Is n=1 ever enough?

Three Dog Night was wrong. One is not the loneliest number. In fact, one is getting a lot of action these days.

As a researcher who primarily works in the qualitative space, the question of whether an n=1 is ever enough really intrigues me. So, for the past few months, I’ve posed the question to colleagues and clients. The answers are rich, diverse, often long, heated conversations.

Simply put, they are:

1. “Never!”
2. “Sure, we do it all the time.”
3. “It depends on the research goals.”

Why even consider this question? Because more and more RFPs ask for a sample plan to accomplish a broad – but thin – recruit, resulting in “skinny qual”. By this I mean filling a cell or quota group with only one or two participants within the broader scope of a research engagement. Our clients are under pressure to satisfy questions about ‘Group A,’ but aren’t given the budget or time to match the request. And so the n size suffers.

Let’s look at the different points of view for a moment: Is n=1 ever enough?

‘Never!’

I notice this is the initial, visceral reaction for most of us, myself included. Some research colleagues might even question the character of anyone who proposes this suggestion. Since qualitative research isn’t statistically representative, n=1 just seems wrong, even to scratch the surface.

Those firmly opposed to n=1 likely have a strong opinion of what a valid n actually is, and they have either deep experience, education or both to validate their point of view. My colleague, Jack Fyock, says n=10 is ideal because it is the size of a legendary social psychology experiment. Yet the requests we get range from 1-36 – and are rarely 10.

The bottom line for this group: Do none, or do more, but n=1 is a disservice to research and will never sufficiently inform your business decision.

‘Sure, we do it all the time’

This is still not a common position, but some are comfortable with it. Reduced budgets and timelines may drive this reality, but changing methods and more advanced targeting techniques may also lead this group to believe they can hone in on the one key participant who actually is the voice of a homogeneous group. For example, if a health care insurer has very clearly defined and well-vetted segments with sufferers of a rare disease, a lot is already known about the participant. Carefully screening one of those sufferers to complete a mobile qualitative diary might be enough to inspire confidence for the decision maker.

What I have anecdotally noticed in this group is an ‘expectation setting gap’ between experienced and non-experienced researchers. Those without experience may either dismiss the learnings altogether or frame the n=1 as a more complete story than it really is. Those with experience can easily frame a valid story of n=1 while clarifying why only one voice is heard.

The bottom line for this group: n=1 is better than n=0 if that one is well-recruited. While it’s not the preferred approach, sometimes research has to be completed quickly and cheaply – n=1 makes sense because, at least, researchers have something to satisfy informational thirst. As long as results are reported clearly, this is a start.

‘It depends on the research goals’

After a visceral first reaction, many researchers land on this fairly safe middle ground. We ask questions like:

“What difference do you think the n=1 will make to your business decision?”
Much of the work we do for the pharmaceutical industry involves early-stage compounds and our client’s need to make a decision whether to pursue or abandon R&D and eventual market entry or exit. These decisions are complex, dynamic and carry high stakes for any company. But, when the decision can mean life or death for millions, the ramifications are exponential. An n=1 could not be enough to inform the entire decision, but it might be a valid, directional part of the bigger picture. For instance, I’ve talked to the one, world-renowned Key Opinion Leader (KOL) in a rare disease state who was able to give my client team a lot of rich ideas to consider. But, at the same time, there was other qualitative, secondary research and forecasting feeding into the same decision.

In these early-stage explorations, qualitative is often one of the starting points, but it is far from the only thing happening. If we, as researchers, understand that n=1 is only one facet of a much larger plan, we can learn and report accordingly.

“Do you mean the capital N or the lower case n?”
This is one of the toughest questions as we think about an approach to recommend. When doing research across a continent, it is not at all uncommon to interview one patient or treater in each country of interest in the EU, Africa or Asia. But that is an uncomfortable number, even if the total patients interviewed add up to a comfortable number. This dilemma also ties back to the overall business and research goals: the request comes most often in the very early stages of product/resource development, when a nascent understanding of the state of a disease or therapeutic area is needed — but rarely (if ever) with concept development.

“How big is the universe?”
If the universe consists of your spouse or a child, it’s pretty easy to say n=1 is all you need. The earlier KOL example is another with a small universe. But in many instances, an accurate universe size is elusive. When conducting ethnography with sufferers of cardiovascular disease across 10 countries, the universe is enormous. Visiting one patient home in the largest city of each country certainly tells us a lot, but it severely skews learnings and limits the possibilities of really understanding the life of the sufferer.

The bottom line for this group: Anything is possible, but sampling decisions must be made with a clear focus on the bigger research picture and business decision.

What is the Magic Number?

Brian DiVita, my former boss who is now director of graduate programs at Aquinas College, often shakes his head at what “passes for research” today. He taught me early that the purpose of research is to reduce the risk in decision making. But it cannot make the decisions for us. That is why I’m still waffling. I’d never recommend n=1 as a first choice. I push back when it is suggested. But that doesn’t mean n=1 isn’t valuable. In fact, it’s all we really ever have: The person sitting across from me when I’m moderating is the single most important voice I can hear in that moment. Their reality is both valid and important.

Our challenge as researchers is to make the appropriate recommendations within the context of the research and make our audience aware of the bias in our sample. Only then can they make their own business decisions.

Nicky Halverson is senior research director at Market Strategies International