 Hello, I'm Rosalyn Edwards, and in this video I'll explain briefly why qualitative research isn't biased. There are two key takeaway messages. Qualitative research isn't biased just because data collection and analysis is not standardised, and qualitative research isn't biased just because it can't be generalised in the quantitative sense. This is a quote from a researcher, Gail Leatherby, from a book she co-authored on objectivity and subjectivity in social research. You may want to pause the video at this point to read it. Gail is saying that her students often misuse or overuse the terms bias, value free and objective when they're talking about qualitative research. She's indicating that she's fed up with the unthinking assumptions being made. It's a really important point that she's making. Bias is often used as a term in social research to cover any form of influence that's felt to provide some form of distortion of research data and findings. Mainly it's used in the quantitative paradigm. In qualitative research, the researcher is an integral part of the process and the final product. Separating the research out from this just isn't possible, and also it's not desirable. The concern for qualitative research is whether the researcher has been critically self-reflective about their own preconceptions, relationship dynamics and analytic focus. Reflexivity and transparency are central to rigor in the qualitative paradigm, so good qualitative research doesn't even try to achieve the personalised opinion-free neutrality. Rather, it articulates the unique value that qualitatively derived knowledge can provide. I'm going to delve into these issues to explain why qualitative research isn't bias taking each of the two key messages in turn. First, qualitative research isn't bias just because data collection and analysis is not standardised. There's no such thing as a view from nowhere. In this sense, some sort of bias is inherited all research and it can't be eradicated or controlled. That's because we're human beings and we're insiders with perspectives on the social worlds that we study. Both quantitative and qualitative data are produced through some form of social interaction, even remotely, between the researcher and the research participants. All research involves subjectivity and interpretation, both qualitative and quantitative. Facts aren't self-generating or self-interpreting. In quantitative research, different research teams may choose to apply equally defensible but different statistical methods of analysis to the same data set to address the same research question. In one study, 29 different teams of quantitative researchers separately analysed a shared data set to answer a simple research question. The statistical analytic approaches they chose to apply varied, and so did the answers they produced. Neither the researchers' prior beliefs about the topic nor their level of expertise explained the variation in outcomes. And the team processes were all pirated for quality of the statistical analysis. So, subjective analytic choices influence analysis of representative data sets, but more than this, steps to eliminate all bias and values in research can create bias of its own. Michael Kelly and clinician colleagues discuss evidence-based approaches to medicine. They argue that Michael's science is value laden. Trying to remove values from a scientific method is impossible. And further, researchers who tried to do this introduced new hidden biases. The clinicians point to the way that the values held by researchers and funders play a role in deciding which questions to ask. They look at the way that values for clinical research just mean that they select particular methods for identifying and appraising research evidence, and how that determines the nature of the knowledge that's produced. So, Kelly and colleagues provide the example of evidence-based medicine. Clinical efficiency and cost-effectiveness need to be demonstrated before new treatments are publicly funded. But as they point out, preferences for efficiency and value for money are value preferences. They're not scientifically neutral and dispassionate observed matters of fact. So, an unacknowledged bias towards utilitarian values is hidden away in the effort to have supposedly objective decision-making criteria. Moving on to the second key message, qualitative research isn't biased just because it can't be generalized in the quantitative sense. Qualitative research explores processes and meanings. It generates generalizable in-depth insights, not facts and figures that are generalizable. It's a misunderstanding to claim that qualitative research lacks generalizability. The statistical types of generalizability that inform quantitative research aren't that inform quantitative research. Those types of generalizability aren't applicable to judge qualitative research. Generalizations can be made from qualitative research thus not in the same way as quantitative results. Brett Smith has suggested four different types of qualitative generalizability. One is naturalistic, where people recognize the research results as capturing their own and others' experiences. Another type is transferability, which is the extent that results are transferable to other settings and contacts. There's also analytic generalizability, where the concepts generated through the research are useful for understanding more widely. It's the theories that are generalizable, not the specific context of populations. And finally, there's intersectional generalizability, which applies to research that records the particulars of historically oppressed or colonized people and communities and their social movements of resistance. And as I said at the start, rigoring qualitative research is being reflexive and making clear the how and why of the research process. So here are some references where you can follow up further on the two key messages about why qualitative research isn't biased. Qualitative research isn't biased because it's not standardized. There's no such thing as a view from nowhere. And it isn't biased because it's not generalizable in the quantitative sense. It's generalizable in terms of recognition, transferability, conceptualization, and purpose within the qualitative paradigm.