An initial test of NRDF's proposed 1994 campaign has been disappointing. Should the agency test the campaign one more time or abandon it? If a test is proposed, what site sample should be taken? The case provides a meaningful managerial context in which to address questions about sample size.