Since the topic of alcohol abuse and its attendant social harms are a continuing part of the public conversation in Alaska, this article will look at the various secondary data sources that can be used in examining effectiveness of alcohol control policies. A number of agencies retain data relevant to the impact of public policies involving alcohol, but not all the data collections are solid—reliable and valid—for research purposes. Considered below are the data collections assembled by the police, public health agencies, and other entities such as the state child protection and transportation departments. In addition to describing each data source, this article discusses each in terms of whether it provides statewide coverage, the extent to which it is complete, and the level of geographic and temporal specificity it provides.
Data collections for examining alcohol control efforts would ideally have the following characteristics: they would be complete; they would measure associated harms at the community level; they would be available for many years; and they would allow for measurement of those harms on a weekly or monthly basis in order to capture the time series effects of changes in policy. In Alaska, four different sources meet these requirements to a useful extent: the Alaska Trauma Registry, death certificate records from the Bureau of Vital Statistics, reports of harm from the Office of Children’s Services, and traffic accident records from the Department of Transportation. Data maintained by law enforcement agencies are, in general, less useful when considered as a means for understanding the impact of public alcohol policies.
Before examining the useful data sources, this article will first consider the flaws in available police statistics.
Police Data Sources
There are two possible sources of police statistics relevant to understanding the harm that results from the use of alcohol in Alaska—the Uniform Crime Reports (UCR) and calls-for-service records kept by individual police agencies. These sources share one major shortcoming: although each is useful for understanding the harms that occur in some jurisdictions in Alaska, neither is particularly useful for making inter-community comparison measurements on a statewide basis.
Uniform Crime Reports
The Uniform Crime Reports are a standardized record of offenses known to the police. For agencies in Alaska, these records are available from the Inter-university Consortium for Political and Social Research (ICPSR) for the years 1975 through 2001. The UCR provided by the ICPSR are available on a month-by-month basis in 27 annual data sets. The UCR includes measures of the so-called Part I offenses—the serious criminal offenses of murder (and non-negligent manslaughter), aggravated assault, forcible rape, robbery, burglary, motor vehicle theft, and larceny theft. Also included are records of the number of simple assaults recorded per police agency per month. For all of these Part I offenses and simple assaults, the UCR record the number of offenses that have come to the attention of the police and are considered to have actually occurred—that is, the reports are not “unfounded.” The Part I offenses are recorded whether or not an arrest is made for the offense.
For less serious offenses—generally referred to as the Part II crimes—the UCR only provide records of police arrests. These Part II crimes include many offenses often associated with the use and abuse of alcoholic beverages: driving under the influence, violations of liquor laws, public drunkenness, disorderly conduct, and vagrancy. The ICPSR has available month-by-month UCR arrest records for police agencies in Alaska only for the years 1998 through 2001. For earlier indicators of arrests for Part II offenses by Alaskan police agencies it is necessary to refer to the annual Crime in Alaska reports issued by the Alaska Criminal Justice Planning Agency for the years 1976 through 1987 or to the annual Crime Reported in Alaska reports issued by the Alaska Department of Public Safety for the years 1988 through 2002. The arrest statistics presented in these two reports are given by individual agency. Available in hard copy through 2001 and online for 2002, the volumes can be obtained at University of Alaska libraries in Anchorage or Fairbanks.
While many of the Part I offenses reported in the UCR may be associated with alcohol use, in general the information recorded does not permit a determination of the extent to which alcohol use is actually involved in crimes occurring in a given jurisdiction. The one exception to this rule is the Supplementary Homicide Reports (SHR) assembled by the FBI. Also available through the ICPSR for the years 1975 through 2001, the SHR provide incident-based information on all criminal homicides known to the police. Included in the SHR are the date of the offense, victim and offender demographics, information on the victim-offender relationship, weapons used, and, for the purposes of understanding the role of alcohol in criminal violence, information on whether the circumstances of the homicide involved a brawl due to the influence of alcohol.
Other than for homicide, most criminologists are reluctant to use UCR data as a measure of how much crime there is. They see the UCR data as lacking measurement validity (because the statistics only measure a fraction of the crime that occurs in society) as well as reliability (because the amount of crime that actually occurs that is eventually measured by the UCR varies from jurisdiction to jurisdiction). In addition to these usual misgivings, there are other reasons why the UCR statistics lack utility when attempting to understand inter-community variations in behaviors often associated with alcohol in Alaska. First of all, the UCR are limited by a lack of geographic specificity in Alaska. Secondly, the Alaska UCR data are somewhat incomplete because many police agencies, especially those from the less urbanized areas of the state, provide information to the FBI on a sporadic basis.
The lack of geographic specificity results from the fact that the reports are made at the level of the individual police agency rather than for specific geopolitical areas. For instance, the UCR contains those crimes known to the North Slope Borough Police Department—a broad level of aggregation. It is impossible to determine which offenses occurred in Barrow, the borough seat, and which offenses occurred in the seven other communities scattered across the approximately 90,000 square mile jurisdiction. This problem is even more conspicuous for those communities policed by the Alaska State Troopers (AST)—about a third (32%) of the state’s population, as shown in Figure 1. Although these crimes are recorded in the UCR as offenses known to the Troopers, most have been reported to the Troopers from one of roughly 90 second-class cities or a similar number of unincorporated villages. Aggregation of these police statistics at a level above that of individual cities and villages masks theoretically important inter-community differences in the amount of harm associated with alcohol.
The other problem with the use of Alaska UCR data is that the data are incomplete on a time series basis. Many agencies, particularly those serving jurisdictions with smaller numbers of residents, report to the FBI only on a sporadic basis. The proportion of the Alaska population covered by police agencies that reported UCR statistics for a complete 12-month-year for the years 1975 through 2001 is shown in Figure 2. Overall, UCR crime statistics were reported to the FBI for a complete 12 months for most of the Alaska population (annual mean = 94%) during that time period. However, when the proportion of the population for which UCR offenses were reported to the FBI is considered for all police agencies except for the state’s two largest departments (i.e., the AST and the Anchorage Police Department), the extent to which the data are incomplete on a time series basis becomes apparent. Excluding the population served by the Troopers or the Anchorage Police Department, between 1975 and 2001, UCR crime statistics were reported to the FBI on a complete 12-month basis by agencies serving roughly four-fifths (annual mean = 78.5%) of the population. Particularly after 1994, these police agencies were even less likely to report.
Illustrating the problem of incomplete UCR data in the ICPSR holdings, the North Slope Borough Police Department (NSBPD) did not report for any months during the three years 1996 through 1998. This gap is particularly unfortunate for those wishing to understand the effects of local alcohol control policies. Residents of the North Slope hub community of Barrow voted to allow alcohol possession in October 1995, then voted to ban possession in February 1996, only to allow possession again in November 1997. A valid examination of the effect these changes had upon alcohol-related violence across the North Slope Borough would have provided compelling evidence regarding the effectiveness (or ineffectiveness) of local prohibition.
Other police agencies serving hub communities in northern and western Alaska are also infrequent in reporting UCR records to the FBI. For example, the Unalaska Department of Public Safety, an agency serving a primary fishing port and the hub for the Aleutian Islands, went 17 years without reporting. The Nome and Kotzebue Police Departments also have not reported UCR statistics for many years. For understanding the effects of alcohol control policies, it is unfortunate that the police agencies in these two communities have not reported recently because they otherwise present potentially interesting comparisons. With similar population sizes and traditional indigenous heritages, one primary difference between the two communities is that Nome allows the local sale of alcohol and has a number of bars while Kotzebue forbids the sale of alcohol within city limits. Without complete UCR data, it is less possible to determine if the differences in alcohol control policies in the two communities translate into differences in rates of violent crimes associated with alcohol use.
Calls-for-Service Records / Incident Reports
Calls-for-service records and incident reports generated in the day-to-day operations of individual departments provide a potential alternative to using the Uniform Crime Reports statistics. As might be expected, the availability of these records varies considerably along the lines of agency size and departmental resources. There are also substantial differences in the validity and reliability of these calls-for-service records.
The availability of calls-for-service records from the agencies (not including police departments responsible for airports and universities) recognized by the Alaska Police Standards Council (APSC) in the latter part of 2003 is mixed. Although no systematic assessment of these agencies’ record-keeping and dissemination capabilities has been performed, it is probably safe to assume that, given the broad range of population sizes they serve, some departments are more able than others to be able to provide data for the purposes of studying violence and other offenses associated with alcohol use. The larger departments in the state, particularly those in Anchorage and Fairbanks, employ computer-assisted dispatch technologies that allow for the production of geographically and temporally specific data records amenable to a broad range of analytical techniques. Use of these records in research is limited mostly by the willingness of the agencies to share the information and by the ability of record-keepers to manipulate proprietary software programs to produce case-based data sets.
The incident reports produced by the Alaska State Troopers are one good example of the records that could be useful in alcohol research. Their records are available for the time period since 1990 from their headquarters in Anchorage, with each record listing the location, date, and offense (if any) for every call for service received by the Troopers. One important benefit of the Troopers’ incident reports is that they allow for an examination of violent behavior and liquor law violations in all areas of the state not covered by municipal police departments. This coverage would include most, but not all, of the Alaska Native villages located in the more isolated areas of the state.
The availability of case-based calls-for-service records from smaller police departments is probably much more limited. An analysis of information drawn from the Census of State and Local Law Enforcement Agencies (CSLLEA) conducted by the Bureau of Justice Statistics in 2000 shows that many small departments in Alaska probably lack the resources to provide calls-for-service information in a format that would be useful for research purposes. For instance, 6 of the 42 APSC member departments responding to the CSLLEA had no full or part-time civilian employees, while 7 departments had only 1 full or part-time civilian employee. Typically in the more isolated areas of the state that are thought to be especially troubled by the harms associated with alcohol use, these departments are rather small, with an average of four sworn officers per department responsible to a mean population of 1195 residents.
Even the calls-for-service records for the larger police agencies more likely to be available for use in alcohol research are potentially beset by problems of measurement validity and reliability. One problem is that these records, as with most police crime statistics, are social artifacts that are often more of an indication of factors such as police presence and variations in the public’s willingness to report offenses than a realistic measure of an underlying amount of criminal behavior.
Public Health Data Sources
An alternative to Alaska police statistics in considering the effects of alcohol control policies are the data available from the state’s public health authorities. The two main data sources are the records of traumatic injuries recorded in the Alaska Trauma Registry and the death certificate records compiled by the Alaska Bureau of Vital Statistics. The primary advantage of the public health data sources is that their records are not subject to the biases of incomplete data and differential reporting associated with the statistics produced by the police.
Alaska Trauma Registry
Maintained within the Alaska Division of Public Health, the Alaska Trauma Registry is a standardized, statewide record of all injuries resulting in a hospital admission or declaration of death in an emergency room. The data in the Alaska Trauma Registry have been gathered since the beginning of 1991 for all 24 acute care hospitals in the state. Information from the Alaska Trauma Registry is currently available in electronic format at the level of the individual trauma case for the years 1991 through 2000. Aggregation of individual trauma cases information in the Alaska Trauma Registry provides for city/village level indicators of harms associated with alcohol use, such as assault, suicide, and unintentional injuries, which are available across multiple time frames. The cases in the Alaska Trauma Registry could be used to consider the effects of a broad range of alcohol control policies and efforts directed toward the reduction of alcohol-related harm. For instance, a city could examine the impact of its programs aimed at reducing underage drinking by following trends in injuries to residents under the age of 21 years.
A broad range of information concerning each individual trauma case is recorded in the Alaska Trauma Registry. This includes information on treatment provided at the injury scene, by those transporting the patient, and by emergency room physicians. It also includes data on case outcomes such as the extent of disability resulting from the injury and whether the case ended in death. For the purposes of studying the effects of efforts to deal with alcohol-related problems, the most important information in the Alaska Trauma Registry involves (1) patient demographics, (2) injury specifics, and (3) alcohol and other drug screening. The demographics in the Alaska Trauma Registry include specifics on each patient’s age, sex, race/ethnicity, and home address. Injury details include the cause-of-injury E-code, measures of injury severity, and records on the date, time, city/village, and place (e.g., home, work, or public building) of injury. Records of the results of blood alcohol content as measured by blood test or breathalyzer are also available for a selected sample of cases.
A great deal of care has been taken to insure that the Alaska Trauma Registry provides a valid representation of traumatic injury in the state. Validation studies have been conducted on three different occasions to determine the extent to which cases of traumatic injury actually were recorded in the registry. Ninety-one percent of potential cases in 11 hospitals in 1994, 87 percent of potential cases in 6 hospitals in 1998, and 90 percent of potential cases in 11 hospitals in 1999 were eventually recorded in the Alaska Trauma Registry database; hence, although there are some gaps, the information that is recorded in the database is generally complete.
Bureau of Vital Statistics Death Certificate Records
Death certificate records compiled by the state Bureau of Vital Statistics (BVS) form another public health data source that can be used to examine the effects of alcohol control policies in Alaska. These records have value as a single data set in examining the impact of control policies and also can be combined with the records of the Alaska Trauma Registry for a complete data set to capture all cases of injury, whether or not they resulted in emergency medical care. Included in the BVS death certificate records are indicators of the dates of death and injury, city/village of injury, city/village of residence, decedent’s age, sex, and race, and the cause-of-injury E-code. They are available on a case-by-case basis with records going back as far as 1980.These death certificate records have been used in a number of studies on the effectiveness of alcohol policies in Alaska. Matthew Berman, a professor with the UAA Institute of Social and Economic Research, used death certificate records to examine the effects of local alcohol prohibition in isolated Alaska Native villages. A similar study, using cases from both the death certificate records and the Alaska Trauma Registry, was conducted by the author of this article to consider the impact of police presence upon traumatic injury in Alaska Native villages where alcohol is prohibited.
There are, however, some limitations to the BVS death certificate records as measures of alcohol-related harms. First, unlike the Alaska Trauma Registry, the death certificate records do not include records for non-Alaska residents. Another problem is that, except for those cases with specific alcohol-related cause-of-death E-codes, the BVS death certificate records do not record whether the death was alcohol or drug-related. It is necessary to examine individual death certificates to obtain this information. The one other shortcoming of the BVS death certificate records is that they are incomplete on a number of measures. For deaths by traumatic injury, the death certificate records are missing the place of injury in a tenth (9.9%) of cases and the date of injury for roughly one of seven cases (14.1%). Otherwise, the records of violent deaths are very comprehensive.
Less Useful Public Health Data Sources
Apart from the Alaska Trauma Registry and the BVS death certificate records, the other statewide data sources that could possibly be used for the study of alcohol-related harmful behaviors are largely incomplete or limited to certain subsets of the population. Among these less useful sources are Emergency Medical Services (EMS) records and hospitalization records. The state only began to catalog EMS records in 2001, and these records are not yet available for analysis. Once available, however, they might be preferable over Alaska Trauma Registry records because they would record those injuries that result in medical care but do not require an overnight admission to the hospital (as is the case with the trauma registry data). There is no statewide set of records for hospitalizations for all Alaskans of all races/ethnicities. Such records would have to be obtained on a hospital-by-hospital basis. However, hospitalization and medical system records for Alaska Natives, roughly 18 percent of the state population, are recorded in a centralized data system—the Resource and Patient Management System (RPMS) kept by the Indian Health Service. The extent to which these records are available for research is unknown. Accessing them would require the approval of the IHS, the Alaska Native Tribal Health Consortium, and other entities.
Other Data Sources
A couple of other data sources providing statewide coverage also allow for inter-community comparisons of problems associated with alcohol abuse. These include records of harm to children from the state Office of Children’s Services (formerly Department of Family and Youth Services) and records of traffic accidents from the state Department of Transportation.
Reports of Harm to Children
The Alaska Office of Children’s Services (OCS) records contain information on reported incidents of harm to children and of substantiated cases of harm to children. These reports and cases are recorded in a statewide, standardized, centralized data system in the OCS office in Juneau. It is possible to obtain village/city level records of reported incidents of harm and of substantiated cases of harm through the OCS data system.
The OCS database contains cases going back to 1990, but according to OCS’s data manager, these reports are reliable only from 1994 forward. Records of cases of neglect, physical abuse, mental injury, abandonment, and sexual abuse are available. The temporal and geographic detail of the OCS reports available for research is limited mainly by concerns over confidentiality and the willingness of the state to make the data available. While it is possible for OCS to produce a case-based data set that would permit the examination of any preferred time at the village/city level, concerns over confidentiality make it more likely that aggregated village/city level reports on a monthly basis for each of the five different types of child mistreatment could be obtained.
Three possible outcome measures for each of the five types of harm to children can be obtained to examine the impact upon child welfare of efforts to deal with alcohol-related problems. The first of these, the reports of harm, include each report of an incident of harm, substantiated or not, that comes to the attention of the OCS. They are reports of incidents and, especially in the larger offices, may include duplicate reports of a single incident. For instance, if a school nurse and a neighbor both reported the same incident of physical abuse for the same child, staff in the larger local OCS offices might not recognize that the reports are based on a single incident, thus inflating the number of incidents. A second outcome measure available from OCS is the substantiated case—those reported incidents of child harm found to be true. The records of substantiated cases are less likely to have duplications and hence are a more valid indicator of actual cases of harm to children. However, as with the records of reports of harm, the records of substantiated cases can include multiple referrals on individual children. For example, there would be two substantiated cases for a child that was harmed in August and then in, say, October. Likewise, there would be two substantiated cases of harm for a child that was both neglected and physically abused. The third outcome measure available from the Alaska OCS—the number of children harmed—gives us less of an idea of how much harm is occurring in a given jurisdiction but does help to avoid the problems of duplicate reports and multiple referrals.
A preliminary analysis of the OCS records of harm to children seems to indicate that the records are reasonably complete. For example, for the reports-of-harm measure the village/city-of-referral variable was missing in only 19 out of nearly 175,000 reports of harm in the 11 year period 1990-2001. During that same period the village/city of referral variable for the substantiated case measure was missing in about 2 percent (1.87%) of 53,586 substantiated cases. Given that the OCS records are systematically gathered by a single state agency, it is sensible to expect similarly high levels of completeness on other variables in the database.
Traffic Accident Records
The other statewide, standardized data source that would be amenable to an analysis of the effects of alcohol policies is the traffic accident records maintained by the Alaska Department of Transportation (DOT). This data source contains records of all reported motor vehicle traffic accidents occurring on any public road in the state. Traffic accident records from as far back as 1977 are recorded by the DOT. Included in the DOT’s Highway Analysis System (HAS) are records on each accident for location (including city/village and mile marker for outside of inhabited areas), date, time, alcohol relatedness, severity (i.e., property damage only, minor injury, major injury, or fatality), pedestrian involvement, and vehicle type (i.e., auto, truck, ATV, snowmachine). The information recorded in the HAS database is provided to the DOT by the state Department of Motor Vehicles (DMV).
The DOT’s traffic accident records would be useful for examining the effects of a number of alcohol policies. Obviously, they would be required for any consideration of policies aimed at reducing drunk driving. They could also be used, for instance, to examine the effects of local regulations of drinking establishment closing times.
The completeness and coverage of the HAS record of motor vehicle accidents is limited by a couple of factors. First, accidents that go unreported to the police and those for which the police fail to notify the DMV are not included in the data. Secondly, the accident database only includes incidents that happen on public roads, so many of the snowmachine and ATV accidents occurring in roadless areas of the state are not included.
Without data in hand to analyze, it is difficult to determine exactly how well the more isolated areas of the state are covered in the accident data. There are, however, a few reasons to believe that motor vehicle accident records receive less than full statewide coverage in the HAS database. One problem is that while many Alaska Native villages have roads used by the general public, not all of these roads are recognized in the state’s records. For instance, only two of the eight villages in the North Slope Borough (Barrow and Point Hope) are represented in the route lists generated by the HAS database. Another indication that there might be less than complete coverage of the isolated areas of the state is that very few traffic accidents are recorded in the HAS database as occurring in isolated places off the main road system or Alaska Marine Highway. According to the latest report on traffic accidents in Alaska, in 2001 only 240 out of 15,273 traffic accidents (1.6%) were reported as happening in non-connected, isolated census areas. It is possible, however, that these small numbers reflect the fact that residents of isolated Alaska Native villages just use motor vehicles much less and therefore have fewer accidents.
To summarize—relatively few statewide, standardized data sources in the state of Alaska are solid and extensive enough to ground a fruitful and valid consideration of the effects of variations in local alcohol policies. Statistics generated by local and state police, where they do exist, are generally inadequate for this purpose. The four data sources that appear to offer potential for the study of the harmful effects of alcohol use in Alaska are the collection maintained by the Alaska Trauma Registry, the Bureau of Vital Statistics death certificate records, the client and incident database at the Office of Children’s Services, and motor vehicle accident records from the Department of Transportation. Each of these databases is centralized, complete, geographically specific to the village/city level, and allows for the measurement of alcohol-related harms in time frames as narrow as the single date. Use of any one of these data sets could facilitate obtaining a clearer understanding of what impact, if any, specific alcohol control policies have upon the harm that often results from alcohol use in Alaska.The research conducted for this paper was made possible by a Graduate Research Training on Alcohol Problems Fellowship, sponsored by the National Institute on Alcohol Abuse and Alcoholism grant number T32AA0724026. This research was also supported by National Institute of Alcohol Abuse and Alcoholism Center grant number AA06282 to the Prevention Research Center, Pacific Institute for Research and Evaluation, Berkeley, California.
Darryl Wood is an associate professor with the Justice Center. He is currently on a sabbatical appointment as a research scientist at the Prevention Research Center, Pacific Institute for Research and Evaluation, Berkeley, California.