Who You Calling Old? Measuring “Elderly” and What It Means for Homicide Research

Abstract

Although Americans are less likely to experience violent crime as they age, research interest in elderly victims of violence is growing. An initial question that has been overlooked concerns how best to measure “elderly.” In the homicide literature, the most common definition is a single category of age 65 and older. With U.S. adults living longer, healthier, and more active lives, use of a single category may no longer adequately capture this heterogeneous population. The present study explores how a multiple-category definition of elderly might inform the study of homicide by identifying patterns that could promote more tailored explanations.

Keywords

elderly < subtypes victimization methodology NIBRS < methodology public policy

Although Americans are less likely to experience violent crime as they age, research interest in elderly victims of violence is growing. One reason for this attention is the so-called “graying” of America, which has been spurred by longer lifespans and the aging of the Baby Boomer generation. The population aged 65 and older is growing at a faster rate than those under 45 (Howden & Meyer, 2011). By 2030, one in five Americans will be over age 65, which translates into more than 72 million people (Vincent & Velkoff, 2010). For the oldest old population, the number of U.S. adults over age 85 is expected to reach 9.6 million in 2030, which is double its numbers from 2003 (He, Sengupta, Velkoff, & DeBarros, 2005).

In the area of homicide research, a small but growing body of work has identified various correlates associated with elderly victims. An initial question that has been overlooked in this literature concerns how to measure “elderly.” Although many researchers define elderly as those aged 65 and older, this age demarcation is not used consistently. The failure to use a standard definition is not surprising given the lack of consensus across disciplines regarding who is elderly. With older adults living longer, healthier, and more active lives, the use of a single category may no longer be adequate to capture this heterogeneous population. Given the growing interest in studying elderly homicide, now is an important time to address how best to define elderly and in particular whether a multiple-category measure could assist in identifying trends, developing tailored explanations, and promoting targeted policies and programs. The present study seeks to explore this issue by examining the current definition of elderly and then assessing a single- versus multiple-category measure of elderly. The implications for future research also are discussed.

Measuring Who Is “Elderly”

“Elderly” is a deceptively complex term to define due to differences in the aging process and the resulting heterogeneity in the older population (Aronow, Fleg, & Pepine, 2011). Considering specific characteristics such as common health, physical, or cognitive abilities rather than using a stark chronological age is considered the best practice for measuring elderly among gerontologists and those in the medical fields (Aronow et al., 2011; Blowers, 2004). Although individualized measures can more validly define elderly and generate more consistent comparisons across studies, researchers also acknowledge that practical needs exist to quickly identify the elderly population (Denton & Spencer, 1999). Here chronological age becomes the default metric. Although using a particular age demarcation provides a convenient definition, it is unclear what this age should be. Researchers have used various ages ranging from 50 to 75 (Blowers, 2004; Brank, 2007; Denton & Spencer, 1999, 2002; Orimo et al., 2006). Divergence arises even among agencies that serve this population (Brank, 2007). For example, the definition of elderly for Adult Protective Services varies by state with most states using either age 60 or 65 to define elderly (Stiegel & Klem, 2007).

Despite this lack of consensus, one of the most common definitions of elderly is age 65 and above (Denton & Spencer, 1999; Orimo et al., 2006). The basis for this age is believed to have its origins in Bismarckian Germany and the designation of 65 as the age citizens could participate in the national pension plan. Age 65 was selected based on the belief that most would die before reaching this age in the 19th century (Brank, 2007; Orimo et al., 2006). More recently, the use of age 65 as the basis for full retirement by the U.S. Social Security Administration likely encouraged defining elderly using this criterion.¹

Given the fact people now are living longer and are more active later in their lives, especially in industrial countries like the United States (He et al., 2005), some researchers question the logic of adhering to age 65 to define elderly (Denton & Spencer, 1999, 2002; Orimo et al., 2006). Denton and Spencer (1999, 2002) compared 40 years of Canadian life table data. Based on longer lifespans, they recommended using age 70 rather than 65 to define elderly (Denton & Spencer, 1999) and 90 rather than 85 to identify the oldest old population (Denton & Spencer, 2002). Orimo and his colleagues examined longitudinal health from Japan to conduct a similar inquiry. Based on their findings of healthier and more active older adults, they suggested changing the definition of elderly to age 75 and above.

In the homicide literature, a fairly clear consensus has formed around using age 65 and above as the measure of elderly. One of the first examples is seen in Wolfgang’s classic 1958 work, Patterns in Criminal Homicide. The vast majority of more recent studies also use age 65 (Abrams, Leon, Tardiff, Marzuk, & Sutherland, 2007; Bachman, 1993; Chu & Kraus, 2004; Fox & Levin, 1991; Kennedy & Silverman, 1990; Nelsen & Huff-Corzine, 1998; Titterington & Reyes, 2010; Weaver, Martin, & Petee, 2004). A few studies use younger ages including 60 (Fazel, Bond, Gulati, & O’Donnell, 2007; Krienert & Walsh, 2010) and 50 (Karch & Nunn, 2011). For all these studies, little explanation is given for the elderly measure that is used. At most, researchers comment on the lack of a standard definition and a reliance on previous studies’ use of age 65 (Bachman, 1993; Nelsen & Huff-Corzine, 1998). Karch and Nunn acknowledge that previous studies use a range of ages (from 50 to 65) and explicitly note their use of 50 as an effort to be overinclusive of fatal elder abuse cases.

If elderly is defined by only a single category, this measure combines a fairly heterogeneous group, no matter what age demarcation is selected. Multiple categories provide a way to balance the convenience of using a chronological age definition with the recognition of differences in the elderly population. The U.S. Bureau of the Census provides a useful illustration of this strategy. Although the Census uses age 65 and above to define the “elderly”, it also uses the additional age subcategories of 65 to 74 (the “young old”), 75 to 84 (the “aged”), and age 85 and above (the “oldest old”; U.S. Bureau of Census, 1996).² Multiple-category definitions of elderly, particularly use of a category to distinguish the “oldest old,” are regularly used in the public health literature. As with elderly, definitions of these elderly subpopulations vary. For the oldest old, the most common age demarcation parallels Census’s use of 85, but ages ranging from 75 to 90 have been used and recommended (Boscoe, 2008; Denton & Spencer, 2002; McLaughlin, Connell, Heeringa, Li, & Roberts, 2010). In the homicide literature, researchers rarely make any distinctions among the elderly.³ Krienert and Walsh (2010) provide one example where they define elderly as those aged 60 and above with the subcategories of 60 to 69, 70 to 79, and 80 and above, but they do not discuss why these particular subcategories are selected.

A multiple-category measure of elderly could advance homicide research by identifying patterns, promoting more nuanced explanations of elderly homicide, and supporting targeted programs and policies. Using subcategories to define the elderly could enable the identification of patterns otherwise masked by a single category, both within the elderly groups and across other nonelderly groups. A multiple-category definition could promote more nuanced explanations by reiterating the heterogeneity of the elderly population. Many explanations of elderly homicide utilize lifestyle or routine activity–based explanations (Kennedy & Silverman, 1990; Nelsen & Huff-Corzine, 1998). In discussing their findings, researchers make broad statements that might not be accurate when defining elderly by a single category. For example, the greater risk of elderly homicides occurring at home is explained as the result of “the elderly rarely leav[ing] their domicile” (Chu & Kraus, 2004, p. 88) and being “less mobile” (Abrams et al., 2007, p. 1669). The decreased use of firearms to kill elderly victims is because “perpetrators may conclude that firearms are not needed to kill frail homebound elderly victims” (Abrams et al., 2007, p. 1669). These studies define elderly as those aged 65 and older, but these explanations depict this group as being comprised of homebound, sickly, isolated individuals.⁴

Subcategories of elderly victims would highlight the different activities of the elderly population, particularly for the “young old” who are increasingly working past retirement age as well as maintaining active lifestyles when they do retire (He et al., 2005; Holder & Clark, 2008). A multiple-category definition also could support programs targeting at-risk elderly populations. Using a single category to define elderly might overemphasize the risks of certain crimes for the young old and underemphasize those for the oldest old. Bachman and Meloy (2008) discuss the variety of programs for the community-dwelling elderly that promote guardianship, assist family caregivers, and help detect victimization. Such programs could use more specific information about risks for different elderly groups to better serve these populations.

As the foregoing discussion highlights, the measurement of elderly is in need of attention by homicide researchers. The present study explores a multiple-category definition of elderly. To assess this definition and examine how it might inform the study of elderly homicide, this multiple-category measure will be compared with a single category to ascertain differences across victim and incident characteristics.

Method

Data

This study uses data from the Federal Bureau of Investigation’s Uniform Crime Reporting Program (UCR). Specifically it relies on data from the UCR’s National Incident-Based Reporting System (NIBRS), which provide the necessary victim and incident details, including incident location and clearance status. This study uses victim-level data from 2007 and 2008 NIBRS, which constitute two of the more recent years of publicly-available data (National Archive of Criminal Justice Data, 2009, 2010). Two years of data are used to expand the number of cases in order to minimize possible variations due to the fact that relatively few homicides are committed against those aged 65 and above. The cases analyzed are all murders and nonnegligent manslaughters (referred to as “murder” for shorthand). For 2007 and 2008, 6,662 murder victims were reported in NIBRS.

One caveat in analyzing NIBRS data is their limited coverage. NIBRS is a substantial departure in the UCR’s crime data collection for law enforcement agencies and it requires a lengthy certification process. As a result, the conversion to NIBRS has been gradual. By 2008, 31 states were NIBRS certified. Within these 31 states, not all agencies submit data in NIBRS format. NIBRS agencies covered approximately 25% of the U.S. population in 2007 and 2008 (JRSA, n.d.). Law enforcement agencies that participate in NIBRS tend to represent smaller population areas. In 2007 and 2008, no agency covering a population of more than 1 million participated in NIBRS. Because participation in NIBRS is voluntary, NIBRS states and law enforcement agencies do not constitute a representative sample of U.S. law enforcement agencies or states. This nonrepresentativeness of NIBRS suggests exercising caution when interpreting the results and generalizing beyond the NIBRS-participating agencies included in this study (but see, Addington, 2008).

Variables

Victim age

To compare definitions, “elderly” is defined using both a single category (65 and older) and three categories (65-74, 75-84, and 85 and older). These age groups are based on the most common definition of elderly in the homicide literature and the elderly subclassifications used by the U.S. Census.⁵ This study also uses the associated Census terms of “young old” (65-74), “aged” (75-84), and “oldest old” (age 85 and older) to distinguish among these groups. The frequencies for these age groups as well as all variables used in this study are presented in Appendix. As indicated in Appendix, the smallest percentage of homicides is seen for those aged 65 and older. Within the elderly category, the young old (65-74) constitute almost half (48%) of all murders against victims aged 65 and older.

Defining the nonelderly comparison groups also presents a challenge, and the literature provides little guidance. In elderly homicide research, some studies use a dichotomous measure of above or below age 65 (i.e., Abrams et al., 2007; Bachman, 1993; Chu & Kraus, 2004; Fazel et al., 2007; Titterington & Reyes, 2010). Others use various categories that typically distinguish young children and teenagers from the young and middle-aged adult populations (i.e., Chu & Kraus, 2004; Kennedy & Silverman, 1990; Nelsen & Huff-Corzine, 1998). The present study follows this latter practice. Since no consistent age categories are used to define the nonelderly population, the following four classifications are generated: 17 and younger (to identify children and teenagers), 18 to 25 (to capture the age group at greatest risk for homicide), 26 to 54 (to categorize the young and middle adult population), and 55 to 64. The 55 to 64 age category corresponds to Census’s designation of the “older” population and is used to explore possible similarities between elderly groups and this older adult group. Future work is needed to ascertain the most appropriate comparison groups.

Victim characteristics

In addition to age, victim characteristics include race, sex, and victim–offender relationship. Victim sex includes male and female. Previous studies present two sets of findings concerning victim sex. When looking across age groups, more elderly women than younger women are murder victims, but when looking among the elderly, more men than women are victims of homicide (Krienert & Walsh, 2010, for a summary). For race, White and non-White victims are compared because of the small number of minority victims who are not African American such as Asians and Native Americans. Findings from previous studies suggest that elderly victims of homicide tend to be White; however, patterns can vary depending on the data used especially for certain city-specific collections (Krienert & Walsh, 2010, for a summary; Titterington & Reyes, 2010).

The third victim characteristic examined is victim–offender relationship. This variable includes six categories: intimate partners, family members, friends/acquaintances, otherwise known, strangers, and unknown.⁶ To minimize the analytical challenges with the large amount of missing data for victim–offender relationship, a category of “unknown” victim–offender relationship is included to capture those cases where the relationship was specified as unknown as well as cases where no information was known about the offender. NIBRS allows multiple victim–offender relationships. Since the vast majority (75%) of cases involves a single relationship, the first NIBRS code is used. Victim–offender relationship is of particular interest given previous findings that elderly homicides tend to involve strangers and family members (Bachman, 1993; Krienert & Walsh, 2010).

Incident characteristics

Four incident characteristics are examined: location, weapon, circumstances, and clearance. Although NIBRS collects a wide range of incident locations, this study dichotomizes location as whether the incident occurred at home or not. Home location is the focus given its high frequency as a crime location for many victims but especially among elderly victims (Krienert & Walsh, 2010; Nelsen & Huff-Corzine, 1998).⁷

For weapon, the categories include: firearm, knife, personal contact, other, and unknown/missing.⁸ Although NIBRS allows for reports of up to three weapons per each offense in the incident, the decision was made to count only the first weapon reported. This decision rule greatly simplifies the analysis and includes the vast majority of weapons (84%) since most of the cases involved only one weapon. Previous studies produce mixed results regarding weapons. Some indicate the majority of elderly homicides occur with a firearm and others suggest blunt objects or personal weapons are more commonly used (see Krienert & Walsh, 2010, for a summary).

Although NIBRS collects several circumstance codes, this study focuses on two in particular: argument-related and felony-related.⁹ Argument-related circumstance is selected as this category is the most commonly reported of all homicide circumstances. This variable is measured using “arguments” coded in the NIBRS circumstance variable. Felony-related circumstances are selected due to prior research suggesting the frequency of murders among the elderly that occur with another crime (Fox & Levin, 1991; Krienert & Walsh, 2010; Nelsen & Huff-Corzine, 1998). To fully capture felony-related murders, this measure includes both victims where the NIBRS circumstance variable is coded “other felony” and victims of a crime in addition to the homicide.

The fourth incident characteristic concerns clearance. For this study, clearances include cases involving either an arrest of at least one offender or an exceptional clearance. Exceptional clearances occur when a suspect is identified but circumstances beyond the law enforcement agency’s control prevent an arrest (such as death of the offender; Federal Bureau of Investigation, 2004). Previous research suggests that age is related to likelihood of clearance. Homicides involving young victims are consistently more likely to be cleared and to be cleared faster than cases involving adults, especially older victims (see Riedel, 2008, for a discussion of the literature; Addington, 2007).

Analyses Conducted

As this study is exploratory in nature, it relies on bivariate analyses to compare initial patterns between use of a single-category and multiple-category definition of elderly.¹⁰ Here contingency tables are used to examine these relationships. To compare specific relationships between and across single- and multiple-category definitions of elderly and other age groups, 95% confidence intervals are reported.¹¹ Adjustments for multiple comparisons (such as Bonferroni corrections) are not applied given the exploratory nature of this work. Due to the relatively small amount of missing data for all variables except victim–offender relationship, complete case analysis is an appropriate approach (Allison, 2002). For victim–offender relationship, an “unknown” category is used to capture these missing cases.

Findings

Table 1 compares victim characteristics by age. The table includes a column using the single-category definition of elderly (aged 65 and above) as well as columns for these same victims regrouped into the three elderly subcategories. As reported in Table 1, substantively different patterns among the elderly subcategories are observed for victim sex and victim–offender relationship, particularly between the oldest old and other two groups. None of these differences, though, is statistically significant. Statistically significant and different patterns are observed when the single- and multiple-category elderly measures are compared with the nonelderly age groups. For victim sex, using the single 65 and older category indicates that 43.8% of elderly homicide victims are female and that this group is statistically significantly larger than the nonelderly groups.¹² When the elderly category is separated into the three groups, different patterns are observed. Comparing across all age groups, the 65-to-74 and 75-to-84-year-old groups are significantly larger and statistically different from all but the youngest (below 17) group. The oldest old group is the only age category where a majority of victims are female and that is statistically significantly larger than all the nonelderly age groups. For victim race, similar patterns are observed across age groups whether elderly is defined using a single- or multiple-category measure. A larger percentage of elderly victims are White as compared to all the nonelderly groups with the exception of the 55- to 64-year-olds.

Table 1.

Victim Characteristics by Age of Victim, NIBRS 2007-2008.

	Victim age					Elderly subcategories
	Under 17 percentage [95% CI] (n)	18-25 percentage [95% CI] (n)	26-54 percentage [95% CI] (n)	55-64 percentage [95% CI] (n)	65 plus percentage [95% CI] (n)	65-74 percentage [95% CI] (n)	75-84 percentage [95% CI] (n)	85 plus percentage [95% CI] (n)
Victim sex
Female	34.5 [31.1-38.2] (239)	16.0 [14.3-17.8] (280)	24.0 [22.5-25.4] (801)	27.7 [23.4-32.4] (105)	43.8 [38.2-49.6] (127)	41.0 [33.1-49.4] (57)	41.4 [32.7-50.8] (46)	60.0 [44.3-73.8] (24)
Total	692	1,751	3,342	379	290	139	111	40
N = 6,454, χ² = 168.6 (p < .05); χ² = 175.3 (p < .05) ^a
Victim race
White	48.7 [45-52.5] (329)	35.9 [33.6-38.2] (619)	48.1 [46.4-49.8] (1,581)	65.5 [60.6-70.1] (247)	74.5 [69.1-79.2] (213)	74.5 [66.5-81.1] (102)	72.7 [63.7-80.2] (80)	79.5 [64-89.4] (31)
Total	675	1,726	3,288	377	286	137	110	39
N = 6,352, χ² = 226.8 (p < .05); χ² = 227.4 (p < .05)^a
Victim–offender relationship
Intimate partner	0.7 [0.3-1.7] (5)	6.8 [5.7-8.1] (119)	14.3 [13.2-15.6] (480)	13.7 [10.6-17.6] (52)	16.6 [12.7-21.3] (48)	17.3 [11.8-24.5] (24)	14.4 [9-22.2] (16)	20.0 [10.3-35.2] (8)
Family	34.8 [31.4-38.4] (243)	2.4 [1.8-3.2] (42)	5.0 [4.3-5.8] (168)	12.4 [9.4-16.1] (47)	21.0 [16.7-26.1] (61)	15.8 [10.6-22.9] (22)	24.3 [17.2-33.1] (27)	30 [17.9-45.7] (12)
Friend/acquaintance	12.2 [9.9-14.8] (85)	20.8 [18.9-22.7] (364)	19.5 [18.2-20.9] (654)	15.8 [12.5-19.9] (60)	15.9 [12.1-20.5] (46)	20.1 [14.3-27.6] (28)	9.9 [5.6-17] (11)	17.5 [8.6-32.4] (7)
Otherwise known	12.9 [10.6-15.6] (90)	7.0 [5.8-8.2] (122)	6.5 [5.7-7.4] (219)	8.4 [6-11.7] (32)	7.9 [5.3-11.6] (23)	9.4 [5.5-15.4] (13)	6.3 [3-12.6] (7)	7.5 [2.4-20.8] (3)
Stranger	6.6 [5-8.7] (46)	12.5 [11.1-14.2] (220)	10.8 [9.8-11.9] (361)	10.3 [7.6-13.8] (39)	10.3 [7.3-14.4] (30)	10.8 [6.6-17.1] (15)	12.6 [7.6-20.2] (14)	2.5 [0.4-15.7] (1)
Unknown/missing	32.8 [29.4-36.4] (229)	50.6 [48.2-52.9] (887)	43.8 [42.1-45.5] (1,465)	39.3 [34.5-44.3] (149)	28.3 [23.4-33.7] (82)	26.6 [19.9-34.6] (37)	32.4 [24.4-41.7] (36)	22.5 [12.1-37.9] (9)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 1008.3 (p < .05)^a

Two χ² statistics are reported. The first is for contingency tables using the “65 plus” category for elderly, and the second is for the tables using the three-category definition for elderly. Due to cells with expected counts less than 5, χ² values are not reported for victim–offender relationship for the contingency table using the three-category definition.

For victim–offender relationship, interesting patterns are observed for the known relationships of family and friends when multiple categories are used to define elderly. For family relationships, defining elderly with one category suggests that 21% are killed by a family member. The over-65 group is significantly different from all the other age groups and is second only to the youngest (under 17) group for having the largest percentage of victims killed by a family member. When using the elderly subcategories, the young old (65-74) are similar to the 55 to 64 age group and statistically different from the other nonelderly groups. Both the 75 to 84 and 85 and older groups differ in a statistically significant way from all the nonelderly age groups except the under 17 group. For homicides committed by friends, using the single 65-and-older definition of elderly suggests no statistically significant differences across age groups. A similar pattern is observed for the young old (65-74) and oldest old (85 and above). For the aged (75-84), a significantly smaller percentage are killed by friends than for the 18 to 25 and 26 to 54 groups.

Table 2 presents the incident characteristics. For home location and clearance, the elderly subcategories do not differ in a statistically significant way from one another. In looking across the age categories for home location, the elderly (measured using above 65) are statistically significantly different from all other age groups and have the largest percentage of victims killed at home. When the elderly subcategories are used, the young old (65-74) and aged (75-84) are larger and significantly different from all the nonelderly age groups with the exception of the 55 to 64 group. The oldest old are only significantly larger from the 18 to 25 and 26 to 54 groups. For clearance, no statistically significant differences are observed when comparing elderly with nonelderly groups when either a single- or multiple-category definition is used.

Table 2.

Incident Characteristics by Age of Victim, NIBRS 2007-2008.

	Victim age					Elderly subcategories
	Under 17 percentage [95% CI] (n)	18-25 percentage [95% CI] (n)	26-54 percentage [95% CI] (n)	55-64 percentage [95% CI] (n)	65 plus percentage [95% CI] (n)	65-74 percentage [95% CI] (n)	75-84 percentage [95% CI] (n)	85 plus percentage [95% CI] (n)
Home location
Home	63.9 [60.3-67.4] (446)	38.9 [36.7-41.2] (683)	51.3 [49.6-53] (1,718)	68.1 [63.2-72.6] (258)	78.6 [73.5-83] (228)	79.9 [72.4-85.7] (111)	76.6 [67.8-83.5] (85)	80 [64.8-89.7] (32)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 280.9 (p < .05); χ² = 281.2 (p < .05)^a
Clearance
Cleared	62 [58.4-65.6] (433)	55.4 [53-57.7] (971)	57.9 [56.2-59.6] (1,938)	59.4 [54.3-64.2] (225)	63.1 [57.4-68.5] (183)	64.7 [56.5-72.2] (90)	64 [54.6-72.3] (71)	55 [39.6-69.5] (22)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 13.1 (p < .05); χ² = 14.4 (p < .05)^a
Argument-related circumstance
Argument related	11 [8.9-13.6] (77)	23 [21.1-25] (403)	25.9 [24.4-27.4] (866)	24 [20-28.6] (91)	14.1 [10.6-18.6] (41)	18.7 [13.1-26.1] (26)	10.8 [6.2-18.1] (12)	7.5 [2.4-20.8] (3)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 85.5 (p < .05); χ² = 88.8 (p < .05)^a
Felony-related circumstance
Felony related	4.4 [3.1-6.2] (31)	6.7 [5.6-7.9] (117)	7.1 [6.3-8] (237)	10.8 [8.1-14.4] (41)	17.6 [13.6-22.4] (51)	18 [12.4-25.3] (25)	18 [11.9-26.3] (20)	15 [6.9-29.6] (6)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 61.3 (p < .05) ^a
Weapon
Firearm	42.3 [38.6-46] (295)	76.8 [74.8-78.7] (1,347)	64.2 [62.6-65.8] (2,149)	45.4 [40.4-50.4] (172)	37.9 [32.5-43.6] (110)	43.2 [35.2-51.5] (60)	36 [27.7-45.4] (40)	25 [14-40.5] (10)
Knife	7.6 [5.8-9.8] (53)	9.5 [8.2-11] (167)	13.7 [12.6-14.9] (458)	16.1 [12.7-20.1] (61)	17.6 [13.6-22.4] (51)	21.6 [15.5-29.2] (30)	15.3 [9.7-23.3] (17)	10 [3.8-23.8] (4)
Personal contact	28.7 [25.4-32.1] (200)	4.6 [3.7-5.7] (81)	10.1 [9.1-11.2] (338)	19.5 [15.8-23.8] (74)	22.1 [17.7-27.2] (64)	15.8 [10.6-22.9] (22)	24.3 [17.2-33.1] (27)	37.5 [24-53.2] (15)
Other	10.9 [8.8-13.4] (76)	4.5 [3.6-5.6] (79)	5.2 [4.5-6] (173)	7.4 [5.1-10.5] (28)	11.7 [8.5-16] (34)	9.3 [5.5-15.4] (13)	10.8 [6.2-18.1] (12)	22.5 [12.1-37.9] (9)
Unknown/missing	10.6 [8.5-13.1] (76)	4.6 [3.7-5.6] (80)	6.8 [6-7.7] (229)	11.6 [8.7-15.2] (44)	10.7 [7.6-14.8] (31)	10.1 [6-16.3] (14)	13.5 [8.3-21.2] (15)	5 [1.2-17.9] (2)
Total	698	1,754	3,347	379	290	139	111	40
N = 6,468, χ² = 589.0 (p < .05)^a

Two χ² statistics are reported. The first is for contingency tables using the “65 plus” category for elderly, and the second is for the tables using the three category definition for elderly. Due to cells with expected counts less than 5, χ² values are not reported for felony-related circumstance or weapon for the contingency table using the three-category definition.

For the murder circumstances, using a single-category definition for elderly suggests that the oldest and youngest age groups have the lowest percentage of argument-related murders. Looking among the multiple categories of elderly, a smaller percentage of argument-related murders are observed for the oldest age groups (75-84 and 85 and older), but these differences are not statistically significant. Comparing these categories of elderly with the nonelderly age groups suggests different patterns than when the single 65 and older measure is used. The young old (65-74) are similar to all nonelderly age groups. The aged (75-84) show the same pattern as when a single-category measure of elderly is used. The oldest old are significantly different only from the 18 to 25 and 26 to 54 groups. For the felony-related circumstance, the 65 and older single category is larger than and significantly different from all age groups but the 55 to 64. When using three elderly categories, this pattern is replicated for all but the oldest old. The oldest old do not differ in a statistically significant way from any other age group with the exception of the youngest (under 17).

For weapons, interesting patterns are observed for knives and personal contact weapons. For knives, defining elderly using a single category indicates that a greater percentage of elderly are killed using knives than the youngest two age groups (under 17 and 18-25) and that the elderly do not differ in a statistically significant way from the other two age groups. For the multiple categories of elderly, a smaller percentage of murders involving knives are observed for the oldest age groups (75-84 and 85 and older) than the young old, but these differences are not statistically significant. Comparing these three categories of elderly with the nonelderly age groups suggests different patterns than when the single 65 and older measure is used. The young old age group has a significantly higher percentage of homicides using a knife than any nonelderly groups with the exception of the 55- to 64-year-olds. No statistically significant differences are observed when comparing the other elderly categories with the nonelderly. For personal contact weapons, defining elderly using a single category indicates that a greater percentage of elderly are killed using these weapons than the 18 to 25 and 26 to 54 groups, but the elderly do not significantly differ from the below 17 or 55 to 64 groups. When three elderly subcategories are compared, a statistically significant and larger percentage of oldest old are killed using personal weapons than the young old (37.5% vs. 15.8%). Looking across the nonelderly age groups, the young old are significantly smaller than the youngest group and larger than the 18 to 25 group. The aged (75-84) show a pattern similar to the single elderly category. A statistically significant and larger percentage of the oldest old are killed using personal weapons than every nonelderly age group with the exception of the youngest (under 17).

Discussion and Conclusions

The motivation for the present research originated from a lack of consensus regarding how to measure “elderly” and ways to capture the heterogeneity of this population. As such, the current study sought to explore how a multiple-category definition of elderly might facilitate homicide research, especially with regard to identifying patterns, developing explanations, and promoting targeted policies and programs. As the findings obtained illustrate, a multiple-category definition of elderly identifies both similarities and differences within elderly and nonelderly age groups. These patterns are otherwise masked by using a single-category definition of elderly. Although the comparisons within the three subcategories of elderly show many similarities, the most common substantive (but not always statistically significant) differences are observed between the young old (65-74) and oldest old (85 and older) groups. A higher percentage of the oldest old victims are female, killed by family members, and killed by personal contact weapons as compared to young old victims. A higher percentage of young old than oldest old victims are killed with knives and in argument-related circumstances. Using a multiple-category definition of elderly also highlights similarities across elderly and nonelderly age groups. Of particular interest are similarities between the youngest and oldest victims of homicide. These similarities include a higher percentage of murders by family members and using personal contact weapons and a lower percentage of argument-related murders than observed for other age groups. These patterns within the elderly categories as well as between the oldest and youngest victims provide future research opportunities to explore these relationships in greater depth.

Based on these patterns, a multiple-category definition of elderly provides some initial age-based explanations. Previous studies have relied on lifestyle/routine activity explanations for elderly homicide. The findings obtained by this study highlight the importance of distinguishing the routine activities between oldest old and young old. As more young old continue in the workforce past traditional retirement age and maintain active lifestyles when they do retire, these differences may become more pronounced. This explanation can be illustrated by the fact that the single-category (over 65) elderly definition masked similarities between the 55 to 64 group and the young old 65 to 74 group, particularly for homicides committed by a family member and occurring at home. Conversely, the similarities between the oldest and youngest victims suggest possible ways to compare lifestyles especially for homicides involving family members and personal contact weapons. Both groups may be more dependent on family members for caregiving and this exposure increases the opportunity for these crimes. In addition to lifestyle/routine activity explanations, previous research suggests that the elderly are at greater risk for serious injury or death because of their increased physical weakness (Chu & Kraus, 2004). Findings that personal contact weapons are more commonly used against the oldest old murder victims suggest that a “frailty hypothesis” might be more applicable for the oldest old category of elderly than the younger old, who are primarily killed using firearms. Finally, these findings confirm demographic patterns of aging. The increased percentage of female victims among the oldest old is likely due to longevity of females and the disproportion of females in this population group.

These patterns and initial explanations can support policies and programs that target at-risk elderly populations. The patterns for the oldest victims (being female, killed by family members, and killed by personal contact or other weapons) suggest similarities with victims of nonfatal elder abuse. National studies indicate that most elder abuse affects females above 80 and is committed by a family member, especially an adult child (Teaser et al., 2006). These findings lend support for programs for the community-dwelling elderly (especially the oldest old) that assist family caregivers and help detect victimization (Bachman & Meloy, 2008). Parallels between the youngest and oldest victims also suggest lessons might be learned from the best practices to prevent child abuse and other violence that could benefit prevention programs and victim services for the elderly.

The ultimate goal for this study is to start a conversation about the best practices for measuring the elderly population.¹³ Although this article presents an important first step, it is not without its limitations. One limitation is the use of NIBRS data. NIBRS data provide relevant information including location and clearance measures, not otherwise readily available, but these data do not cover all murders or all states. Eleven states have over 1 million people aged 65 and older. Not all of those states are included in NIBRS.¹⁴ Another limitation is the fact that a multiple-category definition can result in small numbers and may not always be practical for multivariate analysis. Despite these limitations, this study provides initial support for the utility of a multiple-category definition of elderly. More work is needed to examine this issue and ultimately to achieve consensus on the best practice and a standard for defining the elderly. In addition to more accurately measuring the elderly population, a standard definition would facilitate comparisons across studies and advance knowledge in the field. Now is a particularly important time to develop such a standard given both that attention to studying elderly violent victimization is increasing and the population above age 65 continues to grow in size and diversity.

Footnotes

Appendix

Frequencies for Variables Used

Variable	Frequency (%)
Age
Under 17	698 (10.5)
18-25	1,754 (26.3)
26-54	3,347 (50.2)
55-64	379 (5.7)
65 plus	290 (4.4)
65-74	139 (2.1)
75-84	111 (1.7)
85 plus	40 (0.6)
Unknown/missing	194 (2.9)
Victim characteristics
Sex
Male	5,016 (75.3)
Female	1,602 (24)
Missing	44 (0.7)
Race
White	3,069 (46.1)
Non-White	3,438 (51.6)
Unknown/missing	155 (2.2)
Victim–offender relationship
Intimate partner	711 (10.7)
Family	602 (9.0)
Friend/acquaintance	1,227 (18.4)
Other known	496 (7.4)
Stranger	707 (10.6)
Unknown/missing	2,919 (43.8)
Incident characteristics
Location
Home	3,450 (51.8)
Other location	3,212 (48.2)
Weapon
Firearm	4,164 (62.5)
Knife	801 (12.0)
Personal contact	801 (12.0)
Other	401 (6.0)
Unknown/missing	495 (7.4)
Argument-related circumstance
Argument related	1,500 (22.5)
Not argument related	5,162 (77.5)
Felony-related circumstance
Felony related	483 (7.3)
Not felony related	6,179 (92.7)
Clearance
Cleared	3,854 (57.9)
Not cleared	2,808 (42.1)

N = 6,662.

Acknowledgements

My thanks to Marc Riedel for organizing this special issue. I also thank Marc and the anonymous reviewers as well as Suzanne Perumean-Chaney and Callie Rennison for their thoughtful feedback on an earlier version of this article.

Author’s Note

This work originated from a presentation at the 2011 Homicide Research Working Group Annual Meeting in New Orleans and benefitted from discussions at that meeting.

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Lynn A. Addington is an associate professor in the Department of Justice, Law & Society at American University in Washington, DC. Her research focuses on violent victimization, the measurement of crime, and utilization of national crime statistics. Her publications have appeared in a range of outlets including the American Behavioral Scientist, Journal of Quantitative Criminology and Justice Quarterly. She also is the co-editor (with James P. Lynch) of Understanding Crime Statistics: Revisiting the Divergence of the NCVS and UCR (2007, Cambridge University Press).

References

Abrams

R. C.

Leon

A. C.

Tardiff

Marzuk

P. M.

Sutherland

(2007). “Gray murder”: Characteristics of elderly compared with nonelderly homicide victims in New York City. American Journal of Public Health, 97, 1666-1670.

Addington

L. A.

(2004). The effect of NIBRS reporting on item missing data in murder cases. Homicide Studies, 8, 193-213.

Addington

L. A.

(2007). Hot vs. cold cases: Examining time to clearance for homicides using NIBRS data. Justice Research and Policy, 9, 87-112.

Addington

L. A.

(2008). Assessing the extent of nonresponse bias on NIBRS estimates of violent crime. Journal of Contemporary Criminal Justice, 24, 32-49.

Allison

(2002). Missing data. Thousand Oaks, CA: Sage.

Aronow

W. S.

Fleg

J. L.

Pepine

C. J.

(2011). ACCF/AHA 2011 expert consensus document on hypertension in the elderly. Journal of the American College of Cardiology, 57, 2037-2114.

Bachman

(1993). The double edged sword of violent victimization against the elderly: Patterns of family and stranger perpetration. Journal of Elder Abuse & Neglect, 5, 59-76.

Bachman

Meloy

M. L.

(2008). The epidemiology of violence against the elderly: Implications for primary and secondary intervention. Journal of Contemporary Criminal Justice, 24, 186-197.

Barrett

A. E.

Cantwell

L. E.

(2007). Drawing on stereotypes: Using undergraduates’ sketches of elders as a teaching tool. Educational Gerontology, 33, 327-348.

10.

Blowers

A. N.

(2004). Exploiting the aged in familial settings. In Coston

C. T. M.

(Ed.), Victimizing vulnerable groups: Images of uniquely high-risk crime targets (pp. 71-82). Westport, CT: Praeger.

11.

Boscoe

F. P.

(2008). Subdividing the age group of 85 years and older to improve US disease reporting. American Journal of Public Health, 98, 1167-1170.

12.

Brank

E. M.

(2007). Elder research: Filling an important gap in psychology and law. Behavioral Sciences and the Law, 25, 701-716.

13.

Chu

L. D.

Kraus

J. F.

(2004). Predicting fatal assault among the elderly using the National Incident-Reporting System crime data. Homicide Studies, 8, 71-95.

14.

Denton

F. T.

Spencer

B. G.

(1999). How old is old? Revising the definition based on life table criteria. Mathematical Population Studies, 7, 147-159.

15.

Denton

F. T.

Spencer

B. G.

(2002). Some demographic consequences of revising the definition of old age to reflect future changes in life table probabilities. Canadian Journal on Aging, 21, 349-356.

16.

Fazel

Bond

Gulati

O’Donnell

(2007). Elderly homicide in Chicago: A research note. Behavioral Sciences and the Law, 25, 629-639.

17.

Federal Bureau of Investigation. (2004). Uniform crime reporting handbook. Washington, DC: U.S. Department of Justice. Retrieved from http://www.fbi.gov/ucr/handbook/ucrhandbook04.pdf

18.

Fox

J. A.

Levin

(1991). Homicide against the elderly: A research note. Criminology, 29, 317-327.

19.

Sengupta

Velkoff

V. A.

DeBarros

K. A.

(2005). 65+ in the United States: 2005 (Current Population Reports, P23-209). Washington, DC: U.S. Government Printing Office.

20.

Holder

K. A.

Clark

S. L.

(2008). Working beyond retirement age. Presentation at the American Sociological Association Annual Meeting. Retrieved from http://www.census.gov/hhes/www/laborfor/Working-Beyond-Retirement-Age.pdf

21.

Howden

L. M.

Meyer

J. A.

(2011). Age and sex composition: 2010: 2010 Census Briefs. Washington, DC: US Department of Commerce.

22.

JRSA. (n.d.). Status of NIBRS in the states. Retrieved from: http://www.jrsa.org/ibrrc/background-status/nibrs_states.shtml

23.

Karch

Nunn

K. C.

(2011). Characteristics of elderly and other vulnerable adult victims of homicide by a caregiver: National Violent Death Reporting System—17 US States, 2003-2007. Journal of Interpersonal Violence, 26, 137-157.

24.

Kennedy

L. W.

Silverman

R. A.

(1990). The elderly victim of homicide: An application of the routine activities approach. Sociological Quarterly, 31, 307-319.

25.

Krienert

J. L.

Walsh

J. A.

(2010). Eldercide: A gendered examination of elderly homicide in the United States, 2000-2005. Homicide Studies, 14, 52-71.

26.

Loftin

(1986). The validity of robbery-murder classifications in Baltimore. Violence and Victims, 1, 191-204.

27.

Lyons

(2009). Public perceptions of older people and ageing: A literature review. Dublin, Ireland: National Centre for the Protection of Older People. Retrieved from http://www.ncpop.ie/Year%201%20Reports/Microsoft%20Word%20-%20NCPOP%20Lit%20Rev%201%20IL%2001_12_09Cover%20inc22_01_10.pdf

28.

Maltz

M. D.

(2010). Look before you analyze: Visualizing data in criminal justice. In Piquero

A. P.

Weisburd

(Eds.), Handbook of qualitative criminology. New York, NY: Springer Science + Media LLC.

29.

McLaughlin

S. J.

Connell

C. M.

Heeringa

S. G.

L. W.

Roberts

J. S.

(2010). Successful aging in the United States: Prevalence estimates from a national sample of older adults. Journal of Gerontology, 65, 216-226.

30.

National Archive of Criminal Justice Data. (2009). National incident-based reporting system, 2007: Extract files. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research.

31.

National Archive of Criminal Justice Data. (2010). National incident-based reporting system, 2008: Extract files. Ann Arbor, MI: Interuniversity Consortium for Political and Social Research.

32.

Nelsen

Huff-Corzine

(1998). Strangers in the night: An application of the lifestyle-routine activities approach to elderly homicide victimization. Homicide Studies, 2, 130-159.

33.

Orimo

Ito

Suzuki

Araki

Hosoi

Sawabe

(2006). Reviewing the definition of “elderly.” Geriatrics Gerontology International, 6, 149-158.

34.

Riedel

(2008). Homicide arrest clearances: A review of the literature. Sociology Compass, 2/4, 1145-1164.

35.

SSA. (n.d.). The full retirement age is increasing. Retrieved from http://ssa.gov/pubs/ageincrease.htm

36.

Stiegel

Klem

(2007). Age-threshold eligibility for Adult Protective Services: Comparison chart of criteria by state. Retrieved from http://www.americanbar.org/content/dam/aba/migrated/aging/about/pdfs/Age_Threshold_Eligibility_Criteria_for_Adult_Protective_Services_by_State.authcheckdam.pdf

37.

Taylor

T. J.

Holleran

Topalli

(2009). Racial bias in case processing: Does victim race affect police clearance of violent crime incidents? Justice Quarterly, 26, 562-591.

38.

Teaser

P. B.

Dugar

T. A.

Mendiondo

M. S.

Abner

E. L.

Cecil

K. A.

Otto

J. M.

(2006). The 2004 Survey of State Adult Protective Services: Abuse of adults 60 years of age and older. Washington, DC: National Center on Elder Abuse.

39.

Titterington

V. B.

Reyes

N. C.

(2010). Elder homicide in urban America: An exploratory analysis of Chicago, Houston and Miami. Southwest Journal of Criminal Justice, 6, 228-249.

40.

U.S. Bureau of the Census. (1996). 65+ in the United States, Current Population Reports, Special Studies. Washington, DC: Author.

41.

Vincent

G. K.

Velkoff

V. A.

(2010). The next four decades. The older population in the United States: 2010 to 2050: Population estimates and projections. Washington, DC: Bureau of the Census. Retrieved from http://www.census.gov/prod/2010pubs/p25-1138.pdf

42.

Weaver

G. S.

Martin

C. D.

Petee

T. A.

(2004). Culture, context, and homicide of the elderly. Sociological Inquiry, 74, 2-19.