JOURNAL OF DOMESTIC VIOLENCE, VOLUME 12, NUMBER 4 (1997), p 417-443
INTIMATE VIOLENCE IN CANADA AND THE UNITED STATES: A CROSS-NATIONAL COMPARISON
Elaine Grandin and Eugen Lupri
The University of Calgary, Department of Sociology, Calgary, Alberta, Canada, TZN 1N4.
Using data from the 1985 U.S. National Family Violence Resurvey and the 1986 Canadian National Family Life Survey this paper compares incidence of intimate violence or "common couple violence" (Johnson, 1995) in both countries. As expected, gender symmetry characterizes common couple violence, which is a product of the privatized setting of many American and Canadian households. Although the United Slates exhibits significantly higher rates of societal violent crime than Canada, Canadian women and men were more likely than their American counterparts to use severe intimate violence and to inflict it, as well as minor violence, more often, which is contrary to the culture of violence theory that guided the study. Similarly the higher rates of wife-to-husband severe violence across the life course in both countries are inconsistent with the theory. Several ad hoc explanations are presented to account for these unexpected findings.
A distinction that does not further understanding is no distinction.
Johann Wolfgang Goethe (1828)
Possibly no other issue related to family violence has recently received so much media attention and has generated so much controversy in Canada as the debate over prevalence and seriousness of spousal abuse, including both husband-to-wife and wife-to-husband violence. Although Canadian sociologists, criminologists and feminist scholars came relatively late to the study of family violence, striking similarities exist in the nature of the debate that took place in the United States in the late 1970s and in Canada a decade later.
This paper focuses on two main issues that are integral components of the above debate: the controversial issue of gender symmetry in couple violence and the issue of country differences in prevalence rates of couple violence. The comparative analysis is based on data derived from the 1986 Canadian National Family Life Survey and the 1985 U.S. National Family Violence Resurvey, which we describe later. Our cross-national comparison, we hope, will shed further light on both issues, which we detail below.
THE ISSUE OF GENDER SYMMETRY/ASYMMETRY IN INTIMATE VIOLENCE: COMMON COUPLE VIOLENCE VERSUS PATRIARCHAL TERRORISM
The 1975 U.S. National Family Survey conducted by Straus et al. (1980) and the 1985 U.S. National Family Resurvey conducted by Straus and Gelles (1992b) produced fairly consistent patterns of gender symmetry in spousal abuse. When female and male respondents were asked identical questions based on the Conflict Tactics Scales (Straus 1992b, 1979), women were at least as likely as men, if not more likely, to report committing physically violent acts against their married or cohabiting partners. Similar gender patterns of partner abuse were found in Canadian studies (Bland and Orne, 1988; Brinkerhoff and Lupri, 1988; Brinkerhoff et al., 1992; Kennedy and Dutton, 1989; Lupri, 1990; Sommer, 1994). Feminist scholars have criticized these findings on various theoretical, conceptual, and methodological grounds, but particularly challenged the ensuing quantitative analyses that focused exclusively on acts of spousal abuse, thereby ignoring both the context in which violence arises and the differential power constellations that precipitate specific violent acts. (For a summary of these and other criticisms, see Dobash et al., 1992; for earlier responses to many of those feminist challenges, see Straus, 1992b.) According to feminists, husband abuse is asymmetrical rather than symmetrical, and the motives and effects of women's violence differ sharply from men's. Unlike wife abuse, husband abuse is not viewed as a social problem by most feminist scholars. The debate between the scholars who represent what has become known as the family violence perspective (Straus, 1993) and those who represent the feminist perspective (Kurz, 1993) continues in both countries, as does the controversy surrounding the gender symmetry/asymmetry issue among experts and laypersons alike (see Fekete, 1994).
In a recent comprehensive reanalysis, Johnson (1995) has made an attempt to clarify the controversial issue of gender symmetry and the discrepant findings that are reported in the two literatures. He argues that there are, in fact, two distinct forms of couple violence taking place in American households: "patriarchal terrorism" and "common couple violence" (1995, p. 283).
Patriarchal terrorism is simultaneously a mechanism of men's control over women and a pattern of violence that occurs routinely, develops into escalating coercive cycles of violence, and may produce counterviolence. Along with other modes of control, such as isolation, neglect, and economic dependence, patriarchal terrorism, according to Johnson, differs in its structure and motivation from common couple violence. Patriarchal terrorism is uncovered in qualitative studies of battered women at sites where victims seek help: women's shelters, transition houses, law enforcement agencies, hospital emergency rooms. Rooted historically and culturally in patriarchal ideas of male ownership of their female partners, this type of violence contrasts sharply with common couple violence uncovered through national surveys yielding quantitative data of the kind used in this cross-societal analysis.
One of the major drawbacks of the sample survey approach lies in its inherent limitation to capture the more severe form of patriarchal terrorism. Cross-sectional sample surveys are ill-equipped to trace the coercive processes of escalation, intensity, and reciprocity in severe couple violence over time, although chronicity or frequency of violence can be captured through retrospective self-reports, as we will show later. Johnson's rendering of common couple violence captured in large sample surveys is based on the observation that the structures of patriarchy and its underpinnings have been eroded over the past two decades, albeit unevenly, and that common couple violence is less a product of patriarchy, and more a product of the less-gendered causal processes" (1995, p. 285). Although not lacking in dynamic, common couple violence is tied conceptually to the inevitability of conflict which "occasionally gets out of hand, leading usually to minor forms of violence, and more rarely escalating into serious, sometimes even life-threatening forms of violence'' (Johnson, 1995, p. 285). Indeed, national surveys in the U.S, and Canada have found a strong association between level of conflict between intimate partners and the incidence of common couple violence (Lupri et al., 1994).
Johnson's perceptive distinction between two forms of violence is a conceptual breakthrough because it helps to explain the dramatic difference in gender patterns of violence tapped by the two national U.S. surveys and in the statistics by the research tapped with shelter populations and criminal justice and divorce court data. Johnson's assertion that these two forms of violence represent distinct and nearly nonoverlapping phenomena contributes to a fuller understanding of the complexity of intimate violence and to the resolution of the gender symmetry/asymmetry controversy.
One major purpose of our analysis is an attempt to test Johnson's conceptualization of common couple violence within a cross-national setting and to examine whether what he reported for American households applies equally to Canadian households. Specifically, we tested whether patterns of gender symmetry in common couple violence are similar in both countries. If they are, this finding would lend cross-national validity to Johnson's conceptual distinction. If they are not, how can we explain the different patterns in the two countries? The comparative approach allowed us to maximize the "potential range of variation" in couple violence, which is an essential requirement of theory construction (Lee, 1987, p. 77).
THE ISSUE OF COUNTRY DIFFERENCES IN COMMON COUPLE VIOLENCE: THE CULTURE OF VIOLENCE THEORY
The cross-national data also allowed us to test the culture or subculture of violence theory that holds that violence is unevenly distributed in the social structure (Wolfgang and Ferracuti, 1967, 1972). This is the second major purpose of our study. The differential distribution of violence, within or between societies, is a function of differential cultural norms and values concerning violence (Gelles and Straus, 1979). Thus, all forms of violence reflect effective socialization into the value system and norms of that subculture or that culture as a whole. In applying the culture of violence theory to the family, Gelles and Straus (1979), p. 566) argue that we can view "the family as a training ground for violence, since it is the major unit in transmitting cultural norms." Similarly, Bersani and Chen (1988) view the culture of violence theory as a perspective that embraces both differential association and learning theory.
A prominent extension of the culture of violence thesis is Levinson's (1989, pp. 40-43) general culture pattern model, which suggests that some societies have a basic set of values and beliefs that emphasize aggression and violence. Cross-cultural research provides him with some support for the proposition that family violence will be more common in societies in which other forms of violence are also common. Perhaps the strongest support for the culture pattern theory is research by Masamura (1979), who analyzed comparative data from a worldwide sample of 86 primitive societies, using rigorous methodological techniques in the selection procedure and reliable cross-cultural codes to measure aggression. He found that wife beating was associated with high rates of personal crime, theft, aggression, suicide, homicide, feuding, and warfare. His work gives cross-cultural support to the notion that different kinds of aggression are related to one another because they are all outcomes of an underlying set of beliefs that condone aggression.
Results of this comparative research may not be directly generalizable to the highly developed and industrialized United States and Canada. Nevertheless, findings encouraged us to further pursue the idea that all forms of violence and aggression in society, including intimate violence, may have a common etiology. It is within these theoretical and comparative restrictions that we used the culture-of-violence theory in our macrolevel analysis of the incidence of common couple violence in the two countries.
A PRELIMINARY U.S.-CANADIAN COMPARISON OF MARITAL ABUSE
Although there exists a fairly large body of cross-cultural data on family violence, especially physical wife abuse (for recent work and reviews, see Gelles, 1993; Hoffman et al., 1994; Kumagai and Straus, 1983; Levinson, 1989; Steinmetz, 1987), truly U.S.-Canadian comparisons of intimate violence are nonexistent with the exception of some preliminary research conducted by Steinmetz (1981) in the early 1970s. Steinmetz compared the nature and extent of physical marital violence-based on items in the Conflict Tactics Scales (Straus, 1979)--in six societies: the United States and Canada, as well as Finland, Puerto Rico, British Honduras, and Israel.
On the basis of the data from these six societies, Steinmetz found that the United States had the highest level of societal violence (e.g., serious crime), but that levels of marital violence placed U.S. couples in the middle of the group. Canadian couples were somewhat less likely than their U.S. counterparts to have engaged in marital violence, and Canada's level of societal aggression ranked considerably lower than that established for the United States. An interesting pattern, however, was revealed in frequency and severity with which American and Canadian husbands and wives reportedly inflicted abusive acts onto their intimate partners: the mean frequency of violence scores for wives were found to be higher than those reported for husbands, and Canadian wives outscored American wives. Another observation from Steinmetz's comparative study is worth mentioning: the highest percentage of marital abuse was reported for Finnish husbands and wives; yet Finland as a nation scored consistently lower in the group on all measures of societal violence. Also interesting was the very low frequency of violent acts by each of these Finnish couples.
Steinmetz's research is limited by the use of college students to report on their parents' violent acts, the non-representativeness of her samples, and the failure to precisely define the measures of aggression at the macro level. Yet despite these methodological shortcomings, the study provided us with some theoretical direction.
EMPIRICAL PROPOSITIONS
Although not a "theory" in the more rigorous sense of the word, the culture of violence thesis presents a potentially useful theoretical framework for a cross-national comparison of common couple violence between the United States and Canada, specifically for the proposition that the higher the level of societal violence, the higher the level of common couple violence.
According to criminologist John Hagan (1994, p. 22), the United States and Canada are thought to be very similar because they share an English language and cultural heritage as well as a common border, and are one another's largest trading partners. More importantly, the two nations have similar levels of socioeconomic development, similar high levels of female labor force participation, and a similar high standard of living. Although gender equality has not been realized fully in either Canada or the United States, the status of women is improving. Egalitarian emphases in intimate family and nonfamily relationships have begun to replace traditional gender role attitudes and behaviors in both countries. On tile basis of this similarity in macrolevel indicators of the social structure, we assume that Canada and the United States represent comparable societal units.
Figure 1, which depicts the serious violent crime rates for both countries from 1962 to 1985, illustrates vividly that, at the time of both national surveys, the levels of violent crime in Canada are well below those in the United States, and shows how the difference between the two countries has grown. According to Hagan (1994), incidence of the most serious types of crime in the United States was more than five times the rate for Canada in 1985. In that year, well over 500 such crimes per 100,000 population were committed in the United States, as compared with just over 100 in Canada. In addition, the incidence of the most serious crimes has been rising in the United States in recent years, while the rate has been falling in Canada. Between 1983 and 1985 the rate of serious violent crime rose 3 percent in the United States, but fell 10 percent in Canada.
Overall homicide rates reflect a similar pattern. In 1972, the Canadian rate per 100,000 population was 1.4, as compared with 4.5 in the United States. In 1987 the rates were 2.2 and 8.3, respectively (Hagan, 1991).
Because of the magnitude of the difference in rates of serious violent crime outside the home and because the culture of violence theory assumes that all violence, including family violence, has a common etiology, we hypothesized that American couples would he significantly more likely than their Canadian counterparts to engage in common couple violence. This prediction is based on the assumption that all violence has a "cultural spillover" effect (Feld and Straus, 1992, p. 493), in which the existence of one type of violence, for example, serious violence "in the streets," tends to legitimate and increase the likelihood of other forms, such as common couple violence. In addition, we expected that U.S. couples who abuse their partners would do so more frequently, and that they would indict more serious marital abuse than their Canadian counterparts.
Furthermore, on the basis of propositions derived from the subculture of violence thesis, whereby criminal aggression is negatively related to age (Wolfgang and Ferracuti, 1967, 1972), we expected the perpetrator's age to be related inversely to the level of common couple violence in both countries. As state Wolfgang and Ferracuti (1967)explicitly, "This (cultural) ethos of violence is most prominent in a limited age group, ranging from late adolescence to middle age" (p. 159).
While men are more likely than women to commit serious criminal violence outside the home, patterns of gender symmetry characterize common couple violence in the home, as we discussed earlier. Hence we predicted similar gender patterns of couple violence in both countries.
METHODS AND DATA SOURCES
Following Gelles and Straus (1979), we define violence as an act committed with the intention or perceived intention of physically hurting another person. We limit our comparative study to physical violence because it is highly correlated with psychological abuse, and because psychological violence has been identified as a precursor of physical violence (Stets, 1990; Straus and Sweet, 1992).
Data Collection
The data for this paper were derived from two sources: the 1986 Canadian National Family Life Survey (Lupri, 1989, 1990) and the U.S. National Family Violence Resurvey of 1985, conducted by Straus (1992a) and Straus and Gelles (1992a). The Canadian national survey involved interviewing a probability sample of 1834 women and men, age 18 and older, excluding institutionalized persons and the Native population. At the close of a personal interview, a self-administered questionnaire with a self-addressed stamped envelope was left behind for completion by every respondent who was ever married or cohabiting. The self-administered questionnaire included items tapping the respondent's marital or marriage-like relationships and a modified version of the Conflict Tactics Scales (). UStraus, 1992bsable questionnaires were returned by 652 female and 471 male respondents, or 73.4% of the 1550 eligible ever-married or cohabiting sample members. The nationally representative sample and other aspects of the study are detailed in Lupri (1989).
To make the U.S. data comparable, we used the unweighted national probability sample of 4,032 women and men of the 1985 U.S. National Family Violence Resurvey (Straus and Gelles, 1992a: APPENDIX A, pp. 530-531), who responded to telephone interviews conducted by Louis Harris and Associates. The response rate was 84 percent. Because our analysis involved a comparison of national incidence rates, it includes only those American and Canadian respondents who were married or cohabiting during the 12 months preceding the surveys.
Although the two national surveys are comparable, they differ in two major respects. First, the Canadian data were collected through self-administered questionnaires, as described earlier, whereas the American data were gathered through telephone interviews. Second, to achieve comparability of data, the present analysis is limited to self-reports by perpetrators.
Measures
Common Couple Violence
The questions on violence were based on a modified version of the Conflict Tactics Scales (CTS) developed by Straus (1979) and associates, and were designed to measure the incidence of three general kinds of tactics: reasoning, verbal aggression, and violence. We limit our analysis to the use of the eight items displayed in Table I, which describe the use of physical force and are common to the Canadian and the American national surveys. These acts are ordered in terms of increasing severity and may be categorized as "overall," "minor," or "severe." Respondents were also asked how frequently each act was committed. Frequency was categorized as follows: 0 = never; 1 = once; 2 = twice; 3 = 3 to 5 times; 4 = 6 to 10 times; 5 = 11 times or more.
To maximize comparability, we made two methodological transformations: (1) the item "choked" in the U.S. resurvey was omitted from the analysis because it was not included in the Canadian survey; and (2) it was necessary to collapse response categories 5 and 6 in the American restudy into a single category of 5 (11 times or more).
We performed exploratory factor analysis on the eight violent or potentially physically violent acts listed above, using principal factoring with iterations (Zeller and Carmines, 1980), separately for each of the four subsamples: husband-to-wife violence and wife-to-husband violence in each country. In each case, the factor loadings closely resembled those obtained by Straus (1992b, p. 39); this finding established an adequate degree of validity for the use of the CTS in this comparative study. (The parameters of scale construction are available on request.) The alpha coefficients of reliability (Cronbach, 1967) are reasonably high in each case: Canadian hushand-to-wife, .75; Canadian wife-to-husband, .90; American husband-to-wife, .63; American wife-to-husband, .79. The statistical output for the U.S. sample differs slightly from that reported in , Straus and Gelles (1992b)because the present analysis relies exclusively on self-reports by perpetrators, as mentioned earlier, and omits the reports provided by American victims.
Age
In the absence of longitudinal data, we used respondents' age to examine the incidence of common couple violence across the life course. One must bear in mind the pitfalls of inferring change from cross-sectional studies--that is, the confounding influences of time (age, stage), culture, and socialization.
Limitations and Analysis Strategy of This Study
Like all research, this comparative study suffers from several self-imposed limitations and shortcomings. Our quantitative analysis focuses exclusively on acts of common couple violence, not on the context in which violence arises or on what precipitated specific violent acts. Nor do we deal here with the injuries, pains, suffering, or other negative consequences of any violence between the partners (see Grandin, 1995; Grandin et al., 1997; Straus and Gelles, 1992a). Although we consider such an analytic strategy of utmost importance, it is beyond the scope of the present paper. Our aim is to test cross-nationally a few basic propositions, as detailed earlier; thus we think it best, theoretically and empirically, to limit ourselves to a comparison of national incidence rates of physical violence based on the CTS that have been found to be reliable instruments for cross-cultural studies.
RESULTS
In view of the culture of violence thesis, which posits a positive association between societal aggression and marital violence, and because of the earlier documentation that the United States exhibits significantly higher rates of societal violent crime than Canada, we hypothesized significantly higher rates of common couple violence in the American sample. For both countries we also expected to find gender-balanced patterns of common couple violence, which is a product of a violence-prone culture and the privatized setting of most Canadian and U.S. households (Johnson, 1995). Similarly, we expected couple violence rates to decline over the life course, regardless of country. We used several measures to test our hypotheses; we define and describe each of these measures in the appropriate section.
Cross-National Comparison of Common Couple Violence Rates
Table I presents incidence rates for each of the eight physically violent acts and the frequency of these acts in both the Canadian and the American samples. Following Straus and Gelles (1992b, p. 121), we divide the eight acts into three Minor Violence Acts and five Severe Violence Acts. This distinction is crucial for our comparative analysis, as we show later. The columns titled "Rate" contain the incidence rate (percentage), which reflects whether any violent acts occurred during the 12 months preceding the survey. The columns titled "Freq" contain a chronicity or frequency measure, which reflects the number of times each act was committed during the year.
Table I. Percentage Comparison of Common Couple Violence for Canada (1986) and the United States (1985)
| Canadian | American | Canadian | American | |||||||
| Husband- to-Wife Violence (n=426) |
Husband- to-Wife Violence (n=1622) |
T-Value for Cdn-Amer |
Wife-to- Husband Violence (n=528) |
Wife-to- Husband Violence (n=1889) |
T-Value for Cdn-Amer | |||||
| Type of Violence | Rate | Freq | Rate | Freq | Difference | Rate | Freq | Rate | Freq | Difference |
| A. Minor Violence Acts | ||||||||||
| 1. Threw something at the other | 9.3 | 1.8 | -4.81*** | 12.5 | 4.0 | -5.19*** | ||||
| Once in Past Year | 3.9 | 1.1 | 4.3 | 1.9 | ||||||
| Twice in Past Year | 5.4 | 0.7 | 8.2 | 2.1 | ||||||
| 2. Pushed grabbed or shoved | 11.9 | 9.1 | -1.53 | 13.2 | 9.2 | -2.33* | ||||
| Once in Past Year | 3.8 | 5.1 | 5.6 | 3.9 | ||||||
| Twice in Past Year | 8.1 | 4.0 | 7.6 | 5.3 | ||||||
| 3. Slapped the Other | 4.9 | 2.3 | -2.17* | 7.6 | 3.6 | -3.03*** | ||||
| Once in Past Year | 1.7 | 1.4 | 4.4 | 1.9 | ||||||
| Twice in Past Year | 3.2 | 0.9 | 3.2 | 1.7 | ||||||
| B. Severe Violence Acts | ||||||||||
| 4. Kicked, hit or hit with object | 6.5 | 0.6 | -4.50*** | 6.3 | 2.3 | -3.32*** | ||||
| Once in Past Year | 1.2 | 0.3 | 1.5 | 1.1 | ||||||
| Twice in Past Year | 5.3 | 0.3 | 5.0 | 1.2 | ||||||
| 5. Hit or tried to hit with object | 5.3 | 0.5 | -4.08*** | 9.0 | 3.1 | -4.26*** | ||||
| Once in Past Year | 3.4 | 0.4 | 4.1 | 1.6 | ||||||
| Twice in Past Year | 1.9 | 0.1 | 4.9 | 1.5 | ||||||
| 6. Beat up the Other | 2.5 | 0.3 | -2.67*** | 6.2 | 0.2 | -5.25*** | ||||
| Once in Past Year | 1.0 | 0.1 | 1.7 | 0.1 | ||||||
| Twice in Past Year | 1.5 | 0.2 | 4.5 | 0.1 | ||||||
| 7. Threatened with a knife or gun | 2.1 | 0.0 | -2.78*** | 3.5 | 0.3 | -3.64*** | ||||
| Once in Past Year | 1.2 | 0.0 | 0.0 | 0.1 | ||||||
| Twice in Past Year | 0.9 | 0.0 | 3.4 | 0.2 | ||||||
| 8. Used knife or gun on the other | 0.5 | 0.2 | -0.69 | 0.9 | 0.2 | -1.61 | ||||
| Once in Past Year | 0.0 | 0.1 | 0.1 | 0.1 | ||||||
| Twice in Past Year | 0.5 | 0.1 | 0.8 | 0.1 | ||||||
| C. Violence Indices | ||||||||||
| Overall Violence (1-8) | 18.3 | 10.6 | -3.79*** | 25.6 | 12.2 | -6.44*** | ||||
| Minor-only Violence (1-3) | 9.3 | 9.5 | 0.16 | 11.6 | 8.2 | -2.06* | ||||
| Severe Violence (4-8) | 9.9 | 1.2 | -5.94*** | 15.5 | 4.3 | -6.81* | ||||
As can be seen, a sizable group of Canadian and American women and men admits to having committed physical violent acts, if not assaults, against their intimate partners. This judgment is based on a broad definition of interspousal violence, which includes throwing almost anything at the other partner in anger, grabbing, and pushing and shoving--a very wide net indeed, and one which may include as "violent" many couples who never commit serious or physically harmful abuse. Yet even when these "minor" forms of intimate violence are excluded, levels of violence remain high.
The most noteworthy finding in Table I, however, is the overall observation that Canadian men and women are consistently more likely than their American counterparts to report a commission of a violent act, regardless of the nature of the act. For example, 9.3% of Canadian men reported throwing something at their intimate partners, compared with 1.8% of American men. The same pattern holds for Canadian females: 12.5% of Canadian females reported throwing something at their partners, in contrast to 4% of American females.
Although, as expected, the rates for the more serious violent acts are lower than for the "minor" episodes, the incidence rates for both Canadian men and women remain higher than for their American counterparts. When the specific act measured is "kicked, bit, or hit with a fist," the rate for Canadian men is 6.5%, compared with 0.6% of American men. Similar national patterns prevail among women: Canadian rates exceed those reported by American women (6.3% and 2.3%), respectively.
With the exception of "pushed, grabbed, or shoved" and "used knife or gun," the differences in rates between Canadian and American men are statistically significant at p<=.05 or higher. Similarly, the differences between Canadian and American women are statistically significant at p<=.05 or higher for all items except "used knife or gun."
The Frequency of Specific Acts of Common Couple Violence
Even though Canadian men and women report a higher incidence rate, violent acts may be more pervasive in the American sample when frequency is taken into account. Data presented in Table I differentiate between respondents who reportedly committed acts only once and those who committed acts twice or more.
The probability of committing a violent act only once remains higher among Canadian men than among American men for six of the eight physically violent acts; four of these six are considered acts of "severe violence." When the response is "twice or more," rates for Canadian males remain consistently higher than for American males. That is, the probability of repeated abuse is greater among Canadian males. In fact, for two-thirds (65%) of the Canadian men who committed any violence, it was not an isolated event. This finding is consistent with qualitative research reporting that once violence has occurred, there is a greater likelihood of recurrence and, as observed by Saunders (1986) and by Walker (1984), a higher possibility of escalation. (Without longitudinal data, the escalation thesis cannot be tested here.) It should be noted that the frequency data on American men are somewhat less supportive of the recurrence thesis: for any given act, the likelihood of repetition is lower among American males than among their Canadian counterparts.
Table I also shows the pervasiveness of wife-to-husband violence among Canadian and American couples, a highly controversial issue in both countries and one to which we return below. With the exception of "threatened with a knife or gun" and "used knife or gun," the Canadian women's rates are consistently higher than those of the American women, whether the frequency is "once" or "twice or more." As among Canadian males, a violent act committed by women is typically not an isolated event. The same pattern holds for American women, although it is less pronounced.
In sum, Table I shows clearly that the incidence rates, based on specific violent acts reportedly committed by intimate male and female partners, married or cohabiting, are significantly higher in Canada than in the United States, regardless of the perpetrator's gender. Furthermore, both husband-to-wife and wife-to-husband violence are more pervasive in Canada. With a few exceptions, the frequency data for Canadian men and women and those for American women support the notion that once a violent act was committed, it was more likely to be repeated. The findings displayed in Table I contradict our major hypothesis and do not support the culture of violence thesis.
Minor and Severe Violence
To facilitate data analysis and to establish both continuity and comparability with past research, we constructed three summated scales for the eight physically violent acts. The Overall Violence Index contains percentage of respondents in each subsample who reported committing any one of the eight physically violent acts against their intimate partners during the year preceding the survey. To address the possible objection that the Overall Violence Index is too broad, we computed a Severe Violence Index comprising only those physical acts which have a relatively high probability of causing serious injury or pain (Table I, B). Because the Overall Violence Index includes all the violent acts in the CTS, we also computed a Minor-Only Violence Index, or, as Straus (1983, p. 216) called it, an "Ordinary Violence Index." This identified respondents who engaged in throwing things, pushing, shoving, and slapping (Table I, A) but did not engage in any of the serious physical acts of violence. The computed rates for these three indices are displayed at the bottom of Table I, separately for husband-to-wife and wife-to-husband violence.
Overall rates show that almost one Canadian man in five (18.3%) reported at least one act of physical violence, in contrast to slightly more than one American man in ten (10.6%). This index for women shows that approximately one Canadian woman in four (25.3%) reported committing at least one of the physically violent acts, in contrast to just under one American woman in eight (12.2%). In each case the difference in proportion by country is statistically significant at p<=.005.
As we would expect, the severe violence rates are much lower than the rates based on the Overall Violence Index. Yet almost one Canadian man in 10 (9.9%) reportedly committed at least one of the five serious acts of physical violence, in contrast to about one American man in 100 (1.2%). (The American statistic closely approximates the 1.3% reported by Stets and Straus (1992:156) when self-reports of male perpetrators are the focus.) For Canadian women, the national rate on the Severe Violence Index is more than three times the national rate computed for American women: 15.5% and 4.3%, respectively. As in the overall rates, the national differences for severe violence are statistically significant at p<=.005 for both husband-to-wife and wife-to-husband violence.
Although the Overall and Severe Violence rates are significantly higher among Canadians, the Canadian and American rates are more symmetrical when minor-only violent acts are the focus. Rates for Canadian and American men are almost identical (9.3% and 9.5%), but Canadian women continue to report slightly higher rates of minor violence than their American counterparts (11.6% vs. 8.2%). The exclusive focus on minor acts of violence in this comparative analysis highlights the problematic and misleading nature of the rates based on the Overall Violence Index. Canada's comparatively high Overall Violence Index reflects the magnitude of the serious and severe violent acts among the sampled Canadian women and men.
Multivariate Analysis of Common Couple Violence
To assess the relative contribution of country and gender as predictor variables of common couple violence, we performed a logistic regression analysis. Because age has been found to be one of the most significant predictors of both serious criminal behavior outside of the home (Hagen, 1994; Wolfgang and Ferracuti, 1967, 1972) and common couple violence (Lupri et al., 1994; Straus and Sweet, 1992), we also entered age of perpetrator into the model.
In our multivariate analysis we used two separate indices to measure common couple violence: the Minor-only Index and the Severe Index. These indices were respectively coded as 1 if any of the three minor or any of the five severe acts (see Table I) had been committed, and zero otherwise. Country was dummy coded with zero for the United States and 1 for Canada. Gender was also dummy coded with zero for male and 1 for female. Because of the controversy surrountling the analysis of interaction effects between continuous and noninterval data (Jaccard et al., 1990), age was collapsed into three categories: 18-29 years, 30-44 years, and 45 years and older, and two dummy variables were created with 18-29 years being the reference category in each case.
Table II presents the results of the logistic regression analyses of two models for each of the two types of common couple violence. In each case, the first model regressed the relevant index on country, gender, and age. First-order interaction terms were added in the second model. We begin our interpretation by presenting the results of the multivariate analysis for minor-only couple violence and then follow with the results for severe couple violence.
Minor-Only Couple Violence
In Part A of Table II, the statistical significance (p<=.001) of the model chi-square test associated with Model 1 suggests that the additional variables add a significant proportion to the explained variance in minor-only violence over and above a model of complete independence (DeMaris, 1992). In this model, both country and age are significantly (p<=.001) related to the likelihood that minor-only violence occurred. Other things being equal, being Canadian rather than American increased the odds that minor-only violence occurred by a factor of 1.7215, or 72%. Regardless of country and gender, being middle aged rather than in the youngest age category, decreased the chances that minor-only violence occurred by a factor of .4777, or 52%. When both country and age were controlled, there was no statistically significant difference in the odds of committing minor-only violence by gender. Nevertheless, this finding is of substantive interest and consistent with national studies that report patterns of gender symmetry in common couple violence, as discussed earlier.
Table II. Logistic Regression Results of Common Couple Violence for Canada (1986) and the United States (1985)
| Model 1 | Model 2 | |||
| Variable | B | Exp(b) | b | SE(b) |
| A. Minor-only Violence (N=3895) | ||||
| Intercept | -1.3310**** | -1.3568**** | .1399 | |
| Canada | .5432**** | 1.7215 | .0044 | .2457 |
| Female | .1225 | 1.1303 | .3063 | .1718 |
| Age 30-44 | -.7387**** | .4777 | -.6561**** | .1759 |
| Age 45+ | -1.7948**** | .1662 | -1.7684**** | .2237 |
| Canada � Female | .2151 | .2169 | ||
| Canada � Age 30-44 | .5713* | .2473 | ||
| Canada � Age 45+ | .6433* | .3024 | ||
| Female � Age 30-44 | .3716 | .2179 | ||
| Female � Age 45+ | 03234 | .2768 | ||
| Model chi-square | 226.948**** | 236.489**** | ||
| Degrees of Freedom | 4 | 9 | ||
| R2L | .069 | .072 | ||
| B. Severe Violence (N=4218) | ||||
| Intercept | -3.3140**** | -3.4381**** | .3055 | |
| Canada | 1.6796**** | 5.3634 | 1.1516*** | .3848 |
| Female | .7619**** | 2.1424 | 1.4394**** | .3268 |
| Age 30-44 | -.6332**** | .5309 | -.8741* | .3811 |
| Age 45+ | -1.5493*** | .2124 | -2.9137**** | .6120 |
| Canada : Female | -.6996* | .3624 | ||
| Canada : Age 30-44 | 1.8939**** | .3397 | ||
| Canada : Age 45+ | 2.6611**** | .5510 | ||
| Female : Age 30-44 | -.3973 | .3939 | ||
| Female : Age 45+ | -.2958 | .4895 | ||
| Model chi-square | 225.827**** | 270.502 | ||
| Degrees of Freedom | 4 | 9 | ||
| R | .128 | .154 | ||
Model 2 (Part A) introduces the first-order interaction terms to the equation for minor-only violence. Because being Canadian and being in the youngest age category increased the likelihood that minor-only violence occurred, it is possible that there exists an interaction or joint effect associated with country and age. As can be seen, the addition of the interaction variables at this stage makes a significant contribution to the proportion of variance explained ( p<= .001). The joint effects of being from Canada and in either of the two age categories are statistically significant at p<=.05. If we designate country as the focus variable and age as the moderator variable, the interaction term associated with being in the 30-44 age group and from Canada is .9187 [exp(-.6561 + .5713 * Canada) = .9187; see DeMaris, 1995]. That is, among Canadians the odds of minor-only violence occurring are .9187 as great for those in the middle-aged category, compared to those in the youngest age group. Americans in the 30-44 age group have odds of minor-only violence occurring that are only .5189 times those of Americans in the younger age group. The same pattern holds for both Canadian and American respondents in the 45+ age group.
Following Jaccard et al. (1990), we interpret the main effects in the interaction model as average effects of the independent variable on the dependent variable across values of tile moderator variable. In the present analysis, the very weak positive coefficient associated with being Canadian (.0044) suggests that Canadians, on average, are slightly more likely than their American counterparts to report committing acts of minor-only violence. This association is strongest, however, for the younger age group; hence the observed interaction effect. Although the increased likelihood of minor-only violence occurring in the younger age group is consistent with the subculture theory of violence, the greater probability that minor-only violence occurred in the Canadian sample was unexpected, given the higher U.S. level of societal aggression. Furthermore, the failure to establish statistically significant mean differences in minor-only violence between Canadian and American respondents across all age groups remains inconsistent with the culture of violence theory.
Severe Couple Violence
Part B of Table II presents the results of the estimation of two models for severe couple violence. The model chi-square test (p<=.001) indicates that Model I adds a significant proportion to the variation explained over a model of complete independence, suggesting that at least one of the added factors is important. Tests for the additional coefficients revealed that the added significance can be attributed to the influence of the three independent variables (all p <= .001). When gender and age were controlled, being from Canada as opposed to being from the United Stales increased the odds of committing severe couple violence by a factor of 5.3634, or 436%. Being female increased the odds of severe violence by a factor of 2.1424, or 114%, while being in the 30-44 rather than in the 18-29 age group decreased the odds of severe violence by a factor of .5309, or 47%. While negative coefficients associated with the two older age groups and committing severe couple violence are consistent with the subculture of violence theory, the positive coefficients associated with being Canadian and being female clearly are not.
Part B, Model 2 shows the coefficients after the first-order interaction terms were added to the equation. The chi-square test for additional proportion of variance explained is once again statistically significant, suggesting that the interaction model is a good fit to the cross-national data. While there is no statistically significant interaction effect between gender and age, gender does interact with country (p<=.05). Using country as the focus variable and gender as the moderator variable, the odds of committing severe violence among Canadian females are about twice (2.0955) those of males. Among Americans, the odds of severe violence are about four times (4.2182) as great for females, as opposed to males.
As with minor-only violence, country and age also have a statistically significant joint effect on the odds of severe violence. Americans in the 30-44 age group have odds of severe violence occurring that are only .4172 times those of Americans in the younger age group. The same pattern holds for the older Americans (.0543). Among Canadians, the odds of severe violence occurring are twice (2.7726) as great for those in the middle age group as to those 18 to 29 years of age, whereas those over 45 years of age are less likely to commit severe violence (.7768). As the interpretation of the main effects in the interaction model (below) reveals, the relationship between age and severe violence is in the predicted direction.
When the interaction terms were added, the main effects of country, gender and age remained statistically significant (p<=.05). On average, being Canadian, female and young, increased the odds of committing severe couple violence relative to being American, male and older, respectively. In sum, our multivariate analyses offered little support for the culture of violence theory. Only the inverse relationship between age and common couple violence remained consistent with that theory.
DISCUSSION
In this study of common couple violence in Canada and the United States we examined the incidence rates of minor and severe physically violent acts, as reported by national random samples of legally married or cohabiting women and men, age 18 and older. Lack of comparative data did not allow us to examine the context, motives, and meanings of these abusive acts, nor to explore the personal injury and other harm that they may have caused. Our analysis was guided by Johnson's (1995) distinction of "common couple violence" and "patriarchal terrorism" as well as by several propositions derived from the culture of violence theory. Four sets of findings merit review and discussion here.
Country-to-Country Comparison
First and foremost, based on our bivariate and multivariate analyses, findings of this cross-national study challenge the culture-of-violence thesis on several grounds. Incidence rates of intimate violence were found to be significantly higher in Canada than in the United States, although the latter society is more violent according to the serious crime rates. The higher Canadian rates held whether violence was measured by incidence rates for individual acts, for overall violence, or for severe violence. Furthermore, violence was more pervasive among Canadian than among American couples. "Minor" acts of violence were distributed more evenly in the two countries, an indication that this type of common couple violence is an integral and "normal" element of couples' interaction in a fairly large segment of both populations. This finding is consistent with Johnson's contention that this type of violence is less a product of patriarchy, and more a product of the less gendered processes of couple interactions that rarely escalates into serious forms of violence. Canadian women and men, however, were significantly more likely than their American counterparts to use severe violence and to inflict it, as well as minor violence, more often. The latter findings point to the possibility that the CTS in sample surveys do uncover some forms of "patriarchal terrorism," a notion that Johnson does not seem to entertain given the relatively low rates of severe violence reported in both the 1975 and 1985 American surveys. The Canadian data show clearly, however, that common couple violence can include a fairly large proportion of continuous serious physical abuse that may well overlap with patriarchal terrorism among male perpetrators. Recall that this pattern prevailed in one of ten Canadian couples and was reportedly perpetrated by both men and women. This finding among female perpetrators, however, is not consistent with patriarchal terrorism. Unfortunately, these cross-sectional data did not allow us to test Johnson's escalation thesis.
We can only speculate on the reasons for these results. We suggest several plausible explanations that merit exploration in future cross-national research. We begin with issues that may be related to methodological inconsistencies anti conclude with rival explanations that may be based on cultural differences.
First, methodological differences may account for some of the difference in rates between the two countries. The Canadian data were gathered through a self-administered questionnaire, with no third party present; the American data were gathered through a telephone interview. No third party was at hand during the Americans' interviews, but such a person was present at the other end of the telephone line. Although telephone interviews afford more anonymity than face-to-face interviews (Smith, 1989), the self-administered questionnaire is the most anonymous data collection instrument. Research suggests that people tend to disclose themselves more fully on sensitive issues as anonymity increases (Smith, 1993). Nevertheless, the Canadian incidence rate of overall husband-to-wife violence (18.3) reported in this study closely approximates a comparable national rate of 16.7, which is based on telephone interviews with 12,300 women, aged 18 and over, conducted by Statistics Canada during the early months of 1993 (Statistics Canada, 1993). Similarly, a 1987 telephone survey on the physical abuse of a representative sample of 604 Toronto women by a male intimate partner produced an annual incidence rate of 14.4%, based on the Conflict Tactics Scales (Smith, 1989). A 1986 Alberta telephone survey, based on the CTS, established an annual incidence rate of 11.2% for husband-to-wife physical abuse and a rate of 12.4% for wife-to-husband abuse (Kennedy and Dutton, 1989). In the absence of evidence from a controlled field experiment specifically designed to compare the two methods, it seems warranted to conclude that the incidence rates derived from the 1986 Canadian national study are not to be dismissed on strictly methodological grounds. Canadian studies based on telephone interviews have found incidence rates that are either equal to or higher than the 1985 U.S. national study; no study has produced a comparatively lower Canadian rate of spouse abuse.
Second, the possibility of greater underreporting by the American sample warrants investigation. Possibly Americans have become desensitized to violence because it occurs more often in their everyday life. Thus, acts that Canadians would interpret (and report) as violent may be overlooked by Americans, who may perceive them merely as minor annoyances. This rival explanation, while concerned with a methodological issue, also embraces an element of cultural difference, as does the following speculation.
Third, social desirability may work differently in the two societies. Americans may place a higher value on being perceived as successfully married, may perceive domestic violence more negatively, and hence may be less willing to disclose it.
Fourth, because of the wider availability of handguns and the wider exposure to societal-level violence in the United States, American men and women may feel more justified in venting their frustration and stress outside the family. In contrast, the more stringent laws surrounding the possession of weapons in Canada, as well as the lesser exposure to societal-level violence, may result in Canadian women and men bringing home the frustrations and stress of the work world.
Finally, it is possible that Canadian couples are more insular than their American counterparts. As Lipset (1985) argues, Canadians are more elitist, more ascriptive, and less individualistic than Americans. Does the Canadians' less individualistic orientation manifest itself in greater demands on their intimate partners? Because individualism is highly valued among Americans, American women and men may derive more self-esteem from their activities outside the intimate relationship. Canadians may turn to their intimate partners for much of the emotional and social support that the Americans derive from outsiders; thus Canadian marriages may be more prone to conflict and violence. This notion becomes even more plausible if one considers the greater variety of community, state, and federal prevention programs and services that are available to individuals as well as couples in the United States because of its much larger population. The Americans' higher divorce rate and their much shorter median interval between the first marriage and divorce reflect the greater value placed on individualism in that country.
Gender
The second important finding pertains to the patterns of gender symmetry in common couple violence in both countries. Because the Conflict Tactics Scales were not developed to measure the context in which intimate violence arises and because of the research revealing differences in patterns of underreporting by gender, caution is needed in interpreting the difference in rates by gender. Nevertheless, because female respondents are reporting on their own behavior, one must strive to answer this question: "What element in the intimate relationship causes women to resort to physical violence, given that they have not been socialized to do so?" Our failure to find any significant difference in the minor-only violence rates by gender, regardless of country, suggests that this type of violence is widely accepted as normative in Canada and the United States. Men and women in both countries are equally likely to resort to pushing or slapping their intimate partners, as our bivariate and multivariate analyses revealed.
The significant gender difference in severe violence rates (Table II) among both Canadians and Americans is more difficult to interpret. Without knowing the context, motives, or consequences of the violent acts, one cannot state unequivocally that female violence is identical to male violence. Qualitative research suggests that the physical and psychological consequences of intimate violence are more likely to be greater for female victims. These U.S. studies also report that female perpetrators tend more strongly to act in self-defense or retaliation (Saunders, 1986). However, Sommer's (1994, p. 169) Canadian data show that for the vast majority of women and men, the perpetration of partner abuse was influenced not by self-defense but by other factors (cited in Fekete, 1994, p. 90). Studies on underreporting of sensitive issues by gender reveal that males tend more strongly to underreport, not only in terms of frequency but also in terms of severity. While women may inflict severe violence more often than men, the greater average size and strength of men means that their punches and beatings will produce greater pain, injury, and physical harm (Gelles and Straus, 1988).
Consistent with Steinmetz's findings, Canadian wives are three times more likely than their American counterparts to commit severe violence against their male partners. In the absence of longitudinal and qualitative data that detail the dynamic interpersonal processes of escalating coercive cycles of intimate violence, the high rate of severe violence reported by Canadian women is difficult to explain. Are we observing "victim-precipitated patterns" of counter violence among Canadian wives? Is it possible that Canadian wives perceive a lack of control over their marital relationship and respond with verbal and physical attacks to regain some control? Do men's and women's strategies for control differ? The relationship between gender and control needs further investigation, as Stets (1995) has pointed out recently. Although the frequencies with which these various severe violent acts were committed by Canadian women and men are relatively low compared to those that Johnson (1995) cites from studies based on shelter populations, the magnitude of these rates cannot be ignored.
Age
The third set of findings pertain to the perpetrator's age which was found to be a useful predictor of common couple violence, particularly for the Canadian sample, as the multivariate analyses showed. The inverse relationship between age and common couple violence is consistent with the subculture of violence theory and with the criminological literature on serious crime at the societal level. It persists across the life course for both husband-to-wife and wife-to-husband violence, minor and severe, in both countries. A separate analysis by length of marriage produced a similar consistent decline over the life course (data not shown).
High rates of both minor and severe violence, as reported by younger Canadian and American women and men, are staggering. Most of these younger respondents in both countries are parents, and interparental violence has negative consequences for their children when they reach adulthood (Carrol, 1987). Because social learning in young children depends extensively on the behavioral and symbolic modelling that pervades their daily lives, the family of origin serves as a "training ground" (Straus, 1991) for teaching interpersonal norms that condone verbally and physically assaultive behavior. In this fundamental sense, primary socialization in the family can mediate a violent world, even though not all children who witness violence in the home become abusive adults. Nevertheless, the long-term personal and social implications of these cross-national findings regarding intimate violence among younger couples must not be underestimated.
Violence peaks among the youngest adults and declines throughout the life course, but it remains relatively high (especially minor violence) in the older age groups. More important, the rates remain greater than zero (2 to 5%) among respondents age 65 or older in both countries (data not shown). This finding supports the notion that minor violence persists over the life course because it becomes an integral part of everyday interaction for some couples. Although it is difficult to explore this idea with cross-sectional data, one might assume that these couples redefine their everyday actions in order to make sense of their violent reality, and thereby reinforce it. Thus much elder abuse is actually spousal abuse grown old, as Pillemer and Finkelhor (1988) suggest. It would appear that the phenomenological perspective could be useful in the study of common couple violence at the micro level because it focuses sharply on meanings and understandings as critical components of social action.
Theoretical and Social Implications
Our fourth set of cross-national findings relates to their theoretical and social significance. Johnson's cogent conceptual distinction between common couple violence and patriarchal terrorism provides a valuable theoretical direction for future research. His insistence on distinguishing between violence data that stem from large random surveys and those that are derived from shelter populations, police records, social service agencies, hospital emergencies, and similar settings that deal directly with affected victims contributes to a better understanding of the controversy surrounding the issue of gender symmetry in common couple violence. Undoubtedly, the cross-national data that we have identified as minor-only violence fit Johnson's description of common couple violence in both Canada and the United States. This form of violence may be conceptualized as "essentially normative in many married [Canadian and American] couples because of the high frequency of these behaviors in representative samples" (O'Leary, 1993, p. 17).
However, the data on severe violence, particularly those for Canada, are more difficult to interpret within Johnson's definition of common couple violence. To be sure, we do not want to imply that the Canadian results provide valid information regarding the prevalence or nature of patriarchal terrorism. What we want to argue is that these Canadian survey findings on patterns of severe intimate violence are less uniform and more complex than what Johnson subsumes under common couple violence. In other words, patterns of common couple severe violence uncovered in large-scale surveys embrace multiple forms of physical, verbal, and symbolic aggression that remain to be explained theoretically (Lupri et al., 1994). To lump together minor-only and severe violence into the category of common couple violence tends to trivialize the potentially devastating physical and psychological consequences that some of the life-threatening violent acts may have on the victims. Recall that severe violence prevailed in one of eight Canadian couples and that both women and men reportedly committed these acts. Persistence of some common couple violence across the life course represents a serious social problem in both countries and remains a challenge to family scholars interested in the development of theories that account for all forms of interpersonal violence.
Throughout our presentation and discussion of the cross-national data we have shown that the culture-of-violence theory cannot explain why the rates of common couple violence are higher among our sampled Canadian respondents than among their American counterparts. Contrary to the serious crime statistics at the societal level, females' incidence rates of intimate violence are equal to males' rates, or higher, in both countries. This finding holds across the life course and is inconsistent with the notion of a common etiology for all violence.
The culture-of-violence thesis fails because it neglects to take into account the violence-intimacy paradox--that is, certain structural conditions that set violence in the home apart from violence in the streets. The intimacy-violence paradox embraces such conditions. Bersani and Chen refer to this type of violence as "relationship-specific" situations (1988:69). Unlike perpetrators and victims of violence in the streets, perpetrators and victims in the home are intimate partners, as they are in spousal homicide. Furthermore, in many couples the intimate partners are both the perpetrators and the victims of violence. The etiology of such partner abuse is grounded in intimacy: both the forming and the breaking of the emotional bond are part of the ongoing dialectic of couple interaction because domination, control, and intimacy exist side by side and often become entangled. Perhaps the greatest challenge in understanding common couple violence and devising adequate intervention programs is to recognize that violence and intimacy can coexist in a relationship.
ACKNOWLEDGMENTS
This paper constitutes the revision of a chapter from the first author's doctoral dissertation, Intimate Violence. Asymmetry and Symmetry, Department of Sociology, University of Calgary, 1995. This paper uses data from the "National Family Resurvey" conducted by Richard J. Gelles and Murray A. Straus with funding from the National Institute of Mental Health, grant MH40027. The data set was obtained from the Inter-University Consortium for Political and Social Research. The Canadian National Family Life Survey was conducted by the second author with funding from the Social Sciences and Humanities Research Council of Canada, from the University of Calgary Research Committee, and from an anonymous donor. The authors wish to thank Merlin B. Brinkerhoff, Robert A. Stebbins, Donald L. Mills and anonymous reviewers for earlier comments and suggestions.
REFERENCES
Bersani, C. A., and Chen, Huey-Tsyh (1988). Sociological perspectives in family violence. In Van Hasselt, V. B., Morrison, R. L., Bellack, A. S., and Hersen, M. (eds.), Handbook of Family Violence Plenum Press, New York, pp. 51-86.
Bland, R., and Orne, H. (1988). Family violence and psychiatric disorder. Canad. J. Psychiatry 31: 129-137.
Brinkerhoff, M. B., and Lupri, E. (1988). Interspousal violence. Canad. J. Sociol. 13(4): 407-431.
Brinkerhoff M. B., Grandin, E., and Lupri, E. (1992). Religions involvement and spousal abuse: the Canadian case. J. Scientific Study Rel. 31 (March): 15-31.
Carrol, J. C. (1987). The intergenerational transmission of family violence: the long-term effects of aggressive behavior. Aggress. Behav. 3: 289-299.
Cronbach, L. J. (1967). Coefficient alpha and the internal structure of tests. In Mehrens, W. A., and Ebel, R. J. (eds.), Principles of Educational and Psychological Measurement, Rand McNally, Chicago, pp. 321-367.
DeMaris, A. (1995). A tutorial in logistic regression. J. Marr. Fam. 57(November): 956-968.
DeMaris, A. (1992). Logit Modelling, Practical Applications, Sage, London.
Dobash, R. P., Dobash, R. E., Wilson, M., and Daly, M. (1992). The myth of spousal symmetry in marital violence. Social Probl. 39: 401-421.
Fekete, J. (1994). Moral Panic: Biopolitics Rising (second edition), Robert Davies Publishing, Montreal-Toronto.
Feld, S. L., and Straus, M. A. (1992). Escalation and desistance from wife assault in marriage. In Stralls, M. A.. and Gelles, R. J. (eds.), Physical Violence in American Families, Transaction Publishers, New Brunswick, pp. 489-506.
Gelles, R. J. (1993). Through a sociological lens: Social structure and family violence. In Gelles, R. J., and Loseke, D. R. (eds.), Current Controversies on Family Violence, Sage, Newbury Park, CA, pp. 67-87.
Gelles, R. J., and Straus, M. A. (1979). Determinants of violence in the family: Toward a theoretical integration. In Burr, W. R., Hill, R., Nye, F. I., and Reiss, I. L. (eds.), Contemporary Therories about the Family (volume 1), Free Press, New York, pp. 549-581.
Gelles, R. J., and Straus, M. A. (1988). Intimate Violence, Simon and Schuster, New York.
Grandin, E. (1995). Intimate Violence: Asymmetry and Symmetry, Unpublished Doctoral Dissertation. The University of Calgary, Calgary, Alberta, Canada.
Grandin, E., Lupri, E., and Brinkerhoff, M. B. (1997). Couple violence and psychological distress. Canadian J. Public Health. 88(5).
Hagan, J. (1994). Crime and Disrepute, Pine Forge Press, London.
Hagan, J. (1991). The Disreputable Pleasures: Crime and Deviance in Canada, McGraw-Hill Ryerson, Toronto.
Hoffman, K. L., Demo, D. H., and Edwards, J. N. (1994). Physical wife abuse in a non-Western society: An integrated theoretical approach. J. Marr. Fam. 56(February): 131-146.
Jaccard. J., Robert, T., and K. Wan Choi. (1990). Interaction Effects in Multiple Regression, Sage, Newbury Park.
Johnson, M. P. (1995). Patriarchal terrorism and common couple violence: Two forms of violence against women. J. Marr. Fam. 57(May): 283-294.
Kennedy, L. W., and Dutton, D. G. (1989). The incidence of wife assault in Alberta. Canad. Behav. Sci. 21(1): 40-54.
Kumagai, F., and Straus, M. A. (1983). Conflict resolution tactics in Japan, India, and the USA. J. Comp. Fam. Studies XIV(3): 377-392.
Kurz, D. (1993). Physical assaults by husbands: a major social problem. In Gelles, R. J., and Loseke, D. R. (eds.), Current Controversies on Family Violence, Sage, Newbury Park, pp. 88-103.
Lee, G. R. (1987). Comparative perspectives. In Sussman M. B., and Steinmetz, S. K. (eds.), Handbook of Marriage and the Family, Plenum Press, New York, pp. 59-80.
Levinson, D. (1989). Family-Violence in Cross-cultural Perspective, Sage, Newbury Park.
Lipset, S. M. (1985). Canada and the United States: the cultural dimension. In Doran, C. F., and Segler, J. H. (eds.), Canada and the United Stares: Enduring Friendship, Persistent Stress, Prentice Hall, Englewood Cliffs, NJ, pp. 109-160.
Lupri, E. (1990). Harmonic und Aggression: Uber die dialektik ehelicher gewalt. (Harmony and aggression: on the dialectic of spousal violence.) Kölner Zeitschrift Soziol. Sozialpsychol. 42(3): 474-501.
Lupri, E. (1989). Male violence in the home. Canad. Social Trends 14(Autumn): 19-31.
Lupri, E., Grandin. E., and Brinkerhoff, M. B. (1994). Socioeconomic status and male violence in the Canadian home: A reexamination. Canad. J. Sociol. 19(1): 47-73.
Masamura, W. T. (1979). Wife abuse and other forms of aggression. Victimol. Int. J. 4(1): 46-59.
O'Leary, K. D. (1993). Through a psychological lens: personality traits, personality disorders, and levels of violence. In Gelles, R. J., and Loseke, D. R. (eds.), Current Controversies on Family Violence, Sage, Newbury Park, pp. 7-30.
Pillemer, K., and Finkelhor, D. (1988). The prevalence of elder abuse: a random sample survey. Gerontologist 28(1): 51-57.
Saunders, D. G. (1986). When battered women use violence: husband-abuse or self-defense? Viol. Vict. 1: 47-60.
Smith, M. D. (1993). Enhancing the quality of survey data on violence against women: a feminist approach. Unpublished manuscript.
Smith, M. D. (1989). Woman abuse: The case for surveys by telephone. J. Interpers. Viol. 4(3): 308-324.
Sommer, Reena. (1994). Male and Female Perpetrated Partner Abuse: Testing a Diathesis-Stress Model, University of Manitoba, Winnipeg: Unpublished Doctoral Thesis.
Statistics Canada. (1993). Violence against women survey. The Daily, Statistics Canada Cat. No. 11-001E, Ottawa.
Steinmetz, S. K. (1987). Family violence. In Sussman, M. B., and Steinmetz, S. K. (eds.), Handbook of Marriage and the Family, Plenum Press, New York, pp. 725-765.
Steinmetz, S. K. (1981). A cross-cultural comparison of marital abuse. Sociol. Social Welf. 8: 404-414.
Stets, J. E. (1995). Modelling control in relationships. J. Marr. Fam. 57(May): 489-501.
Stets, J. E. (1990). Verbal and physical aggression in marriage. J. Marr. Fam. 52: 501-514.
Stets, J. E., and Straus, M. A. (1992). Gender differences in reporting marital violence and its medical and psychological consequences. In Straus, M. A., and Gelles, R. J. (eds.), Physical Violence in American Families, Transaction Publishers, New Brunswick, pp. 151-166.
Straus, M. A. (1993). Physical assaults by wives--A major social problem. In Gelles, R. J.. and Loseke, D. R. (eds.), Current Controversies on Family Violence, Sage, Newbury Park, CA, pp. 67-87.
Straus, M. A. (1992a). The national family violence surveys. In Straus, M. A., and Gelles. R. J. (eds.), Physical Violence in American Families, Transaction Publishers, New Brunswick. pp. 3-16.
Straus, M. A. (1992b). Measuring intrafamily conflict and violence: The conflict tactics (CT) scales. In Straus, M. A., and Gelles, R. J. (eds.), Physical Violence in American Families, Transaction Publishers, New Brunswick, pp. 29-47.
Straus, M. A. (1991). Physical violence in American families: incidence rates, causes, and trends. In Knudson, D. A., and Miller, J. L. (eds.), Abused and Battered: Social and Legal Responses to Family Violence, Aldine, New York, pp. 17-34.
Straus, M. A. (1983). Ordinary violence, child abuse, and wife-beating. What do they have in common? In Finkelhor, D., Gelles, R. J., Hotaling, G. T., and Straus, M. A. (eds.), The Dark Side of Families. Current Family Violence Research, Sage, Beverly Hills, pp. 213-234.
Straus, M. A. (1979). Measuring intrafamily conflict and violence: the conflict tactics (CT) scales. J. Marr. Fam. 41(February): 75-88.
Straus, M. A., and Gelles, R. J. (1992a). Physical Violence in American Families, Transaction Publishers, New Brunswick.
Straus, M. A., and Gelles, R. J. (1992b). Social change and change in family violence from 1975 to 1985 as revealed by two national surveys. In Straus, M. A., and Gelles, R. J. (eds.), Physical Violence in American Families, Transaction Publishers, New Brunswick, pp. 113-132.
Straus, M. A., Gelles. R. J.. & Steinmetz, S. K. (1980). Behind Closed Doors: Violence in the American Family, Anchor/Doubleday, New York.
Straus, M. A., and Sweet, S. (1992). Verbal/symbolic aggression in couples: Incidence rates and relationships to personal characteristics. J. Marr. Fam. 54(May): 346-357.
Walker, L. E. (1984). The Battered Woman Syndrome, Springer, New York.
Wolfgang, M. E., and Ferracuti, F. (1967). The Subculture of Violence: Toward on Integrated Theory of Criminology, Travistock, London.
Zeller, R., and Carnlines, E. G. (1980). Measurement in the Social Sciences: The Link Between Theory and Data, Cambridge University Press, Cambridge.