The objective of this book is the practical. To that end, the methods described here are predicated on a specific philosophy of science. This philosophy argues that causal relationships and social processes are a property of the social world, not a property of data we collect about the social world, and thus data can never be used to prove a causal theory. Instead, this philosophy argues that theory can be used to construct hypotheses about cause and that data about the empirical world can be used as evidence to support or fail to support these hypotheses. Based on this philosophy, we argue that new data collection techniques that maximize our ability to test causal hypotheses and provide us with new insights into social cause and consequence are of paramount importance in the social sciences.

The methods we describe are also motivated by our conviction that causal inference based on observational data will always be part of social science research. This may be especially the case for key public policy issues. We argue that because random assignment to important social conditions will continue to be a rare circumstance, we must devise other means of documenting causal relationships. From time to time, a new program or public policy may provide the opportunity to randomly assign some people and not others to a specific new condition.