To some extent fear of crime is thought to be a social problem because it is assumed to constrain social interaction. However, empirical support for the effects of fear on social behavior is mixed. This may reflect two problems. Extant research does not control for the negative effect of social behavior on fear while estimating the effect of fear on social behavior; and research assumes that the effect of fear on social behavior is the same for all people. Using the NCS dataset and simultaneous equation techniques, we address this issue. The analysis suggests that fear and constrained social behavior are part of a positive escalating loop (fear constrains social behavior which increases fear) and that the strength of the loop is contingent on age.