Ethical and responsible behavior in applied empirical research: four essays on academic practices in the social sciences

Pruschak, Gernot; Hopp, Christian (Thesis advisor); Piller, Frank Thomas (Thesis advisor)

Aachen : RWTH Aachen University (2021)
Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2021


Ethical and Responsible Behavior in Applied Empirical Research: Four Essays on Academic Practices in the Social SciencesDissertation by Gernot PruschakAbstractIn a world filled with fake news and alternative facts, public trust in science is of utmost importance. Yet scandals like the cases of Diederik Stapel and Ulrich Lichtenthaler have questioned the integrity of scholars and their research results. To address this issue, several scientists investigated academic (mal)practices like plagiarism, HARKing, authorship misuse and data flexibility. The results were devastating and ignited a credibility crisis, especially in the social sciences. Fortunately, we already can see the light at the end of the tunnel as editors, publishers, research societies and universities have started to introduce techniques and infrastructure that ensure ethical and responsible scholarly behavior. For example, artificial intelligence has enabled plagiarism detection software to not only check for copy-pasting but also for content and reference similarities. Moreover, more and more journals motivate or sometimes even require researchers to pre-register their research hypotheses prior to data collection and/or data analysis to prevent HARKing. In the life sciences, contribution disclosure statements force authors to transparently report the contributions of each researcher involved in a research project. In the social sciences, several articles and editorials highlighted that ensuring replicability by means of transparent reporting and data sharing allows detecting and overcoming flexible and questionable data handling practices.This thesis builds upon the existing body of literature and provides guidance for those academic (mal)practices that have been covered only rudimentarily in the social sciences. Essay 1 addresses the issues of ghost and honorary authorship, the two most infamous forms of authorship misuses. We show that there exists a strong mismatch between actual and hypothetical authorship assignments: While most social scientists in our survey assigned authorship correctly in three hypothetical scenarios, more than every third paper contained at least one honorary author. We conclude that motivational factors like hierarchical pressure force scholars to include low-contributing supervisors and colleagues in their author lists. To overcome this issue, we call for social science journals to follow their counterparts in the life sciences by enforcing contribution disclosure and implementing whistleblowing platforms. Essay 2 asks whether scientific collaborations and authorship teams differ across academic disciplines, geographical regions, working experience and job position. The results indicate that the distribution of multi-authored papers varies substantially even within the social sciences. Furthermore, we highlight that language barriers and infrastructural challenges possess effects on academic collaborations. These findings contain important implications for search and tenure procedures as committees must bear these differences in mind when comparing applicants with different backgrounds. Essay 3 touches upon the need of promoting replicability and replications. We provide a hands-on step-by-step guide on how to conduct a rigor and robust replication. More specifically, we start by directly replicating the results of Kuhn and Weinberger (2005) showing that, for white men, leadership positions in high school correspond to higher wages eleven years later. We then assess the causality of this effect by employing propensity-score matching and three different forms of instrumental variable techniques. Moreover, we move beyond the original sample and investigate the effects for white females and non-whites. We further include data from a follow-up study conducted 50 years later. Our findings highlight that team captainship combined with club presidency induces higher wages eleven and 50 years later for white men, but this effect does not appaer consistently for white females or non-whites. Therefore, leadership interventions should recognize the leadership skills that are already developed in individuals and identify those areas that are in need of further development. In doing so, it is important to be cognizant of diverse aspirations to lead for individuals of different gender and ethnicities.The thorough analyses in Essay 3 were only possible because we had access to the data employed in the original article. Unfortunately, Essay 4 shows that this constitutes rather an exception than the norm. Investigating data sharing among innovation management researchers, we find that only about one third of their datasets are publicly available. We specifically focus on innovation management researchers because we originally expected them to be specifically prone to open data due to them advocating the advantages of openness to firms and other stakeholders. Our results indicate that the identified personal incentives to open data sharing might not outweigh the burden of open data places on individual researchers. Consequently, we call for academic impeti that give more credit to data sharing and for journal policies enforcing data sharing as a mandatory requirement for publication. Overall, this thesis highlights that scholars, journals, research societies and universities must change their habits, incentives and conducts especially in the areas of scientific authorship and data practices to secure scientific integrity. Therefore, we outline detailed solutions for guiding this change to ensure ethical and responsible behavior in applied empirical research. This is the only way we can regain public trust in science.