Beyond the Federal Drug War: A Panel Study of State-Level Powder and Crack Cocaine Laws, 1977–2010

ABSTRACT

Scholars claim that more punitive drug laws are one of the main reasons for the rapid growth of U.S. incarceration rates and related collateral consequences over recent decades. Yet despite the widespread impact of these laws, researchers continue to ignore a seemingly fundamental research question: Why do some states, but not others, enact harsher drug laws? To partly address this gap in the literature I locate the social, political, and economic factors responsible for changes in the severity of U.S. state-level powder and crack cocaine laws from 1977 to 2010. Two-way fixed-effects panel estimates reveal that theoretical accounts based on racial threat, partisan politics, and gendered politics largely explain variation in cocaine law strength within states over this period.

Acknowledgment

I thank Ryan D. King, Paul Bellair, Andrew Martin, Tamara Mix, and the anonymous referees at Sociological Spectrum for their helpful comments.

Notes

¹Several media sources failed to verify that Bias died from a crack cocaine overdose. Bias actually died from a powder cocaine overdose (Stone 2008; U.S. Sentencing Commission 1995).

²The threshold weight that triggered a 5-year mandatory minimum sentence was 500 grams for powder cocaine and 5 grams for crack cocaine. The threshold weight that triggered a 10-year mandatory minimum sentence was 5,000 grams and 50 grams for powder and crack cocaine, respectively (Stone 2008).

³Of the 1,250 state-years between 1986 and 2010, 77% have identical penalties (i.e., 1:1 ratio) for powder and crack cocaine offenses (based on author’s coding of statutes).

⁴Prominent reports include Chriqui et al. (2002), Holden et al. (1988), and Porter and Wright (2011).

⁵Throughout this study, any reference to “cocaine laws” and “cocaine law strength” that does not indicate a specific drug form (i.e., crack or powder) should be understood as a reference to comprehensive cocaine laws and comprehensive cocaine law strength (i.e., crack and powder forms together), respectively. See the Measurement section for details on how I construct the comprehensive cocaine law strength, powder cocaine law strength, and crack cocaine law strength indicators.

⁶The Republican party also is known as the Grand Old Party (GOP).

⁷Alexander (2012) correctly pointed out that calls for “law and order” emerged at the state level in the mid-1950s before gaining national attention during the 1964 presidential bid of Barry Goldwater.

⁸Other examples of possession offenses include weapons and stolen property.

⁹The use of smoking opium by Chinese immigrants also is linked to the enactment of the federal-level Harrison Narcotic Act of 1914 (Musto 1999).

¹⁰Marijuana use by Mexican immigrants also is linked to the enactment of the federal-level Marihuana Tax Act in 1937 (Bonnie and Whitebread 1970; Musto 1999; Peterson 1985).

¹¹Cocaine use among southern blacks also is linked to the enactment of the federal-level Harrison Narcotic Act of 1914 (Musto 1999; Peterson 1985).

¹²The adoption of new cocaine laws, especially in the South, is not particularly surprising given the frequent, widespread publication of sensationalized and stereotypical reports about southern black cocaine users during this era. For example, in describing separate incidents involving black men “who had become crazed by cocaine,” Edward Williams (1914)—a doctor considered an expert in the field of drugs at the time—wrote that black cocaine “fiends” are largely resistant to pain, nearly unstoppable by police, and more accurate and deadly shooters (pp. 247–248). Williams (1914) additionally asserts that a “large portion of the [then] recent wholesale killings in the South … have been the direct result of cocaine and frequently the perpetrators of these crimes have been hitherto inoffensive, law-abiding negroes” (p. 247).

¹³For all practical purposes, crack is the less expensive, more adulterated version of freebase (see Inciardi 1987 for a detailed discussion of these drugs).

¹⁴Increasing the severity of drug laws also is a politically safe and salient issue. People who sell or use drugs in the United States often are politically powerless, unorganized, or few in number (Epstein 1990).

¹⁵President Reagan launched his drug war in 1982, dramatically increased federal funding for law enforcement and drug interdiction, and clearly signaled a shift away from President Richard Nixon’s early 1970’s focus on heroin and drug treatment (Alexander 2012; Musto 1999; Provine 2011).

¹⁶The substantial time and financial costs associated with extending the dependent variables for every state by one year also prohibit an earlier start date. A minimum of 497,000 pieces of information would need to be located, coded, and manually entered into statistical software.

¹⁷The definitions of simple possession without intent, possession with intent, manufacturing, delivery, sale, and trafficking used in this study are provided in Appendix A.

¹⁸Penalties with an effective date on or after July 1 are coded as occurring in next calendar year.

¹⁹I use the alpha command in Stata/SE Version 14.2.

²⁰Ballot initiatives in Arizona (Proposition 200/Drug Medicalization, Prevention, and Control Act) and California (Proposition 36/Substance Abuse and Crime Prevention Act) mandate that most nonviolent, drug (including powder and crack cocaine) possession offenders be given the option of drug treatment in lieu of imprisonment (Drug Policy Alliance 2006; Longshore et al. 2003). Hence, penalties for SPWI of powder and crack cocaine are coded as 0 for the state-years these initiatives are in effect.

²¹The means of all year dummies are identical and would automatically be excluded from the model due to collinearity.

²²Because percentage urban and its square are used to capture urbanization, joint significance tests that take both variables into account must be (and are) referenced throughout the rest of this study.

²³Since the main effect for the South does not change over time, the fixed-effects estimator automatically (and correctly) omits this variable from the model. However, the term resulting from the interaction of South and Republican governor, as well as the main effect for Republican governor, are retained by the estimator because both predictors vary over time.

²⁴As previously mentioned, the association between Republican strength and cocaine law strength outside the South is not significantly different from zero. For informational purposes, however, I plot (but do not further interpret) the predicted values of cocaine law strength for regions outside the South in Figure 3.

²⁵The effects of state spending on social welfare are not significant in models that substitute two-year lags. All other variables, however, remain significant when this longer lag is used.

²⁶Since the total number of predictors in a model should not exceed the number of clusters (k-1) used to calculate the cluster-robust standard errors, many of the alternative models substitute the total crime rate for property and violent crime rates to reclaim a degree of freedom. When possible, alternative explanatory factors are added one at a time and subsequently removed because the simultaneous inclusion of two additional variables violates the aforementioned rule and expectedly results in a missing F-test statistic for the model. The F-test statistic is needed to assess whether the coefficients for the regressors in the model are jointly and significantly different from zero.

²⁷Drug arrest data for several states (especially for Florida, Illinois, Kansas, Montana, South Carolina, and Wisconsin) and years are incomplete or missing, are consequently interpolated or excluded from this analysis, and result in a smaller sample size (N = 1,635). Disaggregated arrest data for opium and cocaine sale and manufacturing are not available. Data for poverty rates also are limited to years after 1976 and result in a smaller sample size (N = 1,617).

²⁸The 11 former Confederate states are Alabama, Arkansas, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee, Texas, and Virginia.

²⁹The 12 Sunbelt states are Alabama, Arizona, California, Florida, Georgia, Louisiana, Mississippi, Nevada, New Mexico, North Carolina, South Carolina, and Texas.

³⁰For instance, I am unable to simultaneously disaggregate the female and Republican legislative strength variables (i.e., % male Republicans, % female Republicans, % male Democrats, and % female Democrats) and still test for previously shown interactions with the U.S. South. The number of variables in the model would exceed the number of clusters (k-1) used to calculate the standard errors.

³¹One-way fixed-effects estimates that separately assess the one-, two-, three-, and four-year lagged effects of a Republican president on comprehensive cocaine laws (models not shown) always yield nonsignificant coefficients for the Republican president variable and significant coefficients for violent crime rates and the tax base. Model AIC statistics for the previous models also indicate poorer model fits than Model 7, this study’s best two-way fixed effects model. A two-way fixed-effects estimate of state cocaine laws (not shown) that includes the presence of a Republican president variable results (as predicted) in a random-year dummy being omitted due to collinearity.

³²In many state-years, the penalties for possession with intent, sale, delivery, and manufacturing for weights greater than or equal to 28 grams are identical.

Additional information

Funding

This research was supported, in part, by a grant from the Criminal Justice Research Center at The Ohio State University.

Notes on contributors

Chad A. Malone

Chad A. Malone is an Assistant Professor in the Department of Sociology at Oklahoma State University. His main line of research addresses the ways in which racial/ethnic, political, and economic factors affect macrolevel social control outcomes over time.