Improving risk prediction accuracy for new soldiers in the U.S. Army by adding self-report survey data to administrative data

Table 3 Odds ratios for univariate and best-fitting models (n = 18,838 men and 2,952 women)^a,b

	Male physical violence perpetration		Male sexual violence perpetration		Female sexual violence victimization
	OR	(95% CI)	OR	(95% CI)	OR	(95% CI)
I. NSS univariate model	2.1	(1.8–2.5)	1.9	(1.6–2.3)	1.9	(1.6–2.1)
II. HADS univariate model	2.5	(2.2–2.8)	1.5	(1.3–1.7)	1.7	(1.3–2.2)
III. Best-fitting model
NSS	1.6	(1.3–1.9)	–	–	1.8	(1.5–2.1)
HADS	2.1	(1.9–2.5)	1.4	(1.2–1.6)	1.3	(1.0–1.8)
NSS*Time (0–12 Mo)	–	–	2.3	(1.8–2.9)	–	–
NSS*Time (13–24 Mo)	–	–	1.7	(1.3–2.1)	–	–
NSS*Time (25+ Mo)	–	–	1.3	(0.8–2.2)	–	–

Abbreviations: OR odds ratio, CI confidence interval, NSS standardized predicted log odds from model based on survey data, HADS standardized predicted log odds from model based on administrative data
^aThe NSS respondents considered here were surveyed between April 2011 and November 2012. Administrative data were available through December 2014 (25-44 months after the survey). The sample size decreased with duration both because of attrition and because of variation in time between survey and end of the follow-up period. The sample included 18,838 men (decreasing to 16,479 by 12 months, 15,306 by 24 months, and 3,729 by 36 months) and 2,952 women (decreasing to 2,300 by 12 months, 2,094 by 24 months, and 687 by 36 months).
^bAll coefficients were estimated controlling for time (number of months in service)

ISSN: 1471-244X