Through survey design, abstract theoretical hypotheses are converted into observable variables, allowing research questions to be embedded in real-world contexts and incorporated into empirical analytical frameworks. Photo: TUCHONG

Survey research plays a foundational role in the development of economics. Its value lies not simply in acquiring data, but in using appropriate methods to address real-world problems and answer three basic questions: “What has happened?” “Why did it happen?” and “How should we respond?” Some studies reduce surveys to questionnaire distribution or data collection, treating them as a purely technical step. Others argue that, amid the rapid growth of big data and administrative records, traditional surveys have become less important. Though different on the surface, both views diminish the foundational role of survey research.

Empirical facts as engine of theoretical evolution

Viewed across the development of economics, empirical data—whether drawn from statistical systems or survey materials—has consistently played an important role in theoretical progress. The advent of representative sampling methods in the late 19th century enabled researchers to make inferences about larger populations from limited samples, providing essential methodological support for empirical research. In the 20th century, as national income accounting systems gradually took shape and household surveys became more widely used, the research trajectory of economics also began to change: analytical approaches based solely on deductive reasoning gave way to explanatory frameworks grounded in empirical facts.

John Maynard Keynes’s formulation of the consumption function, though highly abstract in theoretical form, was inspired by statistical observations of the relationship between income and consumption at the time. Using long-term data, Simon Kuznets later found that the average propensity to consume did not continuously decline as income rose, but remained relatively stable—a finding that clearly diverged from Keynesian expectations.

To reconcile this gap between theory and empirical fact, Milton Friedman and Franco Modigliani proposed the permanent income hypothesis and the life-cycle hypothesis, respectively, offering new explanations for consumer behavior. Empirical data, in other words, does more than test theory; it often catalyzes theoretical adjustment and even reconstruction.

At a more micro level, the connection between survey research and theory is even more direct. Many influential theories did not emerge from a priori assumptions, but gradually took shape through sustained observation and investigation of real-world behavior. In agricultural economics and development economics, for example, Theodore Schultz and later scholars developed the “rational peasant” paradigm through extensive empirical research on farm household production behavior, revising earlier characterizations that had treated farmers simply as irrational actors. This shift not only reshaped understandings of agricultural production, but also redirected research in areas such as technology diffusion and human capital investment.

Michael P. Todaro’s framework for analyzing rural-urban migration was built on a synthesis of the actual characteristics of labor mobility in developing countries, with “expected income differentials” becoming an important basis for explaining migration behavior. Angus Deaton’s systematic research on consumption and poverty, based on household survey data, helped steer development economics toward an analytical path grounded in micro-level behavior.

Survey design as bridge between theory and reality

Survey research also carries significant methodological importance. Its value lies not only in the acquisition of empirical materials, but in translating research questions into analyzable objects. Through survey design, abstract theoretical hypotheses are converted into observable variables, allowing research questions to be embedded in real-world contexts and incorporated into empirical analytical frameworks. In this sense, surveys are not merely subordinate procedural steps attached to the research process; they are a key link running from problem formulation to empirical validation. How a survey is designed largely defines the identification strategy and interpretive boundaries of research findings.

Despite its methodological importance, survey research is still often misunderstood in practice. Three problems are especially common. The first is treating surveys as an isolated stage, focusing only on data acquisition while neglecting their relationship to the research question and theoretical framework. In economic research, questionnaire design is itself a concrete expression of theoretical hypotheses: how variables are defined, how questions are worded, and how samples are selected all directly affect subsequent identification strategies. Without a clear sense of the research problem, even a large-scale survey may fail to produce conclusions with real explanatory force.

The second problem is excessive reliance on data and methods. As econometric tools and computational techniques have advanced, research has become increasingly sophisticated in form, but technical complexity cannot substitute for an understanding of real-world mechanisms. Once institutional contexts and behavioral logics are ignored, even the most elegant model may amount to little more than formal rigor.

The third problem is sample selection bias. Whether a sample is representative directly affects whether research findings can be generalized. Economics has developed a range of methods for dealing with selection bias, but if the survey design itself is flawed, data-quality problems can still fundamentally undermine causal inference. In this sense, a survey is not only a source of data, but part of the research design itself.

Evolving role of survey data in contemporary economics

Over the past several decades, economics has undergone a paradigm shift centered on causal identification, a development that has further highlighted the importance of survey research. The focus of the discipline has gradually progressed from correlations among variables to causal analyses of the effects of policies and institutions. In this process, both natural experiments and randomized controlled trials rely critically on high-quality microdata. The work of Abhijit Banerjee, Esther Duflo, and Michael Kremer, who received the 2019 Nobel Memorial Prize in Economic Sciences, rests on the combination of field experiments and survey data, using meticulous observation of individual behavior to identify policy effects. The rise of this research path has allowed survey research to move beyond its role as a simple means of data collection and assume a more foundational place in identification strategies.

In China, survey-data-driven causal research has likewise become increasingly prominent. In recent years, a series of micro-level databases—including the China Family Panel Studies (CFPS), the China Household Finance Survey (CHFS), and the China Health and Retirement Longitudinal Study (CHARLS)—have been continuously improved and have had a broad impact on academic research. These datasets provide a more granular observational basis for questions concerning income distribution, consumption behavior, financial participation, and related phenomena, enabling relevant theories to be tested more rigorously in the Chinese context.

Research using CHFS data, for example, has found that the impact of housing wealth on consumption operates not only via income effects, but also through asset structure, credit constraints, expectations, and other channels that shape household decision-making. This suggests that, in a wealth structure centered heavily on real estate, some assumptions in traditional life-cycle theory need to be re-examined. Similarly, patterns of intergenerational support revealed by the CHARLS indicate that families may play a more central role in allocating old-age resources than standard models assume. Chinese survey data, then, not only provides new empirical evidence, but also continues to drive adjustments in theoretical frameworks.

Value of survey research in era of big data

Amid the proliferation of big data and administrative records, the value of survey research remains undiminished. Big data offers clear advantages in capturing behavioral traces at scale and with high frequency, but it often provides limited information on critical variables such as preferences, expectations, and institutional constraints—precisely the elements that economic analysis cannot do without. By contrast, survey research can obtain such information through targeted design, giving it a distinctive role in structural analysis and causal identification. It is therefore misleading to treat big data analysis and survey research as substitutes. A more reasonable view is that they complement each other at different levels.

From a longer-term perspective, survey research underpins not only the quality of specific studies, but also the infrastructural foundations for disciplinary development. Both robust theoretical modeling and econometric analysis depend on reliable data. This means that sustained institutional support for high-quality surveys is necessary, while research training should place greater emphasis on problem-oriented thinking and survey design capabilities, so that surveys are genuinely embedded in the research process.

The advancement of economics has never been a matter of abstract models evolving in isolation. It progresses by continuously engaging with real-world problems. Only by grounding inquiry in rigorous, standardized, and sustained survey research can economics develop a Chinese school—one capable of effectively interpreting China’s experience while also proposing theoretically generalizable propositions within broader comparative perspectives. This is both a methodological requirement and an intrinsic direction for the discipline’s future development.

Yin Zhichao is a professor at and vice president of Capital University of Economics and Business.