the glass slipper effect - Spangsdorf-Staal

6 downloads 6 Views 1MB Size Report
pursue top managing positions and creates a perception that the occupational identity is impossible to .... Motivation f
THE GLASS SLIPPER EFFECT: The Influence of Occupational Identity by Association on the Motivation to Pursue a Top Managing Position among Women in Denmark

Sabrina Spangsdorf Liverpool University School of Psychology

Research paper May 2016

CONTENTS Resumé på dansk........................................................................................................................................ 4 1

Abstract .............................................................................................................................................. 7

2

Introduction........................................................................................................................................ 9

3

Literature review .............................................................................................................................. 13

4

5

6

3.1

Occupational identity ............................................................................................................... 13

3.2

Leadership identity ................................................................................................................... 15

3.3

Identity-based motivation ........................................................................................................ 18

3.4

Research question .................................................................................................................... 20

3.5

Aim and objectives ................................................................................................................... 20

Methods ........................................................................................................................................... 22 4.1 4.2

Participants............................................................................................................................... 22 Design ....................................................................................................................................... 22

4.3

Measuring identity ................................................................................................................... 23

4.4

The Bem Sex-Role Inventory .................................................................................................... 23

4.5

Career Aspiration Scale Revised ............................................................................................... 24

4.6

The questionnaire .................................................................................................................... 25

4.7

Data analysis guide ................................................................................................................... 25

Analysis ............................................................................................................................................. 28 5.1

Hypotheses testing ................................................................................................................... 28

5.2

Structural Equation Modelling (SEM) ....................................................................................... 30

5.3

Structural model ....................................................................................................................... 33

5.4

Mediated moderation .............................................................................................................. 35

5.6

Thematic analysis ..................................................................................................................... 36

Conclusion ........................................................................................................................................ 39 6.1

Perceived identity of a top managing position ........................................................................ 39

6.2

Job level of parents and partner .............................................................................................. 40

6.3

Occupational identity by association as a factor...................................................................... 41

6.4

Limitations and future studies ................................................................................................. 42

6.5

Conclusion and practical implications ...................................................................................... 44

References ................................................................................................................................................ 45 Appendix A ............................................................................................................................................... 54

2

Appendix B ............................................................................................................................................... 57

LIST OF FIGURES AND TABLES Table 1 - 25 BSRI-SE items for the perceived identity of a top managing position and for own identity ............... 28 Table 2 - Standardized regression weights – factor loadings of items and error variances ...................................... 31 Table 3 - Model fit ................................................................................................................................................... 32 Table 4 - Measurement properties for study constructs ........................................................................................... 33 Table 5 - Means, standard deviations and correlations between variables ............................................................... 33 Table 6 - Model fit for the hypothesized structural model and alternative models .................................................. 34 Table 7 - Moderated mediation with bootstrapping – Type of industry ................................................................... 36 Table 8 - Thematic analysis ..................................................................................................................................... 37 Figure 1- Hypothesized structural model ................................................................................................................. 27 Figure 2 - Hypothesized model ................................................................................................................................ 34 Figure 3 - Final structural model with OWNID (MAS and FEM) as mediating factor ........................................... 35

3

Resumé på dansk På verdensplan er der generelt få kvinder i topledelse (Grant Thornton International Business Rapport, 2015; Hausmann, Tyson & Zahidi, 2011). I Danmark udgør kvinderne kun 16% af bestyrelsesmedlemmerne og 14% af topleder stillingerne (Hausmann, Tyson & Zahidi, 2011). Både forskning og erfaring viser, at flere kvinder i den øverste ledelse har en positiv effekt på bundlinjen i forhold til indtjening, fastholdelse af kunder og produktudvikling (Armstrong et al, 2010; Credit Suisse, 2014). Danmark har derfor ikke råd til at forsømme eller overse halvdelen af sin talentmasse. Så hvorfor er kvinderne underrepræsenteret i de øverste ledelseslag i danske virksomheder?

Sociale, kulturelle og uddannelsesmæssige barrierer er velkendte faktorer. Men vi ved kun lidt om, hvilken rolle identitet spiller og muligheden for at kunne identificere sig med en bestemt jobfunktion eller rolle. Ashcraft (2013) hævder, at alle former for jobs og stillinger har sin egen identitet bestående af en række centrale kendetegn og egenskaber som associeres med det enkelte job eller rolle. Kendetegnene og egenskaberne er tæt forbundet med de sociale identiteter, som jobbets/stillingens eksisterende udøvere besidder. Det skaber jobs og erhverv, der synes at være "velegnet til visse mennesker og ikke-egnet for andre" (Ashcraft, 2013, s. 78); noget Ashcraft kalder ’occupational identity by association’. Denne rapport har undersøgt, om occupational identity by association er en faktor blandt danske kvinder, og som indirekte bidrager til en underrepræsentation af kvinder i den øverste ledelse. Dette er den første undersøgelse om occupational identity by association i Danmark, og det er også det første forsøg nogensinde på at studere occupational identity by association kvantitativt. Ashcraft bruger udtrykket ’Glass Slipper Effect’ i tråd med lignende udtryk, som bruges om kvinders udfordringer i ledelse, nemlig glass ceiling, glass cliff og glass eskalator. Glass Slipper Effekten er den situation, når kvinder ikke kan identificere sig med en toplederrolle eller som oplever et manglende match mellem sin egen identitet og den identitet som en toplederrolle har fået. Denne identitetskonflikt påvirker kvindernes motivation til at forfølge en toplederkarriere og skaber en opfattelse af, at det ikke er muligt at forene toplederidentiteten med ens egen identitet eller at en toplederposition slet ikke anses for at være en realistisk mulighed; altså at ”glasskoen” ikke passer.

4

Analysen i denne rapport er baseret på spørgeskemabesvarelser fra 1054 danske kvinder mellem 18 og 60 år. Analysen viser, at danske kvinder har en stærk maskulin identitetsopfattelse af en toplederrolle, men at deres egen identitet er langt mere forskelligartet og har et mere lige forhold mellem maskuline og feminine egenskaber. Resultatet af structurel equation modelling analyse viser et stor gab mellem kvindernes egen identitet og deres identitetsopfattelse af en toplederrolle, hvilket påvirker deres motivation til at forfølge en toplederkarriere; jo større gabet er, jo mindre motiverede er kvinderne. Analysen viser også, at kvinder i mandsdominerede brancher oplever en stor overensstemmelse mellem deres egen identitet og den identitet som tillægges en toplederrolle, hvorimod kvinder i kvindedominerede brancher oplever en lav overensstemmelse mellem deres egen identitet og en toplederrolle. Forældrenes jobniveau påvirker ikke motivationen hos kvinderne; til gengæld viser analysen, at hvis en kvinde har en partner, der bestrider en toplederfunktion, så har det en positiv indflydelse på kvindens motivation. Resultaterne viser samlet, at der er behov for langt mere forskelligartede rollemodeller med fokus på både maskuline og feminine træk og egenskaber og mindre fokus på køn som en faktor i sig selv. Resultaterne viser også, at der er behov for at reducere og nedbryde det meget kønsopdelte arbejdsmarked i Danmark, da det virker som en stærk barriere mod at få flere kvinder ind i topledelse.

Vigtigste resultater: 

Der er behov for at synliggøre flere forskelligartede rollemodeller indenfor topledelse med fokus på både maskuline og feminine træk og egenskaber og mindre fokus på køn som en selvstændig faktor. Især er det vigtigt at vise, at en toplederrolle består af flere facetter, herunder både maskuline og feminine egenskaber, da disse findes hos både mænd og kvinder og ikke er relateret til det biologiske køn.



Skoler, uddannelsesinstitutioner, store virksomheder, fagforeninger og ikke mindst de danske medier har en meget vigtig rolle i at fremme flere forskelligartede rollemodeller og ikke bare benytte en kvinde som rollemodel og tro at hun automatisk appellerer til andre kvinder. De fleste identificerer sig ikke nødvendigvis med kønnet hos en rollemodel, men i langt højere grad med de maskuline og feminine egenskaber, som personen udviser.

5



Det meget kønsopdelte arbejdsmarked i Danmark fungerer som en stærk barriere. Ved aktivt at arbejde for at reducere og udligne kønsopdelte brancher, kan man påvirke motivationen positivt hos kvinder til at forfølge en toplederkarriere.



Danske virksomheder bør give flere faktuelle oplysninger om det at være topleder ved at fortælle om de konkrete opgaver, der ligger i rollen, om antallet af arbejdstimer, om den fleksibilitet der ofte følger med en toplederrolle og den frihed som tillader, at man helt kan administrere sin egen tid. Det kan tilsammen være med til at afmystificere rollen og gøre den mere tilgængelig for kvinder, herunder at flere kvinder bedre vil kunne identificere sig med en toplederrolle.

6

1

Abstract

Top managing positions worldwide are occupied by relatively few women (Grant Thornton International Business Report, 2015; Hausmann, Tyson & Zahidi, 2011). In Denmark only 16% of board members and 14% of top managing positions are held by women (Hausmann, Tyson & Zahidi, 2011). Evidence shows that management diversity in general can have a positive effect on firm performance (Armstrong et al., 2010; Credit Suisse, 2014). Denmark simply cannot afford to neglect half of its talent pool. So why are women underrepresented in the highest corporate levels?

Social, cultural and educational barriers are well-known. But little is known about the role of identity and being able to identify with a top managing position and how it might be a piece of the puzzle. Ashcraft (2013) argues that occupational identities and central characteristics of an occupation are associated with the social identities of their practitioners. Such association creates occupations that appear to be “suited for certain people and implausible to others” (Ashcraft, 2013, p. 7-8) what Ashcraft terms ‘occupational identity by association’. This study examined whether occupational identity by association is a factor among Danish women contributing to the underrepresentation of women in senior management. This is the first study of occupational identity by association in Denmark and it is also the first attempt to study occupational identity by association quantitatively.

The ‘Glass Slipper Effect’ is when women are not able to match their own identity with their perceived identity of a top managing position. This conflict of identity affects their motivation to pursue top managing positions and creates a perception that the occupational identity is impossible to incorporate or the occupation is simply not even considered to be an option.

This study is based on survey responses from 1054 Danish women between 18 and 60 years. The analysis shows that Danish women have a strong masculine identity perception of a top managing position, but that their own identity is much more diverse and equal between masculine and feminine identity measures. Results of structural equation modelling suggested a gap between women’s own identity and their perceived identity of a top managing position which affects their motivation to pursue senior management; the greater the gap, the less

7

motivated. The analyses indicated that women in male-dominated industries experience lower incongruence between own identity and perceived identity of a senior role than women in female-dominated industries. The analyses also suggested that job level of parents do not affect motivation, but having a partner in senior management has a positive influence on women’s motivation. The findings indicate that there is a need for more diverse role models with focus on both masculine and feminine traits of a top managing position and less focus on gender. The findings also suggest that there is a need to reduce the highly gender-segregated labour market in Denmark as it acts as a strong barrier.

Practitioner points 

There is a need for more diverse role models with focus on both masculine and feminine traits of a top managing position and less focus on gender.



Schools, educational institutions, large organisations, trade unions and the media have a very important role in promoting more diverse role models and not just use a woman as role model thinking she automatically appeals to other women.



The highly gender-segregated labour market in Denmark acts as a barrier and by actively reducing the high level of gendered industries, more women are likely to pursue and be selected to senior positions.



Organisations should provide more factual information about a top managing role, about the tasks, the number of working hours, the flexibility and freedom to administer your own work schedule as it can make the role more obtainable for women.

8

2

Introduction

Top managing positions worldwide are occupied by relatively few women (Grant Thornton International Business Report, 2015; Hausmann, Tyson & Zahidi, 2011). In Denmark only 16% of board members and 14% of top managing positions are held by women (Hausmann, Tyson & Zahidi, 2011). Evidence shows that management diversity in general can have a positive effect on firm performance (Armstrong et al., 2010; Credit Suisse, 2014). Denmark simply cannot afford to neglect half of its talent pool. So why are women underrepresented in the highest corporate levels?

Social, cultural and educational barriers are well-known (Boatwright & Egidio, 2003; Haveman & Beresford, 2012; Smith, Smith & Verner, 2010). For example, although more women than men have secondary and tertiary education, especially university degrees (EU Commission Report, 2012; Hausmann, Tyson & Zahidi, 2011), the educational field is still highly gender-segregated with fewer women in the fields of Science, Technology, Engineering and Mathematics (STEM) (Bloksgaard, 2011; Holt et al, 2006). Previously, many top managers were recruited from finance and law, but today they are increasingly recruited from STEM fields (Haveman & Beresford, 2012), Holt et al, 2006), therefore the explanation for this gender-segregation may be located in occupational societal identities (Bloksgaard, 2011).

In Denmark women have one of the most equal employment rates in the world. Labour force participation for women sits at a rate of 75% whereas it is at 81% for men (OECD.Stats, 2015). However, Denmark has a very gender-segregated job market (European Commission’s Expert Group on Gender and Employment, 2009) with more women working in traditional female industries such as nursing and education, and men dominating the top managing positions in those occupations (Smith, Smith & Verner, 2010). Suggesting a perception of ‘masculine’ and ‘feminine’ professions exist, and that this perception influences men and women’s choices in the job market (Bloksgaard, 2011).

Women are often not part of the informal networks from which candidates to board and top managing positions are recruited (Linehan, 2001; Schwarts, Luycks & Vignoles, 2011; Timberlake, 2005) and women are often subconsciously discriminated against because they are

9

not perceived as having the requisite masculine characteristics that top management desire (Djøf, 2008). Two Danish studies (Ellersgaard & Larsen, 2010; Larsen, Ellersgaard, & Bernsen, 2015) have examined the elite in Denmark and found that they primarily came from a small number of families, that top positions are often ‘inherited’ from father to son more than from father to daughter and that the men in top positions are married to women from the same circles. Even though networks and being male increases the chances of landing a top managing position, it does not explain why it is primarily men and not women who end up in senior positions. Being male or being part of certain networks cannot independently explain why women are less represented.

The responsibilities related to family life and society in general have been naturally divided between men and women based on physiology. Childbearing abilities make women the primary caretaker and therefore they are traditionally more closely linked to domestic responsibilities (Eagly, Wood & Diekman, 2000; Myers & Twenge, 2013). This division of roles has spilled into the field of management where female behaviours are often categorized as mainly communal in characteristic (Eagly, Wood & Diekman, 2000). Male roles on the other hand are characterized through traits such as confidence and self-assertion as well as having a desire for achievement (Eagly, 1987; Madsen, Jensen, Madsen, Neergaard & Ulhøi, 2007). According to social role theory stereotypic gender roles control behaviour in two ways (Eagly, 1987). Stereotypic gender roles influence the actual behaviour as they drive the expectations that constitute the behaviour and they describe the social beliefs about how men and women should ideally behave. This influences the perception of women in management: if women express agentic characteristics they step out of their expected social role behaviour which is viewed negatively in society, but if they express communal characteristics expected of their gender role, they are viewed as less competent for the role of a top position in management (Ingols & Shapiro, 2015; Karelaia & Guillen, 2011). Thornton (2013) ranked the percentage of women in senior management in all countries, with Denmark only receiving a place as number 28. Countries such as China, Poland, Russia, Georgia and Botswana all have more women in senior management than Denmark. Many of the countries with more than 29% women in senior management score high on collectivism (Hofstede, 2015). Costa, Terracciano and McCrae (2001) report that gender differences are more strongly related to personality traits and are more marked in individualistic cultures. Collectivist behaviour is more determined by

10

social norms and social roles and social roles are related to functions rather than personality (Triandis, 1989). Social roles including gender roles in collectivistic cultures are related to tasks, but in individualistic cultures social roles and gender roles are related to identity.

The aforementioned barriers only address the opportunity for women to enter top managing positions and cannot explain entirely the underrepresentation of women in top management (Boatwright & Egidio, 2003; Savery, 1990). An important factor for wanting a certain position or job type requires a degree of desire and motivation (DeRue, Ashford & Cotton, 2009; Oyserman, 2009). Motivation for a job position or career path is established through identity; that a person is able to identify herself with that job role (Oyserman, 2009; Oyserman & Destin, 2010) and there are limited studies exploring theses processes among women in top managing positions (Schwarts, Luycks & Vignoles, 2011).

Occupational identity and identity-based motivation plays an important role when it comes to desiring top managing positions (Adamson, 2015; Ashcraft, 2013; Oyserman, 2009; Oyserman, 2013; Oyserman & Destin, 2010). Traditionally, occupational identity has been dominated by a unilateral view that people derive identity from their occupation (Phelan & Kinsella, 2009), but Ashcraft (2005, 2013) proposes that occupational identity has a reciprocal relation; that an occupation has an identity of its own derived from associated people, i.e. we associate a certain identity with an occupation based on values, traits, roles, gender, race etc. Such association creates occupations that appear to be “suited for certain people and implausible to others” (Ashcraft, 2013, p. 7-8), what Ashcraft terms ‘occupational identity by association’. Ashcraft (2013) suggests that if women cannot identify with the role of a top managing position, they lose interest in pursuing such careers or view them as impossible to achieve. Based on the phrase ‘if the shoe fits’, Ashcraft developed the term ‘the glass slipper effect’ in line with the glass analogy in women’s studies (glass ceiling, glass cliff and glass escalator). A specified library and online search for occupational identity by association revealed only two studies: one in the US among airline pilots (Ashcraft, 2013) and one in Russia among psychological counsellors (Adamson, 2015). Both studies found that people that enter a specific occupation are those that identify with the collective identity of that occupation. If a top managing position is viewed as an occupation and women are not able to identity with this occupation, it may explain why there is a lack of women in top management

11

and why organisations report that women sometimes decline promotions or simply do not apply for these types of positions in the first place (Desvaux, DevillardHoellinger & Meaney, 2008; Gino, Wilmuth & Brooks, 2015). The aim of this study is to examine the influence of occupational identity by association on the level of motivation to pursue a top managing position among women in Denmark.

12

3

Literature review

3.1

Occupational identity

The study of identity development can be traced from Freud’s early writings through Erikson’s eight stages of development, to Marcia’s empirical findings and operationalization (Schwartz, 2001). Identity, or one’s sense of self, is neither simple nor stable and individuals can hold several identities all related in one way or another (Schwarts, Luycks & Vignoles, 2011). There is no unified definition of what constitutes identity, but professional occupation appears to be a central element of identity and identity formation (Schwartz, 2001). Skorikov and Vondracek (1998) concluded that occupational identity played the leading role in the identity formation among adolescents, well ahead of identities within lifestyle, religion, politics and ideology. In Norway Danielsen et al. (2000) found that work was the primary influence on overall identity regardless of whether the participants had a job, were unemployed or attended college. Although Skorikov and Vondracek’s (1998) was a cross-sectional, instead of a longitudinal, and Danielsen et al’s (2000) participant numbers were small, both studies were conducted in a Western context and reflected earlier findings that occupation in an individualistic culture is related to identity (Triandis, 1989).

Christiansen (1999, 2000) was the first to link theories within occupation and identity focusing on how occupation influences and determines an individual’s personal and social identity. Unruh, Versnel and Kerr (2002) supported the approach of Christiansen by defining occupational identity as “the expression of the physical, affective, cognitive, and spiritual aspects of human nature, in an interaction with the institutional, social, cultural and political dimensions of the environment…” (Unruh, Versnel & Kerr, 2002, p. 12). Despite the fact that Christiansen includes the social context and interaction in the development of occupational identity and that Unruh also includes cultural and political aspects, they all view occupational identity as something that the individual controls. Their dominant discourse is centred round self-efficacy, self-confidence, self-esteem, personal motivators, and personal goals, all of which are viewed as something which the individual is aware of, makes cognitive reflections about, and actively seek to improve, change, adapt, or follow. In most studies occupational identity is defined as a mixture of one’s perception of occupational interests, abilities, goals and values based on one’s competencies and skills (Schwarts, Luycks & Vignoles, 2011). Occupational

13

identity has traditionally been viewed as a personal construct devoid of gender, age, physical appearance, ethnicity, nationality, religion and other characteristics. There is no question that the individual self and thoughts about identity are essential to the development of identity and occupational identity, however, occupational identity cannot be seen as separate from the environment or as arising from within individuals; “occupation is rarely, if ever individual in nature” (Dickie, Cutchin & Humphry, 2006, p. 83). Instead Dickie, et al. (2006) propose a social constructivist approach. Occupational identity is a reciprocal relationship between ‘person’, ‘occupation’ and ‘context’. This social constructivist approach sees occupational identity as a social product based on a reciprocal interplay between culture, language, social relationships, context, time and personality (Berger & Luckman, 1966). According to Burr (2003) each context has a limited number of discourses from which we can develop our own identity. For example, being a nurse at a hospital contains a certain number of discourses such as gender (‘most nurses are female’), the tasks (‘nursing is about caring’), the organization or institution (‘nurses are health care professionals’), or competences (‘nurses know about medicine’). Andsager (2015) showed that such discourses can be projected through the media reporting for example that the Danish media primarily presents women in leadership as unfeminine or as a ‘superwomen’.

Research on occupational segregation (Bolton & Muzio, 2008; Le Feuvre, 2009; Muzio & Tomlinson, 2012; Riska, 2008; Williams, 2013) demonstrates that gender plays a subtle yet strong role in occupations and how we relate to different types of occupations. Ashcraft (2013) argues that collective occupational identities and central characteristics of an occupation are associated with the social identities of their practitioners (whether they are actual or figurative) and are created on a societal level. Such association creates occupations that appear to be “suited for certain people and implausible to others” (Ashcraft, 2013, p. 7-8) what Ashcraft terms ‘occupational identity by association’. Ashcraft (2005) conducted a narrative study among US airline pilots based on 18 interviews and historical data and found that the pilots had chosen their occupation because they identified with several of the collective characteristics of the occupation, and once they were pilots they actively maintained those characteristics confirming their identity. Adamson (2015) found similar results in semi-structured interviews with 26 Russian psychology counsellors. Skinner (2014) used a grounded theory approach to examine whether executive coaching was useful to 11 women in senior roles and found that a core

14

element was that the women used the coaching to form a leadership identity. Ashcraft (2013) proposes the metaphor ‘glass slipper’ to express that occupations are associated with inherent characteristics that seem ‘natural’ to people but that at the same time is only fit for some but not for others. This ‘glass slipper’ approach explains why some occupations are viewed as predominately male and others female creating advantages for those who fit the ‘glass slipper’ and disadvantages for those who do not. The ‘glass slipper’ makes a person evaluate whether he or she fits the identity of an occupation and if the disparity between the occupation and the person’s existing identity is too large, the identity and thus the occupation is considered to be either unnatural, impossible to incorporate or is simply not even considered to be an option.

The evidence for the existence and relevance of occupational identity is rich and substantial, but there is a lack of studies that address occupational identity as more than work, abilities and skills as well as studies that address individual identity, perceived identity of a specific occupation and its effect on motivation. The studies from Ashcraft (2005), Adamson (2015) and Skinner (2014) are explorative and present a good foundation for the study of identity, gender and motivation in relation to specific occupations. Although, their sample numbers are rather small and there is a lack of quantitative representativeness. Compared with Scandinavian countries, which score very high on the femininity dimension, all three studies were conducted in cultures that score high on the cultural dimension of masculinity (Hofstede, 2015). Most research within occupational identity and specific occupations is based in an Anglo-American cultural context. There are a few exceptions such as Russia (Adamson, 2015), Spain (Killeen, Lopez-Zafra & Eagly, 2006) and India (Naqvi, 2011). Again however, this research is based in cultures that score high on masculinity (Hofstede, 2015) something that may well influence the social identities associated with different occupations.

3.2

Leadership identity

Among researchers there is debate whether leadership can be viewed as an occupation of its own. Barker (2010) states that as leadership is related to personal competencies and behaviour and as leadership transcends trades, it is not possible to view leadership as an independent occupation. Khurana (2007) argues that management and thus a leadership role requires skills, knowledge and training similar to other professions and it therefore must be considered to be an

15

occupation of its own. Considering the extensive number of studies on leadership roles, gendertyping of occupations and leadership identity, Khurana’s argument seems to be the strongest. In line with Ashcraft (2013) and Dickie et al.’s (2006) view on occupational identity, DeRue and Ashford (2010) believes that identity construction is a product of reciprocal influences between individual self, attributions, social environments and culture. DeRue and Ashford (2010) criticize the literature of not sufficiently explaining why some individuals internalize a specific occupational identity and others do not. By using leadership as example of an identity, they propose a model of identity construction consisting of three aspects aligning individual internalization, relational recognition and collective endorsement. A leadership identity is constructed through a claiming-granting process in which ‘claiming’ “refers to the actions people take to assert their identity as a leader” and granting refers to “the actions that a person takes to bestow a leader identity onto another person” (DeRue & Ashford, 2010, p. 631).

DeRue and Ashford (2010) have primarily been concerned with organizational settings and identity construction among adults in the workplace. However, identities develop throughout lifespan and some change with different ages or life events such as being a teenager or being a parent (Arnett, 2012). Komives et al (2005, 2006) and Lord, Hall and Halpin (2012) have found that leadership identity not only develops in early adulthood or when having a formal leader position, but that early construction of a leadership identity is often formed as early as in childhood. Experience then either boosts or diminishes early leadership construction through the ‘claiming-granting process’(DeRue & Ashford, 2010). Both Komives et al (2005, 2006) and Lord, Hall and Halpin (2012) state that parents, adults and later on peers, play an important role in the development of a leadership identity, from mentoring and role models to support and encouragement. The more a girl gains social acceptance from parents for taking a leader role (granting), the more likely she will be to feel comfortable with a leadership identity and actively pursuing that path as an adult (claiming) (Karelaia & Guillen, 2011).

Several studies support the development and construction of a leadership identity from a young age. Naqvi (2011) interviewed Indian women managers of which 60% stressed that their parents’ support, encouragement or job level influenced their decision to pursue a management position. Aycan (2004) studied the lack of women in top managing positions in Turkey finding

16

that cultural norms influence gender stereotypes but more importantly they influence women’s self-confidence and identity of self. Aycan’s (2004) suggestion is that support by family can increase self-confidence and the desire to seek management positions. A survey from the Danish union Djøf (2008) supports the findings of Aycan reporting that Danish women lack self-confidence and often lack support from their family, which ultimately stops them from pursuing top managing positions. Yewchuk, Äystö and Schlosser (2001) studied similarities and barriers experienced by women in Canada and Finland, finding that support and encouragement from parents and spouse was the most significant factor for women pursuing top positions. A study of 25,000 Harvard Business School graduates showed that even though women spend more time on household work and childcare compared to their partner, it is not the children who are holding women back from pursuing top managing positions (Ely, Stone & Ammerman, 2014), rather it is their male partner who expects that the couple will prioritize his career over hers. Ely et al (2014) call this second-generation gender bias that subtly and often invisibly influences the behavior of women through cultural assumptions which women internalize. The best way to avoid this is for women to find partners below their employment or educational level and who are less likely to prioritize their own career (Ely et al, 2014). Although the samples of Aycan (2004) and Naqvi (2011) are very small, and the sampling of Djøf (2008) and Yewchuk, Äystö and Schlosser (2001) cannot be easily generalized as they draw on a very selective group of women, these findings present some interesting aspects of the claiming process of leadership identity. Whilst the study by Ely et al (2014) obtained large numbers of US respondents, it is unclear to what extent those findings could be generalized into a Scandinavian context. In DeRue and Ashford’s (DeRue, Ashford & Cotton, 2009) construction model, claiming a leadership identity is not enough. In order to achieve a leadership identity, and a leadership position, other people must also grant you the role either formally, informally or both. Research shows that a leader role, and especially that of a top managing position, is primarily considered male and that agentic traits are required if you want to be a successful leader (Eagly, 2005; Madsen et al, 2007; Schein, 2001). In line with Ashcraft’s (2013) ‘glass slipper’ women are not considered to the same extend as men to be potential candidates to top managing positions and therefore not automatically granted a leadership identity. As social roles are linked to identity in individualistic cultures (Costa, Terracciano & McCrae, 2001; Triandis, 1989), the leadership

17

identity becomes even more about who leaders are – than what they do (Ely & Rhode, 2008; Ibarra, 1999). In a related line of study Cuadrado, Garcia-Ael and Molero (2015) examined gender-typing for managerial roles among 195 Spanish adults and found that masculine traits are valued much more than feminine traits in a successful leader role. A surprising finding was that male-typing of leader roles was stronger for female participants highlighting that women viewed a stronger incongruity between a female role and a leadership role. Similarly, this male-typing and leader roles are reported to exist in Denmark, indicating that it may be a crosscultural phenomenon (Madsen, Jensen, Madsen, Neergaard & Ulhøi, 2007). Both studies support the link between occupational identity and leadership in relation to gender-typing; although, the issue of motivation is only sparsely addressed and in the Danish study only managers participated limiting the generalizability of the study.

Most Danish analyses of women in management and sex-roles are conducted as surveys by trade unions or interest groups and not as part of scientific research (Bonke, 2015; Djøf, 2008; Institut for Konjunktur-Analyse, 2006; Rennison, 2012). Most studies on leadership identity focus on participants that are already managers but fail to include participants that are not managers, thus limiting the generalizability of the results and many studies did not use psychometrically standardized measures. Finally, occupational identity by association has not previously been addressed in Danish studies and we therefore do not know whether occupational identity by association might be a factor contributing to the fact that few women seek top managing positions.

3.3

Identity-based motivation

The identity-based motivation model (IBM theory) assumes that people interpret present and future situations in ways that are congruent with their current identities (Oyserman & Destin, 2010). Most people are able to imagine their future selves, but that does not mean they start acting in ways that will lead them to their future selves (Oyserman, 2015). According to the IBM model the reason for this is located in three aspects: 1) Psychological Relevance which means that the future identity must seem relevant for the current situation, 2) Readiness to Act which means that the strategies that can lead to the future identity must feel identity-congruent with the current identity, and 3) Interpretation of Experienced Difficulty which means that if actions, behaviours and goals feels identity-congruent, they are seen as obtainable and any

18

difficulties only highlights that the behaviour is important and meaningful. If the behaviour feels identity-incongruent it is seen as pointless and ‘not for people like me’ (Oyserman & Destin, 2010). IBM theory is not only concerned with existing identities but also encapsulates future and imagined identities as well as goals and attainment (Elmore & Oyserman, 2012). Elmore and Oyserman (2012) conducted an experiment on 146 sixth graders testing the relationship between gender-congruent identities and motivation and its influence on school identities and school effort. They found a significant effect that supports the model and that subtle cues linking gender and future success not only influences future goals and career aspirations but also current effort on academic tasks. However, the study only tested children and the results can therefore not be generalized to an adult population.

IBM theory can explain studies of stereotype threat, for example, where women are found to perform poorly in math and science tests if their gender is subtly brought to mind (Spencer, Steele & Quinn, 1999; Steele, Spencer & Aronson, 2002). Karelaia and Guillen (2011) examined psychological and motivational consequences of identity conflict between gender identity and leadership identity among 638 women in executive roles. The greater the identity conflict, the greater stress and the lower life satisfaction the women experienced. Conversely, the more leadership experience the women had, the more they identified with the leadership role and the more their perceived identity conflict was reduced.

Based on 311 senior-level students, Kreuzer (1992) concluded that women who rated themselves high on masculinity and low on femininity showed stronger leadership aspirations compared with male students. The results also indicated that mother’s level of education was related to leadership aspirations in women, however, Kreuzer notes that she did not use psychometrically standardized measures for her data collection. In a similar study Karami, Ismail and Sail (2007) used psychometrically valid scales in their examination of 100 Iranian postgraduates in a Malaysian university, and found that both female and male students desired a top management position, but the more masculine (for both genders) students rated themselves to be, the more they desired to be in top management. At the same time the female participants expected that society would not accept them in top management positions and had only an expectation of a middle or lower managerial role. These results once again demonstrate again the role that Hofstede’s (2015) masculinity cultural plays.

19

3.4

Research question

Studies on occupational identity have primarily been focused on women that already belong to a specific occupation and not the general population, and there are only two studies on occupational identity by association both of which are qualitative and studies on identity-based motivation have been conducted among children, but not adults. All studies on occupational identity and motivation have been conducted in cultures that score high on masculinity, but studies in cultures high on femininity may provide a different result. Many studies either have a small number of participants or they have chosen not to use standardized tests or measures making it more difficult to compare the findings. Based on the literature review the following research question will direct this study: Is occupational identity by association a factor in Danish women when considering a top managing position?

3.5

Aim and objectives

The aim of this study is to examine the influence of occupational identity by association on the level of motivation to pursue a top managing position among women in Denmark. As occupational identity by association can refer to many aspects such as gender, race, age, traits, status etc., this study will focus on the masculine and feminine traits associated with own identity and the perceived identity of a top managing position. The objectives are: 1. To determine women’s occupational identity perception of a top managing leadership role. 2. To examine differences and similarities between women’s occupational identity and that of a top managing position and how these differences and similarities affect the level of motivation to pursue a top managing position.

The purpose of this study is to examine the relationship between motivation, personal identity and perceived identity of a top management position. According to the literature review, own identity and the perceived identity of a top managing position are closely related and can equally affect the level of motivation, leading to the alternative directional hypothesis:

20

H1: Own identity and the perceived identity of a top managing position both have an interrelated mediating effect on the level of motivation to seek a top managing position.

The studies from Aycan (2004), Naqvi (2011) and Yewchuk, Äystö and Schlosser (2001) demonstrated that job level and support from parents play a significant role in the motivation and formation of a leadership identity for women. Two Danish studies showed that among the elite sons often follow in the footsteps of their fathers into top managing positions more than the daughters (Ellersgaard & Larsen, 2010; Larsen, Ellersgaard, & Bernsen, 2015). It is therefore relevant to examine the influence of parent’s job level: H2: Respondents with a mother or father who are in a managing position will be more motivated to seek a top managing position.

The Djøf (2008) analysis suggests that when receiving support from their partner, Danish women are more likely to pursue managing positions. Ely et al (2014) demonstrated strong support for the influence of the partner’s job level on women’s career choices. Therefore, a third alternative directional hypothesis focuses on the influence of the partner: H3: Respondents with a partner who are in a managing position will be less motivated to seek a top managing position. Finally, this study aims to investigate Danish women’s perception of a top managing position based on the Bem Sex-Role Inventory items. Several studies have shown that both men and women attach agentic traits to that of a top managing position (Ashcraft, 2013; Brenner, Tomkiewicz & Schein, 1989; Madsen, Jensen, Madsen, Neergaard, & Ulhøi, 2007). Similar results are expected here leading to the fourth hypothesis: H4: The respondents will score masculine BSRI items higher than feminine BSRI items for the perceived identity of a top managing position.

21

4

Methods

4.1

Participants

This study targeted adult Danish women between 18 and 60. The age limit was set at 60 as the retirement age in Denmark is 65-70 leaving a span of 5-10 years for career aspirations. 1308 people answered the questionnaire. Men, women over 60, those who did not agree to the consent form and incomplete questionnaires were removed leaving a total of n=1054. According to the Statistics Denmark (2015) there are about 1,600,000 women in Denmark between 18 and 60. In order to get a representative selection sample size was calculated. The minimum sample size for this study was 384 respondents. The total sample size for this study was n=1054 which gives a margin of error of only 3%. Compared with the data from Statistics Denmark the sample is representative of the Danish female population between 18 to 60 years (Statistics Denmark, 2015a). The demographic data of the participants can be viewed in Appendix A.

The estimated mean age is 39. There is a slight underrepresentation of participants between 18 to 25 (n=75) and 56 to 60 (n=66) compared with national data, but this is not considered to have any effect for the conclusions of this study as the majority of participants represent the working female population. There is a disproportionately large number of participants with a master’s degree compared to the general population (Statistics Denmark, 2015a). This may be due to the spread of the survey via the internet which may not be as widely used among women with vocational or short-cycle educational levels. The type of industry clearly reflects the very gender-segregated job market in Denmark with most women in the educational (n=132) and social/health care sectors (n=132) and public administration (n=313) (Statistics Denmark, 2015a).

4.2

Design

The research design was correlational utilizing cross-sectional survey methodology to assess the relationship between own identity and perceived identity of a top managing position and its influence on the level of motivation. For this study the sampling design was a mixture of convenience sampling and snowballing. The online survey was distributed via email, newsletter, intranet and social media to more than 6,000 potential female participants, and it

22

was estimated that a response rate of 8-10% would provide the required minimum of 387 participants. The advantages of using an online survey and using Internet based channels to reach the participants are that snowballing and convenience sampling can be more effective, it is easier to reach a larger area geographically and a broader scope of women. A typical disadvantage is whether the participants have access to the Internet. This was not viewed as an obstacle in Denmark as almost 100% of the population has access to the Internet from home or from their smart phone (Danmarks Statistik, 2014).

4.3

Measuring identity

For the purposes of this study, identity is defined according to social identity theory (Abdelal, Herrera, Johnston & McDermott, 2009) where traits are ascribed to in-groups and out-groups and a person will be conditioned to behave and internalize selected traits only because these traits describe his in-group. And to role theory (Eagly, Wood & Diekman, 2000) where people adopt the role into their own identity in order to feel role-congruent and thereby intensifies the role (Eagly, Wood & Diekman, 2000). Adopting this view, identity can be expressed in a set of traits or characteristics that are associated with people, situations and roles (Abdelal et al, 2009).

4.4

The Bem Sex-Role Inventory

BSRI applies a trait model to the measurement of identity. Bem (1974) developed the BSRI as a tool to measure gender role perceptions based on masculine and feminine traits. The strength of the BSRI is that masculinity and femininity are regarded as two independent dimensions rather than a continuum; this allows an individual to be high on both dimensions, low on both dimensions or high on one and low on the other (Bem, 1974). This also makes the BSRI useful for measuring the perception of a top managing role as it allows for a broader assessment than just being either masculine or feminine. The BSRI does not measure the actual sex of an occupation or a person, but the self-reported perception of whether an occupation or person is viewed as having traits that are more masculine or more feminine or both.

The validity of the inventory has been debated several times; Hoffman and Borders (2001) delivered strong criticism of the BSRI for lack of validity and reliability, particularly when it comes to measuring femininity and masculinity of different traits. Other researchers have

23

argued that the inventory is valid as long as it is used to test gender-typing (Cuadrado, GarciaAel & Molero, 2015; Gurman & Long, 1992; Holt & Ellis, 1998; Ingols & Shapiro, 2015). Most researchers have found that the BSRI is valid across cultures when a few test items are removed for cultural fit (Carver et al, 2013; Vafaei et al, 2014). Bem (1974) herself measured Cronbach’s α to be .86 for the masculinity scale for men and women combined and .80 for the femininity scale. Holt and Ellis (1998) found α to be .95 for the masculinity scale and .92 for the femininity scale. Persson’s (1999) adapted BSRI-SE showed a Cronbach’s α to be .82 for the masculinity scale and .81 for the femininity scale. In recent years the BSRI have been used to study which traits people associate with leadership roles and leader identities (Cuadrado, Garcia-Ael & Molero, 2015; Grinnell, 2002; Ingols & Shapiro, 2015).

The full version of the BSRI test consists of 60 items (Bem, 1974). The test was developed in an American context and is mostly used in its original language, English. Persson (1999) translated the 60 items into Swedish and conducted t-tests and a factor analysis based on 118 participants. He found that 25 items could be translated and used in a Swedish cultural context. The remaining 35 items were not similarly understood in a Swedish context. The study showed high reliability and psychometric properties in the revised and adapted BSRI-SE version, however, as Persson (1999) points out the BSRI-SE only accounts for 41.4% of total variance so further studies and greater samples are needed. Nevertheless, Persson’s study provides a useful tool for this study as not only has the BSRI items been translated into a language that is very close to Danish, the items have been tested in a culture that shares many similarities with the Danish culture. According to Hofstede (Daun, 1996; Hofstede, 1984) Denmark and Sweden have very equal scores on the dimensions of femininity, individualism and power distance making it possible to transfer the BSRI-SE to a Danish context. Participants responded to the BSRI-SE items using a 7-point Likert scale with endpoints ranging from 1 (never or almost never true) to 7 (always or almost always true). The BSRI-SE Likert scale was measured as intervals.

4.5

Career Aspiration Scale Revised

CAS-R has proven to be a valid and reliable tool when dealing with occupational aspirations and career desires (Gray & O’Brian, 2007; Gregor & O’Brian, 2015; Kim, 2014). Factor analyses have shown that the Achievement and Leadership aspiration subscales create reliable

24

results with Cronbach’s α between .72 to .82 (Gray & O’Brian, 2007) and a Cronbach’s α of .86 when using 5 statements instead of the original 8 statements (Kim, 2014). The CAS was originally developed to assess women’s career aspirations (O’Brian, 1996) and it has since been revised in order to improve its psychometric properties. The CAS-R covers three subscales: Achievement aspirations, Leadership aspirations and Educational aspiration (Gray & O’Brian, 2007; Kim, 2014). The CAS-R Leadership aspirations subscale consists of 8 items measured on a 5-point Likert scale. Gray and O’Brian (2007) and Kim (2014) found that 5 statements in particular account for high test-retest reliability. For statistical reasons the scale is treated as an interval similar to the BSRI-SE, which makes correlation between the scales possible. The CAS-R measures career aspirations to leadership positions in general. As this study focuses on top management levels, the 5 statements were adapted to reflect aspirations to top managing positions. Participants responded to the CAS-R statements using a 5-point Likert scale with endpoints ranging from 1 (strongly disagree) to 5 (strongly agree).

4.6

The questionnaire

The questionnaire consisted of six sections (see Appendix B). Section one included information about age, level of education, employment status, type of industry, marital status and number of children. Section two included the 25 BSRI-SE items measuring the identity of the respondent. Section three included the 5 statements from the CAS-R as well as two questions measuring the respondent’s interest in management positions in general. Section four included information about the level of education and employment status of parents and partner. Section five included the 25 BSRI-SE items measuring the perceived identity of a top managing position. Section six included an open question about the respondent’s opinion on the underrepresentation of women in senior management in Denmark. The respondents were asked to consider a top managing position and score the BSRI-SE items based on their perception. To make sure that the respondents understood what was meant by the expression ‘top managing position’, a definition was provided in the questionnaire.

4.7

Data analysis guide

Given the complexity of the analyses to follow, this section is provided to signpost the reader. 25

As the research question and H1 concern motivation and identity, these are variables than cannot be directly measured or observed, so-called latent variables. Instead they are measured by a number of indicators such as the CAS-R and BSRI-SE. In Structural Equation Modelling (SEM) it is possible to work with latent variables (Byrne, 2001). SEM also allows examination of the relationship between several dependent and independent variables, including moderating variables that may influence the dependent variables. SEM is therefore suited as statistical method for this study. The SEM analysis was based on the theoretical foundation of the literature review as to which factors may affect the variables identity and motivation. The SEM model consists of a measurement model and a structural model. The measurement model tests the measurement items of the latent variables and the structural model maps the relationships between the latent variables. Maximum Likelihood estimation was used to test the hypothesised model. The dependent variable for the hypothesised model was Motivation (MOT) (see Figure 1). Construct validity and construct reliability was measured using Average Variance Extracted, Construct Reliability (CR) and Cronbach’s α.

To examine H2 a two-way between-groups ANOVA was performed to examine the influence of parent’s job on the level of motivation to pursue a top managing position. For H3 independent t-test was used to determine if there was a statistically significant difference between the means in two unrelated groups: participants with a partner in a top managing position versus participants with a partner not in top management. Any differences between the two groups would show up as a difference between the mean levels of motivation. To examine H4 the participants were asked to rate the 25 BSRI-SE items according to their perception of a top managing position. Descriptive analysis would show how the participants have scored the masculine or feminine items.

Finally, a short thematic analysis was conducted to examine the open question in section six of the questionnaire. The results of the thematic analysis would show if identity is an expressed issue among the respondents.

26

Figure 1- Hypothesized structural model OWNID: TOPID: MOT: OWN1: TOP1: Q1: E: D:

Own identity Perceived identity of a top managing position Level of motivation Statement / item 1 (see Appendix B for full list of statements / items) Statement / item 1 (see Appendix B for full list of statements / items) Question 1 (see Appendix B for complete list of the five questions) Measurement error of measurable items Measurement error of latent variables

27

5

Analysis

5.1

Hypotheses testing

As predicted in H4 the mean score for the masculine items combined (M=5.98, SD=.64) is much higher than the mean score for the feminine items combined (M=4.23, SD=.98) when it comes to the perceived identity of a top managing position. The standard deviations are generally small among the masculine items compared to the feminine items indicating that the perception of a top managing position among Danish women is more or less uniform when it comes to the masculine traits (see Table 1). When it comes to scoring their own identity, the mean for the feminine traits are generally higher than for the perception of a top managing perception. The mean and standard deviation for the feminine (M=5.19, SD=.81) and masculine (M=5.13, SD=.86) traits is much more equal when scoring own identity than when scoring the perceived identity of a top managing position. This indicates that Danish women view themselves as having both masculine and feminine traits in equal measures which is in strong contrast to their very one-sided perception of a top managing position. Table 1 - 25 BSRI-SE items for the perceived identity of a top managing position and for own identity N=1054

Perceived identity of top managing position

Own identity

M

SD

M

SD

Self-reliant

6.17

1.02

5.19

1.23

Defends own beliefs

6.12

1.06

5.67

1.12

Assertive

6.34

.89

5.22

1.23

Strong personality

6.25

.99

5.65

1.67

Forceful

6.04

1.08

5.56

1.22

Has leadership abilities

6.51

.91

5.21

1.52

Willing to take risks

6.19

1.00

5.03

1.34

Makes decisions easily

5.85

1.12

5.07

1.42

Self-sufficient

5.61

1.35

5.25

1.35

Dominant

5.10

1.48

4.53

1.49

Masculine

4.33

1.59

3.26

1.54

Willing to take a stand

6.55

.82

5.85

1.10

Act as a leader

6.63

.80

5.17

1.57

Affectionate

3.79

1.34

5.58

1.22

Loyal

5.59

1.41

6.01

1.02

Masculine items

Feminine items

28

Feminine

3.61

1.33

5.06

1.32

Sympathetic

5.22

1.43

5.72

1.03

Sensitive to the needs of others

4.85

1.47

5.67

1.12

Understanding

3.80

1.43

4.56

1.27

Compassionate

4.39

1.41

5.64

1.15

Eager to sooth hurt feelings

3.19

1.44

4.54

1.43

Soft spoken

4.79

1.43

4.50

1.57

Warm

4.28

1.45

5.73

1.17

Tender

4.39

1.38

5.42

1.16

Gentle

2.89

1.34

3.84

1.50

To test the influence of having a mother or father in a top managing position on the level of motivation to pursue a top managing position, a two-way between-groups ANOVA was conducted. None of the variables had skewness or kutosis levels above one, which suggests normality. A 2x2 ANOVA with father’s job level (top job, no top job) and mother’s job level (top job, no top job) as between-groups factors revealed no main effect for father’s job level, F(1, 1050) = .26, p=.61, or mother’s job level, F(1,1050) = .30, p=.59. There was also no interaction effect, F(1, 1050) = .05, p=.83. This suggests that there is no statistically significant influence of either mother’s or father’s job level of the level of motivation to pursue a top managing position among Danish women. H2 can therefore be rejected. An independent t-test was performed for H3. Normality test showed that assumptions were met. The analysis showed that there was a statistically significant difference between participants with a partner in a top managing positions (M=3.30, SD=1.13) and participants with no partner or a partner not in a top managing position (M=2.89, SD=1.20), t(1052) 2.31, p=.01. Scatter and box plots show that the relationship between the two variables (partner’s job level and motivation score) is positive. This suggest that, contrary to the expected outcome in H3, having a partner who has a top managing position has a positive influence on Danish women’s motivation to pursue a top managing position. The null hypothesis can therefore be rejected. The effect size was found to be small with Cohen’s d=0.35 and Hedges’ g=0.34.

29

5.2

Structural Equation Modelling (SEM)

As the labour market in Denmark is highly gender-segregated (Bloksgaard, 2011; Statistics Denmark, 2015), type of industry may act as a moderating factor in the SEM analysis. A oneway ANOVA was conducted to test whether type of industry was significant. Type of industry consisted of 27 categories. These categories were coded into three groups: High masculine (industries with a majority of male employees), Medium masculine (industries with no gender majority) and Low masculine (industries with a majority of female employees). There was a statistically significant difference between the three groups in terms of the level of motivation scores, F(2,1051) = 33.28, p=000. A Tukey post-hoc test revealed that motivation scores were significantly lower among women in female dominated industries (M=2.65, SD=1.15) compared to women in male dominated industries (M=3.27, SD=1.16) or women in nongendered industries (M=3.23, SD=1.20). There was no statistically significant difference between High and Medium industry groups (p=.952). Effect size showed medium effect (ή2=.06).

Prior to estimating the structural model, Confirmatory Factor Analysis (CFA) was conducted to test the model fit of the measurement model. Five latent variables were included in the measurement model. Motivation (MOT) was represented by 5 items, Own identity was divided into masculine and feminine items with OWNID_MAS represented by 13 items, OWNID_FEM represented by 12 items. The perceived identity of a top managing position was equally divided into two latent variables of feminine and masculine variables with TOPID_MAS represented by 13 items and TOPID_FEM represented by 12 items. Assessment of normality showed that 3 items (Leadership abilities, Willing to take a stand, Act as a leader) had severe skewness. Due to the skewness of the data, Bollen-Stine bootstrapping was used in the CFA with 2000 bootstraps.

Convergent validity was examined to ensure that the items only loaded to their specific latent variable. Standardized regression weights showed that 3 items on the latent variable TOPID_MAS, 3 items on the variable TOPID_FEM, 1 item on the variable OWNID_MAS and 4 items on the variable OWNID_FEM were below cut-off point (see Table 2). These items were therefore deleted from the model. The Modification Indices suggested a number of error covariance paths to improve model fit. According to Harrington (2009) measurement error may

30

be caused by method effect such as self-reported methods using Likert scales and be the result of highly similar meaning or close to the meanings of words and phrases which is both the case for this study. Fontayne, Sarrazin and Famose (2000) found that there was high correlation between certain items when adapting the BSRI for use among French teenagers. BianchardFields, Suhrer-Roussel and Hetzog (1994) also found several items to be correlated, especially when perceived as personality traits rather than specific traits for masculinity and femininity, which is the case for this study. The suggested error covariance paths were therefore included in the model. The result was a good fit for the measurement model (see Table 3).

Table 2 - Standardized regression weights – factor loadings of items and error variances TOPID_MASCULINE Item TOPID_1 TOPID_3 TOPID_4 TOPID_6 TOPID_9 TOPID_10 TOPID_13 TOPID_16 TOPID_17 TOPID_19 TOPID_20 TOPID_23 TOPID_24

Factor loading 0,65 0,54 0,70 0,62 0,58 0,64 0,58 0,54 0,45 0,33 0,20 0,70 0,67

Error variance 0,43 0,29 0,49 0,39 0,34 0,41 0,34 0,29 0,20 0,11 0,04 0,49 0,45

TOPID_FEMININE TOPID_2 TOPID_5 TOPID_7 TOPID_8 TOPID_11 TOPID_12 TOPID_14 TOPID_15 TOPID_18 TOPID_21 TOPID_22 TOPID_25

0,73 0,48 0,72 0,87 0,87 0,80 0,73 0,64 0,50 0,78 0,64 0,21

0,53 0,23 0,51 0,76 0,75 0,65 0,53 0,41 0,24 0,61 0,41 0,04

OWNID_MASCULINE OWNID_24 OWNID_23

0,79 0,69

0,62 0,47

31

OWNID_20 OWNID_19 OWNID_17 OWNID_16 OWNID_13 OWNID_10 OWNID_9 OWNID_6 OWNID_4 OWNID_3 OWNID_1

0,24 0,58 0,51 0,59 0,56 0,77 0,54 0,69 0,73 0,56 0,69

0,06 0,34 0,26 0,35 0,32 0,59 0,29 0,47 0,53 0,32 0,48

OWNID_FEMININE OWNID_25 OWNID_22 OWNID_21 OWNID_18 OWNID_15 OWNID_14 OWNID_12 OWNID_11 OWNID_8 OWNID_7 OWNID_5 OWNID_2

0,33 0,59 0,72 0,44 0,48 0,59 0,81 0,82 0,83 0,69 0,35 0,71

0,11 0,34 0,52 0,19 0,23 0,34 0,65 0,67 0,69 0,47 0,13 0,51

MOTIVATION CAS_1 CAS_2 CAS_3 CAS_4 CAS_5

0,90 0,87 0,91 0,92 0,59

0,72 0,82 0,76 0,85 0,35

Table 3 - Model fit X

2

df

p

X / df

2

CFI

RMSEA

SRMR

Initial model fit

7804

1420

.000

5.49

.79

.06

.07

Model fit after Modification Indices

4997

892

.000

5.60

.84

.06

.06

Final model fit

3248

874

.000

3.71

.91

.05

.06

Model fit

32

To further test the robustness of the measurement model, we conducted internal reliability (Cronbach’s alpha), construct reliability (CR) and average variance extracted (AVE) analyses following Hair et al (2010). Overall, the results indicate sufficient construct reliability and discriminant validity of the five latent constructs (Table 4). Table 4 - Measurement properties for study constructs Constructs

Cronbach’s alpha

CR

AVE

.87 .92 .92 .89 .92

.91 .90 .92 .89 .83

.39 .57 .42 .52 .70

TOPID_MASCULINE TOPID_FEMININE OWNID_MASCULINE OWNID_FEMININE MOTIVATION

5.3

Structural model

The means, standard deviations and correlations for all study variables are reported in Table 5. The correlations indicate significant and positive relationships between all variables in the model (Figure 2), as expected.

Table 5 - Means, standard deviations and correlations between variables

TOPID_MAS TOPID_FEM OWNID_MAS OWNID_FEM MOTIVATION

M

SD

TOPID_MAS

TOPID_FEM

OWNID_MAS

OWNID_FEM

MOT

5.98 4.23 5.13 5.19 2.92

.64 .98 .86 .81 1.20

1 .19 .37 .25 .02

.04 1 .22 .28 .13

.14 .05 1 .10 .46

.06 .08 .01 1 -.07

.04 .02 .21 .01 1

N = 1054 Correlation significant p=