Methodology

Methodology

Methodology

This secondary data analysis study will attempt to answer the following research question: For online sexual offenders who have committed crimes against minors (people under 18 years of age) in the United States, do variables on personal history and variables detailing the types of offense committed predict whether or not the offender is a repeat offender? This study will control for the effect of certain demographic variables such as gender, race and offender age at the time of the crime. The researcher hypothesizes that certain personal history characteristics and particular types of offenses will emerge as predictive factors of offender recidivism.

This research involves the use of secondary data, The Second National Juvenile Online Victimization Incidence Study (NJOV-2), collected by Cornell University for publishing via the National Data Archive of Child Abuse and Neglect. The data collected from the NJOV-2 is taken from a representative national survey taken from local, country, state and federal law enforcement agencies quantifying the frequency of online sexual offenses against minors, the characteristics of the offenses, and the demographic information of the offender between 2000 and 2006. The National Data Archive of Child Abuse and Neglect provided de-identified data to the researcher for data analysis. The present was approved by the Institutional Review Board (IRB) at The Catholic University of America, the university with which the researcher is affiliated.

The NJOV-2 data were collected from law enforcement agencies via mail survey and telephone follow-up. From the sample of 2,500 law enforcement agencies that were selected at random for participation in the study, there were 1,051 sexual offense cases reported and ultimately used in the data findings. Although the data are not provided, it is possible that multiple cases were committed by the same offender. There were three inclusion criteria for the NJOV-2: 1) was the offense internet-related, 2) Was the victim under 18 (or legally a “minor”) and 3) Did the arrest happen between July 1, 2000 and June 1, 2001 or during the calendar year of 2006. The inclusion criteria were chosen to fit the population and crimes of interest and the time frame of data collection. If inclusion criteria were met, the case was included for analysis and follow-up telephone interviews were conducted to garner additional detail about the crime.

All participation was voluntary and participating law enforcement agencies were assured that they could cease participation at any time. Identifying information was omitted from the data prior to its release to the researcher. Furthermore, both waves of the study passed review from the New Hampshire University Institutional Review Board and complied with the U.S. Department of Justice’s ethical regulations. Lastly, all research participants were offered the opportunity to be sent the results of the survey once data collection terminated.

In the present study, the independent variables are personal history variables and nature of the crime variables. The dependent variable is offender recidivism. The researcher will control for the effects of race, age, and gender. Personal History variables were extricated from the available NJOV-2 data examining their respective characteristics ancillary to their sexual offense. These data provided contextual non-demographic characteristics of who the offenders were. In other words, personal history variables are an attempt at understanding who commits online sexual crimes against minors in a manner that is more specific than general demographics. For the purposes of the present study, personal history variables include: 1) Did the offender have trouble with drugs and alcohol at the time of the crime? 2) Did the offender have a diagnosed mental illness at the time of the crime? 3)Were there prior arrests for non-sexual offenses, and 4) Did the offender have an illegal job at the time of the crime? These personal history variables were chosen predominantly in response to the gaps in the literature that suggest that personal history and personality may have predictive power over chance of recidivism. They were also chosen due to their availability within the secondary data set. Each item was included as a separate variable with a “yes” or “no” response (0=No, 1=Yes). All other responses such as “not ascertainable,” “do not know,” or “refused” were recoded as missing responses because they did not clearly answer the question posed.

Nature of the crime variables specify the act or infraction that led to the offender’s arrest for internet-related sexual exploitation chargers against minors. The nature of the crime variables was selected from the available data in response to existing empirical findings on sexual offense recidivism. Although findings demonstrate that highly stigmatized crimes are associated with recidivism, the research does not adequately address which crimes, specifically, predict a higher likelihood of offense. In other words, the researcher seeks to understand if certain internet related sexual crimes carry more strain and predictive power over further sexual offenses. Each nature of the crime item was entered as a separate variable and was recoded to dichotomously reflect yes/no responses (0=No, 1=Yes).

In the present study, the dependent variable is offender recidivism. This variable may be defined as whether or not the offender was already a registered offender at the time of the crime. In other words, if the offender had already been convicted of a previous sex crime. The dependent variable was recoded in the present study to become a dichotomous variable that demonstrates a yes/no response (0=No, 1=Yes).

The control variables are demographic characteristics that were collected during the national survey are age at the time of crime commission, sex/gender, and race. The intention of incorporating these control variables is noise-reduction. Controlling for variables increases the probability that dependent variable is varying in accordance with the independent variable without interference from other salient variables. ³³ The variable of offender sex was collected in two categories, males (coded as 1) and females (coded as 0). Age is measured at a continuous level and is normally distributed (mean = 36.61, median = 35, St. Deviation = 13.133, Skewness = .455). Race is a categorical variable that initially contained six racial distinctions. For the purposes of this logistic regression analysis, the variable has been collapsed into two categories to create a more even distribution of cases: white (coded as 1) and non-white (coded as 0). Statistics for the new categorical variable that describes offender race show that 92 percent of offenders are white (n = 972) and 7.1 percent of offenders are non-white (n = 74).

For data analysis, the researcher conducted a hierarchal logistic regression analysis to examine whether variables on personal history and types of offense predict whether or not the offender is a repeat offender, while controlling for the effect of certain demographic variables such as gender, race and offender age. Descriptive analyses were first conducted to show the sample characteristics and goodness of fit between the data and hierarchal logistic regression. Because the dependent variable is dichotomous and yields results that attest to the predictive power of the independent variables, a logistic regression method was indicated. The researcher conducted a three-block hierarchal logistic regression in which control variables were entered into Block 1, nature of the crime variables were entered into Block 2, and the personal history characteristics were entered in Block 3. This was done so that the researcher could better analyze the effect of the independent variables, both as groups and as individual variables, without the noise contributed by the inclusion of the control variables. Ultimately, the results will suggest whether or not the personal history variables and the nature of the crime variables will significantly predict the offender’s likelihood that they have committed previous sexual offenses.

Methodology

Table of Contents