Since the 1970s, the United States has experienced rapid fluctuations in homicide and crime rates, with several notable “booms” and “busts.” During this same time frame, the social, economic, and political climate of the U.S. has shifted as well, with massive increases in incarceration, deindustrialization and the resulting concentration of poverty and joblessness, changes in policing, including community policing and the War on Drugs, changes in family structure, and increases in immigration. These changes, particularly the increase in homicide and violence in American cities in the 1980s and the subsequent 1990s crime decline, resulted in increased empirical attention and scholarly debate about homicide and violent crime trends, as well as what contributes to the shifts in recent trends.

​

Despite a significant body of research in this area, particularly in the past 15 to 20 years, inconsistencies abound and scholars vehemently disagree about the role of factors such as firearms and policing. Complicating our understanding and impeding our ability to make sense of this “crime trends puzzle” is the fact that there has been substantial variability in how these explanations have been tested, particularly with regard to research design and model specifications, making it difficult to draw any firm conclusions.

​

This research systematically reviews the homicide and violent crime trends literature. In order to synthesize findings from this vast body of research, I use meta-analytic techniques and data from studies conducted at the city-level or higher that were published between 1980 and 2016 in order to identify the relative importance of each of the prominent explanations put forth to explain recent crime trends. Explanations assessed include economic conditions, family structure, age structure, immigration, policing, incarceration, drug markets, and firearm prevalence and legislation. Specifically, in this work I estimate an overall average effect size for a number of common predictors capturing each of the key explanations (e.g., absolute deprivation, relative deprivation, and wages are all estimated separately as indicators of economic conditions), and then they are rank-ordered to establish their relative importance. Empirical tests of even the most common explanations exhibit substantial variability in research design and model specification, including differences in operationalization of key constructs, covariates included, unit of analysis, time period covered, and statistical methods used. Therefore, this research also assesses the impact of methodological variation on the obtained results, and considers whether inconsistent results and contradictory conclusions in the extant literature are attributable to variation in the methods employed. 

​

Implications for theory, including the scope conditions under which certain explanations hold, research, including guidelines for future empirical research on crime trends and the key covariates to include, and policy, including recommendations on where to direct resources to prevent and control crime, are discussed.