State of the Field

Much has been written on the topic of educational inquiry, but the literature on educational inquiry and research methods per se continues to suffer from four major limitations. First, because research is viewed largely as a “prescribed” journey in which applying appropriate methods and techniques is the sine qua non of first-rate scholarship, developing meaningful knowledge and understanding—ironically, the goal of inquiry—is not placed at the forefront of inquiry. In turn, most texts on research methods not only fail to address adequately the context of inquiry—including animating purposes and key stakeholders—but also give woefully little consideration to identifying and exploring potentially fruitful research problems. Second, much of the literature fails to help prepare researchers to recognize and address the most fundamental challenges—before, during, and following research—in conducting inquiry. Third, qualitative and quantitative approaches are bifurcated rather than integrated into discussions of inquiry. Fourth, much of the literature advances a “one best way” approach to research that falls short of exploring a wide range of alternative and emerging perspectives on ways in which to enrich inquiry.

It is not clear whether assigning our students to discern from the research literature methods for attaining questions of interest avoids these four deficiencies. The articles and textbooks they choose to read in the pursuit of their goal will not likely reveal the motivations underlying the course of study, the failed attempts and dead ends that are more prevalent than successes in research, or especially the thought processes that lead to successes despite the failures. And even if some of these aspects are dealt with in chapters and articles, it is likely that the presentation will make it seem as if the journey from inception to conclusion was, at least in hindsight, the only logical way in which one could have proceeded.

Inquiry Through a Keyhole: Retroduction

It might seem, from our description of our assignment to our students, that we have asked them to apply the well-known method of induction in the course of coming up with questions of interest from a reading of the literature. For instance, they could amass a set of questions that seem to have motivated various experiments, look for possible patterns in these questions that might have made them interesting to pursue, and assert by induction that the pattern that seems to have held regarding these particular questions holds in general. Or, because there is an extremely large number of ways of characterizing the question of interest in even a single experiment, the students could try to infer which of the many aspects was the key factor and generalize that this would be so in other experiments as well. Or they could find common elements among various explanations of the potential practical or theoretical value of studies and assert by induction a single theoretical explanation of the many claims to importance.

The process of formulating theories by induction from observations has been commonly held to be the generative mechanism of science since Aristotle's Organon. In the preface to Novum Organum, Bacon described a method of induction in which the scientist would question nature without biases or hypotheses and move to generalities in an algorithmic fashion, as if (in Bacon's words) by machinery. Broad (1952) described Baconian induction as “the glory of science” (p. 143). According to Thomas Reid, who popularized the works and methods of Isaac Newton during the 18th century, the success of Newton's theories led to the wide acceptance of Bacon's philosophy, for Reid (1785) claimed that Newton's methods were inductive. Indeed, Newton's famous assertion that he did not formulate hypotheses, implying that his theories were derived inductively from observation, seemed to substantiate Newton's methods as inductive.

Baconian inductive methods were actually more sophisticated than it might seem at first glance. Briefly, Bacon advocated listing all instances in which a characteristic under examination is present as well as listing the concomitant characteristics in these circumstances. He then proposed listing similar circumstances in which the characteristic under examination is not present to see, in contrast, the effects of removing the characteristic (in this manner, conducting a controlled experiment). Finally, he suggested studying the characteristic to be examined in conditions where it is present in varying degrees. This method was called “eliminative induction” by von Wright (1951).

That induction forms a crucial step in the scientific process is a view held long into the 20th century. Indeed, the scientific method as formulated by Braithwaite (1953) was depicted as cyclical, alternating between inductive and deductive phases. The view that science has an inductive phase continues to be surprisingly influential given that it has been strongly contraindicated, even from Bacon's own time. For instance, William Harvey, a contemporary of Bacon's and the person who discovered the body's circulatory [Page xii]system, said that Bacon wrote philosophy like a chancellor. This was interpreted by Nobel Prize winner Richard Feynman to indicate that while Bacon spoke of gathering observations, he omitted the factor of judgment regarding what to observe and what to attend to in gathering observations. And as Broad noted in virtually the same breath as he extolled induction as the glory of science, it is also “the scandal of philosophy” (Broad, 1952, p. 143).

David Hume, a contemporary of Reid, argued devastatingly that no inductive generalizations whatsoever can be logically justified, whether in pursuit of certain or probable knowledge, unless a principle of induction is presumed to hold. This principle itself, which can be variously stated as “the future will resemble the past” or as “nature is uniform,” can be derived only inductively and so begs the question. Of Hume's conclusion, Russell (1945) exclaimed,

William Whewell attempted to explain the growth and stability of scientific knowledge without requiring Baconian induction. In his Novum Organon Renovatum, Whewell (1858) defined induction as the representation of facts with principles rather than as a generalization from facts. He argued that to explain the growth of scientific knowledge, both experience and intuition were required. Most important, he claimed that to account for how scientists discover true principles, science needs guesses. According to Wettersten (1993), Whewell's theory makes clear that “even if we start with poor guesses and treat them critically, we can come to the truth; there are many paths to the truth, but only one goal” (p. 506). Admittedly, the word “guess” is an infelicitous choice. As Medawar (1974) explained,

This view of what could be called a scientific method was also elucidated by Charles Sanders Peirce during the late 19th century. According to Peirce, all inference is either deductive or synthetic, with the latter typology subdivided into induction and what Peirce variously called hypothesis, retroduction, or abduction (Peirce, 1878). As Peirce (1958) described it, when a scientific explanation is sought,

The final step in this process is when the deduced consequences are compared with experimental results.

Feynman (1965) reiterated the notion that scientists employ guesswork, saying that a new law is first guessed, its consequences are then computed assuming that it is a correct guess, and finally the results of the computation are compared with nature. If it disagrees with experiment, then it is wrong, Feynman concluded. Feynman wrote,

For Whewell, Peirce, and Feynman, then, theories and hypotheses are not obtained from observations by induction. Rather, by some imaginative retroductive leap, a hypothesis is guessed that yields the observations as deductive consequences of the hypothesis.

Shank (1990) argued that, based on Peirce's insights into retroduction and his semi-otic theory, the conceptual confusions created in maintaining the apparent qualitative-quantitative dichotomy can be resolved. Shank contended that the same logical processes are involved in a case study examined in a qualitative framework as in the quantitative perspective within which Peirce worked, so that the two perspectives differ in the ways in which they constitute units of analysis, but not in terms of the logic of the analysis. Common to the two perspectives are implementations of deduction and retroduction for the purposes of producing and testing explanatory hypotheses.

As Shank (1990) pointed out, it is not possible to observe and take note of all aspects of the phenomenon under study, and the number of possible explanatory hypotheses that could be imagined is virtually limitless. According to Hoffmann (2000), however, there is a relationship between the context of retroduction and the process of attaining a promising hypothesis, so that the range of explanations is limited by a complex interaction among factors at play in the given circumstance.

So it is retroduction, and not induction, that we are asking our students to perform in the course of generating questions of interest. Furthermore, they must formulate these questions within the limits imposed by aspects of the field of study, including the state of the art of theory, political realities, and funding possibilities, as well as by the circumstances in which the students find themselves, possibly involving the purposes of the studies in terms of the students' academic careers, for whom the studies may be conducted, and the ability of the students.

At a minimum, these actualities raise the question of how best to teach students how to perform retroduction, with the hope that the result of teaching them will be a good question of interest and an able scientist. From at least the time of Piaget and Ausubel during the 1960s to today's constructivism and situated learning, theory suggests that to teach abstract abilities such as retroduction, instruction must actively engage students in the process itself, but in such a manner that they are able to carry out a successful retroductive episode. So we send our students out to the literature to learn about context and limitations, extant theories, and previously tried but failed explanations. This experience can provide students with a powerful schema that will help them to incorporate the desired skills into their own inquiry that we are seeking to model in our respective classrooms.

Medawar (1974) cautioned, however, that scientific papers “actively misrepresent the reasoning that goes into the work” (p. 287) because it seems that publishable papers must be written as if inductions had occurred. It is also not sufficient to listen to what scientists say they do because their opinions will vary so widely. Medawar concluded that “only unstudied evidence will do—and that means listening at a keyhole” (p. 287).

It is our hope that this Handbook, by presenting detailed accounts of successful episodes of scientific activity in a variety of fields, from a variety of perspectives, in a manner that highlights the difficult exploration of ideas and actively involves the readers, will provide Medawar's necessary keyhole through which the readers can peer in the course of experiencing successful retroductive efforts regarding the engagement of ideas.