1-2 The Nature of Evidence: Inductive and Deductive Reasoning

Overview

Different philosophies of science (Chapter 1-1) inform different approaches to research, characterized by different goals, assumptions, and types of data. In this chapter, we discuss how post-positivist, interpretivist, constructivist, and critical philosophies form a foundation for two broad approaches to generating knowledge: quantitative and qualitative research. There are several often-discussed distinctions between these two approaches, stemming from their philosophical roots, specifically, a commitment to objectivity versus subjective interpretation, numbers versus words, generalizability versus transferability, and deductive versus inductive reasoning. While these distinctions often distinguish qualitative and quantitative research, there are always exceptions. These exceptions demonstrate that quantitative and qualitative research approaches have more in common than what superficial descriptions imply.

Vignette

Rayna (they/their) stared at the blinking cursor on her screen. They had been recently invited to revise and resubmit their first qualitative research manuscript. Most of the editor and reviewer comments had been relatively easy to handle, but when Rayna reached Reviewer 2’s comments, they were caught off guard. The reviewer acknowledged that they came from a quantitative background, but then went on to write: “I am worried about the generalizability of this study, with a sample of only 25 residents. Wouldn’t it be better to do a survey to get more perspectives?”

Rayna hadn’t seen any other qualitative studies talk about generalizability, so they weren’t entirely sure how to address this comment. Maybe the study won’t be valuable if the results aren’t generalizable! Reviewer 2 might be right that the sample size is too small! They immediately panic-email one of their co-authors:

With a sigh of relief, Rayna reads through the attached papers (1-8) and continues her mission to craft a response.

Deeper Dive on this Concept

Qualitative and quantitative research are often talked about as two different ways of thinking and generating knowledge. We contrast a reliance on words with a reliance on numbers, a focus on subjectivity with a focus on objectivity. And to some extent these approaches are different, in precisely the ways described. However, many experienced researchers on both sides of the line will tell you that these differences only go so far.

Drawing on Chapter 1-1, there are different philosophies of science that inform researchers and their research products. Generally, quantitative research is associated with post-positivism; researchers seek to be objective and reduce their bias in order to ensure that their results are as close as possible to the truth. Post-positivists, unlike positivists, acknowledge that a singular truth is impossible to define. However, truth can be defined within a spectrum of probability. Quantitative researchers generate and test hypotheses to make conclusions about a theory they have developed. They conduct experiments that generate numerical data that  define the extent to which an observation is “true.” The strength of the numerical data  suggests whether the observation can be generalized beyond the study population. This process is called deductive reasoning because it starts with a general theory that narrows to a hypothesis that is tested against observations that support (or refute) the theory.

Qualitative researchers, on the other hand, tend to be guided by interpretivism, constructivism, critical theory or other perspectives that value subjectivity. These analytic approaches do not address bias because bias assumes a misrepresentation of “truth” during collection or analysis of data. Subjectivity emphasizes the position and perspective assumed during analysis, articulating that there is no external objective truth, uninformed by context. (See Chapter 1-1.) Qualitative methods seek to deeply understand a phenomenon, using the rich meaning provided by words, symbols, traditions, structures, identity constructs and power dynamics (as opposed to simply numbers). Rather than testing a hypothesis, they generate knowledge by inductively generating new insights or theory (i.e. observations are collected and analyzed to build a theory). These insights are contextual, not universal.  Qualitative researchers translate their results within a rich description of context so that readers can assess the similarities and differences between contexts and determine the extent to which the study results are transferable to their setting.

While these distinctions can be helpful in distinguishing quantitative and qualitative research broadly, they also create false divisions. The relationships between these two approaches are more complex, and nuances are important to bear in mind, even for novices, lest we exacerbate hierarchies and divisions between different types of knowledge and evidence.

As an example, the interpretation of quantitative results is not always clear and obvious – findings do not always support or refute a hypothesis. Thus, both qualitative and quantitative researchers need to be attentive to their data. While quantitative research is generally thought to be deductive, quantitative researchers often do a bit of inductive reasoning to find meaning in data that hold surprises. Conversely, qualitative data are stereotypically analyzed inductively, making meaning from the data rather than proving a hypothesis. However, many qualitative researchers apply existing theories or theoretical frameworks, testing the relevance of existing theory in a new context or seeking to explain their data using an existing framework. These approaches are often characterized as deductive qualitative work.

As another example, quantitative researchers use numbers, but these numbers aren’t always meaningful without words. In surveys, interpretation of numerical responses may not be possible without analyzing them alongside free-text responses. And while qualitative researchers rarely use numbers, they do need to think through the frequency with which certain themes appear in their dataset. An idea that only appears once in a qualitative study may have great value to the analysis, but it is also important that researchers acknowledge the views that are most prevalent in a given qualitative dataset.

Vignette Conclusion

Rayna smiled and read over their paper once more. They had included an explanation of qualitative research, and how it’s about depth of information and nuance of experience. The depth of data generated per participant is significant, therefore fewer people are typically recruited in qualitative studies.  Rayna also articulated that while the interpretivist approach she used for analysis isn’t focused on generalizing the results beyond a specific context, they were able to make an argument for how the results can transfer to other, similar contexts. Cal provided a bunch of margin comments in her responses to the reviews, highlighting how savvy Rayna had been in addressing the concerns of Reviewer 2 without betraying their epistemic roots. The editor certainly was right – adding more justification and explanation had made the paper stronger, and would likely help others grow to better understand qualitative work. Now the only challenge left was figuring out how to upload the revised documents to the journal submission portal…

References

1. Wright, S., O’Brien, B. C., Nimmon, L., Law, M., & Mylopoulos, M. (2016). Research Design Considerations. Journal of Graduate Medical Education, 8(1), 97–98. doi: 10.4300/JGME-D-15-00566.1
2. Monteiro S, Sullivan GM, Chan TM. Generalizability theory made simple (r): an introductory primer to G-studies. Journal of graduate medical education. 2019 Aug;11(4):365-70. doi: 10.4300/JGME-D-19-00464.1
3. Goldenberg MJ. On evidence and evidence-based medicine: lessons from the philosophy of science. Social science & medicine. 2006 Jun 1;62(11):2621-32. doi: 10.1016/j.socscimed.2005.11.031
4. Varpio L, MacLeod A. Philosophy of science series: Harnessing the multidisciplinary edge effect by exploring paradigms, ontologies, epistemologies, axiologies, and methodologies. Academic Medicine. 2020 May 1;95(5):686-9. doi: 10.1097/ACM.0000000000003142
5. Morse JM, Mitcham C. Exploring qualitatively-derived concepts: Inductive—deductive pitfalls. International journal of qualitative methods. 2002 Dec;1(4):28-35. https://doi.org/10.1177/160940690200100404
6. Armat MR, Assarroudi A, Rad M, Sharifi H, Heydari A. Inductive and deductive: Ambiguous labels in qualitative content analysis. The Qualitative Report. 2018;23(1):219-21. https://www.proquest.com/docview/2122314268
7. Tavakol M, Sandars J. Quantitative and qualitative methods in medical education research: AMEE Guide No 90: Part I. Medical Teacher. 2014 Sep 1;36(9):746-56. doi: 10.3109/0142159X.2014.915298
8. Tavakol M, Sandars J. Quantitative and qualitative methods in medical education research: AMEE Guide No 90: Part II. Medical teacher. 2014 Oct 1;36(10):838-48. doi: 10.3109/0142159X.2014.915297

About the Authors

name: Renate Kahlke

institution: McMaster University

Renate Kahlke (she/her) is an Assistant Professor within the Division of Education & Innovation, Department of Medicine. She is a scientist within the McMaster Education Research, Innovation and Theory (MERIT) Program.

name: Jonathan Sherbino

institution: McMaster University

Jonathan Sherbino (he/him) is a Professor in the Department of Medicine, Division of Education and Innovation. He is also the Assistant Dean, Health Professions Education Research. He is a researcher and clinician scientist in the McMaster Education Research, Innovation and Theory (MERIT) program.

name: Sandra Monteiro

institution: McMaster University

Sandra Monteiro (she/her) is an Associate Professor within the Department of Medicine, Division of Education and Innovation, Faculty of Health Sciences, McMaster University. She holds a joint appointment within the Department of Health Research Methods, Evidence, and Impact, Faculty of Health Sciences, McMaster University.