How to Cautiously Uncover the “Black Box” of Machine Learning Models for Legislative Scholars

Machine learning models, especially ensemble and tree-based approaches, offer great promise to legislative scholars. However, they are heavily underutilized outside of narrow applications to text and networks. We believe this is because they are difficult to interpret: while the models are extremely flexible, they have been criticized as “black box” techniques due to their difficulty in visualizing the effect of predictors on the outcome of interest. In order to make these models more useful for legislative scholars, we introduce a framework integrating machine learning models with traditional parametric approaches. We then review three interpretative plotting strategies that scholars can use to bring a substantive interpretation to their machine learning models. For each, we explain the plotting strategy, when to use it, and how to interpret it. We then put these plots in action by revisiting two recent articles from Legislative Studies Quarterly.     

When Heads Roll: Assessing the Effectiveness of Leadership Decapitation

Leadership targeting has become a key feature of current counterterrorism policies. Both academics and policy makers have argued that the removal of leaders is an effective strategy in combating terrorism. However, leadership decapitation is not always successful, and existing empirical work is insufficient to account for this variability. As a result, this project answers three primary questions: (1) Under what conditions does leadership decapitation result in the dissolution of a terrorist organization?; (2) Does leadership decapitation increase the likelihood of organizational collapse beyond the baseline rate of collapse for groups over time?; and (3) In cases where decapitation does not result in group collapse, to what extent does it result in organizational degradation and hinder a group's ability to carry about terrorist attacks? I develop a dataset of 298 incidents of leadership targeting from 1945–2004 in order to determine whether and when decapitation is effective. First, I identify the conditions under which decapitation has been successful in bringing about organizational decline. The data show that a group's age, size, and type are critical in identifying when decapitation will cause the cessation of terrorist activity. As an organization grows in size and age, it is much more likely to withstand the removal of its leadership. Organizational type is also significant in understanding the susceptibility of an organization to decapitation. Ideological organizations are most likely to experience a cessation of activity following the removal of leader, while religious organizations are highly resistant to leadership decapitation. Second, I determine whether decapitation is an effective counterterrorism strategy that results in organizational collapse. The data show that decapitation does not increase the likelihood of organizational collapse beyond a baseline rate of collapse for groups over time. Organizations that have not had their leaders removed are more likely to fall apart than those that have undergone a loss of leadership. The marginal utility of decapitation is negative for many groups, particularly for larger, older, religious, and separatist organizations. Finally, I look at the extent to which decapitation results in organizational degradation and hinders a group's ability to carry about terrorist attacks. Case studies illustrate whether decapitation has an effect on the operational capacity of an organization by identifying whether the removal of key leaders changes the number and lethality of attacks. If certain organizations are more resilient than others, it is important to know when decapitation should be effective and when it could lead to counterproductive outcomes. Overall, these findings illustrate the need to develop a new model for evaluating the efficacy of leadership decapitation and for developing effective counterterrorism policies.     

Introduction

SUMMARY

Previously homeless women (17) and men (12) were surveyed on their experience as a caregiver to physically challenged senior citizen clients (16 men, 13 women: age M= 64.7, SD = 16.8). Caregivers had training in service delivery and experience before the survey. More satisfaction than stress from providing care was reported by caregivers. Caregiving satisfaction was significantly positively related to global life satisfaction and a sense of personal vulnerability to being disabled. Caregiving stress was significantly negatively related to perceived knowledge of the disability. Overall, these previously homeless caregivers reported that providing care to the disabled was positive, and would be repeated and recommended to Others.     

Governmental Efforts on Homeland Security and Crime: Public Views and Opinions

This paper examines views of the respondents regarding homeland security and traditional crime in the United States. Using questions from the 2007 Penn State Poll, a sample of 862 Pennsylvanians participated through a telephone interview. Participants were questioned about their concerns regarding the effectiveness of homeland security, their fear of crime (white-collar, property, violent and terrorist attacks). The results revealed that citizens were satisfied with the effectiveness of homeland security since the September 11, 2001, attacks. The results indicate that fear of crime is different for demographics, and we were able to show that those that thought homeland security had been effective increased the likelihood of fear of white-collar crime. We were also able to show demographic differences for national spending on crime. In addition, we were able to show that those who believed that homeland security was effective did not believe that national spending was at the proper level for property, violent, or white-collar crime. The implications of these results are discussed.     

The Mask of Neutrality: Judicial Partisan Calculation and Legislative Redistricting

Do judges ruling on redistricting litigation increase electoral competition in congressional races while simultaneously drawing districts favoring their party's congressional candidates? I offer a novel theory of judicial partisan calculation, arguing that judges draw more competitive districts than legislatures or commissions, but that judge‐drawn districts favor the electoral interests of their copartisans. These claims are reconcilable because judges target districts held by contrapartisan legislators to maximize their copartisans’ fortunes. I find that Democratic judges draw competitive districts by adding Democratic voters to Republican‐held House constituencies. Court‐administered redistricting increases competitiveness, ostensibly due to judicial neutrality. This mask of neutrality, however, conceals sophisticated partisan calculation.     

Challenges to inmate classification in a developing country setting: implications for context-based inmate classification schemes

Inmate classification is a key aspect of prison management and inmate programming. It guides inmate housing assignment, level of supervision and control, and inmate programme placement. The Philippines formally incorporated these principles to modernise its prisons. However, structural, organisational, and cultural challenges posed barriers to successful implementation. Utilising qualitative data gathered through interviews and focus group discussions with prison personnel, inmates, and volunteers in a major prison in the Philippines, this paper investigates the practices that impact on inmate classification and how these affect housing assignment, levels of supervision and control, and inmate programming. Implications towards developing an integrated correctional management in the Philippines, and other similarly situated countries, are discussed.     

Queer Contestations and the Future of a Critical “Queer” Criminology

It is intuitive to include critical criminologists in early conversations about “queering” criminology given that the paradigms and methods of critical criminology can be employed to challenge subordination and inequality in its several dimensions. The first part (and main focus) of this article problematizes this intuition, which is easy to accept at face value, by reflecting backwards and explaining how early influential critical criminological views perpetuated damaging stereotypes and representations of lesbian, gay, bisexual, transgender, and queer (LGBTQ) people as sexual deviants. The second part reflects forward, and discusses the difficulties of carving a space in the discipline for critical queer perspectives. Drawing on critical and critical race theories, this article advocates a relational, intersectional approach to conceptualize sexual orientation and gender identity in criminological theory and research. This approach considers the connections among sexual orientation or gender identity and other differences (e.g., race, ethnicity, nationality, religion, class, gender, etc.) without assuming or attaching fixed meanings to those differences.