Abstract
The recurring, horrific deaths of minority residents at the hands of police officers and vigilantes have led social movements and international protests to amplify the charge that whereas the loss of White lives is seen as tragic, the loss of Black and Hispanic lives is treated as normal, acceptable, and even inevitable. Building on and advancing theories of “color-blind racism,” the authors examine the process by which the news media uphold and reify the devaluation of Black and Hispanic lives through ostensibly race-neutral language, story lines, and cultural narratives. Drawing on an original data set containing all news articles (n = 2,245) written about every homicide victim (n = 762) in Chicago, Illinois, during 2016, the authors use multilevel models to assess the extent to which victims’ race and neighborhood racial composition are associated with the level of attention, or “newsworthiness,” devoted to their deaths. Using two measures of newsworthiness—the amount of coverage and recognition of “complex personhood”—the authors find that victims killed in predominantly Black neighborhoods receive less news coverage than those killed in non-Hispanic White neighborhoods. Those killed in predominantly Black or Hispanic neighborhoods are also less likely to be discussed as multifaceted, complex people. Our analyses underscore the importance of place, especially the racialization of place, in determining which victims are treated as newsworthy. These findings carry important implications for understanding and addressing color-blind racism, news-reporting practices, and territorial stigma in the reproduction of racist ideologies.
In 2020, tens of millions of people across the globe took to the streets to protest the murders of George Floyd, Breonna Taylor, Ahmaud Arbury, and the mounting number of other minority residents killed at the hands of police officers and vigilantes (Brownstein 2020; Searcey and Zucchino 2020). International social movements, most notably Black Lives Matter, have renewed and amplified the charge that only some lives matter in society, whereas others are systematically devalued. Whereas the loss of White lives is seen as tragic, the loss of Black and Hispanic lives is viewed as normal, acceptable, even inevitable. Although the devaluation of Black and Hispanic lives frequently gains widespread attention as a result of high-profile events, these flashpoints are merely the tip of an iceberg of racial domination that remains hidden just under the surface. Indeed, as contemporary race scholars argue, overt violence is no longer the primary tool for maintaining racist systems. Rather, this task is increasingly accomplished through “color-blind racism,” which relies on coded, ostensibly race-neutral language, story lines, and cultural narratives that uphold and reify White supremacy as common sense (Bonilla-Silva 2014, 2015; Ray 2019; Roberts 2012; Wingfield and Feagin 2010). Addressing overt and violent acts of racism thus requires more systematic investigation and intervention into the color-blind racist institutions on which they ultimately rest.
In this study, we analyze the systemic and cultural devaluation of minority lives by measuring the importance the news media afford minority homicides victims compared with non-Hispanic White victims. We draw on an original data set of all news articles (n = 2,245) written about every named homicide victim in Chicago during 2016 (n = 762). Using multilevel models, we assess the extent to which victims’ race and the racial composition of the neighborhoods in which they were killed are associated with the level of attention, or “newsworthiness,” devoted to them. We use two measures of newsworthiness: the amount of coverage victims receive and whether coverage discusses victims through a lens of “complex personhood” (Gordon 2008; Wanzo 2006)—that is, as multidimensional selves with strengths, weaknesses, and/or resilience (Rios 2017; Stuart 2020). We find that victims killed in predominantly Black neighborhoods receive less news coverage than those killed in non-Hispanic White neighborhoods. Those killed in predominantly Black or Hispanic neighborhoods are also less likely to be discussed as multifaceted, complex people.
Our analysis builds on and advances scholarship on color-blind racism, the racialization of place, and racial inequities in news reporting by demonstrating that the relationship between race and newsworthiness is less straightforward than existing literature holds. Rather than existing primarily at the micro level (e.g., in individual interactions), we demonstrate that the devaluation of minority lives fatefully unfolds at the meso level as well, through the racialization and devaluation of minority neighborhoods (Bonilla-Silva 2015; Stuart 2016; Wacquant 2008). As Eduardo Bonilla-Silva (2014) importantly reminds us, racial domination is too often theorized and discussed as an interpersonal process whereby consequential social actors (e.g., news reporters, police officers, government officials) generate or reinforce the current and harmful normalization of whiteness in society as a result of their own biases toward racial minorities, which are often shaped by wider societal biases. On the basis of our results, we expand upon this theoretical work, arguing that such inequality may also be driven by a devaluation of individuals because of their geographic context, in particular the racial makeup of neighborhoods.
Theoretical Framework: The Devaluation of Black and Hispanic Lives through Color-Blind Racism and the Racialization of Place
Given the role of media as one of the most influential discursive systems driving color-blind racism, this study examines the subtle, often invisible ways that news reporting perpetuates the viewpoint that Black and Hispanic lives are less newsworthy, and thus less valuable, than White lives. Previous research defines newsworthiness as the value of an event, a crime, or a person in the eyes of news agencies, public audiences, and broader society (see Chermak 1998; Gruenewald, Pizarro, and Chermak 2009; Lundman 2003; Surette 1998). Although news media devote substantial coverage to homicides, not every victim is covered equally. Such differences in news coverage provide an opportunity to examine not only how race affects victims’ portrayed value but also how the racialization of place structures such portrayals.
Over the past two decades, critical race scholars demonstrated that a fuller understanding of contemporary racial domination requires moving beyond the long-standing focus on prejudice (i.e., discrimination by outwardly racist actors) to more fully examine “racial ideologies”—that is, the hegemonic, racially based frameworks that explain and justify the racial status quo (Bonilla-Silva 2014, 2015; Roberts 2012; Wingfield and Feagin 2010). Arguably the most powerful of such racial ideologies is “color-blind racism” (Bonilla-Silva 2014, 2015). Whereas racism in the Jim Crow era was enforced through coercion and violence, today’s color-blind racism operates through more covert, institutional, and seemingly nonracial practices. At the discursive level, color-blind racism, like all racial ideologies, relies heavily on “racial story lines,” which reproduce particular racial orders and modes of racial domination as common sense (Bonilla-Silva 2014:124). By omitting, underdeveloping, or demonizing people of color in narratives, racial story lines structure how we see and, perhaps even more powerfully, how we do not see racial domination in everyday social phenomena (Bonilla-Silva 2012).
A related literature considers how the racialization of place perpetuates racial inequality (Holloway 2000; Hunter et al. 2016; Stuart 2016). Developing the concept of “territorial stigma,” Loïc Wacquant (2008) argued that the stigma placed on minority neighborhoods causes the public to look upon these places, and in turn their residents, as immoral, dangerous, and generally repulsive. Such stigma derives from public sentiments about particular areas; however, it is also created and reproduced through news media. To fully analyze the role of the news media in devaluing minority lives, it is critical to use a theoretical framework capable of recognizing color-blind racist narratives at the spatial level, which attribute lower value to homicide victims located in stigmatized minority neighborhoods.
Empirically, our study builds upon prior research, which has consistently demonstrated disparities in the newsworthiness of homicides and their victims. Previous studies showed that although some homicides become front-page stories, covered in multiple articles over many days, others receive barely any coverage, sometimes no coverage at all. Given these differences, researchers have used news media distortion analyses to identify which characteristics make a given homicide more newsworthy than others. By compiling details about incidents, suspects, and victims, these studies have compared how different constellations of characteristics influence newsworthiness (Gruenewald, Chermak, and Pizarro 2013; Johnstone, Hawkins, and Michener 1995:869; Schildkraut and Donley 2012).
Previous research has shown that homicides with particularly rare, unusual, or “statistically deviant” characteristics are treated as more newsworthy and thus receive more coverage. For example, homicides involving multiple victims (Paulson 2003; Sorenson, Peterson Manz, and Berk 1998), racial minority victims (Lundman 2003; Paulson 2003; Weiss and Chermak 1998), young victims (Paulson 2003; Sorenson et al. 1998), and those that occur in wealthy neighborhoods (Paulson 2003; Sorenson et al. 1998) typically receive more coverage than homicides without these characteristics. In a study of homicide reports in Indianapolis, Weiss and Chermak (1998) found that although homicides with Black victims are more common, those involving White victims receive more coverage. Another study showed that suburban homicide is more newsworthy than urban homicide (Chermak 1995).
However, these studies are characterized by three main limitations in their assessments of disparities in newsworthiness. First, studies have overlooked how neighborhood characteristics importantly contribute to the newsworthiness of certain victims. Although homicides are spatially patterned (Messner et al. 1999; Stults 2012), we know of no existing study in which neighborhood racial composition was taken into account when assessing the relative newsworthiness of homicide victims. We address this shortcoming by considering how homicide location, particularly the racial makeup of neighborhoods, influences the newsworthiness of victims.
Second, the majority of studies examining the newsworthiness of homicide victims have focused almost exclusively on the amount of coverage victims receive. However, race also influences how an article is written, the focus of the story, and other attributes that can be discerned only through close reading and coding. Taylor and Sorenson (2002) offered one of the very few studies to use a qualitative approach, although they did not focus on race. Concern for such qualitative aspects of newsworthiness is particularly important in the current age of online news. Freed from word- and page-limit restrictions, some online news sources have begun to report on all homicides occurring within their jurisdictions. This makes it even more important to expand existing measures of newsworthiness. We do so by measuring whether reporting acknowledges victims’ “complex personhood.” Developed in Avery Gordon’s (2008) scholarship on representation and empathy, this concept denotes the manner in which readers are exposed to the multidimensional humanity of individuals (also see Rios 2017; Stuart 2020). When audiences are confronted with an individual’s complex personhood, they develop empathy, which “haunts” them for a longer period of time (Gordon 2008).
Third, findings from previous studies of race and newsworthiness of crime victims are limited by the selection and potential bias of news samples. In most studies homicide news is collected from a single newspaper. The drawback of this approach is that newspapers, like all news outlets, vary widely in their reporting practices. Therefore, findings based on a single source may not be reflective of the overall newsworthiness of homicide victims in the same city or locale. We overcome this limitation by collecting all news articles, published by every available news outlet, that reported on 2016 Chicago homicide victims.
Conceptual Framework
Prior theoretical and empirical scholarship describes newsworthiness as affected by a diverse set of factors, including newsroom culture (Gans 1979), organizational policies (Tuchman 1973), and ideological leanings of editors (Fishman 1980). As Clayman and Reisner (1998) advocat, however, measuring the newsworthiness of an empirical occurrence does not require researchers to analyze the entire range of input factors. Instead, scholars should approach newsworthiness as a social fact that can be observed by examining which occurrences actually “end up” as news and how these stories are told. By identifying which victims’ deaths are deemed most newsworthy, we examine how the value of victims’ lives is understood, constructed, and presented in these final news products (Ferrell and Websdale 1999). Our study is guided by four primary hypotheses, which we detail below.
First, the perceived newsworthiness of victims is inseparable from long-standing racial stereotypes and stigmas. The racial stereotyping of crime has been an enduring feature of American culture, with the frequent representation of minorities, particularly Black and Hispanic men, as criminals (see Entman and Rojecki 2001; Welch 2007). Research consistently shows that police officers rely on racialized typifications when making arrest decisions (Beckett et al. 2006), and court officials deploy racial stereotypes in sentencing (Steen, Engen, and Gainey 2005). Such disparities in arrests and punishments reinforce the prevailing myths that minorities are more violent, dangerous, and criminal than White people. Given these consequential stereotypes, an individual victim’s race is likely to shape whether their death is deemed exceptional or expected. We anticipate that Black and Hispanic victims will be seen as more “deserving” of their fate and therefore less newsworthy. Conversely, because White individuals are perceived as less associated with criminal behavior, their victimization will be perceived as “undeserving” and therefore more newsworthy. This provides our first hypothesis:
Hypothesis 1: Black and Hispanic victims will receive less news coverage than White victims.
Although there is ample theoretical justification for an association between individual race and newsworthiness, scholarship on territorial stigma (Stuart 2016; Wacquant 2008), or the “blemish of place,” suggests that the newsworthiness of a given victim may be influenced by the place where the homicide occurred. We anticipate that homicide victims in predominantly Black or Hispanic neighborhoods will be seen as more deserving of their fate and therefore less newsworthy than homicide victims in White neighborhoods. This provides our second hypothesis:
Hypothesis 2: Homicide victims killed in predominantly Black or Hispanic neighborhoods will receive less coverage than those killed in predominantly White neighborhoods.
The potential devaluation of minority lives is not solely exhibited quantitatively by the number of articles a particular homicide victim receives. Rather, an analysis of color-blind, racial story lines requires examining not only whether minority lives are discussed but also how they are discussed. Arguably the most powerful way to express the value of someone’s life is to recognize their complex personhood, that is, to highlight and discuss their social relationships, such as key family and community roles. Indeed, prior research on news coverage finds that interviews with family members are a way of evoking emotional reactions from readers (Chermak 1995). Yet identifying, locating, and interviewing family, friends, and community peers require extensive work on the part of journalists. It is far easier to rely on police reports, which typically do not contain information about victims’ complex personhood. As such, journalists are likely to put more effort into investigating and reporting on victims’ multiple social roles when something about the homicide victims or locations seems more newsworthy. This results in only some victims’ receiving this sort of empathetic treatment. Given this, we anticipate that coverage of minority homicide victims, as well as victims killed in predominantly minority neighborhoods, is less likely to highlight victims’ complex personhood than White victims and victims killed in predominantly White neighborhoods. This provides our final two hypotheses.
Hypothesis 3: Coverage is less likely to recognize the complex personhood of Black and Hispanic victims than White victims.
Hypothesis 4: Coverage is less likely to recognize the complex personhood of victims killed in predominantly Black or Hispanic neighborhoods than victims killed in predominantly White neighborhoods.
Data and Methods
To examine these four research hypotheses, we constructed an original data set composed of every named homicide victim in Chicago in 2016 and the corresponding news stories about each homicide. Anonymized homicide incidents, along with their corresponding dates and locations, are publicly provided by the Chicago Police Department (CPD) and downloadable via the Chicago Data Portal (data.cityofchicago.org). To establish the identity of victims, we cross-referenced the homicide dates and locations listed in the CPD database with three news publications: DNAInfo’s Homicide Timeline (www.dnainfo.com/chicago/2016-chicago-murders), the Chicago Sun-Times’s Homicide Watch (chicago.homicidewatch.org), and the Chicago Tribune’s Homicide Tracker (http://homicides.redeyechicago.com). We established the identities of 762 unique homicide victims and compiled the name, age, race, gender, home address, incident address, cause of death, date of death, and number of victims involved in each homicide incident. We excluded two victims, from the overall 764 homicide victims, in the formal analysis who remained unnamed (“John Doe” and “Jane Doe”) in the news at the time of our data collection, because we were not able to collect information on their news coverage, the key outcome variable. An additional 6 victims were excluded because they were such extreme outliers, leading to an analytic sample of 756 homicide victims. 1
To compile all news articles written about these homicides, we searched each victim’s name in the three major electronic news databases: ProQuest Global Newsroom, Newsbank, and Factiva. We also searched each victim’s name on the websites of three local news outlets: the Chicago Tribune, the Chicago Sun-Times, and DNAInfo. To ensure that no articles were missed, we performed multiple searches in each database with different variations in our search criteria, such as alternative spellings of first and last names and the omission or inclusion of middle names and initials. This produced 5,092 articles that included local, national, and international news outlets, as well as the Associated Press. A close reading and coding of every article revealed four categories of articles that cover the victims in our study: (1) “review articles,” which provide a formulaic summary and list of homicides occurring on a weekly or monthly basis (1,508 articles); (2) “mention articles,” which include the name of one or more victims but ultimately report on a different story, for instance, descriptions of broader crime trends in the city (697 articles); (3) “suspect articles,” which discuss the apprehension, trial, or sentencing of suspects but mention the victim only in passing (642 articles); and (4) “direct articles,” which directly describe the homicide details and discuss the victim (2,245 articles). Direct articles provide the most consistent comparison for the relative value and newsworthiness of individual victims and therefore constitute our analytic sample. We do not include the broader count of all news articles, because doing so risks introducing significant bias into the analysis. For instance, not all victims’ killers were apprehended and thus did not lead to a “suspect article.”
Outcome Variable: Newsworthiness
The newsworthiness of individual homicide victims is expressed by two measures: the amount of coverage and whether the reports recognize the complex personhood of the victim. The amount of news coverage per victim was measured by the number of news articles discussing the victim’s homicide. On average, homicide victims received 2.8 newspaper articles focused on the individual and the homicide incident, with one victim receiving a high of 28 articles. Almost all victims were discussed in at least one article (98.3 percent).
To compile data on the second outcome variable of interest, recognition of complex personhood, we developed a list of qualitative codes. On the basis of previous literature (Gordon 2008; Rios 2017; Stuart 2020), we constructed 10 codes to capture the extent to which an article conveys a victim’s multidimensional human complexity. These codes documented whether the victim was described as (1) spouse or partner, (2) parent or grandparent, (3) aunt or uncle, (4) child or grandchild, (5) brother or sister, (6) niece or nephew, (7) cousin, (8) friend, (9) neighbor, or (10) other community role (e.g. mentor, leader, volunteer).
The manner in which these codes capture the degree of complex personhood afforded to different victims is illustrated in the example below, taken from a news article about a victim named Anthony Heatherly: Anthony Heatherly, of the 5700 block of North Kingsdale Avenue in Sauganash, was taken to Lutheran General Hospital, where he was pronounced dead at 2:15 p.m. Monday, according to Cook County medical examiner’s records. Anthony, who formerly attended Taft High School, played on the Norwood Park school’s football and basketball teams, Anthony was also mourned by
In this example, discussion of Heatherly playing multiple roles—as a friend, grandson, student, and member of an organization—produces key discursive labels that not only communicate Heatherly as a complex, multidimensional human but also his death as a significant loss or tragedy to loved ones and the community. Interviews with loved ones directly convey humanizing language about the victim, like the teacher who said that Anthony was a “nice kid,” and “bright and funny and capable.” On the other hand, articles that do not mention family or community roles tend to focus specifically on the homicide incident itself, detailing what happened but failing to convey the complex personhood of the victim.
To assess the degree of complex personhood afforded to each victim, a single research team member closely read and coded all 2,245 articles in MaxQDA. Adopting a method developed by Taylor and Sorenson (2002), each article was coded dichotomously according to whether each victim was described as possessing at least 1 of the 10 roles listed above, as well as the number of different roles mentioned. In total, only 22.4 percent of victims were described as having at least one family or community role, with roughly only 15 percent having more than one role mentioned. As a result, we measured this outcome dichotomously in the analysis.
Explanatory Variables: Individual- and Neighborhood-Level Race
In this study, we consider both victims’ race and the racial composition of the neighborhood where victims were killed. Information on each victim’s race was provided by Homicide Watch, a news source that provides information on every homicide victim in Chicago. As Table 1 shows, the majority of victims were identified as Black (78 percent), 16 percent as Hispanic, and 4 percent as White. These statistics are quite similar to the CPD official data on the racial makeup of 2016 homicide victims, suggesting that our data collection method seemed to produce a data set consistent with the confidential, raw police data on homicide victims that is unavailable to the public (Kapustin et al. 2017).
Descriptive Information on Victims, Neighborhoods Where Victims Were Killed, and Homicide Incidents.
Because the CPD data from the Chicago Data Portal contained block-level addresses, we were able to geocode the location of each homicide and obtain information on the racial makeup of the 65 Chicago community areas where victims were killed (Social IMPACT Research Center 2015). 2 From there, we used racial demographic information from 2015 U.S. census data, which is only one year prior to the 2016 homicides we study. In Chicago, Black, Hispanic, and White residents constitute the three largest racial groups within neighborhoods. The descriptive statistics for the percentage of victims who were killed in community areas with different racial demographics are included in Table 1. On average, victims were killed in community areas that were made up of 65 percent Black residents, 21 percent Hispanic residents, and 10 percent White residents. Table 1 shows that about 64 percent of victims were killed in neighborhoods that were majority Black, while only 20 percent of victims were killed in majority-Hispanic neighborhoods, and very few (about 4 percent) were killed in majority-White neighborhoods. On the basis of these results, we focused our analyses on whether victims were killed in predominantly Black, Hispanic, or White neighborhoods.
One reasonable question is the extent to which individual race and neighborhood race are related. Table 2 details the bivariate relationship between neighborhood race and individual victim race. The results suggest that, as expected, an individual victim’s race is associated with the racial makeup in the community area in which he or she was killed. For example, about 75 percent of Black victims were killed in neighborhoods composed of 76 percent to 100 percent Black residents, while White and Hispanic victims were not likely to be killed in neighborhoods that were majority Black (9 percent for each). However, there is still enough variation here to justify the inclusion of both victim race and the racial makeup of neighborhoods where victims were killed.
The Racial Makeup of Neighborhoods in Which Victims Were Killed, by Race of Victims.
Control Variables
We include several key control measures—gender, age, the type of homicide, and whether the incident involved multiple homicides—that were used in prior studies, to control for confounders and provide results that are comparable with other findings. We also include a control measure for the neighborhood homicide rate, created by dividing the number of homicides in each neighborhood from our data set by the neighborhood population size, using estimates from 2015 U.S. census data (Social IMPACT Research Center 2015). It is important to control for neighborhood homicide rate to understand the association between neighborhood race and newsworthiness even net of the homicide rate, as this is associated with both newsworthiness and neighborhood racial composition.
Although it may appear useful to control for neighborhood socioeconomic status to better untangle the effect of neighborhood-level race net of socioeconomic status, we found that neighborhood median income is very strongly correlated with race (percentage White in the neighborhood), with a Pearson’s correlation coefficient of .83. As this could result in multicollinearity, we did not include this measure of neighborhood socioeconomic status in our models.
As Table 1 shows, the vast majority of homicide victims were male (91 percent), and more than half were between the ages of 18 and 29 years (55 percent), with a median age of 29 years. Approximately 90 percent of homicides were designated as shootings, 5 percent as stabbings, and 5 percent as assault or some other method. This is in line with statistics produced from CPD administrative data, which show that roughly 90 percent of homicides in 2016 involved firearms (Kapustin et al. 2017). We created a dichotomous variable to indicate whether a victim was killed in an incident with multiple fatalities, to account for the fact that these homicides tend to receive more news coverage. Only about 9 percent of homicides in this study involved multiple fatalities. 3
Analytic Strategy
We used multilevel model techniques to account for the potential variation at the neighborhood level and the fact that victims are nested within neighborhoods. Using ordinary least squares instead could violate the independence assumption of ordinary least squares regression, because the clusters of observations are not independent of each other (Raudenbush and Bryk 2002). Furthermore, multilevel modeling was appropriate because of our interest in parsing out the respective relationships between news coverage with individual race and with neighborhood racial makeup. For the first outcome, the amount of news coverage victims received, we used hierarchical Poisson models, given that the number of articles written about victims was a count variable. 4 For the second outcome of interest, whether coverage acknowledged the complex personhood of the victim, we used a hierarchical logistic regression model. For both of these outcomes, the level 1 models included victim race, and all of the control measures for victim characteristics and homicide incident characteristics. The level 2, level 3, and level 4 models then incorporated the neighborhood race measures as well as the control measure for the neighborhood homicide rate. For both the Poisson and logistic regression multilevel models, we interpreted the unit-specific models, so that we were holding constant the value of the random effect or holding the neighborhood constant.
Results
We start by descriptively examining the relationship between race and newsworthiness and follow with multilevel models to formally display these relationships. Table 3 presents the average number of articles victims received, by victim race and neighborhood racial composition. This table also has information on the percentage of victims for whom reports acknowledged complex personhood, again by victim race and neighborhood racial composition. On average, Black victims received 2.8 news articles, Hispanic victims received 2.6 news articles, and White victims received 3.8 articles. This table shows that these averages by victim race vary by neighborhood racial composition. Both Black and Hispanic victims received the lowest average number of articles when killed in majority-Black neighborhoods and the highest average number of articles when killed in majority-White neighborhoods. Black victims had an average of one article more when killed in majority-White compared with majority-Black neighborhoods. White victims also received the highest average number of articles when killed in majority-White neighborhoods (4.5 articles) compared with those killed in majority-Hispanic (2.9 articles) or majority-Black (3.7 articles) neighborhoods. Descriptively, Black victims killed in majority-White neighborhoods (3.7 articles) received similar averages as White victims killed in majority-Black neighborhoods (3.7 articles).
Amount of News Coverage and Recognition of Complex Personhood in Coverage, by Racial Characteristics.
We find a similar trend in examining which victims were reported on in a way that acknowledged complex personhood. Black and Hispanic victims were least likely to be described as having family and community roles when killed in majority-Black neighborhoods and most likely when killed in majority-White neighborhoods. These results were even more extreme for White victims: only 33.3 percent of White victims killed in majority-Black neighborhoods received coverage that indicated their complex personhood, compared with 83.3 percent of White victims killed in majority-White neighborhoods. While Table 2 shows the correlation between individual and neighborhood race, these results also highlight the importance of considering both victim race and neighborhood racial composition.
Multilevel Outcome Models
Table 4 presents the hierarchical Poisson regression models examining the number of articles victims received by individual race and neighborhood-level race. Model 1 includes only a measure of the individual victim’s race, with White victims as the reference group, and the level 1 victim and homicide incident control measures. The results from model 1 show that on average, Black (coefficient = −0.27, p < .01) and Hispanic (coefficient = −0.35, p < .01) victims received significantly less news coverage compared with White victims, net of controls, and the neighborhoods where victims were killed.
Hierarchical Poisson Regression: Number of Articles Written about Homicide Victims.
p < .05. **p < .01. ***p < .001.
Model 2 adds the level 2 measure for whether victims were killed in majority-Black neighborhoods, as well as a control for the homicide rate of the neighborhood and a control for the percentage of Hispanic residents in the neighborhood. By adding in a continuous measure of the percentage of Hispanic residents in a neighborhood, we are roughly able to interpret the majority-Black indicator to have a reference group of majority, or close to majority, White residents. In model 2, victims killed in majority-Black neighborhoods receive significantly less news coverage than those killed in neighborhoods that have more White residents (coefficient = –0.20, p < .05). Holding all else constant, including victim race, victims killed in predominantly White neighborhoods receive a predicted number of articles of 3.2, while for victims killed in predominantly Black neighborhoods, the predicted number of articles is 2.6, a difference of just over half an article. Furthermore, after taking into account neighborhood racial composition, Black victims no longer receive significantly less coverage than White victims, suggesting that individual victim race matters partially through place: Black victims are more likely to live in neighborhoods with more Black residents than are White victims, and victims killed in those neighborhoods receive less news coverage, on average.
Model 3 again includes the indicator for majority-Black neighborhood but holds the percentage of White residents in the neighborhood constant. By including this continuous measure for the percentage of White residents in a neighborhood, we are roughly able to interpret the majority-Black indicator to have a reference group of majority, or close to majority, Hispanic residents. We see here that victims killed in majority-Black neighborhoods receive significantly less news coverage than those killed in predominantly Hispanic neighborhoods (coefficient = −0.14, p < .05), net of all controls. Thus, victims killed in predominantly Black neighborhoods receive significantly less news coverage than those killed in White or Hispanic neighborhoods.
Next, model 4 includes an indicator for whether victims were killed in majority-Hispanic neighborhoods, and this time the percentage of Black residents is held constant. This means that the reference group for majority Hispanic can be roughly interpreted as majority, or close to majority, White residents. Here, victims killed in majority-Hispanic neighborhoods compared with those with more White residents did not receive significantly different amounts of coverage (coefficient = −0.08, p > 0.05), net of all controls. 5 Even after taking into account Hispanic neighborhood composition in this model, individual Hispanic victims still receive significantly less news coverage than White victims.
In sum, this table shows that Black victims receive significantly less news coverage than White victims, but this is primarily a result of where victims live, with Black victims being more likely to live in predominantly Black neighborhoods and those killed in majority-Black neighborhoods receiving less coverage. Although Hispanic victims also receive significantly less news coverage than White victims, this inequality remains, even after taking into account whether victims were killed in majority-Hispanic neighborhoods, and being killed in a majority-Hispanic neighborhood compared with a majority-White neighborhood is not significantly associated with the number of articles victims received. However, victims killed in neighborhoods with more Black residents received significantly less coverage than those killed in neighborhoods with more Hispanic residents.
Table 5 next presents hierarchical logistic regression models examining whether the homicide victims were discussed in terms of complex personhood. The models in Table 5 are constructed similarly to those in Table 4, such that model 1 includes only the individual victim race characteristics and level 1 control measures; model 2 focuses on the majority-Black neighborhood indicator, holding the percentage of Hispanic residents and neighborhood homicide rate constant; model 3 focuses on the majority-Black neighborhood indicator, holding the percentage of White residents constant; and model 4 focuses on the majority-Hispanic neighborhood indicator, holding the percentage of Black residents constant.
Hierarchical Logistic Regression: Whether Homicide Victims Were Discussed through a Lens of Complex Personhood in Newspaper Articles.
p < .05. **p < .01. ***p < .001.
In model 1, the results show that Black (coefficient = −1.09, p < .01) and Hispanic (coefficient = −1.44, p < .01) victims were significantly less likely than White victims to be discussed as complex, multidimensional humans. The odds ratio for Black compared with White victims is thus .34 (exp[−1.09]), meaning that the odds for Black victims of being discussed in terms of complex personhood are lower by about a factor of 3 than the odds for White victims, net of the controls. The odds ratio for Hispanic compared with White victims is .24, meaning that the odds of Hispanic victims’ being described as having family and community roles are lower by roughly a factor of 4 than the odds for White victims, independent of the control measures. Here, again, we find that female victims, young victims, and those killed in incidents with multiple fatalities are more likely to be discussed as complex, multidimensional humans.
In model 2, we add a measure for whether victims were killed in majority-Black neighborhoods and take into account the percentage Hispanic in the neighborhood and the neighborhood homicide rate. We find that those killed in majority-Black neighborhoods were significantly less likely to be discussed through a lens of complex personhood compared with those killed in a neighborhood with more White residents (coefficient = −1.10, p < .01), net of controls. Holding all controls and victim race constant, the predicted probability of being discussed through a lens of complex personhood for victims killed in majority-White neighborhoods is .35, while the predicted probability for those killed in predominantly Black neighborhoods is .17. This shows that the predicted probability for being discussed as a complex, multidimensional human for those killed in majority-White neighborhoods is about twice as high as for those killed in majority-Black neighborhoods. After taking into account neighborhood racial composition, individual race was no longer significantly associated with the likelihood that a victim would be discussed in a humanizing tone. These results suggest that again, the neighborhood racial composition where victims were killed is a mechanism through which Black victims are disadvantaged compared with White victims for this measure of newsworthiness. Neighborhood race is associated with indicators of complex personhood, and Black victims were more likely to be killed in predominantly Black neighborhoods.
In model 3, we hold the percentage of White residents constant, allowing the reference group for the majority-Black neighborhood indicator to be roughly interpreted as neighborhoods with close to majority Hispanic residents. Here, being killed in a majority-Black neighborhood, compared with a neighborhood with more Hispanic residents, was not significantly associated with the likelihood of being discussed as complex, multidimensional humans (coefficient = −0.22, p > .05). Those killed in majority-Black and majority-Hispanic neighborhoods were equally as likely to be discussed through a lens of complex personhood.
In model 4, we hold the percentage Black in the neighborhood constant and consider the indicator for whether victims were killed in majority-Hispanic neighborhoods, which allows us to roughly interpret the reference group as a neighborhood with more White residents. Here, those killed in majority-Hispanic neighborhoods compared with neighborhoods with more White residents were significantly less likely to be discussed through a lens of complex personhood (coefficient = −0.71, p < .05), net of control measures. In model 4, even after taking into account whether homicide victims were killed in majority-Hispanic neighborhoods, Hispanic victims were still significantly less likely than White victims to be discussed as multidimensional humans. Here, Hispanic victims, net of neighborhood race, were significantly less likely than White victims be discussed through a lens of complex personhood, and those killed in Hispanic neighborhoods, net of victim race, were significantly less likely to be discussed in this way.
Results from Tables 4 and 5 show that Black victims were considered less newsworthy than White victims and that this happened largely through the racialization of place, such that victims killed in majority-Black compared with majority-White neighborhoods were treated as less newsworthy, and Black victims were more likely to be killed in predominantly Black neighborhoods. The results for Hispanic victims consistently showed that Hispanic victims received less coverage and were less likely to be discussed as complex, multidimensional humans than White victims, and this could not be taken into account by considering place and the neighborhood racial composition of neighborhoods where victims were killed. Victims killed in Hispanic neighborhoods, net of victim race, did not receive significantly different amounts of coverage than those killed in White neighborhoods, but they did receive more coverage than those killed in neighborhoods with more Black residents. The results for whether victims were discussed as complex, multidimensional humans were different: those killed in Hispanic compared with White neighborhoods were significantly more likely to be recognized as complex persons, but this was not the case when compared with neighborhoods with more Black residents. Thus, White victims and those killed in predominantly White neighborhoods were consistently treated as more newsworthy than Black victims and those killed in predominantly Black neighborhoods. White victims consistently received more coverage and were more likely to be recognized as complex persons than Hispanic victims. However, after taking into account victim race and other controls, those killed in predominantly White neighborhoods did not receive significantly different amounts of coverage than those killed in predominantly Hispanic neighborhoods, but they were more likely to be discussed as complex, multidimensional individuals.
To understand potential heterogeneity among predominantly Hispanic neighborhoods in newsworthiness of victims, we also examined differences in the number of articles victims received, and the discussion of complex personhood, within the majority-Hispanic neighborhoods. Exploratory analyses show little variation across predominantly Hispanic neighborhoods in these outcomes and that any existing variation is not explained by the percentage foreign born, poverty level, geographic location in the city (e.g., west vs. south), or the broader ethnoracial makeup of such neighborhoods.
Discussion
Recent high-profile, fatal tragedies affecting Black and Hispanic communities have led commentators and emerging social movements to renew long-standing charges that U.S. society, particularly the news media, devalues minority lives and disregards their loss. In this study, we use the relative newsworthiness of homicide victims to empirically measure this process at the individual and neighborhood levels, responding to Bonilla-Silva’s (2015) important call to more fully investigate the racialization of place. Using multilevel modeling and an original data set, this study produced four main findings, each of which supports the claim that the news media generate and reproduce racial story lines that systematically devalue minority lives.
First, victims killed in predominantly Black neighborhoods receive less news coverage than those killed in predominantly White neighborhoods, even after taking into account victim race and the homicide rate of neighborhoods. Second, victims killed in majority-Black or majority-Hispanic neighborhoods were significantly less likely to be discussed as multidimensional humans than those killed in predominantly White neighborhoods, net of all controls. Third, victims killed in predominantly Hispanic neighborhoods received significantly more news coverage than those killed in predominantly Black neighborhoods, but victims killed in Hispanic and Black neighborhoods were equally unlikely to be discussed in terms of complex personhood. Finally, although Black victims received less news coverage and were less likely to be acknowledged as complex, multidimensional humans than White victims, this was primarily a result of where they were killed. Because of persistent racial segregation, most Black Chicago residents reside and are killed in predominantly Black neighborhoods. However, place was unable to fully account for differences in newsworthiness between Hispanic and White victims.
We posit that disparities based on the racialization of place in the newsworthiness of homicide victims result from two interrelated processes operating at the discursive and organizational levels. Long-standing theories suggest that the differences in the relative coverage and tone devoted to those killed in minority communities can be explained by the influence of hegemonic cultural narratives, or “controlling images” (Collins 2002), about minorities and minority neighborhoods. Because society stereotypes Black and, increasingly, Hispanic residents as inherently criminal (see Collins 2002; Rios 2017; Wacquant 2008), news organizations treat homicides in Black and Hispanic neighborhoods as relatively normal, unremarkable, and, as a result, unworthy of extensive or humanizing news coverage. Indeed, as theories of territorial stigma posit, racialized images of dangerousness, immorality, and undeservingness have become firmly attached not only to minorities, but to minority neighborhoods as a whole (Stuart 2016; Wacquant 2008). One of the most novel aspects of this study is our ability to empirically confirm this “jointly and inseparably spatial-cum-racial” process (Wacquant 2008:181), by demonstrating that neighborhood race matters for newsworthiness of homicide victims even after taking into account the homicide rate in different neighborhoods. Furthermore, the impacts of the racialization of place on newsworthiness cannot be explained by the logic of rarity, or the increased newsworthiness of victims from neighborhoods in which relatively few homicides occur. In sum, hegemonic cultural narratives and territorial stigma appear to be primary forces driving the devaluation of minority lives and neighborhoods in the news.
It is important to note that news organizations, particularly traditional print newspapers such as the Chicago Tribune, have experienced drastic budget cuts and layoffs in recent years (Channick 2015). These changes put additional pressure on staff members, which may result in journalists’ resorting to conscious or unconscious racial stereotypes as shortcuts for determining newsworthiness. At the same time, however, contractions among traditional outlets have been accompanied by the rise of online media platforms, which may alleviate some of the spatial and racial disparities in reporting. Online news organizations confront fewer of the word-count limits, article volume constraints, and revenue-related concerns that encourage journalists and editors to prioritize some lives over others. As a result, online outlets provide a vehicle for local communities and progressively minded journalists to dedicate additional articles to minority victims, providing more opportunities to convey victims’ complex personhood. In Chicago, DNAInfo is one such online-only news outlet, explicitly devoted to neighborhood-level reporting. In our analytic sample of 762 victims and 2,245 articles, victims received an average of 1 article in DNAInfo. In contrast, victims received an average of 0.13 articles in the Chicago Tribune, the city’s dominant (and formerly print-only) newspaper. Although virtually all homicide victims in this study received at least some mention, which was not necessarily the case two decades ago (Chermak 1995), discussion of their deaths (and lives) are not evenly distributed across news outlets.
Although the interrelated processes of racialized cultural narratives, territorial stigma, and organizational demands of news organizations help explain the racial disparities in newsworthiness found in this study, this relationship is unlikely to be solely unidirectional. As previous media research demonstrates, racial disparities in who is covered in the news, and the ways they are portrayed in such coverage, feed back into cultural narratives, territorial stigma, and organizational demands. Research on “media exposure effects” consistently finds that the news shapes our view and understanding of the world around us, as well as the ways we act on those understandings (Lundman 2003; Meyers 1997). Most notably, the amount and tone of reporting given to a particular homicide may even influence the scope and resources devoted to subsequent police investigations (Gilchrist 2010). By placing heightened, sustained attention on certain victims and communities over others, news organizations indicate which homicide investigations the citizenry and the state should deem most urgent and important. This not only feeds back into the devaluation of minorities and residents of minority communities; it simultaneously reinforces the perceived higher value of White communities. As structural theories of color-blind racism importantly remind, racial story lines operate much like air pollution; they are hard to see clearly, yet poison all of us. They affect our cognitive maps, our sense of self, even what we choose to do with our bodies. Perhaps most important, they prevent White people from truly empathizing with people of color, fueling opposition to the growing declarations that Black lives matter.
Limitations and Future Research
We identify two limitations that must be considered in light of these findings. First, very few Chicago homicide victims in 2016 were White—only 32 victims total, roughly 4 percent of the total sample. This makes it more difficult to compare the devaluation of Black lives with White lives. Furthermore, there were even fewer White victims killed in predominantly Black neighborhoods. Despite constituting the majority of homicide victims, there were similarly few Black victims killed in predominantly White neighborhoods. Such racial distribution of homicide locations is a direct product of long-standing residential segregation patterns in Chicago and the broader United States. That said, analyzing smaller cities with less rigid segregation brings its own complications, because such cities are unlikely to experience enough homicides necessary for a robust analysis. Second, Chicago experienced a spike in homicides in 2016, increasing by 58 percent from 2015. It was the city’s deadliest year since the 1990s. Although this necessarily increased our sample size and brought the racial disparities into sharper relief, it may have also biased some of the homicide reporting in undetectable ways. In response, we took intentional steps to reduce potential bias by excluding news stories from our analysis that primarily discussed the spike, focused on trends, or generally only mentioned victims in passing. We limit our analysis to only those news stories that directly discuss a particular homicide event or the victim.
One of the most important next steps for future research on the devaluation of minority lives in news reporting is to expand the analysis beyond Chicago to assess whether spatialized racial disparities vary by geographic context. Furthermore, given that cities such as Chicago have experienced notable Black population loss, and that the majority of Americans live in suburbs that have grown poorer and more racially and ethnically diverse over the past three decades (Kneebone and Berube 2013; Parker et al. 2018), studies can also compare the racialized nature of homicide reporting between suburbs and major cities, among different suburbs, and within a single suburban city over the course of the suburbanization of Black and Hispanic residents. Although our analyses revealed little variation in homicide reporting across Hispanic neighborhoods, future work can further untangle how immigration status or nationality, in addition to race, may explain the devaluation and relative newsworthiness of certain neighborhoods and their residents. Researchers can also explore the joint distribution of race/ethnicity and socioeconomic status in determining the perceived value of neighborhoods, as residential segregation leads to both racial and socioeconomic disparities in neighborhoods (Reardon, Fox, and Townsend 2015). Given the strong association of the gender control measure used in this study with both measures of newsworthiness considered, future research may also benefit from analyzing the intersectionality of race and gender (Collins 2002; Wingfield 2013), which has been largely unexplored in the newsworthiness literature. Finally, future research can expand the qualitative component of the analysis. We suggest ethnographic observations of journalists and editors to better understand how their decisions about which victims to cover, and how to cover them, draw upon and reproduce racial ideologies.
Supplemental Material
Whose_Lives_Matter_Blinded_Manuscript_MARKED_REVISIONS – Supplemental material for Whose Lives Matter? Race, Space, and the Devaluation of Homicide Victims in Minority Communities
Supplemental material, Whose_Lives_Matter_Blinded_Manuscript_MARKED_REVISIONS for Whose Lives Matter? Race, Space, and the Devaluation of Homicide Victims in Minority Communities by Kailey White, Forrest Stuart and Shannon L. Morrissey in Sociology of Race and Ethnicity
Footnotes
Acknowledgements
We would like to thank Stephen W. Raudenbush for many helpful comments on this article, particularly pertaining to the multilevel modeling. We thank Nisarg Mehta for providing comments on a draft of this article. We are also grateful to Hunter Westbrook and Meggie Carroll for their research assistance with collecting and coding data. Any remaining errors in this article are those of the authors.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
