DeltaMV

Sampling 101: Who You Talk to Determines What You Learn 

By DeltaMV Insights TeamReviewed by DeltaMV Academy EditorialUpdated 7 min read

Module 2 · 4 of 7Market Research Methodology & Behavioural Science FrameworksAll articles
  1. What Is Behavioural Science, and Why Does It Matter in Healthcare? 
  2. Describe, Understand, Shape: DeltaMV’s Framework for Turning Data Into Decisions 
  3. Designing a Good Research Question (Projects Often Fail Here) 
  4. Sampling 101: Who You Talk to Determines What You Learn 
  5. Make It Emotional, Make It Social, Make It Easy: DeltaMV’s Approach to Behaviour Change 
  6. Health Belief Model & Theory of Planned Behaviour, Explained Simply 
  7. Nudge Theory in Healthcare: Small Changes, Big Impact 

Sampling is the process of choosing which people to include in a study. In healthcare research, where groups are small and hard to reach, this choice shapes everything you learn. This article explains how to define your population, pick a sampling approach, decide how many people you need, and avoid the common traps. 

Key Takeaways

  • A study can only tell you about the people it includes. Even perfect questions give a poor answer if the wrong people are asked. 
  • Start by defining the population, meaning exactly who counts. Then choose a sampling approach to match what you need to learn. 
  • Qualitative studies need enough people to hear the main themes. Quantitative studies need enough to say how common something is. They are sized differently. 
  • Healthcare populations are often small, so a modest sample can cover a large share of the whole group. 
  • This article covers who to talk to. Writing the question is covered in the previous article.

The Process, End to End 

Sampling means choosing a smaller group of people to represent a larger one. The larger group is the population. The smaller group you actually speak to is the sample. 

The aim is simple: what you learn from the sample should tell you something true about the population. That only works if the sample is chosen on purpose, not just whoever is easiest to reach. 

In healthcare, this takes extra care. As Module 1 explains, physicians have little free time, patients may be unwell, and the group you want can be very small. A country may have only a few dozen specialists in one niche area. 

Step by Step 

  1. Define the population. Write down who counts and who doesn’t. “Oncologists” is too broad. “Oncologists who have treated at least five patients with this condition in the last year” is something you can recruit against. Let your research question decide this, not convenience. 
  1. Choose a sampling approach. The right one depends on what you need to do with the results. The table below sets out the main options. 
  1. Decide how many people you need. This depends on whether the study is qualitative or quantitative. See the next section. 
  1. Write a screener. A screener is a short set of questions that checks whether someone qualifies. It turns your population definition into yes/no criteria, and it sets any quotas, such as a mix of hospital and private doctors. 
  1. Recruit and monitor. Watch who is actually coming through. If the first ten interviews are all from one city or one type of clinic, adjust the recruitment early. 
  1. Record the limits. At the end, state clearly who was included and who wasn’t. A reader should know what the findings can and can’t speak for. 

Choosing a Sampling Approach 

Approach How it works Good for Watch out for 
Random Everyone on a full list has an equal chance of being picked. Quantitative studies where a complete list exists. Complete lists of doctors or patients are often not available. 
Stratified Split the population into groups, then sample within each group. Making sure small but important groups are included. Needs enough people in each group. 
Quota Set a target for each group (for example, 20 hospital and 20 private doctors) and fill it with people who qualify. The most common approach in healthcare studies. Within a quota, recruiters may pick the easiest people. 
Purposive Hand-pick people who match specific criteria, such as high-volume prescribers. Qualitative studies where specific voices matter. Choices can reflect the researcher’s assumptions. 
Snowball Ask participants to introduce others like them. Hard-to-reach groups, such as people with a rare condition. Tends to reach people who are well connected. 
Convenience Whoever is easiest to reach. Quick early exploration only. Rarely safe to base a decision on. 

How Many People Do You Need? 

Qualitative studies aim to hear the full range of views. A sample is large enough when new interviews stop producing new themes. In a fairly uniform group, research suggests that most themes appear within about a dozen interviews.1 If you are speaking to several different groups, such as hospital doctors, GPs, and patients, plan for roughly that many in each. 

Quantitative studies aim to say how common something is. Here, more people give a narrower margin of error. At 95% confidence, and for a large population, the approximate margins are: 

Sample size Margin of error (large population) Margin of error (population of 150) 
30 about ±18 points about ±16 points 
50 about ±14 points about ±11 points 
100 about ±10 points about ±6 points 
400 about ±5 points not applicable: more than the whole population 

This is why healthcare research can feel different. If there are only 150 relevant specialists, then speaking to 100 of them covers two-thirds of the whole group, and the answer is much firmer than the same number from a general consumer population. These figures assume the people who answered are typical of those who didn’t. 

One more point. The sample has to be large enough for the comparisons you plan to make, not just the total. A group of 300 looks solid until it is split by market and specialty. Module 1 shows how quickly that can happen. 

Common Mistakes 

  • Only talking to people who are easy to reach. Well-known doctors, existing customers, and enthusiastic respondents are usually the first to say yes. They are rarely typical. 
  • Leaving out the people who left. If you want to know why patients stop treatment, speaking only to current patients will miss the answer entirely. 
  • Defining the group by convenience. “The doctors on our list” is a population defined by your own past activity, not by the question. 
  • Treating sample size as a measure of quality. A large sample of the wrong people is still the wrong sample. 
  • Ignoring small subgroups. Splitting a sample too many ways leaves groups too small to say anything reliable about. 
  • Skipping the screener checks. Without clear criteria, people who don’t really qualify can get through, and their answers dilute the findings. 

A Checklist to Use Right Away 

  • Can you describe, in one sentence, exactly who is in the population? 
  • Does that definition come from the research question, not from who is easy to reach? 
  • Have you chosen a sampling approach and can you say why? 
  • Is the sample size set for the comparisons you need, not just the total? 
  • Does the screener list clear qualifying and excluding criteria? 
  • Have you included people who stopped, switched, or never started, where it matters? 
  • Will the report state who was and wasn’t included? 

The Cheat Sheet

A study can only tell you about the people it includes. Decide exactly who counts, choose a way of picking them that fits your question, and size the sample for the comparisons you need. Don’t only speak to people who are easy to reach, and be open in the report about who was left out. 

A DeltaMV Perspective

A pattern worth illustrating: a brand wants to understand why patients stop its treatment. To save time, it recruits through its own patient support programme. 

The interviews are warm and positive. Patients praise the nurse line and the reminders. The team concludes that the main problem lies elsewhere. 

But everyone in the sample is, by definition, still on treatment and still enrolled in the programme. The people who stopped are not in the room. 

The fix is to broaden the population on purpose: add a group who started treatment and later stopped, even though they are harder to find and slower to recruit. The answers change, and so does the plan. 

Composite/illustrative example; not a specific client or project.

FAQs 

Is a bigger sample always better? 

No. A bigger sample narrows the margin of error, but only if the right people are in it. A small, well-chosen sample often beats a large, poorly chosen one. 

Can twelve interviews really tell me anything? 

Yes, if the goal is to understand the range of views in a fairly similar group. It is not enough to say how many people hold a view. That is what a quantitative study is for. 

What is a screener? 

A short set of questions that decides whether a person qualifies for the study. Good screeners use clear, factual criteria, such as years in practice or recent patients treated. 

What if my population is tiny? 

Then aim to cover as much of it as you can. With a very small group, a census (speaking to nearly everyone) can be more practical than sampling. State the numbers openly. 

The Bottom Line

Who you talk to sets the limit on what you can learn. Define the population from the question, choose an approach on purpose, and be honest in the report about who was and wasn’t included. 

Sources 

  1. Guest, G., Bunce, A., & Johnson, L. (2006). How many interviews are enough? An experiment with data saturation and variability. Field Methods, 18(1), 59–82. https://doi.org/10.1177/1525822X05279903

What are you trying to change?

Let's explore the behavioural insight behind it.

Connect with us