Research values

Science is facing a reproducibility crisis: the findings of many scientific studies across disciplines cannot be reproduced. Causes of irreproducibility are varied and include publication biases, questionable research practices (such as using inappropriate controls, inadequate statistical analyses and p-hacking), poor documentation practices, a lack of accessible raw data, and, in some cases, scientific misconduct and fraud (plenty of papers on this, see e.g. Bausell, Mikayawa and Elliott).

Scientists are developing solutions to address this crisis: increasing research rigour, encouraging Open Science and improving the infrastructure supporting the reuse of scientific data: data should be findable, accessible, interoperable and reproducible according to the FAIR principles.

When conducting research, I aim to embrace these principles and adhere to Open Science practices wherever possible.

To avoid p-hacking, overreliance on arbitrary thresholds and misinterpretation of results, most analyses in my PhD were conducted within a Bayesian framework. Bayesian statistics emphasise parameter estimation and uncertainty quantification rather than frequentist null-hypothesis testing and p-values. I chose to use a Bayesian approach because it provides full posterior distributions of parameter estimates and a more intuitive quantification of uncertainty. While Bayesian statistics have their own limitations, evidence suggests that they can reduce effect size estimation and thereby contribute to alleviating the reproducibility crisis.

To ensure transparency, all data underlying my PhD chapters have been made publicly available via GitHub and NCBI. Almost all analyses were conducted using code-based approaches rather than point-and-click software which is difficult to document. All code has been archived on GitHub, Zenodo and/or Figshare. In addition, all bioinformatic pipelines were implemented using Snakemake, a workflow management system that explicitly defines dependences and execution rules, ensuring analyses can be repeated, shared and adapted across computing environments, thus enhancing reproducibility and interoperability.

Finally, I recognise that science benefits from embracing the diverse perspectives and contributions of researchers across genders, ethnicities, sexual orientations and other social identities. Throughout my PhD, I have sought to make science more inclusive by quantifying the barriers that researchers face, critically assessing the knowledge gaps shaped by the social-political context of sexual selection theory, and ensuring that all my publications are Open Access – helping to bridge the divide between the global North and South (alongside the many other academic, economic and societal advantages of publishing Open Access).