In the tide of intelligent transformation, social experiments have become a vital channel for advanced exploration of social governance programs. Grounded in the universal significance of social experiments as a common methodology, this book reviews the origins, development processes, and classic cases of social experiments. On this basis, it systematically introduces four types of classical methods, which are natural experiments, field experiments, surveys, and computational experiments. It explores the combined paths of social experiments with methods such as formal modeling, qualitative investigation, and real-world research, analyzing potential factors that may influence the effectiveness of social experiments from three levels of macro experimental design, meso process management, and micro psychological mechanisms. In addition, it briefly introduces commonly used mathematical and statistical analysis techniques, such as difference-in-differences, regression discontinuity, matching, synthetic control, and structural equation model, and provides examples. This book can serve as a methodological textbook for disciplines related to social sciences, a reference for conducting AI social experiments in national intelligent social governance experimental bases, and is also suitable for general readers.