Introduction

Making a career choice and getting a good job after graduation are among university students’ greatest concerns and interests (Doygun & Gulec, 2012). Their career decisions affect not only their own futures, but also the economic prosperity of the nation (Kazi & Akhlaq, 2017). While choosing a career is a critical aspect of university students’ lives, they often face difficulties in doing so (Mberia & Midigo, 2018). Due to the need for specialization in workplace, choosing a career is not an easy or a straightforward undertaking (Nyamwange, 2016). Even one’s career decision status, that is, whether one has decided on a career or not, influences work outcomes, as it is known that those who make a career choice are more likely to be committed to their workplaces (Earl & Bright, 2007).

In South Korea, youth unemployment is a critical social issue. A rapidly growing number of college graduates are unemployed due to the slow growth of professional and semi-professional jobs (Choi, 2017). Students typically feel excessive stress and experience difficulty in making career decisions. To address this issue, most universities in South Korea have been providing career services for over a decade to help their students secure jobs (Yang et al., 2012). Nevertheless, most students have been found to have low career decision-making self-efficacy and limited ability to explore employment options (Lee & Hong, 2013).

Career decision-making is a complicated process. Rather than characterizing it based on a single dominant feature, an individual’s career decision-making process is best seen as a multidimensional model (Gati, 2010). There are numerous career theories and models, and no single one is sufficient to explain the broad scope of individuals’ career decision-making. Several theories, such as the person–environment fit theory (Holland, 1997), social cognitive theory (Bandura, 1997), community interaction theory (Law, 1981), and career construction theory (Savickas, 2005), can be used to describe how people manage their career decisions and preparation. Savickas and Porfeli's model (2012) of career adaptability as a multidimensional construct consisting of four aspects (concern, control, curiosity, and confidence) has attracted increased attention to provide an explanation for career decision-making. These four sub-constructs are not interchangeable representations of career adaptability stems, and different dimensions can be differentially related to potential predictors and outcomes (Hirschi & Valero, 2015).

Career adaptability dimensions have been used to explore the relationship with antecedents including adaptiveness (Neureiter & Traut-Mattausch, 2017) and outcome variables such as career planning, career decision-making difficulties, career exploration, and occupational self-efficacy beliefs (Hirschi & Valero, 2015). Career adaptability is also used to predict advancements in one’s career development (Bocciardi et al., 2017). Several studies were implemented to identify career decision processes with samples of adolescents (Hirschi, 2009; Ozdemir & Guneri, 2017). However, few studies have adopted career adaptability as a predictor to discriminate university students’ career decision status (i.e., yes or no). Moreover, few studies were conducted to identify the relative predictive influences among the sub-constructs (concern, control, curiosity, and confidence) with other related variables such as social support, academic major relevance, and university life satisfaction (Park et al., 2018). Therefore, the aim of this study is to identify the relationship between university students’ career decision status and four selected variables: career adaptability, social support, academic major relevance, and university life satisfaction. In addition, as only few studies have revealed the productive accuracy rate of discriminating university students’ career decision status based on the above variables, this study aims to determine how they can be used to accurately discriminate students’ career decisions (hit ratio).

Career adaptability

Researchers have reported that career adaptability plays a role in influencing or enhancing career decisions and career success (Hailbo et al., 2018). It is also identified as a mediating variable in predicting career planning (Taber & Blankemeyer, 2015) and career exploration (Li et al., 2015). Career adaptability also affects career satisfaction and career performance (Zacher, 2014). As per career construction theory, careers are constructed through adaptive strategies that implement an individual’s personality to fit the occupational role in vocational environments (Savickas, 2005). These adaptive strategies are composed of four sub-factors: concern, which refers to the commitment to making choices about one’s career-related future; control, which involves controlling to prepare for one’s vocational future; curiosity, which denotes exploring possible options and indicates interest in the world of work; and confidence, which is related to the strengthening of self-efficacy regarding career-related problem solving and positive attitudes for overcoming obstacles (Savickas & Porfeli, 2012). These four sub-factors can be employed independently under the higher-order construct of career adaptability.

Considering these adaptability resources have clear benefits, it is also important to understand that they are not sufficient for explaining one’s career planning, career exploration, or occupational self-efficacy (Neureiter & Traut-Mattausch, 2017). Park et al. (2018), for instance, include not only the four sub-career adaptability factors as internal factors, but also social support, university life satisfaction, and academic major relevance as external factors to explain university students’ career decisions. Similarly, following their research approach, we added these variables to the research to discriminate university students’ career decision status.

Social support

Social psychology focuses on social interactions. People’s thoughts, feelings, and behaviors are influenced by the actual, imagined, or implied presence of others through social interaction (Allport, 1983). Social psychology consists of two disciplinary fields: social cognition and social influence. The study of social influences explores how one’s emotions, judgments, or behavior are affected by others (Crider et al., 1989). Social support, as one type of social influence, is defined as social interaction aimed to induce positive outcomes (Bianco & Eklund, 2001). It can be an important resource for career-related information, guidance, and advice for young adult students (Seibert et al., 2001). This type of social support is known to positively affect career planning and negatively affect career concerns (Creed et al., 2009). Furthermore, social support is an important resource in coping with stress at a personal, social, academic, and economic level during the university years (Civitci, 2015).

Academic major relevance

An academic major is an academic discipline in which an undergraduate student formally commits to taking up an undergraduate degree. In South Korea, students are usually required to choose their major discipline before entering university. We adopted academic major relevance as one of the discriminant variables in addition to career adaptability, for educational choices, such as choosing a major, should be considered as important career-related decisions (Germeijs et al., 2012). Students perceive their academic major to be relevant to their vocational preparation as well as their personal growth and development (Pisarik & Whelchel, 2018). Educators also believe students’ academic courses connect their career and personal lives (Belet, 2018). For instance, university students tend to change their major when they change their career direction (Willcoxson & Wyndeer, 2010). Furthermore, a student’s academic major affects future job satisfaction through job-field congruence (Wolniak & Pascarella, 2005).

University life satisfaction

A person’s level of satisfaction regarding life circumstances (life satisfaction) is a part of one’s subjective well-being, which depends on it (Brief et al., 1993). Since university students often face social, academic, and economic stress (Schnettler et al., 2017), their life satisfaction level is an important factor that administrative staff and instructors should consider. Students’ satisfaction with school plays a significant role in influencing behavioral engagement, school success, progress, and academic results (Doğan & Çelik, 2014). Previous studies have shown that life and career satisfaction are positively related (Erdogan et al., 2012). Karavdic and Baumann (2014) insisted that career adaptability and career optimism are related to life satisfaction. They further suggested that university and post-university interventions need to be developed to improve life satisfaction. Similarly, Garrison, Lee, and Ali (2016) reported that career identity and life satisfaction of university students in South Korea were related.

Method

Sample and data collection

The study population is comprised of 15,000 undergraduate university students at Konkuk University in Seoul, South Korea. Normally, more than 8,000 students take at least one e-learning course offered by the university every year. The data were collected through an online survey on eCampus, the university’s e-learning management system, and automatically coded in MS Excel. Students who take e-learning courses were encouraged to voluntarily participate in the survey without any reward. In total, 1,297 students (8.65%) joined the survey. Among these students, 569 (approximately 44%) expressed that they had not yet decided on their future career and job. Table 1 presents the demographic information of the sample and the students’ primary sources of career and job information.

Table 1 Sample demographic information
Full size table

Instruments

We used the Career Adapt-Abilities Scale (CAAS), developed by Savickas and Porfeli (2012), which originally contained 24 items. We further included 20 items in the study after factor analysis. The instruments for other independent variables, namely social support, academic major relevance, and university life satisfaction, were developed in accordance with the objectives of the study and prior literature review. The items related to social support and academic major relevance were created based on the study by Park et al. (2018). University life satisfaction items were developed based on the study by Park et al. (2014). All items were measured on a five-point Likert-type scale, from 1 (strongly disagree) to 5 (strongly agree), except for items related to demographic information.

Validity, reliability, means, and standard deviations of construct items

Two factor analyses were implemented to ensure the validity of items. We employed common factor analysis techniques with the varimax rotation method to extract factors. In particular, to identify the common factor structure, a priori criterion was adopted to extract factors (SAS option, nfact = 4), for career adaptability was known to have four sub-factors (career concern, career control, career confidence, and career curiosity). Each of the factors originally consisted of six items. However, we deleted some items, which were less than .40 in accordance with factor loadings. After deleting the items with low factor loading coefficients, we re-implemented factor analysis. Finally, all factor loadings exceeded.45, and all eigenvalues were greater than 1.0, indicating significance (Hair et al., 1998).

The appropriateness of the factor analyses was confirmed by calculating Kaiser’s overall measure of sampling adequacy (MSA) and the root mean square off-diagonal residuals (RMSR). MSAs were .93 and.85. According to the rule of thumb, .8 or above is considered good. Additionally, the RMSR were .02 and .03, which are considered small enough to be appropriate. Tables 2 and 3 present each items’ mean, standard deviation, and reliability coefficient, as well as the factor analysis results of career adaptability and other independent variables.

Table 2 Means, standard deviations, and factor analysis of CN, CT, CU, and CF
Full size table
Table 3 Means, standard deviations, and factor analysis of SS, MR, and US
Full size table

In addition, discriminant validity and multicollinearity were checked by examining the correlations among variables. Since all correlation coefficients were less than 0.85, we concluded that there were sufficient discriminant validity and low multicollinearity (David, 2012). Table 4 shows the correlation coefficients among the variables. The correlation matrix also provides some general ideas about the magnitude of interrelationships among the variables.

Table 4 Correlations among variables
Full size table

Statistical procedures

The online MS Excel data were directly imported into SAS version 9.4. Descriptive statistics such as the mean, standard deviation, frequency, and percentage were calculated. Several appropriate tests required assumptions such as normality and homogeneity to be met. These tests were conducted before the inferential statistical analyses. A discriminant analysis and several t tests were implemented. In addition, two factor analyses and a correlation analysis were conducted for the validity and reliability tests.

Results

Before conducting the discriminant analysis, several t tests were conducted to identify statistical differences in the mean scores of career adaptability sub-factors and the selected variables between the career decision group and the non-career decision group. All mean scores of the career decision group were greater than those of the non-career decision group. The differences were significant in accordance with the t and p values.

We formulated the equation of discriminant function to differentiate career decision groups based on raw canonical coefficients (unstandardized discriminant coefficients), as follows (1):

$$Z\left( {{\text{discriminant}}\;{\text{score}}} \right) = 1.628\left( {{\text{career}}\;{\text{concern}}} \right) + .361\left( {{\text{career}}\;{\text{control}}} \right) - .826\left( {{\text{career}}\;{\text{curiosity}}} \right) + .289\left( {{\text{career}}\;{\text{confidence}}} \right) + .053\left( {{\text{social}}\;{\text{support}}} \right) + .149\left( {{\text{major}}\;{\text{relevance}}} \right) - .221\left( {{\text{university}}\;{\text{life}}\;{\text{satisfaction}}} \right).$$

Since the canonical correlation was .356 and Wilks’ lambda was .873, with F = 26.74 and p < .001, the equation was accepted. Thus, the discriminant function discriminates university students’ career decisions. Table 5 shows the means, standard deviations, and results of the t test and discriminant analysis.

Table 5 Means, standard deviations, and results of the t test and discriminant analysis
Full size table

To identify the relative influence of each independent variable on university students’ career decisions (i.e., yes or no), discriminant loadings (canonical structure correlations) were produced by the discriminant function (see Table 6). Discriminant loadings are suitable for interpreting magnitude, for standardized discriminant coefficients may have considerable instability and deficiencies in the case of multicollinearity (Hair et al., 1998). All loadings were positive and greater than .20. Career concern (.91) was found to be the most discriminant, followed by career control (.55). University life satisfaction (.22) was the least discriminant. Career curiosity (.23) was also identified as having less influence on university students’ career decisions as a career adaptability sub-factor.

Table 6 Discriminant loadings
Full size table

Finally, the overall discriminant model fit was evaluated by examining the predictive accuracy level. The error count estimate was 34.85%. We also present the posterior probability error rate (35.35%) to reduce bias (overfitting) from the error count estimates (Hora & Wilcox, 1982). The posterior error rate produced about 65% of classification accuracy (hit ratio). This predictive accuracy level was acceptable. In general, a hit ratio larger than 25% is acceptable. SAS set 50% as a priori default value of the hit ratio. Table 7 presents the classification matrix of the discriminant function. It includes the number of predictive observations and two types of error estimates classified based on university students’ career decision.

Table 7 Classification matrix for discriminant analysis
Full size table

Discussion and conclusion

Understanding the factors affecting university students’ career decisions is important because their career decision influences both job preparation and future occupational employment (Lee et al., 2019). The results of this research support career construction theory, which provides the basis for career adaptability to predict students’ career decision status. According to Savickas and Porfeli (2012), career adaptability consists of four sub-factors: career concern, career control, career confidence, and career curiosity. The multivariate discriminant function included four sub-factors and selected variables (university satisfaction, academic major relevance, and social support), which were all found to be significant in discriminating the career decision group (yes) from the non-career decision group (no).

Among the variables, “career concerns” was identified as the most discriminate variable, whereas university life satisfaction was identified as the least discriminate. This is consistent with the results of Park et al. (2018), who reported career concern to be the most discriminate variable, for university students consider job preparation to be the most relevant concern. Therefore, career concern was identified as having the most predictive power in comparison with the other career adaptability sub-factors.

Career curiosity had the least predictive power related to career decision-making among the career adaptability sub-factors. This result was similar to that found by Neureiter and Traut-Mattausch (2017). While they reported the correlation coefficient between career curiosity and career planning to be the lowest, they reported the correlation coefficient between career concern and career planning to be the highest. They argued that career concern was related to career planning, career decision-making difficulties, and career exploration. However, career curiosity was related to career exploration, and career control and confidence were not directly related to career planning or career decision-making difficulties. Since this study’s sample was university students having more career curiosity than adults, when compared to middle and high school students, their career curiosity was identified as being less predictive and carrying less discriminant power to influence career decision status.

We added social support, academic major relevance, and university life satisfaction as independent variables to the multivariate function to discriminate university students’ career decision status in addition to career adaptability. Among them, academic major relevance was identified as the most discriminant variable. This might be because, by establishing a relationship between academic major and career relevance, students are provided with an important opportunity to relate their subject matter to possible employment. Social support and university life satisfaction were identified as being weak discriminant variables in predicting university students’ career status. According to Lee and Lee (2018), social support was found to moderately influence adolescents’ career maturity. University students tend to be more independent in comparison with secondary education students in terms of career decisions, which could be a reason.

Career decision self-efficacy was positively related to life satisfaction among university students (Jiang & Hu, 2016). Furthermore, happiness and life satisfaction may play an important role among university students preparing for the transition to work (Karavdic & Baumann, 2014). University students who reported being more satisfied in life were more certain of their career choices (Zhou & Xu, 2013). However, life satisfaction’s weak discriminant magnitude in the study may be due to the sample’s characteristic. Most participants in the study were not resident college students, but commuters, which may be why their university life satisfaction did not strongly influence their career decision. It is generally accepted that the extra-curricular activities and campus facilities of the university positively impact life satisfaction.

Implications

We concluded that the multivariate function was appropriate to discriminate students’ career decision status and that its predictive accuracy was relatively high. Therefore, the function would be helpful to university educational administrators in predicting students’ career decision status. More specifically, newer students (freshmen and sophomores) could be better oriented after identifying their career decision status.

Based on these results, the following recommendations are suggested to educational administrators and career professionals. First, one application of the findings is the development of tailored non-academic programs related to career guidance after identifying students’ career decision status and career adaptability. Specifically, university educational administrators could provide career counseling programs to students in accordance with their scores on the four sub-factors (career concern, career control, career confidence, and career curiosity). Second, educators and managers should make an effort to boost university students’ life satisfaction due to its influence on career decisions. Third, they need to provide differentiated career-related services in accordance with students’ majors. Students who are social sciences majors may require different career services in comparison with those majoring in the natural sciences or engineering. Fourth, as social support, university administrators should encourage and motivate professors and senior students to help newer students make career decisions, for career decisions are influenced by grades and educational background (Shahzad et al., 2018). For example, the career exploration workshops provided by the university, including both professors and their students, could be helpful in improving appropriate career decision-making processes and decreasing career decision-making difficulties. Lastly, career professionals in universities are recommended to develop and implement action plans to increase students’ career adaptability in accordance with their demographic factors, such as university life status, study major, and social supports, for these factors were identified as affecting students’ career decisions and preparedness.

Limitations

His study implemented statistical analyses that included a student sample across all class standings (freshmen, sophomore, junior, and senior students). However, the discriminant and predictive influence of selected variables, including the four sub-factors of career adaptability, would likely have been different had we divided the sample in accordance with their standing. Therefore, it is highly recommended that future research is carried out based on students’ ability to identify the predictive magnitudes of these variables. Finally, the research results were limited to the Konkuk University students in the Seoul campus, which is located in the South Korean capital. Comparative research should be conducted to identify if there are differences with the students in other provinces, including rural areas.