A study of serious adolescent offenders as they transition to adulthood and out of crime

Dimensions of Organizational Functioning - Subject Release

This measure appears in the following time-points: Release.

Related Construct

• Institutional Environment

Description of Measure

Unlike other sections of the codebook which describe a specific measures, this section of the codebook describes a set of computed constructs ("dimensions") that we have generated to summarize the institutional experience along key constructs. A subset of release interview items were used to generate ratings about eight dimensions of organizational functioning. Each of the dimensions reflects general attributes of the setting that are likely to affect the later adjustment of adolescents who spent time in the institution. The eight dimensions are akin to latent constructs that are comprised of two to three components each. The components (a total of 20 across all eight dimensions) are scores from scales incorporated in the release interview or derived during factor analysis. Details regarding the definition of and methods used to develop the dimensions can be found in Mulvey, Schubert & Odgers, 2010. We provide a summary of the dimensions below, however, other sections of the release interview codebook should be consulted for more in-depth information regarding the measures from which the dimension are derived.

There were several steps involved in the development of the eight dimensions of organizational functioning. First, items from scales included in the release interview were grouped under the most relevant of the eight dimensions (safety, institutional order, harshness, caring adults, fairness, antisocial peers, services, and reentry planning) based on face validity. Next, we derived a score or set of scores (components) to reflect each dimension using one of the following methods: (a) summing the number of endorsed items (e.g., to estimate the number of services received or vocational training opportunities within an institution), (b) calculating scores for measures with pre-established scales to correspond with previously validated scoring protocols (e.g., exposure to violence), and (c) creating new scales based on item sets selected to tap into relevant constructs and validating these scales through the application of exploratory and confirmatory factor analytic methods. When this last method was used, we assessed the internal consistency, parameter estimates, overall model fit, and convergent validity of newly created subscales for each dimension. We also tested whether each of the eight dimensions was best represented by a single factor (unidimensional) or multiple components (multidimensional). All structural equation models were evaluated using recommended fit indices, including the root mean square error of approximation (RMSEA), where values less than .08 indicate acceptable fit and values less than .05 indicate good fit, and the confirmatory fit index, where estimates greater than .90 indicate acceptable fit and values greater than .95 indicate good fit (Hu & Bentler, 1999; McDonald & Ho, 2002).

The dimensions were developed using information from 668 release interviews. No study participant contributed more than one interview, thus eliminating the possibility of bias produced by multiple ratings from the same individual. In situations where one study participant provided more than one release interview, only the first interview ratings obtained during the follow-up period were used in the development of the dimensions. The reports came roughly equivalently from the Philadelphia (n = 325) and Phoenix (n = 343) sites and the ratings represented 128 unique institutions.

The inter-correlation coefficients and multilevel reliability coefficients for subscales across the 128 institutions are as follows:

The variables that comprise the dimensions components noted above can be found in various release codebook measures (exclusive of three exceptions, which are noted below).

• Safety
• Fear within the institution (r0fear; available in this section, explained below).
• Exposure to violence (r0vicvtot, Total violence - witnessed; r0victot, Total violence - victim). Found in the "Exposure to Violence" codebook section.
• Institutional order
• Overall organization (r0cl_orgn2; available in this section, explained below).
• Staff connectedness (r0cl_stcon; available in this section, explained below).
• Staff negative behavior (r0_stng; mean of four items. A higher score indicates more negative behavior from staff). Found in the "Sanctioning Practices and Institutional Climate" measure.
• Harshness
• Sanctions (r0sanc_count; count of 10 or 12 items, depending on the version in which the interview was conducted). Found in the "Sanctioning Practices and Institutional Climate" measure.
• Restrictions (r0restrict_f; count of four items. Available only for release interviews NOT conducted on community-based programs). Found in the "Sanctioning Practices and Institutional Climate" measure.
• Punishment costs (r0varcst; count of 18 items). Found in the "Personal and Social Costs and Rewards" measure.
• Caring Adult
• Domains of support (r0cadpre; count across eight domains). Found in the "Caring Adults" measure.
• Diversity of support (r0caddiv; count of unique adults nominated across eight domains). Found in the "Caring Adults" measure.
• Fairness
• Lack of bias (r0bias; mean of three items). Found in the "Procedural Justice" measure.
• Overall fairness (r0fr_gen; mean of three items. A higher score indicates more fairness). Found in the "Sanctioning Practices and Institutional Climate" measure.
• Antisocial Peers
• Peer delinquency - antisocial behavior (r0prbehv; mean of 12 items. Data must be in nine of the 12 items in order to compute a valid mean). Found in the "Peer Delinquency" measure.
• Peer delinquency - antisocial influence (r0prinfl; mean of seven items. Data must be in five of the seven items to receive a score). Found in the "Peer Delinquency" measure.
• Gangs - present/not present (r0gng01; binary marker for the presence of gangs at the release facility). Found in the "Gang Involvement" measure.
• Services
• Mental health services (r0sertot; count across eight services received in the facility). Found in the "Services and Medication" measure.
• Vocational services (r0njbtra; count across eight types of vocational training at the facility). Found in the "Services and Medication" measure.
• Having a primary caregiver (r0ser01; binary marker for having a primary caregiver at the facility). Found in the "Services and Medication" measure.
• Reentry Planning
• Future orientation of the program (r0planning; mean of six items). Found in the "Perceptions of Chances for Success" measure.
• Having a release counselor (r0ser06; binary marker for having a person assigned to help with making arrangements for return to the community). Found in the "Services and Medication" measure.

Three of the dimension components are available in this section because they are calculated in a different manner than the original version that was described in the "Sanctioning Practices and Institutional Climate" measure.

These items assess the participant's perceived safety while at the institution (e.g., "I felt safe when a staff member was present", "I always had to watch my back around other residents"), staff connectedness (e.g., "Staff did a lot of individual counseling", "If a resident did well, the staff would tell him/her so personally"), and overall organization (e.g., "The facility was always neat and clean", "We followed a regular schedule every day"). Participants respond on a 5-point Likert scale ranging from "Strongly disagree" to "Strongly agree".

The following components are available in this section.

• Safety - fear within the institution [r0fear]; mean of three items, two of which are reverse-coded. A higher score indicates a feeling of being more unsafe. In contrast, the original variable (safety within the institution - original; r0psafet_org) is a score based on the mean of six items which are more inclusive than those used to compute perceived fear.
• Institutional order - overall organization [r0cl_orgn2]; mean of three items. A higher score indicates more organization. In contrast, the original variable (organization of the facility - original; r0clim1_org) is a mean of eight items, six of which must contain valid data to receive a score.
• Institutional order - staff connectedness [r0cl_stcon]; mean of ten items. A higher score indicates more organization. In contrast, the original variable (staff connectedness - original; r0clim2_org) uses 11 items.

Update: During the data cleaning phase, 11 of the 668 cases used to develop the dimensions were removed from the dataset because the interview was conducted outside of the window of opportunity allowed for data collection (either more than 60 days after their release from the institution, or after the end of the follow84 timepoint interview) or because more than one interview was mistakenly conducted on the same institution stay. We do not believe that this issue detracts from the validity of the dimensions since it rooted in the inability to match the time period covered by the release interview to the calendar data, not to a problem with the ratings provided by the study participant.

References

• Mulvey, E.P., Schubert, C.A., Odgers, C.A. (2010). A Method of Measuring Organizational Functioning in Juvenile Justice Facilities Using Residential Ratings. Criminal Justice and Behavior, 37, 11, 1255-1277.
• Schubert, C.A., Mulvey, E.P., Loughran, T., Losoya, S. (2012). Perceptions of Institutional Experience and Community Outcomes for Serious Adolescent Offenders. Criminal Justice and Behavior, 39, 71-93.

This Web site is maintained by the Center for Research on Health Care (CRHC) Data Center. Contact the webmaster at dcweb@pitt.edu. All rights reserved.