U. S. Department of Justice I Bureau of Justice Statistics I

Lifetime Likelihood of Victimization

by Herbert Koppel people's perception of the meaning of BJS Analyst March 1987 annual rates with respect to their own The Bureau of Justice Statistics lives. If the Earth revolved around the This report provides estimates of the National Crime Survey provides sun in 180 days, all of our annual crime likelihood that a person will become a annual victimization rates based rates would be halved, but we would not victim of crime during his or her life- on counts of the number of crimes be safer. time, or that a household will be vic- reported and not reported to timized during a 20-year period. This police in the United States. These Calculatinglifetime victimizationrates contrasts with the conventional use of a rates are based on interviews 1-year period in measuring crime and twice a year with about 101,000 For this report, lifetime likelihoods criminal victimization. Most promi- persons in approximately 49,000 of victimization were calculated from nently, the National Crime Survey nationally representative NCS annual victimization rates and life (NCS) surveys a sample of U.S. house- households. Those annual rates, tables published by the National Center holds and publishes annual victimization while of obvious utility to for Health statistics.% The probability rates, and the FBI's Uniform Crime policymakers, researchers, and that a person will be victimized at a Reports (UCR) provide annual rates of statisticians, do not convey to particular age basically depends upon crimes reported to the police. American citizens the full impact (a) the probability that the person is of crime on them nor do they still alive at that age and (b) the prob- Annual victimizationrates answer a critical question for each ability that a person of that age will be individual: What is the possibility victimized. The lifetime likelihood of Annual victimization rates alone do that I will be a crime victim? This personal victimization is derived from not convey the full impact of crime as technical report is designed to the probabilities of being victimized at it affects people. No one would express shed light on this question. the various ages that constitute a life- his or her concern by saying, "I am ter- Steven R. Schlesinger time. A similar method is used for cal- ribly afraid of being mugged between Director culating the long-term likelihood of January and December of this year." crimes against households. People are worried about the possibility that at some tine in their lives they are rape victims annually, the lifetime Because of the assumptions involved will be robbed or raped or assaulted, or chances of suffering a rape are much in the calculations and because the data their homes will be burglarized. greater. derive from a sample survey, the num- bers presented in this report are Annual rates can provide a false Certainly, annual rates are useful. estimates only; they should be inter- sense of security by masking the real A 1-year period is probably optimum preted only as indications of approxi- impact of crime. Upon hearing that the for researchers, statisticians, and plan- mate magnitude, not as exact mea- homicide rate is about 8 to 10 per ners. It encompasses exactly one cycle sures. Essentially they are calculated 100,000 population, one feels safe; of seasonal variation. It is long enough values of lifetime risk rather than des- after all, 1 chance in 10,000 is not very to smooth out the impact of extraordi- criptions of what has been observed. frightening. Actually, however, at nary events (such as a mass murder in a recent homicide rates about 1 of every small city) and to allow time for col- For the purposes of this report, a 133 Americans will become a murder lecting and analyzing data. On the ulifetime'f begins at age 12. This is dic- victim; for black males the propprtion other hand, a 1-year period is short tated by the data available from the is estimated to be 1 of every 30. enough to show changes and trends in NCS, which does not cover persons Similarly, while 16 out of 10,000 women crime rates. The problem lies with younger than 12 years old. Also, "series" victimizations, those repre- no feasible alternative to assuming con- in table 1. Alternatively, the estimate senting three or more similar incidents stant values. can be restricted to the future, by about which the victim could not pro- taking into account only those victimi- vide detail on each event separately, The lifetime likelihood of victirniza- zations that will be incurred in the have been excluded. If series victimi- tion can be interpreted as an estimate remainder of the person's lifetime, zations were included in the calcula- of a person's entire lifetime experience, starting at his or her current age. This - tions as representing one crime inci- starting in years prior to 1975-84 and eliminates errors due to past changes in dent, the total number of annual violent continuing into the future until the per- annual rates. Estimates using the victimizations would increase by ap- son's death. Such an estimate is subject latter option are presented in table 2; proximately 8%. The exclusion of to two sources of error that result from the values in table 1 constitute such series victimizations, then, results in an assuming constant annual rates: (1) the estimates for persons whose current underestimation to some degree of the probable overestimation of victirniza- age is 12. actual probabilities of victimization, tion rates in past years, and (2) the especially with regard to multiple vic- uncertainty regarding future annual timizations. rates. Estimates of this type are shown

Assumption of constant annual rates i Table 1. Lifetime likelihood of victimization The estimates of lifetime likelihood Percent of persons who will be victimized by crime starting at 12 years of age of victimization are derived under the Tnt-1 .".UA assumption that, throughout their life- One or more Number of victimizations times, people in the U.S. have incurred, victimizations One Two Three or more and will continue to incur, criminal victimization at the same annual rates Violent crimes, total* Total population as were observed in the years 1975 Male through 1984. Calculations are based Female upon the average annual victimization White rates for 1975-84 (except that data for Male rape are for 1973-82). Annual victimi- Female zation rates for the population, and for Black any group within the population, are Male assumed to remain constant at the Female average levels recorded for those Violent crimes, completed* years. For example, it is assumed that Total population Male the rate at which white females of age Female 34 become victims of assault was the same in the past and will be the same in White the future as it was in 1975-84. To the Black extent that annual rates were different Rape in the past or may change in the future, Total female White the estimates of lifetime victimization Black are inaccurate. Robbery Total population Of course victimization rates actu- Male ally vary from year to year. Victimiza- Female tion data are not available for years White before the mid-1970s (when the NCS Black was started), but crime rates as repor- Assault ted by UCR indicate that annual vic- Total population timization rates probably were lower Male then than in the period 1975-84. The Female UCR data, however, are not suitable White for estimating lifetime victimization Black because they include only crimes repor- Robbery or asseult ted to the police, and because there is resulting in injury no information about the age, sex, or Total population race of the victims. Personal theft Total population Since 1981, annual victimization Male rates reported by NCS have declined. Female If this trend were to continue, the White Male lifetime victimization probabilities Female would decrease. Black Male 99 5 10 84 Because there is very little usable Female 98 7 15 76 information about past variation in annual victimization rates, and because Note: Data are based on average victimization rates measured by the National Crime Survey for 1975-84, except for rape data, which are based on victimization rates for future changes are unpredictable in 1973-82. All crimes include attempts except where noted. See box on page 4 for crime definitions. both direction and magnitude, there is - Less than 0.5%. * Includes rape, robbery, and assault. Personal victimization rates Table 2. Lifetime likelihood of victimization, by age Based on 1975-84 annual victimiza- Percent of persons who will be victimized by crime starting at various ages tion rates, an estimated 5 out of 6 Total people will be victims of violent crimes One or more Number of victimizations (rape, robbery, and assault), either victimizations One Two Three or more completed or attempted, at least once Violent crimes* during their lives (table 1). (For the Current age precise definitions of the crimes dis- 12 years old , cussed in this report see the "Crime 20 definitions" box on page 4.) About half 30 40 the population will be victimized by 50 violent crime more than once. Males 60 are somewhat more likely to become 7 0 victims than females. Robbery or assault resulting in injury The greatest contrast is in multiple Current age 12 years old victimizations. It is estimated that 20 more than 2 in 5 black males will be 30 victimized by a violent crime three or 40 more times. This is almost double the 50 60 likelihood for black females and more 7 0 than triple the likelihood for white Personal theft females. Current age 12 years old When attempted crimes are exclu- 20 ded, the differences in lifetime victimi- 3 0 40 zation rates between blacks and whites 50 are greater. About 5 in 10 blacks, com- 60 pared with 4 in 10 whites, will suffer a 7 0 completed violent crime. Note: Data are based on average victimizationrates measured by the National Crilne Survey for 1975-84. All crimes include attempts. Seebox on page 4 for crime definitions. Based on 1973-82 rates, nearly 1 out -- Less than 0.5 percent. of 12 females will be the victim of a * lncludes rape, robbery, and assault. completed or attempted rape. The - estimated rate for black females is 1 out of 9. I Table 3. Long-term likelihood of household victimization I Percent of households that w~llbe v~ct~m~zedby It is estimated that 3 in 10 persons crlme over a 20-year per~od will be victims of a completed or at- Total.".-. One or more Number of victimizations tempted robbery during their life- victimizations One Two Three ormore times. The victim's sex and race ap- pear to have a greater effect upon the BWJary lifetime likelihood of victimization for All households Urban 80 32 26 22 robbery than for other crimes. Blacks Suburban 70 36 22 12 are almost twice as likely to be robbed Rural 64 37 19 8 as whites; males are about 70% more Household larceny likely to be robbed than females. All households 89% 24% 27 % 38% Urban 93 19 25 49 About 3 out of 4 people will become Suburban 9 0 2 3 26 4 1 victims of a completed or attempted Rural 82 3 1 27 25 assault; 2 in 5 will be victims of this Motor vehicle tbeft All households 19% 17% 2% - crime at least twice. Males are more Urban 27 23 4 -- likely to be assaulted than females, but Suburban 20 18 2 - the likelihood is nearly the same for Rural 11 10 1 - blacks and whites. Note: Data are based on average victimization ratesmeasured by the National Crime Survey for 1975-84. Only completed crimesare included. See box on page 4 for crime definitions. At 1975-84 victimization rates, an - Less than 0.5%. estimated 2 in 5 persons will be injured * as the result of a robbery or assault; theft do not vary by the sex or race of person of age 60 is only about one- about 1 in 10 will be injured more than the victim. quarter as likely as a 30-year-old to once. (Injuries reported by the NCS become a victim of violent crime range from minor to life threatening.) The likelihood of becoining the vic- during the remainder of his or her tiin of a violent crime in the remainder lifetime. Similarly, those who are 30 According to these estimates, of one's lifetime declines more rapidly are five times likelier than those who nearly everyone will be the victim of a than life expectancy, reflecting declin- are 60 to be injured during a robbery or personal theft at least once, with about ing victimization rates with increasing assault over the remaining course of 7 in 8 persons being victimized three or age (table 2). For example, the average their lives. rnore times. In general, findings for remaining life expectancy at age 60 is nearly half as long as at age 30, but a Estimates of the likelihood of a of, for example, height, weight, left- equall probable for all members of the personal theft drop off much less handedness vs. right-handedness, or sign group3 This appears to be true of rapidly with age than do the estimates of the zodiac). household victimization as well as per- for violent crime victimization. Per- sonal victimization. sonal theft victimization probabilities The assumption of equal victimiza- are approximately proportional to re- tion probabilities introduces an error The procedure for processing NCS maining life expectancies. into the esti~natesof long-term and victimization data does not track an lifetine likelihood of victimization. individual or a household from year to Victimization of households 'This is because, even in population year; consequently, the available infor- groups that have been selected to have mation on multiple victimizations is At 1975-84 victimization rates, over homogeneous characteristics, certain restricted to those that take place a period of 20 years almost 3 out of 4 individuals have been found to be vic- during a period of 1 year or less. These households will suffer a burglary; 3 out timized much &norefrequently than single-year data consistently show the of 8 will suffer more than one (table chance alone would predict. The num- pattern of multiple victimization de- 3). Nearly 9 out of 10 households will ber of people in any group who are vie- scribed above, but they are not suitable be victimized by larceny; more than 1 timized two, three, or more times in a for use in calculating long-term or life- in 3 will be victimized by this crime year is substantially greater than would time probabilities. Because of this, three or more times. About 1 in 5 of be the case if victimization were equal victimization probability was as- all households will be victims of motor sumed in the calculations. vehicle theft. Crime definitions from the National Crime Survey For all of these crimes urban house- Assault--An unlawful ~hvsical involved. holds are more likely to be victimized attack, whether aggravated or than those in the suburbs, and victimi- simple, upon a person, including Motor vehicle theft (completed)-- zation is even less likely in rural attempted or threatened attacks Stealing or unauthorized taking of areas. The difference is greatest for with or without a weapon. Ex- a motor vehicle. motor vehicle theft; urban households cludes rape and attempted rape, as Personal theft--Includes both per- are about 2112 times as likely as rural well as attacks involving theft or sonal larceny with contact and households to be victims of this crime. attempted theft, which are classi- personal larceny without contact. fied as robbery. Severity of Personal larceny with contact is Assumptions and their effects crimes in this category range from theft of purse, wallet, or cash by threats to attacks that result in stealth directly from the person of Although the numbers in this report life-threatening injuries. the victim, but without force or are estimates rather than exact mea- Burglary (completed)--Unlawful or the threat of force, including at- sures, they are based upon reliable data forcible entry of a residence, tempted purse snatching. Personal and established mathematical tech- usually, but not necessarily, larceny without contact is theft or niques. They provide a realistic picture attended by theft. The entry may attempted theft, without direct of the order of magnitude of the like- be by force, such as breaking a contact between victim and offen- lihood of being victimized, if past and window or slashing a screen, or it der, of property or cash from any future victimization rates are near the may be through an unlocked door place other than the victim's home levels of 1975-84. Unavoidably, how- or an open window. As long as the or its immediate vicinity. The ever, such estimates are subject to person entering had no legal right property need not be strictly per- error. Aside from the normal sampling to be present in the structure, a sonal in nature; the act is distin- error associated with the use of annual burglary has occurred. Further- guished from household larceny NCS victimization rates, part of the more, the structure need not be solely by place of occurrence. Ex- error results from assumptions that had the house itself for a household amples of personal larceny without to be made in carrying out the compu- burglary to take place. Illegal contact include the theft of a tations. The assumption of constant entry of a garage, shed, or any briefcase or umbrella from a res- rates of victimization over time was other structure on the premises taurant, a portable radio from the discussed earlier. Another key assump- also constitutes household bur- beach, clothing from an automo- bile parked in a shopping center, a tion is that victimization is equally glary. In fact, burglary does not likely for all members of a population necessarily have to occur on the bicycle from a school yard, food premises. If the breaking and from a shopping cart in front of a group, defined by age, race, and sex. In supermarket, etc. In rare cases, other words, any victimization can, entering occurred in a hotel or in a vacation residence, it still would the victim sees the offender with equal probability, be inflicted upon during the commission of the act. any member of the group. be classified as a burglary for the household whose member or mem- Rape--Carnal knowledge through Equal victimization probabilities bers were staying there at the the use of force or the threat of time. force, including attempts. Statu- In a sense, the assumption of equal Household larceny (completed)-- tory rape (without force) is exclu- probability of victimization is what Theft of property or cash from a ded. gives meaning to victimization rates. residence or its immediate vicin- Robbery--Completed or attempted Subgroups of the population are selec- ity. For a household larceny to theft, directly from a person, of ted to make the assumption as valid as occur within the home itself, the property or cash by force or threat possible, by dividing the population in thief must be someone with a right of force, with or without a accordance with characteristics that to be there, such as a maid, a weapon. are believed to correlate with suscepti- delivery person, or a guest. Forci- bility to victimization. That is why ble entry, attempted forcible Source: Based on Criminal NCS uses breakdowns in terms of age, entry, or unlawful entry are not Victimization in the United States, race, sex, income, and the like (instead 1984, NCJ-100435, pp. 123-26. As a result, the estiinates of long- Notes New releases from BJS terin and lifetime likelihoods of victim- ization probably tend to show a slightly '~anrran. Patrick A. and Christopher Automated fingerprint identification larger nuinber of people and households A. hies; The Risk of Violent crime, systems: Technology and policy issues, being victiinized at least once than is Bureau omceStatistics S~ecial NCJ-104342, 4/87 actually the case. That is, the estima- Report, NCJ-97119, May 1985, p. 2; Imprisonment in four countries, BJS ted victimizations are distributed too Report to the Nation on Crime and Special Report, NCS-103967, 2/87 thinly over too large a portion of the Justice, 2nd edition, Bureau of Violent crime by strangers and non- population. (Note, however, that the Justice Statistics, forthcoming, ch. strangers, BJS Special Report, NCJ- exclusion of series victimizations re- 2. The numbers for the general pop- 103702, 1/87 sults in some degree of underestimation ulation were derived from coroner 1986 directory of automated criminal of the proportion of people and house- and medical examiner reports for the justice information systems, NCJ- holds that are victimized at least once.) years 1978-82. A more refined tech- 102260. 1,000 pp., 1/87 nique, including data for 1983, was e probation aidparole, BJS Bulletin, In a 1-year period, more people used to determine the rate for black NCJ-103683. 1/87 actually incur rnultiple victimizations males. Very similar results were Criminal justice "hot" files: Criminal than would be estimated by the equal obtained by the FBI in a study based justice information policy series, 75 probability assumption. Over a life- on homicides reported to the police pp., NCJ-101850, 1/87 time, however, the reverse apppears to in 1978-80; see Crirne in the United Population density in State prisons, be the case. This is because the equal States 1981 (Washington: USGPO BJS Special Report, NCJ-103204, 12/86 probability assumption considers the 1982),ndix V, pp. 339-40. State and Federal prisoners, 1925-85, probability of a person being victimized Simpler methodology can be used in BJS Bulletin, NCJ-102494, 12/86 in any year to be independent of what dealing with homicide than with BJS telephone contacts '87, BJS happens in other years. But in reality, other crimes because there are no Bulletin, NCJ-102909, 12/86 if a person is inordinately susceptible to multiple victimizations; a person can Data quality policies and proce- being victimized, that susceptibility is be killed only once. dures: Proceedings of a BJS/SEARCH likely to continue from year to year. conference, 82 pp., NCJ-101849, 12/86 2~ictimizationrates were taken Teena e victims 16 pp., NCJ-103138, During a lifetime, the person will incur from the annual Bureau of Justice many more than his or her "fair share" i i+ of victimizations; the multiple Statistics publication Criminal Tracking offenders: White-collar Victimization in the United States victims will tend to be the samepeople crime, BJS Special Report, NCJ- for the years 1975-84, with addi- year after year. 102867.-. .. 11/86 tional data from the NCS. Life capi(al punishment, 1985, BJS tables are from the National Center Bulletin, NCJ-102742, 11/86 Single-year data on multiple victim- for Health Statistics, Vital Statistics izations were used to estimate the of the United States, 1980, Vol. 11, magnitude of the error caused by using the equal probability assumption. Sec. 6, Life Tables. Several methods of calculation were 'sparks, Richard F., Hazel G. Green, employed. It was determined that this and David J. Dodd, Surveying error is of the same order of magnitude Victims (New York: John Wiley and as the errors and uncertainties from Sons, 1977), pp. 88-97; Gottfredson, other causes such as unpredictable Michael K., "On the Etiology of changes in annual victimization rates. Criminal Victimization" and Sparks, Richard F., "Multiple Victimization: Further details on methodology Evidence, Theory, and Future Research," The Journal of Criminal A meti~odologicalappendix with fur- Law and Criminology, vol. 72, no. 2 ther details on the calculation of the 11981). findings presented in this report is available on request from BJS by calling Herbert Koppel at (202) 724- ""On Bureau of Justice Statistics technical reports are written principally by BJS staff. This report was written by Herbert Koppel and was edited by Frank D. Balog. Report production was administered by Marilyn Marbrook, publications unit chief, assisted by Jeanne Harris and Arlene F. James.

March 1987, NCJ-104274