They found that rates of victimization, as measured by NCS data, are systematically related to residual gain scores in UCR data on crimes known to the police and that rates of crimes known to the police, as measured by UCR data, are systematically related to residual gain scores in NCS victimization rates. Menard has criticized these findings, arguing that the degree to which the two data series converge has been exaggerated and that the analysis has limited practical application. Nonetheless, the Blumstein research hypothesizes that both crime measures reflect an objective reality of criminal events filtered through two different measurement processes. The usefulness of either UCR or NCS crime measures as indicators of underlying crime levels increases when the offending component dominates variation in the measure and when the measurement component is essentially random or systematic in ways that can be modeled explicitly. The Blumstein research focuses on partitioning total variations in either measure between time trends and yearly fluctuations and on the influence of UCR crime rates based on crime reporting to the police. UCR crime rates are estimated from models consisting of separate measures of trend and deviation components of NCS data. Blumstein, Cohen, and Rosenfeld conclude that the results are similar when NCS crime rates are regressed on trend and deviation components of UCR crime rates. They believe that Menard fails to recognize their emphasis on isolating and explicitly analyzing both linear time trends and fluctuations around those trends. 4 references and 1 figure