Subject: Drunk Driving Drunk Driving September 1992 NCJ-134728 Full text with tables available from: BJS Justice Statistics Clearinghouse 1-800-732-3277 Box 6000 Rockville, MD 10850 In the 1989 Bureau of Justice Statistics survey of inmates in local jails, offenders charged with or convicted of driving while intoxicated were more than 1 in every 11 inmates. Among convicted inmates, 86% of those serving a sentence for DWI had been sentenced in the past. Almost a third of the DWI inmates had served 3 or more previous sentences in jail or prison. Among all who are arrested for driving while impaired, persons in jail for DWI are more likely to be the serious offenders. During 1989 over 1.7 million drivers were arrested for driving under the influence. In discussing the population of DWI arrestees, this report examines the trends through the 1980's for arrest rates and changes in the age distribution of licensed drivers. The first BJS Special Report on drunk driving, based on the 1983 survey, also presented the trends in rates, the effects of legislative changes before 1986, and the drinking patterns of inmates. That report had observed an early trend in the reduction of drunk driving arrests among persons under age21-- a trend associated with initiatives like raising the legal age to buy alcoholic beverages. While the annual monetary costs to society from drunk driving remain in the billions, this report presents findings that suggest the positive effects of concerted legislative and law enforcement efforts. Steven D. Dillingham, Ph.D. Director Drunk Driving By Robyn L. Cohen BJS Statistician More than half the persons in local jails charged with the offense of driving while intoxicated with alcohol (DWI) in 1989 had prior sentences to incarceration for DWI offenses. About 1 in 6 persons jailed for DWI had served at least three prior sentences in jail or prison for drunk driving. This report examines the characteristics of persons who were confined in local jails in 1989 and who had been charged with DWI. The findings were obtained from the 1989 Survey of Inmates of Local Jails, which gathered extensive data from interviews with a nationally representative sample of 5,675 inmates in 424 jails during the summer of 1989. The sample was drawn to represent an estimated 395,000 jail inmates in 3,312 local jails on June 30, 1989. The Bureau of the Census carried out the interviews for the Bureau of Justice Statistics. This report also analyzes recent trends in arrests for driving under the influence of alcohol or other intoxicants (DUI). Data on arrests for DUI were drawn from Uniform Crime Reports provided by State and local police agencies to the Federal Bureau of Investigation (FBI). Other major findings include the following: *Between 1980 and 1989 the number of arrests nationwide for DUI increased nearly 22%, while the number of licensed drivers increased 14%. *Over the period from 1980 to 1989 the number of DUI arrests per 100,000 licensed drivers grew by nearly 7% from 982 per 100,000 drivers to 1,049. * Since 1983 all States that permitted the sale or purchase of alcoholic beverages to persons under age 21 have phased in new laws raising the minimum age to 21. Per capita arrest rates for DUI for persons age 18 to 20 have decreased by 21%since then -- more than twice the rate of the decrease among those age 21 to 24 (9.9%). *On June 30, 1989, about 9% of all persons confined in local jails were charged with or convicted of DWI. *In 1989, 96% of persons in jail for DWI were male; their median age was 32; and they reflected a racial distribution similar to the adult general population. At the time of their arrest, more than 70% were not living with a spouse and 78% were employed. *Nearly 9 out of 10 jail inmates (86%) charged with or convicted of a DWI offense had a prior sentence to probation, jail, or prison for a DWI offense or other offense. *Of convicted DWI offenders in local jails, 61% reported drinking only beer, about 2% only wine, 18% only liquor, and 20% had been drinking more than one type of beverage prior to their arrest. *When the type and amount of beverages are converted into equivalent units of pure alcohol (ethanol), convicted DWI offenders who reported drinking more than one type of beverage consumed nearly three times the quantity of ethanol of those who drank only beer. *Prior to their arrest for DWI, half of the convicted offenders in jails were estimated to have consumed at least 6 ounces of ethanol (about equal to the alcohol content of 12 bottles of beer) in about 5 hours. About 29% reported that they had consumed at least 11 ounces of ethanol (equivalent to about 22 beers) prior to their arrest. *Those jail inmates convicted of DWI who consumed greater than average quantities of ethanol prior to arrest reported a greater frequency of usual drinking sessions, as reported by the inmates, and greater consumption of alcohol than other inmates. *For DWI offenders sentenced to jail, the median term imposed was 6 months; those with 2 or more prior DWI sentences received sentences that were more than 1 + times as long as first-time DWI offenders. *About 80% of all inmates in jail for DWI who admitted to being an alcoholic had previously been involved in an alcohol abuse treatment program. DUI and DWI defined DUI is the general term for drivers who operate a motor vehicle after having consumed an intoxicant (such as drugs or alcohol); DWI, in this study, specifically refers to jail inmates who were charged with driving while intoxicated by alcohol (usually defined by State law as a specific concentration of alcohol in the blood). Legislative changes As a result of changes in Federal highway legislation in 1983, States began to phase in new laws raising the minimum drinking age, as defined by the minimum age for which purchase of alcoholic beverages is legal. By 1989 all States had a minimum drinking age of 21. Approximately 18 States do not have a minimum age for consumption of alcohol, and 2 States do not have a minimum age for possession of alcohol. (See A Digest of State Alcohol- Highway Safety Related Legislation (Washington: National Highway Traffic Safety Administration, 1991).) In addition, 29 States and the District of Columbia have adopted legislation that requires administratively imposed sanctions for all persons who, when asked, fail or refuse to take a test measuring the presence and concentration of alcohol. (Data were obtained from a survey of the States conducted by the Nebraska State Senate Transportation Committee for yearend 1990.) Such statutes permit law enforcement officers to immediately confiscate the driver's license of persons arrested for DUI who fail or refuse to submit to alcohol testing. Trends in DUI arrests Between 1980 and 1989, the number of arrests for DUI increased 21.7%, compared to an increase of 13.9% in the number of licensed drivers (table 1). Overall, the number of DUI arrests per 100,000 licensed drivers increased 6.8% over the period, from 982 to 1,049. The difference between the arrest rates for those age 18 to 20 and 21 to 24 steadily increased between 1980 and 1989 (figure 1). In 1980, arrest rates were 1,757 and 1,784, respectively, for drivers age 18 to 20 and 21 to 24 - - a difference of 27 per 100,000 drivers. After 1980, the arrest rate of 18-to-20-year-olds decreased while the rate for 21-to-24-year-olds increased. In 1989, the rates per 100,000 drivers were 1,607 and 2,183, respectively -- a difference of 576 per 100,000 drivers. Arrest rates for those age 18 to 20 peaked in 1982; for 21-to-24-year-olds, the rates peaked a year later. The number of DUI arrests of those age 18 to 20 decreased by 33% between 1983 and 1989 (from 216,255 to 144,800). The number of licensed drivers in this age group declined by 15% (from 10.6 million to 9.0 million). More than half of the decline in the number of arrests between 1983 and 1989 among drivers age 18 to 20 (and as much as 22% of the decline in arrests for all ages between 1983 and 1990) could possibly be linked to changes in the drinking age laws. (This estimate was calculated by applying the 1983 arrest rate for those age 18 to 20 (2,043 per 100,000 drivers) to the number of drivers in 1989 (9,009,821) of this age group, producing an estimate of 184,074 arrests in 1989. Actual arrests in 1989 were 144,800 or 39,274 fewer than expected. The overall decline in the number of arrests between 1983 to 1989 was 71,455 (216,255 - 144,800); the percentage of the decline not due to a change in the number of drivers of these ages would be more than half (39,274/71,455). The total decline in the number of arrests for persons of all ages between 1983 and 1989 was 180,947. Thus, as much as 22% of the drop (39,274/180,947) would be attributable to changes in the prevalence of arrests of 18-to-20-year-olds.) Between 1983 and 1989, arrest rates for age groups 21 or older declined at a slower pace than those for drivers age 18 to 20. Arrest rates among licensed drivers age 18 to 20 declined more than twice as fast as arrest rates for those age 21 to 24 between 1983 and 1989 (21% versus 10%). DUI arrests in 1989 In 1989 more than 165 million persons held a driver's license in the United States -- nearly 86% of the population age 16 or over. The FBI estimated that during the same year more than 1.7 million DUI arrests were made by State and local police agencies. In addition 45,555 motor vehicle fatalities occurred; about 49% were probably alcohol-related, according to the National Highway Traffic Safety Administration. (See Fatal Accident Reporting System, 1989: A Decade of Progress, National Highway Traffic Safety Administration, 1990, p.1.) The prevalence of DUI arrests can be viewed in the context of the amount of alcoholic beverages consumed in the United States. The per capita consumption of alcoholic beverages reflected little change from 1980 to 1989, from 28.3 gallons per person to 27.2 gallons. In 1989, the per capita consumption of alcoholic beverages was greater than the per capita consumption of coffee (26.9 gallons per U.S. resident) and milk (26.0 gallons) and was exceeded only by the consumption of soft drinks (41.8 gallons). (Food Consumption, Prices, and Expenditures, 1970-90, Department of Agriculture, Economic Research Service, 1992, p. 63.) The annual consumption of alcoholic beverages based only upon the adult population age 21 or older (all States now impose this age restriction) would equal about 33.7 gallons of beer, 3.0 gallons of wine, and 2.1 gallons of liquor per person. However, individual patterns of consumption vary. It has been estimated that 33% of the adult population accounts for 95% of the alcohol consumed, and 5% of the adult population accounts for 50% of the consumption. (Steve Olson and Dean R. Gerstein, Alcohol in America: Taking Action to Prevent Abuse {Washington: National Academy Press, 1985, p. 13}.) DUI arrests and age Since 1980 arrest rates for DUI have not increased consistently across all age groups. In 1980 those between age 18 and 39 were overrepresented among arrestees, compared to their share of licensed drivers (table 2). Persons age 18 to 20 accounted for 7.2% of drivers but 12.9% of those arrested, about 1 arrest for every 57 drivers. Drivers age 65 or older, by contrast, accounted for 10.7% of drivers but less than 2% of those arrested, about 1 arrest for 714 drivers in this age group. Compared to 1980, data for 1989 reflected declines in arrest rates for drivers under age 21 and over the age of 40. Arrest rates for those age 18 to 20 decreased by approximately 9%, and among drivers age 16 and 17, arrest rates dropped nearly 25% over the period. Drivers between 21 and 39 years old had higher rates of arrest for DUI in 1989 than in 1980. For example, drivers between the ages of 25 and 34 during 1989 experienced rates of arrest about 40% higher than drivers of similar age groups in 1980. For those age 40 to 44, arrest rates were down about 8% from 1980 and each succeeding age group showed a larger percentage decline. Several possible reasons may account for why arrest rates increased in each age group between 21 and 39 and decreased among all other age groups. Although increased enforcement of drinking and driving laws would be expected to affect all age groups to some degree, more stringent enforcement efforts may have been selectively applied to younger age groups. Drinking or driving behavior may also have changed according to different age groups. Arrest rates for new groups of drivers who turned age 18, 19, and 20, and who are fully covered by the new laws, decreased 8.5% from 1980 to 1989. Lower arrest rates may be a reflection of changing drinking behavior in this group as a result of raising the minimum drinking age to 21.National surveys of high school seniors in 1989 indicated less prevalent daily drinking and drinking in the month preceding the survey than did seniors in 1985 and in 1980 (before drinking ages were raised). In addition, a smaller percentage of seniors in 1989, compared to those in 1980, reported engaging in binge drinking (5 or more drinks in a row at least once in the 2 weeks prior to the interview). (See "Drug use among American high school seniors, college students and young adults, 1975-1990," Volume 1: High School Seniors {Rockville, Maryland: National Institute on Drug Abuse, 1991}.) Overall, the surveys of high school seniors document a decline from 1980 to 1989 in the percentage of high school seniors who drank daily or who had drunk in the last 30 days. Senior class 1980 1985 1989 Percent who had drunk in last 30 days 72.0% 65.0% 60.0% Percent who drank daily 6.0 5.0 4.2 Percent with binge drinking 41.2 34.7 33.0 DUI offenders in jail On June 30, 1989, an estimated 395,000 adults were confined in the Nation's 3,312 local jails. An estimated 30,147 (13.8%) were serving sentences after conviction for driving while intoxicated by alcohol or drugs; in 1983, an estimated 10% of the jail population nationwide had been convicted of DUI. Persons held in local jails who were unconvicted but charged with DUI accounted for approximately 2% of the inmate population in both 1983 and 1989. Profile of DWI offenders in jail Among convicted and unconvicted persons in jail for DWI, males predominated, and the racial distribution was more similar to the adult general population than was the case for those jailed for offenses other than DWI (table 3). Inmates charged with DWI were more likely to classify themselves as white and non-Hispanic (68%) than were those jailed for other offenses (36%). Persons jailed for other offenses were more likely to classify themselves as black and non-Hispanic (45%), compared to those charged with DWI (8%). The average age of those jailed for DWI was higher than that of those jailed for other offenses. The median age of the DWI jail inmates was 32, about 5 years older than the median age of those in jail for other crimes. Of those jailed for DWI, about 47% had completed high school and about 20% had completed 8 years or less of education. Compared to 82% of jail inmates charged with offenses other than DWI, about 70% of persons charged with DWI reported that they were not living with a spouse at the time of their arrest: An estimated 35% had never been married, 35% were divorced or separated, and about 2% were widowed. At the time of the arrest, 78% of those charged with DWI were employed, compared to 63% of inmates charged with other offenses. The median annual income of those who had been free for at least 1 year prior to the DWI arrest was $11,000. The median annual income for those inmates charged with other offenses was $6,750. Prior sentences and criminal histories About 86% of persons jailed for DWI had prior convictions for crimes, including DWI, and had been sentenced to probation, jail, or prison (table 4). This percentage was lower among those in jail for crimes other than DWI (76%). An estimated 71% of those in jail for DWI had prior sentences to incarceration; among those in jail for crimes other than DWI, 56% had previously been confined. Compared to those in jail for other crimes, jailed persons charged with DWI were more likely to have been incarcerated three or more times. Convicted DWI offenders in jail were nearly twice as likely to have two or more prior DWI convictions as those unconvicted persons in jail for DWI (table 5). About 52% of persons jailed for DWI had a previous DWI conviction (table 5). These convicted DWI offenders were more likely than other inmates to have been previously convicted of the same crime (table 6). Among those in jail charged with robbery, 22% had a prior robbery conviction with a jail or prison sentence; among those jailed for assault, 17% had a previous assault conviction; among those charged with drug trafficking, 14%; among those charged with burglary, 28%; and among those charged with larceny, 30% had a prior larceny conviction with a jail or prison sentence. Nearly half (45%) of those in jail for DWI had a criminal justice status at the time of their arrest: Alcohol consumption Convicted offenders were asked detailed questions about their consumption of alcoholic beverages prior to their arrest for DWI. Each offender who reported drinking prior to arrest provided the types and amount of beverages consumed and the period spent drinking. Based on these responses, it was possible to convert the amount and type of beverage consumed to a pure alcohol equivalent (ethanol) to estimate total intake. (See Methodology for conversions.) Half of the convicted DWI offenders had consumed at least 6 ounces of ethanol (equivalent to the alcohol content of about 12 beers) prior to arrest (table 7). About 7% of the convicted offenders had consumed less than 2 ounces of ethanol; 34% had consumed between 2 and 5 ounces; 30% between 5 and 11 ounces; and 29% reported consuming the equivalent of 11 or more ounces of ethanol. To consume 11 ounces of ethanol would require drinking the equivalent of about 22 beers. The median length of the drinking session prior to the arrest was 5 hours (table 8). Given the median consumption of 6 ounces of alcohol, this amount of time would imply a rate of consumption equivalent to about 2.4 beers per hour. The average, or mean, ethanol consumption was 8.1 ounces, and the average amount consumed escalated with the number of hours spent drinking. Those who were drinking for 1 hour or less had an average ethanol consumption of 2.2 ounces, compared to those who were drinking for 12 hours or more and consuming 16.4 ounces. Most convicted DWI offenders reported drinking only beer prior to arrest: Percent of DWI offenders drinking Beer only 61% Wine only 2% Liquor only 18% More than one type 20% Amounts consumed prior to arrest varied with the type of beverage. The median ethanol consumption for beer drinkers 5.0 ounces or the equivalent of about 10 beers. Those drinking only liquor prior to arrest consumed a median quantity of ethanol of 8 ounces -- approximately equal to 10 to 11 drinks -- or nearly 60% more ethanol than those who drank only beer. Those who combined different beverages were estimated to have had an intake of ethanol nearly two and a half times that of those who consumed beer only. Those jail inmates convicted of DWI who consumed greater quantities of ethanol prior to arrest reported a greater frequency of typical drinking sessions and more alcohol consumed during those sessions (table 10). Some evidence indicates that as the amount of ethanol consumed prior to arrest increased, the percentage of those who described themselves as usually drinking daily also increased. Among those offenders reporting consumption of 2 ounces or less of ethanol prior to arrest, 18% described themselves as daily drinkers. By contrast, among those who had consumed at least 10 ounces of ethanol prior to arrest, 45% reported daily drinking. Sentencing and DWI Convicted offenders sentenced to jail are not representative of all persons who drive drunk or of those sentenced for DWI. Many first-time offenders may have driven drunk previously but managed to avoid arrest or conviction. Also, a number of DWI offenders are under probation supervision in the community or have received other sanctions. (On December 31, 1989, 37 States reported that 22.6% of the 1,831,432 adult offenders on probation had been convicted of DWI. If applied to the entire probation population of the 50 States and the District of Columbia (2,461,333 probationers), the estimated number of DWI offenders on probation would be more than 556,000 - - an estimate that is substantially greater than the number of DWI offenders in local jails in 1989. See Correctional Populations in the United States, 1989, BJS Report, NCJ-130445, 1991, table 3.10.)The most chronic DWI offenders may have been sentenced to State prisons rather than local jails. Examination of the length of the jail sentence imposed for DWI may be useful since the inmates are more likely to be the chronic and serious offenders for whom the effect of a prior record can be gauged. Of those with two or more prior sentences to jail or prison for DWI, 3 out of every 4 people received a sentence of 4 months or more. The median jail sentence for first-time DWI offenders was 115 days. For those with two or more prior sentences to jail or prison for DWI, the median jail sentence was 181 . Of those jail inmates convicted and sentenced for DWI, 42% (11,901) were given as part of their sentence a special condition or restriction other than jail time, prison, parole, or probation. Half of these 11,901 jail inmates were required to attend an alcohol treatment program. Location of drinking About 4 out of 5 offenders convicted of DWI and serving time in jail reported that they had been drinking with others prior to their arrest. Those who drank with others were more likely to have been drinking in a bar/tavern/inn or a friend's house. Appendix: Estimating blood alcohol concentration (BAC) Blood alcohol concentration (BAC) refers to the number of grams of pure alcohol present in 100 milliliters of blood. The BAC of an individual may be established by a variety of testing procedures including chemical breath analysis, saliva testing, blood testing, urinalysis, or chemical analysis of tissue samples. Calculating the BAC levels of convicted DWI offenders in jail is useful for two reasons. First, estimating blood alcohol concentration serves as a validity check of the self-reported amounts consumed prior to arrest: whether such amounts seem reasonable or even physiologically possible. Second, BAC provides a measure of intoxication that can be compared to other groups of drivers for whom BAC is known -- such as drinking drivers involved in fatal accidents. Blood alcohol concentrations may be affected by numerous factors, including physiological differences, food consumption, the amount of ethanol ingested, and the time elapsed between drinking and testing. Several assumptions underlie the estimates of blood alcohol concentration presented here: 1. An average rate of metabolism was assumed for the jail inmates equivalent to the general population, though such rates are known to vary because of differences in physiology and alcohol tolerance. 2. The self-reported body weight used in the calculations was assumed to be the correct weight for each inmate. 3. The rate of alcohol consumption was assumed to be stable over the drinking session prior to arrest. For example, if 6 ounces of ethanol were consumed during a 5-hour drinking session, the formula assumes that 1.2 ounces of ethanol were consumed per hour. State statutes often define two types of minimum BAC that constitute evidence of intoxication -- "illegal per se" and "presumptive" levels. Presumptive levels of intoxication are generally lower than illegal per se levels and require a different burden of proof to convict an individual of drunk driving. Across the States, illegal per se blood alcohol levels cluster around 0.10, but several States define it as low as 0.08 and others as high as 0.15. Presumptive levels for DWI or DUI may range from 0.05 and up but also cluster at the 0.10 level. The President's Commission on Drunk Driving has recommended that a presumptive BAC of 0.08 be enacted by State legislatures (November 1983). A BAC level above 0.05 is described as "driving while impaired" by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) Clearinghouse on Alcohol Information. The estimated BAC was highest among inmates age 35 to 39 and those age 50 to 64 (appendix table 1). BAC's did not vary greatly based on the number of prior DWI convictions. As with ethanol consumption, BAC's escalated with the number of hours spent drinking. The BAC increased from 0.11 for those who drank 1 hour or less to 0.32 for those who drank for 8 hours. The BAC varied by the type of beverage consumed, although the highest BAC levels were found among those who drank combinations of beverages (0.32). Jail inmates were estimated to have had a median BAC at the time of the DWI arrest of 0.19 and an average (mean) BAC of 0.21 (appendix table 2). The distribution of BAC levels for DWI jail inmates was similar to the BAC levels of drinking drivers involved in fatal accidents in 1989, suggesting that the average degree of intoxication of both groups was similar. Formula for calculating BAC after multiple hours of drinking (Widmark Formula) The National Highway Traffic Safety Administration (NHTSA) has provided a formula for use in this study that permits an estimate of BAC to be made based upon the self-reported drinking behavior before arrest of the jail inmates. The methodology for estimating BAC was supplied by Dr. Alfred J. Farina, research psychologist, Research Division, NHTSA. Methodology A jail is defined as a confinement facility administered by a local government agency that holds persons detained pending adjudication and persons committed after adjudication, usually for sentences of a year or less. Convicted jail inmates are either awaiting sentencing, serving sentences to jail confinement, awaiting transfer to a prison or serving a prison sentence in jail by arrangement with prison authorities. Unconvicted inmates are those who have been unable to obtain pretrial release, those detained pending trial, those on trial at the time the survey was being conducted, and those held for other governmental entities. The 1989 Survey of Inmates in Local Jails was conducted for the Bureau of Justice Statistics by the U.S. Bureau of the Census. Through personal interviews during July, August, and September 1989, data were collected on individual characteristics of jail inmates, current offenses and sentences, criminal histories, and prior alcohol use and treatment. Similar surveys of jail inmates were conducted in 1972, 1978, and 1983. Conversion formulas for ethanol The formulas used for calculating ounces of ethanol and blood alcohol concentration are described on page 8. In cases where extreme outliers or impossible responses were found, data were treated as missing. For the purposes of this report the following conversions were used: 1 ounce of ethanol is equal to -- * 24 ounces of beer (4% alcohol content); * 7 ounces of wine (14% alcohol content); * 2 ounces of liquor (100 proof or 50% alcohol content) Mixed drinks were assumed to contain 1.5 ounces of liquor. However, these conversions are approximations since some beer, wine, or liquor may have a different alcoholic content. Sample design The sample for the 1989 survey was selected from a universe of 3,312 jails enumerated in the 1988 National Jail Census. The sample design was a stratified two-stage selection. In the first stage six separate strata were formed based on the size of the male and female populations. In two strata all jails were selected; in the remaining four strata, a systematic sample of jails was selected proportional to the size of each jail. Overall, a total of 424 local jails were selected. In the second stage, interviewers visited each sampled facility and systematically selected a sample of male and female inmates using predetermined procedures. As a result, approximately 1 of every 70 men were selected and depending on the stratum, 1 of every 14, 15, or 70 women were selected. A total of 5,675 interviews were completed, yielding an overall response rate of 92.3%. Based on the completed interviews, estimates for the entire population were developed using weighting factors derived from the original probability of selection in the sample. These factors were adjusted for variable rates of nonresponse across strata and inmate characteristics. Further adjustments were made to control the survey estimates to counts of jail inmates obtained from the 1988 National Jail Census and the 1989 Sample Survey of Jails. Accuracy of the estimates The accuracy of the estimates presented in this report depends on two types of errors: sampling and nonsampling. Sampling error is variation that may occur by chance because a sample rather than a complete enumeration of the population was conducted. Nonsampling error can be attributed to many sources, such as non-response, differences in the interpretation of questions among inmates, recall difficulties, and processing errors. In any survey the full extent of the nonsampling error is never known. The sampling error, as measured by an estimated standard error, varies by the size of the estimate and the size of the base population. Estimates of the standard errors have been calculated for the 1989 survey of jail inmates, (see appendix table 5). These standard errors may be used to construct confidence intervals around percentages in this report. For example, the 95-percent confidence interval around the percentage of convicted jail inmates in 1989 who had 2 or more prior DWI sentences to jail or prison is 30.0 plus or minus 1.96 times 2.9 (or 24.3% to 35.7%). These standard errors may also be used to test the statistical significance of the difference between two sample statistics by pooling the standard error of the two sample estimates. For example, the standard error of the difference in the percentage of convicted jail inmates in 1989 who reported they had 2 or more prior sentences to jail or prison compared to those unconvicted jail inmates who reported they had 2 or more prior sentences to jail or prison would be 6.4 (or the square root of the sum of the squared standard errors in each category). The 95-percent confidence interval around the difference would be 1.96 times 6.4% (or 12.5%). Since the observed difference of 14.0% (30.0% minus 16.0%) is greater than 12.5%, the difference would be considered significant. Comparisons discussed in this report were determined to be statistically significant at the 95 percent confidence level. Statements of comparison qualified by language such as "some evidence" or "slightly" indicate statistical significance at the 90% level (1.6 standard errors). Because of the sample design, State, local, or other subnational estimates cannot be made. Bureau of Justice Statistics Special Reports are written primarily by BJS staff. This Special Report was written by Robyn L. Cohen, under supervision of Allen J. Beck. Tracy L. Snell and Darrell K. Gilliard reviewed the statistics, and Rhonda Keith produced the graph. Tom Hester edited the report, and Betty Sherman produced the report. Corrections statistics are prepared under the general direction of Lawrence A. Greenfeld. NCJ-134728 September 1992 The Bureau of Justice Statistics is a component of the Office of Justice Programs, which also includes the Bureau of Justice Assistance, the National Institute of Justice, the Office of Juvenile Justice and Delinquency Prevention, and the Office for Victims of Crime. Data used in this report are available from the National Archive of Criminal Justice Data at the University of Michigan, 1-800-999-0960. The dataset is archived as the Survey of Inmates of Local Jails (ICPSR 9419).