HARKing forward: an argument for a polarising research practice

HARKing is hypothesising after results are known. In principle, HARKing is wrong. This is the consensus. Leading academics have called for it to be banned. But what if scholars were allowed, even encouraged, to practise HARKing or PARKing (proposing after results are known) under specific circumstances when such practices could be justified academically and morally?

HARKing or PARKing that leads to scientific discoveries should surely be applauded. That is why here I will argue for accepting, even recommending, that a researcher HARKs or PARKs as long as the scholar is being transparent about it and properly reports it.

HARKing occurs when authors write, rewrite or revise their hypotheses after results are known. Basically, it means acting against scientific principles (cheating, in common parlance). For example, the authors develop original hypotheses, either in line with the extant theory and literature or to challenge it, deviate from it or suggest rethinking it. They predict findings, but the results contradict the expectation, so the authors change their original set of hypotheses to fit the findings.

• How universities can help rebuild trust in science
• Spotlight: Understanding and protecting academic integrity
• How open knowledge strengthens democracies

Through HARKing, then, scholars are concealing the truth. The edited manuscript presents findings that support the rewritten hypotheses, apparently as anticipated according to the literature. In extreme cases, HARKing might also take place when the authors do not develop any hypotheses until they have the results; they then develop hypotheses that will each gain support from the (now known) results. A related type of HARKing is failing to report hypotheses that were rejected, either because no relationships were identified or, worse, if the relationships identified were in the opposite direction to those posited by the hypotheses.

Strong indications are that researchers are increasingly engaging in HARKing rather than a priori theory testing, and this practice is largely condemned. Although HARKing is widespread, it is under-reported, destructive and harmful for scientific enquiry as well as damaging for the credibility of individual scholars and their fields of research. As it stands, HARKing is considered an unacceptable and unethical practice that, in principle, should be avoided in the scientific realm. It acts against academic logic, integrity and ethical guidelines.

HARKing can be found in quantitative studies employing the positivist approach. A comparable case can take place in qualitative work, employing the interpretivist approach. While qualitative papers do not offer a development of hypotheses, they do start with research question(s) and defining the aims. Similarly, as in HARKing, authors might decide to amend their research questions, add new ones or redefine their aims once their data opens avenues to provide answers to different questions or point towards different directions that were not anticipated a priori. Proposing different or amended research questions or aims can be labelled PARKing.

Yet HARKing is not necessarily fraud, though it can be compared to cherry-picking, when data is “picked” to support preferred directions, like a type of “question trolling”.

The argument for HARKing und PARKing

I argue for the use of HARKing, as long as such a use is restricted to certain circumstances, pointing out under what (limited) conditions HARKing should be acceptable, even encouraged. It should be clear, though, that I am absolutely against opportunistic HARKing and PARKing, when scholars run an analysis, see what the results are and then develop a set of hypotheses or propositions to match the findings. It would be like an archer shooting an arrow, then running to draw the target around where the arrow hit in order to win the competition.

Yet academic enquiry is not competition. Science is more important – its role is to develop and create knowledge. Under positivism, scholarship aims to reveal the truth (assuming an absolute truth can be found). Under interpretivism, the need is to faithfully represent the participants’ opinions and be transparent about the interpretation process. Science is about creating knowledge, making discoveries, not necessarily following basic logic of linear predictions based on existing theories. Returning to the archer metaphor, if a hunter shot an arrow towards the bushes and hit game that was only assumed to be there, no one should care if the aim of finding food was achieved.

From the scientific realism perspective, our aim as scientists is to reveal the truth, to generate and test theories, and to establish knowledge that is valid and reliable.Yet the publication convention is that the introduction and results sections must be consistent with one another. If a scholar develops a logic that leads to specific hypotheses based on certain literature, the data and analysis will be limited to examine only this line of enquiry. However, this might mean that possible avenues for knowledge creation are blocked.

Moreover, playing strictly by the rules could lead to a competitive disadvantage because it can lead to missed opportunities to generate unpredictable findings. It neutralises the prospects of gaining from serendipity (which is a proven critical factor in scientific enquiry – the discovery of penicillin being a case in point).

Banning HARKing could hinder scientific enquiry

On many occasions, HARKing is logical, an enabler in the search for truth, and thus may be applied under certain circumstances. Insistence on a rigid ban pushes scholars to be less than truthful in reporting the realities of their research processes and outcomes. Scholarship should not ignore reality nor avoid finding true knowledge.

Ethically and morally, confining scholars to a practice that blocks them from new discoveries is wrong. Full stop. After all, an outright ban on HARKing could mean discarding new knowledge that could be published. The current “solution” to this is misreporting or concealing elements of the process. Following the rule to the letter also poses a barrier to knowledge development. In terms of academic careers, publishing is a critical factor for academic progress, and banning HARKing might hinder promotions. It is, therefore, tempting to bypass or ignore the rule that the academic community cannot practically adhere to.

Unless the system can change to accept HARKing and PARKing under specific circumstances, these practices will continue to take place in a hidden form. This would be unfortunate because such an approach means bending academic moral guidance. Honest and transparent reporting is a key factor – for example, indicating what the original hypotheses were, what new or revised hypotheses (post-hoc) were added and why that was done.

Let us do that when (and only when) it is justified and appropriate.

Yehuda Baruch is a professor of management at Southampton Business School at the University of Southampton.

If you’d like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.

Read the full open-access paper “HARKing can be good for science: why, when, and how c/should we hypothesizing after results are known or proposing research questions after results are known”, published in Human Resource Management Journal.