Developing a risk assessment instrument - Lessons about validity relearned

The research reported was undertaken to develop a risk assessment instrument (RAI) that would inform case decisions about community corrections placements in Pennsylvania. Against the backdrop of this applied research project, the article examines issues related to external validity that often are neglected or glossed over in social science research. As part of this examination, the basic results of the research study are reported and specific issues related to sampling bias are addressed. Data are presented that reflect real impact of such problems. In the concluding discussion it is argued that the experiences in this project reinforce the need to be rigorous about sampling, and also the need to recognize and value different ways of knowing. (C) 2001 Elsevier Science Ltd. All rights reserved.