The facilitating role of gender diversity in innovation is generally accepted, although empirically unresolved, as some researchers have found a link between gender diversity and innovation (PR Newswire US, 2013; Ruiz-Jiménez, Fuentes-Fuentes, & Ruiz-Arroyo, 2014), while others suggest that such findings are context specific (Parrotta, Pozzoli, & Pytlikova, 2014), ‘neither direct nor definitive’ (McMahon, 2010, p. 44). Others, meanwhile, report that groups of women ‘were no more innovative than all-male teams, nor were there any significant differences in the variety of alternative solutions’ they produced (Fila & Purzer, 2014, p. 1405). Some researchers argue that the ‘diversity can be either conducive or detrimental to team innovation’ (Mitchell & Boyle, 2015, p. 873), while Fernández (2015) and Sastre (2014) posit that an inverted-U relationship exist between gender diversity and innovation outputs, but did not find this under all circumstances. The prevalence of mixed findings from diversity studies implies that the business case for the benefits of diversity is not conclusive (Tatli, 2011; Wentling, 2004).

This research is not about gender diversity, but rather about a more contentious issue – that of the differences between men and women actors when engaging in innovation. Previous innovation research has focused on gender differences regarding characteristics and motivations, leadership style, strategic choices, obstacles and results (Pablo-Martí, García-Tabuenca, & Crespo-Espert, 2014). Sonfield, Lussier, Corman and McKinney (2001) state that results of prior research pertaining to gender and innovation are mixed, as is the case with gender diversity. Differences were found on some aspects, such as motivations and intentions, (Pablo-Martí et al., 2014; Sánchez-Escobedo, Díaz-Casero, Díaz-Aunión, & Hernández-Mogollón, 2014), overall satisfaction (Sonfield et al., 2001), commitment to product and service innovation (Pablo-Martí et al., 2014) and resilience (Caňizares & García, 2010). On other aspects, such as the strategies applied (Sonfield et al., 2001) and reasons for success and survival (Pablo-Martí et al., 2014), men and women reported substantially in the same manner. Lee and Marvel (2014) question past research findings which have reported gendered outcomes, and conclude that resource and context characteristics fully mediate the gender-innovation relationship. On the other hand, Kvidal and Ljunggren (2014) confidently report that gender is a non-issue in terms of innovation.

Important to this study is the measurement of innovation and the antecedents thereto. Assessment may play an important role in gender research, as gender is significant in how profiles are perceived (Sánchez-Escobedo, Díaz-Casero, Hernández-Mogollón, & Postigo-Jiménez, 2011). Alsos, Ljunggren and Hytti (2013) claim that, when analysing gender and innovation, it is possible to interpret innovation as a gender-biased phenomenon. When using a gender-aware operationalisation of innovation, no significant difference in innovativeness was found between men and women (Nählinder, Tillmar, & Wigren, 2015). Research on innovation should consider the gender neutrality of the operationalisation used in the study (Nählinder et al., 2015), and Alsos et al. (2013) state that it is imperative to develop and apply new methodological approaches, as well as new operationalisations of innovation and innovators.

This research will focus on such new methodologies. It will not attempt to produce gender-aware operationalisation versions of the standardised measures used in this study, but will rather seek assurance that the measures used are measurement invariant. Measurement invariance (MI) reflects the extent to which an observed score on a measurement is reflective of an individual’s standing on a construct, independent of their group membership (Mellenbergh, 1989; Meredith, 1993; Meredith & Millsap, 1992; Wu, Li, & Zumbo, 2007). Once MI is achieved, substantive gender-based comparisons resulting from the measurement should be done (Salzberger, Newton, & Ewing, 2014).

In this research, gender differences in the relationship between innovation and its antecedents will be assessed, making use of instruments tested to be measurement invariant. This type of research is rare in the organisational behaviour domain, and although gender is often included as a variable in studies (Eagly, Johannesen-Schmidt, & Van Engen, 2003; Reuvers, Van Engen, Vinkenburg, & Wilson-Evered, 2008), only a few researchers engage with the matter of MI (Xu, Wubbena, & Stewart, 2016) before conducting further analyses. In some studies, such as the study conducted by Steyn and De Bruin (2020), it was found that often used measurement instruments are not gender-neutral. The study aims to contribute to management science by specifying the importance of the gender of individual employees when facilitating innovation within the organisation and when applying managerial actions such as implementing the most suitable leadership style and human resources practices, and creating an organisational climate conducive to innovation.

Innovation in the workplace, which could be described as the propensity of an organisation to deviate from conventional industry practices by creating or adopting new products, processes or systems (Overstreet, Hanna, Byrd, Cegielski, & Hazen, 2013), is an essential component for competitiveness and survival (Gunday, Ulusoy, Kilic, & Alpkan, 2011), and considered by many scholars as one of the most important determinants of firm performance (Adegoke, Walumbwa, & Myers, 2012; Durán-Vázquez, Lorenzo-Valdés, & Moreno-Quezada, 2012; Grant, 2012). From a managerial perspective, it is important to identify the antecedents to innovation, differentiate between important and less important drivers of innovation, and manage these drivers in an effective manner (Bigliardi, 2013; Ndregjoni & Elmazi, 2012). According to Yen (2013), the facilitation of innovation is an essential management function of managers, as it is interconnected with organisational performance.

Research findings on the antecedents to innovation within organisations are readily available. Reports indicate that the nature of human resource management (HRM) practices (Sanz-Valle & Jiménez-Jiménez, 2018; Veenendaal, 2015) could predict innovation. Leadership (Atitumpong & Badir, 2017; De Jong & Den Hartog, 2007; Scott & Bruce, 1994) and specific leadership styles (Reuvers et al., 2008; Vargas, 2015) have also been linked to innovation. Along similar lines, organisational culture and climate has been linked to innovation (Baer & Frese, 2003; Goodale, Kuratko, Hornsby, & Covin, 2011; Hornsby, Kuratko, & Zahra, 2002; Lukes & Stephan, 2017; Sethibe & Steyn, 2016). More complex models, involving leadership style, climate and innovation (Sarros, Cooper, & Santora, 2008; Sethibe & Steyn, 2018), HRM, organisational culture and innovation (Al-Bahussin & Elgaraihy, 2013; Fellnhofer, 2018) have been tested, linking these variables.

Within the organisational behaviour context, ‘literature on diversity in organisations is limited’ and more specifically to this study, ‘even fewer studies investigate its impact on innovation’ (Díaz-García, González-Moreno, & Sáez-Martínez, 2013, p. 149). Although there seems to be consensus in the literature that certain antecedents drive innovation, the research on this relationship involving or including gender is limited (Eagly et al., 2003; Reuvers et al., 2008). This scarcity of research focusing on gender is widespread in academic publishing, but peculiar ‘given the centrality of gender in human life’ (Byrne, 2015, n.p.). Most of the research on the antecedents to innovation is void from an individual gender angle. This is worrisome, as gender is often associated with innovation in organisations (Nählinder et al., 2015). An exception to this may be the research of Reuvers et al. (2008) which reports a positive and significant relationship between transformational leadership and innovative work behaviour, and that the gender of the manager moderated the latter relationship, with employees being more innovative when the transformational leadership is displayed by men, compared to women managers. Mahto, McDowell, Kudlats, and Dunne (2018) found that the gender of the manager did not moderate antecedent-innovation relationships, nor did Pretorius, Millard, and Kruger (2005) find any gender moderation. In none of these cases was MI considered. Within the innovation domain, only the previously mentioned Xu et al. (2016) conducted MI analyses before comparing groups, which is the approach to be followed in this research.

In summarising the present literature on antecedents to innovation, it could be said that a multitude of studies specify these variables, and in many cases specify the importance of each variable relative to others. However, research focusing on the gender of those employees who are to be influenced to act innovatively is sparse. This sparsity of research linking antecedents to innovation and also including a gender angle – and specifically the dearth of gender-related MI in such studies – explains the length of this literature review, as well as the importance of conducting original research of this nature.

From a meta-theoretical perspective critical rationalism is at the core of this research, where the researcher starts off with a biased idea (men and women react in the same manner), and tests this idea to confirm or reject its truth (Higgs & Smith, 2006). The notion of the bias idea is that of relativism (Adamopoulos & Lonner, 1994), where correspondence in human experiences of the world is denied. Given this stance, even after finding an answer, true scientists will remain sceptical of the findings, and will guard against falsificationism (see Hung, 1997).

At a theoretical level, general systems theory is applicable, which emphasises ‘wholeness’ and systems which work as a unit of integrated parts (Von Bertalanffy, 1968), incorporative of input-throughput-output reasoning (Kast & Rosenzweig, 1972), where antecedents influence outcomes (see Wright & McMahan, 1992; Wright & Snell, 1998) and a feedback loop where outcomes in turn influence antecedents (see Kast & Rosenzweig, 1972; Von Bertalanffy, 1968). It is a deterministic model (Teece, 2018), as systems and subsystems inevitably respond to each other, and where any change in the system has a knock-on effect across the system (Dostal, Cloete, & Jaros, 2007). In this research such an interchange between gender, antecedents to innovation and innovative work behaviour is thus assumed, and workplace behaviour is perceived as a ‘network or system of sequential and interdependent decisions’ (Cascio & Aguinis, 2014, p. 43). The systems theory is preferred above behavioural or resource-based theories, as these archetypally describe closed systems and use simple linear processes to link inputs to outputs (Shin & Konrad, 2017).

A cross-sectional survey design was used to collect quantitative data. In this section, the population and sampling, the instruments for collecting data, as well as the way the data were analysed, are discussed.

In this study, the relationship between antecedents to innovation and IWB was assessed across gender. In the literature reviewed, many examples empirically linking these antecedents with IWB were found; however, instances of evidence of gender having a differential effect on the relationships are few. This study makes a significant contribution to the body of knowledge regarding the antecedents of IWB and gender, as it presents relational comparisons that are empirically sound, using only measures in which measurement invariance across gender was demonstrated prior to the analyses. In structural equation modelling terms, relations or paths were only tested once the measurement models were proven to be sound. Eliminating such bias before engaging in testing relationships is an aspect not often implemented and represents the primary contribution of this study.

The sample for this study was adequate for the analyses performed, presenting similar numbers of men and women, and resembling the employment statistics reported in the Quarterly Labour Force Survey (Statistics South Africa, 2019). The relatively large sample size necessitated the use of practical significance as indicator, as statistical significance often leads to unwarranted conclusions when using large samples (see Lin, Lucas, & Shmueli, 2013).

Of the seven instruments administered, only four showed MI across gender. Often-used instruments such as the UWES-9 (Schaufeli et al., 2006) and OCS (Allen & Meyer, 1990) were eliminated from the study, given the non-invariance across gender. In the end, only the MLQ (Avolio et al., 1995, 1999), HRPS (Nyawose, 2009; Steyn, 2012), BCEAI (Hornsby et al., 2002; Strydom, 2013) and the IWB (Kleysen & Street, 2001) were included in the study, as these instruments showed MI across gender.

Mean scores and mean score differences across gender were calculated. As these measurements were measurement invariant, the reported differences would present substantive differences (Bialosiewicz, Murphy, & Berry, 2013; Vandenberg & Lance, 2000) between men and women. Practically significant differences were detected in only three of the 18 mean scores calculated. The largest of these presented one-fifth of a standard deviation and related to IWB (see IWB in Table 1). As the IWB measure is invariant, it can be reported that men do report higher levels of innovative work behaviour. Men also experience more management support (see CEAIF1 in Table 1) and, in general, a more conducive environment to being innovative (see CEAIFTOTAL in Table 1). This may be an important finding, should the only difference be the perceived overall climate in the organisation and the level of managerial support men experience. Also important in this context is that on 15 of the 18 means reported, men and women did not differ in their reporting. Men and women thus experience the workplace in a very similar manner.

How the relationships between antecedents and IWB differed along gender lines was central to this research, rather than the mean differences between the genders. Data were thus primarily collected and analysed to answer questions such as the following:

Considering the most simple models (correlations), no gender differences were found in any of the 17 models tested. Thus, all three leadership styles, all seven HRM practices, and all five climate variables related similarly to IWB across gender (see Table 2). When using measurement invariant measures, the relationships between the variables were similar. It is interesting to note that being exposed to transformational leadership correlated the strongest with IWB for men, while in the case of women the strongest correlation was with being exposed to transactional leadership. This finding is remarkable as research focusing on men and women’s leadership styles indicates that men tend to be more transactional and women more transformational (Eagly et al., 2003; Xu et al., 2016). It would thus be fascinating to find out if men in leadership roles evoke IWB in women, and vice versa.

With regard to HRM practices, the fair management of the recruitment process (see HR5App in Table 2) related most strongly with IWB for both men and women. Considering climate, management support related strongly with IWB for men (see CEAIF1 in Table 2), and rewards and reinforcement for women (see CEAIF3 in Table 2). This finding regarding rewards and reinforcement (CEAIF3) and transactional leadership (for women), and management support (CEAIF3) and transformational leadership (for men), could be interpreted as complementing each other, given the nature of the different leadership styles.

The more complex regression models are presented in Table 3. In the leadership style model, and complementing the finding regarding the correlations, transactional leadership played a unique role in predicting IWB among women. With men, on the other hand, transformational leadership, and to a lesser extent transactional leadership, drove IWB. Training and development (see HRG2T&D), and fair management of the recruitment process (HR5App), contributed significantly and uniquely to IWB for both men and women. With regard to climate, the same variables were drivers of IWB, with the exception of organisational boundaries (CEAIF5), which was important to men only. Considering the complex model, HRM practices and climate, more than leadership styles, seem to be unique predictors of IWB. This model was also most predictive (RMen2 = 0.142; RWomen2 = 0.111; RDifference2 = 0.031), accounting for 14.2% and 11.1% of the variance in IWB.

In all the cases, the predictive power of the models (R²) was larger for men. Others (see Sánchez-Escobedo et al., 2014) report the same tendency in a similar setting, reporting that the explanatory power of models linking gender and innovation (entrepreneurial intentions) are more conclusive for men than for women.

Lastly, testing for moderation was performed (see Table 4). In this analysis, the focus is primarily on how the models improve (∆R²), given the introduction of gender and gender-interaction as variables. As can be observed from Table 4, the introduction of gender improved the models by 1.1% (for the first two models), by 0.5% for the next model and by 0.9% for the last model. When gender-interaction was added (Model 2, the moderation model), the models improved by less than 0.1%. Given this, it may be concluded that gender does not moderate any of these relationships. The picture is, however, a little more complex than this would suggest. In the second model, in which (only) gender is added to the regression, gender contributes uniquely and significantly to the variance in IWB. This indicates that gender is not a moderator, as suggested, but rather a predictor of IWB, and thus an antecedent to IWB.

Some reviewers may judge the absence of structural equation modelling in this research as a limitation. While future researchers may choose to take that route, regression modelling was applied in this instance as the authors are of the opinion that the technique is more explicit in detailing the moderation effects.

The complexity of the research was limited because the mediator variables (engagement, Schaufeli et al. 2006; and organisational commitment, Allen & Meyer 1990) were eliminated because they were not measurement invariant across gender. Future researchers are advised to include more such variables when collecting data, both so as to avoid this predicament, and to enable them to test more complex models.

In line with the present custom in South Africa in most formal organisational settings, this research report refers to its respondents as (self-identified) men and women. Contemporary gender identification, however, is more fluid than this might suggest and identification as lesbian, gay, bisexual or transgender (LGBT) can have substantive consequences in the workplace (see Badgett, Lau, Sears, & Ho, 2007; Grant et al., 2011), a factor which may also influence the relationship between innovation and its antecedents. Researchers are encouraged to engage in this complex matter.

A further limitation to the study relates to the sample investigated. Respondents form both public and private organisations reported on their innovative behaviour. As these organisations may fundamentally differ (for example in structure), this element influences the interaction between antecedents and outcomes. As such, future researchers are encouraged to split their samples along these lines and present their results focusing on one or the other.