Declaring employability
A meta-analysis on flexibility, initiative, and GMA
as antecedents of employability
UvA – Amsterdam Business School
MSc in Business Administration – Leadership and Management Track
Author: Leroy Kleijn
Student number: 10894462 Date of submission: 24-06-2016
Version: Final version
First supervisor: Dr. S.T. Mol Second supervisor: Mw. S. Pajic
Statement of originality
This document is written by Student Leroy Kleijn who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
Statement of originality... 2
Abstract ... 5
1. Introduction ... 6
2. Literature review ... 8
2.1 Employability ... 8 2.2 History ... 8 2.3 Defining employability ... 9 2.4 Perceived employability ... 11 2.5 Perspectives employability ... 132.6 Working towards the meta-analysis ... 14
3. Hypotheses development ... 14
3.1 Flexibility ... 14 3.2 Initiative ... 18 3.3 GMA ... 20 3.4 Moderators ... 244. Methods ... 25
4.1 Search strategy ... 264.2 Study selection process ... 26
4.3 Coding system ... 28 4.4 Publication bias ... 28 4.5 Types of variables ... 29 4.6 Summary measures ... 29 4.7 Methods of analyses ... 29 4.8 Moderator analyses ... 29
5. Results ... 30
5.1 Flexibility ... 31 5.2 Initiative ... 33 5.3 GMA ... 366. Discussion ... 38
6.1 Theoretical implications ... 38 6.2 Practical implications ... 39 6.3 Limitations ... 407. Conclusion ... 42
8. Appendices ... 44
Appendix A; Defining the constructs ... 44
Appendix B; Search strategy PsycINFO, OVID ... 45
Appendix C; Coding sheet variables and descriptions ... 45
Appendix D; Used studies in meta-analysis ... 50
Studies Flexibility ... 50
Studies Initiative ... 54
Studies GMA ... 56
Abstract
Employability has become a topic of great scientific interest due to increasing labour market uncertainty. Within the literature on employability it seems that a single clear definition is missing. That is why Thijssen, Heijden, and Rocco (2008, p. 167) described employability as “an attractive but confusing buzzword”. However, the most common definition of employability comes from Hillage and Pollard (1998, p. 2): "the capability to move self-sufficiently within the labour market to realise potential through sustainable employment”. Three antecedents of employability, namely flexibility, initiative, and general mental ability (GMA), were hypothesized to have a positive relationship with employability. The meta-analytic results showed that all hypothesized direct relationships were significant and positive. A meta-analysis was conducted on the relationship of flexibility (N = 12), initiative (N = 7), and GMA (N = 4) with employability. Also age, gender, and employment status were tested as moderators of the hypothesized direct relationships by ANOVA tests and regression tests. The results showed significant evidence for the antecedent role of flexibility (r = .2474; p < .001), initiative (r = .3567; p < .001), and GMA (r = .3948; p = .011) on employability. Also age tested significantly as a moderator of the relationships between initiative and employability (B = -.0153; p = .0394) and GMA and employability (B = -.0496; p < .001). Employment status thus appears to moderate the relationship between initiative and employability (Q = 4.2524; p = .0392). In short, this study recommends a flexible approach towards chances on the labour market, proactive behaviour on the job or in searching for a job (initiative), and intelligence (GMA). These aspects are significantly tested as being predictors of employability. Also, starting as young as possible (i.e. as a student) is preferable.
Key-words: (perceived) employability; career adaptability; flexibility; willingness;
1. Introduction
The contemporary ‘trend’ to reorganize companies (Finn, 2000; Doherty, 1996; Probst, 2003; Worrall, Parkes, & Cooper, 2004) leads to increasing feelings of job insecurity, which means that individuals have an anxious vision about the labour market (Cavanaugh & Noe, 1999; Davy, Kinicki, & Scheck, 1991; Sverke, Hellgren, & Naswall, 2002; Witte, 1999). The situation is perfectly described by a Dutch proverb: ‘A cat in the narrow makes strange jumps.’ It means that constricted situations lead to extraordinary behaviour. In times of organizational change, employees face stressful situations. Instead of accepting the process and experiencing a higher job insecurity, this paper researches desirable behaviour (flexibility and initiative) and characteristics (GMA) to overcome such situations in the best way. This specific situation could also be translated to unemployed people, who need money for i.e. livelihood. A stressful situation could occur when someone is unemployed for a longer period (Jasmer, 2015). In both situations, unemployed and employed, a high employability could lead to less job insecurity or more success in job application.
Due to the growing research on employability the last decades, what leads to a variety of contexts in which the word is used and ambiguity in its meaning, employability get the reputation of being “an attractive but confusing buzzword” (Thijssen et al., 2008, p. 167). Consequently, insights from different studies are hard to integrate into a comprehensive and internally consistent picture. This paper, with the meta-analysis as resource, contributes to the current literature on employability by finding accurate evidence to answer the question: ‘What is the effect of flexibility, initiative, and GMA on employability?’ According to Chatzisarantis and Stoica (2009), meta-analysis is a technique on which researchers can base very accurate conclusions, because results of various studies are compared and corrected on bias, so that they are usable for doing the meta-analysis.
The current labour market requires specific characteristics of people, namely: flexibility and individualization (Berntson, 2008). Flexibility, defined as “the capacity to adapt” (Golden & Powell, 2000, p. 273), is required, because the labour market is constantly changing. Individuals have to develop strategies to overcome uncertainty, since researchers suggested that these organizational changes have an impact on individuals (Ferrie, Shipley, Marmot, Stansfeld, & Smith, 1998; Hellgren, Sverke, & Isaksson, 1999; Morrell, Loan-Clarke, & Wilkinson, 2004; Shapiro & Kirkman, 1999). Proactive behaviour, also known as initiative, is developed as being an essential characteristic in the current labour market (Crant, 2000; Hall, 1996; Hall & Moss, 1998; Seibert, Kraimer, & Crant, 2001). The need for
proactive behaviour, also known as the protean career, where individuals were expected to find out their way through the labour market on their own (Berntson, 2008). Since proactive behaviour is increasingly relevant to success on the labour market, this paper researches initiative as an antecedent of employability. Furthermore, GMA is the third antecedent of employability investigated in this research. The term GMA stands for general mental ability (Deary et al., 1996). Since an individual’s independence is strongly required, individual’s GMA can distinguish between being recruited and staying unemployed (McQuaid & Lindsay, 2005). In other words, GMA is seen as the most important predictor of job performance.
To the best of our knowledge, there is just one meta-analysis conducted on employability. However, that research was specifically oriented on the relationship between employability and career success (Jasmer, 2015). Jasmer (2015) suggested a positive relationship between employability and career success. Thus, the positive effect on career success is an important reason to execute the meta-analysis on employability.
Next to the positive effect on career success, a higher employability also increases the chance to get employed, thus it has beneficial consequences (Berntson & Marklund, 2007; De Cuyper, Mauno, Kinnunen, & Mäkikangas, 2011; Kinnunen, Mäkikangas, Mauno, Siponen, & Nätti, 2011; Kirves, Cuyper, Kinnunen, & Nätti, 2011). However, the assumption that employability has beneficial consequences only applies to the individuals’ perspective. It is imaginable that organizations are not always that satisfied with high employability employees. For instance, because that type of employees is so popular, that they experience high demand to their services. So, as employer you are not assured that the employee is loyal to the company (Baruch, 2001; Berntson, Näswall, & Sverke, 2010). The different perspectives of employability are treated in the literature review.
As occlusion of the introduction, the structure of this study is described. First, the literature on employability is analysed. Some different perspectives are described and compared to each other. Thereafter, the antecedents of interest are described and reviewed. In that section, the hypotheses are developed in light of the prior literature. Next, the used meta-analytic method is described in detail. Also, potential biases are described and ways of minimizing them discussed. The collection of data is treated as well as the process from raw data into coding and finally into a clear SPSS-file. The last part of the method section described the analytical approach. Next, this paper discusses the results of the executed analyses. Thereafter comes the discussion part, which includes the implications and limitations. Finally, the conclusion part answers the research question and also the results are summarized in a few clear recommendations.
2. Literature review
2.1 Employability
Employability is the likelihood of securing employment (Berntson, 2008). Based on the results of Jasmer (2015), it is suggested that employability has a positive effect on career success. The literature also suggests that employability enhances the chance of getting employed (e.g. Berntson & Marklund, 2007; De Cuyper et al., 2011) So, for individuals, the following question arises: how to enhance the employability to take a better position at the labour market?
This paragraph introduces the various definitions of employability that have been used in the literature by working from an historical perspective towards the currently used definitions. Thereby, we shed a light on various perspectives of employability (including that of students, graduates, unemployed, employed, and disabled).
2.2 History
Due to the uncertain labour market, the last decades have shown growing academic interest in employability (Bhola & Dhanawade, 2013; Forrier & Sels, 2003; Harvey, 2001; McQuaid & Lindsay, 2005; Pool & Sewell, 2007; Rothwell & Arnold, 2007; van Dam, 2004). However, the first use of the concept was in the beginning of the 20th century, when the term was used in a more ‘dichotomous’ way, referring to the distinction between employable and unemployable individuals (Berntson, 2008). According to Berntson (2008) the concept was used to get a clear overview of persons that were able to work and persons that were not. In those years, persons with the right age (15-64 years old), good health, and stable family situation were considered able to work (Gazier, 1999). About a half century later, the concept was used in a broadened and more diversified way. Because of the more physically oriented definition of employability in those years, the term employability in the 1960s is also known as ‘socio-medical employability’ or ‘manpower policy employability’ (Berntson, 2008). The main focus in those years was to differentiate between types of individuals and their abilities (McQuaid & Lindsay, 2005). According to De Grip, Van Loo and Sanders (2004), in the 1970s the term employability became important to individuals, because of the growing unemployment. The term shifts back to a company level definition in the 1980s, when employability was defined as the own flexibility of employees (Jasmer, 2015). At the time,
Forrier and Sels (2003) defined employability as the optimization process of employees within organizations, the continual development of fit between employee and organization. In the mid-1980s, the last change occurred, when the concept of employability was broadened into the comprehensive, and there with the more ambiguous, definition of how it is used nowadays (Berntson, 2008). The level of employability is determined by how well an individual fits the requirements in the labour market, for instance based on human skills (e.g. (Forrier & Sels, 2003; Fugate, Kinicki, & Ashforth, 2004; McQuaid & Lindsay, 2005).
2.3 Defining employability
From the 1990s on, the term employability became more important for people, due to increasing organizational changes and the associated uncertainty. This paragraph discusses definitions of employability in the way it is used nowadays. The aim of meta-analysis is to collect as many relevant studies as possible, to be less sensitive to second order sampling, the remaining sampling error after first order meta-analysis (Hagger, 2006). Field (2001) recommends to include at least 30 studies in meta-analysis, although the minimum number of studies is two (Field, 2001). This meta-analysis therefore starts from a broad view on employability.
The interchangeable use of the terms employability, perceived employability and self-perceived employability is remarkable (Rothwell & Arnold, 2007). While comparing studies and the term choice, it is noticeable that the term employability is often used in dated studies (e.g. Baruch, 2001; van Dam, 2004; Fugate et al., 2004; Hartshorn & Sear, 2005; Harvey, 2001; Van der Heijde & Van Der Heijden, 2006; McArdle, Waters, Briscoe, & Hall, 2007) while recent studies more commonly refer to (self-) perceived employability (e.g. De Cuyper et al., 2011; Kinnunen et al., 2011, 2011; Kirves et al., 2011; Kirves, Kinnunen, & Cuyper, 2014; Vanhercke, Cuyper, Peeters, & Witte, 2014).
The most commonly used definition of employability was developed by Hillage and Pollard (1998). They treated employability as an individual’s journey through the labour market in order to realize potential through sustainable employment (Hillage & Pollard, 1998). In their opinion, individual employability depends on four components: (1) the individual’s assets, (2) the deployment of those assets, (3) how they present themselves and (4) the context in which they seek work.
More recently, Fugate et al. (2004) described employability as a person-centered phenomenon that expresses in effectively adaptability behaviour towards work-related
changes. Traditionally, employees’ careers were based on a limited number of jobs (Hall & Mirvis, 1995), while today, many employees are working in a variety of jobs, organizations and sectors during their career (Arthur, 1994; Arthur & Rousseau, 2001; Leana & Rousseau, 2000; Mirvis & Hall, 1994). Due to the changing requirements, individuals face a large number of job transitions (Ashforth, 2000). The person-centered approach to employability, as treated by Crant (2000), guides Fugate et al. (2004) towards their dimensions of employability in terms of (1) career identity, (2) personal adaptability and (3) social and human capital.
In contrast to Fugate et al. (2004), Van der Heijde and Van der Heijden (2006) defined employability from a more competence oriented view. Their definition of employability is as follows: “The continuous fulfilling, acquiring or creating of work through the optimal use of competences” (Van der Heijde & Van Der Heijden, 2006, p. 453). The study reviews a variety of literature on employability, and results in the identification of four key elements of employability: (1) the outcomes of employability are beneficial to both the individual’s career and the firm, (2) employability is advantageous for both an individual’s short-term performance and his or her long-term performance, (3) employability may also include personal elements besides adaptive behaviour and (4) the competence-based view on employability contains both specific and generic competences.
As a result of the comprehensive search for studies, many definitions of employability were found. In the table below (table 1), the definitions of employability are cited which are not treated in the previous paragraphs.
Author(s) Definition employability
(Lefresne, 1999, p. 465) “The probability, for a given group, at a given time, of finding a job or emerging from unemployment.”
(Forrier & Sels, 2003, p. 106) “An individual’s chance of a job in the internal and/or external labour market.” (Sanders & Grip, 2004, p. 76) “The capacity and the willingness to be and
to remain attractive in the labour market, by anticipating changes in tasks and work environment and reacting to these changes in a proactive way.”
(Rothwell & Arnold, 2007, p. 25) “The ability to keep the job one has or to get the job one desires.”
(Berntson, 2008, p. 15) “The individual’s perception of his or her possibilities of getting new, equal, or better employment.”
Table 1, Definitions employability
2.4 Perceived employability
Since Rothwell and Arnold (2007) used the terms ‘employability’, ‘perceived employability’ and ‘self-perceived employability’ interchangeably, all the terms are reviewed by literature to note the similarities and differences. In order to gather as much data as possible for purposes of executing a meta-analysis, we start this meta-analysis from a broad view on employability. In the following section, the literature on perceived employability as well as self-perceived employability is reviewed.
Kirves et al. (2014) wrote a clear overview regarding their conceptualization of perceived employability. They suggested that perceived employability is the individual’s belief about how easy it is to find a new job in a new organization. To clarify the definition, Kirves et al. (2014) highlighted a number of aspects. First, the definition treats the individual’s perception instead of reality and facts. Second, the definition is written from an employed person’s perspective. Against that thought, Hillage and Pollard (1998) argued that perceived employability is relevant in three phases of an individual’s career: after graduation and upon entry into the labour market, as an unemployed person applying for a job, and as an employee seeking a new job. The argument that Kirves et al. (2014) used to justify their focus on employed persons, is the contemporary turbulence and insecurity on the labour market, in the sense of the probability of losing one’s job. Finally, as the definition already suggested, Kirves et al. (2014) focused their definition on jobs outside the individual’s current organization instead of a new employment within the same organization.
The next insight in perceived employability is adopted from Wittekind, Raeder, and Grote (2010), who treated perceived employability as the individual’s perception. However, like the article of Rothwell and Arnold (2007), Wittekind et al. (2010) also interchangeably refer to employability and perceived employability. The main focus of the article is to explain the similarity of the definitions ‘employability’ and ‘perceived employability’. As mentioned, Wittekind et al. (2010) treated perceived employability as the individual’s perception. The employability literature is also mainly focused on the individual’s perspective. Exceptions are, for example, the context in which individuals seek work (Hillage & Pollard, 1998). Also the
contemporary state of the labour market is outside the individual’s control (Hillage & Pollard, 1998; Kirschenbaum & Mano-Negrin, 1999).
Another insight in perceived employability comes from Vanhercke et al. (2014). They based their definition of perceived employability on the study of Berntson and Marklund (2007): “Perceived employability is the individual’s perception of his or her possibilities of obtaining and maintaining employment” (Vanhercke et al., 2014, p. 593). While Berntson and Marklund (2007) only focused on new employment, Vanhercke et al. (2014) also includes the possibility to maintain current employment. Vanhercke et al. (2014) mentioned five important aspects regarding their definition of perceived employability. First, perceived employability is a subjective measurement, since it is the individual’s perception. It means that individuals with similar profiles may perceive their employability differently, for example because of their access to networks. Second, perceived employability refers to possibilities of employment (Berntson & Marklund, 2007). Possibilities integrates personal factors (e.g. curiosity and ambitiousness), structural factors (e.g. networks, career development, and the number of available jobs) (e.g. Eby, Butts, & Lockwood, 2003; Forrier & Sels, 2003; Griffeth, Steel, Allen, & Bryan, 2005; London, 1993; Ng, Eby, Sorensen, & Feldman, 2005) and their interactions (Vanhercke et al., 2014). Third, as mentioned above, the definition of Vanhercke et al. (2014) refers to both obtaining and maintaining employment. Vanhercke et al. (2014) reviewed perceived employability from three perspectives: from graduates’ and students’ perspective (Hillage & Pollard, 1998; Rothwell, Jewell, & Hardie, 2009), from the employed perspective (Berntson & Marklund, 2007; Rothwell & Arnold, 2007) and from the unemployed perspective (Wanberg, Zhu, & Van Hooft, 2010; Westaby & Braithwaite, 2003). As the fourth aspect, the definition of perceived employability refers to employment possibilities. Finally, the term ‘employment’ includes both quantity focus (e.g. the number of jobs available) and quality focus (e.g. the type of jobs available) (Vanhercke et al., 2014).
In their article, Vanhercke et al. (2014) also compared perceived employability to the competence-based approach (Van der Heijde & Van Der Heijden, 2006) and the dispositional approach (Fugate & Kinicki, 2008). The three approaches are similar in that they all take a subjective perspective. They also all include personal factors, structural factors, and their interaction. And third, the three approaches refer to both the internal labour market as well as the external labour market (Vanhercke et al., 2014).
One of the articles using self-perceived employability is the article of Rothwell and Arnold (2007) . However, in the introduction they developed their definition of employability: “the ability to keep the job one has or to get the job one desires” (Rothwell & Arnold, 2007, p.
25). Their use of self-perceived employability seems to be meaningless, in the sense that they did not remark on the difference between both terms.
2.5 Perspectives employability
Besides the different approaches of employability (perceived employability, self-perceived, and employability), studies also differ in the perspectives of employability (employed, unemployed, students, and disabled). This section gives an overview of the different perspectives of employability used in the literature, whereby the focus lies on the individual’s perspective instead of the organization’s perspective. The first treated perspective is the students’ and graduates’ perspective. This perspective could be divided into two phases: post-graduates (Hay & Hodgkinson, 2006; Morse, 2006; Rothwell et al., 2009) and post-graduates (Harvey, 2001). Literature on post-graduates’ employability generally focuses on the current study programme specifics and the acquired skills of the student (Rothwell et al., 2009), while graduates’ employability literature focuses on individuals’ ability to obtain a job (Harvey, 2001).
The unemployed, as the second perspective, are the people who are able to work but have not yet applied to a job. Wanberg et al. (2010) mentioned the importance of the job search process from both the practical as well as the theoretical perspective. Westaby and Braithwaite (2003) argued that enhancing self-efficacy is a key factor to get reemployed as soon as possible, since high self-efficacy is related to lower levels of stress (Gowan, Riordan, & Gatewood, 1999) and employment status (Saks & Ashforth, 1999). Proactivity and strong career identity are important variables according to McArdle et al. (2007).
Third, the employed are the persons who are working in an organization. Thereby, their employability is relevant in the sense of changing jobs within the organization as well as getting a job in another organization. For this group, career, income, family situation, organizational change, and work environment are seen as reasons to change jobs (Berntson & Marklund, 2007). Individual characteristics as, for instance adaptability assets (Fugate et al., 2004), attitudes (Kluytmans & Ott, 1999) and skills (Silla, Gracia, & Peiró, 2005), are of greatest interest to researchers (Berntson & Marklund, 2007).
Finally, the perspective of the disabled persons. The disabled group shows a high unemployment rate, due to bad job search, lack of work histories, and poor work habits (Adelman & Vogel, 1993; Johnson, Greenwood, & Schriner, 1988). Also, the disabled group differs from the employed, unemployed, and students in the sense of changeability between
perspectives. For example, changing from disabled to employed is very unlikely, while exchanges between employed and unemployed occur regularly.
2.6 Working towards the meta-analysis
After the extensive treatment and exposition of ‘employability’, it is time to shape the meta-analysis. The aim of the literature review was to get a clear overview of the different conceptualizations and approaches to the study of employability, the development of the term ‘employability’, and the various perspectives taken. Perceived employability and self-perceived employability are also reviewed by literature to shed a light on the similarities with employability and to justify their use as synonyms. For the purpose of this study, we focused on the perspective of the unemployed, the employed and the students or graduates. Since the disabled group shows a high unemployment rate which deviate from the other groups and the boundaries between unemployed, employed, and students are more permeable than that of disabled, studies on disabled people need a specialized approach and so they are excluded for this meta-analysis.
3. Hypotheses development
The previous section reviewed the employability construct on the basis of literature. While reviewing the approaches of employability, we noticed that individual’s flexibility, GMA, and initiative are acknowledged antecedents of individual’s employability. To formulate the hypotheses, first the antecedents are conceptualized and synonyms are provided (Appendix A). Second, the antecedents are described in specific terms and the relationship with employability is treated. Also the moderator hypotheses are supported by literature. The theoretical support for the moderator hypotheses is reported in a self-contained section. Finally, the hypotheses are formulated and the research model is visualized.
3.1 Flexibility
The next paragraphs are used to extensively describe flexibility, how it is used in the literature and at the end, the hypotheses are formulated. Van der Heijde and Van der Heijden (2006) defined flexibility as ‘person’s creative adaptability’. Acknowledged key-words in flexibility research are (perceived) mobility (Kirves et al., 2014; McQuaid & Lindsay, 2005), willingness (e.g. to move, to change workplace) (Forrier et al., 2009; Kluytmans & Ott, 1999;
Schyns, Torka, & Gössling, 2007; Wittekind et al., 2010), adaptability (Fugate et al., 2004), openness (Salgado, 1997; van Dam, 2004) and flexibility (Van der Heijde & Van Der Heijden, 2006; Van Der Heijden, Boon, Van Der Klink, & Meijs, 2009; Van der Heijden, 2002). Other synonyms and search terms associated with flexibility are mentioned in Appendix A. The aim is to clearly define flexibility in the way it will be used in the meta-analysis. So, literature on flexibility is treated as well as the relationship between flexibility and employability.
As Wittekind et al. (2010) suggested in their article, flexibility is the attitude towards changing situations. Especially in times of great insecurity, and so rapidly changing situations (Finn, 2000), individuals can enhance their chance to get or keep a job by broadening their opportunity search area. In other words, individuals need to behave flexibly towards chances on the labour market, for instance, they have to show readiness to educate themselves in additional skills (van Dam, 2004). Further, Schyns et al. (2007) highlighted the willingness of an individual to meet the needs of the labour market. In contrast to the ideas of this paper, Inkson, Gunz, Ganesh, and Roper (2012) argue that the willingness to move is not only about internal factors. The attractiveness of the specific job can also play a role in the willingness of individuals to be flexible. Related to employability, individual’s willingness to be mobile (Wittekind et al., 2010), willingness to move (Finn, 2000), willingness to change (Finn, 2000) and attitude towards changing situations (Wittekind et al., 2010) are all accepted definitions due to the overlap in definition.
In the literature, individual mobility is also related to employability. In the article of McQuaid and Lindsay (2005), geographical mobility is highlighted as one of the individual factors that influences employability. Another predictor of employability mentioned in this article is wage flexibility. Wage flexibility, as a form of mobility, is researched as a predictor of employability (Åberg, 2001; Bloemen & Stancanelli, 2001). Kirves et al. (2014) treats perceived mobility as an individual’s own perception to what extent he or she is mobile. Thereby, two types of mobility arise: based on situational circumstances and based on individual tendencies. In fact, all the variables underlying flexibility depend on individual circumstances (Hillage & Pollard, 1998). According to Griffeth et al. (2005) a positive relationship between perceived mobility and employability is expected.
Adaptability, as the next term in describing flexibility, shows great overlap with the term willingness. Adaptable people are defined as individuals who are willing and able to change personal factors and to meet specific situational demands (Fugate et al., 2004). In other words, adaptable persons behave flexibly in changing situations. Fugate et al. (2004)
suggested that five individual differences operate as pillars of personal adaptability: optimism, propensity to learn, openness, internal locus of control and generalized self-efficacy. A limitation of the study is the lack of results, due to its descriptive character. However, McArdle et al. (2007) tested adaptability in relation to employability and found a significant positive relationship.
While openness is mostly used as a personal characteristic, openness can also be seen as a dimension of flexibility (Salgado, 1997). Openness is similar to flexibility, because of the flexible behaviour towards changing situations that it entails. Meta-analyses have demonstrated the positive relationship between openness and training success (Barrick & Mount, 1991; Salgado, 1997), however, because of the lack of research on openness and employability, van Dam (2004) researched this relationship and hypothesized a positive relationship between openness and employability. The hypothesis of van Dam (2004) was supported by the measures. Despite the comprehensive character of this study, we only included openness in the sense of the behaviour to changing circumstances and openness to new experiences.
In this paper, flexibility is used as overarching term for the behaviour that people adapt changing situations. Personal flexibility is a well-researched topic by Van der Heijden and colleagues (Van der Heijde & Van Der Heijden, 2006; Van Der Heijden et al., 2009; Van der Heijden, 2002). Flexibility is defined as ‘person’s creative adaptability’, individuals were required to adapt to changes in both work as well as the labour market environment (Van der Heijde & Van Der Heijden, 2006). According to Boudreau et al. (2001) and Fugate et al. (2004), personal flexibility is an important antecedent of employability. Van der Heijde and Van der Heijden (2006) suggested the positive influence of high flexibility, because of the adaptive attitude towards changes. Unfortunately, they did not test the relationship between personal flexibility and employability. However, based on the literature (e.g. Boudreau et al., 2001; Fugate et al., 2004; Van der Heijde & Van Der Heijden, 2006), the relationship is hypothesized as being positive. As seen, the underlying variables of flexibility shows great overlap. The next part works towards the hypotheses on the relationship between flexibility and employability.
In the following, we will provide our arguments why perceived mobility, willingness, adaptability, openness, and flexibility may all be subsumed under a more general flexibility construct. With regard to perceived mobility, Griffeth et al. (2005) suggested a positive relationship between perceived mobility and employability. However, they did not find evidence for that hypothesis. In their longitudinal study, Kirves et al. (2014) tested the
influence of perceived mobility on employability at two moments. Their results have shown that the hypothesized positive relationship between perceived mobility and employability only occurs in T1(r = .120; p < .001), what means that the positive relationship between perceived mobility and employability is only tested significantly for permanent workers and not for temporary workers (T2). Next, Wittekind et al. (2010) researched individual’s willingness to develop new competencies as well as individual’s willingness to change jobs in relation to perceived employability. In line with the expectations, individual’s willingness to change jobs is supported as a predictor of perceived employability (r = .09; p < .01). Based on the analysis of Wittekind et al. (2010), willingness to develop new competencies is not significantly related to perceived employability. However, based on the literature, the expected relationship between willingness items (e.g. willingness to develop new competencies, - to change jobs, - to move) and employability is positive. With regard to adaptability, McArdle et al. (2007) researched proactive personality and boundaryless mindset as pillars of adaptability. For both proactive personality (r = .59; p < .05) as well as boundaryless mindset (r = .70; p < .05) positive relationships with employability are provided. Fugate et al. (2004) suggested a positive relationship between personal adaptability and employability too. For openness, the comprehensive article of van Dam (2004) is guiding. As hypothesized, the results of the analysis show a positive relationship between openness and employability orientation (r = .27; p < .001). Finally, flexibility as overarching topic is also researched in relation to employability. Besides the suggested positive relationship between flexibility and employability (Van der Heijde & Van Der Heijden, 2006), Clarke (2009) found evidence for the relationship by executing a qualitative study based on 20 semi-structured interviews with outplacement employees in Australia. So, based on the literature, we hypothesize a positive relationship between flexibility and employability.
Hypothesis 1: Flexibility is positively related to employability.
Hypothesis 1a: The positive relationship between flexibility and employability is
moderated by age, so that for lower ages, the positive relationship between flexibility and employability is stronger than for higher ages.
Hypothesis 1b: The positive relationship between flexibility and employability is
moderated by gender, so that for men, the positive relationship between flexibility and employability is stronger than for higher women.
Hypothesis 1c: The positive relationship between flexibility and employability is
moderated by employment status, so that for students, the positive relationship between flexibility and employability is stronger than for employees and unemployed.
3.2 Initiative
In contrast with flexibility, personal initiative is more specific in the definition. Thereby, the relationship between personal initiative and employability is a well-researched topic in management and leadership journals. Personal initiative is defined as: “a behaviour syndrome resulting in an individual’s taking an active and self-starting approach to work and going beyond what is formally required in a given job” (Frese, Kring, Soose, & Zempel, 1996, p. 38). The concept is divided into five aspects, where one of the aspects is self-starting and proactive behaviour. Therefore, proactive behaviour is seen as synonym of initiative for the purposes of this study.
The aspects, mentioned by Frese et al. (1996), characterizing personal initiative include: (1) consistency with organization’s mission, (2) long-term focus, (3) goal- and action oriented, (4) is persistent in the face of barriers and setbacks, and (5) self-starting and proactive behaviour. Individual’s initiative is all about undertaking action, and actions are guided by goals (Frese, Fay, Hilburger, Leng, & Tag, 1997). Thereby, initiative needs to be beneficial to the organization (Campbell, 2000), where taking initiative in the sense of stealing is not considered to be part of personal initiative. Initiative is intended to overcome problems which could re-occur in the future, so a method or system is needed to avoid the problem continuously (barriers and setbacks). Instead of waiting till the breakdown occurs again, initiative means that action is undertaken beforehand to avoid such undesirable repetitions. Action-oriented people quickly translate situations into goals and associated actions (Frese et al., 1997). Being self-started, proactive, and autonomous are the keywords in the definition of initiative by van Dam (2004). People with initiative try to improve their labour situation (Gamboa, Gracia, Ripoll, & Peiró, 2009), for instance by making their work more challenging, adjusted, and satisfactory (Frese et al., 1997). In summary, people with initiative are always aware of possible improvements, and instead of being reactive and following orders, they are self-starting and proactive. According to Campbell (2000), initiative and proactive behaviour are interchangeable. Proactive behaviour is the way an individual is acting, to rephrase it, how someone treats problems and opportunities. Initiative is something more situation specific, exemplified in the terms job-focused initiative and
structure-focused initiative (Campbell, 2000). Proactive persons take initiative and individuals with initiative show proactive behaviour.
That the interchangeable use of proactive behaviour and initiative is valid, is demonstrated by the definition of proactivity by Unsworth and Parker (2003). They mentioned that proactivity is behaviour which involve goal-directedness, persistence and long-term focus, largely similar to the definition of initiative (Frese et al., 1996). Proactive behaviour is also treated as an outcome of proactive personality. In this paper, the terms are seen as synonyms, because proactive personality is strongly associated with proactive behaviour (Thomas, Whitman, & Viswesvaran, 2010). In his study, Crant (2000) treated the differences between both perceptions. While proactive personality is seen as an individual’s disposition, proactive behaviour involves actual in-role and extra-role activities which engages in the work environment. Nevertheless, both terms are focused on ‘hands-on’ individuals, in other words, people who undertake self-started action. Proactive persons are also independent, they can change their circumstances directly and intentionally (Bateman & Crant, 1993). According to Thomas, Whitman and Viswesvaran (2010), proactive personality comes from individuals’ need to control and change actual circumstances. Because of the broadness of the term, a variety of behaviours and cognitions are related to proactivity (Seibert, Crant, & Kraimer, 1999). Therefore, Bateman and Crant (1993) developed the proactive personality scale. The scale measures proactive behaviour and since Bateman and Crant (1993) published their study, it became the most widely used measure of proactive behaviour (Thomas et al., 2010).
As already mentioned, the study of Frese et al. (1997) focused on initiative. As one of the hypotheses, they researched the relationship between initiative and employability. A positive relationship is suggested, because people high on initiative reacts better to changing situations, they have long-term goals and they are action oriented. Frese et al. (1997) concluded that people with high initiative (r = .28; p < .01) have a higher chance on employment than those with low initiative (r = -.47; p < .01). Van Dam (2004) also researched the relationship between initiative and employability. Based on 339 questionnaires, she reported a significant correlation between initiative and employability orientation (r = .24; p < .001). Based on these studies, it is likely that an individual’s initiative is positively related to individual’s employability. Potgieter, Coetzee, and Masenge (2012) researched personality attributes in relation to employability attributes. As one of the personality attributes, they treated proactivity. Their results suggested an important role for proactivity in predicting employability (r = .92; p < .001). Thereby, Kirschner, Caniëls, and
Bijker (2012) also suggested a positive relationship between proactivity and employability. To measure employability, they divided employability into five dimensions: professional expertise, anticipation and optimization, willingness to change, corporate sense, and balance. Their results showed significant correlations for professional expertise (r = .22; p < .01), anticipation and optimization (r = .27; p < .01), willingness to change (r = .31; p < .01), and corporate sense (r = .35; p < .01). In short, also for the relationship between proactive behaviour and employability, a positive relationship is expected.
Hypothesis 2: Initiative is positively related to employability.
Hypothesis 2a: The positive relationship between initiative and employability is
moderated by age, so that for lower ages, the positive relationship between initiative and employability is stronger than for higher ages.
Hypothesis 2b: The positive relationship between initiative and employability is
moderated by gender, so that for lower men, the positive relationship between initiative and employability is stronger than for higher women.
Hypothesis 2c: The positive relationship between initiative and employability is
moderated by employment status, so that for lower students, the positive relationship between initiative and employability is stronger than for employees and unemployed.
3.3 GMA
The third researched antecedent of employability is GMA, first described by Spearman (1904). In relation to employability, we expect that GMA plays an important role. GMA (also known as Spearman’s g factor) stands for general intelligence (Deary et al., 1996). Spearman (1904) suggested that GMA gives an overview of the general intelligence of a person, based on a battery of mental tests. In this section, GMA is explained and the development of the construct is described shortly. Next, this paper discusses intelligence, cognitive ability, g factor, and IQ as synonyms of GMA. Last, the relationship between GMA and employability is reviewed so that a hypothesis can be formulated.
Early in the twentieth century, Spearman developed a theory called the g factor (1904). By the evidence that different ability tests are highly correlated, Spearman (1904) suggested that high scores on a particular intelligence test match with high scores on other intelligence tests. The idea is illustrated in Figure 1, where the blue ovals are the different intelligence test scores which are all correlated with the g factor of that particular individual. Spearman
invented factor analysis, which enables researchers to calculate test loadings in the factor that is common to all the tests in the analysis (Jensen, 1998). In contrast to the first decades, the validity and generalizability of some tests is questioned, in view of the differentiation hypothesis (Deary & Pagliari, 1991; Detterman & Daniel, 1989; Spearman, 1927, 1929). However, so far there is no evidence that the IQ test has to be rejected from the g factor theory of Spearman (Deary et al., 1996).
Intelligence is the first underlying test of GMA described in this paper. According to Albus (1991), intelligence produces successful behaviour. It is not trainable, but it is seen as a result of natural selection. Emotional intelligence is the type of intelligence most researched as antecedent of employability in the contemporary scientific literature (e.g. Maynard, 2003; Pool & Qualter, 2012). Before making a distinction between intelligence and emotional intelligence, intelligence is explained a bit further. Intelligence is the behaviour to sense the environment, to make decisions, and to undertake action (Albus, 1991). Like mentioned earlier, intelligence is congenital behaviour. Higher intelligence may include the ability to react adequately to specific circumstances, and to represent and apply the available knowledge in the best way. Further, intelligence can be seen as the integration of knowledge and feedback in order to generate effective and purposeful action towards goals (Albus, 1991). Since emotional intelligence is a topic of higher interest in the contemporary scientific literature, emotional intelligence is also reviewed to highlight the differences with intelligence. Mayer and Geher (1996) paid attention to the difference between general intelligence and emotional intelligence. Where general intelligence is defined as the overall level of intellectual attainment and ability (Matarazzo, 1972; Ree & Earles, 1992), emotional intelligence is the capacity to recognize emotions (Mayer & Geher, 1996). Mayer, Caruso, and Salovey (1999) suggested that emotional intelligence is like traditional intelligence. They
based their conclusion on four arguments: (1) it could be measured as an ability, (2) emotional intelligence is a sizeable and measurable construct, (3) different defined emotional intelligence tasks were positively correlated with one general factor, and (4) emotional intelligence increases with age (Mayer et al., 1999). Since emotional intelligence is seen as a part of traditional intelligence (Schutte et al., 1998), there is no need to exclude emotional intelligence from the meta-analysis.
Secondly, cognitive ability is explained as part of GMA. Cognitive ability is defined as: “things that people are capable of doing” (Mayer & Massa, 2003, p. 833). In their article, Mayer and Massa (2003) related cognitive ability to the visualize-verbalizer hypothesis, the fact that some people are more textual / verbal minded and some people are more visual minded. By making a distinction between cognitive ability, cognitive style (i.e. thinking with words or images), and learning preferences (i.e. learning with text or images), the visualize-verbalizer dimension is conceptualized and the nature of the dimension is clarified.
Where GMA is used as overarching term of all the tests, the next paragraph describes GMA as individual test. In the literature, GMA is defined as the combination of two, three or more specific aptitudes that measures specific abilities (e.g. verbal, numerical, spatial) by a variety of items (Schmidt, 2002). When the letters GMA are decomposed and written, the word general already suggests a role as overall score. Thereby, the literature on GMA suggested that GMA plays an important role in predicting life outcomes, from criminal behaviour to the ability to take the public transport (Gottfredson, 1997; Lubinski & Humphreys, 1997; Schmidt & Hunter, 2004). GMA can be used interchangeably with the g factor, because Schmidt and Hunter (2004) suggested that the larger a specific measure loads on the general factor of mental ability, the stronger the correlation with the g factor. Other interesting findings from the literature are the stability of GMA over periods longer than 65 years (Deary, Whalley, Lemmon, Crawford, & Starr, 2000) and the genetic basis of GMA (Bouchard, 1998; McGue & Bouchard, 1998).
Finally, the intelligence quotient (IQ) is included in the GMA construct. IQ is a score to indicate human intelligence (Wang, Wee, Suk, Tang, & Shen, 2015). Wang et al. (2015) suggested that IQ test-scores are varying across different occasions, nevertheless the IQ test is seen as statistically valid (Shuttleworth-Edwards et al., 2004; Woodberry, Giuliano, & Seidman, 2008). A limitation of the IQ test is the questionnaire-based character, whereby it is (almost) not applicable to infants or young children (Wang et al., 2015). After all, intelligence, cognitive ability, g factor and IQ are used as synonyms for GMA. Therefore, one hypothesis is developed regarding the relationship between GMA and employability.
The relationship between GMA (and the underlying constructs) and employability is well researched, in particular the relationship between emotional intelligence and employability (e.g. Coetzee & Harry, 2014; Koen, Klehe, & Van Vianen, 2013; Ohme & Zacher, 2015). In a small cross-sectional research, Maynard (2003) found a significant relationship between the level of emotional intelligence and employability (r = .364; p < .001). Additionally, Coetzee and Harry (2014) also found a significant correlation between emotional intelligence and employability (r = .63; p < .001). However, because of the specific group of respondents, Native African female early career employees of three large African-based financial call centers, Coetzee and Harry mentioned that the results cannot be generalized to other ethnicities and ages. Besides the specific research on emotional intelligence, Raza and Carpenter (1987) investigated a model which also included the relationship between intelligence and employability. In total, they found a correlation of .66 (p < .001) between intelligence and employability. Their study made a distinction between male intelligence (r = .66; p < .001) and female intelligence (r = .65; p < .001) in relation to employability, but the results did not show a significant difference in gender. Other interesting correlations regarding the relationship between GMA and employability are found in mental ability researches. Ohme and Zacher (2015) researched mental ability by four items: learning, problem solving, information processing and reasoning. Results of the analysis shows that mental ability is significantly correlated with job performance (r = .58; p < .001). As a limitation, job performance is not seen as proxy of employability. On the other hand, this study result clarifies the direction and general effect of mental ability. The positive relationship between GMA and job performance is also highlighted by Judge, Higgins, Thoresen, and Barrick (1999). The study of Judge et al. (1999) goes one step further by hypothesize that GMA is positively related to extrinsic career success. In this study, extrinsic career success is defined as relatively objective and observable outcomes such as pay or ascendancy. The purpose to review this article is the same as for the study of Ohme and Zacher (2015); it creates an idea of GMA’s influence on other variables. Judge et al. (1999) found a significant positive correlation between GMA and extrinsic career success (r = .53; p < .01). Overall, we expect a positive relationship between GMA and employability.
Hypothesis 3: GMA is positively related to employability.
Hypothesis 3a: The positive relationship between GMA and employability is
moderated by age, so that for lower ages, the positive relationship between GMA and employability is stronger than for higher ages.
Hypothesis 3b: The positive relationship between GMA and employability is
moderated by gender, so that for lower men, the positive relationship between GMA and employability is stronger than for higher women.
Hypothesis 3c: The positive relationship between GMA and employability is
moderated by employment status, so that for lower students, the positive relationship between GMA and employability is stronger than for employees and unemployed.
3.4 Moderators
The hypothesized moderators came forth from the homogeneity analysis of the relationships between the hypothesized antecedents (flexibility, initiative and GMA) and employability. Based on the homogeneity analysis, it was assumed that there would be some moderators. Because of the lack of possible moderators in the coding sheet, age, gender, and employment status were selected as being a possible moderator. For every moderator, the literature is reviewed to suggest the direction of the hypothesis. The study of Van der Heijden, de Lange, Demerouti, & Van der Heijde (2009) suggested positive career outcomes for people younger than 40 years old compared to people above 40 years old. In that study, a clear difference occurred between self-rated employability and supervisor-rated employability. Where the effect of age on employability was positive for the self-rated results, the effect of age on employability was significantly negative (p < .001) based on the supervisor-rated results. These outcomes could be interpreted as, despite people will not admit, for employers a person became less interesting the older he or she is. As a footnote, the study of Van der Heijden et al. (2009) analysed the moderating effect of age on the relationship between career success and employability. For the purposes of this study, it is assumed that age is one of the moderators of the relationships between the hypothesized antecedents and employability, and that age has a negative effect on the hypothesized relationships.
Next, gender is also analysed as having an effect on employability (Van der Heijden et al., 2009). Huddy and Terkildsen (1993) shared that suggestion. They focused on the importance of showing ‘male qualities’ to get eligible for the top jobs in an organization. To getting a promotion, women candidates have to work harder than male candidates, so the suggested gender stereotyping is confirmed by the research of Huddy and Terkildsen (1993). Where Huddy and Terkildsen (1993) analysed gender as having a direct effect on the chance of gaining national and executive-level office jobs, for the purposes of this study, it is assumed that gender could possibly influence the relationships between the hypothesized
antecedents and employability. For the last hypothesized moderator, employment status, no theoretical support is found. Although, the study of Wright, Bushnik, Hammond, Novack, and Wood (2006) suggested higher employability ratings for employed and students. Despite the analysed relationship between employment status and employability, the study lacks on theoretical background and implications of the findings. In this study, the employability of students is hypothesized as having the most positive influence on the relationship between the hypothesized antecedents and employability. That suggestion is built on the suggested negative relationship between age and employability.
4. Methods
This study is developed as part of a comprehensive meta-analysis on employability. A project group of five master students (Angel Izekor, Gian Lee, Kirsten de Haas, and João Pedro) and two supervisors (Dr. Stefan Mol and Sofija Pajic) was composited. As a collaboration, we developed the theoretical framework of the comprehensive meta-analysis on employability, so that we could all end up with writing our own master thesis on some relationships of interest. Within the project group, decisions were made about the way of documentation, the conceptualization of employability, the conceptualization of antecedents, the conceptualization of consequences, inclusion criteria, the format of the coding sheet, and the way of analyzing the data. We decided on documentation in Google Drive, where we were all qualified to work on documents and add files.
4.1 Search strategy
In collaboration with the UvA library, in particular Janneke Staaks and Bjorn Witlox, a comprehensive search strategy was developed (Appendix B). The search strategy includes the search terms and the specific study requirements. Only the search strategy of PsycINFO is presented in Appendix B, but that one is the main search in this meta-analysis and can be seen as fundamental for the other electronic databases. Only electronic searches are used in the search for studies, because it resulted in a great amount of studies. Due to a lack of time to scan and code more articles, we agreed on working on the dataset of articles from electronic searches. Finally, the following electronic databases are screened: PsycINFO (March 21, 2016), Business Source Premier (March 21, 2016), ERIC (March 21, 2016), Medline (March 21, 2016), Web of Science (March 21, 2016), Sociological abstracts (March 21, 2016), and Scopus (March 21, 2016). The search in electronic databases resulted in 12764 articles which are distributed among the databases as listed in figure 3.
PsycINFO 2779 results
Business Source Premier 1619 results
ERIC 1179 results
Medline 649 results
Web of Science 2571 results
Sociological abstracts 622 results
Scopus 3345 results
Total 12764 results
Figure 3, Study search results per database
4.2 Study selection process
The first phase was the development of the search strategy, described in the former paragraph. As mentioned, the second phase of study selection was the first scanning round, where articles were removed based on certain inclusion criteria: publication year of the article > 1980, English language, correlation coefficients mentioned in the article, and mean age of the sample > 18 years old. This paper treats studies, finished after 1980, because from that decades on the term ‘employability’ was defined broadly and shows consistency with employability how it is used nowadays (e.g. Berntson, 2008; Gazier, 1999; Hillage & Pollard, 1998; Lefresne, 1999). To achieve an agreement on which articles to select and which articles
to reject, a trial of 50 articles was organized in the project team and the articles were discussed on their relevance and usefulness. After the first scanning round, we decided on stricter inclusion criteria: exclude criminal contexts and at least one correlation between an antecedent (or synonym) / outcome (or synonym) and employability (or proxy) is mentioned. By the end of the second selection round, there were 1201 articles left. Due to the experience of the project group in coding articles and the time limit to finish the theses, 240 randomly chosen articles were coded and included in the meta-analysis. A coding training was provided, and some unclear articles were discussed to achieve coding agreement. In the coding phase, there were still articles removed from the database due to a lack of relevance or correlations. Regarding the inter-rater reliability, the studies were coded with 10% overlap among coders. The reliability of the pilot coding was calculated using cross-correlation coefficient (Cohen's kappa) among pairs of raters. Inter-coder agreement in the pilot ranged from .80 to 1.00, indicating a relatively high level of agreement. The coders generally agreed about exclusion of the studies in which the disagreements occurred. Commonly, the disagreements referred to coding longitudinal studies, coding multiple effect sizes per study, or coding some of the dimensions or employability (e.g. protean career). The articles were coded on the basis of a coding sheet, developed in Excel. The content of the coding sheet, extracted from the articles, is described in the next paragraph. Finally, this specific meta-analytic study contains twelve coded studies. In figure 4, a flow diagram of the study selection process is illustrated.
4.3 Coding system
The coding sheet was meant as guidance to extract all the relevant information out of the articles. The content of the coding sheet can be found in Appendix C. First, the articles were coded on basic information. Basic information includes: Study ID/DOI, ISSN Number, Title, Author(s), Publication year, Publication language, Country of publication, Publication status, Publication type, Research design, and the Link to the article. Second, the sample was coded by: Actual sample size, Response rate, Mean sample age, standard deviation (SD) of sample age, % of males in the sample, Organizational context, Target population, Contract type, and Number of studies within overall study. Third, the employability dimension was coded. This part included: Employability definition, Employability dimension, Employability sub-dimension, Employability label, Employability measure, Measure reference, Number of items, Number of scale points, Reliability, Mean, and SD. As fourth, the outcome / consequences (Category of outcome construct, Subjective vs. objective outcome, Outcome sub-dimension, Outcome sub-sub dimension, Outcome label, Outcome measure, Measure reference, Number of items, Number of scale points, Reliability, Mean, SD, Correlation with employability, Significance, Specification of relationship with employability, and Hypothesized/non-hypothesized) and predictor / correlate (Category of predictor/correlate, Predictor sub-dimension, Predictor sub-sub dimension, Predictor label, Predictor measure, Measure reference, Number of items, number of scale points, Reliability, Mean, SD, Correlation with employability, Significance, Hypothesized/non-hypothesized, and specification of relationship with employability) were coded. Finally, the page number and the name of the person who coded the article were mentioned. When an articles described more than one study, we agreed to note it down as Study 1.1, Study 1.2 etc. Also every relationship / correlation was coded in its own row, which was noticed down as Study 1 R1, Study 1 R2 etc.
4.4 Publication bias
Publication bias is the phenomenon that studies with non-significant findings end up more often as unpublished study compared to studies which showed statistical significance (Rosenthal, 1995). It could lead to an overly optimistic picture of the relationship between two variables. For the purposes of this study, a correction on publication bias was not required, this paragraph functions as a note that the results could be overly optimistic.
4.5 Types of variables
In this study, the primary outcome variable is employability. Thereby, we listed some proxies of employability that were also accepted in the meta-analysis: career adaptability, human capital, social capital, and boundaryless career. The predictor variables are initiative, flexibility, and GMA. The moderator variables are age (mean age of sample), gender (% of males in the sample), and employment status (employed, unemployed or student).
4.6 Summary measures
For this meta-analysis, effect sizes were computed on the basis of sample size and correlation. To analyse the effect sizes of interest (initiative, flexibility, and GMA as antecedents of employability and age, gender, and employment status as moderators of the hypothesized relationships), the SPSS macros of Lipsey and Wilson (2001) were used.
4.7 Methods of analyses
First, homogeneity analyses were conducted to determine if the effect sizes were constant across studies (Lipsey & Wilson, 2001). When these analyses show significant results, it can be concluded that the sample of studies is heterogeneous and so moderator effects would be present. Second, the collected data is analysed by a fixed effects model and a random effects model. Where the random effect model is focused on using the total number of included studies in the meta-analysis, the fixed effect model is more participants oriented. In that model, the total number of participants are used to execute significance tests. The random effects model is better generalizable, the fixed effect model shows greater statistical power (Rosenthal, 1995). It depends on the heterogeneity of the sample of studies which model to choose. A heterogeneous sample of studies leads to the random effects model, a homogeneous sample of studies leads to the fixed effects model (Lipsey & Wilson, 2001). Since all the analysed hypotheses showed heterogeneity, the results were based on the random effects model. This is further elaborated in the analysis and results section.
4.8 Moderator analyses
A moderator analysis can be conducted when the sample of studies is heterogeneous. Because the homogeneity analyses of the hypothesized direct relationships showed that each meta-analysis is based on a heterogeneous sample of primary studies, a number of likely
moderators were examined. The following moderators will be analysed: age, gender and employment status. For age, the mean sample age is coded, for gender, the percentage of males in the sample is coded, and for employment status, the employment status of the sample is coded. In the analysis part, two moderator tests were applied, an ANOVA test on the moderating role of employment status (as categorical variable) and a regression test on the moderating role of age and gender (as continuous variables).
4.9 Preparation for analysis
Before the analyses were conducted, some corrections were applied and three separate databases were developed in SPSS. The first step was to filter the entire database (Excel-sheet with all the coded studies) on studies with correlations of interest. For the purposes of clearness, from that moment on, all the relevant lines (correlations) for a specific hypothesis were copied and thereafter merged into a separate Excel-file. The second issue in order to prepare the dataset for analysis was to correct for independent effect sizes (Geyskens, Krishnan, Steenkamp, & Cunha, 2008). Where Marín-Martínez and Sánchez-Meca (1999) were sceptic about the averaging method, they concluded that this method is sufficient for the purposes of meta-analysis. In Appendix D, the averaged effect sizes are given for independent effect sizes from the same study as well as the separate effect sizes of relationships within that study. The formula for averaging was adopted from Borenstein, Hedges, Higgins, and Rothstein (2009). The third step of preparation was the export from Excel to SPSS. Every hypothesis had a separate SPSS-file which reported Study ID, sample size, correlation with employability, reliability of the predictor, reliability of the employability measure, effect size, study weight, moderator gender, moderator age and moderator employment status. The effect size and study weight needed to be computed. To compute a correlation into an effect size, an online calculator tool was used (Wilson, 2001). The study weight (inverse variance weight) was easily computed by subtracting three from the sample size. As mentioned, in Appendix D, all the meta-analyzed articles per hypothesis are briefly summarized in a table.
5. Results
This section shows the outcomes of the analyses. Also the interpretation of the outcomes is discussed. The outcomes are displayed per hypothesized antecedent. First, the results of the hypothesized direct relationship are shown. Thereafter, the outcomes of the moderator analyses are discussed. For the ANOVA and regression tests, the ‘Full-information maximum
