Chapter 7. Quantity and quality — two kinds of attention

Key points

• Data comes in two kinds: quantitative (numbers) and qualitative (accounts)
• Numbers are strong at generalising; accounts are strong at digging into "why"
• AI accelerates the collection and organising of both, and combining them is strongest

What numbers alone let slip

So far we have mostly talked about measuring in numbers. In reality, a great deal of important information cannot be turned into a number.

Why does this person not keep going with a programme that ought to work? Why is this service chosen despite being the weaker one? The answers to those "why" questions are not in the scores. They are in what people say.

Social science divides how information is handled broadly in two.

• Quantitative research — handling numerical data statistically, aiming at generalisation. Objectivity and reproducibility come first.
• Qualitative research — reading words, behaviour and context, digging into how people make meaning. Diversity and context come first.

They are not opposed; they complement

There is no need to line up numbers against accounts. They complete each other.

• Quantitative data captures what, and how much, broadly and objectively (mean satisfaction, retention rate)
• Qualitative data captures why, deeply and in context (the account of someone who kept going)

Identify the low-satisfaction group quantitatively in a survey (broad grasp), then interview that group about why they are dissatisfied (deep understanding). This movement from broad to deep produces the richest insight.

Is this not a familiar structure for a therapist? Assessment scores alone do not tell you the patient. Only combined with the history and their account of daily life does the whole picture appear. You have trained in integrating quantity and quality, in the clinic.

AI has transformed qualitative work

What AI has changed most in recent years is exactly this: handling qualitative data.

Transcribing interviews, organising free-text responses, analysing social media posts — all of it used to take enormous time. It was territory where "I know I should, but there are not enough hands". AI changed that completely.

• transcription of recordings
• extracting shared themes from large volumes of free text
• analysing sentiment and topics in social posts and reviews

AI makes all of it dramatically faster. Information given up on because it could not be turned into numbers is now within reach.

Going further: automatic collection with APIs and databases

Distributing surveys and typing data in are not the only ways to collect. As an extension of Coding and Building Systems, you can use APIs to gather data from the web automatically.

Further, if you prepare a back end — a Google Sheet, or a database such as Supabase — and build a mechanism that collects and accumulates data directly there, everything downstream becomes dramatically easier. With AI, the back-end code for such API connections and database links is surprisingly simple to implement. Widening your development knowledge a little, to build something that accumulates data by itself, is well worth considering.

From the next chapter we look concretely at collecting quantitative and qualitative data.

A prompt to try

You are a researcher who knows mixed methods well. I want to investigate (a theme — e.g. retention among users of self-funded rehabilitation). Propose a concrete research flow: how to use a quantitative approach and a qualitative approach so that I grasp things broadly and understand them deeply. Include the order — what to look at quantitatively first, and whom to interview afterwards.

Chapter 7 summary

• Data comes in two kinds: quantitative and qualitative
• Numbers answer "what and how much"; accounts answer "why"; they complement each other
• AI has transformed qualitative work; information once given up on is now within reach
• Automatic collection via APIs and a back end (Supabase and the like) makes everything downstream easier