Homicides in Chicago, 1965-1995 (Chicago Homicide Dataset), Victim-Level Data and Offender-Level Data
About the Homicides in Chicago Data Set (ICPSR 6399)
The Chicago Homicide Data Set (CHD), distributed by ICPSR as Homicides in Chicago, 1965-1995 (ICPSR 6399), contains detailed information on every homicide in Chicago police records from 1965 to 1995. To learn more about this study, read the overview of the CHD project (PDF 492K) and the ICPSR study description.
To help users better understand this data collection, Dr. Carolyn Rebecca Block, Dr. Richard L. Block, and the Illinois Criminal Justice Information Authority (ICJIA), the principal investigators for this study, have agreed to post answers to Frequently Asked Questions on our website.
Please contact Dr. Block if you have other questions about the study or if you would like further clarification of the answers provided on this Web page. Before submitting new questions, please read all of the questions and answers on this page.
Frequently Asked Questions
Q: What is the difference between the structure of the victim-level data and the offender-level data?
A: The Chicago Homicide Dataset (CHD) consists of two data files, a victim-level file and an offender-level file.
The victim-level data consists of 23,817 records, with one record per victim. For example, one arson homicide incident with ten victims would be represented ten times in the victim-level file. Although offender demographic and relationship variables are included in the victim-level for up to five offenders per victim, you should use the offender-level file to conduct analysis of offenders. There are two reasons why the victim-level file is not accurate for offender-level analysis. 1) For incidents with more than five offenders, information is not included for the additional offenders. 2) For incidents with multiple victims, each offender is double (or triple) counted, depending on the number of victims.
The offender-level data file contains 26,030 records, with one record per offender. It includes a record for each offender known to the police, with no limit on the number of offenders per victim or incident. (The actual maximum in this dataset is 11 offenders.) In addition, there is no "double counting" for incidents with multiple offenders. Each offender appears once per incident. Although victim demographic and relationship information is included for the first victim, you should use the victim-level file to conduct analysis of victims.
Users should be aware that the offender-level CHD contains only homicide incidents in which at least one offender is known to the police. The CHD is periodically updated, and any new information resulting from continuing police investigation is added to both victim-level and offender-level files. However, unsolved homicide incidents will appear in the victim-level file but not in the offender-level file.
In summary, only the offender-level CHD is appropriate for analysis of offender issues, for example, calculating the population-based risk of becoming an offender, or analyzing the proportion of offenders who are juvenile. Only the victim-level CHD is appropriate for analysis of victim issues, for example, calculating the population-based risk of becoming a victim, or analyzing the proportion of victims who are juvenile. It is also possible to use both datasets together because each victim record and each offender record contains identifying numbers that can be used to link one file to the other.
Q: I have downloaded the victim and offender datasets a few times. I seem to be making a mistake because every time I download the offender data it has about 26,000 cases and the victim data always has about 23,000 cases.
A: You have the correct numbers. The number of records differ for several reasons. First, there can be multiple victims per offender, and multiple offenders per victim. Second, for some homicide victims, no offender is ever identified. Therefore, for these victims, there is no offender record in the offender-level CHD file.
Q: It's so complicated to use the victim-level and offender-level files. Why can't I just analyze those homicides that had a single victim and a single offender?
A: If you analyze only single victim/single offender cases, your results will apply only to other single victim/single offender homicides, not to homicides as a whole. If you write a paper that implies that your results apply to all homicides, you would be seriously misleading your audience. A substantial number of homicide incidents have multiple offenders and/or multiple victims, and these homicides are systematically different from single victim/single offender homicides, in aspects such as the offender's age and gender, and the homicide circumstances. Street gang-related homicides and familicides are just two of the many examples. Therefore, in most cases, it would be erroneous to base an analysis on a selection of single victim/single offender homicides.
Q: What homicides are included in the CHD?
A: The CHD contains all murders known to the Chicago police from 1965, with considerable information about each murder. It is based on the Murder Analysis Report (MAR) maintained by the Crime Analysis Unit of the Chicago Police Department (CPD). It includes all homicides known to the police, not only those where an offender is apprehended and convicted, but also homicides that never end in arrest (for example, homicide-suicides), homicides that are later determined to have been justifiable at the prosecutorial or court level, and homicides where the offender is eventually found not guilty. Because the CHD includes murders known to the police, it does not contain cases of concealed death, unless the death was later discovered. However, when such cases are discovered, even years after the death, they are added to the CHD.
Q: Does the CHD include homicides in which the offender commits suicide?
A: Yes. While it is true that suicide is more common among Caucasian (and Latino) males after killing their current or former intimate partner than among African-American male perpetrators, investigators actually are able to rule the death as a homicide and the deceased male as the perpetrator of the homicide. On the other hand, for obvious reasons, correctional datasets of homicide offenders would not include offenders who committed suicide.
Q: What kind of data does the CHD contain on offenders, victims, means of death, and motives?
A:. The CHD is a police-level dataset and reflects information that is important to police investigation. Working closely with the CPD, the principal investigators have added information and detail to the MAR codes, using the narrative. (Note that the narrative does not appear in the archived data.) The CHD provides considerable detail about each homicide. For example, you can separate out elderly victims (defined however you want), and look at weapon, offender's motive, and many other variables. There is also a code for mercy killings and for murder-suicide pacts, and a separate variable for murder-suicide.
Q: Is the CHD updated to reflect newly-discovered information about the homicides?
A: Yes, when the principal investigators add additional years to the CHD, they also review all earlier homicides and edit the data to reflect currently available information. For example, if a homicide has been "cleared" since it was originally coded, the record will be edited to reflect new data about the offender(s) and any updated information about motive or relationship. In addition, new fields were added as the dataset evolved over the years, but these fields were also coded for all the earlier homicides. Almost all codes are consistent back to 1965. A few, for example prior arrest, are not available in 1965 but are coded for all later years. In addition to adding any new information about each homicide, investigators also add new records to the CHD when homicides are discovered at a later date. This may happen after a lengthy investigation (for example, with an arson homicide), when a victim dies months or years after the lethal incident, or when a body is discovered some time after the date of the homicide.
Q: Does the dataset include a Chicago Police Department (CPD) case identification number that could be linked with other data from the CPD?
A: Identifiers have been removed from the archived data. We maintain the original data, but we do not publicly release the data with identifiers.
Q: Does the dataset include detailed information about firearms used in homicides (e.g., make, model, and serial number)?
A: The CHD contains limited information about firearms, such as the type of firearm and caliber. It does not contain serial numbers. For more information, see the weapons variables in the codebook for victim-level data (page 63).
Q: How are the weapon-related data gathered?
A: Weapon in the CHD refers to the cause of death. It is coded on the MAR, and there is often a note in the narrative about the specific cause of death, according to the Medical Examiner of Cook County. Where there was more than one weapon used, the weapon coded is the cause of death. Where there was no weapon, the specific means of death, such as strangulation, "hands and feet" (beating), or "thrown from a height" is coded. In some cases, the cause of death was strangulation or smothering, and the weapon listed is "nylon stocking" or "pillow." There are also specific types of poisoning. For more information, see the CHD codebook.
Q: In the Chicago Homicide Dataset, does the semi-/fully-automatic category for the variable WEAPON (p. 63 of the codebook) contain mostly handguns or long guns (i.e. rifles, shotguns) and is there any way of teasing out which ones are which?
A: If you look at the detailed weapon variables, WEAPCAL and WAUTOMAT, you will see the information we have about the type of automatic or semi-automatic firearm. If you ask someone in your police department or another firearm expert, he or she may be able to tell you the chance that the weapon was a handgun or a rifle, given the available detail. You should also know that we have cleaned the CHD weapon data, and the next archived version will contain some corrections.
Q: What kind of geographic variables does the study contain?
A: The CHD is completely geocoded to address, but for confidentiality reasons, neither the specific address fields nor the X- and Y-coordinates are present in the archived data. Instead, the principal investigators have linked the X- and Y-coordinates to the corresponding Census Tract and Chicago Community Area. In addition, the CPD codes for Police Area and Police District are included from the MAR.
Note that there are relationships among these geographic codes. Generally, the first two digits of the Census Tract are the numeric code for the Chicago Community Area (there are a few exceptions; users should check). Also, the Chicago Police Beat is a four digit number, with the first two digits the Police District (there are multiple beats in a district). In later years, the third digit of the Chicago Police Beat represents the Police Sector (there are multiple beats within a sector). However, there is no relationship between Police Districts, Areas or Beats and Census Tract geography. Note also that there was no change in the Census Tract boundaries in Chicago for the entire period covered by the Chicago Homicide Dataset. Therefore, the tracts are coded consistently throughout. However, the Chicago Police Department did change the boundaries of Police Districts and Areas over time, and added new ones. The incident District and Area is taken from the MAR, and therefore, is not geographically consistent over time. The updated version of the CHD will contain a variable that geocodes each homicide location to the 1993 map of Districts, Areas and Beats (which will, of course, be geographically consistent over time). The Police District and Police Area codes correctly represent the specific Police Department entity that handled the case.
Q: Is there information about social class of victims and offenders in the CHD? If not, are there addresses of victims and offenders, so that we can get neighborhood social class?
A: There is no occupation, income or education information in the CHD, about the victim or the offender. There are, however, codes for Census Tract and Community Area. For more information about demographics in the Chicago Community Areas, see the Community Area Fact Books for 1960, 1970, 1980, 1990, and 2000. Demographics by Census Tract are available from other ICPSR data sets.
Q: Is it possible to identify Latino homicides in the Chicago Homicide dataset?
A: Yes, the racial/ethnic group of all victims and offenders are coded as non-Hispanic black, non-Hispanic white, Latino, and other. In the archived dataset, these codes are consistent from 1965. The original codes in the MAR for some years included different codes, such as codes for specific ethnic groups. Since these codes changed over the years, we have re-coded the racial/ethnic group variables to be consistent over time.
Q: Does the CHD contain day of death data?
A: Yes, there are separate fields for date of injury (year, month, day, day of week, time of day) and date of death (year, month, day). In addition, there is a code for "Book Year," the year that the homicide was recorded in the MAR. In the case of a long investigation, or a homicide discovered well after the incident, the Book Year and the INJYEAR (injury year) will differ.
A: How are the time of death data gathered?
Q: Death time was entered into the MAR by a succession of Crime Analysis Detectives assigned to the job over the years. In some years, however, the forms did not include death time. Also, for some reason, death time (and death date) were not always copied over from the original data files to the archived data files. That is why, in the archived data, death time is missing for some years. We are trying to correct this for the next version of the CHD.
Q: Can these data be used to study homicides of intimate partners, children, parents, or other family members?
A: The dataset has a wealth of information on this, including many aspects of modus operandi. The data contain variables that indicate the relationship between each victim and each offender.
Some researchers, such as Margo Wilson and Martin Daly, have also used the CHD to study familicide. (See: Daly, Martin; Wilson, Margo, "Killing the Competition." Human Nature. 1990, 1, (1), 83 - 109).
Dr. Block also used the CHD for the intimate partner violence study, Community Crime Prevention and Intimate Violence in Chicago, 1995-1998 (ICPSR 3437).
Q: In CHD relationship codes, is common law father/child separate from father or stepfather? Is there a way to determine whether the offender co-resides with the victim? For example, for cases in which the mother's boyfriend is the offender and a child is the victim, can I determine whether the boyfriend lived in the same household as the child? In those cases, which code would I use for the relationship of victim to offender?
A: The CHD does not have a relationship code for "common law father/child." In cases where the man is living with the child's mother, and the MAR indicates "boyfriend," the relationship is coded as stepfather. In cases where the man is not living with the child's mother, or there is no information about co-residence, the relationship is coded as mother's boyfriend. In some cases, where it is not clear whether the man has a romantic relationship with the child's mother, the MAR has a relationship that is not a family member, but the man lives with the family, the relationship is coded as "lives together as family." In cases where we have no information about living arrangements or about the relationship of the man to the child's mother, but the narrative clearly indicates that the man was caring for the child during the incident, the relationship is coded "caretaker, other or unknown." In cases where we have no information about living arrangements or about whether the man is caring for the child, and the MAR indicates that the mother knew the man, the relationship is coded "acquaintance of parent, other or unknown." In the next version of the CHD, there will be a field for "co-reside."
Q: Does the CHD include any homicides where fathers killed their children and the childrens' mother and then killed themselves?
A: Yes the CHD contains such cases. Use the victim-level file, look for multiple-victim incidents in which the victim relationship was a female intimate partner and young children were also killed, and look for cases in which the offender committed suicide. Be careful about the offender's relationship to the children. Biological versus step fatherhood is coded where known, but it is not always known.
Q: What is the difference between the CHD codes "child abuse (causative factor)" and "child abuse (type of homicide syndrome)." Does one refer to a history of child abuse?
A: Data in the CHD refer to the specific lethal incident. Where Child Abuse is coded as a causative factor, the child was killed while being physically abused. Some young children may be killed in gang crossfire, or in a home invasion robbery, or in many other circumstances that would not be coded child abuse. We do not know whether there was a history of child abuse.
Q: Is there literature on homicide committed by the elderly that is connected to this study?
A: Publications on the CHD include population-based risk rates up to age 85. See, for example, "Lethal Violence in the Chicago Latino Community: 1965-1989." pp. 267-343 in Homicide: The Victim-Offender Connection, edited by Anna Victoria Wilson. Cincinnati: Anderson Publishing Co., 1993.
Q: Does this study contain data on child-on-child homicides and whether these homicides led to official policy responses?
A: The Chicago Homicide Dataset does not contain this kind of "system outcome" detail. It does, however, contain basic offense information about 22 homicides occurring from 1965 to 1995, in which at least one offender was 11 years or younger. Updated figures show one other case in 1994, and two in 1993. The 22 homicides were committed by a single offender in 19 cases, two offenders in one case, five offenders in one case, and six offenders in one case. In total, there were 29 offenders age 11 or younger and six older offenders (age 12 to age 19) in the 22 homicides. Of these 29, the youngest was age 5, two were age 7, eight were age 9, seven were age 10 and seven were age 11.
Q: I have a question regarding the CHD and coding of the victim's age. For those cases in which the child (victim) is under 23 months, is it possible to find out their specific age (i.e., 3 hrs old, 1 week, or 6 months)?
A: This information is not in the currently archived CHD, but it is being added to the next version.
Q: Is it possible to get age of arrested suspect by year for sexual assault of male and female cases for the archived Chicago data?
A: Yes. Use the offender-level CHD. You will have the age of each offender. Under MOTIVSEX, select either "sexual assault of male" or "sexual assault of female." If you want only "arrested suspects," select "cleared by arrest" under CLEARED. Unfortunately, the archived CHD does not indicate which of multiple offenders was arrested. However, this information will be in the next version.
Q: Can I use the CHD to look at trends over time?
A: In both the victim-level or the offender-level CHD, you can look at data by year, by month within year, or by day of the month within month within year, keeping in mind that homicide (fortunately) is rare, and the number per day or even per month will be small.
Q: Does the CHD contain details of the homicide, such as number of times stabbed or location of the injury?
A: We have recently begun to collect more detailed injury data, but the archived dataset doesn't have it yet.
Q: In the variables on the prior arrest record of the victim and the offender(s), what is the meaning of "connected?"
A: "Connected" means that the person has a prior arrest record that includes at least one arrest for a violent offense, such as assault, robbery, rape or homicide. Sometimes the specific prior arrest charges are noted in the narrative, but they are not reflected in the archived data.
Q: How is the CAUSFACT field in the CHD coded?
A: CAUSFACT is initially coded directly from "Causative Factor" in the MAR. The principal investigators then check the narrative, and add a second causative factor (CAUSFAC2), and more detailed categories where indicated. The rule of thumb for coding Causative Factor is that it must be relevant to the homicide incident. For example, if the narrative mentions that the victim was a prostitute, drug dealer, or gang member, but there is no indication that the victim's status was related to the lethal incident, it is not coded. For example, a domestic fight between a prostitute and her boyfriend would be coded as any other domestic altercation, but if the boyfriend is her pimp and beats her to death for refusing to work, then the Causative Factors should be "Pimp doing business" and "Extortion."
Q: How is the CAUSFACT category "Mental Illness" coded?
A: "Mental Illness" is coded on the MAR when police investigation shows that the mental illness of the offender contributed to the homicide. Narratives in typical cases state that the offender had "a history of mental illness" or that the offender "heard voices" telling him to kill. The code is not based on any medical examination of the offender.
Q: The data identify some offenders as cartage thieves. What is cartage theft?
A: Cartage theft occurs when someone steals goods off a truck.
Q: Does NACJD have other studies that provide homicide-related data about specific cities?
A: Studies in the NACJD holdings that provide homicide-related data for specific cities include:
Los Angeles Homicides, 1830-2001 (ICPSR 3680)
Nature and Patterns of Homicide in Eight American Cities, 1978 (ICPSR 8936) -- the eight cities are Philadelphia, Newark, Chicago, St. Louis, Memphis, Dallas, Oakland, and Ashton
Social Order in Middletown, 1932-1975 (ICPSR 9058)
The following city-specific studies provide homicide rates in regard to homicide:
Crime Changes in Baltimore for 1970-1994 (ICPSR 2352)
-rates for 1970-1992
Anticipating and Combating Community Decay and Crime in Washington, DC, and Cleveland, Ohio, 1980-1990 (ICPSR 6486)
- rates for Washington, DC for 1980, 1983, 1988, and 1990
- Cleveland for 1980 through 1989
Testing Theories of Criminality and Victimization in Seattle, 1960-1990 (ICPSR 9741)
- rates for 1960-1990
Q: Is there an update to the Chicago Homicide Dataset?
A: We have added the years through 2000 and are currently preparing them for archiving.
Q: Can you provide a list of publications related to the CHD?
A: Please see the citations database for this study.
Contact Information for Principal Investigator
Carolyn Rebecca Block, Ph.D.
Illinois Criminal Justice Information Authority
120 South Riverside Plaza
Chicago, IL 60606
FAX: (312) 793-8422