The Competition–Violence Hypothesis: Sex, Marriage, and Male Aggression

Abstract

Sexually active men, who are not in a monogamous relationship, may be at a greater risk for violence than men who are sexually active within monogamous relationships and men who are not sexually active. The current study examines changes in sexual behavior and violence in adolescence to early adulthood. Data on male (n = 4,597) and female (n = 5,523) respondents were drawn from four waves of the National Longitudinal Study of Adolescent and Adult Health (Add Health). HLM regression models indicate that men who transition to a monogamous, or less competitive, mode of sexual behavior (fewer partners since last wave), reduce their risk for violence. The same results were not replicated for females. Further, results were not accounted for by marital status or other more readily accepted explanations of violence. Findings suggest that competition for sex be further examined as a potential cause of male violence.

Keywords:

• competition
• marriage
• sex
• violence

Acknowledgments

Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design.

Notes

This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill.

¹ Harris, K.M., C.T. Halpern, E. Whitsel, J. Hussey, J. Tabor, P. Entzel, and J.R. Udry. 2009. The National Longitudinal Study of Adolescent to Adult Health: Research Design [WWW document]. URL: http://www.cpc.unc.edu/projects/addhealth/design. The sampling weight variables provided by Add Health were not used in this study due to the use of male and female subsamples and the implementation of hierarchical linear models that account for the complex error structure in the data. Missing cases for certain variables constituted less than two percent of the analytic sample. The mean was imputed for the census-block variables, percent below the poverty line and percent never married, while zero was imputed for all other variables with missing cases and was interpreted as “no evidence for the behavior in question.”

² The multilevel/mixed regression is a two-level model, expressed below as a composite of level 1 and level 2.

Composite Model:

The notation is the stochastic part of the model and represents the random variance components (ζ) for time, the intercept, and the error (ε), at level 1. The notation is the structural part of the model and represents a vector of fixed effects and the intercept for the ith respondent at the jth time of measurement, at level 2 (see Singer & Willett, 2003).

³ The results from the HLM models were compared to results from a GEE logit model (general estimating equation), fit for repeated observations as well as to a general linear model for estimating fixed effects. The “genmod” SAS procedure produced the GEE model estimates. The SAS “general linear model” procedure was used to estimate the fixed effects regression parameters. The SAS “mixed” procedure was used to estimate the HLM regression parameters shown in Tables 2 and 3. Results from these alternative models were nearly identical to that of the HLM models.

Additional information

Notes on contributors

Patrick M. Seffrin

Seffrin is an associate professor of sociology at Marywood University. His research interests include crime and violence, alcohol, drugs and drug dealing, and the influence of social and cultural factors.