Gender-fair language and diversity labels in German job ads impacting job attraction: findings from two experimental studies Open Access

Organizations continuously face challenges including globalization, evolving customer needs, technological advancements and intense competition for finding qualified candidates (Beechler and Woodward, 2009). Attracting human capital is crucial for organizational success, making employer branding initiatives essential. Employer branding helps companies stand out by communicating unique benefits to potential employees. Employer branding research shows that perceptions of a company’s Corporate Social Responsibility (CSR) practices such as environmental efforts, community relations and diversity have the largest impact (Backhaus et al., 2002; Theurer et al., 2018) and influence its attractiveness to individuals seeking employment (Dögl and Holtbrügge, 2014). Embracing workplace diversity fosters a positive corporate culture where all employees feel valued and respected. Theurer et al. (2018) suggest that research should examine the relative significance of employer image attributes to accurately capture the branding dimension, along with ensuring consistency and clarity. Consumers are increasingly willing to support or oppose brands based on their principles regarding diversity, equity and inclusion practices (Ferraro et al., 2023). By ensuring fair employment practices and equal career opportunities regardless of an individual’s sexual orientation, gender identity or expression, organizations can build an inclusive workplace that promotes fairness, respect, innovation and productivity. Distinctive policies and certifications identifiable through seals or labels communicate the organization's values, enhancing its image to attract potential customers or job seekers. Incorporating CSR labels that represent diversity and equality in job ads can enhance the perception of an organization and positively influence consumers purchase intention, brand image and increase recruiting efforts. Moreover, employing gender-fair language may increase employee attractiveness and, therefore, enhance a higher number of job seekers who might not have considered applying to the organization.

Recruitment research has identified connections between recruitment activities, organizational image (such as CSR) and job seeker outcomes that influence employer attractiveness (Yu and Cable, 2012). Enhancing the employer brand through CSR activities is a complex matter, particularly when effectively communicating these initiatives in job advertisements. In consumer research, certification standards are often signaled by visible labels or seals (Jahn et al., 2005; Kleiss and Waiguny, 2021), which serve to reassure consumers of quality, compliance or ethical practices.

Attracting human capital can be challenging, making talent acquisition a critical factor for organizational success. Companies must enhance their attractiveness as employers by providing enticing compensation packages, fostering a positive work environment that includes diversity measures and offering ample career development opportunities. Employers need to initiate communicating what differentiates them from other employers through the inclusion of CSR initiatives to increase attractiveness and generate interest among potential employees (Puncheva-Michelotti et al., 2018; Turban and Cable, 2003).

By having an inclusion and diversity policy, organizations demonstrate their commitment to being a diverse and inclusive employer. Employers should incorporate a diversity statement in job advertisements or a link to the organization's diversity webpage so people can find out more about the organization's commitment. A lesbian, gay, bisexual, transgender, queer and intersexual, +additional identities (LGBTQI+)-friendly corporate climate is one of the most important criteria when LGBTQI+ individuals are selecting a future employer (de Vries et al., 2020). However, diversity also relates to the language used; in languages where nouns are related to gender, the form of a gender-fair language is essential to also display the CSR efforts of a company (Hodel et al., 2017; Sczesny et al., 2016).

The aim of this study is to investigate the relationship between CSR activities, especially related to diversity through different labels in job advertisements and using gender-fair language; hence, if employee attractiveness increases when the organization is shown to be a diversity-friendly company. While there is research on gender-fair language in job advertisements and linguistic gender-fairness in forms across cultures and branches, there is less research on the use of labels and gender-fair language to enhance employee attractiveness. Therefore, this paper contributes by investigating if gender-fair language and diversity labels, especially with the inclusion of the LGBTQI+ community, influence employee attractiveness when applied to job advertisements and how they interact. It provides an outlook on how organizations can strategically make inclusive job advertisements to foster diversity and inclusiveness in the workplace and how diversity-inclusive language and the presence of diversity labels affect the attitude towards the job as well as behavioral intentions to engage with the job advertiser and apply for the job.

CSR is a tool that aims to go beyond formal contracts and employment, and it adds value to employees' working lives. A crucial factor of CSR is the equal treatment of employees regardless of gender, age, race, religion or sexual orientation (Uddin et al., 2008). Having a diverse workforce can benefit the company and the working environment; employees from diverse backgrounds bring fresh views and diverse input that can result in advanced growth (KsiężaK and Fischbach, 2017). In the past years, sexual orientation and identity have received increasing importance in organizations and society. Numerous studies took place concerning the LGBTQI+ community in German-speaking countries. In a study from the Johannes Gutenberg University of Mainz in cooperation with the dating platform “Gayromeo” and the “Gay Happiness Index,” 21% of most gay men referred to their workplace as being “rather homophobic”, and 32% experienced real discrimination in their working life (Lemke et al., 2015). In the study commissioned by the work council Vienna (Weinwurm-Wilhelm, 2021) regarding sexual orientation and identity in the workplace, insights were gathered from over 1,200 LGBTQI+ employees; many chose not to disclose their identity at work due to potential job-related disadvantages; specifically, 44% expressed fear of gossip and insult, 34% preferred not to draw attention to themselves and 30% were concerned about jeopardizing their good relationships with colleagues, and many say their sexual orientation is a private matter, which deserves privacy (Schönherr and Zandonella, 2017). Although the European Commission Eurobarometer showed improvements in discrimination between 2007 and 2014 (Eurobarometer, 2015), in the study by Schönherr and Zandonella (2017), 30% of those questioned have experienced professional disadvantage including missing recognition, fewer opportunities for promotion, scheduling of work hours and not having vacation time approved.

Trans* genders seem to be discriminated against the most. More than 1/3 of the trans*- gender individuals did not receive a job based on their orientation, 26% lost their jobs and 41% quit their jobs on their own due to incriminating working situations (Schönherr and Zandonella, 2017). Studies consistently show that the LGBTQI+ community selects industries and organizations that include diversity management, where prohibitions of operational discrimination have been implemented (Köllen, 2010). LGBTQI+ people are an above-average group in the labor market, more likely to be self-employed and less likely to work as blue-collar workers. LGBTQI+ people are highly educated compared to the heterosexual population, which makes them an attractive employee (de Vries et al., 2020).

The rights of lesbian, gay, bisexual and transgender persons have been increasingly recognized and addressed through various developments. The European Union, the Council of Europe and the United Nations have reinforced standards on equality, making it illegal to discriminate against people based on their sexual orientation and identity, thereby ensuring equal treatment for all members of society. According to the EU-wide protection, it is illegal to discriminate in employment and occupation, which can be found in the Employment Equality Directive of the European Commission (EC) (Directive 2000/78/EC) and the Gender Equality Directive (Directive 2006/54/EC).

One challenge in recruiting is creating a superior value proposition to convince people to work for and stay within the organization. Employers must find ways of making jobs appealing, challenging and worthwhile to encourage highly qualified individuals to apply (Suff, 2006; Wilden et al., 2010).

Individuals establish their own identities within the context of social collectives, and these affiliations serve to safeguard groups based on shared attributes (Tajfel, 1978; Tajfel et al., 1979), leading to more favorable attitudes, trust and positive emotions towards groups with similarities (Ashforth and Mael, 1989). LGBTQI+ inclusive advertisements signal to customers that an organization values diversity and inclusivity, resulting in improved perceptions and intention among customers towards the organization (Cunningham and Melton, 2014). When organizations actively engage in LGBTQI+ supporting activities, they send signals to individuals that the company is LGBTQI+ friendly and may increase customer attraction (Ro and Khan, 2022) and perceptions and intentions to stay.

In many languages, linguistic asymmetries exist in reference to women and men. Previous research has analyzed numerous gender asymmetries that reflect and support the traditional gender hierarchy. Hellinger and Bußmann (2015) analyzed thirty languages from various language families including Arabic, Chinese, English, Finish, Hindi, Turkish and Swahili. A fundamental asymmetry lies in the masculine generics; in English, “he” can be used when gender is not relevant. In contrast, feminine forms do not work generically and refer to women only (Hellinger and Bußmann, 2015). In German, on the other hand, masculine role nouns are used for mixed groups of men and women. Past research showed that masculine generics evoke a male bias in representations (Stahlberg et al., 2011). German is considered a grammatical gender language, where personal nouns have a grammatical gender that normally corresponds to the gender of the person, compared to natural gender language like in English, where most personal nouns are gender neutral (Hellinger and Bußmann, 2015).

The wording in a job advertisement can also reflect “who” is invited to apply for the job. Gender-fair language aims at reducing gender stereotypes and discrimination, and several guidelines have been established to promote the use of inclusive language and avoid gender-specific forms such as “he” and “she.” The first guidelines for gender-fair German were already issued in the 1980s (Guentherodt et al., 1980).

Research has revealed that the utilization of gender-fair or gender-inclusive language has an impact on girls’ interest in various occupations. Specifically, employing gender-fair language in descriptions of stereotypically male-dominated jobs enhances the mental accessibility of female job holders and pushes girls’ interest in pursuing careers in these fields (Vervecken et al., 2013). Additionally, gender-biased language has particularly impacted women in the hiring process, especially when it comes to job advertisements. The presence of gender-specific references discouraged women from applying to roles traditionally perceived as male-dominated (Bem and Bem, 1973), and job advertisements using “he” instead of “she” decreased female USA students identification with a job and their pursuit to apply (Stout and Dasgupta, 2011). It was observed that women were considered less suitable than men when applying for leadership positions where the job was advertised with a masculine job title (Askehave and Zethsen, 2014; Horvath and Sczesny, 2015). Gender-fair language is more frequently used in countries with a high level of gender equality, e.g. Austria and/or German-speaking countries (Hodel et al., 2017). Legal requirements in the German-speaking countries require neutralization and feminization and/or a combination of linguistic forms used in job advertisements, such as m/f for male/female, in German m/w, männlich/weiblich (Hellinger and Bußmann, 2015), and recently, a diverse option (d) is added (G77/2018, 2018). A study in Austria shows that gender-fair job titles (70%) are more frequent than gender-specific (21%) (Hodel et al., 2017).

Yet, using male and female forms might not cover all gender identities. The gender-star * indicates that this acronym includes further orientations and gender identities (de Vries et al., 2020). Using the gender-star * (or other inclusive forms) should enhance a gender-neutral and inclusive gender-fair language (Schönherr and Zandonella, 2017). Ensuring wording policies and procedures are gender neutral (e.g. paternity and maternity) shared parental leave policies should be inclusive of same-gender partners and non-binary people to demonstrate inclusion is taken seriously by an employer (Ali, 2021). Table 1 shows the various forms of gender in the German language. The use of culturally sensitive language, respecting confidentiality concerns and educating oneself about the impacts of sexual minority stress are examples to help promote cultural competence (Moreno et al., 2020).

However, besides advancements in gender-fair language, recent developments make the use of language more of an ideology (Kotthoff, 2020). There is a group that argues that the German generic masculinum is not intended to prefer men, and therefore, it is just a development of the language. The more feminist-driven groups advocate for the feminization or a binary-oriented use of people-related terms and thus the male and a female form or using a capital I within the text used to represent females, whereas gender-inclusive groups that argue gender is an identity rather than biological advocate for gender-neutral language with an inclusive representation using asterisk, colon or underscores (Kotthoff, 2020). The following table shows examples for each in German.

Yet, more ideologists debate on gender-fair language in politics and media (Kotthoff, 2020) is happening, and we even find a range of discussions in the academic realm, e.g. see the discussion based on the opinions of Stöber (2021), Hanitzsch (2021) and Klimmt (2021). Especially the gender-* or similar approaches lead to heated discussions and reference to decreased legibility of the text (Jäckle, 2022; Pöschko and Prieler, 2018). Having a publisher’s pro or contra, while some use gender-inclusive language, some are still using more neutral forms, but generally the generic masculinum (Zylka and Grimberg, 2021). A very recent question of the day of the large panel provider YouGov in Germany revealed that 50% of the participants see gendering text as extremely unimportant and 19% as unimportant (YouGov, 2023). Another study by YouGov showed that also in internal communication (e.g. Liebes Team vs Liebe Kollegen/dear team vs dear colleagues (only male form)), only a small number of 14% support gendered or gender-neutral language (Zeit, 2021). A large study (Jäckle, 2022) found that the majority oppose the inclusive gendered language with the gender-star *. Despite a shift in the younger generation (Jäckle, 2022; YouGov, 2023), which is more accepting, we can still assume that generic forms or the binary gendered form will perform better than using the gender *.

Research has shown that potential employees compare the image of organizations with their own values, needs and personalities. If there is a match with one's own values, then the organization becomes attractive; therefore, it is important to create a positive image to increase employer attractiveness (Sivertzen et al., 2013) and willingness to apply for the job (Puncheva-Michelotti et al., 2018). Potential job seekers consider job advertisements that contain larger amounts of information more credible and attractive (Allen et al., 2004). Enhancing interest in a position is achievable through the inclusion of position-specific information, which can assist applicants in determining whether there exists a person-organization fit (Roberson et al., 2005). Yet, CSR is complicated to communicate in a short- and- specific information format, like a job description. Drawing on findings from consumer behavior research, the demonstration of the importance of certificates and labels based on standards is noticeable (Carrero and Valor, 2012; Jahn et al., 2005).

Signaling theory (Spence, 1978) is commonly used to explain how applicant attraction to a recruiting organization may or can be influenced by information or signals about organizational characteristics revealed during recruitment activities (Celani and Singh, 2011; Connelly et al., 2011). Applicants understand many recruitment-related activities as signals of unknown organizational characteristics (Collins and Han, 2004; Turban and Cable, 2003) and recruiter characteristics and/or behavior (Rynes, 1991). Signaling theory (Spence, 1978) suggests that when decision-makers are faced with uncertainty and incomplete information, they use the information they have as the basis for inferences about missing information. Signaling is the idea that one party conveys some information about itself to another party. It is useful for describing behavior when two parties (individuals or organizations) have access to different information (information asymmetry). The two parties could get around the problem of asymmetric information by having one party send a signal that would reveal some piece of relevant information to the other party. That party would then interpret the signal and adjust their purchasing behavior accordingly. A job applicant might engage in behaviors to reduce information asymmetry that hampers the selection ability of prospective employers. A perceivable action that is intended evolves to indicate an otherwise non-perceivable quality about the signalers’ environment (Connelly et al., 2011; Spence, 1978). The purpose of the signal is to indicate a certain quality or characteristic (Celani and Singh, 2011).

Signals gained through the inclusion of CSR information in recruitment advertisements taken from CSR labels can contribute to an applicant's ability to assess the match between the organization and the individual. Therefore, in this study, a variety of labels were used to send signals to prospective job seekers to determine whether CSR label(s) will lead to an increased attractiveness of an organization compared to a company with no label. Rynes (1991) suggested that recruitment activities serve as signals to applicants about unknown organizational attributes, and applicants may interpret a signal positively or negatively and generalize the assessment or other aspects of the organization.

However, if a label specifically emphasizes gender-inclusive (LGBTQI+ inclusive) attributes, perceiving such a signal in the job advertisement or company profile will set an expectation of language congruence with other attributes. Congruence (Mandler, 2014) is related to expectations and relevance (Heckler and Childers, 1992), influencing the perceived fit based on activated or pre-set expectations of the recipient. A range of studies in several different domains of marketing and communications found that incongruity led to a more negative evaluation (Belanche et al., 2021; Dens et al., 2008; Myers et al., 2014) as expectations are not met. Thus, when the company features the LGBTQI+ label, it will set relevance for the topic and an expectation that the other communication fits. Thus, using generic masculinum or the male–female-gendered language should not perform better than the inclusive-gendered language anymore if the LGBTQI+ label activates the expectations and therefore would make the texts incongruent.

To investigate our hypotheses, we have carried out two experimental studies following an experimental vignette study design (Aguinis and Bradley, 2014). Our focus of the studies is on attitudes towards the job advertiser and the intentions to apply. An experimental study can establish a causal relationship between the independent variables of gender-inclusive language and diversity labels, testing this relationship with high internal validity. A vignette study in our context can effectively mirror reality, as job advertisements are typically the primary source of information for prospective applicants during the initial stages of the job search.

Ethical approval was waived for these experiments by the Ethics Committee of the Province of Lower Austria (Ref: GS1-EK-12/763-2023), as no direct interaction between researchers and participants have been planned. Furthermore, the studies are fully anonymous. Participants have consented to the studies by first accepting it on the micro working platform and then through information provided on the study’s front page, which stated that completing the survey constitutes approval for the use of data. Additionally, non-completion or withdrawal due to discomfort would results in the exclusion of their data from the sample. Furthermore, we informed participants about the attention checks in the study that not passing them will lead to immediate screening out. For ethical reasons, the attention check was already implemented in one of the first set of questions to avoid participants spending too much time on the survey.

Design: To test our first two hypotheses, we employed a three-diversity language (generic masculinum vs binary gendered vs inclusive gendered) by two labels (LGBTQI+ label present vs LGBTQI+ label not present). We created a fictitious aviation service company (Aviazon) and showed a general profile with a text ( Appendix 1). The labels have been professionally designed by a graphic designer and were displayed on the left side. The text used the generic masculinum in German (e.g. Verkäufer/salesmen) or a binary gendered text (e.g. Verkäufer/salesmen or Verkäuferin/sales women) or an inclusive gendered text using an asterisk (Verkäufer*in salesperson). We asked respondents in the experiment whether they remember the labels. In the LBGTQI+ condition, 73.49% of the respondents could recall either the LGBTQI+ label or a diversity related label, while in the no-label conditions, 37% recall such labels, which yield a significant difference (χ²(df = 2) = 22.805, p < 0.001). For the gendered language, we assessed whether participants remember the text to be “gendered” or “addressing men and women” ^[1]. In the generic masculinum condition, (33.3%) significantly (χ²(df = 4) = 41.909, p < 0.001) less participants report the gendering of text compared to the binary gender text (55.8%) and the inclusive gendered text (73.1%).

Measures: Attitude towards the advertising company (ATT = 4.53, sd = 0.907, α = 0.919 and ω = 0.926) was measured on a six-point semantic differential with three items (bad-good, unappealing-appealing and unattractive-attractive) based on common brand attitude measures (Spears and Singh, 2004) as we want to assess employer branding. We adapted the items from the intentions to buy branded items (Spears and Singh, 2004) for our context resulting in four items (measured on a six-point scale: total disagreement to total agreement) describing the most common activities on job platforms (applying, searching for additional information, searching for job offers, saving the company in the favorites) to mirror intentions to engage with the job advertiser (INT = 4.10, sd = 1.220, α = 0.884 and ω = 0.887). Both dependent variables show satisfying values for Cronbach’s alpha as well as McDonald's omega (Wulff et al., 2023). To further check discriminant validity, we performed a factor analysis, confirming the two constructs and subsequently calculated the Fornell–Larcker criterion (Fornell and Larcker, 1981). For both measures, the square root of the Average Variance Extracted (AVE) was considerably larger than the correlation coefficient. ATT sqrt-AVE = 0.813 and INT sqrt-AVE = 0.884. The correlation between the two constructs is 0.598; thus, the two constructs are discriminant, as the square rooted AVEs are larger than the correlation.

Since the openness for diversity in workplaces is an important trait, we used the 15-item dealing with diversity in an organizational context/5 factors (DWD-O5) scale from Pietzonka and Kolb (Pietzonka and Kolb, 2021) to measure this variable. The responses rated a on a six-point scale (from total disagreement to total agreement), were collapsed into a composite indicator (DWD-O5 = 4.40, sd = 0.709, α = 0.838, ω = 0.826), which was subsequently entered as a covariate into the final model. Furthermore, we measured how rural the area the respondents were living in (see sample description in Table 2) and used this along with the age in years as additional covariates.

Procedure: German participants were invited on a micro working platform to complete the online experiment for a payment of 60cts. After a brief introduction, the participants got randomly assigned to one of the six experimental groups and the company profile of the fictitious company was displayed along with the labels. Following, respondents had to complete the attitude and intentions questions followed by the DWD-O5 scale. Manipulation checks followed the scales, and the experiment closed by asking some demographic questions. A screening question was included to assess whether participants were reading the instructions (Oppenheimer et al., 2009). Participants who did not pass this check were ineligible to proceed with completing the questionnaire.

Sample: 218 participants completed the online experiment. To improve data quality, the relative answering speed was calculated (Spence, 1978). Furthermore, as not all questions were mandatory, participants who left any question blank were excluded from the analysis. The final sample therefore consisted of 164 participants (50. % male, 48.2% female and 1.8% non-binary) with an average age of 36.90 years. Table 2 reports more details on the sample showing that more than 50% of the sample has a university degree but also that about 28% have vocational training. Furthermore, the distribution how rural or urban people live shows that about 22% live in rather rural conditions, about 43% in semi urban conditions and about 35% in urban-metropolitan areas. We checked for differences between the experimental groups for the demographic data, which yielded no significant differences.

Results: We calculated a MANCOVA to test our hypotheses with ATT and INT as dependent variables and DWD-O5, age and how urban people live as covariates. The omnibus tests of the MANCOVA show a significant effect of the language (Roy’s largest root F_(2, 15) = 3.304, p = 0.039 and ƞ² = 0.041) as well as a marginal significant interaction effect of the label with the language (Roy’s largest root F_{(2, 155)} = 2.620, p = 0.076 and ƞ² = 0.033). These results of the multiple test show that for some conditions significant differences to other conditions exist with weak to moderate effect sizes, which allows a further interpretation of the subsequent single tests.

For both dependent variables, the DWD-O5 scale had a considerable influence, while age and urbanity did not. The MANCOVA’s subsequent univariate tests for attitude did not find any significant main or interaction effects, rejecting H1a and H2a.

However, the univariate test showed that the overall MANCOVA effects are due to differences in the intentions, which show a significant (F_(2,164) = 3.304, p = 0.039 and ƞ² = 0.041) main effect for language. Single comparison (SPSS Bonferroni adjusted for multiple testing) of the conditional groups indicated that the inclusive gendered text as expected in H1b performed least (INT_inclgen = 3.75) compared to the generic masculinum (INT_masc = 4.22 and p = 0.09) and the binary gendered text (INT_bingen = 4.27 and p = 0.06).

Furthermore, a marginally significant interaction effect for INT was found (F_(2,164) = 2.604, p = 0.077 and ƞ² = 0.033). Contrast analyses (Bonferroni adjusted) revealed that while there was no significant difference between the language conditions if the LGBTQI+ label was present (p = 0.518), however, in case of the non-presence of the LGTBQI+ label, a significant difference was observed (F_(1,164) = 5.022, p = 0.008 and ƞ² = 0.061). Single comparisons (Bonferroni adjusted) for the three language conditions without displaying a label show that the masculinum text performs better than the inclusive gendered version (INT_masc = 4.25 vs INT_inclgen = 3.62 and p = 0.134); however, it is non-significant. The binary gendered version (INT_bingen = 4.63) performed significantly better than the inclusive gendered text (INT_inclgen = 3.62 and p = 0.006). In contrast, the difference for the binary gendered text between the LGBTQI+ label condition and the non-label condition is significant (F_(1,164) = 6.711 and p = 0.023). This leads to a confirmation of H2b that the presence of the LBTQI label does mitigate the negative effect of the inclusive gendered text, as in the non-present condition, we could clearly observe a negative effect of the inclusive gendered text. Figure 1 illustrates the relationship.

Results and discussion: Our data show that the negative effect of the inclusive gendered language can be mitigated by the label, as it is congruent with the label shown. However, given the weak effect size, it’s possible that this outcome might simply be due to the activation of associations. Thus, in the next study, we are investigating whether the label is a real signal with a positive effect, as it conveys a certain narrative or message or if only a congruence effect with any LGBTQI+ visual would happen. Thus, we propose the following hypothesis.

Design: To further investigate whether just any diversity-related information (e.g. just a rainbow symbol) or the label with accreditation numbers in particular affects the attractiveness of a prospective employer or hiring agency, we employed a three-label (none vs rainbow heart, LGBTQI+ label) vs three-language (generic masculinum, binary gendered and inclusive gendered).

As we saw in the first study that the dependent measures might also relate to the industry (in this case, airport handling company), we decided to create a fictitious recruitment processing organization (RPO: Hire Pro) with no indication for any industries. Based on the nine conditions, Instagram post vignettes were created, as Instagram is a popular advertising platform for job offers. Yet, we drastically reduced the text, as Instagram usually does not include large text elements (see the appendix for examples for Study 2).

Language was manipulated again in the German text through the masculine form, the binary gendered form and the inclusive gendered form. Using a manipulation check at the end of the study showed that the manipulation was successful, as significantly more participants stated that the text includes males and females in (masculinum: 24%, binary-gendered: 49%, inclusive gendered: 40%, χ² (df = 2) = 23.886 and p < 0.001). Another manipulation check asking directly whether the text was perceived as gendered also showed a significant difference (masculinum: 6%, binary-gendered: 22%, inclusive-gendered: 45%, χ² (df = 2) = 67.524 and p < 0.001).

The label was manipulated either by using no-label information, a rainbow heart sticker and the accredited label placed in the same position as the rainbow heart. We measured whether participants remembered any information about the workplaces that specifically catered to LGBTQI+ individuals in their recruitment ads. This revealed a significant difference between the conditional groups, demonstrating the success of the manipulation (none: 16%, rainbow visual: 33%, label: 50%, χ² (2) = 39.949 and p < 0.001).

Measures: As we transitioned from a direct organization hiring to a recruitment process outsourcing agency, we adapted the measures accordingly. The measures for the attitude remained the same, measured on a six-point semantic differential scale (ATT = 4.476, sd = 0.959, α = 0.910 and ω = 0.913). Since RPO and Instagram were used as platforms, we split the intention measures. Additionally, we adapted the intention measures to suit each platform. With a single item, we assessed the direct intention to apply if the RPO posts a relevant job (INT-APPLY = 4.20 and sd = 1.198) intentions towards finding further information were assessed with three items on a six-point scale (total disagreement to total agreement) (INT-INFO = 3.73, sd = 1.198, α = 0.865 and ω = 0.867). As we developed the dependent variables based on brand attitude and intention measures (Spears and Singh, 2004), we again checked for the discriminant validity. The square root of the AVE is for all three constructs and/or measures higher than the highest correlation (Fornell and Larcker, 1981)

Again we measured the DWD-O5 scale (Pietzonka and Kolb, 2021) and collapsed them as suggested into a composite indicator (DWD-O5 = 4.39, sd = 0.734, α = 0.870 and ω = 0.868) as well as how rural or urban people live.

Procedure: Participants were anonymously invited via a micro working platform for a payment of 40cts. After a brief instruction explaining that the study focuses on the attractiveness of job ads, participants were presented with one of the nine conditions. They answered the dependent variables, followed by the DWD-O5 and scales on social and economic conservatism as fillers and the demographics. A screening question was included to determine if participants were reading the instructions (Oppenheimer et al., 2009). Participants who did not pass this check were ineligible to proceed with completing the questionnaire.

Sample: A total of 585 participants from Germany successfully completed the study. As in Study 1, we calculated the average answering speeds (Leiner, 2019) and discarded participants who answered at a very high pace. Furthermore, as all questions were mandatory, participants who left any question blank were excluded from the analysis. The final sample therefore consisted of 475 participants (51.4% male, 47.4% female, 0.6% non-binary and 0.6% no answer) with an average age of 36.69 years and sd = 10.320. The following table shows the educational spread as well as how rural participants live. In summary, the sample corresponds largely with the distributions of study one, which makes them comparable. We checked for differences in age, gender, education and living area and did not find significant differences between these groups (see Table 3).

Results: To test our hypotheses, a 3×3 MANCOVA was conducted with attitudes and intentions as the dependent variables, language and label as independent variables and DWD-O5, age and urban living conditions as covariates. The omnibus tests of the MANCOVA show a significant effect of the language (Roy’s largest root F_{(3, 475)} = 3.270, p = 0.021 and ƞ² = 0.021) as well as an interaction effect of the label with the language (Roy’s largest root F_{(4, 475)} = 3.322, p = 0.011 and ƞ² = 0.028). These results show that for some conditions, significant differences to other conditions exist.

Investigating the single effects for the language, we did not find a significant difference on the between-subject tests, albeit a significant omnibus test. This is likely due to the corrections for multiple testing.

Investigating the interaction effect, we found a significant interaction for intention to apply (F_(4,463) = 2.833, p = 0.024 and ƞ² = 0.024), while the other two dependent variables show a similar pattern like in Figure 2 but were not significant. The calculated contrast analyses (Bonferroni adjusted) show that the differences are in the masculine-gendered language conditions (F_(2,475) = 5.525 p = 0.004 and ƞ² = 0.023). There were no significant differences in application intentions between the conditions with no label and those with a label (INT-APPLY_none = 4.15 vs. INT-APPLY_label = 4.47, diff = 0.321 and p = 0.385), the difference between the no-label masculine-generic condition and the rainbow visual masculine generic condition was marginally significant (INT-APPLY_none = 4.15, INT-APPLY_rainbow = 3.80, diff = 0.351 and p = 0.276). The largest observed differences were between the rainbow visual and the label in the intentions to apply (INT-APPLY_rainbow = 3.80, INT-APPLY_label = 4.47, diff = 0.671 and p = 0.003). Figure 2 summarizes these differences.

Contrasting (Bonferroni adjusted) on the level of the label, we observe significant interaction effects for the rainbow visual (F_(4,475) = 3.838, p = 0.022 and ƞ² = 0.016). The single comparisons revealed differences for the masculinum language in comparison to the binary-gendered language (INT-APPLY_masculinum = 3.80 vs. INT-APPLY_binary = 4.29, diff = 0.489 and p = 0.024) as well as between the masculinum and the inclusive gendered language (INT-APPLY_masculinum = 3.80 vs. INT-APPLY_inclusive = 4.31, diff = 0.513 and p = 0.016).

In summary, albeit revealing weak effects, we can demonstrate that the label and the rainbow visual had significant and different effects. While the rainbow visual lead to the least favorable intentions to apply in the masculinum condition, there was no difference between the no-label condition and the label condition. Therefore, just adding any LGBTQI+-related visual does not have the same quality as with a label.

As a result, H3(b) is partly confirmed. We can see that the label is recognized as just another visual information (rainbow heart) that did not perform better, and in the case of the masculine language, it performed even worse. This supports the idea that labels are not just a matter of congruence with the general text, but they also offer genuinely valuable and beneficial information.

The main purpose of this study was to identify the impact of employer branding strategies, particularly using gender-fair language and the inclusion of diversity indicators, which were displayed with labels and visuals referring to LGBTQI+ in job advertisements. The aim was to investigate if implementing these elements into recruitment efforts would enhance the attractiveness of jobs for potential job seekers by changing attitude towards the job advertiser and the intentions to engage with the advertiser. These elements act as strategic signals that communicate an organizations value, contributing significantly to the transfer of its image to prospective employees, specifically targeted to LGBTQI+ individuals, as they represent a valuable segment on the labor market as a group with above-average education levels and entrepreneurial tendencies (de Vries et al., 2020). Additionally, we investigated the interaction between visual representations and the use of language in German job ads, as there has been limited evidence on combined effects of gender-inclusive language and diversity signals in employer branding.

As expected, across two studies, we show that the binary gendered language as well as the generic masculinum language performed better for the intentions towards the recruiter in Study 1, whereas in Study 2, again the masculine conditions lead to higher mean scores, but the differences were not that clear. This could be due to a reduction of the text in Study 2 or due to the weak to medium effect sizes lacking statistical power to observe such subtle effects. These subtle effects might also be reflected in the answers of our manipulation checks as about one third to a quarter of the participants could not directly recall the label or the gendered language. Besides maybe neglecting the gendered language for ideological reasons (Kotthoff, 2020), we might also see that job ads are more likely peripherally processed (Petty and Briñol, 2011) and therefore might not have such a strong impact on the attitude towards the job advertisers employer brand.

Another explanation could be that part of the sample was collected during the pride month, which might create more understanding, but a test to detect any differences based on the collation time of the data did not yield any influences. Yet, what we can see is that recruiters might influence this negative effect of the inclusive gendered language (with the gender *) by adding information that implies LGBTQI+ inclusivity and therefore changing respondents' expectations as a masculine- or binary-gendered language would conflict. Thus, as gendered language becomes a more widespread requirement, we suggest that job advertisers might use labels but also other visuals to improve the perception of their ads, despite the rather negative appeal of gendered language in society as adding more signals is mitigating the negative effect of the gendered language.

It is important to recognize the complexity of German’s grammatical gender system. Non-native speakers might face challenges in interpreting gender-neutral or gender-inclusive job titles such as Verkäufer/Verkäuferin (salesmen/saleswomen), which can create uncertainty about whether they can apply. Therefore, providing clarifying statements such as m/w/d ensures that all genders are welcome to make postings more inclusive for all applicants. Another option and avenue for research would be the application of gender-neutral terms like “Verkaufpersonal” (sales personnel), which does not conflict with ideological issues with language and corresponds more with terms from languages that do not require gendering.

Promoting and supporting diversity in the workplace represents good people management and demonstrates the organization’s dedication to valuing every individual, creating a fair, respectful and innovative environment and contributing to employee well-being and engagement. This inclusive environment empowers individuals to bring out their full potential, attracts prospective employees and fosters a culture of openness, which is also highly relevant in the race for talent. By incorporating these elements in job advertisements, organizations can align their branding strategies with contemporary values of inclusivity and fairness, which not only attract a diverse talent pool of highly skilled candidates but also reinforce the organizations reputation as a socially responsible and forward-thinking employer. As the competition for top talent intensifies, prioritizing inclusivity and diversity is not just an ethical choice but a strategic necessity for organizational success.

The data were collected on a micro working platform with a slightly higher proportion of male respondents. Also, the education appears quite high in comparison, which might impact results. Furthermore, collecting data via micro-working platforms might incur participants reading instructions as well as manipulations with little attention. A recent meta-analysis on the use of micro working in comparison to traditional samples showed that there are no significant differences in means or variances; however, the MTurk differs in certain conditions (Keith et al., 2022). We applied a rigorous data clearing based on an instructional manipulation test (Oppenheimer et al., 2009), which showed no difference in other studies between traditional samples and micro working samples (Paas et al., 2018) and a reaction time based on the relative speed index (Leiner, 2019), and we see that this improved the manipulation checks as well as the results in comparison to the overall collated sample. Yet, there might be additional screening or attention checks needed to further purify the data. Subsequent studies could enhance the validity of micro-working platforms by replicating these findings with a more representative sample.

Looking into our results, we detect that the effect sizes are weak to medium, which could be due to a small sample given the effects of language or label. Future studies, therefore, might consider a more statistically power-driven approach to sample size (Lakens, 2022) rather than our heuristic approach to sample size based on common practices in communications and business experimental research.

Additionally, in Study 2, the text was shorter, which might have led to a less-perceived gendered language; thus, replicating Study 2 with more gendered text might overcome this issue. Another interesting effect would be to investigate whether males and females in comparison to non-binary identifying persons perceive these ads differently; however, our sample size does not allow to check that accurately. Therefore, future studies might make a particular comparison.

Translation (we added here for better illustration a male-female gendered English version following the German logic): Aviazion GmbH: Aviazion is the specialist regarding aviation services. Be a part of our team of specialists, which provides ground services, facility services, catering and maintenance for airlines and airports as well as sales. Are you the born salesman*woman? You love the challenge as aviation technician*ess? Are you a specialist*ess in creating fine meals? Are you an electrician*ess, plumber*ess, metal-worker*ess? Are you an experienced consultant*ess? You have a university degree and are looking for a position as Manger*ess? Then do not hesitate to check our job offers. You can expect fair working conditions, an above-average salary an awesome team, multiple career options and many other benefits.

Translation (we added here for better illustration a male-gendered English version): Aviazion GmbH: Aviazion is the specialist regarding aviation services. Be a part of our team of specialists, which provides ground services, facility services, catering and maintenance for airlines and airports as well as sales. Are you the born salesman? You love the challenge as aviation technician? Are you a specialist in creating fine meals? Are you an electrician, plumber, metal-worker? Are you an experienced consultant? You have a university degree and are looking for a position as Manger? Then do not hesitate to check our job offers. You can expect fair working conditions, an above-average salary an awesome team, multiple career options and many other benefits.

Translation (we added here for better illustration a male-gendered English version): hire.rpo “We are looking for you as a specialist” – Your entry to administration, management, marketing, IT, engineering, facility, sales, customer services, health and social work”. Since its inception, HireRPO successfully recruits young professionals in various industries. You are looking for a new challenge? We have something suiting for specialists.

Translation (we added here for better illustration a binary male-female English version): hire.rpo “We are looking for you as a specialist (m) or specialist (f)” – Your entry to administration, management, marketing, IT, engineering, facility, sales, customer services, health and social work”. Since its inception, HireRPO successfully recruits young professionals in various industries. You are looking for a new challenge? We have something suiting for specialists (m) and specialists (f).