Newspaper Coverage of Intimate Partner Violence: Skewing Representations of Risk

Abstract

How media portray intimate partner violence (IPV) has implications for public perceptions and social policy. Therefore, to better understand these portrayals, this study content analyzes a nationally representative sample of newspaper coverage of IPV over a two-year-period and compares this coverage to epidemiological data in order to examine the implications of the discrepancies between coverage and social reality. Stratified media outlets across the country were used to obtain a representative sample of daily newspapers based on their designated market areas, resulting in 395 IPV-related articles. Results show that newspaper framing of IPV tends to be heavily skewed toward episodic framing. In addition, there are significant differences between our data and epidemiological estimates, particularly in the coverage of homicide and use of alcohol and illegal drugs, which may skew public perceptions of risk. Implications for public perceptions and social policy are discussed.

How media portray issues and incidents concerning intimate partner violence (IPV) may well have important implications for public policy governing responses to IPV as well as, perhaps, beliefs and perceptions regarding norms of acceptable behavior within intimate relationships. Certainly, prior research suggests that media coverage of issues with health and legal implications can influence public policy response, and through such response can impact individual behavior (e.g., Sotirovic, 2003; Yanovitzky & Bennett, 1999). Moreover, Kozol (1995) argues that, by denying the normalcy of the act, media representations of IPV may allow the public to distance themselves from the issue while simultaneously reproducing “popular assumptions about public and private spheres as well as ideals of race and gender that are embedded in American national discourses” (pp. 648-9).

To better understand how media portray IPV and resultant implications for public perceptions and social policy, the study presented here content analyzes a nationally representative sample of newspaper coverage of IPV over a two year period and compares this coverage to epidemiological data, examining implications of discrepancies between coverage and social reality. Notably, the research presented here builds on previous content analyses in this area in several important ways. First, by using a nationally representative sample rather than a small number of regional newspapers, this study allows for more accurate comparisons to epidemiological data which has also been derived from nationally representative samples. Second, the coding scheme employed is more extensive, thus providing a more comprehensive picture of those involved in and the nature of the IPV incident itself, as well as more exhaustive categories analyzing the potential reasons behind the attack (including both victim-blaming and perpetrator-mitigating categories) and the possible framing devices indicating social or thematic coverage. Lastly, Thorson (2006) argues that crime and violence are most appropriately presented as public health issues and that presenting violence as a purely criminal issue may misrepresent health risks. Through the strengths highlighted in points one and two above, this content analysis is able to provide the critical first step needed to answer the call for research presented by Thorson; specifically, how portrayals of violence impact the health of individuals and communities and influence health policy. The findings from this study are more generalizable than in previous studies and can guide future research into the effects of newspaper coverage of IPV.

Intimate Partner Violence

The Centers for Disease Control and Prevention (CDC) defines IPV as “physical, sexual, or psychological harm by a current of former partner or spouse [which] can occur among heterosexual or same-sex couples and does not require sexual intimacy” (2006). IPV includes behaviors that range from a single occurrence to chronic abuse, with varying levels of severity (CDC). According to Tjaden and Thoennes (2000), approximately 5.3 million incidents of IPV occur in the U.S. each year among women 18 years of age and older. Among men, the occurrence is 3.2 million. Similarly, a national study found that 29% of women and 22% of men reported experiencing physical, sexual, or psychological IPV in their lifetime (Coker et al., 2002). IPV results in costs exceeding $5.8 billion each year, with the present value of expected lifetime earnings lost by IPV homicide victims at $892.7 million, which is an average of more than $713,000 per fatality (CDC, 2003).

The CDC has been a leader in establishing IPV as a public health issue, founding the Division of Violence Prevention within the National Center for Injury Prevention and Control in 1994 (Graffunder, Noonan, Cox, & Wheaton, 2004), whose foci includes “hidden health burdens” such as violence against women and other forms of violence that “occur in private domains and receive far less public attention” (p. 6). Moreover, the CDC has named identifying social norms that support IPV and ways to change them, along with studying the role of substance use as precursors to IPV perpetration as specific research priorities (Graffunder et al., 2004). Understanding how media portray IPV is an important component of identifying social norms surrounding IPV.

Though public perceptions have begun to shift in recent decades, IPV has historically been viewed as an individual or family problem (Carlson & Worden, 2005). IPV began being understood as a social problem with legal ramifications in the 1970s (Carlson & Worden, 2005), and progress has continued to be made as more forms of IPV, such as marital rape, have been recognized as illegal actions. Despite this progress, Worden and Carlson (2005) still found in their telephone survey of 200 randomly selected adults from each of six communities in New York that most people do not view IPV as an issue grounded in society or culture, but as one whose causes are contained within the context of individuals and relationships.

According to Gracia (2004), one of the factors that influences the rate of IPV is the lack of social responsibility attributed to the issue. Currently, there is a silence surrounding IPV that indicates social tolerance of the crime. Therefore, public opinion plays an important role in determining how IPV is understood and addressed. In fact, Taylor and Sorenson (2005) found people reported that the responsibility of finding a solution to an IPV incident should fall on the victim alone (31%) or both individuals involved (52%). When asked what should be done, only 12% mentioned engaging formal authorities. This indicates that there is a perception among people that IPV is an issue that victims should handle as a personal matter rather than something with which society as a whole also should be concerned. At a societal level, there could be increased support for hotlines, safe houses, and willingness to grant and enforce injunctions, as well as more encouragement for attitudes supportive of victims. Furthermore, better IPV coverage could improve the social climate so that people would begin to question the fundamental norms of male-female relationships which can contribute to IPV situations. From this perspective, by failing to put IPV incidents in the context of an overall problem, news media discard the “social roots of violence against women and absolves the larger society of any obligation to end it” (Carll, 2003, p. 1603).

Public Health Perspective

By taking a public health perspective, community-based support as opposed to individual responses are encouraged. Stalans and Lurigio (1995) describe the importance of public opinion in this endeavor as such: “Ordinary community members play an important role in how society as a whole and the criminal justice system handle domestic violence situations…. Crimes are actions that violate social norms about appropriate behavior. Criminal laws reflect social values and are sometimes created to reinforce them” (p. 399). By viewing IPV as a public health concern, society will be more likely to view this issue as a collective problem to be solved as opposed to specific instances that should be handled by the individuals involved (Hammond, Whitaker, Lutzker, Mercy, & Chin, 2006). Greater awareness and concern might, for example, lead to stronger laws and greater law enforcement.

Hammond et al. (2006) distinguish the role of public health from the role of the criminal justice system as one that focuses on the fundamental research questions surrounding violence and aggression including its causes and how to prevent it, not as the system that deals with the persons charged with violent behaviors. Dorfman, Woodruff, Chavez, and Wallack (1997) emphasize the importance of defining violence as a public health issue affecting society and not just a criminal issue affecting individuals because how the problem is defined determines the solutions available. For example, when an issue is presented as isolated cases of deviant behavior, then the interpretation might be that little can be done. On the other hand, if the issue is presented as a pervasive social problem, then resources might be directed toward ameliorating the problem through education, prevention, greater protection, and granting and enforcing protective injunctions. By reinforcing the idea that IPV is a societal public health issue, policymakers can take a proactive stance rather than the more common reactive response to IPV (Hammond et al., 2006).

Role of the Media in Perceptions of Social Risks

Given the ambiguous perceptions of a public health issue such as IPV, it is important to consider the socializing agents that help create the public’s perception of the issue. One potential source of information is media, especially since media influence the shaping of conceptions of reality by providing consistent social acceptance (Bandura, 1986). The extent of coverage of IPV as well as its framing can affect attributions of both blame for IPV incidents and of responsibility for addressing IPV; these attributions, in turn, can impact support for funding, safe houses, legal sanctions for perpetrators and protections for victims, and so forth. Given this potential impact, it is important to understand first, mechanisms through which media wields its influence, and second, how media portray IPV.

Media influence

Media exposure influences perceptions of social reality and social risks. Media effects also are mediated by other social agents (e.g., family, peers, and criminal justice system) who influence the social context in which individuals make decisions (Yanovitzky & Bennett, 1999). Yanovitzky and Bennett assert that this social context of formal and informal sanctions influence behavioral decisions in important ways. For example, the social acceptability of drunk driving has impacted the success of public health interventions aimed at curbing this behavior. Given this, Yanovitzky and Bennett argue that media “may be more effective in influencing behavior through their impact on social institutions,” and suggest that public health communication campaigns focus more on social institutions rather than at-risk individuals (p. 446).

Haider-Markel and Joslyn (2001) assert that media depictions of events can impact the causal reasoning employed by the public by emphasizing particular attitudinal or behavioral linkages of cause and effect. Therefore, coverage and portrayals of IPV in media are especially important because they help create the perceptions of risk and willingness to support an investment of social resources in prevention and intervention. As Combs and Slovic (1979) explain, if media representations are biased, then people’s perceptions of the issues also will be biased. Because news media both provide information about social problems and can imply various causes or consequences of events (Sotirovic, 2003), how issues are framed has significance for how people understand social phenomena.

Media framing

Gamson (1989) defines a frame as “a central organizing idea for making sense of relevant events and suggesting what is at issue” (p. 157). The efficacy of frames is found in their ability to make certain elements and perspectives more salient, thereby increasing the chances that certain schemas of interpretation will be evoked. Entman asserts, “The power of a frame can be as great as that of language itself” (p. 55). Frames exist in the properties of news narratives, thus encouraging certain interpretations and understandings of issues. It should be noted, as Sieffe (2003) explains, that it is the repetitiveness with which these frames are presented to the public that slowly shapes the way an issue is seen.

The more frequently an issue is framed in a particular way, the more likely it is for people to adopt media’s frame for it. For example, Sieff (2003) examined the negative frames that tend to pervade media coverage of mental illness, and argues that these negative frames help perpetuate negative attitudes towards those with mental illness, particularly because most people do not have direct experience with someone experiencing a mental illness and instead rely on media representations to form their opinions.

In addition, the framing of health issues that become legal issues also can have an impact on policy. In fact, as Iyengar and Simon (1993) explain, frames have the ability to change how individuals attribute responsibility of a problem. In other words, frames can determine if the public thinks about IPV as something the individuals involved in specific instances need to work through or if it is a larger issue that requires a community-wide effort to alleviate the problem. In their examination of news coverage of a secondhand smoke injury claim, Wakefield, Smith and Chapman (2005) argue that legal cases are especially important because they provide the opportunity for a social issue to be framed in a personalized way, with such a frame being more powerful than epidemiological or scientific frames.

One way to think about how issues are framed is in the context of episodic and thematic frames. According to Iyengar (1991), episodic framing tends to focus on the individual, whereas thematic focuses on society’s role in addition to the individual. Similarly, episodic frames tend to rely on individual explanations, whereas thematic rely on social explanations. Sotirovic (2003) describes individual explanations as those that focus on the personality, disposition, or motivational states of the people involved, whereas social explanations focus on circumstances and situational forces. This particular type of framing has important implications for how public health issues are perceived and can affect the type of attributions the public makes regarding events (Sotirovic, 2003). For example, Dorfman and colleagues (1997) examined youth and violence on local television news and found that episodic framing occurred with more than five times the frequency of thematic framing. The authors caution that if “news continues to report on violence primarily through crime stories isolated from their social context, the chance for widespread support for public health solutions to violence will be diminished” (p. 1311).

In the study presented here, episodic frames are defined as those in which a specific instance of IPV is portrayed, while a thematic frame includes a social context for the IPV incident. Examples of ways in which a thematic frame can be constructed are the inclusion of information about victimization rates, elements of the criminal justice process, government response to IPV, or a public health perspective addressing prevention or the health consequences of IPV. By expanding previously used conceptualizations of social coverage (e.g., Maxwell et al., 2000), this study allows for a better examination of the hypothesis put forth by McManus and Dorfman (2003); specifically, that as views of IPV shift from being a personal matter to a public matter, the nature of newspaper coverage will also evolve to include information from a public health perspective.

Media portrayals of IPV

Having addressed the mechanisms through which media can influence individuals, we now examine how the media portrays IPV specifically. As discussed earlier, media has the ability to determine how an issue is perceived by the public. Maxwell, Huxford, Borum, and Hornik (2000) argue that media coverage plays a significant role in the varying perceptions of IPV as a social issue over time. Carll (2003) argues that media influences and reinforces the stereotypes of women as the victims of violence. What is more, these stereotypes become part of everyday life as well as integral in the judicial system. Because media and news in particular play such a role in shaping the attitudes of society, distortions and misrepresentations mislead the public.

How then is IPV portrayed by media? In their study of media coverage of IPV, Maxwell and colleagues (2000) found that most articles were framed in a way that left the victim responsible for ending the violence in the relationship. Moreover, the authors found that social factors perpetuating violence were largely ignored. Even after the much publicized OJ Simpson case, whose prolonged coverage created opportunities for media to frame the story is a variety of ways, reports of IPV remained largely based on individual incidents. This may lead to societal attitudes that are victim-blaming and unsupportive of intervention from social structures, despite the need for victims of crime to be protected. In fact, Maxwell et al. (2000) assert that by covering specific incidents of IPV, media moves the responsibility to solve the problem from society to the individual.

A more recent study of newspaper coverage of IPV in Washington State found similar results (Bullock & Cubert, 2002). In their analyses of articles, the authors found an overwhelming percentage of the articles were single incident reports which did not leave room for a discussion of the problem on a larger scale. However, they did find a handful of articles which did put the incident into a frame that looks at the issue on a societal level. The authors report that this is an indication that journalists have the ability to frame IPV incidents from a societal perspective and still remain within the current reporting norms. To this end, McManus and Dorfman (2003) have recommended that journalists include risk factors for IPV, point readers toward local resources, avoid myths and stereotypes, and apply the research that exists on IPV in their articles.

Interconnected with many of the stereotypes that pervade stories about IPV is the issue of gender. In her critique of various forms of media, Kozol (1995) argues that “representational strategies used by the mainstream media protect male power from the threat posed by the exposure of these crimes,” with portrayals typically “focused on the women involved, either blaming them for the abuse or championing them as lone heroines fighting lone villains” (p. 648). Kozol argues further that media representations “often reinforce national ideologies through their focus on individualism and free will … attacking the specific problem but not larger structural forces ignores crucial factors of race, class, and sexuality in ways that reproduce, rather than question, American national ideals” (p. 665). Furthermore, Meyers (1994) claims that the coverage of IPV serves to perpetuate the past ideology that men should dominate women. Although these studies make a valuable contribution to the literature, there have not been any studies that systematically explore the nature of newspapers coverage of IPV with an extensive nationally representative sample. Given this, we propose the following research question:

RQ1: A) Is this nationally representative sample consistent with regional findings of episodic being more dominant than thematic coverage? B) How large is the differential between episodic and thematic coverage? C) With respect to episodic coverage, what are the characteristics of incidents, perpetrators, and victims that are reported? D) What are the predominant news frames for thematic coverage?

Not only are we interested in how media portray IPV, but we are also interested in how these portrayals compare to epidemiological data. As discussed above, how media portray IPV has important implications. For example, if emotional and verbal forms of IPV are underrepresented, this may lead the public to believe that these forms of violence are not included in the definition of IPV. This, in turn, may lead to a lack of awareness of resources available for victims of these forms of violence, as well as a general lack of public support for legal protections from non-physically violent perpetrators. What is more, as explained in Combs and Slovic (1979), by neglecting to report an issue, like IPV, newspapers fail to bring attention to an issue that needs societal action in order to help reduce the potential problem. Without coverage of these problems, society may not understand the potential risk. Skewed media representations may also inadvertently discourage IPV victims from seeking help. For example, over-representation of homicide may discourage victims from leaving abusive relationships because it reinforces that leaving is when they are most likely to be killed. Continually presenting stories where IPV victims are killed without placing the act in a social context emphasizing protections and resources available could be particularly problematic from a public health perspective. As such, we propose the following research question:

RQ2: How does newspaper coverage of IPV compare to epidemiological data?

In addition, we are interested in several specific contextual features of portrayals of IPV discussed next.

Contextual features

The characteristics of the IPV incident which newspapers choose to include, such as the relationship between the perpetrator and victim and whether there is a history of IPV in the relationship, can have important implications. As Sotirovic (2003) explains, the amount of contextual information available is vital to individuals’ understanding of social issues, without which the ability to develop alternative explanations for events may be restricted. Several studies have examined the impact of contextual features of IPV incidents on support for different responses by the criminal justice system. The response of the criminal justice system is important because it impacts perceptions of social norms (Salazar, Baker, Price, & Carlin, 2003). Hilton (1993) found that the severity of the IPV injury and whether it was a first offense influenced public views about whether the criminal justice system should intervene. Notably, those incidents portrayed as first offenses were viewed as less severe and less likely to occur again. Similarly, Stalans and Lurigio (1995) found that the public preferred dismissal of cases where the victim did not receive an injury in the current dispute. This perspective underscores the dominant view of IPV as physical violence and minimizes the severity of the consequences of psychological, emotional, and economic forms of abuse. Given that physical forms of IPV tend to be the only ones that conform to journalistic ideals of “newsworthiness”, we advance the following hypothesis:

H1: Rates of news coverage of IPV as compared to coverage of other violent acts in our sample will be less than the equivalent comparison using epidemiological data.

Another contextual feature of particular interest is substance use. There is general agreement that alcohol plays a significant role in various forms of violence, including IPV (Leonard, 2005). However, there is not widespread agreement about the specific role that alcohol plays or the nature of appropriate causal claims. In his editorial summary of the literature, Leonard found some studies argue that alcohol is used to excuse the violent behavior (Zubretsky & Digirolamo), that alcohol consumption increases the risk for IPV (Lipsey, Wilson, Cohen, & Derzon, 1997), and/or that alcohol may increase the severity of the IPV incident (Fals-Stewart, Leonard, & Birchler, in press). Moreover, previous research indicates that coverage of alcohol’s role in violent crime is under-reported relative to epidemiological data (Slater, Long, & Ford, in press). Given the ambiguity with which the role of alcohol can be interpreted, how media frames this contextual feature can have important implications. To examine this, the following research hypothesis is proposed:

H2: The rate at which alcohol is acknowledged as a contributing factor in news coverage will be less than epidemiological estimates of rates of IPV in which alcohol use is a contributing factor.

Method

Sample

Stratified media outlets across the country were used to obtain a representative sample of daily newspapers based on their designated market areas (DMA^®). According to Standard Rate and Data Service (2000), this is the most widely used approach to defining electronic media markets. There are 210 DMAs^® covering the continental United States, Hawaii, and parts of Alaska (Nielsen Media), which are particularly useful as sampling units because they define markets at the local level, thus providing a way for researchers to select both local TV programs and daily newspapers from the same geographic area (Long, Slater, Boiarsky, Stapel, & Keefe, 2005).

News outlets were sampled such that their probability of inclusion in the sample was approximately proportional to their circulation or reach; specifically, we divided the country’s DMAs^® into six strata, with each stratum consisting of approximately 1/6^th of all U.S. households). The sample was restricted to six strata because with six strata, the stratum for the largest markets includes one from the West Coast (Los Angeles), one from the Midwest (Chicago), and one from the East Coast (New York City). This approach permitted reasonable regional representation in all strata and assured reasonable homogeneity of market size within each of the strata (Long, et al., 2005).

Researchers created one constructed month each from 2002 and 2003 to avoid the problems associated with using random samples of media content (Riffe, Aust, & Lacy, 1993; Riffe, Lacy, Nagovan, & Burkum, 1996; Riffe, Lacy, & Fico, 1998). In addition, each year’s constructed month consisted of one constructed week from each season to address any seasonal variation in media coverage. On each day of the constructed month in 2002, one DMA^® was randomly selected from each of the six strata. The same DMAs^® in 2003 as in 2002 were used because of the large effort involved in constructing the list of daily newspapers in any given DMA^®. Regardless of the use of the same DMAs^® in both years, the individual media outlets in a selected DMA^® were randomly chosen each year.

IPV Story Selection

From this sample, all articles were coded for whether the story contained mention of domestic violence as part of a larger study (see Slater, Long, & Ford, in press for a complete description). For this study, this designation was further refined to include only those articles that mentioned intimate partner violence specifically (i.e., does not include stories about violence between family members other than intimate partners) for a final sample size of 395 articles.

Story-Level Coding

For each story selected, the following variables were coded: primary news frame (episodic vs. thematic), specific types of thematic framing, potential reason behind attack (e.g., victim infidelity, argument, money), types of IPV (e.g., physical, sexual, emotional, verbal, economic), where the IPV incident occurred, whether the perpetrator was arrested and the outcome (e.g., charged with a crime, convicted), history of IPV within the relationship, perpetrator and victim demographics, relationship between perpetrator and victim, criminal and personal history information about perpetrator and victim, and the involvement of alcohol or drugs.

Two coders were trained to identify the elements described above. Intercoder reliability was assessed on 20% of the sample with kappas ranging from . 100–1.00. The average kappa across categories was .772, with only seven categories having a kappa below .500. Kappas were low or could not be computed for some categories because of the small number of times a category appeared in the dataset (e.g., reasons for staying in the relationship, specific types of emotional abuse). Landis and Koch (1977) state that kappas in the .41 to .60 range indicate moderate agreement, kappas in the .61 to .80 range indicate substantial agreement, and kappas above .80 indicate almost perfect agreement.

Epidemiological Comparisons

To analyze the research question and hypotheses requiring epidemiological comparisons, data are taken from the current statistics provided by the CDC whose sources include the Bureau of Justice Statistics National Crime Victimization Survey (NCVS), the National Violence Against Women (NVAW) survey, and the FBI's Uniform Crime Reporting program (UCR). More specifically, Rennison and Welchan (2000) and Rennison (2003) use data from the NCVS and the FBI's Supplementary Homicide Reports. NCVS data are from a nationally representative sample of 77,200 households comprising nearly 134,000 persons in the United States. Tjaden and Thoennes (2000) use data from the NVAW, whose sample was drawn by random-digit dialing from households with a telephone in the 50 States and the District of Columbia for a total of 8,000 women and 8,005 men age 18 and older. Fox and Zawitz (2004) and Greenfeld (1998) both use data from the UCR whose data comes directly from reports that law enforcement agencies submit information to the FBI monthly. Additional details about these data sources can be found in the articles cited.

Results

Research Question One

Research question one asked several broad questions about how newspapers portray IPV. Of principal interest was the type of primary frame employed, whether it was consistent with previous regionally-based research suggesting more episodic than thematic framing, and the magnitude of this difference. Results were consistent with previous research (e.g., Maxwell et al., 2000), with episodic coverage being dominant. Overall, episodic framing was used in 88.3% of the articles, while thematic was used for 11.7%. Among articles that had episodic as their primary frame, 19.9% also contained at least one element of thematic framing.

Research question one also asked what are the predominant news frames for thematic coverage? Of the articles containing any thematic framing elements (a total of 29.6 % of articles mentioning IPV), the most common (present in 42.5%, or 7.8% of the whole sample) thematic frame addressed government responses to IPV including policies and laws. A public health perspective such as prevention or social and psychological impact was included in 39.7% (7.3% of the whole sample); whereas 31.5% (5.8% of the whole sample) addressed community response and 28.8% (5.3% of the whole sample) provided information on IPV resources.

In addition to framing, research question one was also interested in the characteristics of the incidents reported because things such as whether a reason or justification is offered for the IPV incident reported can be an important priming device. The most common reasons behind the IPV incident offered and the specific type of IPV reported are presented in Table 1. The primary victim was killed in 56.7% of the stories and at least one person was killed in 58.5%. The majority of articles (91.6%) did not give an indication of a history of IPV in the relationship.

Table 1.

Select Contextual Characteristics

	Percent of total articles (n = 395)
Reason behind the IPV incident*
Argument	13.2%
Money	8.9%
Victim infidelity	6.9%
Alcohol use	4.9%
New romantic interest (perpetrator)	4.6%
Stress	3.7%
Emotional duress	3.4%
Anger	2.9%
Jealousy	2.0%
Drug use	1.7%
Specific type of IPV**
Physical	96.1%
Verbal	7.6%
Emotional	1.8%
Sexual	1.5%

^*Columns do not sum to 100% because any given article may contain one or more reason or none at all.

^**Columns do not sum to 100% because any given article may contain reports of more than one type of IPV.

Lastly, contextual features addressing the personal behaviors of both the perpetrator and victim were of interest in research question one. Characteristics of the perpetrators and victims, including sex, social status, and personal history are presented in Table 2. Notably, no same-sex incidences were reported. Race was unknown in the majority of cases for both perpetrator (78.5%) and victims (92.7%). The relationship between the perpetrator and victim was spouse in 62.0%, separated 14.6%, dating 13.3%, cohabiting 3.9%, and engaged 2.7%. The perpetrator’s picture was included in 21.7% of the articles, while only 6.6% contained a picture of the victim. Only three articles contained a picture of the victim when the victim was not killed in the incident. During the IPV incident, the perpetrator was under the influence of alcohol in 6.0% of the stories and under the influence of drugs in 2.1%. For victims, 1.3% were under the influence of alcohol and 0.3% under the influence of drugs. However, 9.9% of stories mentioned a history of alcohol use for perpetrators and 2.2% for victims. Finally, in fewer than 3% of all articles did the victim rationalize the incident itself or being/staying in the relationship, or did anyone else question why the victim was in the relationship.

Table 2.

Perpetrator and Victim Characteristics

	Perpetrator	Victim
Male	80.7%	16.2%
Female	19.3%	83.8%
Social status
Sports figures	8.4%	3.2%
Public figures	18.1%	3.5%
Locally known	0.9%	0.9%
Not famous	72.6%	92.4%
Personal history information*
Alcohol use	9.9%	2.2%
Drug use	3.6%	1.9%
Criminal history	9.3%	3.2%
Infidelity	9.0%	7.9%
Sexual activity	3.6%	1.9%
Violent behavior	7.2%	0.9%
Mental illness	2.1%	0.3%
Socially Deviant	4.5%	0.3%

^*Columns do not sum to 100% because any given article may contain one or more pieces of personal information or none at all.

Research Question Two

Research question two asked how newspaper coverage of IPV compares to epidemiological data on its prevalence in the United States. Compared to all violence stories in the national sample of newspaper articles, 8.2% of stories were related to IPV. Between 1993 and 1998, IPV accounted for 22% of violent crime against women and 3% of the violence against men (Rennison & Welchan, 2000). Most IPV physical assaults are relatively minor (Tjaden & Thoennes, 2000); however, in this sample 13.6% of physical assault were minor (e.g., slap, hit, shove), while 47.3% were severe (e.g., stab, shoot, strangle), though not all articles specified the specific type of physical violence. The majority of victims (83.8%) were females in our content analysis, similar to 85% in the general population (Rennison, 2003). Of the victims who were killed in our sample, 85.8% were female and 14.2% male. Nationally, the proportion is 76% female and 24% male (Fox & Zawitz, 2004), meaning that male homicide is significantly underreported in our sample (z = −3.42, p < .01).

Hypothesis one

We proposed that news coverage of IPV compared to coverage of other violent acts will be less than the rates of IPV to other types of assaults as estimated epidemiologically. This hypothesis received partial support. The IPV victim was killed in 56.7% of stories in the sample, which is much higher than CDC (2003) estimates of approximately 1,300 deaths per year. However, deaths in IPV stories accounted for 9.4% of all homicides in the national sample, which is similar (z = 0.96, ns) to the 11% of homicide victims that are killed by an intimate partner according to U.S. Department of Justice homicide trends data collected from 1976 to 2002 (Fox & Zawitz, 2004).

Hypothesis two

We proposed that the rate at which alcohol is acknowledged as a contributing factor will be less than epidemiological estimates of rates of IPV in which alcohol use is a contributing factor. This hypothesis was supported. Specifically, we found that the perpetrator was under the influence of alcohol in 6.0% of the stories and under the influence of drugs in 2.1%. The U.S. Department of Justice (Greenfeld, 1998) reports that perpetrators were under the influence of alcohol in 55% of all violent crimes committed between intimates, under the influence of drugs in 9%, and alcohol or drugs in 12%. This means that the influence of both alcohol (z = 18.46, p < .001) and drug (z = 4.52, p < .001) use are significantly underreported in our sample. For violent victimizations occurring between spouses, 65% involved only alcohol, 5% involved only drugs, and 11% involved both (Greenfeld, 1998).

Discussion

We began this paper with the argument that how media portray a particular health issue has important implications for how it is perceived by the public as a public health concern, as well as perceptions of how society should respond. The analyses conducted here show that newspaper framing of IPV tends to be heavily skewed toward episodic framing which focuses on the individual and tends to ignore the larger social context within which IPV occurs. Our finding that episodic framing dominated coverage is consistent with previous research examining domestic violence which suggests that coverage focuses on the individual and tends to ignore social factors that help perpetuate violence (Maxwell, Huxford, Borum & Hornik, 2000). By portraying IPV as an individual or relationship issue without discussing the underlying causes of IPV, victims may be more likely to feel blamed for their own victimization. For example, some of the primary reasons why many cases of IPV do not get reported are the prevalence of beliefs such as privacy of the family, victim blaming attitudes, and the imbalance of power that exists between men and women in society (Gracia, 2006). Importantly, the more an individual feels held responsible for his or her own victimization, the less likely the individual is to seek help and the more likely it is that social tolerance for the action will be maintained (Gracia, 2006).

Moreover, the relative lack of coverage of resources available in the community may exacerbate perceptions of isolation and inability to receive help. Only 5.3% of the articles provided any information about IPV resources such as a hotline, website, shelters, or other forms of victim services such as advocacy and counseling. By omitting any mention of societal resources, especially ones that are available everywhere such as national hotlines, these stories leave the victim as the sole party responsible for finding a solution outside of the police arresting the perpetrator and decrease public knowledge about these important services. IPV resources are significant because they indicate that there is a widespread problem that requires intervention outside of the individual. By ignoring this perspective, media perpetuate the belief that IPV is an individual problem to be dealt with by the victim alone. Correspondingly, Sotirovic (2003) asserts, “People learn certain ways of explaining social events through their exposure to media. Excessive personalization of news diverts attention away from social causes” (p. 134). From a public health vantage point, an opportunity to increase utilization of protective resources and decrease physical risk is lost.

Not only was thematic not the dominant frame present in the majority of articles, only about 20.0% of the episodically framed articles contained any elements of thematic framing. The most common thematic frame involved government responses to IPV, including policies and laws. Although this was the most common thematic element, it was still present in only 7.8% of the total articles. The sparse attention given to this perspective is problematic for a variety of reasons. Predominantly, this indicates that IPV policy is not receiving much attention. The news media serve as an indicator of the public agenda, and by not covering IPV, this may indicate to policymakers that IPV is not an issue that it important to the public and should not receive much of their attention.

In fact, the limited coverage of IPV could have more imperative policy implications. Yanovitzky (2002) explains that media attention to issues is important in gaining the attention at a policy level because policy makers use the media to determine which issues the public cares about. If an issue like IPV is not receiving much attention, policymakers could infer that the subject is not a top priority for the public. If this is the case, then there is little encouragement for policymakers to begin tackling the issue. Therefore, in order for strides to be made at a policy level, it is important that media coverage reflects the public’s concern about the issue.

Epidemiological Comparisons

In some ways, newspaper coverage of IPV does not significantly differ from epidemiological reports. For example, when the number of homicides due to IPV was compared to the number of homicides present across all articles in the national sample, the percentage was similar to the national rate of homicides due to IPV. Conversely, when the number of stories involving an IPV victim being killed is compared to only the sample of IPV articles, the percentage of homicide victims due to IPV is over-represented. The CDC (2003) reports that IPV results in approximately 2 million injuries each year, but only 1,300 deaths; however, the victim was killed in 56.7% of IPV stories in this sample. Newspapers over-reporting incidents of IPV where the victim is killed is consistent with previous research stating that media coverage of homicide is disproportionate with its relative contribution as an overall cause of death (Frost, Frank, & Maibach, 1997).

While this type of coverage is understandable as homicide is more “newsworthy” than other forms of IPV, the extent of the discrepancy between newspaper coverage and epidemiological reports has important implications. Not only does this over-representation of the most severe form of IPV skew representation of risk, it also may intimidate other victims of IPV from leaving violent relationships because it reinforces that this is when the chances of being killed are the greatest. Indeed, 14.6% of the victims reported on in our sample were separated from their intimate partner when the attack occurred. Moreover, this type of over-representation strengthens the conceptualization of IPV as having a physical dimension only, and the effects of other forms of victimization go unacknowledged despite the severity of their effects.

Perhaps the most significant discrepancy between prevalence in our sample and epidemiological data is found in the area of substance use. According to the U.S. Department of Justice (1998), two-thirds of IPV victims report alcohol being a factor in the incident. Among IPV incidents involving spouses, the perpetrator was reported to have been drinking in 3 out of 4 cases. However, in our sample, we found that the perpetrator was reported as being under the influence of alcohol in only 6.0% of the stories and under the influence of drugs in only 2.1%. Although some argue that alcohol is used to excuse violent behavior (Zubretsky & Digirolamo, 1996), not presenting alcohol’s role accurately in the news media may have negative public health consequences. As discussed earlier, if a person is getting most, if not all, of their information about IPV from media, because of this lack of reporting, individuals might be led to the perception that alcohol or drug use is not related to IPV. On the contrary, alcohol consumption is an important risk factor for IPV (Lipsey et al., 1997) that may increase the severity of the IPV incident (Fals-Stewart, Leonard, & Birchler, in press); media ignoring this crucial factor hinders the efforts of public health agencies working toward understanding and eliminating the effects of alcohol on violent behavior.

Strengths and Limitations

Apart from its numerous strengths, there are limitations worth noting. First, the sample was created during the time that the Robert Blake arrest and trial was occurring. Articles with him as the alleged perpetrator comprised 10.9% of the total sample. While this may have skewed the sample, we do not believe this is a serious limitation. If anything, the Robert Blake trial increased coverage of IPV overall in the media; thus our findings can be interpreted as conservative estimates of coverage. Moreover, in their analysis of how the O.J. Simpson case shaped reporting of domestic violence, Maxwell et al. (2000) found that the nature of coverage was relatively unaffected by this high-profile case. A second limitation is the use of only newspaper articles. To speak definitively on news media portrayals of IPV, future research should also look at how televised news covers IPV. A third limitation relates to the epidemiological data used. Because these data come from government sources, the time frame from which the data were collected was not always within the same time frame from which our sample was drawn. However, we do not feel this had a substantial effect on our results because, first, the epidemiological data used tends to be relatively stable over the span of five to ten years, and second, most of our results were highly significant such that slight variations in the epidemiological data would be unlikely to change our results to the extent that they become non-significant. Moreover, the epidemiological data used is the most authoritative data that is currently available and the data upon which national policy is based. Finally, intercoder reliability for certain categories was lower than ideal, though imperfect reliability is unsurprising given the challenge of coding such latent content (Neuendorf, 2002).

Despite these limitations, this study represents one of the most comprehensive examinations of newspaper coverage of IPV to date, examining not only the framing devices used, but also the contextual characteristics of the IPV event and inclusive information about both perpetrator and victim. It is differentiated further from other content analyses by the nationally representative sample employed. In addition, this study distinguishes between domestic violence in general and IPV specifically, which makes the results more comparable to epidemiological data. Given the strength of the sample, these are the most authoritative results regarding newspaper coverage of IPV presently available.

Conclusion

The patterns found raise a variety of concerns, as noted. However, we can from these data only infer the possibility of negative effects of such coverage patterns. The content analysis presented here provides the foundation for several programmatic lines of research. For example, future research should examine the effects of the newspaper coverage described here on public attributions of responsibility, both for the incident occurring and for finding a solution, as well as on attitudes toward victims. Future research should also examine the effects such portrayals have on public support for a public health response to the IPV epidemic, including support for the funding and implementation of various prevention efforts and community-based supports for victims. Other possibilities include an examination of a possible contagion effect for those IPV victims contemplating leaving the abusive relationship or seeking help, or an examination of how coverage of issues with both crime and health implications, such as IPV, compares to coverage of issues that are perceived only in terms of health. Having the quality of data present in this study makes it possible for those researchers interested in the connections between media, health, and policy to pursue lines of research in this area that can produce actionable results which may very well lead to improvements in the health of society.