What is the Philosophy of Research? ―Rethinking the Present of Research from Its Origins―③

Continuing from last time, we asked Mr. Akira Kajiyama, who has long been involved in research across academia and industry, and Mr. Hideki Katahira, a leader in marketing spanning data analysis to branding, to identify major trends in the research world and the issues and challenges emerging within it. They also explored the very essence of what research is, examining its origins, and discussed the philosophy and positioning of research within future marketing activities.

---

Increased Freedom in Research

Katahira: This is starting to sound like old folks reminiscing about "the good old days" (laughs), so let's shift gears a bit. What do you think is needed to address this situation? Please offer advice on that front, or share any areas where things have improved.

Kajiya: It would be regrettable if professional traditions were fading in the field of research. However, since the problems are clear, I believe they can be solved if those involved in research share a sense of crisis and provide thorough training to newcomers. In fact, I hope we'll see more research techniques emerging that offer a distinct flavor from the paradigm of inferential statistics.

For instance, while interpretive approaches may have limitations in reproducibility and repeatability, they can vividly capture the individuality of people and groups more effectively than mass surveys. I hope this level of approach is further elevated, and I'd also like suggestions on how to apply it in the workplace.

Regarding Bayesian statistics mentioned earlier, Bayesian filters for removing spam emails gained attention in the online world. However, there are various other models like Bayesian networks and hierarchical Bayesian models using MCMC. While the algorithms and computational processes are complex and beyond my expertise, members of the Marketing Science Society are analyzing POS data and log analysis data, reporting their findings one after another. I believe this will become more widespread.

Several years ago, a Dentsu Inc. researcher analyzed advertising effectiveness using state space models. While complex, the essence of this model is continuously tracking the state of something in motion using new observations. What surprised me was the sharpness of its predictions. When I worked at a newspaper company, my job involved analyzing and reporting monthly data on circulation figures and advertising column inches. The biggest challenge was that while I could decompose the time-series data into individual fluctuations, I couldn't combine them to make predictions. I often wished this model had existed back then. State space models seem to be primarily used and successful in the natural sciences, and I hope they will also prove useful for solving problems in marketing and advertising.

Previously, grasping mass trends was sufficient, but recently, needs are increasingly localized. Weather forecasts are a prime example; people want predictions for their specific town, not just their region. Interestingly, online forecasts leverage site visitors as contributors, enhancing accuracy by combining institutional weather probabilities with subjective user insights. This emphasis on the "individual" extends to academia: while traditional economics focused on aggregate market movements, behavioral economics analyzes individual actions, and neuroeconomics measures localized blood flow and electromagnetic activity in the brain. In marketing, attention is shifting from consumer commonalities to their heterogeneity. Analysis is becoming individualized and localized, yet a broader perspective is simultaneously required. Surveys must evolve to align with this trend. From a survey philosophy perspective, the purpose of research seems to be moving from approaching some distant "truth" toward solving individual problems, and the social utility of surveys is increasingly being questioned.

Katahira: When I think of surveys, random sampling immediately comes to mind. While it remains one of the main businesses for survey companies, its effectiveness is declining. At the same time, we're entering an era where data on various human behaviors is almost freely available. I suspect this is where marketing science will truly come into its own. Moreover, as the focus shifts to humans, if the purpose of the survey – what we aim to learn – is clear, the freedom to combine different methods and determine the cost required to gain knowledge from the information is incomparable to just ten years ago.

Kajiya: Recently, systems using wearable devices attached to people seem to be developed to measure the movements of customers and staff in stores. They combine devices that measure physiological responses with location beacons to track people's movements and work content. I think the development of ICT has truly ushered in an era where all kinds of research is possible. Of course, there are personal information issues, so obtaining consent from the people who become samples is necessary.

Katahira: Actually, the people putting this into practical use right now aren't from market research or marketing science backgrounds. They're applied engineers. They're launching small ventures and doing astonishing things.

Kajiya: Is this becoming a business?

Katahira: I recently heard about someone who originally worked in nuclear engineering. They're advancing research showing that combining about three types of data reveals individual human behavior.

Kajiyama: I'd be curious about the accuracy, but it sounds like a massive advancement of old mesh data.

Katahira: Exactly. For example, understanding how drugs work—by treating cells as individual units and applying certain rules, you can reproduce outcomes. Fundamentally, I believe anything is possible if done properly. But here too, the crucial question remains: what do we want to know?

Rethinking the fundamentals of research

Kajiya: So hypothesis-generating power is becoming increasingly essential.

Katahira: Exactly. I strongly feel that aspects like hypothesis-building ability, principles, and philosophy are becoming increasingly prominent.

Kajiya: In these fields of empirical analysis, engineering sensibilities are also required. Future researchers must keep pace with new knowledge and technological trends. Your book on Zeami emphasizes "the beginner's mind in old age" – we shouldn't just pin our hopes on the young (laughs).

Also, regarding the recently discussed big data, there are various challenges from a research perspective, including defining what constitutes "big." The research methods and analytical approaches should differ between structured data like online shopping or banking data and unstructured data like SNS chats or GPS. We need to clearly define the purpose of using the data and extract only the truly useful information from vast amounts in real time. In that sense, the power of science seems increasingly vital.

Katahira: Speaking of marketing science, it's incredibly deep. If the implications don't resonate with practitioners, it's meaningless.

Kajiya: That's certainly true.

Katahira: Getting published in top-tier American journals is great for your career, but when you ask how much happier Japanese consumers will be as a result, you might as well not bother (laughs).

Professionalism is advancing in the US, but it's increasingly detached from the actual field. Japan is still better in that regard. In Japan, the proportion of practitioners in the Marketing Science Society has increased significantly, while in the US, it's steadily decreasing. People are increasingly migrating to fields like quantitative biology, quantitative psychology, or mathematics because they're easier to publish papers in. Consequently, it becomes irrelevant where or how such research is applied, further distancing it from the reality of human behavior. With greater freedom in research environments, I hope we see the emergence of skilled craftsmen or heroes who can effectively harness big data.

Kajiya: Research methods include empirical approaches and interpretive approaches, but I haven't seen simulations, commonly used in engineering, applied much. New research methods will be needed going forward, and alongside interest in human behavior, a spirit of challenge toward new fields will become crucial.

Katahira: In marketing, simulations using multi-agent systems are emerging to analyze shopping behavior, but practical implementation is still ahead. I believe it's also vital to return to basics and reexamine questionnaire surveys. Actually, about a year ago, we conducted our own survey. It focused on word-of-mouth when searching for restaurants for dining out. The findings revealed that, generally, others' evaluations don't work; what does work is factual information.

Essentially, factual information like the menu, appearance, opening hours, and location. We learned this through actual practice, and since it was validated as an empirical hypothesis, it reinforced that traditional questionnaire surveys still have value. This could potentially offer a solution for research firms grappling with the challenge of ensuring respondent quality.

Moreover, with just one survey form, you must ask precisely what you want to know in a simple and accurate manner, within budget and deadlines. This really hones the skills of the researcher. After all, just a slight difference in auxiliary verbs can completely change the nuance. That's why survey professionals must be craftsmen and masters of their craft.

Kaji: Japanese business places immense emphasis on the quality of manufacturing and services, and that principle certainly applies to the world of research too.

Katahira: I feel that applies to everything. People chase after specialized studies and skip the absolute fundamentals. That's why I want to start a seminar for working adults to read the classics. Perhaps the world of research, too, has reached a point where it needs to revisit its origins. Thank you very much for today.

[End]

---

*The full text is available on the Yoshida Hideo Memorial Foundation website.