APPLYING SOCIAL SCIENCE DATA TOOLS TO CRIMINAL JUSTICE POLITICS, ADMINISTRATION, AND RESEARCH: THE CASE OF ALABAMA Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include proprietary or classified information. ________________________________ John Bennet Wright Certificate of Approval: _________________________ ________________________ James H. Gundlach Charles E. Faupel Profesor Profesor Sociology Sociology _________________________ ________________________ Greg S. Weaver Stephen L. McFarland Associate Professor Dean Sociology Graduate School APPLYING SOCIAL SCIENCE DATA TOOLS TO CRIMINAL JUSTICE POLITICS, ADMINISTRATION, AND RESEARCH: THE CASE OF ALABAMA John Bennet Wright A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Arts Auburn, Alabama December 16, 2005 iii APPLYING SOCIAL SCIENCE DATA TOOLS TO CRIMINAL JUSTICE POLITICS, ADMINISTRATION, AND RESEARCH: THE CASE OF ALABAMA John Bennet Wright Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ________________ Signature of Author ________________ Date of Graduation iv THESIS ABSTRACT APPLYING SOCIAL SCIENCE DATA TOOLS TO CRIMINAL JUSTICE POLITICS, ADMINISTRATION, AND RESEARCH: THE CASE OF ALABAMA John Bennet Wright Master of Arts, December 16, 2005 (B.S., Auburn University, 1998) 78 Typed Pages Directed by James H. Gundlach Information from a variety of criminal justice and demographic sources can be merged to create a data file useful for a variety of audiences. Politicians, administrators, and academics would be beneficiaries of analyses performed as the result of completing a source of data outlined in this work. Current problems with administrative files are detailed to give perspective on the scope of work that can currently be performed and what is possible if inherent deficiencies in the files were modified. The determination of crimes as violent or not for administrative reasons is detailed to show discrepancies in the two most widely used definitions of violent crime in Alabama. Recently passed child sex offender legislation is analyzed to provide a backdrop for some of the drawbacks in the current system of housing pertinent data on v offenders and victims in Alabama. A multivariate analysis reveals some of the impacts of specific variables on recidivism of inmates in the Alabama Department of Corrections. Offenders in the custody of the Alabama Department of Corrections as of October 31, 2004 were used for analyzing recidivism in the study. The population at this time was 26,179 inmates. vi American Sociological Review JMP 4.0 SPSS 12.0 BBEDIT EXCEL 2000 WORD 2000 vii TABLE OF CONTENTS List of Tables???????????????????????????..?. Chapter 1: Introduction??????????????????????.???. Chapter 2: Background???????????????????.??????. Measuring Violence for Administrative Purposes?????????.??? Politics of Sexual Offenders????????????????????... Social Science Research Example One: Unemployment and Recidivism??? Social Science Research Example Two: Education and Crime??????... Social Science Research Example Three: Marital Status and Crime??.??. Chapter 3: Analyses????????????????????????.?? Two measures of Violent Offenders: Distortions of Reality?????.??.. Politics of Sexual Offenders: Fox Folly Example???????????... Social Science Research Example One: Unemployment and Recidivism.??. Social Science Research Examples Two: Education and Crime???? ?? Social Science Research Example Three: Marital Status and Crime????.. Whole Model on Recidivism????????????????..???. Chapter 4: Conclusion??????????????????????....??. References?????????????????????????????? Appendix A?????????????????????????????? viii 1 4 4 7 8 19 24 26 26 28 35 36 41 45 50 55 62 viii 6 6 12 18 19 24 25 36 38 39 40 43 44 45 46 47 47 48 LIST OF TABLES Table 1: Frequency Distribution of violent crimes by ASC definition?????...?... Table 2: Frequency Distribution of violent crimes by DOC definition???????.. Table 3: Distribution of Unemployment Rates in Alabama Counties in 2004????. Table 4: Frequency Distribution of AIS Suffixes of DOC Inmates????????.. Table 5: Frequency Distribution of Dummy Variable for Recidivism???????. Table 6: Frequency Distribution of Educational Attainment ??????????.. Table 7: Frequency Distribution of Marital Status of Inmate Population?????? Table 8: Relationship of Unemployment on Recidivism???????????..... Table 9: Impact of Education on Commissions of Violent Crime????????? Table 10: Impact of Education on Recidivism????????????????. Table 11: Means for Unemployment Rate by Education Level?????????.. Table 12: Impact of Marital Status on Commission of Violent Offenses????...?. Table 13: Impact of Marital Status on Recidivism?????????????...? Table 14: Logistic Regression Whole Model Significance and Strength?????.? Table 15: Impact of Race on Whole Model for Recidivism?????????...?. Table 16: Impact of Committing Violent Crime on Whole Model for Recidivism ...?. Table 17: Impact of Unemployment Rates on Whole Model for Recidivism????.. Table 18: Impact of Education on Whole Model for Recidivism???????...?. 1 CHAPTER 1 INTRODUCTION This thesis will document the construction of and show the use of a multi-purpose data set on the total incarcerated population of Alabama as of October 31, 2004. The study draws on several official sources including court, corrections, and probation and parole records. The data set will also be structured in a manner to allow inmate data to be merged with other data sets including contextual social variables such as employment and crime rates. In this thesis, I will also illustrate how the data set can be used for criminal justice system administration, providing information for political decision makers, and social research on crime. The goal of this project will be to overcome several problems identified in the current system of using criminal justice system data to meet the needs of criminal justice system administration, political leaders, and scholars. There are several variables that are routinely collected and recorded for populations of inmates. The prevalent shortcoming of the array of information is the manner in which it is coded, stored, and housed. Most collections of inmate data generally involve the collection of basic demographic and offense variables. The recording of other data fields varies widely by the interested parties or agencies. Data for inmates cover a wide spectrum of issues but is generally limited to specific tables and sources. This thesis demonstrates how data from a variety of sources can be combined to 2 provide more useful combinations of variables to be analyzed for different purposes. This thesis also illustrates how such data are useful for administrative, political, and social research purposes. Compiling such a complex data set presents several problems. One is the dispersion of variables among many different administrative units. Multiple data sets, within the same agency, sometimes contain the same variables and other times they do not. Within the data sets, the same variable also may be coded differently. For example, the Alabama Department of Corrections maintains multiple data sets compiled for different purposes with only unique identifiers used in common. Some of these data sets contain information on demographic variables, others on classification issues, still others on sentencing information, and yet others contain data on transfers and releases. The third problem is the data are maintained in several incompatible computer systems with different means of restricting access. The fourth problem is that some important variables are not computerized at all and can only be measured by a physical examination of paper records. A fifth problem is that administrative units are protective of and unwilling to share much of the data they have produced. In general these problems can be traced to a history of compiling data needed for specific reporting and administrative goals without regard to future data needs. Overcoming these problems can lead to the production of a data set that combines variables from several sources on the total population of incarcerated offenders. Such a data set should include variables allowing inmate data to be merged with other social and economic data as well as national criminal justice data programs. 3 A data set that combines the features of an entire population of incarcerated offenders along with featured social variables creates a rich source of data for investigators. Such data could be used to more efficiently produce reports for administrative and political decision making purposes as well as address more complex and richer social research issues. 4 CHAPTER 2 BACKGROUND In addition to constructing the data set described in the introduction, this study will illustrate how these data can be used for administrative, political decision making, and social research purposes. In this section, I will describe the background for three illustrations to follow. Measuring Violence for Administrative Purposes First, for the administrative example, an examination of the classification of crimes as violent and non-violent will be conducted. The problematic nature of how crimes are classified as violent or non-violent is illustrated by using two different sets of criteria. The two different sets of criteria come from the Alabama Department of Corrections and the Alabama Sentencing Commission. The complete list of criteria for both definitions of violent offenses is listed in Appendix A. There are two primary differences between the sets of criteria that result in more crimes being classified as violent under the Alabama Sentencing Commission?s list of criteria. First, the Alabama Sentencing Commission classifies more types of crime as violent than does the Alabama Department of Corrections. For example, the Alabama 5 Sentencing Commission classifies the crime of Sexual Torture, a Class A felony, as a violent crime, whereas the Department of Corrections does not. Essentially, the Alabama Sentencing Commission classifies any crime as violent if it entails any of the criteria for violent crime listed in Appendix A7. Second, the Alabama Sentencing Commission classifies any attempt, conspiracy, or solicitation of any of the crimes as a violent crime. The Alabama Department of Corrections does not. The definition of violent offenses under the Alabama Sentencing Commission criteria contains more crimes as violent due in large part to discussions with political decision makers who suggested these crimes be classified as violent. District attorneys and victims advocates were also instrumental in classifying more offenses as violent in the Sentencing Commission?s definition. The rationale for deeming more offenses as violent under the Commission?s definition was if harm could be done to society, as well as or in lieu of the immediate victim, then the offense should be categorized as violent. The ASC_violent variable is a dummy variable that stipulates if the controlling offense for the current incarceration episode is considered to be violent by the Alabama Sentencing Commission. The complete listing of offenses that are considered to be violent by the Alabama Sentencing Commission can be found in Appendix A7. The Alabama Sentencing Commission?s definition of violent crime is used rather than the Alabama Department of Corrections? definition because the Sentencing Commission uses a more updated and more encompassing definition of violent crime. The Alabama Department of Corrections list of crimes considered to be violent comes from actual admissions rather than all possible offenses. The offense of Sexual Torture is not classified as a violent offense by the Department of Corrections because this admission 6 was the first time an offender was admitted to the Department for this particular offense. Rather than looking at all possible offenses to determine what offenses are considered violent, the Department of Corrections only looks at offenses that offenders are admitted for when determining what crimes will be included on their list of violent offenses. The Alabama Department of Corrections complete listing of violent crimes can also be found in Appendix A8. The following distributions show the numbers of crimes considered to be violent using both definitions. The Alabama Sentencing Commission definition of violent crime indicates about 62 percent of inmates are currently imprisoned for violent offenses where the Department of Corrections definition shows about 54 percent of inmates currently have violent offenses. Table 1 shows the distribution of crimes considered to be violent by the Alabama Sentencing Commission definition and Table 2 shows the distribution of crimes considered to violent by the Alabama Department of Corrections definition. Table 1: Frequency Distribution of violent crimes by ASC definition ASC definition of violent crime crime frequency percentage not violent 9877 37.82 violent 16240 62.18 TOTAL 26117 Table 2: Frequency Distribution of violent crimes by DOC definition DOC definition of violent crime crime frequency percentage not violent 11954 45.77 violent 14163 54.23 TOTAL 26117 7 Politics of Sexual Offenders Next I will lay the groundwork for illustrating how the data can be used for political decision-making. The O?Reilly Factor show on Fox News devoted an episode to child sex offender laws. Bill O?Reilly, the host of the show, asserted that Alabama, among other states, did not care about child sex offender laws and said punishments in Alabama were low compared to other states. While the source of the data was not revealed by the show or the host, the airing of the show precipitated a movement by the Attorney General?s office in Alabama to strengthen child sex offender laws. It was never demonstrated that the laws or punishments in Alabama were weaker or less restrictive than similar measures in other states, but nonetheless the movement for harsher laws and penalties was put in motion. An analysis of the process and results of proposed legislation stemming from this single episode of the O?Reilly show is contained in the following chapter with a detailed description. This example also illustrates a shortcoming in the lack of data fields recorded in agencies? data and how it is coded and recorded. The age of sex offenders? victims is not a detailed record in the Department of Corrections data sets. The specific age of the sex offense victims is not identified in the data sets, rather only if the victim is 17 years of age or under. The omission of victim age produced a situation that necessitated the use of projections based partly on data from outside state agencies. Instead of relying on actual data showing the number of admissions to prison for sex offenders with a victim under a specific age, projections were required to approximate the number of admissions to prison for these offenses. 8 The amount of time that offenders in the Department of Corrections serve for offenses is also not readily available or easy to calculate across different classes of felonies and different offenses. The projections, mentioned above, were then used to forecast the discrepancy in time served currently and under the proposed legislation. In order to provide a required fiscal assessment of the legislation, complex analyses had to be performed to accommodate for the lacking data. Social Science Research Example One: Unemployment and Recidivism Some researchers and agencies have used various measures to predict and explain incarceration patterns from a broad, national level but few have attempted to use county level aggregate data to analyze a specified inmate population. One of the emerging research goals regarding crime and punishment in the United States is the study of recidivism among criminals/offenders. Many researchers focus not just on the reasons individuals commit particular crimes, but the correlates, and patterns of offenders who repeatedly commit criminal acts. Recidivism is defined and measured in a variety of fashions in criminology and criminal justice research often making cross comparisons of studies extremely difficult. Recidivism is generally measured in three different ways: whether someone is arrested for a subsequent crime, whether someone is convicted of a subsequent crime, or whether someone is incarcerated for a subsequent crime. Studies analyzing crime and recidivism often overlap in the utilization of common variables. Unemployment figures, racial/ethnic composition of areas, educational attainment, and prevalence of certain types of crimes are variables that are 9 often used when researching recidivism and patterns of criminality. These variables are often conceptualized and/or measured differently across studies influencing the interpretations of results from bodies of research. Combining these variables together for a contextual analysis of a statewide incarcerated population at the county level would constitute a new examination of the data. Unemployment has often been associated with criminal activity as well as other social problems. Some researchers have utilized local employment figures in conjunction with other variables to analyze crime figures of a specified geographic area (Schmitt 1956). Schmitt noted that unemployment was not only linked to criminal activity in his research, but unemployment was also highly correlated with the other social problems detailed in his study. The vast majority of literature shows that unemployment is directly linked to increased levels of criminal activity with one notable exception. Galster and Scaturo (1985) found that higher unemployment rates actually resulted in lowered court commitments to prison when looking at the 50 United States as a whole unit. The one caveat to this study however, was that when controlling for geographical region, the relationship between unemployment and incarceration did not hold for the southern states. For the southern states, higher unemployment rates were indicators of increased prison admissions and the strength of this relationship was stronger than the opposite relationship found for the entire country treated as a whole. Other research in the field incorporates unemployment and other variables to examine inmate population growth and imprisonment rates (Grimes and Rogers 1999; Quimet and Tremblay 1996). The Grimes and Rogers study ?presents the empirical estimation of an econometric model of the flow of prisoners into a state correctional 10 system.? This model controlled for unemployment and poverty rates, probability of admission into the correctional system, presence of law enforcement and a sentencing structure for the study area. This study indicated that out of other factors, unemployment was a contributing factor to the growing inmate problem. Quimet and Tremblay (1996) noted that many research projects did not address the punitive level of the state correctional system and wanted to know if this distinction would have an impact on the imprisonment rate. It was found that although unemployment was not the strongest predictor of imprisonment, it did have an impact on the overall imprisonment rate. Some researchers have used unemployment data as their main independent variable when researching crime (Cappell and Sykes 1991; Ekland-Olson, Kelly and Eisenberg 1992). Cappell and Sykes used unemployment figures in examining changes in commitment rates to United States prisons between 1933 and 1985 and found that unemployment did significantly impact both crime and prison commitments. Ekland- Olson, Kelly, and Eisenberg also used unemployment figures to compare incarceration and crime in Texas and California in the 1980s. The authors note that although the incarceration policies of these two states differ in the time period, violent crime rate trends remained constant. The authors assert that unemployment rates could be explanatory factors when describing the relationship. Carlson and Michalowski (1999) also analyzed the relationship between unemployment and imprisonment. Their analysis covers the time period from 1933 until 1992 and finds that the relationship between imprisonment and unemployment is historically contingent in the United States. The authors argue that ?qualitative relations among capital, labor, and the state are reflected in quantitative changes in the relationship 11 between rates of unemployment and imprisonment.? (pg. 217) The findings from this study revealed that the dynamics between the unemployment rate and imprisonment is largely dependent on which time period within the study window was being studied. Some time periods showed the relationship between unemployment and imprisonment was significant while other time periods revealed relationships of much less strength. Unemployment has also been used as part of a ?social stress model? in the study of imprisonment rates (Selke 2003). This social stress model incorporated unemployment and other factors commonly linked to increased rates of imprisonment. Once again, along with other factors, unemployment figures were found to be an important component in explaining crime and imprisonment rates. The unemployment variable has also been included in views combining sociology, criminology, and economics to more clearly see the link between poverty and crime (Berk, Lenihan and Rossi 1980). Unemployment benefits extended to those prisoners just released from prison were found to be a significant factor in reducing crime after release. The unemployment variable, whether viewed singularly or in conjunction with other variables, has been used in a great deal of social research regarding crime and incarceration. The constant theme in the literature appears to be that unemployment plays an important role in crime and incarceration rates. Unemployment has been used and measured differently throughout the studies and consistently yields important findings in the field with the exception of one study (Chiricos 1987). Chiricos did not find the correlation between unemployment and crime rates to be as strong or significant as the other analyses. Galster and Scaturo (1985) found the results to be significant, however the nature of the relationship was opposite that of the other studies. 12 Unemployment information has widely been used in social research analyzing criminality. By grouping offenders at a county level, unemployment figures aggregated at the county level can be used to view the relationship between unemployment and selected crime components. Consistent with findings in the literature, the hypothesis for analyzing the relationship between unemployment and recidivism is areas of higher unemployment will reveal increases in recidivism. The unemployment variable is a measure of unemployment at the county level in Alabama. The figure given for each county is a mean of the unemployment rates for the twelve months in the 2004 calendar year. Table 3 is presented below showing the quantile distribution of the 2004 unemployment variable. The mean for the variable was 5.75 with a standard deviation of 1.76. Table 3 shows that 25% of Alabama counties have unemployment rates of 4.8 or under, 50% of counties have unemployment rates of 5.1 or under, and 75% of counties have unemployment rates of 6.4 or under. Approximately one half of the counties in Alabama have unemployment rates between 4.8 and 6.4 with the other half falling as low 2.3 and ranging as high as 15.7. Table 3: Distribution of Unemployment Rates in Alabama Counties in 2004 quantile Unemployment rate 100% 15.7 99.5% 14.0 97.5% 11.3 90.0% 7.9 75.0% 6.4 50.0% 5.1 25.0% 4.8 10.0% 4.2 2.5% 3.7 0.5% 2.3 0.0% 2.3 13 Table 3 shows the distribution of unemployment rates throughout Alabama for 2004 and demonstrates the variability of unemployment in the state. If there is a link between unemployment and recidivism in Alabama, recidivism should also fluctuate with unemployment figures. Recidivism is a topic that has recently received much attention in the social sciences and is increasingly garnering more attention at a national level. In 2002 alone, over 600,000 prisoners left state and federal prisons (Visher & Travis 2003). One of the largest recidivism studies indicates that nearly 68% of those released from prison will recidivate within 3 years (Bureau of Justice Statistics 2002). This study consisted of over 270,000 released prisoners from 15 states represented over 67% of inmates released in 1994. This work was compared to another recidivism study from prisoners released in 1983, also conducted by the Bureau of Justice Statistics, and found that the recidivism rate had significantly increased from 1983 to 1994. Recidivism, as previously mentioned, has been defined and measured in many different ways and therefore recidivism research covers a broad spectrum of topics and methodologies. Recidivism research has studied juveniles and adults, drug and alcohol treatment programs, gender and racial components, and severity of arrest among other topics. One article did conduct a contextual analysis as a basis for studying recidivism (Weidner, Frase and Schultz 2005). However this research emphasized the effects of case-level factors for studying jurisdictional differences in severity of sentencing. This study only used three variables in the analysis; the use of sentencing guidelines, level of crime committed, and racial composition of community to demonstrate the impact of contextual factors on imprisonment decisions. 14 Research in recidivism often focuses on the likelihood of re-offending based on previous arrest history (Smith & Gartin 1989). Gartin and Smith were interested in examining specific deterrence and the likelihood of committing further criminal acts. Others interested in deterrence have focused on the level of ties to conventional society of individuals and their offending habits (Dejong 1997). Dejong found that for those with few ties to the community, longer imprisonment terms were predictors of longer periods between arrests. The research also indicated those who were sentenced to incarceration and who were first time offenders were more likely to be arrested again. Many researchers have focused recidivism research on juvenile populations (Bernburg & Krohn 2003; Dean, Brame & Piquero 1996). Bernburg and Krohn?s study showed that early official intervention in juveniles resulted in increased levels of crime in early adulthood. They speculated that this happens because the intervention has negative effects on the educational attainment and employment histories of the youth. This idea is particularly interesting because it would imply that mechanisms designed to curb criminal behavior actual increase the behavior in a young population in a manner not planned. The study conducted by Dean, Brame, & Piquero demonstrated that there was no difference in offending patterns in youth regardless of whether or not they had an adjudication at an early or a late age. If the results of Dean, Brame, & Piquero are correct, the onset of criminal behavior of juveniles at a particular age would not be as important on future criminality as just simply offending while at a juvenile age. Results from these studies would be of great interest in conducting longitudinal analyses of offenders from juvenile ages through adulthood. 15 There is also evidence that drug offenders sentenced to prison are more likely to commit crime when released than those drug offenders sentenced to probation. Spohn & Holleran (2002) found no evidence that incarceration decreases recidivism among drug offenders and they further found that imprisonment could have a more profound criminogenic effect on drug offenders than non-drug offenders. This could be due to the effectiveness of treatment programs in prisons compared to ones on the outside of the prisons and /or the continued interaction with drug users while incarcerated. Gender differences in re-offending patterns have been studied and significant results have been detected (Benda 2005). Benda found that men are more likely to re- offend for different reasons than women. Men were likely to commit future crimes because of criminal peer associations, weapons, alcohol and aggressiveness. Job satisfaction and educational attainment were factors that were also more likely to impact males than females. One study focused on recidivism differences in inmates released from public or private prisons (Bales, Bedard, Quinn, Ensley & Holley 2005). A central claim to the establishment of the private prison industry was that these facilities would reduce recidivism among those inmates who were released. This study found no support for that claim. The research showed no significant differences in recidivism for adult males, adult females, or youthful offender males. Little research has been done combining numerous variables in a single model for recidivism. One study has compiled an analysis using age and employment to try and explain variations in recidivism (Uggen 2000). Transitions such as employment can be life-course variants (Sampson & Laub 1993; Warr 1998). Sampson and Laub?s research 16 indicates that certain factors, including employment and education, may be turning points for individuals and transition them from crime to legitimate means of financial opportunity. Uggen?s research found that work appears to be a turning point for reducing criminal activity for those over the age of 26 but not for those under the age of 26. One of the other apparent weaknesses in the literature is the lack of data containing a sufficient number of variables for a large group of incarcerated offenders so as to develop a powerful multivariate model. As demonstrated in the previous review of the literature, the vast majority of work on incarcerated offenders has involved treatment of a limited number of variables and has not been focused on an entire system of imprisoned offenders. The inherent difficulty involved in pulling together a study of the aforementioned group is the compilation of data and information for all the offenders. The measure of recidivism used for this study was the number of times an offender had been admitted to the Alabama Department of Corrections. The Department of Corrections measures this by assigning inmates a suffix (ais_suf) to the ais number. The suffix may actually indicate one of two things: (1) the number of times an inmate has been admitted to the Alabama Department of Corrections; or (2) a certain category of inmate. Inmates with no suffix are first time admissions to the Department of Corrections. For analysis purposes, inmates with no suffixes were given AA as the suffix. An inmate with the ais_suf A would be someone admitted for the second time, an inmate with the ais_suf B would be someone admitted for the third time and so on. These letter codes for the suffix indicate the number of admissions until the alphabet reaches the letter P. There are then six letters that indicate different types of classifications of inmates. The suffix of Q is a code used when either the number of admissions or the 17 classification of the offender is unknown. The suffix of P indicates that the offender has a prior record of being on parole in the past in Alabama. An inmate with a suffix of R indicates that this offender is restricted from participating in work release programs. A suffix of S denotes that the inmate is a sex offender. The S suffix indicates that the inmate is currently, or has in the past, served time for a sexual offense. The suffix of X is given to those offenders who have life sentences without the possibility of parole. Finally the suffix of Y shows that an offender is barred from the parole process in the state. The ais_suf serves as the only measure of recidivism supplied on an inmate?s record. The only other way to document prior felony history is a complicated process and is explained at the conclusion of the study. The suffix can indicate if an inmate has previously been admitted to the Alabama Department of Corrections in the past. The use of the suffix also has some limitations. If a P, Q, R, S, X or Y suffix is used, it is not always possible to decipher if the offender has any prior felony convictions. An inmate with a P suffix obviously has prior offenses since they have been previously been on parole, but one can not tell the extent of the criminal history simply by viewing the P suffix. The Q suffix was only utilized five times in the process so it is not considered to be a significant obstacle when viewing recidivism at this level. The suffixes of R, S, X, and Y can be used for first-time or repeat offenders. It is not always possible to determine if an offender is a recidivist if the offender has one of these four suffixes. For this reason, inmates with one of these four suffixes were excluded from recidivism analyses. The following is a distribution of the incarcerated population and associated ais suffixes. 18 Table 4: Frequency Distribution of AIS Suffixes of DOC Inmates ais_suf frequency percentage AA 11682 44.62 A 5007 19.13 B 1887 7.21 C 530 2.02 D 129 .49 E 26 .10 F 3 .01 G 2 .01 I 1 .00 P 29 .11 Q 5 .02 R 2084 7.96 S 3400 12.99 X 1381 5.28 Y 13 .05 TOTAL 26179 Table 4 shows that almost 45 percent of inmates were incarcerated in the Department of Corrections for the first time and another 19 percent had been admitted only twice. These two figures are conservative estimates of the first and second time admissions based on the problems of determining admission counts because of the dual use of some of the suffixes addressed above. Many analyses in criminology and social research have focused primarily on whether or not an offender was a repeat offender. The creation of a dummy variable based on the ais suffix allows analyses to look at these two different categories of offenders. The ais_dummy variable is a variable that measures if the offender is a recidivist. This is measured as a dummy variable with a code of 0 indicating the offender is a first time admittee to the Alabama Department of Corrections, and a code of 1 19 showing the offender is a recidivist and is serving at least his/her second incarceration. Offenders with an ais suffix of Q, R, S, X, and Y were excluded from this distribution because it is not always possible to determine if they are recidivists or not. With the exclusion of the offenders mentioned above, Table 5 indicates a little over 60 percent of the offenders were classified as first time offenders. Table 5: Frequency Distribution of Dummy Variable for Recidivism ais_dummy frequency percentage 0 11682 60.54 1 7614 39.46 TOTAL 19296 Social Science Research Example Two: Education and Crime The link between education and crime has been explored and investigated in previous research but little attention has been devoted to any possible relationship between educational attainment and recidivism. There has been a great deal of research conducted in the field of education within the correctional system and the associated impact that it has had on inmates and treatment programs. Little work has been done in the sociology or criminology fields exploring the link between educational attainment and future criminality among offenders. Most of the work done on education and crime asserts that increased education will decrease criminality with two notable exceptions (Grogger 1998; Tauchen, Witte and Griesinger 1994). These two studies found there to be no significant link between education and crime. These studies pointed out that other 20 factors, including unemployment, played powerful roles in criminal activity but that education was of minimal impact. Some research on educational attainment has revealed that it is not as highly correlated with other social problems as other variables (Schmitt 1956). Schmitt?s study found both educational attainment and suicide rates to have very low rates of correlation with forms of social disorganization in Hawaii. Others in the correctional field insist that educational attainment must be included in any meaningful study to accurately assess the success of the public or private prison system (Gaes 2005). Gaes? work centers around comparing the effectiveness of public and private prisons. He asserts that educational attainment, along with staff safety, disease control, and inmate misconduct are central to evaluating the success of any prison. Some researchers have attempted to address the gap in research by using education as a causal link in the commission of crime (Lochner and Moretti 2004). Lochner and Moretti lay out five reasons that educational attainment may affect criminality of individuals. First, increased schooling increases earning potential, making crime a costly economic venture. Second, punishment for those with more education is more costly than to those with less education because of the time out of the labor market. Third, more education could possibly increase patience of the student creating more awareness of future punishments. Fourth, increased education can affect the mindset of someone so that they are less inclined to break any laws. Fifth, compulsory education can keep children?s time occupied keeping them away from crime early, possibly lowering the probability of them committing crime later on in life. Lochner and Moretti contend that increased education will lead to less criminal activity early and later on in someone?s life. 21 Some research has focused specifically on the relationship between education and likelihood of recidivism (Ulmer 2001). This work demonstrated that education was significantly related to both the likelihood of being arrested and the severity of the offense an offender was arrested for committing. Other study in the area has involved looking at juveniles and their education/delinquency relationship (Hannon 2003; Hansen 2003). Hansen studied young males with different levels of educational attainment to try and decipher if evident relationships existed. Hannon looked at whether or not poverty had an impact on the relationship between educational attainment and delinquency. This study found that delinquency is less consequential for the educational attainment of disadvantaged youth than it is for advantaged youth. This analysis outlined how advantaged youth are more likely than disadvantaged youth to be set back in the educational system for committing delinquent acts. Delinquency is perceived to be far less common in advantaged communities so when a juvenile is labeled as a delinquent in an advantaged area, the juvenile is punished in a manner that may ostracize him/her from the educational cohort they were in before being labeled a delinquent. Hirschi (1969) also noted that when individuals have stakes in conformity, this influences their behavior. Hirschi said this type of commitment could lower criminal activity, particularly delinquency. Study done on incarceration has also shown that, even while controlling for other variables, education greatly contributes to the likelihood of incarceration (Pettit and Western 2004). Pettit and Western studied lifetime risks of imprisonment for black and white males by constructing a model and found that incarceration risks are ?highly 22 stratified? by education levels. This model showed that increased education led directly to lower chances of imprisonment. It is commonly believed that access to opportunities and resources decreases the likelihood of engaging in criminal acts. This theory has met with mixed results in some cases involving educational attainment (LaFree, Drass, and O?Day 1992; LaFree and Drass 1996). The 1996 study reveals findings that education and crime are contingent upon intraracial income disparity. For African Americans, increases in educational attainment are associated with rising arrest rates, but only during times of growing income inequality and for whites, increases in educational attainment are only associated with reduced crime rates during times of declining inequality. The 1992 study indicates that access to opportunities has the anticipated effect for whites but not blacks. Increased opportunity to resources resulted in lower levels of crime for whites but did not have a significant impact for blacks. Numerous studies have been conducted attempting to measure the results of educational programs with the correctional system. These studies generally seek to determine if the education provided while incarcerated is successful when the inmate is reintegrated into the community after release (Fabelo, 2002; Vacca 2004; Wilson, Gallagher & MacKenzie 2003). These studies indicate that education can have a lowering influence on criminal activity once released from incarceration, but that the education also needs to take the form of programs that will help inmates successfully reenter the community. The majority of the body of work pertaining to education and crime overwhelmingly indicates that increased education manifests itself in lowered levels of 23 crime and incarceration. Following results found in previous research, three hypotheses will be used to analyze the relationship between educational attainment and crime: (1) increases in educational attainment will be associated with a lower likelihood of committing violent offenses; (2) increases in educational attainment will be associated with lower recidivism; and (3) offenders with higher education will come from counties with lower unemployment rates. Education has been viewed as possibly one of the most important variables in attempting to explain crime rates and determine which populations would be more or less likely to engage in criminal activities. The education recode variable is a measure of the level of completed educational attainment. Inmates reported what the highest level of completed educational attainment was and then this information was recoded into this education variable with four categories. Table 6 below illustrates the distribution of educational attainment across the inmate population. A code of 1 indicates the inmates had no education, a code of 2 corresponds with an education somewhere between 1 st and 11 th grade, a code of 3 goes with high school graduates or inmates who hold GEDs, and a code of 4 represents inmates who have at least had some college. Nearly 60 percent of the inmate population has not finished a high school or an equivalency test. Only 8 percent of the inmate population has attended any college at all. Approximately 5 percent of the total population indicated they had no formal schooling. 24 Table 6: Frequency Distribution of Educational Attainment level of education frequency percentage 1 1302 4.98 2 14323 54.75 3 8345 31.90 4 2191 8.38 TOTAL 26161 Social Science Research Example Three: Marital Status and Crime Marital status has often been incorporated into efforts to explain stabilizing life factors that among other things, may reduce the likelihood of criminality. Numerous studies have shown that marital status is associated with recidivism (Landis, Mercer, and Wolff 1969, Petersilia 1985, Petersilia and Turner 1990). Benedict and Huff-Corzine (1997) did not find marital status to be as significant in their research, but they noted that their analysis was restricted exclusively to property offenders and did not include other offenders with drug or violent offenses. Lauritsen (2001) found that martial status is also an indicator of victimization in results from the National Crime Victimization Study. The inclusion of this variable allows analyses to be performed measuring any possible impact marital status has had on the inmate population?s criminal behavior. Two hypotheses will be used to analyze the The marital recode variable is a modified version of the original marital status variable found in the current inmate file data table. Inmates who were divorced but remarried were classified into the married category as were inmates who were widowed and remarried. Finally the unknown and other categories were combined to finish the recoding of the marital status variable. Table 7 shows that 25 76 percent of the inmate population were single and only a little over 10 percent of the inmates were married at the time of incarceration. Table 7: Frequency Distribution of Marital Status of Inmate Population marital status frequency percentage common law 324 1.24 divorced 2354 8.99 married 2646 10.11 other/unknown 151 .58 separated 596 2.28 single 19894 75.99 widow(er) 214 .82 TOTAL 26179 26 CHAPTER 3 ANALYSES In this chapter I will use the data set on the total inmate population of Alabama to describe and compare the development of two measures of violent offenders, present the cost analysis of proposed legislation increasing sentences for convicted sex offenders, examine the effect of education, marital status, and crime, and analyze the effect of unemployment on recidivism while controlling for race and education. Two measures of Violent Offenders: Distortions of Reality Two different definitions of determining whether or not a crime is considered violent were introduced in the background chapter. The two different definitions introduced were that of the Alabama Sentencing Commission and the Alabama Department of Corrections. Both are substantially different from what you would expect the man or woman on the street to classify as a violent crime. Due to the different definitions of determining if an offense is categorized as violent or non-violent, the Alabama Sentencing Commission and the Alabama Department of Corrections have different figures for the percentage of inmates incarcerated for violent offenses. Using the Alabama Sentencing Commission definition 27 of violent crime, 62.12% of the inmates are incarcerated for violent offenses, and if the Alabama Department of Corrections definition is used, the figure falls to 54.23%. As previously mentioned, the definition used by the Alabama Sentencing Commission includes attempts, conspiracies, and solicitations of all listed offenses as violent where the Alabama Department of Corrections does not. There are 38 different crimes, involving 2208 inmates, that the Alabama Sentencing Commission classifies as violent that the Alabama Department of Corrections does not. There are two primary categories of offenses that account for a large portion of the discrepancy between the two definitions. All drug trafficking offenses are considered to be violent by the Sentencing Commission and the Department of Corrections does not consider any of these to be violent. These trafficking cases account for 26.00% of the cases that are considered to be violent by the Sentencing Commission but not by the Department of Corrections. The larger discrepancy between the two definitions involves the offense of burglary. Burglary I is considered to be violent by both definitions, but only the Sentencing Commission classifies Burglary 2 nd and 3 rd degree offenses as violent. Burglary, in the 2 nd and 3 rd degrees, occurs when someone enters or unlawfully remains in building with intent to commit a crime in that location. The distinction between 2 nd and 3 rd degree burglary is an armed offender, or an offender that threatens with a deadly instrument or causes physical harm would be guilty of Burglary 2 nd degree. The inclusion of the other burglary offenses accounts for 70.02% of the remaining cases categorized as violent by the Sentencing Commission but not the Department of Corrections. These two groups of offenses together account for 96.02% of the difference between the two definitions. 28 The process of deciding what is and what is not a violent offense in these two different entities varied. The operational definition of violent crime arrived at by the Sentencing Commission included input from a variety of sources including district attorneys. The district attorneys voiced their opinion that more cases should be considered violent even without the use or threat of force or a weapon. The Department of Corrections definition leaves off crimes that many people outside of the Department of Corrections may consider to be violent. Both of the definitions include and exclude offenses that others outside of the field may consider to be violent or non-violent. Politics of Sexual Offenders: Fox Folly Example Bill O?Reilly?s show on child sex offender laws generated a lot of publicity, particularly in states identified by him to be soft on child sex offenders. Alabama was one of the states O?Reilly identified and urged people to voice their displeasure to the governor. O?Reilly did not divulge any information on what his presentation was based nor did he outline the laws or punishments he detailed on the air. The result was a large number of Alabama residents pressuring for harsher laws and punishments for child sex offenders. The show aired on Fox News and the reaction was very swift. Within days of the show airing, political pressure began to mount to initiate new legislation in reaction to the perceived leniency on child sex offenders. The Alabama Department of Corrections is consistently among the lowest funded Department of Corrections in the nation, and is 29 consistently among the most overcrowded. Without any analysis into the effects of this legislation, the movement gained steam and political clout. I prepared an initial analysis detailing the projected cost of the Sex Offender Notification Bill on the Alabama Department of Corrections (DOC). This bill strengthens punishments for sex offenses where the victim is less than twelve years of age. The punishments apply to Class A, B, and C felonies. This analysis is a conservative estimate of the additional costs to be incurred by DOC for housing inmates affected by this bill for longer periods of time. This analysis is a conservative estimate for three primary reasons. The impact of offenders sentenced to split sentences was not factored into the projections. A split sentence is a sentence where the offender is initially ordered to serve either 3 or 5 years in the prison system, and then the rest of the sentence is probation. All split sentences were removed before any analysis took place because from the data entry it was not always possible to tell if the sentence term was the sentence term or the number of years imposed as the split portion confinement. Of the 1908 sentenced cases for the 902 sex offenders in the data file, a little over 7% of the sentences were split sentences. Due to the limited number of splits that would be a part of the projected 58 admissions involving a child sex offense where the victim was under the age of 12, and the complexity of determining the length of the few number of split sentences, these sentences were not included in the analysis. However, the exclusion of split sentences should be taken into account when considering our conservative cost estimate because an offender that could be given a split sentence under current law (maximum to serve as 3 or 5 years) will now be required to serve a substantially longer 30 incarceration time - a minimum mandatory of 20 years for a Class A felony offender and a minimum of 10 years for a Class B felony offender, for an increase in time served of between 7 and 17 years. The impact of parole on time served was not factored in due to constraints with data. Data indicating the level of variation in sentence length as a result of parole was not available so this variable was also not included in the analysis. Finally, the cost figure used for housing a DOC inmate per day remained constant over the specified time periods and was not adjusted for inflation or possible increases in costs for new staff or construction of new facilities. Cases with sentences of 999 years were also excluded from this analysis. In order to make a projection on the future costs to DOC for this bill, it was first necessary to estimate the number of offenders that would be subject to the enhanced punishments provided by the bill. Figures on the number of admissions to DOC for sex offenses with a victim under the age of twelve were not available. Figures from DOC were provided for the number of admissions for sex offenses with a victim under the age of seventeen for 2000-2004. A data request was made to the Alabama Criminal Justice Center for the number of reported incidents of sex offenses with victims under the age of 12 and the number of reported incidents of sex offenses with victims under the age of 17 for 2000-2004. The percentage of DOC admissions for sex offenses with a victim under the age of 17 compared to the total number of reported incidents of sex offenses with a victim under the age of 17 for each year (2000-2004) was calculated. This percentage was then used to estimate the number of admissions to DOC for sex offenses with a victim under the age of 12. The above percentage was multiplied by the number of 31 reported incidents for sex offenses with a victim under the age of 12 to arrive at an estimate of the number of DOC admissions for sex offenses with a victim under the age of 12. This calculation was performed for each individual year from 2000-2004. The mean number of projected admissions for sex offenses with a victim under the age of 12 was 58. This figure is the number used as the estimate of the number of offenders to enter DOC in 2006 for sex offenses with a victim under the age of 12. The next step in the process was to determine what percentage of projected admissions to DOC for sex offenses with a victim under the age of 12 would fit into Class A, Class B, or Class C felony categories. A list of offenders currently serving time for a sex offense with a victim under 17 was obtained from DOC. This list contained offenders with a single sex offense and offenders that had multiple sex offenses. For identification purposes the most serious sexual offense for offenders with multiple cases, only the case with highest class of felony was kept. This resulted in a data file with only one record for each offender. 43.3% of the offenders had a Class A felony as their most serious offense, 28.3% of the offenders had a Class B felony as their most serious offense, and 28.3% of the offenders had a Class C felony as their most serious offense. These percentages were then applied to the projected number of 58 admissions to DOC for 2006 with sex offenses with a victim under the age of 12. The expected number of admissions to DOC for Class A felonies with sex offenses with a victim under the age of 12 was 25.114, and 16.414 for both Class B & Class C felonies with sex offenses with a victim under the age of 12. The decimals can be used, even though the figures represent humans, because we are using the figure as an estimate for a period of time and not necessarily a single year. 32 The projected number of Class A felony sex offenses with a victim under the age of 12 was 25.114. Class A felons are not eligible for goodtime, and parole considerations were not part of this analysis, so Class A felons sentences were viewed as day for day in terms of expected time to serve. The mandatory minimum for a sex offense involving a victim under the age of 12 is 20 years in the proposed legislation. Therefore Class A felons with a sentence of 20 years or more would not be influenced by the proposed mandatory minimum assuming day for day time served. The analysis indicated that 54.7% of Class A felony sex offenders had sentences under 20 years in length. Of the 25.114 admissions who were Class A sex felons, 13.74 would have sentences under 20 years. The mean sentence length, of those Class A felony sex offenders with sentences under 20 years, was 13.57 years. Assuming day for day punishment, 13.57 years equates to 164.88 months. The mandatory minimum is 20 years (240 months) so the difference between the mandatory minimum and the mean sentence for those sentenced to fewer than 20 years is 75.12 months. The anticipated additional cost to DOC for the Class A sex offenders admitted in 2006 who would be expected to be impacted by the mandatory minimum through the 20 year period can be calculated as follows: 13.74 (expected people to be impacted) X 75.12 (average anticipated increase in time served in months) X 1002.64 (cost of housing a DOC inmate for one month) _______ $1,034,873 The projected number of admissions in 2006 of Class B offenders with sex offenses with a victim under the age of 12 is 16.414 offenders. The mean sentence for 33 Class B sex offenders was 11.21 years or 134.52 months. The Alabama Sentencing Commission?s Initial Report to the Legislature in 2002 provides a schedule to determine expected time to serve based on sentences 15 years or less, or sentences above 15 years that are consecutive. The expected time to serve on a 11.21 year sentence is 42.72 months. The minimum mandatory sentence for a Class B sex offense with a victim under the age of 12 is 10 years or 120 months. The difference between the mandatory minimum confinement term and the average expected time to serve for each inmate is 77.28 months. The anticipated additional cost to DOC for the Class B sex offenders admitted in 2006 who would be expected to be impacted by the mandatory minimum through the 10 year period can be calculated as follows: 16.414 (expected people to be impacted) X 77.28 (average anticipated increase in time served in months) X 1002.64 (cost of housing a DOC inmate for a month) ______ $1,271,822 The projected number of admissions in 2006 of Class C offenders with sex offenses with a victim under the age of 12 is 16.414. For the analysis involving the Class C felony sex offenders, six records were deleted because the sentence was over 20 years. These records were deleted because the over 20 year sentences could only be the result of habitual offender status being applied to the offender and not a result of punishment only relating to the single offense. The proposed legislation would not initiate a mandatory minimum for Class C felony sex offenders with a victim under the age of 12 but would restrict these offenders from probation, parole, split sentences, or good time. These 34 offenders would be serving day for day prison terms. The mean sentence length for felony Class C sex offenders was 8.99 years or 107.88 months. The expected time to serve for a sentence of 8.99 years (107.88 months) is 35.15 months. The difference between the sentence (would be time served under new bill) and expected time to serve on a 8.99 year sentence is 72.73 months. The anticipated additional cost to DOC for the Class C sex offenders admitted in 2006 who would be expected to be impacted by the new restrictions on time to be served on sentences for the duration of their sentences can be calculated as follows: 16.414 (expected people to be impacted) X 72.73 (average anticipated increase in time served in months) X 1002.64 (cost of housing a DOC inmate for a month) ______ $1,196,941 The conservative estimate for additional costs to the Department of Corrections for each year of admissions of sex offenders with victims under the age of 12 would total $3,503,636. This cost is the conservative projected additional cost over the 20 year period for each year of child sex offenders with victims under the age of 12 admissions. This figure illustrates the cost per year cohort of child sex offender admissions that misinformation can lead to when supported by powerful political decision makers. 35 Social Science Research Example One: Unemployment and Recidivism Hypothesis: Areas of Higher Unemployment will Reveal Increases in Recidivism As noted in the background section, unemployment can be used as a variable of interest to many scholars and social science researchers regarding the relationship with crime. The unemployment variable can be used to provide a contextual element to the analysis of crime and repeat offenders. Unemployment has previously been mentioned in the literature to have many possible effects on the prevalence and trends of crime. Can unemployment be used as a predictor variable to estimate the level of recidivism? The literature suggests that areas of high unemployment experience higher rates of criminal activity than do areas that have lower rates of unemployment. If this information is used as basis to view the data in this study, then offenders from areas (counties) with higher unemployment rates should be more likely to be recidivists than offenders from areas with low unemployment rates. When the unemployment rate of a county is used to predict the likelihood of recidivism, the hypothesis is the probability of being a recidivist should decline as the unemployment rate declines as well. Table 8 indicates the relationship between unemployment rates and the likelihood of recidivism. This relationship is significant at the p<.008 level. Unemployment rates and the likelihood of recidivism are significantly related, however the direction of the relationship is the opposite of the predicted direction. The nature of the relationship shows that as unemployment rates increase, the probability of recidivism actually declines. As the unemployment rate rises by 1% , the probability of being a recidivist decreases by .00712. The strength of the relationship is weak, but significant nonetheless 36 because of the large sample size used in this analysis. The findings in Table 8 contradict the original hypothesis. Table 8: Relationship of Unemployment on Recidivism Prob>ChiSquare Wald .008 RSquare(U) .00712 Predicted probability Unemployment=0 Unemployment=1 .43497 .42785 Social Science Research Example Two: Education and Crime Hypothesis 1: Increases in Educational Attainment will be Associated with a Lower Likelihood of Committing Violent Offenses Hypothesis 2: Increases in Educational Attainment will be Associated with Lower Recidivism Hypothesis 3: Offenders with Higher Education will come from Counties with Lower Unemployment Rates The literature suggests that as education increases, the likelihood of committing violent offenses should decrease. It was suggested in some of the literature that people with more education have more to lose from committing crime and would stand to lose more for more serious or violent offenses. If education impacts the commission of violent crime as expected, those people with more education should commit less violent crime than those with less education. One would expect that as education level increases that the likelihood of committing a violent offense would decrease. Table 9 uses the 37 educational recode variable and the Alabama Sentencing Commission definition of violent crime (see Appendix A) to examine this relationship. Table 9 shows the relationship between educational attainment and the commission of violent crime to be significant at the p<.01 level. There are two interesting relationships that emerge from this table. The first relationship is that people with no education are far less likely to be imprisoned for a violent offense than other educational levels. 24.54% of inmates with no education are serving prison sentences for violent offenses compared to 62.68% of those not finishing high school, 66.97% of inmates with high school degrees, and 63.12% of inmates with at least some college. These results effectively go against the original hypothesis and show that increased education does not reduce the likelihood of being incarcerated for violent offenses. The only significant relationship emerging from the table is those inmates with no education are significantly less likely to be in incarcerated for violent offenses than those with more education. The next relationship that is of interest is actually that there is no relationship between the two variables for the other educational levels. Excluding those inmates with no education, the commission of violent offenses does not appear to be dependent upon educational level whatsoever. The original hypothesis about this relationship is discounted and one element of contrary evidence has been found in the data. 38 Table 9: Impact of Education on Commission of Violent Crime No education < 12 th grade HS grad college Not violent 981 75.46% 5331 37.32% 2751 33.03% 807 36.88% 9870 Violent 319 24.54% 8952 62.68% 5577 66.97% 1381 63.12% 16229 1300 14283 8328 2188 26099 *p<.01 Cramers V=.1822 The literature suggests that people with more education will not commit as much crime as less educated people because the effects of the criminal behavior are more costly to them as opposed to less educated people. If this is true, then more educated people should have lower amounts of recidivism than lower educated groups of people. Offenders with higher amounts of education should then be assumed to be largely first time offenders as compared to lower education groups where one would expect to find more repeat offenders. The working hypothesis for this analysis will be that inmates with less education would be more likely to be recidivists. Table 10 offers varying levels of support for this hypothesis. The results from Table 10 are significant at the p<.01 level. 8.29 percent of offenders with no education were recidivists, 44.04 percent of offenders with less than a 12 th grade education were recidivists, 38.93 percent of offenders with high school diplomas were recidivists, and 34.87 percent of offenders with at least some college were recidivists. There is a gradual decline in the recidivism percentages from the less than 12 th grade education level down to the some college level. However, the lowest percentage, by a large margin, was the 39 offenders with no education category. This group was 5 times less likely to be recidivists than the offenders that had some college. While there was evidence that increased education was associated with decreased levels of recidivism, the lowest education level exhibited the lowest recidivism percentage of all the categories. Table 10: Impact of Education on Recidivism No education <12 th grade HS grad college 1st time admission 1150 91.71% 5968 55.96% 3570 61.07% 992 65.13% 11680 recidivist 104 8.29% 4696 44.04% 2276 38.93% 531 34.87% 7607 1254 10664 5846 1523 19287 *p<.01 Cramers V=.1790 There is a large body of literature that asserts that unemployment is related to criminality. One of the ways to measure the impact of unemployment on crime at the county level is to view the level of education of each offender and then look at the unemployment rate of the county where the offender committed the crime. Most of the literature suggests that areas of higher unemployment will have more crime. People with higher amounts of education are expected to come from areas with lower unemployment rates. Does this relationship remain for the incarcerated population as well? If the hypothesis is correct, as educational attainment increases the unemployment rate of the counties of the offenders should decrease. Table 11 shows the relationship between educational attainment and the unemployment rate at the county level from which the offender committed the offense for which he/she was incarcerated. The findings reported in Table 11 are significant at the p<.01 level. The hypothesis that lower unemployment rates would be associated with higher educated inmates is supported by the results in Table 11. Offenders with no 40 education came from counties with a mean unemployment rate of 5.90, offenders with less than 12 th grade education came from counties with a mean unemployment rate of 5.79, offenders with a high school diploma came from counties with a mean unemployment rate of 5.70, and offenders with at least some college education came from counties with a mean unemployment rate of 5.55. Each advance in the educational attainment of offenders resulted in lowering unemployment rates. The hypothesis that more educated offenders would come from counties with lower unemployment rates is supported. The strength of the relationship is given by a R square of .0020 indicating that 0.20 percent of the variation in the county unemployment rate can be explained by the educational attainment level of the offender. Offenders with higher education do seem to come from counties with lower unemployment rates. The mean unemployment rates do decrease with additional education, however the unemployment rates do not reduce drastically. Table11: Means for Unemployment Rate by Education Level Level of education Frequency Mean Unemployment Rate No education 1302 5.90 < 12 th grade 14323 5.79 HS grad 8345 5.70 college 2191 5.55 *p<.01 R2=.001968 41 Social Science Research Example Three: Marital Status and Crime Hypothesis 1: Married Offenders will be Less Likely to Commit Violent Offenses Hypothesis 2: Married Offenders will be Less Likely to be Recidivists Marital status was earlier discussed as a possible stabilizing factor for individuals. If indeed marriage was a stabilizing factor in life course events, then it would follow that individuals who were married would be less likely to commit crime and therefore be incarcerated and singles would be more likely to be serving time for offenses. Table 12 shows the relationship between marital status and the commission of violent offenses as defined by the Alabama Sentencing Commission. Table 12 shows the relationship between marital status and the commission of violent crime. Marital status is used as a predictor variable for the commission of violent offenses. The hypothesis used for analysis for this table would be that married offenders should show a lower probability of committing a violent offense than other offenders. The relationship between marital status and the commission of a violent offense in Table 12 is significant at the p<.01 level. The interesting relationship that emerges out of this table is one category of marital status is significantly related to the likelihood of committing a violent offense. Those in the widow(er) category are more likely than the other categories to commit violent offenses. The categories of the marital status variable show similar percentages committing violent offenses with the exception of the widow(er) category. Categories other than widow(er) show a percentage range between 59.57 and 62.21 committing a violent offenses. The widow(er) category shows 77.57 percent committed a 42 violent offense. 62.21 percent of single offenders committed a violent offense, 61.34 percent of separated offenders committed a violent offense, 61.59 percent of unknown marital status committed a violent offense, 62.10 of married offenders committed a violent offense, 61.24 percent of divorced offenders committed a violent offense, and 59.57 percent of common law offenders were currently imprisoned for a violent offense. The results from Table 12 do not provide support for the hypothesis supplied for the relationship between the two variables. The only relationship that emerged from the table was that those in the widow(er) were significantly more likely to commit a violent offense than offenders in other categories of the marital status variable. Evidence for the initial hypothesis is not present and the hypothesis is rejected. If marital status provides a stabilizing factor, recidivism should be lower among those offenders who are married and higher among those who are single. Table 13 shows the relationship between marital status and recidivism. The results from Table 13 are significant at the p<.01 level. The results from Table 13 do not lend support to the hypothesis that married people are less likely to be recidivists or that singles would be more likely to be recidivists. 28.36 percent of singles were recidivists compared to 31.52 percent of married offenders that were recidivists. The category with the highest recidivism percentage was common law marriage and the category with the lowest recidivism percentage was the widow(er) category. There is no evidence from this data to conclude that marriage has a stabilizing factor that inhibits repeat crime. Table12: Impact of Marital Status on Commission of Violent Offenses P<.0006 Cramers V=.0300 Common law Divorced Married Other/ Unknown Separated Single Widow(er) Non- Violent Offense 131 40.43 909 38.73 998 37.90 58 38.41 230 38.66 7503 37.79 48 22.43 9877 Violent Offense 193 59.57 1438 61.27 1635 62.10 93 61.59 365 61.34 12350 62.21 166 77.57 16240 324 2347 2633 151 595 19853 214 26117 43 Table 13: Impact of Marital Status on Recidivism P<.0001 CramersV=.0471 Common law Divorced Married Other/ Unknown Separated Single Widow(er) Non- recidivist 195 60.19 1629 69.20 1812 68.48 99 65.56 398 66.78 14253 71.64 179 83.64 18565 Recidivist 129 39.81 725 30.80 834 31.52 52 34.44 198 33.22 5641 28.36 35 16.36 7614 324 2354 2646 151 596 19894 214 26179 44 45 Whole Model on Recidivism One of the apparent gaps in the available literature in the field of sociology and criminology is the appearance of few multivariate quantitative attempts to predict or explain recidivism. The variables used in previous analyses in this study: unemployment, educational attainment, and commission of violent crime will now be joined with race to form a larger model to view recidivism. These four variables will serve as independent variables to see what, if any, controlling impact they have on recidivism among the incarcerated population. The resulting model was a logistic regression predicting the probability of an offender being a recidivist based on categories of the four independent variables used in the analysis. Table 14 shows the significance and strength of the model as a whole. The model was significant at the p<.0001 level and had a low strength. The strength of the model is weak but is significant because of the large number of cases used in the study. Table14: Logistic Regression Whole Model Significance and Strength Prob>Chi Square <.0001 R Square(U) .0626 Each of the four predictor variables used in the analyses will be analyzed to the view the nature of the relationship between them and the probability of recidivism. Race is a controversial issue in virtually all social science topic areas, and especially in the criminal justice field. Does race have an impact on recidivism in this model? Table15 shows the impact and significance values of race on the probability of recidivism. Race was significant at the p<.001 level, but the impact of race on recidivism 46 is somewhat controlled for by the other variables used in the model. For the analysis only whites and blacks were used. The other three categories of race were excluded from the analysis as a result of very small sample sizes. Table15: Impact of Race on Whole Model for Recidivism Prob>Chi Square Wald .0000 Predicted probability Black .41536 Predicted probability White .36137 The probability of being a recidivist is only .05399 higher for blacks than it is for whites. In the background section of this work, it was noted that a large portion of the inmate population was black. Loading other variables into the model minimizes the impact that race has on recidivism. This is evidence that other social variables can partly control for the effect of race on recidivism. Does the controlling offense that the offender is currently serving time for explain any part of the relationship with recidivism? The Alabama Sentencing Commission definition of violent crimes was used again to determine if the inmate was currently incarcerated for a violent offense. Table 16 indicates that this variable is significant at the p<.0001 level and is a significant factor in the model. Table 16 indicates the expected probability values to be a recidivist based on whether or not the offender was currently incarcerated for a violent crime or not. The results show that if the offender is incarcerated for a violent offense he/she, the probability of the offender is a recidivist decreases .16798. This result may seem surprising, but is likely the result of longer sentences mandated for violent offenses resulting in a shorter span to commit other offenses. 47 Table16: Impact of Committing Violent Crime on Whole Model for Recidivism Prob>Chi Square Wald .0000 Predicted Probability for Violent Offense .31456 Predicted Probability for Non-Violent Offense .48254 The unemployment variable was previously used in an analysis and the impact was found to be contrary to the expected result. The same unexpected result holds true in the whole model test as well. Table 17 shows the significance level of county unemployment rates in the model and also indicates the expected probabilities of being a recidivist if a county with a unemployment rate of 6.5 and a county with an unemployment rate of 7.5. Table17: Impact of Unemployment Rates on Whole Model for Recidivism Prob>Chi Square Wald .0010 Predicted Probability for County with Unemployment Rate of 6.5 .40396 Predicted Probability for County with Unemployment Rate of 7.5 .38551 The results of Table17 are significant at the p=.001 level and indicate that as unemployment rises, the probability of being a recidivist actually decreases. The nature of the relationship is that as the unemployment rate rises by 1.00, the probability of being a recidivist decreases by .01845. Educational attainment has been used in examining the commission of violent crime, unemployment rates, and has been used in a bivariate analysis of recidivism. Does the addition of the other variables in the model change the impact that education has on recidivism? The whole model logistic regression indicates that the inclusion of the other variables does not change the relationship between level of education and the probability 48 of being a recidivist among the inmate population. Table 18 shows that the educational attainment variable is significant at the P<.001 level. Table 18 also indicates the expected probability values of being a recidivist based on educational attainment categories. Table18: Impact of Education on Whole Model for Recidivism Prob>Chi Square Wald .0000 Predicted Probability for No Education .08307 Predicted Probability for <12 th grade .44051 Predicted Probability for HS grad .38860 Predicted Probability for college .34960 The effect of education on recidivism in the whole model virtually mirrors the bivariate relationship. The expected probabilities in the whole model are nearly identical to the percentages found in the bivariate table. Both the bivariate table and the whole model indicate that inmates with no education are decidedly less likely to be recidivists than other offenders falling in the other educational categories. In both the model and the bivariate table, inmates having less than a 12 th grade education are shown to have the highest likelihood of being a recidivist. The relationship for the other two educational categories is consistent in both tables as well showing that inmates possessing a high school degree or the equivalency are more likely to be recidivists than those who have at least attended some college. The whole model shows the impact of education on recidivism was not mitigated by the addition of the other independent variables. The impact of unemployment rates on recidivism was again shown to have the opposite impact on recidivism that was previously stated in another previous analysis as a hypothesis. The impact of race on recidivism appears to have been somewhat controlled for by the addition of the other 49 variables. The whole model approach is an important approach in examining the relationship between the variables to establish any possible change in impact on the recidivism variable. 50 CHAPTER 4 CONCLUSION The available data on inmates in the custody of the Alabama Department of Corrections is primarily used and recorded for administrative purposes. Many parties outside of the Department of Corrections have research interests, data requests, and policy questions that could be met more effectively met if the data collected on the inmate population was organized and collected more efficiently. Criminal justice administrators, political leaders, and scholars would be in a better position to analyze data and make policy decisions by having access to a more unified and comprehensive compilation of data than is currently available. The inclusion of certain social variables to a larger, unified data set would also allow more complex analyses to be performed on the inmate population. The addition of these social variables would be of particular interest to scholars and other criminal justice researchers to identify possible relationships and trends in the data. One of the major components of the classification process of offenders is identifying if the offense committed is deemed to be violent or not. The two definitions primarily used in Alabama to determine if an offense is violent are the Alabama Sentencing Commission?s definition and the Alabama Department of Corrections? definition. Depending on which definition is used, the percentage of offenders considered violent can fluctuate considerably. 51 The variability between the two definitions of violent crime can be almost entirely explained by the difference of classification of a relatively small number of offenses. The Sentencing Commission?s definition includes Burglary 2 nd & 3 rd Degrees and Trafficking offenses, whereas the definition of the Department of Corrections does not include these offenses as violent. The difference with the classification of these offenses accounts for over 96 percent of the discrepancy between the two definitions. Another important distinction in the inclusion of violent offenses with the two different definitions is the Department of Corrections definition only includes offenses for which an offender has previously been incarcerated. The Alabama Sentencing Commission?s definition uses all possible offenses as a frame for selecting violent offenses from, but the Department of Corrections only includes offenses deemed violent on a discretionary basis when an offender is admitted for an offense. An example of how the current system of measuring, storing, and compiling data in the criminal justice system in Alabama is hindering research and policy decisions is the sex offender legislation put in motion by Bill O?Reilly?s piece on what were deemed to be soft laws and punishments in Alabama for sex offenders with child victims. Currently the Department of Corrections only the records the age of victim as under 17 and 17 and over. The new proposed legislation dealt with sex offenders with victims under the age of 12. Analysis using records and data from outside agencies was required to make a projection of the number of offenders who would be admitted yearly for these specific offenses. Additional analysis was performed to forecast the number of offenders in different felony groups and the anticipated difference in time served due to harsher penalties. The analysis revealed that the proposed legislation would cost 52 approximately 3.5 million dollars extra for each group of sex offenders admitted yearly over the course of their incarceration than is budgeted for current incarcerations. The complexity of the analysis could have cut down, greatly increasing the certainty of the results, if the records kept on offenders and offenses were altered to better accommodate data analyses. Policy decisions based on the analyses would then be based on more accurate and complete information. To illustrate the dynamic between unemployment and recidivism, a dummy variable for recidivism was used in conjunction with county level unemployment rates to determine the nature of the relationship. There is a significant relationship between the variables, but the results indicated that as unemployment rates increase, the likelihood of being a recidivist actually decreases. This result contradicted the original hypothesis that lowering unemployment rates would be accompanied by a lower probability of recidivism. Educational attainment has often been linked to criminal activity. This work examined if educational attainment was related to the commission of violent offenses and recidivism. The educational attainment variable was found to be associated with both, however some unexpected results emerged from the data. Offenders with no education were the least likely to be a recidivist and also the least likely group to have committed a violent offense. This finding was in direct opposition to the hypotheses proposed for both relationships. Very little, if any, relationship existed between other educational levels and the commission of violent crimes also contradicting the original hypothesis. Increased educational levels, other than those with no education, were associated with a lower likelihood of being a recidivist 53 providing support for the hypothesis that increased education would have a negative impact on the probability committing subsequent criminal acts. The relationship between marital status and the commission of violent offenses and recidivism also produced an interesting finding. The hypotheses for these analyses asserted that individuals who are married would be less likely to be recidivists and less likely to commit violent crimes than singles. Results from both of these analyses did not support that claim, but did reveal other findings. Widow(er)s were the marital group most likely to commit a violent offense and the least likely to be recidivists. This marital group was the only category that exhibited any measurable differences in the commission of violent offenses and differed the most, by a large margin, in the likelihood of being recidivists. The data indicated that marriage was not a stabilizing factor in criminal activity or the severity of offense. One of the areas in criminal justice research receiving attention on a continuing basis is recidivism. The variable of recidivism in the Department of Corrections is coded in a manner that is not exclusive resulting in difficulties interpreting prior criminal history of inmates. Any analysis using recidivism, as measured by the Department of Corrections, as a variable must have proper caveats attached to properly identify problems in the measurement of the variable and interpretation of results. Analysis using the recidivism variable in the Department of Corrections data is difficult but can be done by carefully selecting cases and properly interpreting results. In order to gauge the effects of unemployment rates, educational attainment, commission of a violent offense, and race on the likelihood of recidivism, all four variables were used as independent variables in a single model analysis using 54 recidivism as the dependent variable. The strength of using all the variables in a single model is the detection of controlling effects that the variables may produce on each other. All four of the independent variables were found to be significant factors in the model. The whole model itself was found to have a significant relationship existing, however the strength of the model was weak. Evidence was found in the analysis of the results that the introduction of variables into the model mitigated the strength of one variable, but not another variable in the model. Race was a significant factor in the model, but the effects of race were somewhat controlled for with the introduction of the other variables into the model. The introduction of all the variables into the model did not change the impact of educational attainment on recidivism. The relationship between educational attainment and recidivism remained almost identical to the bivariate relationship between the two variables. Offenders with no education were still the most likely not to be recidivists, while the likelihood of being a recidivist declined with successive advances in educational attainment. 55 REFERENCES Bales, William D., Laura E. Bedard, Susan T. Quinn, David T. Ensley, and Glen P. Holley. 2005. ?Recidivism of Public and Private State Prison Inmates in Florida.? Criminology & Public Policy 4(1): 57-83. Benda, Brent B. 2005. ?Gender Differences in Life-Course Theory of Recidivsim: A Survival Analysis.? International Journal of Offender Therapy & Comparative Criminology 49(3): 325-343. Benedict, W. Reed and Lin Huff-Corzine. 1997. ?Return to the scene of the punishment: Recidivism of adult male property offenders on felony probation, 1986-1989.? Journal of Research in Crime and Delinquency 34(2): 2-18. Berk, Richard A., Kenneth J. Lenihan, and Peter H. Rossi. 1980. ?Crime and Poverty: Some Experimental Evidence from Ex-Offenders.? American Sociological Review 45(5): 766-787. 56 Bernberg, Jon Gunnar and Marvin D. Krohn. 2003. ?Labeling, Life Chances, and Adult Crime: the Direct and Indirect Effects of Official Intervention in Adolescence on Crime in Early Adulthood.? Criminology 41(4): 1287-1319. Bureau of Justice Statistics, U.S. Department of Justice, Office of Justice Programs. 2002. ?Recidivism of Prisoners Released in 1994.? Washington , DC: Bureau of Justice Statistics. Retrieved May 1, 2004 (http://www.ojp.usdoj.gov/bjs). Cappell, C. & Sykes, G. 1991. ?Prison commitments, crime, and unemployment: a theoretical and empirical specification for the United States, 1933-1985.? Journal of Quantitative Criminology 7: 155-199. Carlson, Susan M. and Raymond J. Michalowski. 1999. ?Unemployment, imprisonment, and social structures of accumulation: historical contingency in the Rusche-Kirchheimer hypothesis.? Criminology 37(2): 217-250. Chiricos, T. 1987. ?Rates of Crime and Unemployment: An Analysis of Aggreagate Research.? Social Problems 34(2): 187-211. Dean, Charles W., Robert Brame, and Alex R. Piquero. 1996. ?Criminal Propensities, Discrete Groups of Offenders, and Persistence in Crime.? Criminology 34(4): 547-575. 57 Dejong, Christina. 1997. ?Survival Analysis and Specific Deterrence: Integrating Theoretical and Empirical Models of Recidivism.? Criminology 35(4): 561-576. Ekland-Olson, Sheldon, William R. Kelly, and Michael Eisenberg. 1992. ?Crime and incarceration: some comparative findings from the 1980s.? Crime and Delinquency 38(3): 392-416. Fabelo, Tony. 2002. ?The Impact of Prison Education on Community Reintegration of Inmates: The Texas Case.? Journal of Correctional Education 53(3): 106-111. Gaes, Gerald G. 2005. ?Prison Privatization in Florida: Promise, Premise, and Performance.? Criminology & Public Policy 4(1): 83-89. Galster, George C. and Laure A. Scaturo. 1985. ?The U.S. Criminal Justice System: Unemployment and the Severity of Punishment.? Journal of Research in Crime and Delinquency 22(2): 163-190. Grimes, Paul W. and Kevin E. Rogers. 1999. ?Truth-in-sentencing, law enforcement, and inmate population growth.? Journal of Socio-Economics 28(6): 745-758. Grogger, J. 1998. ?Market Wages and Youth Crime.? Journal of Labor Economics 16(4): 756-791. 58 Hannon, Lance. 2003. ?Poverty, Delinquency, and Educational Attainment: Cumulative Disadvantage or Disadvantage Saturation?? Sociological Inquiry 73(4): 575-595. Hansen, Kristine. 2003. ?Education and the Crime-Age Profile.? British Journal of Criminology 43(1): 141-169. Hirschi, Travis. 1969. Causes of Delinquency. Berkeley, CA. University of California Press. LaFree, Gary, Kris A. Drass, and Patrick O?Day. 1992. ?Race and Crime in Postwar America: Determinants of African-American and White Rates, 1957-1988.? Criminology 30(2):157-188. LaFree, Gary and Kriss A. Drass. 1996. ?The Effect of Changes in Intraracial Income Inequality and Educational Attainment on Changes in Arrest Rates for African Americans and Whites, 1957 to 1990.? American Sociological Review 61(4): 612-633. Landis, Judson R., Jams K. Mercer, and Carole E. Wolff. 1969. ?Success and Failure of Adult Probationers in California.? Journal of Research in Crime and Delinquency 6: 34-40. 59 Lauritsen, J. L. 2001. ?The Social Ecology of Violent Victimization: Individual And Contextual Effects in the NCVS.? Journal of Quantitative Criminology 17(1): 3-33. Lochner, Lance and Enrico Moretti. 2004. ?The Effect of Education on Crime: Evidence from Prison Inmates, Arrests, and Self-Reports.? American Economic Review 94(1): 155-190. Petersilia, Joan. 1985. Probation and Felony Offenders. Washington, DC. U.S. Department of Justice. Petersilia, Joan and Susan Turner. 1990. ?Comparing Intensive and Regular Supervision for High-Risk Probationers: Early Results from an Experiment in California.? Crime & Delinquency 36: 87-111. Pettit, B and B. Western. 2004. ?Mass Imprisonment and the Life Course: Race and Class Inequality in U.S. Incarceration.? American Sociological Review 69(2): 151-170. Quimet, M. & Tremblay, P. 1996. ?A Normative Theory of the Relationship Between Crime Rates and Imprisonment Rates: An Analysis of the Penal Behavior of U.S. States from 1972-1992.? Journal of Research in Crime and Delinquency 33(1): 109-126. 60 Sampson, Robert J. and John H. Laub. 1993. Crime in the Making: Pathways and Turning Points through Life. Cambridge, MA: Harvard University Press. Selke, William. 2003. ?A Test of Social Stress Theory as Applied to the Study of Imprisonment Rates.? Prison Journal 83: 426-442. Schmitt, Robert C. 1956. ?Intercorrelations of Social Problems Rates in Honolulu.? American Sociological Review 21(5): 617-620. Smith, Douglas and Patrick R. Gartin. 1989. ?Specifying Specific Deterrence: the Influence of Arrest on Future Criminal Activity.? American Sociological Review 54(1): 94-107. Spohn, Cassia and David Holleran. 2002. ?The Effect of Imprisonment on Recidivism Rates of Felony Offenders: a Focus on Drug Offenders.? Criminology 40(2): 329-358. Tauchen, H., A.D. Witte, and H. Griesinger. 1994. ?Criminal Deterrence: Revisiting the Issue with a Birth Cohort.? Review of Economics and Statistics 76(3): 399-412. Uggen, C. 2000. ?Work as a Turning Point in the Life Course of Criminals: a Duration Model of Age, Employment, and Recidivism.? American Sociological Review 67: 529-546. 61 Ulmer, Jeffrey T. 2001. ?Intermediate Sanctions: A Comparative Analysis of the Probability and Severity of Recidivism.? Sociological Inquiry 71(2): 164. Vacca, James S. 2004. ?Educated Prisoners are Less Likely to Return to Prison.? Journal of Correctional Education 55(4): 297-306. Visher, Christy A. and Jeremy Travis. 2003. ?Transitions from Prison to Community: Understanding Individual Pathways.? Annual Review of Sociology 29: 89-114. Warr, Mark. 1998. ?Life-Course Transitions and Desistance from Crime.? Criminology 36: 183-216. Weidner, Robert R., Richard S. Frase, and Jennifer S. Schultz. 2005. ?The Impact of Contextual Factors on the Decision to Imprison in Large Urban Jurisdictions: A Multilevel Analysis.? Crime & Delinquency 51(3): 400-425. Wilson, D., C. Gallagher, & D. MacKenzie. 2003. ?Corrections-Based Education, Vocation, and Work Programs for Adult Offenders.? Journal of Research in Crime and Delinquency 37: 347-368. 62 APPENDIX A A1-A13 Codebook & Description of Unemployment Variable A1 Alabama Institutional Serial Suffix (ais suffix) AA 1 st admission to DOC A 2 nd admission to DOC B 3 rd admission to DOC C 4 th admission to DOC D 5 th admission to DOC E 6 th admission to DOC F 7 th admission to DOC G 8 th admission to DOC P Has previous parole record in Alabama Q Unknown R Restricted from work release programs S Sex offender X Life without parole sentence Y Barred From Parole A2 Sex M Male F Female A3 Race A Asian B Black I Indian U Unknown W White 63 A4 County Where Crime was Committed (commit county) Autauga Autauga Baldwin Baldwin Barbour Barbour Bessemer Jefferson Bibb Bibb Blount Blount Bullock Bullock Butler Butler Calhoun Calhoun Chambers Chambers Cherokee Cherokee Chilton Chilton Choctaw Choctaw Clarke Clarke Clay Clay Cleburne Cleburne Coffee Coffee Colbert Colbert Conecuh Conecuh Coosa Coosa Covington Covington Crenshaw Crenshaw Cullman Cullman Dale Dale Dallas Dallas Dekalb Dekalb Elmore Elmore Escambia Escambia Etowah Etowah Fayette Fayette Franklin Franklin Geneva Geneva Greene Greene Hale Hale Henry Henry Houston Houston Jackson Jackson Jefferson Jefferson Lamar Lamar Lauderdale Lauderdale Lawrence Lawrence Lee Lee Limestone Limestone Lowndes Lowndes Macon Macon Madison Madison Marengo Marengo Marion Marion Marshall Marshall Mobile Mobile Monroe Monroe Montgomery Montgomery Morgan Morgan Perry Perry Pickens Pickens Pike Pike Randolph Randolph Russell Russell Shelby Shelby St. Clair St. Clair Sumter Sumter Talladega Talladega Tallapoosa Tallapoosa Tuscaloosa Tuscaloosa Walker Walker Washington Washington Wilcox Wilcox Winston Winston A5 Offense (Offense for which offender is currently incarcerated) ?OFFENSES-OTHER (Class ??A?? FELONY)? ?OFFENSES PROPERTY-THEFT (??C?? FELONY)? ?OFFENSES-DRUGS (CLASS ??A?? FELONY)? ?OFFENSES-DRUGS (CLASS ??B?? FELONY)? ?OFFENSES-DRUGS (CLASS ??C?? FELONY)? ?OFFENSES-PBLC ORDER/SAFETY (CLASS ??C?? FELONY)? ?OFFENSES-PERSON (CLASS ??A?? FELONY)? ?OFFENSES-PERSON (CLASS ??B?? FELONY)? ?OFFENSES-PERSON (CLASS ??C?? FELONY)? AGGRAVATED STALKING AGGRAVATED MURDER ALTERING/POSS ALTERED ID MARK PISTOL ARSON I ARSON II ASSAULT I 64 ASSAULT II ASSAULT/INTENT RAVISH ATTEMPT TO COMMIT CONT SUB/INVOLVE A FELONY ATTEMPT TO COMMIT CONT SUB/INVOLVE B FELONY ATTEMPT ASSAULT I ATTEMPTED ARSON ATTEMPTED ARSON II ATTEMPTED BURGLARY I ATTEMPTED BURGLARY II ATTEMPTED KIDNAPPING I ATTEMPTED KIDNAPPING II ATTEMPTED MURDER ATTEMPTED RAPE I ATTEMPTED RAPE II ATTEMPTED ROBBERY I ATTEMPTED SODOMY I ATTEMPTED THEFT OF PROPERTY I BAIL JUMPING I BIGAMY BIGAMY (OLD CODE/PRIOR JAN 1980) BRCSP BRIBING A WITNESS BRING INTO ST STLN PROP/OBT FP II BRING STOLEN PROP INTO STATE I BRING STOLEN PROP INTO STATE II BURGLARY I BURGLARY I (OLD CODE/PRIOR JAN 80) BURGLARY II BURGLARY II (OLD CODE/PRIOR JAN 80) BURGLARY III BURGLARY OF A MOTOR VEHICLE CARNAL KNOWLEDGE/UNDER 12 CHILD ABUSE CONSPIRACY TO COMMIT MURDER CONSPIRACY TO COMMIT ROBBERY I CONT SUB CRIM SOLICT/INVOLVE A FEL CRIM CONS TO COMT CONT SUB/INV A FEL CRIM CONS TO COMT CONT SUB/INV B FEL CRIM NEG HOMCDE A 09/30/88 CRIM POSS FORGED INSTR I CRIM POSS FORGED INSTR II CRIMINAL MISCHIEF I CRIMINAL TAMPERING I CRUELTY TO ANIMALS DISCHARGE GUN OCC BLDG/VEH DISCHARGE GUN UNOCC BLDG DIST CONTROL SUBSTANCE DOMESTIC VIOLENCE I DOMESTIC VIOLENCE II ENTICING CHILD ENTER/IMMORAL PURPOSE ESC/ATT TO ESCAPE FROM PEN OR GUARD ESCAPE I ESCAPE II ESCAPE III EXTORTION II FAIL ATTACH STAMP ON MARIJUANA FELONY DUI FORGERY I FORGERY II FORGERY II (OLD CODE/PRIOR JAN 1980) FRAUD USE/REPRES CREDIT CARD GRAND LARCENY HINDERING PROSECUTION I ILL POSS/FRAUD USE-CREDIT CARD IMPERSONATING PEACE OFFICER INCEST INTENT TO SET FIRE TO FOREST LANDS INTEFERENCE WITH CUSTODY A/07/18/83 INTIMIDATING A WITNESS KIDNAPPING I KIDNAPPING II LEAVING SCENE ACCIDENT W/INJURY MANSLAUGHTER MANSLAUGHTER I MANU CTRL SUBSTANCE MURDER MURDER I MURDER II OBSTRCT OF JUSTICE-FALSE IDENTITY OBSTRUCTION OF JUSTICE < 2001-312 OBTAIN CTRL SUBS FRAUDULENTLY OFFENSES-OTHER (CLASS B FEL) OFFENSES PROPERTY-DAMAGE/INTRUSION OFFENSES-OTHER CLASS C FLNY PARENT/GUARDIAN PERMIT CHILD PROD OBSCEN PERJURY I POSS CHEMICALS/INTENT TO MANUFACTURE POSS CONTROL SUBSTANCE POSS MARIJUANA I POSS OBSCNTY OF PRSN UNDER 17 POSS OF BURGLAR?S TOOLS POSS PISTOL AFTER CONVICT VIOLENCE POSS PRECURSOR CHEM/NO LICENSE POSS/SELL SHRT BRRL RIFLE/SHTGN POSSESSION OF ANHYDROUS AMMONIA PROD OBSCENE MATTER UNDER 17 YEARS PROM PRISON CONTRABAND II PROMOTING PRISON CONTRANBAND I RAPE RAPE I RAPE II RECV STOLEN PROPERTY I RECV STOLEN PROPERTY II RENDERING A FALSE ALARM ROBBERY ROBBERY I ROBBERY II ROBBERY III SALE CONT SUB > 18 YR TO < 18 YR SEXUAL ABUSE I SEXUAL TORTURE SODOMY I SODOMY II SOLICITING A CHILD BY COMPUTER STALKING TERRORIST THREATS THEFT BY DECEPTION I THEFT BY DECEPTION II THEFT OF INDENTITY I THEFT OF LOST PROPERTY I THEFT OF LOST PROPERTY II THEFT OF PROPERTY FRAUD LEASE/RENTAL THEFT OF PROPERTY I THEFT OF PROPERTY II THEFT OF SERVICES II THROW/SHOOT INTO OCCUPIED VEHICLE TRAF AMPHETAMINES 1-10 KG TRAF AMPHETAMINES 28-500 GM TRAF AMPHETAMINES 500G-1 KG TRAF CANNABIS 100-500 LBS TRAF CANNABIS 2.2-100 LBS TRAF CANNABIS 500-1000 LBS TRAF CANNABIS < 2000 LBS TRAF COCAINE 1-10 KG 65 TRAF COCAINE 28-500 GRAM TRAF COCAINE 500G ? 1 KG TRAF COCAINE < 200 GRMS TRAF COCAINE < 400 GRMS TRAF COCAINE > 400 GRMS TRAF ENTERPRISE 1 ST CONVICT TRAF METHAMPHETAMINES TRAF OPIUM/MORPH/HEROIN 4-14 GMS TRAF OPIUM/MORPH/HEROIN 14-28 GM TRAF OPIUM/MORPH/HEROIN 4-14 GM TRAF OPIUM/MORPH/HEROIN < 14 GRAMS TRAF OPIUM/MORPH/HEROIN > 56 GM TRAFFIC STOLEN IDENTITIES UNAUTH USE VEH/FORCE UNLAW BREAKING AND ENTERING VEHICLE VEHICULAR HOMICIDE VIO OF SECURITIES ACT OF ALABAMA VIO OF STATE ETHICS LAW VIOLATION OF SEX OFF REGIS LAW WILFULLY CORRUPT AND FALSE SWEARING YOUTHFUL OFFENDER ACT A6 Education Level NONE no education completed 1 st grade completed 1 st grade 2 nd grade completed 2 nd grade 3 rd grade completed 3 rd grade 4 th grade completed 4 th grade 5 th grade completed 5 th grade 6 th grade completed 6 th grade 7 th grade completed 7 th grade 8 th grade completed 8 th grade 9 th grade completed 9 th grade 10 th grade completed 10 th grade 11 th grade completed 11 th grade 12 th grade, GED or HED high school diploma, GED or high school equivalency SOME COLLEGE taken some college courses SOME COLLEGE grad college graduate MA completed masters degree PHD completed doctorate degree UNKNOWN level of education not specified A7 Alabama Sentencing Commission Violent Definition 0 not considered a violent offense 1 considered a violent offense The definition of what is or is not a violent offense for this analysis will be delineated by the Alabama Sentencing Commission definition of violent offenses contained in the Alabama Sentencing Reform Act of 2003. The following offenses are considered by the Alabama Sentencing Commission to be violent and: 1. Capital murder 2. Murder 3. Manslaughter 66 4. Criminally negligent homicide 5. Assault I 6. Assault II 7. Compelling street gang membership 8. Kidnapping I 9. Kidnapping II 10. Rape I 11. Rape II 12. Sodomy I 13. Sodomy II 14. Sexual torture 15. Sexual Abuse I 16. Enticing a child to enter a vehicle for immoral purposes 17. Stalking 18. Aggravated Stalking 19. Soliciting a child by computer 20. Domestic violence I 21. Domestic violence II 22. Burglary I (unless the offender enters the dwelling without a weapon or other dangerous instrument and does not use or threaten to use a weapon or dangerous instrument against another person during the commission of the offense). 23. Burglary II 24. Burglary III (if the intent is to commit a violent offense). 25. Arson I 26. Criminal possession of explosives 27. Extortion I 28. Robbery I 29. Robbery II 30. Robbery III 31. Pharmacy robbery 32. Terrorist threats 33. Escape I 34. Promoting prison contraband I 35. Intimidating a witness 36. Intimidating a juror 37. Treason 38. Discharging a weapon into an occupied building, dwelling, automobile, etc. 39. Promoting prostitution I 40. Production of obscene matter involving a minor 41. Trafficking 42. Child abuse 43. Elder abuse 44. Terrorism 45. Hindering prosecution for terrorism 67 46. Any substantially similar offense for which an Alabama offender has been convicted under prior Alabama law or the law of any other state, the District of Columbia, the United States, or any of the territories of the United States. The basis for defining these offenses as violent is that each offense meets at least one of the following criteria: 1. Has an element, the use, attempted use, or threatened use of a deadly weapon or dangerous instrument or physical force against the person of another. 2. Involves a substantial risk of physical injury against the person of another. 3. Is a nonconsensual sex offense. 4. Is particularly reprehensible. Any attempt, conspiracy, or solicitation to commit a violent offense shall be considered a violent offense. A8 Department of Corrections Violent Definition 0 not considered a violent offense 1 considered a violent offense The crimes considered to be violent by the Alabama Department of Corrections are outlined in the following list: 1. Aggravated Murder 2. Aggravated Stalking 3. Arson I 4. Assault I 5. Assault II 6. Assault Officer-Deadly Instrument 7. Assault/Intent to Murder 8. Assault/Intent Ravish 9. Assault/Intent Rob 10. Attempt to Commit Assault I 11. Attempted Burglary I 12. Attempted Kidnapping I 13. Attempted Kidnapping II 14. Attempted Murder 15. Attempted Rape I 16. Attempted Rape II 17. Burglary I 18. Burglary I (old code/prior to January 1980) 19. Capital Murder 20. Carnal Knowledge/Over 12-Under 16 68 21. Carnal Knowledge/Under 12 22. Child Abuse 23. Conspiracy to Commit Murder 24. Conspiracy to Commit Robbery 25. Crime Against Nature/Sodomy/Beast 26. Criminal Possession of Explosives 27. Criminally Negligent Homicide 28. Discharge Gun into Occupied Building/Vehicle 29. Discharge Gun into Unoccupied Building 30. Enticing Child to Enter for Immoral Purpose 31. Escape or Attempt to Escape from Penitentiary of Guard 32. Escape I 33. Incest 34. Intimidating a Witness 35. Kidnapping-Generally 36. Kidnapping I 37. Kidnapping II 38. Manslaughter 39. Manslaughter I 40. Murder 41. Offenses-Other (Class ?A? Felony) 42. Offenses-Person (Class ?A? Felony) 43. Parent Permit Child-Obscene Matter 44. Permitting or Facilitating Escape I 45. Possession Obscenity of Person Under 17 46. Possession Pistol after Conviction of Violence 47. Producing Obscene Matter Under 17 48. Promoting Prison Contraband I 49. Rape 50. Rape I 51. Rape II 52. Robbery 53. Robbery I 54. Robbery II 55. Robbery III 56. Sexual Abuse I 57. Sodomy I 58. Sodomy II 59. Soliciting a Child by Computer 60. Stalking 61. Throw/Shoot into Occupied Vehicle 62. Unauthorized Use of Vehicle by Force 63. Violation of Sex Registration Law 69 A9 Education Recode 1 no completed education 2 1 st grade-11 th grade completion 3 high school diploma, GED or high school equivalency 4 at least some college A10 Alabama Institutional Serial Recode 1st 1 st admission to DOC 2nd 2 nd admission to DOC 3rd 3 rd admission to DOC 4th 4 th admission to DOC 5th 5 th admission to DOC 6th 6 th admission to DOC 7th 7 th admission to DOC 8th 8 th admission to DOC lwop life without parole sentence no parole barred from parole restricted restricted from work release programs sex sex offender (current or previously) unknown unknown A11 Marital Recode common law common law marriage divorced divorced married married, those divorce and remarried and those widowed and remarried other/unknown other/unknown (specified other in file) separated separated (married) single single (never married) widow(er) widow(er) A12 Sentence Length The sentence imposed to the offender in years. 70 A13 Unemployment Rate These figures were gathered from the Bureau of Labor Statistics homepage (www.bls.gov). ?State and Local Unemployment Rates? was selected under the heading of ?Employment and Unemployment?. At the top of the menu bar of the page you are directed to, select ?Get Detailed Statistics?. In the ?Employment and Unemployment? section of choices, ?Local Area Unemployment Statistics? is the location of the unemployment rates. ?Create Customized Tables (Multiple Screens)? should be chosen and then Alabama was chosen as the census region. Areatype F, ?Counties and equivalents?, reveals all counties in Alabama as options for analysis and all counties were selected for this analysis. Unemployment rate is then chosen as the variable to retrieve data for and the unemployment rates were not seasonally adjusted. These results were then transferred to BBEdit and arranged by alphabetizing the corresponding county and then joined into the JMP dataset containing the other variables.