Corruption as social phenomenon is studied by a variety of disciplines　― anthropology, criminology, development theory, economics, political science, psychology, sociology. Each of them has developed its own scientific traditions associated with different methods of collecting and analyzing empirical data, such as participant observation in anthropology or sociological interviews.In this paper we describe experimental methods, which in recent years have emerged as an important tool for interdisciplinary social analysis. In the first part of our work, we describe and confront experimental methods to other tools of social analysis, and characterize their relative strengths and weaknesses. In the second part, we turn to the experience of using experiments to study corruption and show their power and applicability whenever other methods are powerless. These conclusions are illustrated by the material of a unique laboratory study involving police officers of middle rank as experimental subjects, conducted by A. Belianin and L. Kosals in Moscow. We model corruption as a public good for potentially corrupt participants and measure their attitude towards this institution under different experimental conditions, especially those when its existence is pointless from individual and social rationality viewpoint. Our results speak in favour of combination of various research tools and methodologies for the study of complex social phenomena.