Positive Administrative Control: Using Social Exchange to Assess Managerial Impacts on Inmate Misconduct

Abstract

Positive administrative control uses social exchange theory to explain how management impacts institutional rates of inmate disorder. This study proffers that prison managers can use their relationship with staff to influence them to act in accordance with agency desires, which in turn affects institutional disorder rates. Using data from the Federal Bureau of Prisons, the model demonstrates an indirect connection exists between management/staff relationships and disorder. However, two of the scales used to explain positive administrative controls’ sway—the leader-member exchange (LMX) and perceived organizational support (POS)-had opposing effects. Increases in the quality of the leader-member exchange (LMX) was associated with decreases in misconduct rates, while increases in perceived organizational support (POS) was associated with increases in misconduct rates.

Keywords:

• administrative control
• social exchange theory
• relational leadership
• prison management
• institutional disorder

Notes

¹ Recent scholarship, such as Skarbek (2014), are important for the study of the sociological order in prison, but doesn’t represent an advance in disorder theory as much as it is a continuance of the proposition that prison staff have limited influence on the social order and do not significantly impact disorder rates.

² DiIulio (1987) examined three different organizational processes: Texas’ control model, Michigan’s responsibility model, and California’s mixed model.

³ Again, the conclusions of the SSRC must be considered in light of the fact that it was not until the early 1970s that multiple site studies were being conducted, and most of those used proxies for management, such as security level and institution size. They assumed differences in inmate experiences were due to managerial style.

⁴ The Sykesian model is a term used to describe Gresham Sykes (1958) assertion that his analysis “has the initial, presumptive advantage of analyzing the disturbances in the prison not as isolated, fortuitous events but as an integral part of the nature of confinement” (1958, p. 129), but Dilulio (1991) points out there is no evidence that there was any tightening of regulations leading up to the two riots Sykes (1958) studied.

⁵ Useem and Kimball (1989) analyzed five riots: Attica, 1971; Joliet, 1975; Santa Fe, 1980; three in Michigan in 1981; and Moundsville, WV, 1986, to show that breakdowns in administrative control preceded significant disorder.

⁶ Interestingly, few if any studies contradicted Useem and Kimball’s (1989) assertions, other than to use differing administrative breakdown variables to demonstrate that coercive measures are antithetical to controlling inmate behavior—see McCorkle et al. (1995) for an example of this type of analysis.

⁷ See earlier discussions by Cressey (1959), McClerry, (1961), Useem and Kimball (1989), and Bottoms (1999) for further explanations of the importance of equity for garnering a sense of legitimacy.

⁸ This is true for staff–inmate relationships and for management–staff relationships.

⁹ Transformational leadership is a term for a set of leader behaviors that help create an environment conducive to high-quality leader-member exchange relationships (Dulebohn, Bommer, Liden, Brouer, & Ferris, 2012), which is rooted in social exchange theory (Blau, 1964; Homans, 1958). However, researchers have shown that effective leaders engage is both transformational and transactional leadership (Avolio, Bass, & Jung, 1999).

¹⁰ Also known as contingent reward behavior.

¹¹ Wayne et al. (1997) showed that POS and LMX have separate constructs that developed independently and have different antecedents, but together form an integrated model of social exchange.

¹² It should be noted that POS primarily impacts affective organizational commitment (Rhoades & Eisenberger, 2002), since it is based on an exchange relationship. Organizational commitment also has normative and continuance elements. Continuance commitment has been shown to be related to POS, (Eisenberger et al., 1986; Wayne et al., 1997), but normative commitment has differing antecedents and while not exclusive from the other components is not entirely affected by POS (Meyer & Allen, 1991).

¹³ 2008 was chosen for analysis because in 2009 the BOP initiated a Special Management program for highly disruptive inmates. This program removed inmates with severe and/or repetitive discipline, along with influential inmates in disruptive groups, from mainline institutions and placed them in a program institution designed to deter them from committing further misconduct while at the same time reducing misconduct at other institutions. This may have resulted in reduced variability in rates of misconduct between institutions.

¹⁴ There are important differences between male and female offenders that demonstrate specific skills are needed for female inmates: male offenders commit nearly twice as many violent offenses as female offenders, female offenders are more likely to be incarcerated for a drug offense, and female offenders are much more likely to have been the victim of physical, emotional, and/or sexual abuse (Bloom, Chesney-Lind, & Owen, 1994; Brennan & Austin, 1997; Cranford & Williams, 1998; Wright, Salisbury, & Van Voorhis, 2007). Positive administrative control can be used in this context, but the explanations were beyond the scope of this introductory study.

¹⁵ Administrative facilities are those with transient populations, such as federal jails and hospitals. They house inmates of all security levels in one building.

¹⁶ In 2008, the PSCS was handed out to 12,247 of 36,029 staff. 7,997 responded to the survey. (response rate: 65.3%). See the appendix for further information.

¹⁷ Questions in the PSCS that ask staff about the quality of the relationship with their supervisors do not differentiate between supervisors. Many staff report to more than one supervisor. Because of this, management is analyzed as one unit.

¹⁸ The Bureau of Prisons uses a misconduct incidence rate per 5,000 inmates, so this study continued the practice. There is no one reason the BOP chose a rate per 5,000 inmates, but it does allow for a more nuanced examination of differences between institutions and eliminates disparities due to institution size.

¹⁹ Subject to specific exceptions based on sentencing type—United States Department of Justice, Federal Bureau of Prisons, 2011.

²⁰ Although more recent scholarship supports the use of oblique rotations to remove the uncorrelated factor restriction imposed by orthogonal methods (Fabrigar, Wegener, MacCallum, & Strahan, 1999), in this study all the questions loaded on the same factors using both methods, save for two that were very close between factor 1 (LMX) and factor 2 (POS)—“My supervisor asks my opinion when a work-related problem arises,” and “my supervisor engages me in the planning process.” The two items in question loaded on the POS scale using oblique rotation, but loaded on the LMX scale using orthogonal rotation. However, these questions were more closely aligned with questions that loaded on an LMX factor in previous studies (Wayne et al., 1997), and since there was only a .05 difference between factors 1 and 2 for “my supervisor asks my opinion” and .09 for “my supervisor engages me in the planning process,” the orthogonal rotation results were used.

²¹ Conducting this type of exploratory factor analysis, where responses from a questionnaire are organized into scales that are then used to test hypotheses, is consistent with existing research (Rousseau & Tijoriwala, 1999). A factor analysis was chosen over a principle components analysis because the intent was to understand the scales’ latent structure, and common factor loadings are very accurate when the data fits the assumptions of factor analysis (Gorsuch, 1990; Widaman, 1993).

²² Even though the eigenvalue of organizational flexibility and communication was below 1.0 (violating Kaiser’s, 1958 rule), it was not viewed as being overly relevant for determining the most parsimonious number of factors because Kaiser’s rule is not consistently accurate (Gorsuch, 1997).

²³ Even so, Gaes and McGuire (1985) cautioned that their results do not challenge the putative relationship between age and criminal behavior.

²⁴ Other measures of violence were available, such as history of violence within the past 5 years, 10 years, etc., but they included violence committed while incarcerated, which would have put institutional violence on both sides of the regression.

²⁵ The Bureau of Prisons uses GED completion rates as a managerial benchmark.

²⁶ Inmates are assigned to a security threat group based on an assessment of their involvement in one or more of 91 different organizations deemed by the Bureau of Prisons to be an organized threat to the security and order of a prison. The process of determining what groups are considered security threat groups and the mechanism for determining if an inmate is a member, suspect, or associate of a security threat group is not public information.

²⁷ Two institutions did not report any inmate grievances, resulting in an N of 83 for that measure. Since all formal grievances are cataloged and tracked, it is likely these institutions did not have any grievances to report.

²⁸ Because the correlation between inmates incarcerated for a crime of violence was significant with every other variable in the study except for the positive administrative control measures and average sentence, the models were initially run without violent inmates, and again with them included.

²⁹ For reasons that will be explained later, robust error terms will be used in this study to control for heteroskedasticity (Wooldridge, 2009).

³⁰ Tthe results for the total misconduct full model were significant: (X ²(1) = 20.39; p > X ² = .000.

³¹ Bivariate correlations were not presented, but are available upon request.

³² Non-serious non-violent misconduct was not modeled for violent inmates due to its poor performance in the first set of regressions.

³³ The bivariate correlations were not presented, but the relationship between organizational communication/flexibility and average number of grievances was −.23, p < .05. The table is available upon request.

Additional information

Notes on contributors

Rodger C. Benefiel

Rodger C. Benefiel is an assistant professor of criminal justice at Bloomsburg University. He has over 24 years’ experience working in corrections. His research interests include criminal justice administration, staff/inmate relationships, recidivism, reintegration, restrictive housing, gangs, and labor/management relations.