Abstract: With the expanding of social science experiments, relevant researches are fruitful and booming within almost every sub-fields. This paper, based on the historical review of the experiment method, explores the development context, application procedure and context, and influence to social sciences brought by such an approach. Furthermore, by comparing the application of various experiment operation techniques and quasi-experiment analysis models, we get an overall view upon the experiment method employed in social sciences at home and abroad, make a prediction of its development prospect and further lay a theoretical foundation for the further application. The authors tentatively present the differences of regression results of different causal inference methods, such as DIDD model, PSM-DID model, placebo test and synthesis control. Yet, there are probably opposite results during the data mining process when we analyze the same topic or same data under diverse quasi-experiment models. This finding reminds us that the exploration of causality is attributed not so much to the single experiment method as to validation of multiple models within quantitative research. Nevertheless, this article does not intend to provoke a “paradigm dispute” between the traditional qualitative and quantitative methods, or set up an “iterative debate” between the traditional quantitative models and the new model. We are more trying to improve the reliability and validity of our research results under the guidance of inclusive methodological thinking, so as to improve the display of social science and make better contributions to the improvement and perfection of economic and political policies.

Key words: experimental method, multi-model comparison, application scenarios, reliability, validity