Study
Slep and colleagues (2020) used a randomized experimental design (at the air base level) to compare the outcomes of individuals on Air Force bases which had been randomly assigned to deliver New Orientation for Reducing Threats to Health from Secretive-problems That Affect Readiness (NORTH STAR) versus individuals on Air Force bases that were randomly assigned to the control group.
To recruit participants, Air Force prevention teams met with behavioral points of contact from Air Force Major Commands. The Air Force points of contact then met with installation-level counterparts to determine interest. Installations from Major Commands were accepted until the study reached the required number of installations. Of the 79 Air Force installations with Community Action Teams (CATs), 24 volunteered and enrolled in the study.
Data for this study came from the Community Assessments that were administered to all active-duty members and their spouses on the participating bases in 2006 and 2008. From April 2006 to June 2006, 16,020 active-duty members and 4,833 spouses at the 24 participating bases completed the online Community Assessments. From April to June 2008, 16,998 active-duty members and 3,410 spouses at the participating bases completed the online Community Assessments. Analyses including individual-level outcomes were restricted to active-duty members (n = 33,018). Analyses including family outcomes were limited to individuals who were in romantic relationships or had children; thus, this analysis included both active-duty members and spouses (n = 35,297). However, in cases in which both the active-duty member and the spouse participated in the survey, the spouse was selected for analysis.
The active-duty-only sample was 73.2 percent male, with an average age of 32 years. Sixty-seven percent were married, 53 percent were parents, and 22 percent were officers. The active-duty member or spouse sample was 58 percent male, with an average age of 33. Eighty-five percent were married, 65 percent were parents, and 24 percent were officers.
Assigning the bases to NORTH STAR versus the control group was done using Microsoft Excel's random number generator. The control group received enhanced feedback; in other words, control bases were sent detailed feedback of their Community Assessment results. This also was provided to NORTH STAR bases; however, in the control condition no additional training or explanation of the results was provided.
The outcomes of interest included secretive problems, hazardous drinking, controlled prescription drug misuse, suicidality, clinically significant interpersonal violence (IPV) and child abuse (emotional and physical), and cumulative risk (however, the CrimeSolutions review of this study focused on the impact of NORTH STAR on hazardous drinking, controlled prescription drug misuse, suicidality, physical clinically significant IPV, physical clinically significant child abuse, emotional clinically significant IPV, and emotional clinically significant child abuse). Hazardous drinking was measured using the Alcohol use Disorders Identification Test, which is a 10-item self-report measure of alcohol dependence. Controlled prescription drug misuse was measured by participants completing a checklist of commonly abused controlled prescription medications, indicating their frequency of use. Suicidality was measured with four items from the Youth Risk Behavior Survey. Finally, physical clinically significant IPV, physical clinically significant child abuse, emotional clinically significant IPV, and emotional clinically significant child abuse were measured using the Family Maltreatment measure.
The study used repeated cross-sectional surveys that sampled each participating community at multiple time points. Multilevel analysis with robust maximum likelihood estimation was used to measure the impact of NORTH STAR in independent samples of people within each base in 2006 and 2008. The unit of analysis was individuals. Individual-level outcomes were analyzed using the active-duty sample, while family-level outcomes (physical clinically significant IPV, physical clinically significant child abuse, emotional clinically significant IPV, and emotional clinically significant child abuse) were measured with the active-duty members and spouse sample. No subgroup analysis was conducted.
Study
Slep and colleagues (2021) used the same randomized experimental design as Study 1 (Slep et al. 2020); however, in Study 2, the unit of analysis was bases, rather than individuals. The study compared the outcomes of Air Force bases assigned to NORTH STAR (n = 12 bases) with bases that were assigned to the control group (n = 12 bases). As such, the data, the sample, the randomization procedure, and the outcomes of interest are the same in Study 1 and Study 2. The primary difference between the studies is that analysis of the outcome data was conducted at the individual-level in Study 1 and conducted at the base-level in Study 2.
To measure the outcomes of interest, the study used repeated cross-sectional surveys that sampled each participating community at multiple time points. Logistic regression models were used to estimate bases? changes in outcomes. Individual-level outcomes were analyzed using the active-duty sample, while family-level outcomes (physical clinically significant IPV, physical clinically significant child abuse, emotional clinically significant IPV, and emotional clinically significant child abuse) were measured with the active-duty members and spouse sample. No subgroup analysis was conducted.