When the average person thinks about domestic violence they likely think of a male perpetrator and a female victim and many of us assume that the numbers support that idea. This article reviews the most comprehensive analysis of the data on intimate partner violence (IPV) to date, The Partner Abuse State of Knowledge project (PASK) .
The PASK project was a massive review by 42 scholars and 70 research assistants at 20 universities and research institutions which summarized the results from 1700 different studies.
I found the data which shows the differences of violence between the sexes, of mainly heterosexual couples, this data supports the idea that in general IPV is close to symmetrical among each sex, even when considering data from patriarchal countries.
A Bit on the Idea that PV is Male-Dominated
The idea that violence is mainly male perpetration and female victimization is associated with mainly, though not exclusively, with feminist writers, usually regarding partner violence as a consequence of patriarchy or patriarchal values, and the domination of males over females, this idea is called “Feminist theory of intimate violence” . Feminist researchers normally use study samples selected for high levels of partner violence by men, such as women from refuges or violent men on treatment programs (although there is also data crime surveys).
And this idea has been accepted by some major institutions, for example, the Duluth model (a program developed to reduce domestic violence against women, and the most common batterer intervention program used in the United States) is based on the feminist theory that “domestic violence is the result of patriarchal ideology in which men are encouraged and expected to control their partners”. 
The feminist theory of intimate violence has been widely criticized and disputed… Now, onto the PASK results-
The Bi-directional and Uni-directional Data
The authors reviewed 50 studies which reported rates of bi-directional versus uni-directional. As shown in Figure 1, for almost all categories (apart from military and male treatment samples), most of the violence was perpetrated by both the male and female in the relationship, followed by female on male violence, followed by male on female violence.
This indicates that bi-directional violence is a common IPV pattern, if not the most common and suggests that women play a larger role in the occurrence of IPV than people care to admit.
The Worldwide Abuse Data
They also looked at 73 studies from 49 countries containing data on both male and female IPV. There were 117 direct comparisons across gender for physical PV, 54 comparisons made for psychological abuse and 19 direct comparisons were made for sexual PV.
The percentage of partner abuse that was higher for female perpetration/male victimizations compared to male perpetration/female victimization, or was the same is shown in Figure 2.
Note: the only category which had the same amount of female to male violence and male to female violence was physical IPV.
It should be said that in many comparisons, the differences were slight. When the close percentages are taken into account, then the overall percentage of adult general IPV that is symmetrical – comparable across gender, or higher in the direction of male victimization/female perpetration – constitutes the majority of IPV throughout the world.
What about self-defense?
When talking about domestic violence between the sexes you often hear about women almost always committing any partner violence in self-defense, for example, the influential World Health Organization report on violence states that “Where violence by women occurs it is more likely to be in the form of self-defense.”  So let’s see how that claim holds up in this review.
Ten papers containing gender-specific statistical analyses were reviewed, five indicated that women were significantly more likely to report self-defense as a motive for perpetration than men. Four papers did not find statistically significant gender differences, and one paper reported that men were more likely to report this motive than women.
For non-perpetrator samples, the rates of self-defense reported by men ranged from 0% to 21%, and for women, the range was 5% to 35%. The highest rates of reported self-defense motives (50% for men, 65.4% for women) came from samples of perpetrators, who may have reasons to overestimate this motive so we will take that result with a heap of salt. The authors also mention that it might be particularly difficult for highly masculine males to admit to perpetrating violence in self-defense, as this admission implies vulnerability. So there is not as of yet enough data to provide conclusive results, that being said it appears that (assuming the respondents were telling the truth) the majority of violence, both male and female is not in self-defense.
 All 17 manuscripts of the PASK project can be accessed here: http://www.domesticviolenceresearch.org/17-full-pask-manuscripts-and-tables-of-summarized-studies/
A summary of the PASK findings can be found here: http://www.domesticviolenceresearch.org/domestic-violence-facts-and-statistics-at-a-glance/
 Krug, E. G., Dahlberg, L. L., Mercy, J. A., Zwi, A. B., & Lozano, R. (2002). World report on violence and health. Geneva, Switzerland: World Health Organization.
 Wayne Bennett; Kären Hess (2006). Criminal Investigation (8th ed.). Cengage Learning. p. 281.
 Claire Houston, How Feminist Theory Became (Criminal) Law: Tracing the Path to Mandatory Criminal Intervention in Domestic Violence Cases, 21 Mich. J. Gender & L. 217 (2014).
 Dutton, D. and Nicholls, T. (2005). The gender paradigm in domestic violence research and theory: Part 1—The conflict of theory and data. Aggression and Violent Behavior, 10(6), pp.680-714.
(Also See: Dutton, D. and Nicholls, T. (2006). Corrigendum to “The gender paradigm in domestic violence research and theory: Part 1—The conflict of theory and data”. Aggression and Violent Behavior, 11(6), p.664.)