In Defense of Qualitative Research

Contributed by:

Brad Davidson, Ph.D. General Manager Ogilvy ­CommonHealth Insights & Analytics, part of Ogilvy ­CommonHealth Worldwide

NOTE: The content below contains the first few paragraphs of the printed article and the titles of the sidebars and boxes, if applicable.

“The science of statistics is the chief instrumentality through which the progress of civilization is now measured, and by which its development hereafter will be largely controlled.” — S. N. D. North, Director of the Census, 1903-1909 The Rise of Statistics It is possible to quantify what people do, but nearly impossible to quantify why; nevertheless, we live in an age in which quantitative approaches to understanding the world reign supreme. Universities, the traditional bailiwicks of research, are well on their way toward marginalizing the humanities and social sciences in favor of the more computational, quantitative — and lucrative — mathematically based fields; economics is being defined by the sorts of statistical approaches found in “Freakonomics;” and even baseball has turned toward statisticians for their insight into how best to select the players for a winning team, (see: Moneyball). A good friend assures me that many in the field of Natural Language Processing have become convinced that statistical approaches are the best for approximating natural language with a computer, and Siri and Watson are the proof of this. And we in the pharmaceutical industry live and die by trial results that prove that new drugs work, or that they don’t, to the most precise p-value available. This point of view has become so dominant that even some qualitative researchers have ceded that quantitative research is more robust. They wouldn’t say it that way, of course, but they say as much every time they begin a report with the words, “This report is qualitative and as such is not projectable…” However, properly executed qualitative research is projectable; it’s just what it predicts that begs the question. This, then, is the age of quantification, where Amazon and Google can predict our future behavior by tracking our tendencies in purchasing and search, and where we seem to have replaced Twain’s quote about statistics with a new one — there are lies, damn lies, and qualitative research findings. With the success of quantitative approaches to all sorts of intractable problems (such as, how can the Red Sox overcome a century-old curse and win the World Series?), and with the power of statistics to discern patterns that would not be identifiable otherwise, how can a qualitative researcher hope to convince brand managers that in lieu of fielding an online quant survey, they might want to talk to a few people first? If we as qualitative researchers can’t even defend ourselves, why should anyone else? The Irreplaceability of ­Qualitative Insights One of the key issues seems to be one of perceived rigor. While quantitative research requires that we be able to write. Quantitative research reads out in absolutes, giving the world a mathematical precision that it lacks in reality (but that is reassuring nonetheless when one is deciding how to allocate thousands, or millions, of dollars); but good qualitative results should make the complex simple, and in some cases seem obvious; this type of research can seem like something that occurred to the researcher while he was taking a shower, or after she viewed a few focus groups. The underlying work that is done to create the findings is obscured, and so qualitative research can seem easy, or even lazy (it’s not). Another key issue is one of belief. When one reads the academic discussions about qualitative versus quantitative research, very quickly words like “ontology” and “epistemology” appear. This is the language not of scientific fact but rather of scientific belief — it is the language of religion and philosophy. If you fundamentally believe that human behavior is reducible to numerical certainties, this is your belief, not your science — and your bias toward qualitative research will be just that, a bias. Those of our clients who come, say, from the sales side of work, where careers are made and broken by the numbers, may have a hard time embracing an approach that does not quantify the thing we are examining (whether this is patient motivations, physician segmentations, or the nature of the “scientific story” we are telling about a new molecular entity). Somehow, in all of this, “quantifiable” has become synonymous with “scientific” and “unquantifiable” with “unscientific.” However, scientific discovery depends on a number of key elements, including a strict accounting for all available data, hypothesis-driven work, a willingness to fit theories to facts (and not vice versa), and transparency in reporting on experimental or observed findings, assumptions, and limitations — nowhere does it say that a science is defined by math. In fact, there are some types of data that do not lend themselves to quantification and statistical analysis, and if we want to study them, we have to handle these data using qualitative methods and techniques; included in this set of data are nearly all forms of human behavior, especially those that deal with communication, beliefs, attitudes, emotions, and the impact of these on decision-making processes. In other words, numbers and statistics tell us what is likely to happen in a given instance, with certain actors in a certain context, but they can never, ever tell us why people are going to act that way, nor can they tell us what people believe they are achieving through their behavior at that moment. The most interesting human behaviors, and the ones that are most likely to give us a window into how we might influence behavior in a positive direction, are the ones that are either contradictory to what is stated or contradictory to other forms of behavior. For example, why would someone go through the trouble of picking up a script if he never takes the medication? Why would someone who was injured and is now in a wheelchair, who states she “accepts” her chair, refuse to modify her house to accommodate her physical limitations? The answers to these questions are not simple, and they are not, at their core, quantifiable. They deal with values and beliefs, with emotions and with fundamental issues of personal identity, which are not things that can be assigned a meaningful or stable numeric value. A clear numeric answer is easier to act upon than a qualified one; and yet, if that clear answer bears little resemblance to some meaningful aspect of the question being answered, it’s not terribly useful. Adherence is a perfect example of this in our own field of interest, medicine. In order for us to understand, for example, how some patients can know that a therapy is important to their ongoing health, and yet still not take it, we have to be able to understand their deeper motivations and belief systems — not just catalogue what these patients do, but also try to understand why they do it, and determine what they believe they are accomplishing when they do it, perhaps, in this case, avoiding “unnecessary” exposure to medication. Quantitative methods of correlation help us determine the degree of impact price may have on nonadherence, but these studies fall far, far short of telling us why the bulk of patients stop taking chronic medications at three to six months from initiation. We simply cannot quantify our way into comprehension in this case — we need tools that discern patterns of belief and internal value conflicts to make sense of the behavior. The Role of Qualitative Research “You can’t fix by analysis what you bungled by design.” — Professors Light, Singer and Willett One chronic complaint about qualitative research is really a complaint about qualitative research design. We know from massive experience that asking people if they like something, or if they are likely to do something based on something else, is a very poor predictor of what they actually will like or do. Focus groups that ask marketing opinions of patients or doctors are no more useful than asking marketers about what it’s like to practice medicine. Just because we have an opinion doesn’t mean it’s a good, or educated, one. The biggest flaw in qualitative research in the medical advertising field is that it often asks questions that are beyond the expertise of the people being asked. People with diabetes can tell us what it’s like to have diabetes, not what sorts of diabetes advertising will make them act in a new way. What they can tell us is what a word or image means to them in the context of diabetes, and from there it is up to us to determine what that meaning tells us about what they are thinking, and what that implies for the work we want to do. When we ask of the world, not just what, but why, we see that numerical models fall short of explaining or illuminating why humans behave the way they do; this in turn affects our ability to create good communications materials that will change behavior in positive ways. Medicine is a numerically driven science, and qualitative researchers within that science have always had to face a “soft science bias” in presenting their results. That does not mean, however, that qualitative research should be abandoned. The debate about validity of qualitative versus quantitative research is very old. In general, it is accepted that qualitative and quantitative research are mutually reinforcing, the former helping interpret what the numbers mean, and the latter helping to validate and give perspective. At best, they feed each other: quantitative research identifies patterns that qualitative needs to explore, and qualitative research helps craft meaningful quantitative survey instruments. In other words, it is a false, and unnecessary, argument. Just because we can quantify something does not make it better at measuring the world than something we can’t quantify. How we talk and act is quantifiable, but why we do so is not. For this reason, I believe qualitative researchers should stop making apologies for the data we handle and the conclusions we draw simply because these data are not — and cannot be — quantified. Instead, we should pay close attention to the types of findings that only qualitative research can uncover, so that we can add insight into why as well as what and how often to the questions we ask. Brad Davidson, Ph.D. General Manager Ogilvy ­CommonHealth Insights & Analytics, part of Ogilvy ­CommonHealth Worldwide In Defense of ­ Qualitative Research “Research is to see what everybody else has seen, and to think what nobody else has thought.” — Albert Szent-Gyorgyi, Nobel Laureate Properly executed qualitative research is projectable; what begs the question is what, exactly, we can predict with it, a question that is rarely asked and even more rarely answered in pharmaceutical research. Ogilvy CommonHealth Worldwide, part of the Ogilvy & Mather network and a WPP company, represents the largest assembly of creative talent in the world of healthcare communications with 65 offices across 36 countries. { For more information visit

Posted in:

Post a Comment

You must be logged in to post a Comment.