MENU TITLE: Client Evaluations of Gang Services: . An Analysis of Client Responses Gathered During the National Youth Gang Suppression and Intervention Project Series: OJJDP Published: May 1990 22 pages 48,625 bytes Figures, charts, forms, and tables are not included in this ASCII plain-text file. To view this document in its entirety, order a photocopy from the NCJRS library (or borrow it through interlibrary loan). Call 800-851-3420 for information and fees. David Curry, Consultant (West Virginia University) with Irving A. Spergel, Principal Investigator Ron L. Chance, Project Director and Jae Kim, Research Assistant National Youth Gang Suppression & Intervention Project The School of Social Service Administration The University of Chicago 969 E. 60th Street Chicago, IL 60637 In Cooperation With Office of Juvenile Justice and Delinquency Prevention U.S. Department of Justice Washington, DC 21531 PURPOSE As a supplement to the National Youth Gang Suppression and Intervention Research and Development Project, we gathered data from former gang members and clients of gang programs. Our purpose is to discover what services these clients of gang programs have received, how effective they felt these services to be, and, if applicable, what their reasons for leaving the gang were. Also of interest is any differences that may exist among program client experiences and any association of these differences with race and ethnicity, sex, and gang membership status. A copy of the survey instrument is attached as an Appendix. SOURCES OF DATA Our data were obtained from six program sites and two conferences as shown in Table 1. It is best to describe each source of data separately. More detailed descriptions of the programs at each site are found in Spergel and Chance (1990). The complete transcripts of the two Social Service Administration Symposia are available from the National Youth Gang Information Center (NYGIC), funded by the Office of Juvenile Justice and Delinquency Prevention. Tables 2 through 7 summarize the data reported by source below. Columbus, Ohio The first set of data consists of ten program clients from Columbus, Ohio. Seven are former gang members, two have never been a member of a gang, and for one this information is missing. One of these youths is a black female. Seven (70 percent) are black. One of these, a male, is black non-African American. Two are Mexicans. For one, racial/ethnic information is missing. The mean age for these respondents is 18.6 but ages range from 13 to 27. For the five of these respondents who provide information on highest grade of schooling, the average is 9.2. The highest level of educational achievement reported is eleventh grade. Fort Wayne, Indiana The two respondents from Fort Wayne, Indiana, are from the Coalition for Youth Services (CYS). Both of these respondents are African American former gang members. One is male, the other female. One is 19, and the other 20. Neither provided educational information. East Los Angeles, California Our largest number of respondents comes from programs in East Los Angeles. The 40 client respondents from the East LA Skills Center include 16 present gang members (40 percent), 8 former gang members (20 percent), and 16 youths who report never belonging to a gang. Fourteen of these respondents (35 percent) are females. The East LA Skills center client respondents are predominately Hispanic (38 or 95 percent). Two of the East LA Skills Center respondents are white. Only one (2.6 percent) of the Hispanics is Puerto Rican, and five are "other" Hispanic (13.2 percent). The majority (32 or 84.2 percent) are Mexican. The East LA Skills Center clients are relatively young. Their average age is 16.2. This average is reduced to 15.8 if one respondent, a 31-year-old female who reports never being in a gang is dropped. Their average years of education is comparably low at 9.7 years with very little deviation. The standard deviation is less than one grade level. An additional source of East Los Angeles data is eleven Gang Violence Reduction Project (GVRP) staff members. Eight of these (72.7 percent) report being former gang members. Two did not answer this item. One (a 41 year-old male) reports currently being a member of a gang. Four (36.4 percent) of these staff members are females. One staff member is black, and one is white. The remaining nine (81.8 percent) are Hispanic. As with the East Los Angeles clients, a majority of these respondents (seven or 77.8 percent) are Mexican. The other two are "other" Hispanic. The average age of these respondents is older (32.7), as might be expected. The average level of education is less than a high school diploma, though seven (63.6 percent) of the respondents have completed high school. Chino, California Thirteen clients of Chino Youth Services completed survey instruments. Seven (53.8 percent) are former gang members, and five (38.5 percent) are present gang members. One does not provide information on gang status. All are males. Eleven respondents are Mexican (84.6 percent). One of the other two is black, and the other white. The average age is 17.6. One former gang member at age 27 is much older than the others. For the eleven respondents who report their education, the average grade completed is 9.6. El Monte, California Another thirteen respondents were obtained from agencies in El Monte, California. A majority of these (7 or 53.8 percent) are former gang members. Five (38.5 percent) report still being active members in gangs, and, for the remaining one, this information is missing. All of the El Monte respondents are Hispanic with all but one of them (92.3 percent) being Mexican. The final respondent is "other" Hispanic. The respondents from El Monte are older than those from the other sites with an average age of 23.6 (for the 12 reporting age). The age range is from 15 to 32. The average years of education for the El Monte respondents is 9.9. Two report having a high school education. Ethan Allen School for Boys Twenty-five cases were obtained from Ethan Allen School for Boys in Wisconsin. Details on the school can be found in Spergel and Chance (1990). Of these youths, ten (40 percent) report that they are currently gang members. Another ten (40 percent) are former gang members. Three report never having been a member of a gang, and two did not provide the information. Since the information on this group of respondents was obtained from a section of the correctional facility designated for hard core gang youth, it may prove useful to regard self-reported answers on gang status from this subpopulation of respondents with some skepticism. Twenty (80 percent) of the Ethan Allen respondents are black. Two (8 percent) are Puerto Rican, and two are Mexican. The other respondent from Ethan Allen is "other". The Ethan Allen respondents are rather homogenous in age with an average of 16.6 and a standard deviation of a little over a year. We did not obtain highest grade completed on the Ethan Allen youths. Special Gang Symposiums Two special symposia of former gang members were held at the School of Social Service Administration (SSA), University of Chicago in 1989. While the principal purpose of these symposia was gathering the qualitative information reported in two National Youth Gang Project documents, participants were also asked to respond to the survey instrument analyzed here. All ten respondents from the SSA symposia are former gang members. The respondents from the first symposium which focused on Hispanic former gang members include four Puerto Ricans, one Mexican, and one black. The four respondents from the second symposium are all black. One respondent in the second symposium is female. The average age for first symposium respondents is 32.5. The average years of education is 10.8. One of the first symposium respondents reports a college education. The average age for second symposium respondents is 29.5. The average years of education is 11.3. One of the second symposium respondents reports completing three years of college. SERVICES Respondents were asked "Which [Service] Have You Ever Received as a Member of a Gang?" This question was followed with the list of services shown in Table 8. Table 8 shows the portion of total respondents reporting that they had received a particular service. Beside each service respondents were asked, "How helpful in reducing your participation in criminal gang activity was this?" They were offered four scores to choose from -- 1=Very, 2=Some, 3=Not Much, and 4=Not. Table 8 also shows the average score and the percentage of respondents who rated each service. The last two columns of Table 8 show the ranking of each service by use and helpfulness rating. Sports and recreation services are the services most commonly reported by our respondents with 60 percent reporting having, at some time, received such service. Sports and recreation services are rated by respondents as being most helpful to them in reducing their participation in criminal behavior. After this initial uniformity in agreement on the most often provided service proving most helpful, there is divergence between the rankings of what services are more often received and what services are most helpful in reducing gang criminality. Individual counseling, while the second most frequently provided service, is rated the third most helpful. Job placement is the tenth most frequently received service (by 38 percent of the respondents), but is rated as being second most helpful. Supervision (parole or probation) is the third most frequently received service but in terms of helpfulness ranking it places twelfth. Work visits is the least reported service received, but it is ranked thirteenth in helpfulness. Such overall calculations may, however, not be especially useful given the diversity of our total population of respondents. Before drawing conclusions, we must first break our respondents into subpopulations and note the patterns of responses specific to each group. SITE DIFFERENCES First, let us examine differences by site. Table 9 shows the site specific survey results. The GVRP employees and the respondents from the two symposia are omitted from this analysis by sites. It should also be remembered that there are only two respondents from Fort Wayne. In order to compare helpfulness rankings across sites when individuals report and rank varied individual sets of service experiences, we examine our data from several perspectives. The measures of percent reporting each service in each location and mean helpfulness ranking for each service provide the greatest detail by site and service. For instance, 100 percent of respondents from Columbus and Chino report receiving individual counseling. Among the Ethan Allen respondents, 61 percent report receiving this service compared with 45 percent and 32 percent respectively for East LA Skills Center and El Monte. The respondents from Chino Youth Services on the average rank their individual counseling as most helpful in reducing their involvement in gang crime with a mean helpfulness rating of 1.23. Columbus is next with an average helpfulness rating of 1.70 followed by El Monte and Ethan Allen respondents tied at an average helpfulness rating of 2.0. Individual counseling among East LA Skills Center respondents is rated least helpful for the six sites. (The one respondent from Fort Wayne who reports receiving individual counseling rates this service as "very" helpful, but it's impossible to say anything about a program from only two client responses.) As one more example, job placement is reported by 75 percent of the respondents from Columbus and El Monte, but only 25 percent of those from Chino and 23 percent of East LA skills Center report receiving this service. In terms of helpfulness, the Chino respondents all rate job placement as very helpful with El Monte following with a mean of 1.56 and Columbus and East LA Skills Center each at 1.86. From Table 9, it is evident that for certain services, no respondents at specific sites report having ever received them. For instance, no respondents in Columbus or El Monte report having received protection from violence. Table 10 summarizes the distribution of service reports across the six sites. Ten of the services are reported by respondents at all six sites and seven more at all but one site. Only one service, housing/shelter, is reported by less than four of the sites. Two client-level summary measures are also computed by site. The first is average number of services received per client. The second is average helpfulness rating as computed per client for all services that the client received. It should be remembered that neither of these statistics take into account what specific services are included in their computation. From Table 11, it can be seen that, on the average, clients from Chino Youth Services report receiving more services than those at any other site. The numbers in superscript beside each mean identify those other sites for which Tukey's least significant difference test indicates that the difference is significant at the 0.05 level. The average number of services reported by clients from Chino, Columbus, and Ethan Allen are significantly greater than the average number of services reported by clients from East LA and El Monte. On average, respondents from East LA Skills Center rate the services that they have received as least helpful. The average helpfulness rating for East LA Skills Center respondents is significantly lower (a larger numeric value) than the average helpfulness ratings for respondents from Chino and Ethan Allen. The helpfulness rating for El Monte is different from that for Chino at the 0.05 level. Two service-level summary measures that we compute are the average proportion of clients reporting receipt of a service at each site and the average helpfulness rating as it is computed across all offered services. From these measures we compute for each site the correlation between the proportion receiving a particular service and how helpful that service is perceived by site respondents. This measure presented in Table 11 can be thought of as a measure of how much the services deemed most helpful by clients are those that clients receive. Since our helpfulness measure runs from one to four, a significant relationship between those services provided to most clients (a proportion) and those services rated as most helpful in reducing gang crime should have a negative sign. For example, the -.911 correlation between proportion of clients receiving services and the helpfulness ratings of those services for Columbus indicates that those services that respondents are most likely to report receiving in Columbus are also those services that respondents consider to have been most helpful in reducing their involvement in gang crime. Besides Columbus, the only other site where a high correlation between the services most likely to be received by respondents and the helpfulness of those services is found in East LA Skills Center. While, on the average, respondents from Chino Youth Services are most likely to rate services as more helpful, respondents from Chino are least likely to receive the services they consider most helpful. The Wilcoxon matched pairs comparison test is a nonparametric test that allows us to compare two sets of rankings. On a service by service basis, the test, as we apply it, compares offered services, first in terms of proportion of clients reporting being provided that service and second by the helpfulness rating of that service. Tables 12 and 13 show the significance of the differences between each site respectively by proportion receiving services and rating of helpfulness of services. Unlike the averages of number of services and helpfulness rankings shown in Table 11, these pairwise comparisons of sites do take into account differences in specific service mix and patterns of helpfulness ratings among services. In Table 12, the patterns of service receipt at Chino is significantly different from the patterns of service receipt at Columbus, East LA Skills Center, and El Monte, but not significantly different from the pattern of service receipt at Ethan Allen. The patterns of service receipt at East LA and El Monte are significantly different from the comparable patterns at the other sites. The service receipt pattern at Columbus is very different from those of Chino and El Monte, while the difference between Ethan Allen and Columbus on this comparison is not statistically significant. (The results for the two respondents from Fort Wayne are included for informational but not analytic purposes.) From Table 13, it is seen that there are significant differences among sites in how clients rate services in terms of helpfulness. The helpfulness ranking of services by Columbus clients is very different from those of Chino and Ethan Allen respondents. Only the differences in helpfulness rankings of services between Columbus and East IA Skills Center and between Ethan Allen and El Monte are not significant at the 0.05 level. These results indicate that there are substantial differences between sites in terms of pattern of service receipt and how helpful clients perceive specific services to be. SERVICE DIFFERENCES BY GANG INVOLVEMENT As we have noted above, our respondents represent different stages of gang involvement. Here we will examine the degree to which respondents in different levels of gang involvement report receiving and regard as helpful our list of 22 services. Non-Gang Members As Table 2 above shows, 21 or 16.9 percent of our respondents report never having been in a youth gang. Table 14 presents the survey results for nineteen of these respondents. The two excluded respondents include one from East LA who reports being a 31-year-old female and one for whom age is missing. Without these two respondents, all of the "never-beens" included in our survey are less than 18 years of age. Eleven are boys, and eight are girls. It may be appropriate to regard services provided to adolescents who have never been gang members as prevention as opposed to intervention efforts. Present Gang Members From Table 2 above, 37 or 29.8 percent of our respondents identify themselves as presently gang members. Ignoring one 41-year-old GVRP worker who claims current gang membership, we compute percentage reporting receipt of each service and helpfulness rating of each service for these 36 male current gang members in Table 15. Former Gang Members Our largest group by gang membership status is former gang members. Here we break former gang members into three groupings: young former members (under 21), older former members, and female former members. The reported service use and average helpfulness rating for each service for each of these three subgroups of former members are shown in Tables 16 through 18. Comparisons by Membership Status Rather than run 22 analyses of variance and 22 contingency table analyses to test hypotheses for specific services and specific subpopulations of respondents, we again look at general differences in number of services received (regardless of service) and average helpfulness rating for services received. Then we look for general differences between subpopulations in service receipt patterns and ranking of service helpfulness in Tables 20 and 21 using, as above, a series of Wilcoxon matched pairs tests. Again, the tests in Tables 20 and 21 can be thought of as tests for differences between the pairs of rankings displayed in Tables 14 to 18. In Table 19, we can see that there are more differences between subpopulations in average number of services received than in helpfulness ratings. Young former gang members report receiving the most services. The average is 9.1. This is significantly different from the average number of services reported by non-members and older former members. Young former gang members report receiving almost three times as many services as non-gang members. On the average, present gang members report having received twice as many services as non-members. Former female gang members report receiving an average of 6.2 services each, and, in this, they do not significantly differ from any of the other subpopulations. The non-members and present members have lower average helpfulness ratings, but none of the differences in helpfulness ratings among these groups is significant at the 0.05 level. As above, the correlation between the proportion of respondents reporting a service and its helpfulness rating serves as a measure of the degree to which the services clients receive are those that they consider most helpful. Since the helpfulness ranking runs from 1 to 4 (very helpful to not helpful), a negative correlation indicates that clients report receiving more of the services that they rate-as more helpful. Only for older former gang members is this relationship not significant at the 0.05 level. From Table 20, we can note several differences in reported service receipt patterns and evaluation of services by gang status. Non- gang members and former female gang members report service receipt patterns that are not significantly different from each other but are significantly different from the other subpopulations. The reported pattern of service receipt for present gang members and older former gang members is not significantly different, but, in terms of the patterns of services received, these two subpopulations significantly differ from the other three. Younger former gang members report a service receipt pattern that is significantly different from the other four subpopulations of clients. Table 21 shows that clients rank services significantly differently when their gang involvement status is controlled. The service evaluation pattern of young former gang members is significantly different from all other groups except older former gang members. Older former gang members evaluate services differently from non-members and female former gang members. Present members evaluate services in a ranking that is significantly different from those of young former and female former gang members. Female former gang members evaluate services significantly differently from all groups except the non-members. The difference between how non- members and young former members rank services appears to be the greatest for any two gang status groups. SERVICE DIFFERENCES BY RACE/ETHNICITY Another concern in comparing client evaluations of gang program services is whether there exist differences in either service receipt or service helpfulness by race or ethnicity. Black Respondents Black respondents are broken into two groups based on selfreported status as black African- American or "other" black. African American is the designation chosen by 25 of our respondents. Excluding one African American who reports completing three years of college, their average reported education is 9.11 years. Their average age is 21.6 years. Three (12 percent) of the African American respondents are female. Table 22 shows that recreation and sports services are reported as those most frequently received (by 78 percent) and those considered most helpful by African American respondents. Eleven of our black respondents report themselves to be black other than African American. The vast majority of these eleven are ten respondents from Ethan Allen. All are male. While all are missing answers to our age and education questions, we may assume that the group approaches the average age reported for other Ethan Allen respondents and that they are getting an educational experience similar to the others who did respond. As is the case for African American respondents, Table 23 shows that most of them (80 percent) report receiving sports and recreation services in gang programs and that they rate it as the kind of service that has been most helpful to them in reducing their involvement in gang-related crime. Puerto Rican Respondents Of our seven Puerto Rican respondents, information on four were obtained from our first SSA symposium. Two more are from Ethan Allen and another from East LA. Our Puerto Rican respondents are much older than any other group with an average age of 30.3 years. Excluding the one Puerto Rican respondent with a college degree, their average education is the lowest for an ethnic category at 9.0 years. All seven are male. In terms of services received, the Puerto Rican respondents differ from other ethnic groups. The most commonly reported service (by 5 respondents or 71 percent) is protection from violence. While sports and recreation services are the second most frequently reported type of service received for the seven respondents, they are not ranked very high (twelfth) in terms of helpfulness. Mexican Respondents The largest ethnic subpopulation among our respondents is Mexican Americans with an n of 67. Thirteen (19.4 percent) of them are females. More than half of them (58.2 percent) are from East LA Skills Center with 32 clients and 7 GVRP staff members identifying themselves as Mexican- American. Twenty-three more are from Chino and El Monte, making all but five of this subpopulation from California. Their average education level is 9.95 years. Their average age is 19.89 years. The reported receipt and helpfulness assessment of programs for Mexican respondents differs from those of black and Puerto Rican respondents. The most frequent service reported as received is supervision (parole or probation) by 56 percent of the subpopulation. While the average helpfulness ranking of this service is 2.00 (some help), it ranks eleventh in terms of helpfulness among the twenty-two types of service. Recreation and sports is the second most frequently reported received service (by 50 percent of the respondents), but it is ranked eighth in terms of helpfulness. Protection from violence (received by 30 percent,- of these respondents) ranks first in terms of helpfulness rating among Mexican-American respondents. Other Hispanic Respondents Only eight Hispanic respondents indicated that they are other than Puerto Rican or Mexican. All but one (from El Monte) of these respondents are from East LA Skills Center (5 clients and 2 GVRP staff). The average age of this subpopulation is 22.86 years. Their average level of education is higher than that of blacks, Puerto Ricans, and Mexicans at 10.43 years. The most frequently reported type of service received among other Hispanics is individual counseling (by 63 percent) followed by tutoring and protection from violence (each by 50 percent). Four services are rated as "very helpful" by all eight of these respondents tutoring, protection from violence, drug prevention and treatment, and work visits. Individual counseling, the most frequently received service, ranks sixth in helpfulness among this subpopulation. White Respondents Only four white respondents are represented in our survey of gang program clients. Three of them are from East LA Skills Center (one a GVRP staff member) and the other from Chino. The average age of the whites is 21.75 years. Their average educational level is the highest among the racial and ethnic subgroups at 10.5 years. All four are male. Three of the four report receiving recreation and sports services and job training making these the most frequently reported services. For job placement and housing and shelter, one respondent each report receipt and each rates the service as "very helpful." Table 28 shows the mean number of services reported and the mean helpfulness ratings of services received by race and ethnicity. The average number of services reported is greatest for blacks -- African Americans and others -- and lowest for Puerto Ricans. The differences in average number of services received are significant between all three Hispanic groups and African Americans. The difference between other blacks and Puerto Ricans is also significant at the 0.05 level. The range of differences in average helpfulness rating for services received across ethnic groups is less. Other Hispanics rate the services they receive as most helpful, while Mexicans rate the services they receive as least helpful. Only the difference between these two groups is significant at the 0.05 level. As noted above in comparing sites and respondents of different levels of gang involvement, the negative correlation of proportion receiving a service and the average helpfulness rating for that service can be used as an indication of the degree to which respondents are receiving services that they rate as more helpful. Only the relationship between service receipt and helpfulness is not significantly different from zero for black other respondents. This may be the result of the lack of this same relationship for Ethan Allen respondents in Table 11 and the fact that 91 percent of the black other respondents are incarcerated there. The relationship between service receipt and perceived helpfulness is highest for whites and African Americans. Tables 29 and 30 show the results of applying the Wilcoxon matched pairs tests to each pair of ethnic subpopulations in order to compare patterns of service receipt and helpfulness ranking. As can be seen in Table 29, the differences in service receipt reported by blacks and Mexican-Americans can be seen to be significantly different from those of the total populations. The difference in service receipt between African Americans and other blacks is not significant at the 0.05 level. The differences in service receipt among our three Hispanic subgroups are not significant at the 0.05 level. The differences between both groups of black respondents and the three Hispanic groups are statistically significant. The differences between whites and both groups of blacks are not significant. Whites are significantly different in their reported service receipt from all three subgroups of Hispanics. Significant results from paired comparisons of helpfulness ratings by service are less common. Only Puerto Ricans are significantly different from all respondents. The differences between Puerto Ricans and both groups of black respondents are significant at the 0.05 level. So is the difference between other blacks and other Hispanics. All of these results are significant only at the 0.05 level.Puerto Ricans rate services as more or less helpful differently than do blacks and otherHispanics. LEAVING THE GANG: RETROSPECTIVE REASONS As a group our 59 former gang members come from all 9 of our sources of data. They represent each of our racial/ethnic subpopulations. The largest representation is Mexican (24 or 40.7 percent) followed by African Americans (17 or 28.8 percent). The number of other blacks (6), Puerto Ricans (5), and other Hispanics (5) are comparably equal. There are two white former gang members included. Forty-seven (79.7 percent) are males. Former gang members were asked to indicate all reasons for leaving the gang from among the dozen possible reasons shown in Table 31. An additional item invited respondents to list "other reasons for leaving the gang." Only one of the few responses that were received was mentioned more than once. Having children was mentioned twice. Table 31 shows the number and proportion of former gang members reporting each suggested reason for leaving the gang. The most commonly reported reason for leaving the gang is simply getting older (42.4 percent) followed by getting arrested (37.3 percent)and becoming tired of violence (35.6 percent). The leastfrequently reported reason for leaving the gang is becoming too involved in drug sales (5.1 percent). The fact that three respondents reported even this least reported of reasons for leaving the gang is indicative of the viability of all of our options. When making comparisons between male and female former gang members, it is important to note that none of the females report leaving the gang because of arrest or jail. LEAVING THE GANG AND RACE/ETHNICITY Based on patterns of homogeneity that we perceived in our analysis of more specific race and ethnicity categories, we examine leaving the gang for black versus Hispanic gang members. For this purpose, we classify African Americans and other blacks as "black" and Mexican-Americans, Puerto Ricans,and other Hispanics as Hispanic. Table 32 shows the breakdown of black and Hispanic respondents by gang membership status. For the respondents in our survey, blacks are significantly more likely than Hispanic clients to report no longer being a gang member. It is important to not that there is no significant age difference between blacks and Hispanics, overall, in our study. In fact, blacks tend to be slightly younger (19.7 years) that Hispanics (20.5 years). The means for age, number of services reported, and the helpfulness rating for blacks and Hispanics still in the gang and not in the gang are shown in Table 33. The average ages for blacks and for Hispanics still in the gang and not in the gang are significantly different at the 0.01 level. The difference in the average number of services received by black and Hispanic former gang members is significant at the 0.01 level. None of the other differences in these pairs of means is significant at the 0.05 level. A general linear model analysis of variance reveals that only the difference in number of services between blacks and Hispanics is significant when all three variables are examined in a multivariate context. Even though differences are not statistically significant, it is interesting to note that black former gang members report receiving more services than present gang members. Hispanic former gang members report receiving fewer services than Hispanics who are still in the gang. Black former gang members rate services received slightly less helpful than blacks still in the gang. Hispanic former gang members rate services as more helpful than Hispanics still in the gang. Table 34 shows reported reasons for leaving the gang by black and Hispanic respondents. A greater percentage of blacks are significantly more likely to report leaving the gang due to being arrested. Though the differences are not significant at the 0.05 level, greater percentages of black former gang members report leaving the gang as a result of getting a job, going to jail, religious motivation, and growing older. Former black gang members are much more likely to report being influenced by the violence of gang life in their decision to leave. Comparatively fewer Hispanic gang members report fear of violence or simply being tired of violence as reasons for leaving the gang. Blacks are not as likely to report drug involvement as a reason for leaving the gang. Only one black reports leaving the gang due to drug use, and no blacks report leaving the gang due to becoming involved in drug sales. There is some evidence that either of these behaviors are more easily tolerated within the structure of the black youth gang than in the Hispanic (Spergel 1990). CONDITIONS OF SERVICE RECEIPT AND HELPFULNESS EVALUATION Above we found differences in service receipt and service helpfulness evaluations by site, ethnicity, and gang status. While these differences at the univariate level of analysis are of interest to our evaluation, we now perform multivariate analyses to determine what factors account for differences when other variables are controlled. In Table 11, we noted significant differences in our summary measures for number of services received and average helpfulness ratings for services received by site. Here we ask whether those differences might be the result of other characteristics of the client pools from each site. Two variables that might account for variations in service receipt and helpfulness evaluation by site are the age and level of education of clients. Neither of these variables, however, is significantly related to either average number of services received per client or average helpfulness rating of services received as is seen in Table 35. (Note that we only had these measures for 71 of our respondents.) Hence, it is not possible that age or education can account for differences in service receipt and average helpfulness ratings that we found for other factors including site. Table 36 reports the results of general linear model analyses of variance for number of services received and average helpfulness rating of services received. Only the five program sites (excluding Fort Wayne) are included in the analysis. Only the general categories black and Hispanic are included as measures of the effect of race and ethnicity. For both dependent variables, the main effects for site remain significant parts of the model. For number of services received, there is no evidence of significant interactions between any of the dependent variables. In addition to the significant effect for site, the main effects for gang status and ethnicity are also significant components of the analysis of variance for number of services received. While the main effect for site is significant at the 0.001 level for average helpfulness rating, neither the main effect for race/ethnicity and gang status are significant at the 0.05 level. There is, however, a significant interaction effect between gang status and site. The nature of this interaction can be seen in Table 37. There are actually five sources of interaction between site and gang status. Three of them, though, are the result of the one female former gang member in Columbus and her very low (high, numerically) helpfulness rating for the services she reports receiving. One of the other two sources of interaction involves non-members and young former male members. Non-members in East LA Skills Center and Ethan Allen rate the services that they have received as more helpful than do younger former members. On the other hand, non- members in Columbus rate the services that they have received as less helpful than do younger former members there. The other source of interaction is between present and younger former gang members across sites. In East LA, young former gang members evaluate the services that they receive as less helpful than do present gang members. At Ethan Allen, Chino, and El Monte, it is the younger former gang members who rate the services that they have received as more helpful in comparison to present gang members. It seems reasonable to conclude that these interactions between site and gang status are not of a kind that would allow us to ignore the significant main effect of site noted in Table 36. Still, we must close this section with a caution that our measures, number of services received and average helpfulness rating for services received, do not take into account the nature of the specific combination of services received nor the intensity of the services received. PREDICTING LEAVING THE GANG Given our data, we must use caution in building a regression model of leaving the gang. We do have present gang members from four of our program sites plus additional former gang members from our other site, the GVRP staff respondents, and our symposia participants. The resulting model of a logistic regression analysis is no surprise but captures well the frustrations faced by gang program evaluators in general (Spergel 1990). As Table 38 shows, only age is a significant predictor of having left the gang even though a sizable number of younger former gang members are included in the analysis.1 It is very important to note that neither number of services received and average helpfulness rating are significant predictors of leaving the gang. Also, a number of services and average helpfulness of services are not significantly correlated with any of the reasons for leaving the gang, with or without racial/ethnic controls. SUMMARY Clients and former gang members were surveyed at six program sites and two symposia in an effort to discover what services they report receiving and how helpful they feel those services to be in reducing their involvement in criminal gang activity. Of the 124 respondents, almost half (59 or 47.6 percent) are former gang members, the remainder are present members of gangs and presumably persons at-risk for becoming gang members. About a fifth (17.7 percent) are females. A majority of the respondents (66.1 percent) are Hispanic, and 29.1 percent are black. The largest subgroup of Hispanics are Mexican-Americans (81.7 percent of Hispanics). For all respondents, the most commonly reported services are recreation and sports (by 60 percent). Recreation and sports are also the services most highly ranked in terms of helpfulness by our respondents at an aggregate level. The service reported second most helpful is job placement, but it is reported as the tenth most frequently received. There are significant differences in the proportion of respondents receiving services by program site. In general, clients from Chino and Columbus report receiving more services than respondents from East LA Skills Center and El Monte. Ethan Allen falls somewhere in between. Also, Chino respondents rate services received as significantly more helpful than do respondents from East LA Skills Center and El Monte. However, when the relationship between the proportion of clients receiving specific services and the perceived helpfulness of services is analyzed, Chino is the site where respondents are least likely to receive the services that are rated as most helpful. Respondents from Columbus and East LA Skills Center are most likely to report receiving those services that they evaluate as most helpful, even though, in general, these two groups of respondents do not necessarily rate all the services that they receive as helpful. A comparison of the complete ranking orders of service receipt and helpfulness rankings by site indicates that examining the rankings for each site is merited. When we examine service receipt and helpfulness evaluation by respondent involvement in gang activity, we again find significant differences among groups. Male former gang members, under age 21, report receiving significantly more services than non-members and older former gang members. Present gang members report receiving significantly more services than non-members. The differences in average helpfulness rating are not statistically significant. only former male gang members over 21 are not more likely to have received the services that they consider most helpful. Again comparisons of complete service rankings by gang membership status indicates that differences in proportion of a subgroup receiving the service and the helpfulness ratings for services significantly differ across gang membership status subgroups. Comparisons of service experience by race and ethnicity reveal, in general, that Hispanics report receiving fewer services than blacks. On average, Puerto Rican respondents report receiving fewest number of services and feel that the services that they have received are least helpful. Other Hispanic respondents rate the services that they have received as most helpful. While black other respondents tend to report receiving more services, they are the only ethnic subpopulation for whom services received are not significantly related to how helpful respondents feel these services to be. Comparison of overall service and helpfulness rankings by race and ethnicity subgroups indicates that there is not as much diversity in overall rankings by race and ethnicity as there is by site or gang membership status. The most noteworthy difference in helpfulness rankings is between Puerto Ricans and other racial and ethnic groups. When we look at differences between black and Hispanic clients we see that our respondents include significantly a larger proportion of black former as compared to present gang members. There are also different patterns in the number of services received and the average helpfulness ratings of services received across the two groups of respondents. When we look at reasons for leaving the gang, blacks are more likely to report being influenced by arrest and violence in their leaving the gang than are Hispanics. Hispanics are more likely to attribute their leaving the gang to drug involvement. In general, the two most commonly reported reasons for leaving the gang are getting older (42.4 percent) and getting arrested (37.3 percent). Female former gang members do not report getting arrested or going to jail as reasons for leaving the gang. An analysis of number of services received shows that there is no interaction between site, race and ethnicity, and gang membership status in explaining number of services received by clients. The direct effects of each of these variables is significant in explaining the variation of number of services received. A mild interaction between site and gang membership status is found in explaining variation in average helpfulness rating for services received. Only the direct effect of site is significant in explaining variation in average helpfulness rating. Neither number of services received and average helpfulness rating are significant predictors of leaving the gang. A logistic regression analysis reveals that only age of respondent significantly predicts whether or not the respondent is still an active gang member. REFERENCES National Youth Gang Suppression and Intervention Project, "Symposium of Former Hispanic (Mostly Puerto Rican) Gang Influentials, " Chicago: School of Social Service Administration, University of Chicago, May 27, 1989. National Youth Gang Suppression and Intervention Project, "Symposium of Former African-American Gang Influentials," Chicago: School of Social Service Administration, University of Chicago, August 26, 1989. Spergel, Irving A., "Youth Gangs: Continuity and Change," in Michael Tonry and Norval Morris (editors), Crime & Justice: A Review of Research, Volume 14, Chicago: University of Chicago Press, 1990. Spergel, Irving A., and Ron L. Chance, "Community and Institutional Responses to the Youth Gang Problem: Case Studies Based on Field Visits and Other Materials," Chicago: National Youth Gang Suppression and Intervention Project, School of Social Service Administration, University of Chicago, January 1990. ENDNOTE: 1. Since our data is exploratory in nature, we conducted a forward stepwise logistic regression analysis in which our 22 services were allowed to serve as potential predictors of having left the gang. Only two of our services variables actually enter logistic regression models as predictors -- supervision/parole and assistance with school problems. Both of these variables have negative coefficients meaning that their impact on leaving the gang as measured here is in the opposite direction of that intended. APPENDIX As part of a national youth gang research and program development project, we are conducting an assessment of organizations and programs that deal with gang-related problems in this area. You will not be identified as an individual in any reports that are generated from this information. Your CONFIDENTIALITY is assured. No names of any individual person should appear on this form.