This study investigates the relationship between young adults’ graduate employability capacities (Coetzee, 2010), self-esteem (Battle, 1992) and career adaptability (Savickas & Porfeli, 2012). There appears to be a lack of research that investigates the association between graduate employability capacities, self-esteem and career adaptability, particularly in South Africa. The work of De Guzman and Choi (2013) and the study by Coetzee et al. (2015) are of relevance to this particular study, seeing that they have explored the associations between graduate employability capacities and career adaptability. The study by De Guzman and Choi (2013) explored the associations between employability skills (communication, problem-solving and teamwork skills) and career adaptability among technical secondary school learners. The study by Coetzee et al. (2015) identified important associations between employability capacities such as problem-solving and decision-making skills, interactive skills, goal-directed behaviour and continuous learning orientation, as well as career adaptability among human resource professionals in their early careers.

This study extends the work of De Guzman and Choi (2013) and Coetzee et al. (2015) by focusing on young adults in South African further education and training (FET) colleges, while furthermore adding self-esteem to the possible associations. Canonical correlation analysis, in particular, was employed in this study with the aim to establish (1) whether a significant relationship exists between graduate employability capacities, self-esteem and career adaptability capacities and (2) the variables that have the greatest influence on this relationship.

Graduate employability capacities, self-esteem and career adaptability are discussed in this section. The discussion of these three variables is followed by a section on theoretical integration.

In the literature, there is no full consensus as to what constitutes graduate employability capacities. Furthermore, the term ‘graduate employability capacities’ is used interchangeably with ‘graduateness skills and attributes’, ‘graduate attributes’, ‘employability attributes’, ‘employability capacities’ and ‘employability skills’. The term ‘graduate employability capacities’ is used for the purposes of this study.

Graduate employability capacities refer to those skills and attributes that are produced by an institution of higher education, which reflect the quality of personal growth and intellectual development of graduates and the relevance they bring to the workplace (Coetzee, 2012; Griesel & Parker, 2009; Steur, Jansen & Hofman, 2012). Regarded as the most critical skills in the current global job market, graduate employability capacities refer to those attributes that stimulate adaptability within revolutionising situations (Bezuidenhout, 2011; Potgieter, 2012).

Coetzee’s (2012) model is utilised in this study because of its relevance to South African graduates. This study also incorporates the work of Barrie (2004) and Steur et al. (2012). Coetzee (2012) identified eight core skills and attributes that constitute the graduate employability capacities of young adults, which fit into three attitudinal domains: (1) scholarship: problem-solving and decision-making skills, analytical thinking skills and enterprising skills; (2) global and moral citizenship: ethical and responsible behaviour, presenting and applying information skills and interactive skills; and (3) lifelong learning: goal-directed behaviour and an orientation towards continuous learning.

The 21st-century economy is progressing at an exponential rate, which emphasises the increasing requirement for employees to be highly qualified and highly skilled (Goodman & Tredway, 2016). Individuals who possess well-developed graduate employability capacities are most likely to engage in adaptive reasoning, action and influence and to amplify their ability to gain and to maintain suitable and sustainable employment, thereby advantaging themselves in the search for employment (Bezuidenhout, 2011; Coetzee, 2012; Fugate, Kinicki & Ashforth, 2004; Yorke & Knight, 2006).

Self-esteem is defined as a socially constructed emotion that reveals insights of a person’s various self-concepts and self-images, which are grounded on the psychological requirement for being accepted and affiliated with a social unit (Battle, 1992; Baumeister & Leary, 1995; Hewitt, 2002; Maslow, 1970). Self-esteem also refers to the thirst for accurate and successful functioning, capability and accomplishment when compared with other members of one’s social unit (Battle, 1992; Baumeister & Leary, 1995; Hewitt, 2002; Maslow, 1970). Battle (1992) proposed three dimensions to self-esteem, namely general self-esteem (an individual’s overall perceptions and feelings related to his or her worth), social self-esteem (an individual’s perception and feelings related to the quality of his or her relationships with peers) and personal self-esteem (an individual’s most intimate perceptions and feelings of self).

Positive self-esteem indicates that a person fits in well with the social world, is competent and able to meet the challenges of the world, is ready to participate in life within this social context and is able to balance social demands and personal desires (Hewitt, 1998; Scheff, 1990). This is then aligned with the purpose of career construction theory (Savickas, 2013), which requires a person to adapt to the changes in the social environment in order to ensure a good person–environment compatibility.

It is believed that individuals with low levels of self-esteem are more likely to display poor psychological adjustment when faced with stressful situations (Heimpel, Wood, Marshall & Brown, 2012; Zeigler-Hill & Wallace, 2012). A general way to increase people’s self-esteem is to teach them problem-solving skills that they can apply to a range of difficulties and challenges, thereby enabling them to better adjust to complexities in their environment (Mruk, 2013). Problem-solving skills have been identified as a key attribute that will improve one’s employability (Coetzee, 2012; Griesel & Parker, 2009).

One of the key components of career construction theory, namely, career adaptability, refers to a set of psychosocial resources. These resources assist in job search prospects and the formation of other career improvement opportunities, allowing a person to fit into the favoured work contexts by participating confidently in the work role (Ambiel, Carvalho, Martins & Tofoli, 2016; Duffy, 2010; Savickas, 2013; Tolentino et al., 2013). Career adaptability capacities are believed to assist individuals in coping with work-related transitions because of unanticipated changes in work and work-related environments (Savickas & Porfeli, 2012; Tolentino et al., 2013).

Career adaptability has been described with reference to four different dimensions that represent the coping mechanisms people employ to incorporate their self-concept into their vocational role and environment (Nota, Ginevra, Santilli & Soresi, 2014; Savickas & Porfeli, 2012). These four dimensions are (1) concern (concern for one’s future), (2) control (the extent to which an individual employs control over his or her future), (3) curiosity (demonstrating curiosity in the procurement and search for self- and occupational information) and (4) confidence (the extent to which individuals display a sense of self-esteem that will enable them to conquer hurdles that appear as they attempt to fulfil career goals) (Del Corso, 2013; Savickas & Porfeli, 2012). These dimensions of career adaptability have been shown to signify general adaptive resources and strategies required at different career transitions throughout the course of life (Savickas, 2013).

Career concern describes the degree to which employees are future oriented and proactive in their preparation for impending career tasks and challenges (Savickas & Porfeli, 2012). Being concerned about one’s future requires one to be alert, engaged, prepared and responsive (Savickas, 2013). Career control describes the degree to which employees accept personal responsibility pertaining to influencing their development and work environment by displaying self-discipline, effort and determination (Savickas & Porfeli, 2012). Individuals with a low sense of career control may display career indecision and struggle to cope with the 21st-century work environment (Coetzee et al., 2015; Del Corso, 2013).

Career curiosity involves individuals exploring possible future selves and opportunities and thinking about how they may influence various work roles and environments (Savickas & Porfeli, 2012). In order to adjust to changes in the working world, individuals need to be inquisitive and adventurous and take risks (Savickas, 2013). Career confidence describes individuals’ beliefs that they can transform their career goals into reality and successfully rise above any obstacles (Savickas & Porfeli, 2012). The way in which individuals deal with the stressors they face throughout their lifetime, both occupationally and personally, displays their career confidence (Del Corso, 2013).

When faced with the turbulent working world of the 21st century, young adults who possess career adaptability resources are more likely to flourish (Maree, 2012; Savickas & Porfeli, 2012). Young adults need to learn to adapt to, to become accustomed to and to delight in constant change (Blewitt, 2010). Employees are required to accept responsibility for their careers and, therefore, greater emphasis is placed on psychological, career-related meta-capacities such as graduate employability capacities, career adaptability and a high concept of self (self-esteem) (Ferreira & Coetzee, 2010).

Graduate employability capacities, self-esteem and career adaptability are important self-supervised psychosocial resources for individuals’ continuous learning and employability (Bezuidenhout, 2011; Coetzee, 2012). Being career adaptable involves engaging in lifelong learning to acquire new skills and procedures, transferring skills within and between contexts, effectively dealing with any uncertainty, viewing new situations as opportunities rather than obstacles and being self-aware and reflecting on one’s own actions (Creed, Macpherson & Hood, 2010). Young adult graduates need to have a willingness to learn, which constitutes lifelong learning, and this willingness or enthusiasm to learn arises primarily from a good level of self-esteem (Ceccatelli & Battista, 2012; Coetzee, 2012).

As separate parts, graduate employability capacities, self-esteem and career adaptability are crucial to enhancing young adults’ employability. The purpose of this research is to investigate whether graduate employability capacities and self-esteem, when combined, will influence young adults’ career adaptability and ultimately affect their employability. More specifically, the research question posed in this study is: Do graduate employability capacities and self-esteem predict and explain young adults’ career adaptability?

It is believed that through the exploration of this research question, career counsellors can focus on these capacities in enhancing young adults’ career adaptability, thereby helping them to be more adaptable to the turbulent world of work and, therefore, more employable.

A cross-sectional, quantitative research approach was followed to collect primary data. This took the form of a paper-based, self-report questionnaire to allow for a statistical analysis of the data and for associations to be made between the variables.

A non-probability convenience sample (N = 332) of undergraduate young adult learners contributed to this study. The participants (aged 18–29) were enrolled at two South African FET colleges; 98.5% of the participants were black people and 62% were women. In total, 95% of the sample was represented by unemployed individuals.

Graduate employability capacities were measured using the graduateness skills and attributes scale (GSAS) (Coetzee, 2010). The GSAS is a self-rating multifactorial scale that is measured on a six-point Likert scale (1 = never; 6 = always) and consists of 64 items in total. There are eight subscales, namely problem-solving and decision-making skills (eight items, e.g. ‘I make quick but clear decisions that spur others on toward action’), enterprising skills (nine items, e.g. ‘I prefer to work under my own direction’), analytical thinking skills (four items, e.g. ‘I feel confident in my ability to draw insightful conclusions from numerical data’), interactive skills (16 items, e.g. ‘I can communicate my viewpoints with clarity and fluency in English’), presenting and applying information skills (five items, e.g. ‘I can write my ideas and opinions clearly to convince my audience’), ethical and responsible behaviour (five items, e.g. ‘I accept responsibility for the results of my decisions and actions’), goal-directed behaviour (10 items, e.g. ‘I develop plans for specific goals and tasks’) and continuous learning orientation (seven items, e.g. ‘I monitor my performance against deadlines and milestones’). Coetzee (2010) confirmed the construct’s convergent and discriminant validity through the use of exploratory factor analysis and correlational analysis. In terms of this study, the overall GSAS obtained a reliability coefficient of 0.95, while the internal consistency reliability coefficients for the subscales ranged between 0.58 (ethical and responsible behaviour) and 0.85 (interactive skills).

Self-esteem was measured using the Culture-Free Self-esteem Inventory for Adults (CFSEI 2-AD) (Battle, 1992). The CFSEI 2-AD is a self-rating scale that consists of 40 items (e.g. ‘I can do most things as well as others’, ‘I am usually successful when I attempt important tasks or assignments’ and ‘I often feel that I am not good at all’). A six-point Likert scale (1 = strongly disagree; 6 = strongly agree) was used to record the responses of the participants to each of the 40 items. There are four subscales, namely general self-esteem (16 items, e.g. ‘I am happy most of the time’), personal self-esteem (eight items, e.g. ‘I am as nice looking as most people’), social self-esteem (eight items, e.g. ‘Most people I know like me’) and lie items (eight items, e.g. ‘I have taken something that does not belong to me’). The lie items can be referred to as a subtest that indicates defensiveness. Battle (1992) confirmed content validity of the CFSEI 2-AD through the development of a construct definition of self-esteem and through the recoding of items meant to encompass all aspects of the construct. Internal consistency for the factor analysis proved the data to be significant (0.81) (Battle, 1992). In terms of this study, the overall CFSEI 2-AD obtained a reliability coefficient of 0.82, while the internal consistency reliability for the subscales ranged between 0.31 (social self-esteem) and 0.74 (general self-esteem and personal self-esteem). It seems that a low internal consistency reliability value is generally reported in respect of social self-esteem, as other researchers have reported the following figures: 0.57 (Brooke, 1995), 0.38 (Ogle, 2012) and 0.57 (Potgieter, 2012).

Career adaptability capacities were measured using the Career Adapt-Abilities Scale (CAAS) (Savickas & Porfeli, 2012). The CAAS is a self-rating multifactorial scale that is measured on a five-point Likert scale (1 = not strong; 5 = strongest) and consists of 24 items divided into 4 subscales: concern (six items, e.g. ‘Realising that today’s choices shape my future’), control (six items, e.g. ‘Keeping upbeat’), curiosity (six items, e.g. ‘Exploring my surroundings’) and confidence (six items, e.g. ‘Working up to my ability’). Within the South African context, Maree (2012) confirmed the instrument’s construct and structural validity through the use of confirmatory factor analysis, where he found that the internal consistency reliability of the overall CAAS was 0.91 and the reliability of the subscales ranged from 0.71 (control) to 0.80 (confidence). In terms of this study, the overall CAAS obtained a reliability coefficient of 0.91, while the internal consistency reliability coefficients for the subscales ranged between 0.68 (concern) and 0.78 (curiosity).

Paper-based questionnaires were distributed to participants at the beginning of one of their classes and were collected at the end of the class. Respondents were given approximately 20–25 min to complete the questionnaires. A total of 355 students completed the questionnaires, after which 332 questionnaires were identified as usable for the purpose of the study (N = 332). Thus, a response rate of 93.5% was achieved. Data were coded, cleaned and captured.

Descriptive statistics (means, standard deviations and Cronbach’s alpha coefficients), Pearson product-moment correlations and canonical correlation analysis were employed in the study using JMP software (from SAS). In interpreting Pearson’s correlations (r), Cohen’s (1992) practical significance was used, that is, r < 0.10 (small practical effect), r < 0.30 (moderate practical effect) and r > 0.50 (large practical effect). In order to counter multicollinearity, concerns were set at r > 0.90 (Hair, Black, Babin & Anderson, 2010). With regard to the canonical statistics, self-esteem and graduate employability capacities were combined to form the skills scores (independent variable) and career adaptability (dependent variable).

The following function was computed: graduate employability capacities and self-esteem (independent variable) and career adaptability (dependent variable). Wilks’s lambda chi-square test was performed to test for the significance, r²-type effect size (yielded by 1 – 0.λ), of the overall canonical correlation between the independent and dependent variates of a canonical function. In order to counter the probability of a Type I error, the significance value for interpreting the results was set at a 95% confidence interval level (Fp ≤ 0.05). The cut-off criteria for canonical factorial loadings (≥0.30) were used to interpret the relative importance of the canonical structure correlations or loadings in deriving the canonical variate constructs (Hair et al., 2010). The redundancy index was also considered in assessing the magnitude of the overall correlational relationships between the two variates of a canonical function and the practical significance of the predictive ability of the canonical relationship (Hair et al., 2010). The squared canonical correlation (Rc²) values of ≤ 0.12 (small practical effect), ≥ 0.13 ≤ 0.25 (medium practical effect) and ≥ 0.26 (large practical effect) (Fp ≤ 0.05) (Cohen, 1992) were also considered in the interpretation of the magnitude of the practical significance of the results.

Table 1 displays the reliability coefficients that indicate the acceptable internal consistency reliability of the three scales and the subscales. The overall scale reliability coefficient of the GSAS (α = 0.95), the overall self-esteem scale (α = 0.82) and the overall career adaptability scale (α = 0.91) were very high, implying strong overall consistency for the three scales.

Also displayed in Table 1 are the correlations between the GSAS, the CFSEI 2-AD and the CAAS variables, which ranged from r ≥ 0.21 (small practical effect) to r ≥ 0.62 (large practical effect), indicating significant positive associations between self-esteem, graduate employability capacities and career adaptability variables. The GSAS subscales displayed high correlations with the overall GSAS scale (r ≥ 0.67 ≤ 0.94). The CAAS subscales also correlated strongly with the overall CAAS scale (r ≥ 0.80 ≤ 0.88).

The significant intercorrelations between the GSAS, the CFSEI 2-AD and the CAAS variables were moderate to large in effect size (r ≥ 0.15 ≤ 0.40; p ≤ 0.01), which implies that multicollinearity did not affect the interpretation of the canonical correlation results. The association between gender and the CFSEI 2-AD, the GSAS and the CAAS was mostly insignificant and small in practical effect and was therefore regarded as negligible.

Canonical correlation analysis was used to study the multivariate relationships between the eight GSAS scores (problem-solving and decision-making skills, enterprising skills, analytical thinking skills, interactive skills, presenting and applying information skills, ethical and responsible behaviour, goal-directed behaviour and continuous learning orientation), the three CFSEI 2-AD scores (general self-esteem, personal self-esteem and social self-esteem) and the four CAAS scores (career concern, career control, career curiosity and career confidence). The GSAS scores and the CFSEI 2-AD scores were treated as independent variables and were combined to form the skills set variable, and the CAAS scores were treated as dependent variables.

Table 2 confirms that the full model was significant, using Wilks’s multivariate (Wilks’s lambda, λ, = 0.49, function 1: Fp = 5.73; p = 0.0001). Only the first function was significant and contributed 45% (Rc² = 0.45) of the overall explained variation, relative to the first function. The full model r²-type effect size (yielded by 1 – 0.λ) was 0.51 (large practical effect), indicating that the full model explained an adequate proportion – about 51% – of the variance shared between the two variable sets. The redundancy index (Table 2) indicated that the skill variate (graduate employability capacities and self-esteem) explained 31% (Rc² = 0.31; large practical effect) of the overall variance in the career adaptability canonical variate variables and was able to predict 45% (large practical effect) of the proportion of the overall variance in the dependent canonical variate construct (career adaptability).

Table 2 further displays that the four CAAS variables were all loaded positively with the career adaptability canonical construct and were assumed to represent high career adaptability. Career concern (Rc = 0.70) displayed the lowest loading in comparison to career control (Rc = 0.76), curiosity (Rc = 0.87) and confidence (Rc = 0.96).

Social self-esteem (Rc = 0.36) displayed the highest loading with the skills canonical construct when compared with the other self-esteem capacities. Ethical and responsible behaviour (Rc = 0.57) displayed the lowest loading with the skills canonical construct when compared with the other graduate employability capacities such as analytical thinking skills (Rc = 0.62), presenting and applying information skills (Rc = 0.72), goal-directed behaviour (Rc = 0.78), enterprising skills (Rc = 0.83), interactive skills (Rc = 0.88) and continuous learning orientation (Rc = 0.88).

Table 2 further displays that when 0.30 was used as the cut-off value for interpreting the results, the self-esteem and graduate employability capacities positively contributed in explaining the variance of all the career adaptability variables: career concern (Rc = 0.47; 22%), career control (Rc = 0.51; 26%), career curiosity (Rc = 0.58; 34%) and career confidence (Rc = 0.64; 42%). The career adaptability variables contributed positively in explaining the variance of all the graduate employability capacities: interactive skills (Rc = 0.59; 35%), continuous learning orientation (Rc = 0.59; 35%), problem-solving skills (Rc = 0.57; 32%), enterprising skills (Rc = 0.56; 31%), goal-directed behaviour (Rc = 0.53; 28%), presenting and applying information skills (Rc = 0.48; 23%), analytical thinking skills (Rc = 0.42; 18%) and ethical and responsible behaviour (Rc = 0.39; 15%).

Certain limitations associated with the current study must be noted. Owing to the small sample size and the sampling strategy utilised, the findings cannot be generalised to the South African young adult population. The interpretations of the findings of this study are limited by the correlational design and by the cross-sectional nature of the data, implying that no interpretations can be made relating to the causality of the associations of the variables. The study is also limited to a particular moment in time. According to Hair et al. (2010), canonical correlation analysis is a maximisation technique and researchers should be open to the possibility of overestimating when interpreting the results because of the intensification of the linear composites. As a result of the self-report nature of the study, the possibility of common-method bias should be noted.

Future research could reproduce this study, using a more representative sample, especially in terms of age, race and gender. Future longitudinal studies could also clarify the relationships between graduate employability capacities, self-esteem and career adaptability as reported in this study and may also supply evidence of the way in which these variables evolve over time as individuals progress from one transition to the next.

This study proposes that young adults’ career adaptability may be improved through the development of their graduate employability capacities. Industrial psychologists and career counsellors should engage in interventions to assist young adults in enhancing their graduate employability capacities in order to enhance their career adaptability and, consequently, their employability.

Individuals who measure high on career adaptability are sought after by employers because they respond to transitions better than those who measure low on career adaptability. Career development interventions should concentrate on supporting young adults in formulating their employability capacities in order to increase their confidence and self-efficacy in demonstrating their ability to manage their careers and employability (Del Corso, 2013). Career adaptability resources can be enhanced through the facilitation of career interventions such as time-perspective workshops that promote future orientation and planfulness (concern), information-seeking activities (curiosity), self-esteem enhancement (confidence) and decision-making training (control) (Savickas, 2013; Tolentino et al., 2013).