Quali vs Quanti: How to use them? Part II
In the previous post, we have discussed that both qualitative and quantitative techniques should be employed to obtain the best possible results. Part 1 was exploring how we can enrich our words with numbers. Now, we will see how we enrich our numbers with words.
Enrich your numbers with words
1. Use qualitative techniques to develop a hypothesis and build a model
All quantitative work starts with a mental model. Sometimes, this mental model is very explicit, and the researcher will put forward its underlying assumptions. Other times, the model is more concealed. Nonetheless, a model is always present. Even if you use random control trial data, you will still make assumptions by selecting which variables you want to collect. Conceptually, qualitative knowledge is thus the perfect place to start quantitative research.
2. Account for large deviations from the norm
As seen in part 1, it is convenient to have a good idea of the base rate for what you are studying. But there can be too much of a good thing. If you stick too close to your base rate, you might miss unusual deviations to the norm (aka Black Swans). And anticipation is more about detecting large and infrequent deviations from the norm than regular events. Here, qualitative work can be useful in the form of thought experiments (link) such as semi-factuals (even-if), counterfactuals (what if), and prefactuals (if-then). These tools are instrumental in scenario building. Moreover, concept-building makes anticipations much more agile.
This approach works best when re-combined with another quantitative approach. For instance, scenarios about how and when a coup will happen could be generated qualitatively by a team of experts. However, the different scenarios would be quantitatively evaluated (through an anticipation market) to discard the ones that make no sense. This way, decision-makers can concentrate on and prepare for the most likely coups.
Conclusion: Qualitative (words) and quantitative (numbers) are often opposed. However, an analyst who wishes to be “less wrong” should use the two. An effective approach to mix them is to i) quantify your qualitative work to detect trends within a large amount of qualitative pieces ii) apply the logical rigor and the precision of numbers to your qualitative pieces and iii) use qualitative (expertise) approaches to design a minima model and detect large deviations from the norm.
In a soon-to-come blog post, we will also explore a third way, which is often forgotten in the quali vs quanti war: visualization (drawing).