagree to 7 = strongly agree. Higher scores for the total scale indicate higher subjective underemployment. The estimated internal consistencies for subscales were underpayment (α = .97), status discrepancy (α = .96), hours discrepancy (α = .95), field (α = .95), involuntary temporary work (α = .97), and poverty wage (α = .96) in the scale development study. Allan et al. (2017) conducted CFA and found the correlational model to best explain the data. They also found that scores to correlate negatively with work-related well-being indicators, such as job satisfaction, work satisfaction, coworker satisfaction, pay satisfaction, supervision satisfaction, and meaningful work, when examining the incremental validity.

We translated the SUS based on established translation procedures (e.g., Ægisdóttir et al., 2008). The first author and an expert reviewer, both bilingual and bicultural from being educated in the United States and Korea, independently translated the SUS to minimize bias and misunderstanding. They then compared their translations and reached a consensus. A third expert reviewer, a bilingual Korean counseling psychologist, then back-translated the Korean items into English. After reaching linguistic equivalence, three additional counseling psychologists responded to the original and translated scales, respectively, and concluded that the K-SUS items appropriately reflected subjective underemployment. They also rated the degree to which items reflected the meaning of the original SUS and were culturally appropriate on a 5-point scale, and all items received average scores greater than three. Two Korean speakers outside of psychology read all items and confirmed that the items were easy to understand. In this study, estimated internal consistencies for the six subscales and total scale were pay (α = .96), status (α = .91), field (α = .92), hours (α = .92), involuntary temporary work (α = .92), poverty wage employment (α = .94), and the total score (α = .95). The K-SUS is presented in Appendix C in the article or Online Appendix B.

Perceived overqualification

We measured perceived overqualification with the Korean Perceived Overqualification Scale (K-POQ; Roh & Kim, 2009), originally developed by Johnsone et al. (2002). The original POQ consists of 10 items answered on a 5-point scale ranging from 1 = strongly disagree to 5 = strongly agree. Roh and Kim (2009) excluded each item of two dimensions due to their low internal consistencies in their validation process. Example items include “My formal education overqualifies me for my present job” and “Some continuing education related to my job would improve my job performance.” Roh and Kim (2009) found scale scores to have good internal consistency (e.g., α = .81) and to negatively correlate in the expected directions with the ratio of required education to the actual education and the occurrence of change from a permanent job to a temporary job. In the current study, the estimated internal consistency was .79 for the overall scale.

Korean pay satisfaction

We assessed pay satisfaction with the Korean Pay Satisfaction Questionnaire (K-PSQ; S. Kim et al., 2008) that validated the original PSQ (Heneman & Schwab, 1985). Participants responded on a 7-point Likert-type scale with the anchors 1 = very dissatisfied and 5 = very satisfied. The scale consists of three factors, including pay level satisfaction, benefits satisfaction, and pay structure/administration satisfaction. Example items include “Are you satisfied with your current pay?” “Are you satisfied with your benefit package?” and “Are you satisfied with the company’s pay structure?” In the validation study, reliability coefficients were .94 (pay level satisfaction), .91 (benefits satisfaction), and .84 (pay-structure/administration satisfaction). S. Kim et al. (2008) found the scale to negatively correlate with self-efficacy and positively correlate with pay-for-performance perceptions. Our study used the total score for this measure, and the estimated internal consistency was .97.

Meaningful work

We used the 10-item Korean Work as Meaning Inventory (K-WAMI; Choi & Lee, 2017) to measure the degree to which participants regarded their work as meaningful. Choi and Lee (2017) translated and validated the original WAMI (Steger et al., 2012). Participants responded on a 7-point Likert-type scale ranging from 1 = strongly disagree to 7 = strongly agree. Sample items for the WAMI include “I have found a meaningful career” and “The work I do serves a greater purpose.” Choi and Lee (2017) found scores from the K-WAMI to positively correlate with calling, life satisfaction, and life meaning and found scale scores to have good internal consistencies (e.g., α = .91). In present study, the estimated internal consistency was α = .94.

Recognition

We adopted the 5-item Recognition subscale in the Work Meaning Inventory, developed by Korean researchers (Tak et al., 2015), because recognition is a domain that Korean people experience as meaningful (Tak et al., 2015; Tak et al., 2017). Participants responded on a 7-point Likert-type scale ranging from 1 = not very important to 7 = very important. Examples of the Recognition are “It is important to get recognition at work” and “It is important to be acknowledged as competent at work.” In the initial study, the scale correlated in expected directions with meaning in life, life satisfaction, psychological well-being, and mental health. The estimated internal consistency of the original recognition was .88 (Tak et al., 2015), and we found the estimated internal consistency of α = .90 in the present study.

P-J fit

The 3-item P-J fit (Tak, 2011) was developed in Korea, and we used it to assess the Korean P-J fit. Participants responded on a 5-point Likert-type scale ranging from 1 = strongly disagree to 5 = strongly agree. A sample item includes “My job fits my interests.” Tak (2011) found higher P-J fit in permanent employees and to negatively correlate with turnover intention in Korean context. Tak (2011) found an estimated internal consistency of α = .84, and the estimated internal consistency for the present study was α = .89.

Job insecurity

We used the 6-item Job Insecurity subscale in the Occupational Stress Scale for Korean (KOSS) employees (Chang et al., 2005). Developers of the KOSS conducted qualitative research first to incorporate Korean working environments (Chang et al., 2005). Participants responded on a 4-point Likert-type scale with the anchors 1 = very disagree and 4 = very agree. Sample items for the job insecurity include “My future is uncertain because the current situation of my company is unstable” and “It is possible to lose my job within 2 years.” In the scale development study, the scale scores correlated in expected directions with poor organizational system, job stress, and work demand. The estimated internal consistency of the job insecurity in the scale development study was α = .61 (Chang et al., 2005), whereas the estimated internal consistency was α = .71 in the present study.

Job satisfaction

We used the 5-item Hedonic Job Satisfaction subscale of the Korean Global Hedonic and Eudaimonic Job Satisfaction (K-GHEJS; Song et al., 2018), translated from the GHEJS (Rothausen, 2014). Participants responded on a 7-point Likert-type scale with the anchors 1 = very disagree and 7 = very agree. Sample items include “I am happy in my job” and “I enjoy my job.” In the validation study, scholars translated items and confirmed two factors of hedonic and eudaimonic job satisfaction and found the scale to correlate in the expected directions with happiness and eudaimonic well-being (Song et al., 2018). The estimated internal consistency in the validation study was α = .94, and the estimated internal consistency in the present study was α = .96.

Life satisfaction

We adopted the 5-item Korean version of the Satisfaction With Life Scale (K-SWLS; Lim, 2012), translated from the SWLS (Diener et al., 1985). Participants were asked to respond on a 7-point Likert-type scale ranging from 1 = strongly disagree and 7 = strongly agree. Example items include “In most ways my life is close to my ideal” and “I am satisfied with my life.” In the validation study, scholars translated items and found the scale correlated in expected directions with emotional, social, and psychological well-being. Cronbach’s α were .91 in police officers, .84 in college students, and .86 in high school students in Korea. In the present study, the estimated internal consistency was α = .86.

Procedure

We created a survey link and posted it on job-related forums on Korean social networking sites, including Facebook, Daum, and Naver. Participants had to (a) be over the age of 18 years, (b) work at least part-time, (c) not be a full-time student, and (d) reside in South Korea. The posting included an introduction of the researchers and the present study, the participant inclusion criteria, and the estimated time to complete the survey. The survey link allowed participants to access an informed consent form and the instruments. Recent research has concluded that sampling through social networking sites results in valid data (e.g., Alshaikh et al., 2014). For participating, we randomly entered participants into a draw for 1 of the 25 ₩5,000 coffee shop gift cards. For the first part, participants completed the demographic questionnaires and the translated SUS. We selected approximately half the sample (n = 328) to complete the validity scales; however, the whole sample completed the K-SUS for CFA. There were no differences between the whole sample and the sample that completed the validity scales on any demographic variables or average K-SUS scores based on t test results. In total, 819 participants accessed the survey. However, 1 participant disagreed with research participation, 195 participants did not provide any data, 41 participants only provided demographic data, and 155 participants did not respond correctly to three attention check items. We removed all these cases, which resulted in the final sample size of 427.

Analysis Plan

We conducted CFA to examine how the original 37-item, six-factor SUS structure fit with Korean working adults. We evaluated correlational, single-factor, higher order, and bifactor models in accordance with the original SUS development study. To evaluate these models, we used Mplus Version 7.4 with robust maximum likelihood estimation, which effectively addresses issues of nonnormality (Muthén & Muthén, 2012). The correlational model had all factors correlate with one another. The single factor had all indicators load on a single subjective underemployment factor. The higher order model had a higher order subjective underemployment factor that all subfactors loaded onto and all indicators load on their factors. Finally, the bifactor model had all indictors load simultaneously onto their subfactors and a general subjective underemployment factor.

We evaluated the CFA model–data fit based on conventional model fit statistics, such as change in χ², the comparative fit index (CFI), and the combination of two-index criteria, including the root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR; Hu & Bentler, 1999). Specifically, we used two-index criteria for acceptable fit, including the combination of SRMR < .09 and RMSEA < .06 with a maximum upper bound of the 90% confidence interval (CI) of .10 or the combination of CFI > .96 and SRMR < .09 (Browne & Cudek, 1993; Hu & Bentler, 1999). To compare nonnested models, we used the Akaike information criterion (AIC), which considers both goodness-of-fit and parsimony. Regarding the AIC, lower values are better, but there are no absolute cutoff values (Hooper et al., 2008). If a bifactor model shows the best fit to the data, researchers can calculate bifactor indices to explore how much variance can be accounted for the subscales and the general factor, respectively (Dueber, 2017; Rodriguez et al., 2016).

Second, we performed invariance tests to explore whether the instrument demonstrated factor invariance across differ groups. Before testing measurement invariance, we created two categories per group for gender, age, income, and employment status. For gender, we compared men versus women, and for employment status, we compared full-time employment versus part-time employment. For age and income, we split the variables at the mean to create two categorical groups. To test measurement invariance, we constrained parameters across groups in configural, metric, and scalar models and observed whether fit declined. The configural model examines whether different groups share the same model structure (Meredith, 1993; Riordan & Vandenberg, 1994). The metric model constrains the configuration of variables and all factor loadings to be the same across groups. Finally, the scalar model constrains the configuration, factor loadings, and the indicator intercepts for each group to be the same across groups (Cheung & Rensvold, 2002). Using a χ² difference test is a common way to evaluate statistical differences among the models. However, χ² is overly sensitive with large samples sizes, so scholars recommend using changes in CFI > .010 and RMSEA > .015 to evaluate the practical significance of the change in model fit (Cheung & Rensvold, 2002).

Results

Preliminary Analyses

We evaluated the statistical assumptions for structural equation modeling. All study variables had appropriate levels of univariate normality with skewness ranging from −.83 to .85 and kurtosis ranging from −1.24 to 1.20 (Weston & Gore, 2006). Visually inspected histograms and boxplots appeared normally distributed. To assess outliers, we examined cases that fell outside the distance between the third and first quartile multiplied by 2.2 (Hoaglin & Iglewicz, 1987), and we found no variables over the multiplied range. We also found no variables with extreme z-scores (Shiffler, 1988). All variables had significant linear relations with each other, and visually inspected scatterplots did not provide evidence for nonlinear relations.

Regarding missing data, 34.9% of participants’ responses on underpayment Item 7 were missed due to an item initially missing on the survey. We found that missing data on Item 7 were not related to any study or demographic variable, concluding that the data were missing completely at random. Except for this issue, 2.8% (n = 12) of the sample that only completed the K-SUS and 4.2% (n = 18) of the sample that completed the all scales had missing data on at least one of study variables. To assess whether there was a pattern of missing data, we created a dummy code variable representing those with and without missing data and correlated it with each study and demographic variable. The dummy code was not related to any study or demographic variable, and we found data missing completely at random, justifying our use of full information maximum likelihood (Schlomer et al., 2010). All variance inflation factors for each variable were between one and three, indicating that there was no evidence of multicollinearity issues (Menard, 1995).

Factor Structures

CFA

We conducted CFA to evaluate single, higher order, correlational, and bifactor models. The single-factor model had poor fit to the data, χ²(629) = 7,255.884, p < .001, CFI = .39, AIC = 55,379.508, RMSEA = .16, 90% CI [.15, .16], and SRMR = .17. The higher order model also had poor fit to the data, χ²(623) = 1,849.265, p < .001, CFI = .89, AIC = 48,462.686, RMSEA = .07, 90% CI [.06, .07], and SRMR = .09. All items loaded on factors, ranging from .63 to .93, p < .001. All subfactors loaded on the higher factor, ranging from .33 to .91, p < .001. The correlational model had poor fit to the data, χ²(614) = 1,796.836, p < .001, CFI = .89, AIC = 48,419.733, RMSEA = .07, 90% CI [.06, .07], and SRMR = .07. All items loaded on factors, ranging from .63 to .93. Finally, the bifactor model had acceptable fit to the data, χ²(592) = 1,570.172, p < .001, CFI = .91, AIC = 48,138.800, RMSEA = .06, 90% CI [.06, .07], and SRMR = .07. Therefore, we retained this model as the best-fitting model (see Online Appendix A). All indicators loaded on their factors individually, and their factor loadings included hours (.70–.84), underpayment (.44–.65), field (.60–.87), status (.28–.75), temporary work (.60–.82), and poverty wage employment (.14–.56). In addition, all indicators significantly loaded on the general factor: hours (.12–.34), underpay (.66–.79), field (.27–.42), status (.38–.68), temporary work (.37–.44), and poverty wage (.69–.85).

Next, due to the bifactor model having the best fit, we calculated bifactor indices to investigate internal structural reliabilities, such as omega (ω), omega hierarchical (ωH), and explained common variance (ECV). The ω for the total score including the general factors and the subscale factors was .98, meaning that the factors explained 98% of the variance in the total score and 2% was attributable to error. The ωs for the subscales were underpayment (ω = .96), field (ω = .93), temporary work (ω = .93), hours (ω = .92), status (ω = .92), and poverty wage employment (ω = .95). We also examined ωH estimates to identify how much the variance is attributable to the general factor. Statistically, the procedure regards the subjective underemployment subscales as error. ωH for the subjective underemployment total was .77. Comparing ω (.98) and ωH (.77), we can see that 79% of the reliable variance in the subjective underemployment total score (.77/.98 = .79) is attributable to the general factor and 21% is attributable to the subscale factors. The ωHs for the subscales were underpay (ωH = .36), field (ωH = .77), temporary work (ωH = .70), hours (ωH = .86), status (ωH = .52), and poverty wage (ωH = .20). Additionally, we examined ECV to investigate how much of the common variance is attributable to the general factor versus the six factors. The ECV was .41, meaning that 41% of the common variance is attributable to the general factor and 59% is attributable to the six subscales. The ECVs for subfactors included underpay (.08), field (.13), temporary work (.12), hours (.14), status (.09), and poverty wage (.04), meaning that poverty wage, underpay, and status are more reflected by the general factor. Then, we identified item explained common variance (IECV) to assess unidimensionality at the individual item level. The average value was .40, ranging from .02 to .98. Items with IECV loadings >.85 (Stucky & Edelen, 2015) were Item Status 1, Item Poverty Wage 4, and Poverty Wage 5. The average relative measurement parameter bias across the items was <.001. Finally, we examined the percentage of uncontaminated correlations (PUC), the percentage of covariances that only reflects variance from the general dimension. The PUC for the current model was .86.

Invariance Models

Gender

We performed invariance tests on the bifactor model because it had the best fit to the data. The configural model for gender had borderline acceptable fit, χ²(1,184) = 2,782.89, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.08, .08], and SRMR = .08, and the fit slightly declined for the metric model, χ²(1,251) = 2,889.34, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.08, .08], and SRMR = .08. Although the models had a significantly different χ², Δχ²(67) = 106.45, p < .01, the change in CFI was not substantial, ΔCFI < .01. Finally, the fit also slightly declined for the scalar model, χ²(1,281) = 2,934.15, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.08, .08], and SRMR = .08. Despite a significant difference in χ², Δχ²(30) = 44.80, p < .05, the change in CFI was not substantial, ΔCFI < .01. Therefore, we determined that factor structure and indicator intercepts were invariant across the two gender groups.

Age

The configural model for age had a borderline acceptable fit, χ²(1,184) = 2,748.30, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.08, .08], and SRMR = .08. The fit slightly declined for the metric model, χ²(1,251) = 2,857.83, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.08, .08], and SRMR = .08. The models had a significantly different χ², Δχ²(67) = 109.54, p < .001; however, the change in CFI was not substantial, ΔCFI < .01. Next, the fit slightly declined for the scalar model, χ²(1,281) = 2,914.38, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.07, .08], and SRMR = .08. Although the metric and scalar models had a significantly different χ², Δχ²(30) = 56.54, p < .01, the change in CFI was not substantial, ΔCFI < .01. Thus, we concluded that factor structure and indicator intercepts were invariant across the two age groups.

Income

The configural model for income had an acceptable fit, χ²(1,184) = 2,565.86, p < .001, CFI = .90, TLI = .89, RMSEA = .08, 90% CI [.07, .08], and SRMR = .07. The fit slightly declined for the metric model, χ²(1,251) = 2,688.64, p < .001, CFI = .90, TLI = .89, RMSEA = .08, 90% CI [.07, .08], and SRMR = .09. The models had a significantly different χ², Δχ²(67) = 122.77, p < .001; however, the change in CFI was not substantial, ΔCFI < .01. Fit was also similar for the scalar model, χ²(1,281) = 2,711.79, p < .001, CFI = .89, TLI = .89, RMSEA = .08, 90% CI [.07, .08], and SRMR = .09, and the scalar model was not significantly different from the metric model, χ²(30) = 23.15, p = .81, CFI < 0.01. Therefore, we concluded that factor structure and indicator intercepts were invariant across the two income groups.

Employment status

The configural model for employment status had an acceptable fit, χ²(1,184) = 2,605.12, p < .001, CFI = .90, TLI = .89, RMSEA = .08, 90% CI [.07, .08], and SRMR = .07. The fit slightly declined for the metric model, χ²(1,251) = 2,755.58, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.07, .08], and SRMR = .09. The models had a significantly different χ², Δχ²(67) = 150.46, p < .001; however, the change in CFI was not substantial, ΔCFI < .01. Finally, the fit slightly declined for the scalar model, χ²(1,281) = 2,859.04, p < .001, CFI = .89, TLI = .88, RMSEA = .08, 90% CI [.07, .08], and SRMR = .09. The models had a significantly different χ², Δχ²(30) = 103.46, p < .001; however, the change in CFI was not substantial, ΔCFI < .01. Therefore, we determined that factor structure and indicator intercepts were invariant across the full-time and part-time groups.

Validity Estimates

As the original SUS development study investigated, perceived overqualification significantly correlated with all underemployment subscales (underpayment: .35, status: .32, hours: .17, temporary work: .17, field: .37, poverty wage: .43, and total score: .37). Additionally, underpayment related to pay satisfaction (−.63) and job insecurity (.37), and poverty wage employment related to pay satisfaction (−.68) and job insecurity (.41). Status related to pay satisfaction (.42), perceived overqualification (.32), and job insecurity (.17), respectively. Temporary work was related to P-J fit (−.14) and job insecurity (.36), and field was associated with pay satisfaction (−.19), P-J fit (−.43), and job insecurity (.33).

Life satisfaction showed negative associations with all underemployment subscales except for hour-based underemployment (underpayment: −.29, status: −.20, hours: −.06, temporary work: −.32, field: −.48, poverty wage: −.47, and total score: −.49). As for career-related well-being constructs, meaningful work and job satisfaction had negative correlations with field (−.43 and −.47), poverty (−.36 and −.30), and the total subjective underemployment (−.37 and −.23), respectively. Recognition positively related to underpayment (.27) and status (.26) but was not related to other subscales. Table 1 depicts the remaining correlations between the underemployment subscales and the study variables.

Table 1.
Correlations and Descriptive Statistics for the Underemployment Subscales.

Underemployment Subscales Pay Satisfaction P-J Fit POQ Job Insecurity Meaningful Work Recognition Life Satisfaction Job Satisfaction

Underpay −0.63* −0.12 0.35 0.37 −0.19 0.27* −0.29 −0.14

Status −0.42 −0.1 0.32 0.17 −0.06 0.26 −0.20 0.02

Hours −0.08 −0.13* 0.17 0.12 −0.07 −0.05 −0.06 0.04

Temporary work −0.32 −0.14* 0.17* 0.36 −0.12 0.03 −0.32 −0.11

Field −0.19 −0.43 0.37 0.33 −0.43 −0.11 −0.48 −0.47

Poverty −0.68 −0.31 0.43 0.41 −0.36 0.06 −0.47 −0.30

Total −0.65 −0.33 0.37 0.42 −0.37 0.13 −0.49 −0.23*

M 59.86 10.17 20.64 14.10 46.89 25.84 19.89 22.13

SD 21.47 2.89 5.55 3.52 12.22 5.34 5.99 7.04

Note. P-J fit = person–job fit; POQ = perceived overqualification.

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

Discussion

The purpose of the present study was to translate the SUS and determine whether the K-SUS was reliable and valid in Korean context. In contrast to the SUS development study, the current study revealed that the bifactor was the best fit to the data. Additionally, invariance models suggested that the K-SUS’s internal structure was applicable across different gender, age, income, and employment status. Finally, given that scores from the K-SUS correlated in the expected direction with a number of related constructs and well-being indicators, results suggest that the K-SUS is a valid measure of subjective underemployment in a Korean context.

Factor Structure

Our data showed the bifactor model was the best fit to the data, but the fit indices were borderline acceptable. This may have been because the lower factor loadings of hours-based underemployment (.12–.34) harmed the general model fit, but we expected this result due to the mix of underemployment and overemployment in Korea. For instance, there are involuntary part-time workers and workers who overwork, consistent with previous studies and current governmental effort to control the tendency. Specifically, annual working hours in Korea were ranked third internationally, and the Korean government provides compensation for mental distress related to long working hours (W. Kim et al., 2016; K.-H. Lee et al., 2013). Notably, items of the hours-based underemployment loaded onto the subfactor well (.70–.84). This suggests that although our items explain hours-based underemployment well, the unique combination of underemployment and overemployment in Korea may have led to lower model fit.

The finding that the bifactor model was the best fit to the data diverges from the result of the original scale development study, which supported the correlational model (Allan et al., 2017). Scholars suggest that a general factor ECV < .70 with a PUC < .80 suggest a multidimensional model (Quinn, 2014; Rodriguez et al., 2016). Our data show mixed results. Specifically, the general factor had a high PUC, a high ωH, and a low average relative measurement parameter bias. These suggest unidimensional structure, whereas the ωs of subfactors and low ECV suggest a multidimensional structure. However, ECV becomes less important when PUC is higher than .80 (Rodriguez et al., 2016), suggesting that K-SUS is fairly unidimensional, but the subscales retain their individual meanings. Therefore, future scholars could use the total scale score as a unidimensional model but also use subfactors to understand unique constructs. If they want to use the general total scale, they could (a) use the scale within a latent framework, (b) use the general factor by loading subscales onto the general factor, or (c) calculate the mean of the total K-SUS and subtracting it from each subscale mean score (i.e., ipsative scoring).

To understand the general factor, underpayment and poverty wage employment’s factor loadings strongly loaded onto the general factor. In addition, ECVs for subfactors and IECVs propose that the general factor likely captures poverty wage, underpay, and status underemployment. It suggests that Korean underemployment may primarily reflect financial hardship and the following psychological comparison about their status. This is in line with other findings that a substantial amount of people are in working poverty or receive low pay in Korea (Hwang & Lee, 2012; A. Kim & Lee, 2014). In addition, the high factor loadings of underpayment and poverty wage employment propose that Korean underemployed workers may be located at the intersection between financial-related underemployment and other types of underemployment. For example, temporary workers have lower wages compared to in the past, and their wages are lower than those with permanent contracts, which may lead people to compare with those with permanent contracts (H. Lee & Lee, 2015). In short, the general underemployment factor may reflect the current economic context and related psychological comparisons in Korea.

Invariance Tests

We conducted measurement invariance test across gender, age, income, and employment status. We did not find any significant changes in fit across groups, indicating that participants tend to respond to the scale indicators in a similar way regardless of their groups, and these responses form equally valid structures. Despite the scale measurement invariance, creating two arbitrary groups may obscure interpretation. Specifically, splitting the group into the full-time and the part-time work could remove important information. For example, part-time workers’ experiences vary depending on their intention of work as either voluntary or involuntary (Allan et al., 2019). In addition, dividing income and age at the mean may lose information that exists in continuous variables. Therefore, future studies should test measurement invariance with larger samples sizes and among more specific groupings of employment, income, and age.

Construct Validity

The facets of underemployment mostly showed moderate to strong correlation with validity variables as expected. Consistent with the results of the original SUS development study, underpayment and poverty wage employment had large negative correlations with pay satisfaction, and poverty wage employment had large correlations with pay satisfaction. These consistent results suggest that poverty wage employment and pay dissatisfaction can be located under the similar financial constructs. All subfactors also had weak to moderate correlations with perceived overqualification, which corroborates the finding of the original SUS study that perceived overqualification could be an overarching variable subjective underemployment. All dimensions except for the hours also had moderate to large negative correlations with pay satisfaction, indicating that Korean subjective underemployment generally captures financial hardships. Job insecurity had moderate positive correlations with all subfactors except for hours, proposing that job insecurity could be a relevant construct in Korean context. P-J fit showed weak to moderate negative correlations with all dimensions except for underpayment and hours, and the correlation was the highest with field. It indicates that the field mismatch in field may reflect the poorer P-J fit. In short, evidence supports the construct validity of the K-SUS.

Links among the underemployment total score and meaningful work, job satisfaction, and life satisfaction corroborate previous studies result that underemployment is associated with general well-being (e.g., Allan et al., 2017). This also corroborates previous findings that irregular workers are unlikely to experience meaningfulness at work (Tak et al., 2015). Specifically, field underemployment and poverty wage employment were negatively associated with career-related well-being constructs such as job satisfaction and meaningfulness at work. This is consistent with previous research supporting that field underemployment negatively relates to positive job outcomes (Allan et al., 2017; Feldman & Turnley, 1995). The negative relation between poverty wage employment and both job satisfaction and meaningfulness at work is in line with previous research, showing that Korean workers with low income are more likely to experience lower job satisfaction (B. Lee, 2013). Surprisingly, recognition, a subfactor of meaningfulness at work in Korea, was positively associated with underpayment and status subfactors. Pay and status underemployed workers may alternatively want to seek societal reward at work because their employment statuses could easily frustrate their desires such as unsatisfactory pay and positions. They may also want to be acknowledged at work because getting recognition at work is in process of fulfilling social connection needs (Tak et al., 2017). However, even though they pursue social acknowledgment, they are less likely to feel connected and more located in vulnerable social statuses (T. Kim & Allan, 2019).

Contrary to career-related well-being indicators, all subfactors had negative correlations with life satisfaction, showing that underemployed workers are likely to experience lower well-being. This finding is in line with previous studies showing the relation between psychological distress and diverse dimensions of underemployment such as precarious employment and temporary work (W. Kim et al., 2016; Park, 2017).

Implications for Practice

Findings from the present study are relevant to counselors in practice. First, findings highlight the domains of underemployment and how they relate to well-being. Counselors can assess the extent to which their clients experience underemployment by using the general total score and may target specific dimensions to intervene by considering specific subscales’ scores. For example, clearly underemployment is relevant to clients’ well-being and functioning, so assessing whether clients are in working poverty during intakes is critical for appropriately developing case conceptualizations and interventions. Second, underemployed workers may internalize unsuccessful attempts at obtaining better work or stigma related to holding undesirable work statues, leading to feelings of demoralization or helplessness. Helping clients understand the systematic causes of their employment situation by reflecting on each subscale may reduce self-blame and help them move toward greater agency and empowerment.

The K-SUS may be also useful for psychologists to promote social advocacy. For example, our findings revealed that most domains of underemployment were negatively related to life satisfaction. This suggests that diverse forms of current labor markets, such as outsourcing and temporary contract works, might be psychologically threatening to employees. Psychologists can advocate to increase social safety nets for underemployed workers and restructure employment-related policies. Policy makers in Korea attempted to address well-being of underemployed workers by addressing hours underemployment and by targeting large companies that provide higher wages (W. Kim et al., 2016; K.-H. Lee et al., 2013; Umeda, 2018). However, our results suggest that financial-related and field underemployment are negatively associated with work-related well-being. Therefore, psychologists and other vocational professionals may guide government planners to more focus on financial-related underemployment and pay attention to smaller companies with lower wages.

Limitations and Future Directions

First, we used CFA by considering theories, previous findings, and Korean cultural backgrounds. However, substantial amount of missing data and lower factor loadings of hours-based underemployment seemed to contribute to the marginal model fit. Future studies may collect more data to more accurately identify the model fit of the scale. If the model fit is consistently questionable, scholars may want to modify items to increase model fit and explore whether modification relates to other components of underemployment and other variables.

Second, the original SUS tested the constructs of underemployment by adding perceived overqualification as an extra dimension of underemployment. Our study also found a positive correlation between subjective underemployment and perceived overqualification. However, we did not include perceived overqualification as another dimension, which does not measure underemployment comprehensively. Third, the current study aimed to understand how underemployed individuals interact with their contextual factors such as unstable economic structures and P-J fit. However, we did not identify the relations between subjective underemployment and individuals’ traits. For example, other research has established a link between personality traits and perceived overqualification tendency (e.g., Liu & Wang, 2012), suggesting individual dispositions may affect the experience subjective underemployment. Therefore, future studies can incorporate personality variables to extend the underemployment literature. Finally, the K-SUS likely captures financial-related hardships, so future studies should investigate the moderating impact of abundant financial resources on the relation between subjective underemployment and mental health.

Supplemental Material

Supplemental Material, Appendix_A - Context and Validation of the Korean Subjective Underemployment Scale (K-SUS): 한국의 불완전 취업 실태 및 주관적인 불완전 취업 척도 타당화

Supplemental Material, Appendix_A for Context and Validation of the Korean Subjective Underemployment Scale (K-SUS): 한국의 불완전 취업 실태 및 주관적인 불완전 취업 척도 타당화 by Taewon Kim and Blake A. Allan in Journal of Career Assessment

Supplemental Material

Supplemental Material, Appendix_B - Context and Validation of the Korean Subjective Underemployment Scale (K-SUS): 한국의 불완전 취업 실태 및 주관적인 불완전 취업 척도 타당화

Supplemental Material, Appendix_B for Context and Validation of the Korean Subjective Underemployment Scale (K-SUS): 한국의 불완전 취업 실태 및 주관적인 불완전 취업 척도 타당화 by Taewon Kim and Blake A. Allan in Journal of Career Assessment

Underemployment Subscales	Pay Satisfaction	P-J Fit	POQ	Job Insecurity	Meaningful Work	Recognition	Life Satisfaction	Job Satisfaction
Underpay	−0.63*	−0.12	0.35**	0.37**	−0.19	0.27*	−0.29**	−0.14
Status	−0.42**	−0.1	0.32**	0.17**	−0.06	0.26**	−0.20**	0.02
Hours	−0.08	−0.13*	0.17**	0.12	−0.07	−0.05	−0.06	0.04
Temporary work	−0.32**	−0.14*	0.17*	0.36**	−0.12	0.03	−0.32**	−0.11
Field	−0.19**	−0.43**	0.37**	0.33**	−0.43**	−0.11	−0.48**	−0.47**
Poverty	−0.68**	−0.31**	0.43**	0.41**	−0.36**	0.06	−0.47**	−0.30**
Total	−0.65**	−0.33**	0.37**	0.42**	−0.37**	0.13	−0.49**	−0.23*
M	59.86	10.17	20.64	14.10	46.89	25.84	19.89	22.13
SD	21.47	2.89	5.55	3.52	12.22	5.34	5.99	7.04

Footnotes

Appendix C

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

Taewon Kim

Supplemental Material

Supplemental material for this article is available online.

References

Ægisdóttir

Gerstein

L. H.

Çinarbas

D. C.

(2008). Methodological issues in cross-cultural counseling research: Equivalence, bias, and translations. The Counseling Psychologist, 36, 188–219. https://doi.org/10.1177/0011000007305384

Allan

B. A.

Kim

Liu

T. Y.

Deemer

E. D.

(2019). Moderators of involuntary part-time work and life satisfaction: A latent deprivation approach. Professional Psychology: Research and Practice, 51, 257. https://doi.org/10.1037/pro0000268

Allan

B. A.

Tay

Sterling

H. M.

(2017). Construction and validation of the Subjective Underemployment Scales (SUS). Journal of Vocational Behavior, 99, 93–106. https://doi.org/10.1016/j.jvb.2017.01.001

Alshaikh

Ramzan

Rawaf

Majeed

(2014). Social network sites as a mode to collect health data: A systematic review. Journal of Medical Internet Research, 16(7), 136–143. https://doi.org/10.2196/jmir.3050

Benach

Vives

Amable

Vanroelen

Tarafa

Muntaner

(2014). Precarious employment: Understanding an emerging social determinant of health.Annual Review of Public Health,35229–35253. https://doi.org/10.1146/annurev-publhealth-032013-182500

Blustein

D. L.

(2008). The role of work in psychological health and well-being: A conceptual, historical, and public policy perspective. American Psychologist, 63(4), 228–240. https://doi.org/10.1037/0003-066X.63.4.228

Browne

M. W.

Cudek

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136–162). Sage

Burris

B. H.

(1983). The human effects of underemployment. Social Problems, 31(1), 96–110. https://doi.org/10.1525/sp.1983.31.1.03a00070

Campbell

(1981). The sense of well-being in America: Recent patterns and trends. McGraw-Hill.

10.

Cassidy

Wright

(2008). Graduate employment status and health: A longitudinal analysis of the transition from student. Social Psychology of Education, 11(2), 181–191. https://doi.org/10.1007/s11218-007-9043-x

11.

Chang

S. J.

Koh

S. B.

Kang

Kim

S. A.

Kang

M. G.

Lee

C. G.

Chung

J. J.

Cho

J. J.

Son

Chae

C. H.

Kim

J. W.

Kim

J. I

Kim

H. S.

Roh

S. C.

Park

J. B.

Woo

J. M.

Kim

S. Y.

Kim

J. Y.

… Son

D. K.

(2005). Developing an occupational stress scale for Korean employees. Korean Journal of Occupational and Environmental Medicine, 17(4), 297–317. https://doi.org/10.35371/kjoem.2005.17.4.297

12.

Cheung

G. W.

Rensvold

R. B.

(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233–255. https://doi.org/10.1207/S15328007SEM0902_5

13.

Choi

H. G.

Lee

J. M.

(2017). Validation of the Korean version of working as meaning inventory. The Korean Journal of Social and Personality Psychology, 31(4), 1–25. https://doi.org/10.21193/kjspp.2017.31.4.001

14.

Diener

Emmons

R. A.

Larsen

R. J.

Griffin

(1985). The Satisfaction With Life Scale. Journal of Personality Assessment, 49(1), 71–75. https://doi.org/10.1207/s15327752jpa4901_13

15.

Dooley

Prause

Ham-Rowbottom

K. A.

(2000). Underemployment and depression: Longitudinal relationships. Journal of Health and Social Behavior, 41, 421–436. https://doi.org/10.2307/2676295

16.

Dueber

D. M.

(2017). Bifactor indices calculator: A Microsoft Excel-based tool to calculate various indices relevant to bifactor CFA models. https://doi.org/10.13023/edp.tool.01

17.

Feldman

D. C.

(1996). The nature, antecedents and consequences of underemployment. Journal of Management, 22(3), 385–407. https://doi.org/10.1177/014920639602200302

18.

Feldman

D. C.

Turnley

W. H.

(1995). Underemployment among recent business college graduates. Journal of Organizational Behavior, 16(S1), 691–706. https://doi.org/10.1002/job.4030160708

19.

Friedland

D. S.

Price

R. H.

(2003). Underemployment: Consequences for the health and well-being of workers. American Journal of Community Psychology, 32(1/2), 33–45. https://doi.org/10.1023/a:1025638705649

20.

Heneman

H. G.

Schwab

D. P.

(1985). Pay satisfaction: Its multidimensional nature and measurement. International Journal of Psychology, 20(1), 129–141. https://doi.org/10.1080/00207598508246743

21.

Hoaglin

D. C.

Iglewicz

(1987). Fine-tuning some resistant rules for outlier labeling. Journal of the American Statistical Association, 82(400), 1147–1149. https://doi.org/10.2307/2289392

22.

Hooper

Coughlan

Mullen

M. R.

(2008). Structural equation modelling: Guidelines for determining model fit. Electronic Journal of Business Research Methods, 6(1), 53–60.

23.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. https://doi.org/10.1080/10705519909540118

24.

Hwang

D. S.

Lee

B. H.

(2012). Low wages and policy options in the Republic of Korea: Are policies working? International Labour Review, 151(3), 243–259. https://doi.org/10.1111/j.1564-913x.2012.00147.x

25.

International Labor Organization. (2017). World employment and social outlook: Trends 2017. Author.

26.

Jahoda

(1981). Work, employment, and unemployment: Values, theories, and approaches in social research. American Psychologist, 36(2), 184–191. https://doi.org/10.1037/0003-066X.36.2.184

27.

Jang

(2009). Social comparison and adjustment in two cultures. Korean Journal of Social and Personality Psychology, 23(2), 157–169. https://doi.org/10.21193/kjspp.2009.23.2.009

28.

Johnson

W. R.

Morrow

P. C.

Johnson

G. J.

(2002). An evaluation of a perceived overqualification scale across work settings. The Journal of Psychology, 136(4), 425–441. https://doi.org/10.1080/00223980209604169

29.

Kim

Lee

(2014). Neoliberalism and insecure employment in Korea: Emergence of the working poor and worsening socio-economic polarization. Korea Observer, 45(2), 255–273.

30.

Kim

Koo

M. J.

Lee

H. E.

Won

Y. L.

Song

(2019). Overwork-related disorders and recent improvement of national policy in South Korea. Journal of Occupational Health, 61(4), 288–296. https://doi.org/10.1002/1348-9585.12060

31.

Kim

Mone

M. A.

Kim

(2008). Relationships among self-efficacy, pay-for-performance perceptions, and pay satisfaction: A Korean examination. Human Performance, 21(2), 158–179. https://doi.org/10.1080/08959280801917727

32.

Kim

Allan

B. A.

(2019). Underemployment and meaningful work: The role of psychological needs. Journal of Career Assessment, 28, 76–90. https://doi.org/10.1177/1069072718824004

33.

Kim

Park

E. C.

Lee

T. H.

Kim

T. H.

(2016). Effect of working hours and precarious employment on depressive symptoms in South Korean employees: A longitudinal study. Occupational and Environmental Medicine, 73(12). 816–822. https://doi.org/10.1136/oemed-2016-103553

34.

Kim

(2016). Size and trend of irregular work. Korea Labor & Society Institute, 2016, 4, 1–36.

35.

Lee

(2013). Job and life satisfaction of nonstandard workers in South Korea. Workplace Health & Safety, 61(8), 355–363. https://doi.org/10.1177/216507991306100806

36.

Lee

D. H.

Lee

S. M.

(2000). The differences in job attitudes and performance between permanent and contingent workers. Korean Journal of Business Administration, 24, 249–272.

37.

Lee

(2015). The impact of offshoring on temporary workers: Evidence on wages from South Korea. Review of World Economics, 151(3), 555–587. https://doi.org/10.1007/s10290-015-0215-z

38.

Lee

K. H.

Kim

J. E.

Kim

Y. K.

Kang

D. M.

Yun

M. J.

Park

S. G.

Song

J. S

Lee

S. G.

(2013). Long working hours and emotional well-being in Korean manufacturing industry employees. Annals of Occupational and Environmental Medicine, 25(1), 38. https://doi.org/10.1186/2052-4374-25-38

39.

Lee

(2016). Analysis of statistics on college graduates finding major-related jobs. KRIVET Issue Brief, 91, 1–5.

40.

Lim

(2012). Psychometric properties of the Satisfaction With Life Scale among Korean police officers, university students, and adolescents. Korean Journal of General Psychology, 31(3), 877–896.

41.

Liu

Wang

(2012). Perceived overqualification: A review and recommendations for research and practice. In Perrewé

P. L.

Halbesleben

J. R. B.

Rosen

C. C.

(Eds.), The role of the economic crisis on occupational stress and well-being (pp. 1–42). Emerald Group Publishing Limited.

42.

Maynard

D. C.

Feldman

D. C.

(2011). Underemployment: Psychological, economic, and social challenges. Springer.

43.

Maynard

D. C.

Joseph

T. A.

Maynard

A. M.

(2006). Underemployment, job attitudes, and turnover intentions. Journal of Organizational Behavior, 27(4), 509–536. https://doi.org/10.1002/job.389

44.

McGoldrick

Robst

(1996). Gender differences in overeducation: A test of the theory of differential overqualification. The American Economic Review, 86(2), 280–284. https://doi.org/10.1080/000368496328858

45.

McGowan

M. A.

Andrews

(2015). Skill mismatch and public policy in OECD countries. OECD Economics Department Working Papers. https://doi.org/10.1787/5js1pzw9lnwk-en

46.

McGuinness

(2006). Overeducation in the labour market. Journal of Economic Surveys, 20, 387–418. https://doi.org/10.1111/j.0950-0804.2006.00284.x

47.

Menard

(1995). Applied logistic regression analysis: Sage University series on quantitative applications in the social sciences. Sage.

48.

Meredith

(1993). Measurement invariance, factor analysis and factorial invariance. Psychometrika, 58, 525–543. https://doi.org/10.1007/bf02294825

49.

Muthén

L. K.

Muthén

B. O.

(2012). Mplus user’s guide (7th ed.). Author.

50.

OECD. (2015). OECD skills strategy diagnostic report Korea 2015. http://www.oecd.org/skills/nationalskillsstrategies/Diagnostic-report-Korea.pdf

51.

OECD. (2018a). Hours worked (indicator). Author. https://data.oecd.org/emp/hours-worked.htm

52.

OECD. (2018b). Earnings dispersion and incidence of high and low pay. https://www.oecd.org/els/emp/employment-outlook-statistical-annex.htm

53.

Park

(2017). The relationships among job satisfaction, life satisfaction, and organizational commitment: Focused on the mediating effects of life satisfaction and moderating effects of employment status (permanent/contingent worker). Quarterly Journal of Labor Policy, 17(1), 27–54.

54.

Quinn

H. O.

(2014). Bifactor models, explained common variance (ECV), and the usefulness of scores from unidimensional item response theory analyses [Unpublished master’s thesis]. The University of North Carolina at Chapel Hill, NC.

55.

Riordan

C. M.

Vandenberg

R. J.

(1994). A central question in cross-cultural research: Do employees of different cultures interpret work-related measures in an equivalent manner? Journal of Management, 20, 643–671. https://doi.org/10.1016/0149-2063(94)90007-8

56.

Rodriguez

Reise

S. P.

Haviland

M. G.

(2016). Evaluating bifactor models: Calculating and interpreting statistical indices. Psychological Methods, 21, 137–150. https://doi.org/10.1037/met0000045

57.

Roh

Kim

(2009). A multilateral approach in the measurement of underemployment. Korean Journal of Industrial and Organizational Psychology, 22(3), 421–446. https://doi.org/10.24230/ksiop.22.3.200908.421

58.

Rothausen

T. J.

(2014). Eudaimonic job satisfaction: Exploring global and facet satisfaction with fulfilled job purposes [Doctoral dissertation]. University of St. Thomas, MN. https://ir.stthomas.edu/cgi/viewcontent.cgi?article=1001&context=ocbmgmtwp

59.

Schlomer

G. L.

Bauman

Card

N. A.

(2010). Best practices for missing data management in counseling psychology. Journal of Counseling psychology, 57(1), 1–10. https://doi.org/10.1037/a0018082

60.

Seth

M. J.

(2002). Education fever: Society, politics, and the pursuit of schooling in South Korea. University of Hawaii Press. https://doi.org/10.1515/9780824862305-011

61.

Shiffler

R. E.

(1988). Maximum Z scores and outliers. The American Statistician, 42(1), 79–80. https://doi.org/10.1080/00031305.1988.10475530

62.

Shin

(2009). An empirical analysis of the wage differentials between regular and irregular workers in Korean labor markets. Korean Society, 10(2), 93–123.

63.

Song

Kim

Lee

(2018). Validation of the Korean version of Global Hedonia-Eudaimonia Job Satisfaction Scale: A study on domestic application of a measurement for happiness in the social welfare profession. Korean Journal of Social Welfare Studies, 49(1), 191–219. https://doi.org/10.16999/kasws.2018.49.1.191

64.

Sorensen

C. W.

(1994). Success and education in South Korea. Comparative Education Review, 38(1), 10–35. https://doi.org/10.1086/447223

65.

Statistics Korea. (2018). Explore Korea through Statistics 2018. http://kostat.go.kr/portal/korea/kor_nw/1/3/1/index.board?bmode=read&aSeq=372640&pageNo=1&rowNum=10&amSeq=&sTarget=&sTxt=

66.

Steger

M. F.

Dik

B. J.

Duffy

R. D.

(2012). Measuring meaningful work: The Work and Meaning Inventory (WAMI). Journal of Career Assessment, 20, 322–337. https://doi.org/10.1177/1069072711436160

67.

Stucky

B. D.

Edelen

M. O.

(2015). Using hierarchical IRT models to create unidimensional measures from multidimensional data. In Reise

S. P.

Revicki

D. A.

(Eds.), Handbook of item response theory modeling: Applications to typical performance assessment, 183–206. Routledge.

68.

Tak

(2011). Relationships between various person–environment fit types and employee withdrawal behavior: A longitudinal study. Journal of Vocational Behavior, 78(2), 315–320. https://doi.org/10.1016/j.jvb.2010.11.006

69.

Tak

Seo

H. J.

Kim

H. S.

Nam

D. Y.

Jung

H. J.

Kwon

Kim

Jung

(2015). Development and validation of the Work Meaning Inventory. The Korean Journal of Industrial and Organizational Psychology, 28(3), 357–372.

70.

Tak

Seo

H. J.

Woo

Sim

(2017). A test of construct validity of the work meaning inventory: Based on employees. Korean Journal of Industrial and Organizational Psychology, 30(3), 357–372. https://doi.org/10.24230/ksiop.30.3.201708.357

71.

Umeda

(2018, May 18). South Korea: Labor standards law amended to limit workweek. Global Legal Monitor. http://www.loc.gov/law/foreignnews/article/south-korea-labor-standards-law-amended-to-limit-workweek/

72.

Weston

Gore

P. A.

(2006). A brief guide to structural equation modeling. The Counseling Psychologist, 34, 719–751. https://doi.org/10.1177/001100000628345

73.

White

Lehman

D. R.

(2005). Culture and social comparison seeking: The role of self-motives. Personality and Social Psychology Bulletin, 31(2), 232–242. https://doi.org/10.1177/0146167204271326

74.

Zvonkovic

A. M.

(1988). Underemployment: Individual and marital adjustment to income loss. Lifestyles, 9(2), 161–178. https://doi.org/10.1007/bf00986937

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB

0.25 MB