In the current study, a comprehensive review of the literature was conducted and 50 studies (n = 48 empirical studies; n= 1 meta-analysis; n = 1 book chapter) that reported rates of bi-directional versus uni-directional violence were uncovered using a variety of search engines and key terms.  Included studies were published in 1990 or later, appeared in peer-reviewed journals, and contained empirical data. These studies were then categorized by the nature of the sample they assessed as follows: large population samples, smaller community samples, University/school samples, treatment seeking samples, legal/criminal justice related samples, and samples assessing the relationships of gay, lesbian, and bi-sexual individuals (GLB). Each table contains the rates of bi-directional and uni-directional violence. As reported, among uni-directional violence rates, male-to-female (MFPV) and female-to-male (FMPV) percentages were included or were derived on the basis of data contained within the article. Also, in the corresponding publication, specific additional results were calculated to determine the overall rate of IPV in each sample type. These rates were then weighted by the sample sizes from which they were derived.

Among epidemiological/population samples, the average weighted rate of IPV reported was 16.3% (22.1% unweighted). Using weighted averages, among those reporting IPV, 57.9% of the IPV reported was bi-directional. Of the remaining 42.1% that was reported as uni-directional IPV, 13.8% was MFPV, 28.3% was FMPV, and the ratio of uni-directional FMPV to MFPV was 2.05 weighted (2.02 unweighted).

Among community samples, the average weighted rate of IPV reported was 47.0%. Using weighted averages, among those reporting IPV, 59.6% was bi-directional. Of the remaining 40.4% that was reported as uni-directional IPV, 17.5% was MFPV, 22.9% was FMPV, and the ratio of uni-directional FMPV to MFPV was 1.30 weighted (1.98 unweighted).

Among school and college samples, the average weighted rate of IPV reported was 39.2%. Using weighted averages, among those reporting IPV, 51.9% was bi-directional. Of the remaining 48.1% that was reported as uni-directional IPV, 16.2% was MFPV, 31.9% was FMPV, and the ratio of uni-directional FMPV to MFPV was 1.96 weighted (2.18 unweighted).

Among legal or female-oriented clinical/treatment seeking samples that were not associated with the military, the average weighted rate of IPV reported was 70.6%. Using weighted averages, among those reporting IPV, 72.3% was bi-directional. Of the remaining 27.7% that was reported as uni-directional IPV, 13.3% was MFPV, 14.4% was FMPV, and the ratio of uni-directional FMPV to MFPV was 1.09 weighted (1.07 unweighted).

Finally, among IPV treatment samples within the U.S. Military and at-risk males samples, the average weighted rate of IPV reported was 99.9%%. Using weighted averages, among those reporting IPV, 39.3% was bi-directional. Of the remaining 60.7% that was reported as uni-directional IPV, 43.4% was MFPV, 17.3% was FMPV, and the ratio of uni-directional FMPV to MFPV was .40 weighted (.33 unweighted).

When using unweighted overall rates of IPV, the samples differed significantly. Follow-up comparisons revealed that the overall rate of IPV in school samples did not differ significantly from large population and community samples. All other comparisons of overall rates of violence revealed significant differences with the epidemiological studies yielding the lowest reported rates of IPV and the legal/criminal justice samples yielding the highest rates. When examining rates of bi-directional versus unidirectional IPV, the unweighted bi-directional IPV rates ranged from 49.2% to 69.7% whereas the uni-directional rates ranged from 30.3% to 50.4%. Bi-directional IPV is a prevalent phenomenon. These findings held even though the prevalence of violence differed greatly among the samples studied. This robust result suggests that the role of women in violent relationships is important to consider, even if all aspects of women’s perpetration of IPV are not symmetrical to men’s perpetration of IPV. Moreover, bidirectional IPV was the most common type of violence for all sample types except the legal/treatment seeking samples of military or at-risk men. However, even among this sample type the uni-directional rate was only slightly higher as compared to the rate of bi-directional violence (50.4% vs. 49.2%, respectively). This again suggests that clinicians and researchers in all settings should be routinely assessing for both perpetration and victimization.

Furthermore, statistical analyses indicated that among the five sample types, the unweighted percentages of bi-directional as compared to uni-directional IPV did not differ significantly. Thus, rates of uni-directional and bi-directional IPV were similar across all five sample types. However, significant differences did emerge within the group of unidirectional IPV perpetrators, such that a significantly higher rate of uni-directional MFPV (38% unweighted) and a significantly lower rate of FMPV (12.4% unweighted) was found among the military treatment/legal samples of males as compared to all other sample types. Post hoc analyses revealed that these rates did not differ from the female-oriented treatment/legal samples. But, in female-oriented treatment/legal samples, the rate of FMPV (15.7% unweighted) was significantly lower as compared to the rate of FMPV in every other sample type. Thus, the ratio of uni-directional female-to-male (FMPV) compared to male-to-female (MFPV) IPV differed significantly among samples with higher rates of female-perpetrated unidirectional violence found in four of the five sample types considered. A higher ratio of male-to-female unidirectional violence was only found in criminal justice/legal studies that relied on police reports of IPV perpetration and/or in samples drawn from the U.S. military. Competing explanations for the differing gender-based ratios were offered in the discussion section of the published manuscript. These considerations need to be systematically varied and tested empirically in order to fully understand differences in the reported expression of IPV across samples and settings.

In further exploratory analyses, differences in the extent of bi-directionality in the expression of IPV were not found across samples of gay, lesbian, or bi-sexual individuals; however, among a limited number of studies, rates of bi-directional violence were found to vary significantly by race/ethnicity. Specifically, only 50.9% of IPV was bi-directional according to White reporters, as compared to 49.0% among Hispanic reporters, and 61.8% among Black reporters. Significantly different ratios of FMPV to MFPV were also obtained across the ethnic groups such that the ratio was 2.27 for Black reporters, 2.26 for White reporters, and 1.34 for Hispanic reporters. However, it must be noted that these ratios differ dramatically from those reported above when the sample is drawn from the military (0.61 for Black reporters, 0.50 for White reporters, and 0.00 for Hispanic reporters). These findings warrant additional investigation.

In conclusion, our results demonstrate the amount of overall IPV differed significantly among samples, but the percent that was categorized as bi-directional did not. This indicates that bi-directional violence is a common IPV pattern and suggests that women play a larger role in the occurrence of IPV than previously thought. Such findings have considerable implications for assessment, legal, intervention, and preventive efforts. It is suggested that if one resolution of the gender symmetry/asymmetry debate is to argue that there are subtypes of male and female intimate partner violence perpetrators, or that there are different patterns of violence amongst relationships characterized by IPV (Johnson, 2005; Johnson, 2006), researchers and clinicians will need to work to together to determine how to reliably and meaningfully make these determinations in ways that will facilitate our ability to effectively prevent and treat all types of IPV.