Document Type


Journal/Book Title/Conference

The Journal of Wildlife Management






John Wiley & Sons, Inc.

Publication Date


First Page


Last Page



Traditionally, most scientists accepted reductionist and mechanistic approaches as the rigorous way to do science. Sells et al. (2018) recently raised the argument about reliability in wildlife science. Chamberlin (1890), Platt (1964), Romesburg (1981, 1991, 2009), and Williams (1997) were rightly referenced as very influential papers. My intention in this letter is not to refute the essence of the Sells et al. (2018) commentary but to add seldom addressed but important aspects that influence the attainment of rigor and certainty in wildlife studies. The elements of a rigorous approach (i.e., strong inference) as described by Platt (1964) included devising alternative hypotheses, devising ≥1 crucial experiments that will exclude ≥1 of the hypotheses, and carrying out the experiment to get a clean result. The process was then repeated using logical inductive trees (i.e., a continually bifurcated statement hypotheses approach) to obtain the essential cause for the effect. Platt (1964) agreed with Popper (1959) that science advanced only by disproof. He argued that this was a hard doctrine and leads to disputations between scientists, but that Chamberlin's (1890) method of multiple working hypotheses helped to remove that difficulty. Platt (1964) emphasized inductive inference and crucial and critical experiments whereby alternate hypotheses are refuted. Romesburg (1981) explained that in wildlife biology, induction (reliable associations) and retroduction (developing hypotheses) were the basis for almost all wildlife research but were not sufficient. He proposed the hypothetical‐deductive (H‐D) method as a more reliable approach. Citing Harvey (1969), and Popper (1962), Romesburg (1981:294) explained that “Starting with the research hypothesis, usually obtained by retroduction, predictions are made about other classes of facts that should be true if the research hypothesis is actually true.” The hypothesis is then tested indirectly by using logic to deduce one or more test consequences (Romesburg 2014). Data are then collected in a statistical framework. Romesburg (1981) distinguished between a research hypothesis (i.e., a conjecture about some process) versus a statistical hypothesis (i.e., a conjecture about classes of facts encompassed by the process). Williams (1997) clearly explained the differences between necessary and sufficient causation and gave examples of the coherent logic both entailed. He summarized that the science endeavor included theory, hypotheses, predictions, observations, and comparison of predictions against data, and argued that inductive and deductive logic were required for testing hypotheses. Importantly, Williams (1997:1014) recognized that wildlife biology often involves simultaneous complementary explanatory factors, requiring “the framing of many scientifically interesting issues about cause and effect in terms of the relative contribution of multiple causal factors.” Over the years, many others have addressed the issue of rigor and reliability in the Journal of Wildlife Management (JWM) and the Wildlife Society Bulletin (WSB) either directly (McNab 1983, Eberhardt 1988, Anderson 2001) or indirectly (Steidl et al. 1997, Guthery et al. 2001). This is not a complete list and is limited primarily to JWM and WSB but gives an idea of the wide interest in achieving reliable results from wildlife studies.


This is the pre-peer reviewed version of the following article: Bissonette, J. A. (2019), Additional thoughts on rigor in wildlife science: Unappreciated impediments. Jour.Wild. Mgmt., 83: 1017-1021. doi:10.1002/jwmg.21677, which has been published in final form at This article may be used for non-commercial purposes in accordance with Wiley Terms and Conditions for Use of Self-Archived Versions.