Overcoming citation bias is necessary for true inclusivity in Plant Science

Dear Editor,

Despite concerted efforts in recent years to achieve equity in science and academia, both implicit and explicit biases persist and remain a concern (Llorens et al. 2021; Windsor and Crawford 2021; Graves et al. 2022; Kozlowski et al. 2022). Women and people of color continue to be a minority and face uphill battles to achieve equity with their majority peers. Although proactive attempts toward hiring and recruiting have resulted in some improvement in the representation of minorities at academic institutions, industry, and journal editorial boards, these still do not reflect the actual societal composition. Moreover, women and other minorities remain significantly underrepresented in leadership roles (Grandizio et al. 2020; Tiwana et al. 2021; Dai et al. 2022; Lerman et al. 2022). The situation has been described as a “chute and ladder” problem (Cardel et al. 2020; Grandizio et al. 2020; Tiwana et al. 2021; Dai et al. 2022; Lerman et al. 2022), where throughout their careers, women and other minorities encounter multiple chutes, which bring them down and not as many ladders that can pull them up. In this letter, we want to draw the reader's attention to one such “chute,” citation bias.

The term “citation bias” has been used in several different contexts, and in each one, it results in unintended consequences. There are geographical citation biases, where research conducted in or published from specific geographical locations receives less (e.g. India and Russia) or more (e.g. Germany and The Netherlands) citations (Paris et al. 1998; Gomez et al. 2022). There are biases specifically related to an idea or a positive result. It has been well-documented that research findings that support a particularly exciting idea are easier to publish and are cited significantly more often than the findings that do not support the idea or lack such correlations (Ravnskov 1992; Duyx et al. 2017, 2019a, 2019b; Urlings et al. 2019a, 2019b; Karst et al. 2023). This has had substantial effects on specific research fields and has even resulted in misleading scientists and the public. One historical example is the field of eugenics that permeated the United States in the early part of the last century (Graves et al. 2022). A recent example relates to the seemingly positive effects of common mycorrhizal networks and their role in forest management. Several studies and popular media claim that these networks are used for resource allocation, for which there is little scientific evidence (Karst et al. 2023). A third and potentially more prevalent is gender- or race-based citation bias, where research from women or other minorities is cited less often than the researchers who represent the nonminority population (Chatterjee and Werner 2021; Llorens et al. 2021; Ni et al. 2021; Iancarelli et al. 2022; Lerman et al. 2022; Ray et al. 2022; Wu 2023). This third “bias” is the focus of this letter.

We became interested in raising this concern based on our personal experiences. We are woman scientists of color and first-generation immigrants. We are what most people will see as “success stories.” Despite many challenges, we have managed to secure academic positions, we serve on editorial boards of journals, and we have been running independent research programs and leading productive laboratories in the field of our choice for some time now. Our day-to-day interactions with colleagues and co-workers are incredibly positive, and the idea that we might still face discrimination seems far-fetched. However, we continue to hit the chutes, a particularly glaring one being continuously under-cited. With every relevant paper published in our respective fields, we are reminded of how so many scientists and co-workers in our respective fields ignore our scientific contributions, whether deliberately or unconsciously. On specific occasions, where it seems to be an oversight, we have reached out to the authors to remind them of our existing work, but in many cases, there is apparent intent to ignore our work. In several instances, authors have acknowledged that our work was overlooked due to oversight, although to our knowledge, no effort was made to rectify the omission.

Citations are a recognition of the impact and significance of our work by our peers. However, more than positive reinforcement, citations also affect our careers. They may be a critical deciding factor in promotion and retention, funding, networking, invitations to meetings and workshops, and our perceived scientific worth (van den Besselaar and Sandström 2017; Dworkin et al. 2020a; Hofstra et al. 2020; Llorens et al. 2021; Smith et al. 2022; Wu 2023). While we have pondered over this for some time and have fought the urge of self-doubt (maybe it is not real), oversensitivity (maybe we are overreacting), or apathy (just ignore it and do your work), there are several studies and large-scale analyses that confirm that the problem exists and is actually getting worse (Nielsen and Andersen 2021). In our personal conversations with scientists who identify themselves as gender, ethnic, or racial minorities, we have not come across a single individual who has not felt this bias.

These personal experiences are supported by several empirical analyses across fields ranging from medical sciences to physics and from communication to ecology and evolution (Fox and Paine 2019; Benjamens et al. 2020; Bertolero et al., 2020; Dworkin et al. 2020b; Urlings et al. 2021; Wang et al. 2021; Iancarelli et al. 2022; Kane et al. 2022; Teich et al. 2022; Zurn et al. 2022). Our field of Plant Science is no different, as demonstrated by a recent study (Marks et al. 2023). An analysis of almost 300,000 articles published in the 20-yr span of 2000 to 2021 revealed that the publication rate tracked with national affluence, leading to a striking imbalance in the global distribution of plant science publication, with richer, northern countries publishing the bulk of articles. When controlling for the journal impact factor, papers from the Global South were found to receive fewer than half the number of citations received by those with authors from the Global North. The study authors note that this disturbing trend has been persistent for the past 20 yr. Given this clear geographical bias in citations, it was, therefore, not surprising that there was also glaring gender citation bias, with papers by first authors with names normatively associated with femininity receiving, on average, 5 citations fewer than those whose first authors had names normatively associated with masculinity (Marks et al. 2023). Again, this concerning difference in citation rates was persistent over the 20-yr study period.

So, what can we do as a community to minimize this bias? Just as with anything else, all of us first need to acknowledge that the problem exists. Under-citation of specific groups of people affects not only their growth trajectories, but also those of the trainees and next generation of scientists who work with them. It contributes to a vicious circle and eventually affects the community at large by limiting the diversity of the scientific workforce (van den Besselaar and Sandström 2017). Here, we make 5 recommendations for countering citation bias.

The first is to be aware of this problem in our own writing and citations. Are we citing the most-cited reviews, just because we assume that they contain all relevant information? Are we citing the most recognized names in the field because we have those names memorized? Are we citing people in our networks because we are familiar with their work? Are we self-citing excessively? We recognize that in some instances, a great, authoritative review needs to be cited, and self-citation is required to build the foundation for the work to be discussed, but these should not preclude including relevant work by other scientists in the field. Presuming that much of such bias is unintentional, self-awareness, and self-correction will certainly help improve the current situation. In addition, we strongly believe that making the next generation of scientists aware of such biases and including discussion about these biases as a part of their scientific ethics training will have a positive effect in the future.

Another point of concern is that when discussing science endemic to specific parts of the world, particularly the global south and other developing countries, many scientists tend to cite some high-level report published by international organizations (e.g. UN Climate Report) and not the original work by local scientists. In many instances, these high-level reports are easily accessible, and it may take more time to find the relevant primary research. As a second step, therefore, conscious efforts should be made to cite the original research. The scientists who produce ground-level data for such information, often from the Global South or lower-income countries, deserve to be acknowledged for their work.

A third level of scrutiny can be applied at the level of reviewers and editors of scientific journals. Having a diverse pool of reviewers and a diverse editorial board is helpful, in general. However, they may also scan for whether the citations include multiple groups of scientists and not only from the authors’ network of people. Similarly, self-citation is always obvious, and excessive self-citation should be pointed out in review reports or editorial comments.

Another suggestion is to deliberately raise awareness of citation bias by incorporating this topic in discussions at scientific conferences and meetings and in other settings where professional development or diversity, equity, and inclusivity training or workshops are conducted.

Finally, several journals (e.g. Cell) have now started the practice where the authors can fill out a diversity and inclusion form when submitting a manuscript, where they can identify the diversity and inclusion in the list of authors and note whether diversity and inclusion was considered in relevant reference citations. This should be widely applied, as it may lead the authors to pause for a moment, take a closer look at the work they are citing, and avoid unintentional oversight of relevant work. The availability of specifically developed language model system tools (King 2023) may be able to assist authors with this exercise in the future.

While these may not solve the long-standing problems per se, these are certainly steps in the right direction for creating self-awareness in biased citation practices.

Funding

The research in Pandey’s laboratory is supported by the National Science Foundation (grants MCB-2207012 and EF URoL-1921724). Research in Burch-Smith’s laboratory is supported by the National Science Foundation (MCB-2210127).

Data availability

No data were generated in this work.