Early-career researchers in Australia report dissatisfaction, bullying and questionable research practices. We discuss how this may contribute to the replication crisis and suggest local and international strategies to improve the industry.
Australian early-career researchers (ECRs) are struggling. Internationally, there have been concerning reports from researchers in the science, technology, engineering, mathematics and medicine (STEMM) disciplines regarding poor job security and career prospects, as well as discontentment with the general work environment.
The Australian ECR experience
Regrettably, there are numerous problems with the Australian ECR workplace and the academic research industry. To better understand the workplace for Australian STEMM ECRs (defined here as those up to 10 years postdoctoral), we conducted surveys that included questions related to job satisfaction, workplace pressures, bullying and questionable research practices. Our surveys received 658 responses in 2019 (ref. 1) and 530 responses in 2022 (non-peer-reviewed preprint2).
ECRs almost unanimously indicated a “love of research”. However, many are struggling and most intend to leave academia (Fig. 1a). ECRs reported frequent bullying and questionable research practices, and the majority agree with the statement that “this is a poor time for a young person to start in this career” (Fig. 1a,b). Inadequate job security, a lack of funding or a lack of independent positions are some of the reasons that ECRs felt motivated to leave academia in 2019. In 2022, personal wellbeing was also a major concern and over half of respondents indicated that they were considering leaving because of depression, anxiety or other mental health concerns related to their work.
Precarious employment makes ECRs vulnerable to bullying and questionable research practices. ECRs reported a lack of support from institutional superiors and frequently indicated they had been a victim of bullying or harassment from those in a position of power (Fig. 1b). Of the ECRs who experienced bullying, supervisors were the perpetrator in nearly half of cases, and other senior colleagues or peers were implicated in the remaining instances.
The effects of questionable research practices are worryingly common among the ECRs who we surveyed, and almost half of them did not believe that their institute would act on a complaint. Many respondents also felt that they themselves would suffer negative consequences if they were to make a complaint.
The Australian government recently reviewed recommendations that suggested the establishment of a National Research Integrity Office. For now, the government has made no commitment to establishing an integrity office, commenting that they were concerned about the damage that public reporting could inflict on the reputations of universities. Of course, if the federal government is more concerned about university reputation than oversight, then ECRs are rational to assume that their concerns may be ignored.
We reason that many workplace challenges in academia are related to supply versus demand pressures that are associated with both people and resources, as well as complications that are associated with using publications as the primary currency in academia. Through this economic lens, we suggest strategies to improve both the local and the international academic research industry.
The Australian ECR environment
Why do ECRs tolerate these work conditions? Because an oversupply of PhD graduates means tough competition for jobs, and that when ECRs are worn out they will be quickly replaced by younger, less-expensive new graduates.
Australian universities graduate more PhDs than the average of nations in the Organisation for Economic Co-operation and Development (OECD), yet Australia ranks 21st (and below average) of OECD nations for percentage of GDP investment into research and development (R&D) (2019 data, published in 2021). The number of Australian PhD graduates has increased exponentially since the 1990s, whereas academic positions have increased only in a modest linear manner. Limited academic research jobs and less-developed advanced industries that typically employ highly skilled workers means that there is a disconnect between the number of Australian PhD graduates and the workforce demand for some skillsets.
In Australia, the federal government allocates block grant funding to institutes using a calculation that accounts for their successful winning of competitive grant income (25%), attracting engagement income (25%) and for higher degree completions by research students (50%). Effectively, the largest portion of block grant funding (and access to cheap labour (that is, PhD students)) incentivizes Australian universities to enrol large numbers of PhDs. Equally concerning is that only a PhD completion yields a dividend, potentially incentivizing institutes to overlook candidate quality in an effort to capture the completion dividend.
The Australian government’s report ‘2030, Prosperity through Innovation’ identifies the need to train a future workforce to lead innovations that will support our standard of living as the population ages and the resource boom eases. We agree that it is critical to look to the future and to establish an appropriately skilled workforce. However, it may be equally counterproductive to train highly specialized PhD graduates that are not needed in Australia — potentially reducing the earning potential of these talented individuals as well as their capacity to contribute to other industries. Intertwined with this national economic loss is the personal struggle of PhD candidates and ECRs who may suffer considerable personal distress and loss because of poor career stability and little training for alternative careers. This is reflected in our survey: ECRs were motivated to leave academia owing to poor job security (Fig. 1a) but expressed concern that they were not well equipped for employment in alternative industries. These challenges have been echoed by PhD candidates around the world.
Publication currency is being devalued
Perhaps many of the challenges in the research ecosystem could be tolerated if we were confident that this ecosystem was delivering high-quality science. However, this is not necessarily the case. Instead, as indicated above (Fig. 1c), ECRs in Australia have felt pressure to manipulate manuscripts at various stages and reported knowing fellow faculty who had fabricated data, plagiarized data, falsified data, selectively dropped datasets from analysis without transparent explanation and trialled iterative statistical analysis until finding a model that yielded a ‘significant’ result (Fig. 1d). Many ECRs stated that they sometimes or often observed behaviours that are likely to contribute to the replication crisis.
This is an international problem: in 2016, Nature published data3 from 1,576 scientists that indicated that more than 70% had failed to reproduce another scientist’s experiments and that most (more than 60%) assumed that reproducibility challenges were associated with the pressure to publish and selective reporting. Further, high-profile studies have demonstrated that data presented in our most prestigious journals are often not reproducible4, and a fascinating economic analysis has estimated that nonreproducible pre-clinical research cost the USA US$28 billion per year5.
Publications are the primary currency of academia but there is no universal audit system to validate reproducibility. Peer review can identify flaws in logic arguments or in data analysis, but it cannot identify data fabrication or selective reporting. Some argue that, with time, science is self-correcting, but the scientific industry has expanded rapidly and it is unclear how effective or efficient this self-correction process is. Michael Park and colleagues recently published an analysis suggesting that, despite an “exponential growth in the volume of new scientific and technological knowledge”, “progress is slowing in several major fields”6. They argue that the “observed declines are unlikely to be driven by changes in the quality of published science” as patterns hold in even when analysis focuses on data derived from Nature, Science and Proceedings of the National Academy of Sciences USA (PNAS)6. We share their concerns but suggest the consideration that contemporary data may not necessarily be reliable, even in Nature, Science and PNAS. An alternative explanation might be that data in Nature, Science and PNAS are similar to the bulk of the literature and are sometimes compromised. In one high-profile replication study, the authors state that “the reproducibility of published data did not significantly correlate with journal impact factors”7. Yet, those on the academic job market with a paper in Cell, Nature or Science received offers for 11% of applications, whereas those without a Cell, Nature or Science paper received offers for only 2% of applications8 — a single paper in Cell, Nature or Science changes a career trajectory, thus amplifying the incentives for bad behaviour.
Reform Australian research funding
With the aim of better aligning supply and demand in Australia, we suggest that R&D expenditure be increased (where feasible) and that federal block grant funding (50%) for PhDs be distributed in a manner that incentivizes institutions to target graduate employment rather than simply graduate completions. Effectively, the 50% block grant funding would be allocated on the basis of graduate employment statistics. Australian PhD programmes, unlike those in North America, do not currently include a substantial advanced coursework component. Programmes might be adapted to include broader training if graduate employment were a fundable metric. This approach might also function to reduce Australian PhD intake in fields in which employment is challenging, making entry more selective and, potentially, (on average) leading to higher-quality, more-rounded graduates. Holistically, it may be better for both individuals and the nation to simply increase the barrier for entry into some PhD programmes — instead of the current barrier, which is a limited capacity for research employment following PhD graduation. Critics may argue that all education is beneficial and that there should be no barriers to entry into PhD programmes, but public resources are finite and a mature discussion is needed about how to use limited resources.
Encourage mentorship and PhD employment
Intertwined with the redistribution of block grant funding to align with graduate employability, institutional culture changes are required to help PhD graduates and ECRs to navigate career paths. In the case of Australia, even though precarious ECR employment is a well-known problem, our survey data indicated that not all supervisors even discuss career aspirations with ECRs — and even fewer discuss skill development and alternative career options. If graduate employment outcomes are made a priority, Australian institutes might consider extending the nominally 3-year PhD training programme to make room for additional, broader training and focus on quality research. Within institutes, laboratory head, tenure or academic promotions could in part be based on 360° reviews that consider data from current and past students or ECRs, including the appropriateness of their supervision and the employment outcomes of former trainees. Laboratory heads carry an enormous burden, often including having to find a salary for themselves. For this approach to work, institutes would be obligated to account for this time investment in academic workload allocations and to ensure that programmes were in place to maximize the employment prospects for graduates. Science has become increasingly complex, and it is likely that nations that desire to remain competitive will need to adopt a longer-term view on investment into research projects and the development of productive researchers.
Fund replication studies
To address the replication crisis, we believe that academic research requires an audit process; it is surprising that a multibillion-dollar industry has evolved without one. We propose that a portion (perhaps 10–20%) of public research funding be allocated to replication studies, and that this be rolled out on an international scale. Replication studies could be performed at independent research organizations or contracted out to specialist laboratories. For some laboratories, this contract work might augment funding used to pursue original research. Every year, all publications funded by a specific organization would be entered into a lottery for replication. Entries could be weighted, with publications in higher-impact journals having a greater weighting. Weighting might be modified with time as replication data accumulated or weighting might be biased to target areas of intense public interest. Funding agencies would expect authors to cooperate and assist replication efforts, and we anticipate that many authors would be enthusiastic to see their results validated. Similarly, we anticipate that most authors would want to avoid replication failure and that this would motivate greater care in the evaluation and publication of datasets. Finally, integrated into the replication crisis challenge is the tendency of researchers to overstate the potential impact of their work9,10. Replication study publications could include a brief independent, and probably anonymous, editorial providing a tempered assessment of the scientific and clinical impact of the results. This process would aim to improve the reliability of published data and the straightforward communication of their impact — effectively, this would stabilize the depreciating value of the publication as currency.
Effectively, we are arguing for slower, more systematic training and science. In Table 1 we suggest solutions, for discussion, that are aimed at improving the ECR research environment in Australia and internationally.
For Australia, the economics of training many PhD graduates and ECRs for careers that do not exist is irrational. Internationally, spending billions on science that is not being validated is irrational. The stress of the COVID-19 pandemic exacerbated conditions for Australian ECRs, exposing flaws in academic environments, just as COVID-19 stresses exposed vulnerabilities in many systems around the world. The positives are that the pandemic has demonstrated the value of STEMM research, that it is often possible to find viable solutions and that researchers are the right people to drive these changes.
Christian, K., Johnstone, C., Larkins, J.-A., Wright, W. & Doran, M. R. eLife 10, e60613 (2021).
Christian, K., Larkins, J.-a. & Doran, M. R. Preprint at https://doi.org/10.1101/2022.12.06.519378 (2022).
Baker, M. Nature 533, 452–454 (2016).
Begley, C. G. & Ellis, L. M. Nature 483, 531–533 (2012).
Freedman, L. P., Cockburn, I. M. & Simcoe, T. S. PLoS Biol. 13, e1002165 (2015).
Park, M., Leahey, E. & Funk, R. J. Nature 613, 138–144 (2023).
Prinz, F., Schlange, T. & Asadullah, K. Nat. Rev. Drug Discov. 10, 712 (2011).
Fernandes, J. D. et al. eLife 9, e54097 (2020).
Ioannidis, J. P. A. Amer. Stat. 73(supp. 1), 20–25 (2019).
Boutron, I. & Ravaud, P. Proc. Natl Acad. Sci. USA 115, 2613–2619 (2018).
The authors thank K. Futrega for assistance with the figure, and M.R.D. thanks A. Davies for engaging discussions about replication studies.
M.R.D. is currently employed in industry in the USA (AstraZeneca), but having worked in academic STEMM in Australia for many years, until 2022, continues his interest in the welfare of ECRs. The other authors declare no competing interests.
About this article
Cite this article
Christian, K., Larkins, Ja. & Doran, M.R. We must improve conditions and options for Australian ECRs. Nat Hum Behav 7, 1038–1041 (2023). https://doi.org/10.1038/s41562-023-01621-w