Objectives. To describe the characteristics of homeless and low-income preschool-aged children, and to identify family and environmental determinants of their behavior.
Methods. An unmatched case-control design was used to recruit a sample of sheltered homeless families and a comparison group of low-income housed families who were never homeless in Worcester, Massachusetts. Seventy-seven sheltered homeless and 90 low-income housed mothers with preschool-age children were assessed using a comprehensive interview protocol. Information about mothers' housing, income, service use, par-enting practices, and children's father was obtained. Data about children's background, health, and life events were included. Standardized instruments were administered to assess mothers' mental health and their children's behavior. Comparisons of homeless and low-income housed families were used to describe the sample of 167 preschoolers. Multiple linear regression was used to examine the association of various stressors, such as homelessness, and family factors with their behavior.
Results. Although homeless preschoolers were significantly more likely to have experienced stressful life events, undergone a care and protection investigation, and been placed in foster care when compared with low income preschoolers, differences in adverse behaviors were minimal. Although homeless children scored higher than housed children on the internalizing, externalizing, and total problem score on the Child Behavior Checklist (CBCL) (52.5 vs 49.9, 54.8 vs 51.2, and 54.4 vs 51.1, respectively), approximately equal numbers of children from both groups scored in the clinical range. With regard to determinants of behavior, mothers' emotional status was one of the strongest independent predictors of negative behavioral outcomes on both subscales. Foster care placement and death of a child's friend were predictors of adverse internalizing behavioral outcomes on the CBCL. After controlling for housing status, parenting practices, child's age, child's history of physical abuse, and specific life stressors predicted adverse externalizing behavioral outcomes. For both subscales, housing status and behavior were only marginally associated in the multivariate model.
Conclusions. Both homeless and low-income children experienced significant adversity in their lives, with homeless preschool children facing more stress. However, differences in behavior as measured by the CBCL were minimal. Mothers' emotional status, in addition to various stressors, strongly predict children's negative outcomes for both CBCL subscales. These findings emphasize the importance of preventive family-oriented interventions that address the needs of preschoolers and their mothers.
Almost 1 in 5 American children are growing up in poverty—and the situation is worsening. Since 1973, the US economy has been growing at a slower rate on average than at any time since the Civil War, and with it has come a decline in wages and opportunities for most citizens. These trends have led to a redistribution of wealth: by 1993, the top 20% of American households received 48.9% of the total aggregate income, whereas those in the bottom 20% shared only 3.6%.1 Reflecting the widening gap between rich and poor, the proportion of children being raised in extreme poverty—at or below 50% of the poverty line—doubled between 1975 and 1993. This subgroup comprises 22% of all poor and near-poor children.
Consistent with these trends, homelessness is far more prevalent than ever before. Over the past decade, families with children have now become the fastest growing segment of the homeless population and comprise 36.5% of the overall numbers. An estimated 100 000 children live in shelters on a given night and approximately 500 000 are homeless each year.2 The majority of these children are 6 years old or less. Many do not have the resources to ensure their healthy growth and development and suffer many negative consequences. Poverty exacerbates illness and disability, and is frequently associated with hunger, teenage pregnancy, delinquency, dropping out of school, and later difficulties associated with becoming self-supporting.3
Most researchers studying the characteristics and needs of homeless children4-14 have reported that homeless preschoolers have health problems,3,4 developmental delays,4-8and emotional and behavioral difficulties4-7,10 at greater rates than their housed counterparts. However, these findings have been inconsistent, with some reports failing to identify differences in development or behavior.10,11 These inconsistencies may be accounted for by study limitations that include differing definitions and methodologies, small sample size, lack of a comparison group, and limited questions.
This article describes 77 sheltered homeless and 90 low-income housed preschoolers and their parents who were part of a larger study—the Worcester Family Research Project, which was designed to address limitations of earlier studies. Using an unmatched case-control design, the larger project enrolled 220 homeless and 216 low-income housed female-headed families and their 627 dependent children in order to study the characteristics of the mothers and the risks for family homelessness.1,15 Information about the children were gathered by direct assessment and through maternal report, while data about the fathers were limited to maternal report. Information about mothers were collected directly across a range of domains, including housing and income histories, childhood experiences, violent victimization, social supports, and mental and physical health.1
This article, which focuses on children aged 2 years, 6 months to 5 years, 11 months, has two major goals: 1) to compare the characteristics of homeless and low-income housed children and their parents; and 2) to identify the determinants of adverse behavioral outcomes among these children, including housing status as one independent variable among many. For these goals, the data from the case control study are viewed as a cross-section of families, stratified by housing status. Using multivariate regression, we explore the relationship of maternal and paternal factors as well as various stressors, including homelessness, to child outcomes.
The conceptual framework for this study was based on an ecological model that views the child as embedded within a set of interdependent systems ranging from the immediate family, to the school and neighborhood, and ultimately to the broader culture and society.16,17 These systems vary in their impact on development depending on an individual's life stage.18Especially for young children, the family system has a profound influence on social, behavioral, and emotional development.19 Because our sample consists exclusively of single-parent mothers, we hypothesized that the dominant family influence on the child would be the mother.
An ecological model emphasizes external influences on behavior as well as the reciprocal effects of the parent-child relationship. This mutual influence is particularly pronounced for low-income families because the stressors associated with poverty can create a feedback loop: low-income single mother's are severely stressed by poverty and may feel emotionally exhausted, leading to detrimental parenting practices, which can then adversely affect the child's well-being and behavior. The child's behavior may, in turn, cycle back to further stress the mother.
Using an ecological framework, this article describes the characteristics of homeless and low-income housed preschoolers and some of the determinants of their behavior. Improved understanding of extremely poor preschoolers' needs, and more importantly, of factors that influence their behavioral status are critical to formulating responsive policies and programs.
Enrolling Homeless Families
We used an unmatched case-control design for recruiting a sample of sheltered homeless families and a comparison group of low-income housed (never homeless) families in Worcester, Massachusetts. Located in central Massachusetts, Worcester has a population of 169 000. Unlike most small to mid-sized cities, Worcester has a large Hispanic population of Puerto Rican descent and fewer blacks.20Because more than 95% of families receiving Aid to Families With Dependent Children (AFDC) and approximately 95% of homeless families in Massachusetts are headed by women, only female-headed families with children between the ages of 3 months and 17 years who were living with their mothers were enrolled.21
We recruited a total of 220 homeless families from all nine of Worcester's emergency shelters and transitional housing facilities serving homeless families, as well as from two welfare hotels (3.2% of total families) between August 1992 and July 1995. Study staff asked families who had been in shelter for at least seven days to participate in a multi-session interview. Out of the 361 families approached for enrollment, 102 refused to participate and another 39 did not complete all four interviews.
Enrolling Poor Housed Families
A comparison group of 216 families from never homeless female-headed families receiving AFDC who came to the Worcester Department of Public Welfare (DPW) office was enrolled. During the first half of the enrollment period, project staff enrolled only women coming to DPW to redetermine their eligibility for benefits. Because appearing for redetermination was required, we were able to approach potential participants in a random manner. Due to changes in DPW policy that reduced the flow of women coming in for redetermination, during the second half of the enrollment period we also recruited women who came to DPW for other reasons, which were documented. Out of 539 women approached, 141 were disqualified for previous homelessness, and an additional 178 refused to participate. A comparison of participants and nonparticipants suggested that we had successfully recruited a random cross-section of housed female AFDC recipients living in Worcester.
Interviewing Mothers and Children
Mothers from both the homeless and housed groups participated in face-to-face interviews conducted over three to four sessions and lasting a total of approximately 10 hours. For homeless women, each shelter provided a private interview space. Housed women were interviewed in their homes or at a community-based project office. Informed consent was obtained to interview all family members. As an incentive to participate, mothers received vouchers redeemable for merchandise at local stores while children received an age-appropriate toy or voucher.
Mothers completed a structured interview about themselves along with an interview lasting about 1 hour for each child in the study. We obtained information about the mother's demographic characteristics, factors in her childhood and adulthood that might influence the risk for and consequences of homelessness, service needs, parenting practices, and data about the fathers. Information about children's background, health, life events, and service use patterns was also gathered. Instruments were chosen and modified with sensitivity to those who are extremely poor and currently homeless, and to cultural issues–particularly for Spanish-speaking participants.
Children between the ages of 3 months and 17 years who were currently living with their mother were eligible for assessment. Protocols for directly assessing children were developmentally appropriate for three age groups: 1) infants and toddlers (ages 3 months to 30 months); 2) preschool children (ages 2 years, 6 months to 5 years, 11 months) who are the focus of this article; and 3) school-age children and adolescents (ages 6 through 17 years). For children in the two youngest age groups, only one child per family was randomly selected in each age group for assessment.
Criteria for instrument selection included use with homeless, high-risk, or minority populations; use in large-scale community studies in which diverse socioeconomic subgroups were sampled; ease of administration in shelters; and proven reliability and validity. All instrument questions were translated into Spanish by bilingual and bicultural translators. Spanish versions of the Child Behavior Check List (CBCL) and Structured Clinical Interview for the Diagnostic and Statistical Manual-III-Revised, (DSM-III-R), Non-Patient Version were used.
Assessment of Mothers
Housing, income, education, jobs, family structure, life events, and information about childhood, adolescence, and service utilization was gathered using an extended and modified Personal History Form,22 an instrument designed for use with homeless and low-income persons. The Personal Assessment of Social Support23 was used to elicit mothers' supports and resource base. The Personal Assessment of Social Support was developed and validated with families of developmentally at-risk preschoolers, particularly families living in poverty. Mothers were asked to name up to seven persons who played a role in their lives (either positive or negative) and then asked questions regarding the quality of these seven relationships.
Information about the mothers' experiences of physical and sexual victimization across the lifespan was obtained using the Conflict Tactics Scale.24 Women were also asked about physical and sexual assaults by nonintimates.
Lifetime prevalence of Axis-I DSM-III-R mental and substance use disorders was assessed using the Structured Clinical Interview for DSM-III-R, Non-Patient Version.25 The following diagnoses were assessed: mood disorders, anxiety disorders (including posttraumatic stress disorder), somatoform disorders, eating disorders, and alcohol and other drug abuse or dependence. In addition, each mother's current symptomatology and distress was assessed using the Symptom Checklist 90 (SCL-90).26 This self-reported instrument provides a current profile across nine symptom dimensions: somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. The Global Severity Index (GSI) is a summary score that combines a symptom count with a measure of distress to produce an overall rating. Physical and sexual abuse across the lifespan was assessed using the Conflict Tactics Scale.27
The Parenting Practices Scale28 measures the nature and quality of parents' interactions with their preschool children. This self-report instrument, which yields a summary score, consists of 34 items that evaluate parental behaviors and cognitions. Scores range from 0 to 190, with higher scores representing more favorable parenting practices. The scale was developed with low-income parents, has good reliability and validity, and was significantly associated with measures of parent's psychological well-being and children's behavior.
Information on each child's father was gathered through a structured interview administered to the mother. Questions included information about the father's age, education, employment status, payment of child support, and problems. Positive responses to eight problems (alcohol abuse, drug abuse, legal problems, mental illness, physical illness, unemployment, family problems, violence toward women) were summed to give a total problem count.
Assessment of Child Behavior Problems and Life Events
To evaluate emotional and behavioral problems, the age-appropriate version of the CBCL was administered to the mothers. The CBCL is comprised of 118 symptoms rated by the mothers on a 3-point scale. Scores are derived for total problems and for the composite internalizing and externalizing global scores. For children ages 4 to 6, the internalizing global score is comprised of the anxiety/depression, withdrawn behavior, and somatic complaints syndrome scales. The externalizing score is derived from items belonging to the delinquent behavior and aggressive behavior syndrome scales. Social, thought, attention, and sex problems were neither internalizing or externalizing. For children ages 2 to 3, the internalizing scale consists of withdrawn and anxiety/depression syndrome scales, and the externalizing consists of destructive and aggressive syndrome scales. Sleep problems and somatic problems were neither internalizing or externalizing.
Internalizing and externalizing global scores can be calculated for children in this age range and are comparable to such scores among older children. Raw scores on the syndrome and global scales can be converted into T-scores that are based on normative data. The CBCL has been shown to have high reliability and criterion-related validity.29,30
Stressors experienced by each child were assessed with Masten's Life Event Questionnaire filled out by the mother (Masten AS, Neeman J, Andenas S. Life events and adjustment in adolescents: the significance of event independence, desirability, and chronicity. J Res Adolescence. 1994;4:71–97). This instrument consists of 39 items describing a chronic or discrete event (eg, “Your child witnessed something violent happen at school or in the neighborhood,” or “During the past year, at least one parent became seriously ill or was injured”). Mothers respond yes or no to each event. Positive responses are summed to give a cumulative stress score. In addition, events were examined separately to identify those that contributed significantly to poor behavioral outcomes.
Comparisons of homeless and housed families were used initially to describe this sample. Preschoolers of the homeless and housed sample were compared using t tests and Wilcoxon rank-sum tests for continuous variables and χ2 tests for discrete variables. Similar comparisons were made between the homeless and housed mothers and fathers of the preschoolers.
Multiple linear regression was used to examine the association of behavior problems in preschool children with hypothesized predictors suggested by the ecological model, including homelessness. Two regression models were developed, one each for the internalizing and externalizing global scores of the CBCL. Pearson product-moment correlations between each scale and the characteristics of the children and their mothers and fathers were computed to nominate predictors for the regression analysis. The multivariate model developed for each scale included important background factors. Covariates for age, gender, and ethnicity of the preschoolers were included in all models under consideration. Housing status was also included in all models to reflect the study design (children were sampled according to this covariate). To maximize power, the two regression models were developed using the SCL-90 GSI. Predictors that did not remain significant when controlling for background covariates and other predictors were eliminated. Because each family in the sample had only one preschooler who was assessed, there was no need to adjust the standard errors orP values from the regressions for within-family correlation. The final regression models were estimated separately for girls and boys to assess the interactive effects of gender with other predictors. In addition, the final regression models were refitted, using each of the subscales of the SCL-90 in place of the GSI to assess how results varied across symptom dimensions.
Regression was also used to investigate the effects of shelter environment on behavior problems in homeless preschoolers. The number of weeks at a shelter before interview and indicator variables for the Worcester shelters were used to summarize the potential variation in the preschoolers' experiences of homelessness. These predictors were added to the final model developed for each scale of the CBCL using homeless and housed children, but omitting housing status.
Homeless vs Housed Comparison
Characteristics of the Mothers Who Had Preschool-aged Children (Table 1)
The entire sample of 77 homeless and 90 low-income housed mothers with preschool-age children were assessed. Their average age was approximately 27 years and they had 2.7 children. More than one-third of both groups are Puerto Rican and approximately 57% had never been married. The homeless had resided in Worcester for almost 6 years less than the housed (P < .0005). Both groups were living below the federally established poverty level, but the homeless earned approximately $2452 less than the housed. Both groups had multiple sources of income that they used to supplement their AFDC grant. Homeless mothers tended to be less educated and had more limited work histories. During the prior 2 years the homeless had moved 3.4 times compared with 2.0 times among the housed. Compared with the overall sample of mothers in the study (N = 436), the mothers of preschoolers (N = 167) tended to be slightly older, were more often black, had more children, were less poor, and had been on AFDC for longer periods.
Both groups had fragmented support networks, with the homeless significantly more socially isolated than the housed. An astounding 84.9% of the homeless and 85.2% of the housed had been victims of violence at some point in their lives. Forty-five percent had been sexually abused as children and 65.2% had been severely physically assaulted by intimate partners in adulthood. Consistent with these high rates of violent victimization, posttraumatic stress disorder, major depressive disorder, and substance abuse were disproportionately represented compared with a general female population.31Scores on the SCL-90 confirm the high levels of distress experienced by these mothers. Significantly more homeless mothers than housed had talked to doctors about their emotional problems, and had been hospitalized for mental illness and/or substance abuse. Approximately one-third of both groups had made suicide attempts. Physical health was also compromised, especially considering the groups' young age. With regard to parenting practices, both groups scored in a range comparable with a normative sample of low income parents (138 for normative sample parents).15
Characteristics of Fathers With Preschool-aged Children (Table 1)
The average age of the child's father was almost 30 years. Only 5% were currently married to the mothers of their children. Approximately 10% were living with their child. In contrast, nearly one-third of both homeless and housed had no relationship with the child's mother. The fathers had a mean of approximately 11 years of education. About 46% were now working, but only a minority were paying child support. Surprisingly, some of the fathers had considerable contact with their children. However, 25.3% of the fathers with homeless children and 30.3% of those with housed children never saw them.
As reported by the mothers, the fathers who visited 2 to 3 times a month had an average of 1.5 problems (out of a possible 8). Approximately 40% of both homeless and housed mothers reported that their child's father was violent toward women. Furthermore, 45.4% of the homeless fathers and 46.7% of the housed had drug/alcohol problems. In addition, 37.1% of the homeless and 27.4% of the housed fathers were currently in jail or on probation.
Approximately three-fifths of the preschoolers were boys and two-fifths were girls. Their average age was 4.0 years. Reflecting the larger sample, about one-third were Puerto Rican, with more whites among the housed and more blacks among the homeless. In the previous year, the homeless children had moved three times more frequently than the housed. In this period, 25.7% of the homeless had moved four or more times compared with only 3.4% among the housed (P < .001). However, since birth, 75.0% of the homeless had moved three or more times compared with 36.0% among the housed (P < .001).
According to the Masten Life Event Questionnaire, homeless children had experienced more stressful life events in the past year than their housed counterparts (7.4 events vs 5.8 events, P = .007). Not surprisingly, homeless children were significantly more likely to move, be evicted, be placed in another home, or have a sibling placed in another home. Housed children, on the other hand, were more likely to experience increased parental absence due to work.
Overall, almost twice as many homeless compared with the housed (40.3% vs 21.1%, P < .007) had undergone a care and protection investigation (known as a 51A in Massachusetts); currently, many more homeless had open 51As (20.8 vs 2.2, P < .001). Children with 51As were more residentially unstable, tended to be in foster care or other placements four times as often, were more commonly physically and sexually abused, had fathers with substance abuse problems, and had multiple serious life stressors compared with those children without 51As.
Despite the high rates of violent victimization among the mothers, a small but similar percentage among both homeless and housed had reported that their children had been sexually abused (5.3% vs 4.6), physically abused (10.5% vs 6.8%), or had witnessed violence in school or in the neighborhood (3.9% vs 4.5%).
With regard to physical health of the children, the mothers reported that only 10.5% among the homeless and 5.5% among the housed had fair or poor health ratings. Twice as many homeless children compared with housed had been medically hospitalized in the past year. Almost all the children had a regular source of medical care.
On the CBCL (see Table 3), the homeless children scored significantly higher than housed children on the externalizing global scores. Although homeless children had higher scores on both the internalizing global score and the total problem score. These differences were not statistically significant. In all cases, the differences were not clinically important. The proportion of homeless and housed children in the clinical range for all three global scores (T = 64 or greater) was approximately equal.
Determinants of Behavior
Impact of Mothers and Fathers Characteristics on the Children: CBCL
Table 4 shows that after controlling for housing status, the child's age and ethnicity, mother's scores on the SCL-90 GSI and the Parent Practices Scale, child's history of physical abuse, and three life stressors were significant predictors of preschoolers' scores on the externalizing scale of the CBCL. Although all subscales of the SCL-90 were individually predictive, no single subscale, with the exception of the psychotic symptom cluster, was more strongly associated with poor outcomes than any other subscale. A 1 standard deviation (SD) change in SCL-90 GSI score (SD = .61) was associated with a 3.12 change in the preschoolers' externalizing scores, compared with a −3.83 change associated with a 1 SD change in the Parent Practices Scale score (SD = 15). After controlling for other significant predictors, the presence of physical abuse was associated with a 6-point increase in CBCL externalizing score; the 95% confidence interval of (1.15, 10.85) reflected the uncertainty due to the small number of preschoolers experiencing physical abuse. More dramatically, the death of a childhood friend, experienced only by 2 preschoolers, was associated with an 11-point increase in externalizing score. This predictor was retained in the model despite a Pvalue less than the traditional .05 cutpoint to account for the unique experience of these 2 children.
Mother's score on the SCL-90 GSI was also a significant predictor for preschoolers' scores on the internalizing scale of the CBCL (see Table5). A history of foster care placement was also a significant predictor after controlling for child's age, gender and ethnicity, death of a childhood friend, and housing status. The effects of the these two predictors are comparable (a 1 SD change in SCL-90 is associated with a 5.22 change in externalizing score), although both are less than one half of the change associated with the death of a childhood friend. Regressions on internalizing scores for girls and boys showed that housing status was only a significant predictor of boys' internalizing scores; the change in boys' internalizing scores associated with homelessness was estimated to be 4.7 points (P = .018) compared with −.88 for girls (P = .74). Although statistical power is limited for the interaction between housing status and gender (ie, the difference of the effect of homelessness for boys and girls), it is included in the final model to properly account for these gender differences. Regressions within homeless preschoolers showed that neither duration of homelessness nor the specific shelter location were significant predictors of either CBCL scales.
To our knowledge, this is the first epidemiologic study to date that describes the needs of sheltered homeless and low-income housed preschool age children, and relates parental and environmental variables to their well-being by developing multivariate predictive model. Other researchers investigating these issues in low-income school-aged children32 have concluded that adverse behavioral outcomes were more common in homeless children compared with housed, and were associated with “greater stress exposure and fewer resources.”33 Using data from the Worcester Family Project, this study explored similar questions in sheltered homeless and low-income housed children between the ages of 2½ years and 6 years (mean 4.2 years).
Our study documented that homeless preschoolers were more likely to have experienced stressful life events, undergone a care and protection investigation, and been placed in foster care when compared with their housed counterparts. On the CBCL, homeless children had slightly higher total problem and global internalizing scores and significantly higher global externalizing scores. However, a similar proportion of both groups were in the clinical range. It should be noted that researchers have reported that children from lower socioeconomic groups score higher than average on the CBCL and attributed this discrepancy to the risks associated with poverty and to being raised by single mothers who are more likely to be depressed and anxious.34
Previously, authors have speculated about the relative contribution of various risk factors to adverse behavioral outcomes in low-income preschoolers, but did not conduct multivariate analyses to determine the impact of maternal and paternal functioning and the effect of environmental factors (eg, housing status)35 on these children. Using the CBCL as the dependent variable, we found that mothers' emotional status, as measured on the GSI of the SCL-90, was a powerful independent predictor of adverse outcomes on the CBCL for both homeless and housed preschoolers, explaining 21.6% and 26.1% of the variation in the externalizing and internalizing scales, respectively (after controlling for other predictors). However, mothers who are distressed often report more symptoms in their children and we were unable to distinguish this measurement/reporting effect from the clinical effect. For this sample, although all subscales on the SCL-90 were highly predictive of negative outcomes on the CBCL, when included in the multivariate models, only the psychotic symptom cluster was associated with larger increases relative to other subscales. This lack of association with any specific symptom dimension suggests that reporting bias is probably small.
Because adverse behavioral outcomes in preschoolers are highly associated with the emotional status of their mothers, it is important to understand the reality of these families' lives. The homeless and housed preschoolers in our sample were being raised in extreme poverty by young single mothers who were intensely stressed, socially isolated, and commonly victims of severe abuse and assault. The homeless mothers tended to be living in slightly more adverse circumstances than their housed counterparts: they moved more frequently, were less educated, worked infrequently, and were more often victims of violence. As adults, they were more often hospitalized for emotional problems and/or substance abuse.
Some of the fathers of both homeless and housed preschoolers saw them frequently, but only a small number paid child support and many were abusive and/or had substance abuse problems. Despite some support from the fathers, the majority of low-income mothers in our study were faced with the daunting task of raising preschoolers alone with few resources, and very limited social and economic margin.
In addition to the influence of parenting figures on behavior in preschoolers, various stressors were also independent predictors of poor outcomes. For example, on the internalizing scale, we found that three factors predicted worse outcomes. First, the child's placement in foster care was an independent predictor. We can speculate that separations from parents may not only lead to grief, terror, and feelings of abandonment but can also compromise the child's capacity to form secure attachments36,37—leading to more total problems on the CBCL, and depression, anxiety, and withdrawal. Researchers have also reported that children in foster care have high rates of developmental delays and difficulty functioning.38
Second, the traumatic impact of the death of a friend was highly predictive of worse internalizing and externalizing scores.
Third, although housing status, duration of homelessness, and gender alone were not independent predictors of adverse internalizing outcomes, we developed an interaction term between gender and housing status: homeless boys had significantly worse scores in contrast to girls. It is possible that this gender effect may be due to a differential relationship between mothers and their sons as compared with mothers and their daughters.
With regard to the externalizing scale, in addition to the GSI of the SCL-90 in the mothers, various demographic factors, parenting practices, and life events independently predicted worse outcomes. Older children were more likely to manifest aggressive, acting-out behaviors. Puerto Rican preschoolers were less likely than children from all other ethnic groups to have high externalizing scores, a factor perhaps related to cultural values and childbearing practices.39
We also documented that more favorable parenting practices were associated with lower externalizing scores. The literature has discussed how unfavorable practices, such as physical punishment and inconsistent discipline, may contribute to aggressive and antisocial behaviors in children.40,41 It is easy to understand how parenting quality can be compromised for some mothers who have endured traumatic childhoods and, in adulthood, struggle with social isolation, emotional distress, and extreme poverty.
In addition, children who were physically abused were more likely to have higher externalizing scores. Previous studies support this finding and have shown that child maltreatment is associated with diminished social competence and poorer social problem-solving skills.42 Finally, various stressors, such as death of a childhood friend and parental conflict were also predictive of more externalizing behaviors. Surprisingly, family financial problems were associated with better scores; however, additional data analyses indicated that these monetary difficulties were generally associated with break-up of a relationship with a male partner, many of whom had significant problems.
Neither homelessness nor residential instability (as measured by number of moves) were strong predictors of adverse outcomes. However, as discussed in the data analysis section, we were constrained by the study design because we could not look at measures of residential instability, such as number of moves, independent of housing status in building the model. We found that housing status was only marginally significant for externalizing behavior and only significant for boys on the internalizing scale. The lack of association between housing status and adverse preschool behavioral outcomes highlights the primacy of the mother-child relationship for younger children. Furthermore, research describing the impact on school-aged children has shown that the effect of housing status increases with age (Masten et al33 and J. Buckner et al, unpublished manuscript).
Various other limitations should be kept in mind when considering the findings. The study sample largely reflects the findings from small and mid-size cities that have a disproportionate representation of Puerto Rican children. Because the study sample was small and the fathers were not interviewed directly, these data may have omitted significant information that may have affected the children. Most important, this study was cross-sectional and longitudinal research is necessary to more definitively determine the relative contributions of parental and environmental factors to child outcomes. Furthermore, the CBCL does not have normative data on specific minority groups, especially Hispanics of Puerto Rican descent.
Our findings highlight the importance of developing preventive family-oriented interventions that consider both the needs of the mothers and their preschool-age children. Effective programs must tailor specific strategies that help mothers climb out of poverty, build self-esteem and supportive relationships, prevent violence and cope with its sequelae, and interact positively with their children. In addition, we must expand our notion of intervention to not only include the treatment of problems, but also to develop preventive programs that promote child competence and self-esteem.43 Furthermore, approaches that decrease children's exposure to risk factors and enhance protective experiences will reduce adverse outcomes. For example, programs that prevent violence as well as preserve family integrity by supporting the real challenges faced by single low-income mothers will also greatly benefit children.
The current federal climate poses great threats to the well-being of our nations' youngest citizens. The proposed cuts in anti-poverty programs that provide the most basic survival assistance for vulnerable children and their families, such as AFDC and food stamps, jeopardize the safety net that assists children in need. State and federal policies that aim to provide affordable housing and income security, prevent violence and cope with its sequelae, and eliminate child poverty, not dismantle child and family supports, must accompany program intervention efforts. Only in this way, will all young children have opportunities for healthy and productive lives.
- Received July 18, 1996.
- Accepted November 7, 1996.
Reprint requests to (E.L.B.) Better Homes Fund, 181 Wells Ave, Newton, MA 02159.
- AFDC =
- Aid to Families With Dependent Children •
- DPW =
- Department of Public Welfare •
- CBCL =
- Child Behavior Check List •
- SCL-90 =
- Symptom Checklist •
- GSI =
- Global Severity Index •
- SD =
- standard deviation
- ↵Institute of Medicine, National Academy of Science. Homelessness, Health, and Human Needs. Washington, DC: National Academy Press; 1988
- ↵Hamburg DA. Today's Children. Creating a Future for a Generation in Crisis. New York, NY: Random House; 1992:48
- Rafferty Y,
- Shinn M
- Molnar J,
- Rath W
- Miller D,
- Lin E
- Wood D,
- Valdez RB,
- Hayaski T,
- Shen A
- Bassuk E,
- Rosenberg L
- Whitman B,
- Accardo P,
- Bryert M,
- Kendagor R
- ↵Molnar J, Rath WR, Klein TP, Lowe C, Hartmann A. Ill fares the land: the consequences of homelessness and chronic poverty for children and families in New York City. New York, NY: Bank Street College of Education; 1991
- ↵Brofenbrenner U. The Ecology of Human Development. Cambridge, MA: Harvard University Press; 1979
- ↵Kellam S, Branch J, Agrawal K, Ensminger M. Mental Health and Going to School: The Woodlawn Program of Assessment, Early Intervention, and Evaluation. Chicago, IL: University of Chicago Press; 1975
- ↵US Bureau of the Census. 1990 Census of Population: Metropolitan Areas. Washington, DC: US Government Printing Office; 1990
- ↵Barrow S, Hellman F, Lovell A, et al. Personal History Form, (PHF, 7/85). New York, NY: New York State Psychological Institute; 1985
- ↵Dunst CJ, Trivette CM. Personal Assessment of Social Support Scale. Family, Infant and Preschool Program. Morgantown, NC: Western Carolina Center; 1988
- Straus MA
- ↵Spitzer RL, Williams JBW, Gibben M, First MB. Structured Clinical Interview for DSM-III-R Non-Patient Edition (SCID-NP, Version 1.0). Washington, DC: American Psychiatric Press; 1990
- ↵Derogatis LR. Symptom Checklist-90-R: Administration, Scoring, and Procedures Manual. Minneapolis, MN: National Computer Systems, Inc; 1994
- Strause MA
- ↵Achenbach TM. Manual for the Child Behavior Checklist/4–18 and 1991 Profile. Burlington, VT: University of Vermont Department of Psychiatry; 1991
- ↵Achenbach TM. Manual for the Child Behavior Checklist/2–3 and 1992 Profile. Burlington, VT: University of Vermont Department of Psychiatry; 1992
- Raadal M,
- Odont, Milgrom P, Cauce AM, Mancl L
- ↵Blacher J. When Their's No Place Like Home. Options for Children Living Apart from Their Natural Families. Baltimore, MD: Paul H. Brookes Publishing Co; 1994
- ↵Costin LB. Child Welfare: Policies and Practices. 2nd ed. New York, NY: McGraw-Hill; 1979
- ↵Wasserman GA, Brunelli SA, Rauh VA, Alvarado LE. The cultural context of adolescent child-rearing in three groups of urban minority mothers. In: Lamberty G, Cole Garcia L, eds. Puerto Rican Women and Children. Issues in Health, Growth, and Development. New York, NY: Plenum Press; 1994:137–160
- Levendonsky AA,
- Okun A,
- Parker JG
- Copyright © 1997 American Academy of Pediatrics