What is the difference between Qual and Quant?
This is a perennial question and it came up in conversation earlier today. So, here is another go at producing an answer.
The first thing I would stress is that we need to focus on Qual Analysis and Quant Analysis, because the data used is neither a sufficient or necessary criterion.
Let’s start with Quant Analysis because I think the definition is narrower. Quantitative Analysis is where numbers are produced that meaningfully describe something of interest from data. In general, an algorithm is used to produce the output. For example, to find the mean, the values are added up and divided by the number of values used – that is an algorithm. To an extent, we can think of Quant Analysis as ‘discovering’ things (i.e. it is a form of post-positivism). In theory, but often not in practice, if you change the analyst in a quant project you will get the same numbers*.
In Qualitative Analysis the analyst/researcher/insighter combines the data with their insight to join the dots to ‘create’ meaning (i.e. it is a form of Constructivism/Interpretivism). The most common reason for using Qualitative Analysis is that the problem being addressed can’t be answered by simply applying an algorithm to generate numbers. Note, if you change the researcher, you will normally (to some extent) change the interpretation – because the additional information they bring and their specific way of seeing patterns will differ.
Are qual and quant becoming the same thing? No! People quote the example of applying tools such as AI to photos, videos, text etc to produce analyses such as sentiment. This is simply quant analysis. It might be very good, it might fully answer the problem, but it is an algorithm generating numbers.
Is qual unique in addressing the ‘Why?’. Not definitively. For example, you might do an ethnographic study to gain a description of what is happening, but it might not answer the why. Conversely, most of the behavioural economics findings from Kahneman, Tversky, Ariely, Thaler et al (that explain why people do things they do) are based on quantitative experiments and observations. But it is true that why is usually a question asked by qual, and why is often not the key element of quant analysis.
What about AI that could take images, text, videos etc and which produce written answers to the insight problem, drawing on knowledge not included in the data being reviewed. That might be qual, or it might be a new category, perhaps AQ (artificial qual) – but it wouldn’t be quant.
Do all clients need to understand the difference between qual and quant? No! Some clients need to know (because they are making the decision), other clients don’t need to know (but they are interested), and other clients do not need to know and don’t want to know (they just want to know the research is valid and the recommendations useful). However, the person conducting the research, in particular the person choosing the method and interpreting the results, needs to understand the strengths and limitations of the method selected, irrespective of whether it is qual or quant.
* At the top of this post, I said that changing the quant analyst on the project often changes the numbers, which is not what you might expect with quant. That is because a large number of subjective decisions are made by the analyst, including whose data to include/exclude, whether any data transformations are needed, and how to apply iterative analyses (such as cluster analysis). Even decisions about whether to focus on the mean or median are decisions that draw on the analyst’s preferences and experiences.