Reviewed by
Katarzyna Abramczuk
Institute of Sociology, University of Warsaw
When sociologists read about data-centric social sciences they instantly become suspicious. We were taught that social science is about understanding the nature of how human collectives are functioning. It usually implies a theory-based research. Sato, however, proposes that we should concentrate on data. The data-centric science method, as he puts it, starts from acquiring data in a specific field. Next, a technique is searched for an adequate interpretation of the data. Sato proposes to think about this process in terms of a PDCA (plan-do-check-adjust) cycle known also as Deming cycle.The book is organized so as to give the reader a good understanding of the whole proposed process. It consists of three clearly separated parts and a short concluding part devoted to future work. The first part is an extensive introduction into the core topics of the book. Sato starts with a 50 page long chapter where the basic concepts such as data, measurement, models, or collective behavior are explained. Then he outlines the basic framework, i.e. he lists and elaborates on all the consecutive steps required when conducting a data-centric investigation. In the second part of the book Sato moves on to introduce the most popular tools used in the data-centric research. This methodological part contains two chapters. The first one is devoted to mathematical expressions. These include statistical methods, time series analysis, network analysis, and spatial analysis. The second methodological chapter concentrates on technical issues such as a data acquisition process, database server handling, and available software. The third part of the book is about case studies where the author offers five examples: (i.) risk assessment in foreign exchange markets, (ii.) segmentation of foreign exchange markets, (iii.) analysis of hotel booking data, (iv.) international air travels, and (v.) energy consumption.
The book offers a good summary of the growing scientific field that emerged largely from an interest of disciplines such as information science or physics in topics traditionally analyzed by social sciences and as such it shares both the strengths and weaknesses of this field.
Three points are among the strengths. First, Sato has broad knowledge and experience with formal models and techniques that can be used to analyze social data. The methodological part of the book offers an introduction to the chosen methods. It is appropriate even for readers who have no or only little experience with them, because even the most basic terms are meticulously defined. If one seeks to grasp the basic idea and learn about the possible types of social data analysis, this introduction seems sufficient. However, the narrative is rather formal. Good mathematical training and further readings are necessary to fully understand the potential and properties of the described tools.
The second strength of the book results from the author’s concentration on thinking from the basis of data. It led him to describing explicitly and in detail the process of data acquisition. Once again, this can be useful for researchers who never approached a similar subject and want to learn what to expect when collecting data electronically. Concurrently we gain insight into what Sato calls a cyber-physical link between data and data-generating mechanisms.
The final merit of Sato’s work is the subject itself. It is important to realize and promote social sciences based on empirical knowledge. Well practiced it can help us to understand modern societies. We live in a data enriched world. The technological development created massive amounts of information on almost every type of human activity. In order to use it adequately we need appropriate tools. The tools, however, are not enough. It is the theory and interpretation that can lead to the creation of new services, policies and goods.
This is the moment where we get confronted with the problem of “Applied Data-Centric Social Sciences”. In chapter two Sato lists the steps required by a data-centric investigation. He starts with a problem definition and ends with interpretation and decision making. These points seem crucial for our research to step beyond a simple formal exercise. However, when reading the examples provided in the third part of the book, it is often difficult to pinpoint the purpose of the given analysis. Even more problematic is what we should learn from it. Sato’s cycle seems to lack clear indicators of successful data interpretation which leaves us with an impression of insufficiency.
This problem is a problem of the whole field of data-centric social science. When dealing with social phenomena it rarely reaches for an output of traditional social sciences and if so, it is often superficial and somewhat detached from the research problem at hand. Such an example is visible in the introduction chapter, where the author refers to Max Weber talking about different types of meaning and understanding. The concepts are described shortly but their implications for the following argument are at best vague. Generally speaking, the presentation of social problems in the book is incomplete. This in turn leads to the aforementioned missing indicators of successful data interpretation and makes it impossible to state clearly what we learned from the analysis performed.
In summary, the book presents a fairly good introduction into an important and promising research field, but it is mostly a formal and technical one. It would benefit from a more careful definition of research problems, a better description of the properties of the tools, and a more thorough effort towards understanding the data we analyze.
Return to Contents
of this issue