The value of social network analysis is critically dependent on the comprehensive and reliable identification of actors and their relationships. We compare regional knowledge networks based on different types of data sources, namely, co-patents, co-publications, and publicly subsidised collaborative Research and Development projects. Moreover, by combining these three data sources, we construct a multilayer network that provides a comprehensive picture of intraregional interactions. By comparing the networks based on the data sources, we address the problems of coverage and selection bias. We observe that using only one data source leads to a severe underestimation of regional knowledge interactions, especially those of private sector firms and independent researchers. The key role of universities that connect many regional actors is identified in all three types of data.