Gender-based citation differences in speech–language pathology

Gender inequalities are well documented in academia and typically favour male researchers (Huang et al. 2020; Ross et al. 2022). For example, female researchers compared to male researchers frequently fare worse regarding work recognition, compensation, grant funding outcomes, teaching evaluations, hiring, or tenure promotion. Moreover, despite increasing awareness of gender biases in academia and initiatives to address them, necessary change appears to be slow and substantial inequality remains (Llorens et al. 2021).

In this context, undercitation of female scholars is a particularly detrimental type of gender difference that has profound negative effects on academic visibility and career advancement (Andersen et al. 2020; Dworkin et al. 2020; Llorens et al. 2021). Several recent studies have demonstrated that reference lists of scientific publications tend to include more papers with men as first and last authors (i.e., prominent authorship positions, frequently indicating the lead author and supervisor of the study) than one would expect based on the gender distribution of authors in specific fields. This pattern has consistently been shown across several fields, including neuroscience (Dworkin et al. 2020; Fulvio et al. 2021), physics (Teich et al. 2022), astronomy (Caplar et al. 2017), international relations (Maliniak et al. 2013), and political science (Sa et al. 2020). These studies suggest that male overcitation is a pervasive and universal pattern in science, with widespread negative implications for female scholars.

However, previous studies on citation inequality in science have mainly investigated fields with author distributions strongly skewed toward male researchers. In these fields, sociological theory predicts that a combination of implicit or explicit favouritism towards similar group members (e.g., individuals of the same gender, particular by individuals in influential positions; McPherson et al. 2001) and systemic factors (Dworkin et al. 2020; Llorens et al. 2021; Ross et al. 2022), contribute to overcitation of male scientists.

Therefore, the present study investigated citation inequality in a field, which has traditionally been more accessible to female scientists, i.e., speech–language pathology (SLP). For example, males make up only 2.5% of speech and language pathologists in the United Kingdom (Litosseliti and Leadbeater 2013), and a recent survey in 13 different countries revealed that on average, only 5% of registered speech–language therapists are male (range: 2%–13%; Speechguys 2017). Similarly, the percentage of male academics registered with different professional or scientific SLP organizations in 2021 ranged from 4.5%–22% (these numbers were acquired in personal communications with representatives of Speech Pathology Australia, American Speech–Language-Hearing Association, The Academy of Aphasia, and The German Association for Aphasia Research and Treatment). This field-specific pattern of gender distribution, and the resulting gender author distribution in SLP publications (see below), allows investigating whether (male) citation imbalance is a universal phenomenon, even in fields that are less “dominated” by males, or whether citation patterns are influenced by the author gender distribution in specific fields.

A total of 8607 publications and 66 495 citations were included in the respective analyses (Fig. 1). A categorical gender breakdown of authorship in the investigated journals from 1958 to 2021 is visualized in Fig. 2, showing that the vast majority of publications in SLP in 1958 (69.2%) were authored by MM author teams. Over time, this proportion gradually decreased to 23.4% in 2021. Publications with at least one female author increased over time and in 2021, 21.2% of the publications were WM-authored, whereas 19.1% were MW-authored. WW-authored publications increased to 36.3%. Hence, 76.6% of publications had at least one female author in 2021. For a detailed overview of the categorical gender breakdown in individual journals and available time frames see Supplementary Fig. 1.

Our results show that the author gender distribution in SLP publications is currently more balanced compared to previously investigated fields of science that are largely dominated by male authors (Dworkin et al. 2020; Fulvio et al. 2021; Wang et al. 2021; Teich et al. 2022). Moreover, for the last decade, our results demonstrate a citation pattern in SLP that (a) overall tends to favor female authors, (b) persists after controlling for potential confounding factors, but to a lesser extent, and (c) is particularly strong for publications involving female first and last author teams. While it is common that male scientists receive more citations than expected in many fields of science (Llorens et al. 2021), our results emphasize the contribution of author gender distributions in specific sub-fields of science.

Previous studies interrogating citation inequity have highlighted a pervasive pattern of undercitation of female scholars (Maliniak et al. 2013; Caplar et al. 2017; Dworkin et al. 2020; Sa et al. 2020; Fulvio et al. 2021; Wang et al. 2021; Teich et al. 2022). However, these studies have mainly investigated fields where author distributions are strongly skewed towards men (e.g., neuroscience, physics, or astronomy). Even in recent years, the majority of publications in these fields were contributed by male authors. The most dramatic example comes from contemporary physics, where only 3.6% of articles were contributed by WW author teams and only 33% involved at least one female first or senior author (Teich et al. 2022). Indeed, similar author team distributions have been reported in all previously investigated fields to date. For example, in neuroscience, 78–90% of publications involved male authors, either as first or senior author, or both (Dworkin et al. 2020; Fulvio et al. 2021). In contrast, the present study investigated for the first time citation patterns in a field where the majority of recent publications have women in prominent authorship positions (i.e., ∼75% of publications with a woman as first and/or senior author over the last two decades, including ∼30% WW author teams, compared to ∼65% of publications with a woman as first and/or senior author over the last two decades; see Fig. 2). For this specific field, we demonstrate that author teams involving at least one female author are citing publications authored by female first and last author teams (WW) 8.5% more than expected. This suggests that the degree of undercitation of women (and potential negative downstream effects on visibility and career advancement) is at least partly rooted in authorship distributions of specific fields of science, rather than being a universal pattern.

Nonetheless, the present research only considered one specific field of science that differs substantially from previously investigated fields (e.g., neuroscience, physics) and we did not consider other potentially relevant factors that may contribute to field specific citation practices like field size, subfield structure, or journal impact factors. However, by keeping the analytical approach consistent with previous publications (Dworkin et al. 2020; Fulvio et al. 2021; Wang et al. 2021; Teich et al. 2022), we aimed to facilitate direct “historical” comparisons between these specific subfields. To establish a causal link between author gender distributions and gendered citation practices, future research is needed that directly compares fields with varying author gender distributions and also considers potential field-specific idiosyncrasies in citation practices.

Three additional limitations need to be acknowledged: As in previous studies (Dworkin et al. 2020; Fulvio et al. 2021; Wang et al. 2021; Teich et al. 2022), first names of authors were identified using a probabilistic and binary gender algorithm, which may not reflect the actual self-identification of authors. However, because we used the same algorithm as in previous publications, the overall results of this study are still comparable to previous work. Moreover, in line with previous publications, we only considered first and last author gender in our study. While SLP currently follows standard authorship conventions (lead and last authors as most prominent positions), senior authorship in earlier publications may have followed a different convention (e.g., indicated by the second author). This potential confound may have affected the historical trends in authorship analysis, especially the earlier years that are reported.

Finally, actual gender self-identification may be unknown to the citing authors and obscured by citing authors’ assumptions of gender identity. It also needs to be noted that relative to the large number of publications/citations that were included in some previous studies (e.g., physics, neuroscience), we investigated a much smaller field of science (SLP), which produces relatively fewer publications/citations. This also prevented investigation of temporal trends in authorship across the past decade, which would have allowed more direct investigation of whether changes in authorship distributions over time result in changes in gendered citation inequality. However, no clear pattern emerged from the temporal analysis (not reported) and wide error bars suggested that no inference can be made from these analyses.

Nonetheless, the overall pattern of the results in a field that is largely dominated by female scientists, are consistent with those reported previously in fields dominated by men (with widely varying numbers of analyzed publications, N = 2.069–1.067.276; Dworkin et al. 2020; Fulvio et al. 2021; Wang et al. 2021; Teich et al. 2022), even though the direction of the effects (favouring female vs. male scientists, respectively) was different. This suggests that citation differences depend at least to some degree on authorship distribution in specific fields, which also has implications for potential interventions aimed at reducing gender-based inequality in science.

Specifically, while the exact mechanisms underlying gender-based citation inequity are currently unknown and the present study was not designed to pinpoint the many potential individual and systemic factors that mediate the observed citations below expectations of female or male authors (for discussions see Dworkin et al. 2020; Llorens et al. 2021), the present study highlights the contribution of a common human factor to citation bias in science (i.e., “homophily”; McPherson et al. 2001). This concept describes the tendency of individuals to associate, bond, or (implicitly or explicitly) favour other individuals that are similar with regard to age, race, and social status, but also gender. Within this framework, it would be expected that scientific fields dominated (either with regard to numbers or degree of influence on a given field) by men or women alike show a bias towards citing work (or “appreciation”; Ross et al. 2022) of their own “gender in-group”. In the broader context of the current literature on disadvantages female scholars are faced with (Llorens et al. 2021), this theory provides a framework to further investigate currently reported patterns of citation differences and to establish causal links between general or field-specific factors and gendered citation disparity. Such analyses will also provide context for studying cumulative bias against individuals perceived as belonging to multiple marginalized groups (e.g., black women). However, the present analyses were focused specifically on studying gender-based citation inequality and the comparison with previous publications using methods comparable to those previously used with author gender distributions skewed towards male scientists.

Indeed, while the current data suggest that male citation imbalance may be reduced in fields with more balanced authorship distributions, other types of disparities and biases may certainly have a negative impact on career advancement of women. For example, a recent review that focused on implicit gender biases in communication science and disorders provided an overview of why more men advance in academic careers than expected in this academic subfield (Rogus-Pulia et al. 2018). It was suggested that this imbalance is explained by several implicit gender stereotypes (i.e., favouring men in hiring and evaluation processes, backlashes for agentic behaviour in women, and bias in funding), which are also likely applicable to SLP. However, implicit stereotypes are not at odds with our findings, because citations are a necessary but not sufficient condition for academic career advancement. Furthermore, the review emphasizes the importance of the intersection of gender and race, showing specific stereotypes against Black and White women (Rogus-Pulia et al. 2018). The intersectionality of biases was also emphasized in a recent work discussing the effect of inequality in peer review in communication science and disorders reducing the likelihood of publication, and in turn, citation rates in marginalized groups (Girolamo et al. 2022). This is augmented by the fact that marginalized groups are more likely to investigate topics dealing with diversity, which are also less likely to be published (Girolamo et al. 2022). In addition, a recent citation analysis in neuroscience used a deep learning algorithm to generate categories based on race/ethnicity (e.g., White, Hispanic, Asian, and Black) using first name probabilities and showed that White individuals were cited more than expected. Moreover, they also demonstrated an intersection of gender and race/ethnicity regarding citation patterns, with gender as primary explanatory factor (i.e., White men were most overcited and men of color less overcited, but still cited more than expected; Bertolero et al. 2020). Hence, citation analyses based on naming probabilities are suitable to detect multiple inequities and future studies are needed to further analyse the complex and multifactorial causes underlying disadvantages of marginalized groups in science.

In sum, our results highlight the contribution of field-specific author gender distribution to citation inequity. Implementing effective measures to increase the number and influence of under-represented individuals in specific fields of science (e.g., by connecting them to influential scholars; Verhoeven et al. 2020) may contribute to reducing downstream disadvantages regarding visibility and career advancement.