Volume 1, Issue 2, January 1999 ISSN 1096-4886 http://www.westerncriminology.org/Western_Criminology_Review.htm © 1999, The Western Criminology Review. All Rights Reserved.

Appendix B

This case study examines San Jose and San Francisco, California, during 1985-98. The two cities are of similar size and lie fifty miles apart in the San Francisco Bay area. Although they have similarities and differences (see Table B-1), for the purpose of this study the cities differ in one major way: San Francisco vigorously enforced its youth curfew in the late 1980s and then reduced enforcement to near zero in the 1990s. In contrast, San Jose enforced curfew very little in the late 1980s but increased enforcement in the mid-1990s. Thus, these neighboring cities are a laboratory for studying curfews. If curfews have a major impact on youth crime and/or safety, it should correspond to the divergent curfew enforcement trends of the two cities.

Characteristic		San Francisco	San Jose
Total population	1985	720,000	720,400
	1990	723,959	782,248
	1997	777,400	885,000
Population age 10-17	1990	49,922	83,658
Race/ethnicity (in percent)	White	20.8	39.3
	Hispanic	20.7	34.1
	Black	16.8	5.7
	Asian/Native	42.3	21.5
Total percent in poverty		12.7	9.3
Percent of youth in poverty		18.6	13.1
Median family income	1990	$40,561	$50,281

Source: Demographic Research Unit. 1985-97. California Statistical Abstract (1990). Sacramento, CA: Department of Finance, Table B-4; US Bureau of Census. 1990. Census of Population and Housing. San Francisco-Oakland-San Jose, California. Tables 1-5, 19.


DATA SOURCES AND METHOD

Complete statistics for crimes reported to police, arrests for curfews and other crimes, and categories of violent death involving youth are available for the study period (1985-June 1998 for crime; 1985-96 for deaths). These provide a variety of measures for evaluating curfews. Two of these have been used in the main study, above: reported crimes and arrests. These data have been taken from the California Criminal Justice Statistics Center (CJSC). Moreover, we also use combined reported crimes and arrests.¹

Moreover, we supplement our analysis with data on violent deaths. Complete annual statistics on violent deaths (those from traffic accidents, other accidents, suicides, homicides, and undetermined intent) are available for each city from the California Vital Statistics Bureau. The strength of this measure is consistency and completeness regarding serious behavior. The weakness is that the figures do not indicate time, place, or other charateristics of deaths that would permit refined analysis of whether deaths are affected by curfew.

For each of these measures, rates are calculated per 100,000 youth age 10-17 and adults age 18-69 (the age groups used by the CJSC to define "youth" and "adult") for each city and year, using population figures from the U.S. Bureau of the Census (1990) and Demographic Research Unit, California Department of Finance (intercensal years).

The rate of each indicator is compared to curfew arrest rates by year for each city for the 1985-97 time period.Three kinds of trends are examined. The first is absolute youth crime and death rates among 10-17 year-olds. The second is net youth crime and death rates, which is computed by dividing youth rates by adult rates. The latter measure factors out trends other than curfews that affect crime and death rates among both youth and adults. Correlations for both measures are repeated, using arrest rates adjusted for annual reported-crime clearance rates. No "net" measure is available for reported crimes since no ages are given; the "absolute" measure is crimes reported per 100,000 population of all ages. Finally, another analysis compares net curfew arrest rates in each year with the net crime and death rates the next year. The purpose of this "lagged" comparison is to determine whether the effects of curfew enforcement are delayed. The raw counts, rates, and correlations (discussed below) are shown in Appendix C.

RESULTS

A total of sixty separate correlations were calculated: five for each type of crime, three for each type of reported crime, and one for violent deaths for absolute, for net, and for lagged rates for each city, which sum to forty-eight (net rates do not apply to reported crimes); and absolute and net correlations for each of the three reported crimes for both cities, which sum to twelve. As noted earlier, a correlation of -.50 to -1.00 indicates a significant reduction in crime following adoption of a curfew; a correlation of .50 to 1.00 a significant increase in crime following the curfew; a correlation between -.50 and .50 indicates no effect that can be distinguished from chance alone.

The measures provide reasonably consistent results to make this generalization: San Francisco did not experience an increase in youth crime, crimes reported to police, or teenage violent death associated with the dismantling of its once active curfew arrest policy from the late 1980s to the 1990s. Moreover, San Jose did not experience a decline in youth crime, crimes reported to police, or violent death associated with its sharp increase in curfew arrests in the mid-1990s.

Only four significant correlations are found for any crime or death category, and these are in opposite directions. San Jose experienced a significant increase in youth property crime arrests and a significant decrease in violent crimes reported to police coincident with increased curfew arrests. San Francisco experienced significant increases in all youth offenses, and a significant, lagged decrease in youth violent crime arrests, coincident with a decrease in curfew arrests. For sixty correlations tested at a significance level of .05, we would expect three (five percent of sixty) to be significant by chance alone. The fact that the four significant correlations are in opposite directions indicates that they are measuring random, not real, effects. In addition, the sixty non-significant correlations also display a random pattern: thirty-three are positive, twenty-seven negative. This is an almost picture-perfect, "no effect" result.

There is a significant racial/ethnic difference in curfew enforcement in San Jose that can not be explained by each group's contribution to crime (Table B-2). More specifically, Hispanic youth are responsible for fifty-one percent of San Jose's youth crime but they comprise seventy-two percent of its curfew arrests. Hispanic youth are twice as likely as white youth, and three times more likely than black or Asian youth, to be arrested for curfew violations than their contribution to the city's other crime would lead one to expect. Hispanic youth are five times more likely to be arrested for curfew violations than their percentage of the youth population would predict.

Source: California Criminal Justice Statistics Center, California Criminal Justice Profiles, cities of San Francisco, San Jose, 1985-97; Demographic Research Unit, California Department of Finance.

DISCUSSION

Even the strongest levels of curfew enforcement, as San Jose implemented, do not appear to reduce juvenile crime, improve juvenile safety, or reduce a city's crime volume. Conversely, even total abandonment of curfew enforcement, as San Francisco accomplished, is not associated with increases in crime or violent death. This "study of extremes" reaches only the blandest of "no effect" findings.

Imagine if anecdotal assertions that curfews are ineffective were accepted by law enforcement, politicians, and the media in the same way that anecdotes about the supposed value of curfews are accepted now. We might see reports (based on available statistics) such as these:

• In 1994, San Jose had no curfew arrests; in 1996, 989. But violent crime among youth rose from 686 in 1994 to 691 in 1996, juvenile felony offenses rose from 2,456 to 2,553, and violent deaths among youth jumped 50 percent, from 8 in 1994 to 12 in 1996. Therefore, increasing curfew enforcement increases juvenile crime and violent death.
  
• From 1996 to 1997, San Jose police drastically reduced curfew enforcement, from 989 arrests to just 82. The number of juvenile violent crimes plunged from 691 to 583, property crimes fell from 1,047 to 988, and crimes reported to police plummeted by 2,000, from 34,942 in 1996 to 32,990 in 1997. Therefore, reducing curfew enforcement cuts juvenile crime.
  
• In 1988, San Francisco police made 596 curfew arrests; in 1997, zero. From 1988 to 1997, San Francisco's juvenile felonies declined from 2,829 to 2,155, and all crimes reported to police fell sharply, from 58,260 to 43,875. Sharp declines continued in the first eight months of 1998. This is more evidence that eliminating curfews reduces youth crime.
  

Even though all the above statements are factual, according to city police reports, the flaw in such logic is that only certain crimes, time periods, and cities in isolation are cited (in this hypothetical example, those most favorable to make the case against curfews). Yet, this is exactly the kind of selective strategy used to promote curfews by many law enforcement agencies and political authorities and reported in the news media. By picking and choosing which crime, time period, city, or neighborhood to cite, proponents and opponents of curfews can make any assertion they wish.

Our general study finds no beneficial effects of stronger curfew enforcement on youth crime or violent death. This more detailed case study of San Jose and San Francisco also finds no effect. It would be just as valid to say, "not having a curfew reduces juvenile crime and improves juvenile safety" as the other way around.

If cities were considering curfews for adults, the analysis would be exhaustive, scientific, and strongly court-scrutinized. We believe younger citizens' rights deserve an analysis that is just as careful, especially those of racial/ethnic groups who seem to be disproportionately arrested in areas of high curfew enforcement relative to their contribution to other crime.

1 . This measure uses crimes reported to police to adjust arrest rates by year. It is computed by dividing age, crime, year, and city-specific annual crime arrest rates by their respective clearance rates. In effect, this measure assigns uncleared crimes (that is, crimes for which no arrest is made) to juveniles and adults based on the proportion indicated by arrests of juveniles and adults for those same crimes. The strength of this measure is that it maximizes available information to produce an estimate of the total crimes committed by each age group for each city and year regardless of whether an arrest occured. The weakness of the measure is that the ages of the persons who commit uncleared crimes may not-- and in fact, probably do not--reflect the ages of persons arrested for those same crimes. In practice, this measure probably overstates juvenile crime, since FBI clearance data indicate that adults commit substantially more crimes per offender than juveniles do. The fact that the correlation formula used to evaluate each measure only compares annual changes in rates, rather than the rates themselves, mitigates this problem considerably. For the purposes of this study, this "combined measure" is probably the best to evaluate curfew effects.