Journal of Criminal Justice, Vol. 27, No. 4, pp. 309–320, 1999 Copyright © 1999 Elsevier Science Ltd Printed in the USA. All rights reserved 0047-2352/99 $–see front matter
PII S0047-2352(99)00007-0
FAMILY STRUCTURE AND RACE IN A SAMPLE OF CRIMINAL OFFENDERS
Vic W. Bumphus Department of Police Studies Eastern Kentucky University Richmond, Kentucky 40475
James F. Anderson Department of Sociology, Criminal Justice, and Criminology University of Missouri at Kansas City Kansas City, Missouri 64110-2499
ABSTRACT Seen through social control and social learning perspectives, family structure is assumed to have little effect on criminal behavior, while family functioning factors have comparatively strong effects. This investigation posited that family structure, when measured dichotomously (“intact” versus “broken” homes), misrepresents contemporary family arrangements, especially with regard to racial and cultural differences. The study utilized a multidimensional measurement of family structure and relates it to several categories of criminal behavior. Research findings suggested that traditional measures of family structure relate more to the criminal patterns of Caucasians than to those of African Americans or Hispanics. © 1999 Elsevier Science Ltd. All rights reserved.
INTRODUCTION Criminological theory has long recognized the importance of familial characteristics as they relate to crime and delinquency (Brownfield and Sorenson, 1993; Cernkovich and Giordano, 1987; Hirschi, 1969; Johnson, 1986; Nye, 1958; Rosenbaum, 1989). Early considerations of family influences on delinquency suggested
that the traditional family was less criminogenic than others (Moynihan, 1965; Nye, 1958; Shaw and McKay, 1942; Wadsworth, 1979). In recent years, however, theoretical and empirical studies have reported that family functioning dimensions (i.e., attachment, cohesion, conflict, and involvement) are more closely associated with crime and deviance (Cernkovich and Giordano, 1987; McCord, 1991; Van Voorhis et al.,
309
310
V. W. BUMPHUS and J. F. ANDERSON
1988; Wells and Rankin 1988). The substantial influences of social learning and social control theories have no doubt fueled contemporary research. As Van Voorhis et al. (1988, p. 258) conclude, “bad homes not broken homes, it seems, place youths at risk.” The importance of family functioning is well established in relation to delinquent and criminal behavior. This does not mean that family structure (i.e., various dimensions of parental presence in the home) factors have no utility in explaining deviance. There are several compelling reasons for examining family structure in relation to crime. First, informal dimensions of social control have a more direct effect on delinquent behavior than formal control mechanisms (Grasmick, Bursik, and Kinsey, Unpbl.). This implies that family functioning is influenced by the size and extension or structure of the family unit. In this sense, individuals may consider intimate others who might react negatively to their actions, which may influence their criminal behavior. Second, empirical studies of family structure have utilized (mostly) unidimensional family constructs. The model approach has been to differentiate intact and broken homes, indicating the absence of at least one parent (Johnson, 1986). This dichotomy fails to represent contemporary family structure, especially regarding racial and cultural differences. Third, family structure and family functioning factors may be interrelated in important ways. One conclusion of recent scholarship on family structure is that it has little or no association with criminal behavior (Johnson, 1986; Van Voorhis et al., 1988). If this research is valid, it indicates that intact, as well as broken, homes share dysfunctional dimensions that are related to criminal behavior. The presence and number of positive or negative role models become critical factors that influence crime. The research reported in this investigation addressed three important questions: (1) Is family structure measured in multidimensional terms related to youthful offending? (2) Is family structure related to different categories of offending by race? and (3) Does family structure and crime vary significantly by racial status?
FAMILY STRUCTURE AND DELINQUENT BEHAVIOR In the 1960s, the influential Moynihan Report concluded that broken homes are directly associated with delinquent behavior, especially for African American adolescents (Moynihan, 1965). Despite this, Chilton and Markle (1972) and Austin (1978) found that broken homes correlate more with delinquency among Caucasian youths. Johnson (1986) suggested that, while studies of juveniles processed through the juvenile justice system support the contention that broken homes have an impact on delinquency, studies utilizing self-reported data have found few or no direct effects of family structure on juvenile delinquency. Social processing and control theorists report that in homes characterized by an absent parent, conflict and tension, or a lack of familial love and support, youths reared in these environments will be more susceptible to crime-producing forces (Lincoln and Straus, 1985; Loeber and Loeber 1986). Matsueda and Heimer (1987), examining the association between broken homes and crime, tested differential association and control theory. They reported that, for African Americans, broken homes foster an excess of definitions favorable to delinquency. The effect is found to be greater among African Americans who live in broken homes than among their Caucasian counterparts. African Americans living in broken homes located in criminogenic neighborhoods would be more likely than those in crime-free neighborhoods to have associations with delinquent and criminal peers who provide reasons favorable to engaging in delinquency and crime. Matsueda and Heimer (1987) attributed the causes of delinquency to broken homes, unsupervised family life, and social disorganization. As for social control theory, they argued that it predicts greater delinquency in broken homes rather than in intact homes because youths from broken homes have weak attachments to parents. Since the father is usually absent, such families are less likely to provide needed supervision and discipline. Glueck and Glueck (1950), McCord and McCord (1959), and Goetting (1994) argued that
Family Structure and Criminal Offenders
an established relationship exists between family structure and delinquency or criminal behavior. Studies have found that adolescents living with biological parents have lower rates of delinquency and crime than those living in broken or reconstituted homes. Single parents may be unable to provide the same amounts of supervision, monitoring, and consistent discipline that can be provided by two-parent families (Gottfredson and Hirschi, 1990; Wilson and Herrnstein, 1985). Absent fathers place a greater economic strain on the family, thereby forcing mothers to join the workforce. As a result, mothers are unable to provide adequate supervision for monitoring children (Gottfredson and Hirschi, 1990). Nye (1958) considered the loss of direct control due to maternal employment as causally related to delinquency, while others contend that a lack of maternal supervision is strongly related to delinquency (Glueck and Glueck, 1950; Hirschi 1969). Wilson (1984) argued that a disproportionate amount of African American involvement in crime can be explained by the absence of fathers. Miller (1958) contended that female-headed households lead to crime because mothers are left with the responsibility of rearing African American male youths. Many lower-class African American youths, as a result, are preoccupied with proving their toughness and masculinity and, as a reaction to maternal control, these adolescents engage in crime and physical violence. Stepparents report a lack of parental “attachments” toward stepchildren. Hirschi (1969, p. 242) found that for both Caucasian and African American boys, “only those living with step or foster fathers are more likely than children from ‘intact’ homes to be delinquent.” Austin (1992) contended that the stigma of abandonment, illegitimacy, and divorce leaves youths feeling disgraced, often causing emotional problems that lead to crime. McCord (1991) contended that single parent families are not more criminogenic than two-parent families. Austin (1992) found no relationship between female family headship and delinquency, and suggested that when it comes to instilling nondelinquent attitudes, female-headed households are just as effective as two-parent households. A review of the literature reveals inconsistent claims about family structure and its impact
311
on juvenile delinquency and crime. Despite some apparent contradictions, researchers have found that family structure has a differential impact on crime by race. Research on family structure and delinquency has involved two major methodological shortcomings (Johnson, 1986; Lincoln and Straus, 1985; Loeber and Loeber, 1986; Van Voorhis et al., 1988; Wells and Rankin, 1988). First, family structure variables measured dichotomously fail to represent contemporary family structures, especially with reference to certain ethnic and minority groups. It can be argued that family traditionalism by race and ethnicity suggests important differences that have theoretical and empirical implications. The impact of the nuclear family on criminal behavior could be considerably different from that of the extended family. The extent to which minorities reside within extended family systems is a subject worthy of theoretical consideration. Staples noted that: Kinship bonds have always been important to the Black population. In African societies, kinship was and is the basis of social organization. During the period of slavery, many of the bondsmen were organized into an extended family system based on biological and nonbiological standards. Most research studies of Black kinship networks generally indicate that they are more extensive and significant to the Black community than to the White community. (Staples, 1994, p. 154)
Second, as Johnson (1986) suggested, family structure and family functioning variables have been offered as competing perspectives as they relate to crime and delinquency. Johnson (1986, p. 66) stated that “researchers have generally overlooked the interrelationships among other variables thought to be factors in the etiology of delinquent behavior in their own right.” A case can be made for studying family structure and functioning both independently and simultaneously. The present study only considered the influence that family structure has on crime by race. This research examined family structure and criminal behavior through self-reported data from a sample of incarcerated offenders. The analysis presented here is similar to that used by Johnson (1986). Where most (contemporary) re-
312
V. W. BUMPHUS and J. F. ANDERSON
searchers have utilized a general population sample, this investigation used an offender sample.
DATA AND METHOD A random sample of 639 participants was selected from a population of 1,220 offenders who were present at Courts Regimented Intensive Probation Program (CRIPP) from 1991 to 1994 in Humble, Texas (Anderson, 1995). This research measured the impact of family structure and race on self-reported levels of youthful offending. The data were collected over a fouryear period. To determine the impact of family structure on criminal processing and self-reported criminal behavior, this study relied on two instruments: a 91-item survey (Harris County [TX] Lifestyles Survey) given to participants pre- and post-boot camp participation; and the Justice Information Management System (JIMS) for analyzing (corroborating) offender involvement in crime. The random sample of 639 for this analysis was chosen from the entire population of participants who had graduated at the beginning of the study.1 Subjects The characteristics of the sample are found in Table 1. They revealed that the average age of participants was 19.5 years. African Americans were disproportionately represented at 44.4 percent. Caucasians comprised nearly 35 percent (34.9). Hispanics totaled nearly 20 percent (19.6). Other racial categories comprised only 1.1 percent, therefore, the “other” race category respondents were excluded from the analysis that examined racial differences. Approximately half (49.7 percent) of the sample reported being reared by both parents. Another 37.5 percent were reared by their mothers only. Very few respondents (5.4 percent) reported being reared by their fathers only, or in some other family arrangement (7.4 percent). Slightly over 50 percent (n 5 311) of the respondents reported being incarcerated as a result of committing property offenses. The second most common type of conviction was for committing drug offenses, at 30.4 percent. The
remaining 19.2 percent were convicted for engaging in personal offenses. On average, respondents reported their first arrest at 16.7 years, with 3.5 lifetime arrests. With regard to self-reported crime variables, respondents averaged 1.1 and .53 lifetime felonies and misdemeanors, respectively. Respondents had spent an average of 11.3 months on probation and .19 years in prison. Measures The survey instrument used to collect data from CRIPP participants focused on offender lifestyles and family backgrounds. It attempted to assess graduates’ attitudes toward CRIPP and its impact on recidivism. Six programmatic areas were examined: (1) perception of boot camp staff, (2) drug and alcohol counseling, (3) perception of future opportunities, (4) general perception of the boot camp program, (5) impulsivity and self-control, and (6) family arrangement. This investigation focused on aspects of the last area—family arrangement. The Harris County (TX) Lifestyle Survey was designed for pre- and post-measures of items from the programmatic areas. To determine the type of household in which respondents grew up, respondents were asked to select one of the following questions: (1) Were you raised by mother and father? (2) Were you raised by your mother only? (3) Were you raised by your father only? or (4) Were you raised by other family members or a guardian? Other questions were used to obtain rates of past criminal behavior: (1) Did you commit a misdemeanor before boot camp? (2) Did you commit a felony before boot camp? (3) List the number of before (prior incarceration) misdemeanors; (4) List the number of before (prior incarceration) felonies; (5) List the number of years or months in prison; (6) List the number of years or months on probation; and (7) List the number of times arrested. The effect of family structure on criminal behavior was assessed by the use of four family structure categories: (1) raised by mother and father (MOMDAD), (2) raised only by mother (MOMONLY), (3) raised only by father (DADONLY), and (4) raised by some other family member (OTHERFAM). All respondents matched one of the family structure categories.
313
Family Structure and Criminal Offenders
TABLE 1 DEMOGRAPHIC CHARACTERISTICS OF THE SAMPLE, FAMILY STRUCTURE CATEGORIES, AND SELF-REPORTED CRIME DESCRIPTIVES
Variables Age* 17–19 20–22 23–25 Gender Male Race African American Hispanic Caucasian Other Family structure* Mother and father Mother only Father only Other family Current offense type Drug Property Personal First arrest age (Arst_Age) Number of arrests (Arst_Num) Self-reported felonies (BFEL) Self-reported misdemeanors (BMISD) Prison months Probation months
(N)
%
376 177 79
59.4 28.1 12.5
639
Mean
SD
Range
19.5
2.2
8.0
16.7
2.4
15.0
3.5
3.6
39.0
100
284 125 223 7
44.4 19.6 34.9 1.1
314 237 34 47
49.7 37.5 5.4 7.4
187 311 118
30.4 50.5 19.2
1.1 .53 .19 11.3
.61
4.0
.96 2.3 19.6
8.0 48.0 120.0
*Both the age and family structure categories have seven missing observations; therefore, percentages in these categories are valid ones.
The family structure variables for the bivariate analysis were “dummy coded” for each of the four groups. Living with mother and father, mother only, father only, and other family member were all coded 1, representing the stated category, and 0, representing some other status. The self-reported crime variables, BFEL (number of before incarceration felony offenses) and BMISD (number of before incarceration misdemeanor offenses), were numeric values that indicated the actual number reported by each participant. The age variable was also represented by a numeric value. In the multiple discriminant analysis, the family structure variable was categorical (Group 4 5 raised by mother and father; Group 3 5 mother only; Group 2 5 father only; and Group 1 5 some other family member). The PROBATION variable represented the number of months respondents re-
ported being on probation (over his lifetime). The PRISON variable measured the number of self-reported years in prison. The means for probation and prison periods measured average months and years, respectively. Discriminant analysis is appropriate when dependent variables are measured categorically and independent variables, metrically. Using this statistical procedure, dependent nominal variables can include three or more attributes. In multiple discriminant analysis, the main objective is to identify group differences (Hair et al., 1995; Klecka 1980). The dependent categorical variables (family structure and race) used in the analysis as mean differences were observed in the metric crime variables (i.e., first arrest age, number of arrests, self-reported misdemeanor and felonies, and self-reported months of prison and probation).
314
V. W. BUMPHUS and J. F. ANDERSON
Multiple discriminant analysis requires approximately twenty observations in each independent group. In addition, dependent groups should have substantially more observations than the number of independent variables (Hair et al., 1995). In all cases, the dependent groups numbered twenty or more and outnumbered the independent variables in the model. There was, however, substantial variance across the family structure and race categories. This variance may have given larger groups a higher chance of discriminant classification. When group variances vary too greatly, one acceptable approach is to randomly reduce larger groups proportionate to smaller ones (Hair et al., 1995). There were a substantial number of observations in most of the dependent groups, so researchers opted to go forth without reducing dependent groups in the sample. Investigators interpreted the findings related to the family structure groups, other family and father only, with due caution. For the analysis (Table 7) concerning race, the seven observations representing some other racial category were excluded from the analysis. This investigation used JIMS to corroborate self-reported crime (see Anderson 1995). It tracked the frequency of reoffending by boot camp participants in Harris County, Texas. The JIMS is a database that contains criminal histories of offenders who have been processed by the Harris County Criminal Justice Institutional Division. Containing more than 42 million records, it is considered the most extensive database system in the nation. This system operates by assigning each offender a system number that remains from incarceration to disposition.
FINDINGS First, researchers examined bivariate relationships between family structure and race categories included in the sample. Second, researchers conducted a multiple discriminant analysis controlling for family structure variables and race as related to mean differences in self-reported offending, arrest, and criminal sanctioning (parole and prison). Bivariate Analysis Table 2 depicts the bivariate associations between family structure, age, and self-reported crime variables. The only significant correlations in the self-reported misdemeanor category (BMISD) are age and number of arrests (ARST-NUM) indicating a positive relationship between these variables and this crime category. Number of arrests is positively correlated with being raised by father only. Self-reported felonies (BFEL) also correlate positively with age. In addition, this category of offending correlates significantly with one of the family structure variables. The significant correlation between BFEL and other family suggests that those reared by other family members reported fewer felonies than other groups in the sample. The remaining bivariate correlations were expected due to dummy coding of categorical family structure variables, which would logically result in significant relationships. Taken together, the bivariate correlations were weak with regard to the strength of associations.
TABLE 2 CORRELATIONS FOR SELF-REPORTED CRIME, AGE, AND FAMILY STRUCTURE VARIABLES
BMISD
BFEL
BMISD 1.000 2.014 BFEL 2.014 1.000 AGE .169** .124** ARST_NUM .109** .047 DADONLY .035 2.017 MOMONLY 2.024 .035 MOMDAD .015 .024 OTHERFAM 2.014 2.097*
AGE .169** .124** 1.000 .032 2.005 .077 2.061 2.022
Note: See text for abbreviation explanations. *p , .05; **p , .01 (2-tailed significance).
ARST_NUM DADONLY MOMONLY MOMDAD OTHERFAM .109** .047 .032 1.000 .080* 2.009 .001 2.053
.035 2.017 2.005 .080* 1.000 2.185** 2.237** 2.068
2.024 .035 .077 2.009 2.185** 1.000 2.770** 2.220**
.015 .024 2.061 .001 2.237** 2.770** 1.000 2.282**
2.014 2.097* 2.022 2.053 2.068 2.220** 2.282** 1.000
315
Family Structure and Criminal Offenders
TABLE 3 MULTIPLE DISCRIMINANT ANALYSIS EQUALITY TEST OF GROUP MEANS BY FAMILY STRUCTURE
Variable
Mean
SD
Wilks’ Lambda
F
Significance
X1 ARST_NUM X2 BFEL X3 BMISD X4 Probation X5 Prison X6 Age X7 ARST_AGE
3.5 1.1 .5 11.6 .2 19.5 16.9
3.7 .6 .9 19.7 2.3 2.2 2.4
.99125 .98953 .99789 .99295 .98373 .99376 .99518
1.8000 2.1595 .4319 1.4488 3.3730 1.2814 .9886
.14 .09 .73 .22 .01* .27 .39
Group differences1 PRISON Group 1 Group 2 Group 3 Group 4
x 5 .23 x 5 1.41* x 5 .05 x 5 .16
Note: See text for abbreviation explanations. Wilks’ lambda and univariate F ratio with 3 and 612 degrees of freedom. 1 Group 1 5 other family (n 5 47); group 2 5 father only (n 5 34); group 3 5 mother only (n 5 237); group 4 5 mother/father (n 5 314). * p , .01.
Multiple Discriminant Analysis The multiple discriminant analysis, reported in Tables 3 through 7, shows means and standard deviations of self-reported offending, criminal sanctioning, and age. Table 3 indicates general sample mean differences in reference to the independent variables across the four family structure groups (4 5 mother/father; 3 5 mother only; 2 5 father only; and 1 5 other family member). Only one mean is significantly different for the general sample for average
self-reported prison years. This statistic reveals a significant mean difference between those in family structure Group 2 (father only) and the remaining family structure groups (x 5 1.41). The means for prison years are significantly lower in family structure groups 1 (x 5 .23), 3 (x 5 .05), and 4 (x 5 .16). All models were abandoned because none of the factors in the discriminant analysis (for any model) loaded after the second step in the procedure. The equality of means tests approximate analysis of variance findings.
TABLE 4 MULTIPLE DISCRIMINANT ANALYSIS EQUALITY TEST OF GROUP MEANS BY FAMILY COMPOSITION (CAUCASIAN SUBSAMPLE, n 5 223)
Variable
Mean
SD
Wilks’ Lambda
F
Significance
X1 ARST—NUM X2 BFEL X3 BMISD X4 PROBATION X5 PRISON X6 AGE X7 ARST—AGE
4.0 1.1 .5 12.1 .1 19.7 16.8
4.7 .6 .9 18.6 1.3 2.3 2.4
.94885 .96131 .96104 .99327 .99017 .98634 .99598
3.7194 2.7768 2.7970 .4676 .6853 .9552 .2785
.01** .04* .04* .70 .56 .41 .84
ARST—NUM differences1
Group Group 1 Group 2 Group 3 Group 4
BFEL x 5 2.3 x 5 7.7** x 5 4.1 x 5 3.5
x 5 .67* x 5 1.2 x 5 1.1 x 5 1.1
BMISD x 5 .13 x 5 .92* x 5 .38 x 5 .60
Note: See text for abbreviation explanations. Wilks’ lambda and univariate F ratio with 3 and 207 degrees of freedom. Group 1 5 other family; group 2 5 father only; group 3 5 mother only; Group 4 5 mother/father. *p , .05; ** p , .01.
1
316
V. W. BUMPHUS and J. F. ANDERSON
TABLE 5 MULTIPLE DISCRIMINANT ANALYSIS EQUALITY TEST OF GROUP MEANS BY FAMILY STRUCTURE (HISPANIC SUBSAMPLE, n 5 125)
Variable
Mean
SD
Wilks’ Lambda
F
Significance
X1 ARST—NUM X2 BFEL X3 BMISD X4 PROBATON X5 PRISON X6 AGE X7 ARST—AGE
2.9 .99 .58 11.2 .47 19.3 17.3
2.78 .64 .90 22.5 4.4 2.1 2.4
.99895 .94897 .99051 .97978 .67854 .97134 .98114
.0405 2.0792 .3705 .7980 18.3187 1.1410 .7432
.98 .10 .77 .49 .00* .33 .52
Group differences1 PRISON Group 1 Group 2 Group 3 Group 4
x 5 .00 x 5 .16* x 5 .17 x 5 .00
Note: See text for abbreviation explanations. Wilks’ lambda and univariate F ratio with 3 and 116 degrees of freedom. 1 5 other family; group 2 5 father only; group 3 5 mother only; group 4 5 mother/father. *p , .01. 1Group
Table 4 summarizes a multiple discriminant test for Caucasian respondents by family structure. There are significant differences across this subsample in terms of self-reported felonies, misdemeanors, and arrests. Family structure Group 2 (father only) is observed to have a significantly higher rate of self-reported arrests (x 5 7.7, compared to xs of 3.5, 4.1, and 2.3 for Groups 4, 3, and 1, respectively).2 The same pattern exists for misdemeanors, which indicates that Group 2 had mean misdemeanor rates higher than the other family structure groups. Consistent with the bivariate analysis, Group 1 (other family) participants self-reported felonies at a significantly lower rate. The mean for other family is .67, while comparable means are 1.2 (for father only), 1.1 (for mother only), and 1.1 (for mother and father).
For the Hispanic subsample, Table 5 reveals one significant mean difference, which suggests that, as with the general sample, family structure Group 2 (father only) exhibits a significantly higher rate of self-reported prison years (group 2, x 5 16.0, subsample x 5 .46). No other means within the Hispanic subsample reveal significant differences. As shown in Table 6, there are no significant mean differences with any family structure group for the African American subsample, suggesting that family structure has comparatively less influence of the criminal behavior of African American offenders. The final discriminant analysis test for equality of means substitutes race (and excludes family structure) and utilizes the independent crime factor list. This analysis, shown in Table
TABLE 6 MULTIPLE DISCRIMINANT ANALYSIS EQUALITY TEST OF GROUP MEANS BY FAMILY STRUCTURE (AFRICAN AMERICAN SUBSAMPLE, n 5 284)
Variable
Mean
SD
Wilks’ Lambda
F
Significance
X1 ARST—NUM X2 BFEL X3 BMISD X4 PROBATION X5 PRISON X6 AGE X7 ARST—AGE
3.5 1.2 .5 11.6 .1 19.5 16.9
3.1 .6 1.1 19.4 1.6 2.0 2.4
.99060 .99319 .99265 .98622 .99323 .97465 .99244
.8664 .6262 .6766 1.2761 .6226 2.3758 .6957
.45 .59 .56 .28 .60 .07 .55
Note: See text for abbreviation explanations. Wilks’ lambda and univariate F ratio with 3 and 274 degrees of freedom.
317
Family Structure and Criminal Offenders
TABLE 7 MULTIPLE DISCRIMINANT ANALYSIS EQUALITY TEST OF GROUP MEANS BY RACE
Variable
Mean
SD
Wilks’ Lambda
F
Significance
X1 ARST—NUM X2 BFEL X3 BMISD X4 PROBATON X5 PRISON X6 AGE X7 ARST—AGE
3.5 1.1 .5 11.6 .2 19.5 16.9
3.7 .6 .9 19.7 2.4 2.1 2.4
.98916 .98966 .99904 .99978 .99685 .99627 .99571
3.3543 3.1982 .2935 .0680 .9669 1.1447 1.3172
.03* .04* .74 .93 .38 .31 .26
ARST-NUM Group differences1 Group 1 Group 2 Group 3
BFEL x 5 3.9* x 5 2.9 x 5 3.4
x 5 1.0 x 5 1.0 x 5 1.2*
Note: See text for abbreviation explanations. Wilks’ Lambda and univariate F-ratio with 2 and 612 degrees of freedom. Group 1 5 Caucasian; group 2 5 Hispanic; group 3 5 African American. *p , .05.
1
7, was conducted to establish whether there were overall differences in mean rates by race for self-reported crime (BFEL, BMISD, ARSTNUM), PRISON, PROBATION, AGE, and family structure. The analysis reveals differences among group means for ARST-NUM and BFEL. The number of arrests observation indicates that Caucasian youths in the sample report significantly higher rates of arrest (x 5 3.9) than either Hispanic (x 5 2.9), or African American youths (x 5 3.4). With regard to self-reported felonies, African American youths (x 5 1.2) report significantly higher rates than either Caucasian (x 5 1.1) or Hispanic youths (x 5 1.0). With the exception of self-reported felonies and arrests, no significant mean differences exist for the sample by race. In recent years, there has been a growing reliance on crime data collected by law enforcement agencies. Official crime data only contain crimes reported to police, therefore, these data (e.g., Uniform Crime Reports) underestimate the true nature and extent of crime. As a result, many scholars believe that self-reported and victimization data are more accurate measures of crime (Doerner and Lab, 1995). There are mean differences in offending by race in the sample, though the differences do not approach the levels reported in official statistics. The selfreported data in the present research was corroborated by official criminal records.
DISCUSSION What does the analysis reveal about family structure and crime? First, the findings suggest that a multidimensional family structure construct reveals variation in its impact across race and crime categories. Similar to Johnson (1986), this research found that criminality may be greater in intact families or in arrangements with at least one parental figure. For example, Johnson found that delinquency is more common in Caucasian family structures composed of a mother and a stepfather. He argued that this could be due to the lack of attachment between boys and their stepfathers. Adolescents resent stepfathers and the absence of their biological fathers, so they demonstrate their antagonism by engaging in criminal behavior. The findings from this research contradict conventional literature measuring family structure in terms of intact or broken homes. This research conceded that some of the respondents reared by other family members (family structure Group 1) may have separated from the nuclear arrangement because of family discord. In addition, some of these individuals may reside in other family groupings because of necessity (e.g., death of a parent, abandonment, parental neglect, or incompetence). For persons separated from parental attachments due to family discord, the reconstituted family structure most
318
V. W. BUMPHUS and J. F. ANDERSON
likely influenced family functioning in terms of assessing offender patterns. This research found that individuals living in other family arrangements reported less serious crime, an observation that contradicts reports indicating intact versus broken homes as more deviant. This study showed that the conventional measurement of family structure may be inadequate. In terms of criminal behavior, the dichotomy of being reared in some other family arrangement or with at least one parent is a more useful measure of family structure than intact versus broken homes. The broader implications of this research, however, call for a more comprehensive measurement of family structure. This research showed that individuals reared by both parents were equally or more deviant with respect to felonies than individuals reared in other family arrangements. Much of the social control literature suggested that parental control and supervision have an impact on delinquency and crime. Family structures that encompass more appropriate parental figures may produce fewer criminals. Collins (1991) has noted the tradition of “other mothers and fathers” in minority communities. In addition, scholars have noted that the crime level in societies characterized by extended family kinships is much lower than in other, more diverse, industrial societies (Scott and Black, 1989). The cultural diversity within American society necessitates a more comprehensive measurement of family structure. In answer to the first question posed by this research (Does a multidimensional measurement of family structure have an impact on criminal behavior?), this research finds that family structure may have an impact on crime regardless of whether youths reside in nuclear or traditional family arrangements. This study also suggests that family structure measured with “at least one parent” or some “other family” arrangement may be more appropriate than conventional measurements. Is family structure related to different categories of crime by race? Discriminant analysis findings indicate that traditional family structure variables may be more important in explaining crime among Caucasian youth than for other racial or cultural groups (see Austin, 1978; Chilton
and Markle, 1972). This study suggests that family structure has a significantly different impact on crime by race. Youthful offending in the sample is more associated with Caucasian persons who were reared by their fathers. These youth were more likely to commit fewer felony violations and had higher arrest and misdemeanor rates. Among the other race categories, fatherheaded households were linked to Hispanics’ probability of going to prison. Caution should be exercised with regard to the interpretation of mean differences for this observation due to the comparatively small number of respondents who had been incarcerated in the sample. Family structure has been considered central to criminal behavior among minority adolescents. Family structure may have an impact on the criminal behaviors of these groups, though conventional measures of family structure may apply more to Caucasian than to minority youths. Previous measurements (the present research included) have fallen short of representing racially and culturally relevant dimensions of family, especially the extended kinship groupings most often associated with minorities in contemporary society (Scott and Black, 1989). Having taken into account the majority of current research on family structure and crime, criminal behavior is said to exist at comparable levels in conventional, as well as nonconventional families. This perspective implies that the composition of the family may be an intervening or influencing variable in explaining levels of criminal behavior. This research has noted the centrality of social learning and social bonding theories in explaining criminal behavior, though this study provides support for expanding the measurement of family structure respective to both theories. Social bonding theory gives primary consideration to the span of control in families. The extended family, being no more criminal, may influence crime in terms of the number of parental figures within familial groupings, providing more caregiver control. This inference was based on an assumption of comparability in terms of criminal behavior across family structure categories. Social learning theory places primary emphasis on familial relationships in terms of modeling
319
Family Structure and Criminal Offenders
and imitation. If bad or inappropriate role models (e.g., parents or siblings) are found in either traditional and nontraditional families, positive (or negative) role models or intimates in extended, larger familial groupings may better explain the relationship between family structure and family functioning factors. These research findings differ somewhat from Johnson (1986), who utilized a similar methodology. First, Johnson found that family structure is related to race but not to criminal behavior. On the other hand, this research found that family structure is related to the criminal behavior of Caucasians in the sample. Johnson (1986) concluded that because of the small number in the sample of the “father only” group, no inferences could be made. This research had a comparatively low number of respondents in the father only group, though the observations were well distributed across race categories (with 6, 3, and 6 percent; Caucasian, Hispanic, and African American, respectively). These findings show that Caucasian youths who reside with their fathers are more criminal (having significantly more arrest and misdemeanor charges) than their Hispanic and African American counterparts. With regard to the third question posed by this research (Does family structure and crime vary significantly by race?), the findings suggest variation for Caucasians, less variation for Hispanics (only in terms of prison rates), and no significant variation for African American youths. This observation implies that family structure, as measured by varying components of parental presence, is associated with criminal behavior among Caucasian youth. The inference is that family structure measurements for both African Americans and Hispanics would benefit from some essential modifications.
CONCLUSION Feminist scholars have articulated the need to construct gender-relevant behavioral models. These scholars are beginning to make some headway. This research advances a similar plea, which is the need to construct culturally and racially relevant measures of family structure. Traditional familial structures may say a great deal
about crime among Caucasian youths, but considerably less about the criminal patterns of minority youths. As Staples (1994) noted, extended family networks may be more common to minority communities. The composition of the family, as it relates to the number of adult caregivers may, therefore, be a valid approach to understanding delinquent and criminal behavior. Research efforts should resist categorizing structure and composition and utilize all household individuals in analysis of family. Family structure may be an important variable in understanding youthful offending, but the rather elastic nature of measuring family structure might be likened to “comparing apples to oranges.”
NOTES 1. The sample utilized for this research was reduced from 653 to 639 since 14 cases had considerable missing data. 2. The number of cases in the sample where juveniles reported being raised by their fathers only was low, though the cases were distributed proportionately by race with 6, 3, and 6 percent falling in the Caucasian, Hispanic, and African American categories, respectively.
REFERENCES Anderson, J. F. (1995). A follow-up investigation on participants of the Harris County, Texas Courts Regimented Intensive Probation Program: CRIPP. Ph.D. dissertation. Huntsville, TX: Sam Houston State University. Austin, R. L. (1978). Race, father absence, and female. Criminology 15:487–504. Austin, R. L. (1992). Race, female headship, and delinquency: A longitudinal analysis. Justice Quarterly 9(4):585–607. Brownfield, D., and Sorenson, A. M. (1993). Sibship size and sibling delinquency. Deviant Behavior 15:45–61. Cernkovich, C., and Giordano, P. C. (1987). Family relationships and delinquency. Criminology 25(2):295–319. Chilton, R. J., and Markle, G. E. (1972). Family disruption, delinquent conduct, and the effect of subclassification. American Sociological Review 37:93–9. Collins, P. H. (1991). Black feminist thought. New York: Routledge Press. Doerner, W., and Lab, S. P. (1995). Victimology. Cincinnati, Ohio: Anderson Publishing Co. Glueck, S., and Glueck, E. (1950). Unraveling juvenile delinquency. Cambridge, MA: Harvard University Press. Goetting, A. (1994). The parenting crime connection. Journal of Primary Prevention 14:167–84. Gottfredson, M. R., and Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press. Grasmick, H., Bursik, R., and Kinsey, K. (Unpbl.). Shame and embarrassment as deterrents to noncompliance with the law: The case of anti-littering campaign. Paper pre-
320
V. W. BUMPHUS and J. F. ANDERSON
sented at the annual meeting (Nov. 1990). Baltimore, MD: American Society of Criminology. Hair, J., Anderson, R., Tatham, R., and Black, W. (1995). Multivariate data analysis. 4th ed. Paramus, NJ: Prentice-Hall. Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University of California Press. Johnson, J. E. (1986). Family structure and delinquency: General patterns and gender differences. Criminology 24(1):65–84. Klecka, W. (1980). Discriminant analysis. Beverly Hills, CA: Sage Publications. Lincoln, J., and Straus, M. (1985). Crime and the family. Springfield, IL: Charles C. Thomas Press. Loeber, R., and Loeber, M. S. (1986). Family factors as correlates and predictors of juvenile conduct problems and delinquency. Crime and Justice: An Annual Review of Research 7:29–149. Matsueda, R. L., and Heimer, K. (1987). Race, family structure, and delinquency: A test of differential association and social control theories. American Sociological Review 52:826–40. McCord, J. (1991). Family relationships, juvenile delinquency, and adult criminality. Criminology 29(3):397–413. McCord, W., and McCord, J. (1959). Origins of crime: A new evaluation of the Cambridge-Somerville study. New York: Columbia University Press. Miller, W. B. (1958). Lower-class culture as a generating
milieu of gang delinquency. Journal of Social Issues 14:5–19. Moynihan, D. P. (1965). The Negro family: The case for national action. Office of Policy Planning and Research. Washington, DC: U.S. Department of Labor. Nye, F. I. (1958). Family relationships and delinquent behavior. New York: John Wiley & Sons. Rosenbaum, J. L. (1989). Family dysfunction and female delinquency. Crime and Delinquency 35(1):31–44. Scott, J. W., and Black, A. (1989). Deep structures of African American family life: Female and male kin networks. Western Journal of Black Studies. 13:17–24. Shaw, C. R., and McKay, H. D. (1942). Delinquency areas. Chicago, IL: University of Chicago Press. Staples, R. (1994). The Black family: Essays and studies. 5th ed. Belmont, CA: Wadsworth Publishing Company. Van Voorhis, P., Cullen, F. T., Mathers, R. A., and Garner, C. C. (1988). The impact of family structure and quality on delinquency: A comparative and functional variables. Criminology 26(2):235–60. Wadsworth, M. (1979). Roots of delinquency. New York: Barnes and Noble. Wells, L., and Rankin, J. (1988). Direct parental controls and delinquency. Criminology 26:263–85. Wilson, J. Q., and Herrnstein, R. J. (1985). Crime and human nature. New York: Simon and Schuster. Wilson, W. J. (1984). The Black underclass. Wilson Quarterly 8(2):88–99.