We create standards, codes, checklists, and best practices to separate good research from poor research. However, we will likely do a bad job unless we recognise the differences between the theoretical underpinnings of different research methods.
The Quantitative Position
Whilst there are variations, the most common epistemological position underpinning quant is the post-positivist position, which Steven Hawking and Leonard Mlodinow described as ‘model-dependent realism’ (The Grand Design). In essence, this holds that the researcher’s job is to discover, describe, and measure phenomena that exist using the precepts of science.
Because the phenomena are assumed to exist, terms such as representativity, accuracy, and unbiased are vital issues. Researchers know that in all non-trivial cases, it is impossible to be representative, accurate, and unbiased, but the researcher seeks to identify, minimise and try to account for any shortcomings in these three. A code that makes quantitative research more representative, unbiased, and accurate is good.
One of the model-dependent characteristics of the world that quant researchers occupy is the need to accept that the phenomena they measure start as qualitative entities. These qualitative entities are made countable by a process of operationalising the data. For example, when we seek to measure satisfaction, we deal with non-stable, non-numeric, poorly defined mental states in the minds of our participants. We are operationalising qualitative fuzziness into a model-dependent quantity by asking people to use a 0-to-10 scale to say how likely they are to recommend this service.
The Qualitative Position
There is a wide range of qualitative epistemologies, for example, critical, post-modern and emancipatory. However, I shall focus on a mainstream one, the constructionist or interpretivism position. The constructionist position assumes that we create things as opposed to discovering them. The constructionist researcher embraces their subjectivity to help create sense and meaning from what they experience.
In the commercial world of market research, the key test of a piece of work is its usefulness. The concept of accuracy is hard to define in commercial qualitative research. The story created by the researcher accurately describes what the researcher perceived and constructed, but it may not be helpful. Consider the example of qual research being used to find a way to improve a TV commercial. Three researchers might come up with three quite different suggestions, each of which might be equally useful. But, if one of the three researchers was more useful than the other two, we would not say it was more ‘accurate’, but we would say it was better.
Talking to the right people is just as crucial in qualitative research as in quantitative research. But, in qualitative research, the term representative is less apposite. For example, we are often interested in edge or liminal cases in qual. In a study looking at people’s challenges with etting food that matches their needs, we might focus on those with multiple needs, e.g., vegan and gluten-free, or kosher and high protein. By contrast, a quantitative study would normally try to reflect a population’s full range of experiences, i.e., be representative.
Codes governing qualitative research, therefore, need to focus less on being representative, unbiased, and accurate and more on how the user can assess the likelihood of the research being helpful. For example, asking what methodology is being used, the practitioner’s experience with that method and the field of research, the rationale for selecting people to research, and how to triangulate these findings with other sources.