Lee Shumow Northern Illinois University

Deborah Lowe Vandell University of Wisconsin -- Madison

Kyungseok Kang Inha University

Urban low-income 5th-graders participated in a school-choice study. Families utilizing choice schools (N = 73) were more likely to be African American, lower-income, and high-risk neighborhood residents than families whose children attended assigned schools (N = 100). Firm-responsive parenting, family togetherness, and family supportiveness also were linked positively to utilization of choice. Parent involvement in children's schooling was higher in neighborhood schools. School choice positively predicted children's mathematics achievement and school orientation. Parents who chose schools rated the teachers as practicing more parent-involvement strategies than parents of assigned students, but teachers reported equal practices. Parent ratings of school quality did not differ between conditions, nor did teachers or parents report better relationships in either condition.

School choice refers to an educational policy that provides a broad spectrum of options whereby parents can select the schools their children attend. These options include magnet and open-enrollment public schools as well as private schools. Magnet public schools serve students from diverse geographical areas, depend on voluntary enrollment, and have some special offerings designed to attract parents and students ( Blank, 1984). Open enrollment involves children crossing attendance boundaries to attend public schools in neighborhoods other than the one in which they reside ( Maddaus, 1988). Private schools, funded by tuition or tuition vouchers, are yet another form of school choice. Because low-income parents are less able than middleincome parents to make residential choices based on the school in the neighborhood or to pay for private-school tuition costs ( Darling-Hammond & Kirby, 1985; Williams, Hancher, & Hutner, 1983), open-enrollment plans have been developed to allow parents to place their children in a particular school even when they are precluded by economic circumstances from living in the school's neighborhood attendance boundaries. 

Lee Shumow, Department of Educational Psychology, Counseling, and Special Education (EPCSE), Northern Illinois University; Deborah Lowe Vandell, Department of Educational Psychology, University of Wisconsin -- Madison; Kyungseok Kang, Department of Education, Inha University, Korea.

A version of this article was presented at the 1995 annual meeting of the American Educational Research Association, San Francisco. This research was supported by a grant from the Spencer Foundation.

We greatly appreciate the helpful critiques provided by Jill Posner and the assistance during data collection provided by Rob Rosenthal and Sarah Richardson.

Correspondence concerning this article should be addressed to Lee Shumow, Department of EPCSE, Northern Illinois University, De Kalb, Illinois60115. Electronic mail may be sent via Internet to This e-mail address is being protected from spam bots, you need JavaScript enabled to view it

 

A critical question being raised by researchers and policymakers is the impact of school choice on children's academic achievement ( Camegie Foundation for the Advancement of Teaching , 1992; Clune, 1990; Coleman, 1992). There are several reasons why school choice might be related to academic achievement. First, families are empowered to avoid schools that they perceive to be of low-quality ( Wells, 1990). Second, parents are able to select a school with resources that match individual children's needs. Third, parent involvement might be expected to increase in schools the parent has selected ( Wells, 1990). Juxtaposed to the potential benefits of school choice, attending a neighborhood-assigned school may contribute to children having better academic performance if the proximity to the neighborhood school fosters parental involvement with the school.

Much of the research investigating school choice has focused on magnet schools originally designed to encourage voluntary desegregation. Inconsistent results regarding student achievement are reported. For example, in an evaluation completed in the Milwaukee public schools (MPSs), Mitchell ( 1989) found that magnet school children performed better academically than their counterparts in assigned schools. In an analysis of African American youth attending magnet schools in a large New York school district, however, Singletary ( 1993) found little evidence of such differences.

One possible explanation for these discrepant findings is differential selection of the magnet school experience by families. Some ( Lowe & Miner, 1992; Wells, 1990) have argued that it is the relatively more educated and affluent parents who take advantage of choice plans. Consistent with this argument is a British study by Willms and Echols ( 1992; cited in Cookson, 1994) that found "creaming" to

occur as children from more advantaged homes utilized choice plans. However, two other studies ( Coleman, Schiller, & Schneider, 1991; Witte, 1991) reported that low-income minority families were more likely to avail themselves of choice opportunities than were higher economic status and Caucasian families, although among the low-income group the least educated parents were less likely to participate in choice plans. Both sets of findings illustrate the importance of assessing the extent to which family characteristics are linked to utilization of school choice because these selection factors and not attending the choice school per se, may account for differences in children's school achievement.

Only limited research is available examining relationships between open enrollment and children's school achievement. Contrary to finding hypothesized associations with children's academic achievement, these studies have failed to find differences in children's performance. For example, one early experimental program ( Capell, 1978) involving 14 schools found no differences in reading achievement over a 3-year period as a result of parents choosing schools outside of the neighborhood, after controlling for pretest reading scores, race, and lunch subsidization status. A more recent study found that reading levels of elementary school children participating in an open-enrollment plan did not differ from those of children attending either magnet or integrated-neighborhood schools, after controlling for subsidized lunch status ( Easton, Bennett, & Seymore, 1987). It is possible, however, that differences would have been reported if more extensive child outcomes were considered.

In the current study, we asked if demographic characteristics such as income, educational level, race, and neighborhood risk are related to utilization of a school-choice option. In addition, school choice is examined in relation to family psychological characteristics of parental involvement in schooling, parental socialization practices, and family togetherness and supportiveness to determine if these family processes and experiences are associated with school choice.

There is an extensive research literature suggesting that parent involvement in academic activities is an important contributor to children's school achievement ( Bempechat, 1990). Proponents of school choice have assumed that parents who choose schools are more involved in all aspects of their children's education ( Cookson, 1994). In support of this assumption, Mitchell ( 1989) noted that children attending magnet schools came from homes with more parental involvement in education than children who attended neighborhood schools. Similarly, low-income parents participating in a private-school voucher program in Milwaukee had higher parental involvement at home and at school than did the control group of parents ( Witte, 1991). However, in at least one study ( Bauch, 1988), parent involvement measured by school visits was comparable in low-income urban private and public schools. In the current study, we hypothesized that parental involvement is greater when school choice is in effect. Furthermore, this parental involvement in their children's school is expected to contribute positively to children's academic achievement and orientation.

A second psychological characteristic investigated in the current study is parental socialization practices, which have been found to be related to children's school adjustment. For example, firm-but-responsive parenting is associated with elementary-school children having fewer behavior problems, better social adjustment, and more academic success ( Baumrind, 1989; Clark, 1983; Grolnick & Ryan, 1989), whereas harsh parenting is linked to children having poorer academic grades and more conduct problems ( Dodge, Pettit, & Bates, 1994; Wentzel, Feldman, & Weinberger , 1991). Firm-but-responsive parents establish reasonable rules, set attainable standards, explain reasons for their behavior, and are responsive to their children's needs. We hypothesized that firm-but-responsive child rearing is related to utilization of school choice because choice might represent a concrete way by which low-income parents can meet their children's needs.

A third factor that may be associated with school choice and children's development is family time spent together doing activities. In one study of low-income inner city adolescents, this type of family time was found to promote academic and behavioral development ( Luthar & Zigler, 1991). In the current study, we test the hypothesis that parents who spend more time with their children are likely to reflect that family orientation by actively choosing schools for their children to attend.

In addition to being related to family factors, the utilization of school choice plans may be influenced by children's competence. A negative selection factor would occur if parents move a child from an assigned school because the child is having academic or social problems. A positive selection influence would occur if parents believe that their academically achieving child needs the stimulation associated with attending what the parents view as an academically superior school. Witte, Bailey, and Thom ( 1992) found that children who subsequently were enrolled in an experimental voucher program had lower academic achievement and greater school conduct problems than a comparison group who remained in their assigned public schools. In order to examine the issue in the current study of fifth graders, their standardized test scores, report card marks, and conduct ratings from third grade are investigated to determine if children moving to choice schools differed significantly from children who remained in assigned schools.

A final issue to be examined is the possibility that school choice promotes better relationships between home and school. Contact and cooperation between homes and schools have been highlighted as important for low-income and working-class families ( Espinoza, 1988). According to one analysis ( Slaughter-Defoe, Nakagawa, Takanishi, & Johnson, 1990), "mutual respect and cooperation between families and schools appears to have much to do with situations in which African-American children are successfully 'making it' academically and socially in school" (p. 373). In a large-scale study in Maryland ( Epstein, 1986), the majority of parents reported that teachers rarely or never attempted to involve them in their children's education. Practices that teachers can use to involve parents include

asking them to help with homework, providing parents with information, inviting parents to school, and requesting information and help from parents ( Becker & Epstein, 1982; Kanter, Ginsburg, & Milne, 1986). Although individual teachers vary in their use of these practices, we examine whether there are more overall teacher attempts to involve parents in choice schools. We expect that active selection of a school choice option is associated with better relations between home and school.

In summary, this study has three goals. The first goal is to compare the demographic and psychological characteristics of families who utilize different school arrangements. We investigate, within a low-income sample, whether children attending choice schools come from families with more resources such as income, education, neighborhood quality, parental involvement, family support, and family activities. We also examine whether firm-but-responsive parenting strategies are associated with school choice. In addition, potential child selection factors that predated school choice are determined. Second, we compare children's academic achievement and orientation to school (defined as children's academic self-competence, school work habits, and conduct grades) between choice and assigned schools, controlling for selection factors. Based on prior findings, we expect that children's reading and math achievement along with children's school orientation are positively associated with school choice, even after family demographic and psychological factors are controlled. Finally, we investigate differences in home-school relations between choice and assigned schools.

All parents of third graders (N = 529) in nine elementary schools within the MPS system were sent letters describing a 3-year study. The study was framed in an ecological perspective and parents were told that we were interested in different facets of their children's daily experiences and how these experiences influenced school achievement and social development. Because of an interest in children's after-school experiences, four schools housed or were located near community after-school programs. The nine schools also were selected because they had some of the highest proportions of children who qualified for lunch subsidies within the school district and served primarily two ethnic groups (African American and Caucasian children).

Sixty-four percent of the contacted families returned forms, and 58% (N = 307) stated their interest in the study. Virtually all (99.8%) of the families returning the forms reported household incomes below the median income of families in Wisconsin. The demographic characteristics of the families who were willing to participate did not differ from those who chose not to participate. In addition, demographic characteristics of the responders did not differ from the overall demographic profile of their particular schools, suggesting that families who returned the forms did not differ from those who did not return the forms. From the pool of willing families, sufficient resources were available to study 216 children (mean age = 9.1 years, SD = 0.5). There were no differences between those children who were selected as participants and those who were not selected in terms of gender, race,

lunch subsidy, or school transfer during the year. All children who were in self-care (n = 15), formal programs (n = 34), or informal adult-supervised care (n = 45) after school were included in the final sample. A subsample of children (n = 121) who were in mother-care after school was selected from a pool of available mother-care children in a way that ensured similar distributions by gender and race. About half of the 216 children selected were African American (n = 103), the remainder was Caucasian; almost half of the children were boys (n = 100). Fifty-five percent of the children lived in single-parent households. Nearly 60% of the sample qualified for subsidized school lunches.

There was considerable geographic mobility during the course of the study; 194 children continued with the study through fifth grade. Participant attrition was not related to child gender, race, family structure, or family income. In fifth grade, children attended 43 different schools (28 MPSs, 5 out-of-district public schools, 7 parochial schools, 3 special education schools).

School choice became a salient policy issue affecting the ecology of the families and children during the course of the study. Considerable attention was focused on this issue by community activists, mass media, and the MPS district during this time, and we were aware that some study parents were actively investigating school-choice options. As a result, when the children were in fifth grade, we examined factors associated with school choice for those children who continued attending regular public school programs in the Milwaukee area. Children who were attending parochial or special education schools were excluded from the school-choice analyses in order to control for variability that might be attributed to religious or organic issues.

The resulting sample for the school choice analyses consisted of 173 children. By parent report, 84 children were African American and 89 were Caucasian. The sample included 94 girls and 79 boys. Average maternal educational level was high school graduation. Mean per capita income (in 1993 dollars) was $6,596 (SD = $4,108). Table 1 presents a detailed profile of the fifth-grade sample including a neighborhood risk score (defined below) for children's residence.

Of the 100 children classified as attending assigned schools, 92 attended their neighborhood schools while 8 were assigned by the district to a school outside their neighborhood. Twenty-three parents of assigned school children reported selecting their housing based on the school.

The Milwaukee schools have been restructured to maximize parental choice, thus the choice category in this study was broad and represented a diversity of schools. Parents had the option of applying to magnet schools, to approximately 20 neighborhood MPSs outside their attendance area but within a designated region (limited open enrollment), to any MPS (via appeal), to a suburban exchange program, or to a private school "voucher" experiment. Of the 73 children classified as attending choice schools, some (n = 33) attended magnet schools. These magnet schools offered a range of programs designed to develop multiple intelligences, creative arts, and academics, as well as a Waldorf school and an open classroom design. Other magnet schools that offer individually guided education, environmental, Montessori, language immersion, or ungraded programs were not selected by our families. Other children (n = 32) classified as participating in choice schools used the limited open-enrollment plan in which children attend traditional neighborhood schools outside of their own neighborhoods. Other parents (n = 8) appealed to have their children attend a particular school outside of their neighborhood. Reasons for appeals included family moves after which the parent wished to keep the child in the previous neighborhood school, proximity to parent employment, and extended family in the area of the school. None of the parents in our study chose to participate in the suburban exchange or in the private-school voucher trial operating in Milwaukee. As a result of open enrollment and appeal, nine schools in this study included both choice and neighborhood assigned children.

Table 2 displays indicators of educational effectiveness for schools classified as magnet, assigned, and open enrollment. Open-enrollment schools are usually attended by both assigned and choice children since they were neighborhood public schools chosen by at least some parents. The percentage of children scoring above national norms (e.g., 50th percentile) on standardized reading subtests, teacher characteristics such as percentages of teachers with more than 6 years experience and with advanced degrees, and descriptive school statistics such as expenditures per student, percentage of students receiving free lunch, percentage of mobility, and average percentage of child attendance are included as indicators. The median of these indicators is displayed in the table by school type. Differences between schools on these indicators were tested with the Kruskal-Wallis one-way analysis of variance by ranks. No differences were found.

The adult who reported primary parental responsibilities was asked to complete a phone interview and a set of questionnaires that were mailed back to the project office. In almost all cases, this person was the child's mother, although a few grandmothers, aunts, fathers, and foster mothers provided information. In-person interviews at the family's residence were completed if the family could not be reached by telephone. Teachers and children also provided information that will be described below.

School choice. We operationalized school choice as a policy that allows parents to participate in selecting the school their children will attend ( Cookson, 1994). Parents were asked to indicate why their children attended the particular school in which they were currently enrolled. These options included assignment by the district to the neighborhood school, assignment by the district to a school other than neighborhood school, application by parent (all choice schools require a completed application), appeal by parent, or choice of private school by parent. In the current study, the first two cases were grouped together into an assigned school category. Application or appeal was designated as school choice. This grouping strategy was justified empirically as well as conceptually. No significant differences were found for the three forms of school choice in terms of variables investigated in the current study (family demographics, family psychological characteristics, child school achievement).

Parents were asked if they selected their housing so that their child could attend a particular neighborhood school. Families who reported choosing their housing based on the neighborhood school (n = 23) did not differ from those assigned families who did not choose their housing for the school (n = 77) in terms of their demographic or psychological characteristics, nor did their children differ on achievement. Consequently, we aggregated these families into an assigned school group.

As a reliability check, an administrator at each school, who had access to the child's records, was interviewed about the children's enrollment status. In five cases of conflicting reports between

parent and school official, the situation was investigated and resolved by the research staff.

Parent reports. In addition to information regarding school choice, parents provided information about where they lived, their race, family structure (1-parent vs. 2-parent households), education level, and family income.

Parents completed questionnaires describing their socialization practices. The Raising Children Checklist is a 30-item self-report inventory of socialization practices (Posner, Shumow & Vandell, 1993). One of the three subscales (firm-responsiveness) was relevant for the current study. These 9 items were rated by parents using 4-point response scales, ranging from 1 = definitely false to 4 = definitely true. The subscale has a mean of 3.35 and standard deviation of 0.31 (Cronbach's α = .76). Sample items included "Do you say something positive to your child when he/she does something you like?" "Do you make rules which take your child's needs into consideration?" "Do you explain the reasons for the rules you make?" Do you try to show understanding for your child's feelings when you punish him/her for misbehaving?"

The parents completed a questionnaire about home-school relationships that was developed for this study. One subscale (8 items) asked parents to use 5-point ratings (1 = never; 2 = once or twice; 3 = three to four times; 4 = pretty often; 5 = almost every day) to describe how often teachers attempted to involve them in their children's schooling in the past month. These items included how often the teacher sent home written information, talked to the parent about the child, requested information about the child, invited parents to visit the classroom, and asked parents to help with school work. Cronbach's alpha for this parental report was .83. A summary score was computed by averaging the parent's responses (M = 2.4, SD = 0.7).

Another subscale (8 items) had parents rate the quality of their relationship with the teacher using 5-point ratings (1 = strongly disagree to 5 = strongly agree). Sample items included "I feel comfortable talking to this teacher"; "I feel this teacher and I are partners" (Cronbach's α = .96). The mean for parent reports of the parent-teacher relationship was 4.3 (SD = 0.7).

Finally, parents were asked to rate the quality of the school (M = 3.9, SD = 0.9) compared to other schools in the state on a scale of 1 = very poor to 5 = very good.

Child report. Participation in family activities was measured by asking children how often they engaged in such things as attending church, going on outings, and playing sports with their parents. This family activity report (M = 2.96, SD = 0.74) was based on 10 items (Cronbach's a = .82) derived from a larger measure of child activities ( Brown & Clasen, 1985). Children indicated how often in the past month they engaged in particular activities with their family using 5-point ratings: 1 = never, 2 = once or twice, 3 = three or four times, 4 = pretty often, or 5 = almost every day.

Children's perceptions of family supportiveness (M = 4.48, SD = 0.53) were obtained from a subscale of the Social Support Appraisals Scale ( Dubow & Ullman, 1989). The 12-item scale (Cronbach's a = .85) uses 5-point ratings (ranging from 1 = never to 5 = always). Sample items included "Do you feel your family is there when you need them?" and "Do you have a hard time talking to your family (reverse scored)?"

Teacher report. Teachers were asked to rate maternal (or custodial adult) involvement in schooling (Cronbach's a = .86; M = 3.74, SD = 0.73). They rated parent involvement practices that they would have knowledge of such as parent attendance at school functions, child's completion of reading at home, child's completion of homework, and child's discussion of an educational experience or outing with family. Nine items were included on the scale. Five-point ratings ranged from 1 = strongly disagree to 5 = strongly agree. In addition, a "don't know" option was available.

Teachers completed a questionnaire about home-school relationships, analogous to the questionnaire completed by parents. Teachers used 5-point ratings (1 = never; 2 = once or twice; 3 = three or four times; 4 = pretty often; 5 = almost every day) to describe their attempts to involve parents in their children's schooling during the past month. The eight items included how often the teacher sent home written information, talked to the parent about the child, requested information about the child, invited parents to visit the classroom, and asked parents to help with school work. Cronbach's alpha for the teacher report was .83. The mean score was 2.2 (SD = 0.7).

Another subscale (eight items) on the questionnaire had teachers rate the quality of their relationship with the parent using 5-point ratings (1 = strongly disagree to 5 = strongly agree). Sample items included "I feel comfortable talking to this mother"; "I feel this mother and I are partners." Cronbach's alpha equaled .96 for this teacher report. The mean for the teacher report of the teacherparent relationship was 4.0 (SD = 0.9).

Neighborhood risk. Children's addresses were keyed to census tracts. Information was obtained from the 1990 U.S. Census, including (a) proportion of female-headed households in the tract (M = 24%, SD = 18), (b) median years of education from persons over 18 years of age in the tract (M = 11.9 years, SD 0.9), and (c) median household income within the tract (M $22,530, SD = 8,484). In addition, information was obtained from the Milwaukee Police Department regarding the number of crimes reported against persons (homicide, rape, aggravated battery, other battery, and vice) within each tract for the calendar year ending in December of the children's fifth-grade year (M = 108.6, SD = 89.7). Principal component analysis was conducted to reduce this data. A single factor, which we called neighborhood risk, accounted for 76% of the variance in the data. Median income and educational levels were negatively weighted. Each child was assigned the component score for the risk within their neighborhood.

Some measures of children's academic performance were obtained from school records. The Wisconsin Third Grade Reading Test (M = 32.03, SD = 7.28), a 40-point standardized test of reading comprehension developed by the Wisconsin Department of Public Instruction, was administered to all third-grade students by the schools. The Iowa Test of Basic Skills (ITBS) was administered by the school district in the spring of fifth grade (mean percentile score = 53.2, SD = 27.7 for reading; mean percentile score = 53.9, SD = 30.0 for math). Report card grades were obtained for marking periods throughout the school year in both third and fifth grade. Grades were translated to a numerical equivalent and average grade points for reading and mathematics were calculated. The correlation between fifth-grade reading marks and the ITBS reading score was .57 (p < .001). The correlation between fifth-grade math marks and the ITBS math scale was .72 (p > .001).

In order to reduce the number of substantive analyses and to create a more reliable index, a reading achievement composite was created by averaging the standardized reading grade point for the fifth grade and the standardized ITBS reading score. Likewise, a math achievement composite was created by averaging the standardized math grade and standardized ITBS math score. (It should be noted that separate analyses of report card grades and achievement test scores yielded the same pattern of findings.)

Information regarding children's school orientation was ob

tained from children's self-appraisal of school competence, teacher reports of children's work habits, and report card conduct grades. Children's self-appraisal of school competence (M = 3.89, SD = 0.6; Cronbach's a = .80) was a 9-item scale from the SelfDescription Questionnaire ( Marsh, Smith, & Barnes, 1983), a measure validated for use with low-income inner-city elementary school children. The questionnaire utilized 5-point ratings ranging from 1 = totally false to 5 = totally true. Sample items included "I enjoy school work," "I learn quickly in school," and "I'm good at all school subjects."

Teachers rated children's work habits using a 7-item subscale from Santrock and Warshak ( 1979) Behavioral Adjustment Rating Scale. Items were rated using a 5-point rating scale ranging from 1 = always to 5 = never. Sample items included "This child is easily distracted," "This child is disorganized," and "This child quits working on a task when problems arise." In the current sample, a mean rating of 3.6 was obtained (SD = 0.8). Cronbach's alpha for the scale is .91.

The mean of all conduct grades (M = 1.86, SD = 0.18) was computed for the same marking period as the academic report grades. Conduct grades of "1" were recorded by the teacher if the child required improvement, otherwise conduct was considered satisfactory, and rated "2." The conduct marks consisted of study habits (e.g., following directions, work completion, independence, works well with others, class participation), and personal development (e.g., observes rules, shows self-control, accepts responsibility, respects rights of others, and works and plays safely). Pearson correlations from these three indices were conduct grades with teachers' report of work habits (r = .61), conduct grades with child's self-report of school competence (r = .20), and teachers' work habit ratings with child self-report of school competence (r = .37). The ratings were standardized and averaged to create a composite indicator of children's school orientation.

Three questions are investigated in these analyses: (a) Are family demographic, family psychological, and child prior adjustment variables associated with the utilization of school choice? (b) Are children's reading achievement, math achievement, and school orientation associated with school choice? and (c) Do home-school relations differ between school type?

Possible demographic and psychological variables associated with the utilization of school choice were investigated. The demographic variables considered were race (African American or Caucasian), child's gender, family structure (1-parent vs. 2-parent households), maternal education, reported family income, and neighborhood risk. Family psychological characteristics were firm-butresponsive parenting strategies, parental school involvement, family activities, and family supportiveness. Differences associated with categorical variables were tested using the chi-square test; t tests were used to test continuous variables.

As shown in Table 3, significant demographic and psychological differences were associated with the utilization of school choice. African American parents were more likely to select choice schools than were Caucasian parents. Children living in high-risk neighborhoods were more likely than children not living in high-risk neighborhoods to attend

a choice school. Single-parent households were more likely to use choice schools. Psychological differences associated with school choice also were obtained. Choice families were reported to spend more time engaged in activities with their children and to be more supportive of their children. Firmbut-responsive parenting was more common in families using choice arrangements.

Two differences favoring assigned schools were obtained. Per capita income was higher in families using assigned schools. Teachers reported that parents were more involved in assigned schools. No differences were obtained on maternal education or child gender to distinguish characteristics of families participating in school choice.

As Table 4 shows, some associations were found between the family demographic and psychological characteristics. Not surprisingly, family income was associated negatively with neighborhood risk and positively with two-parent households. Income was positively related to parent involvement in schooling. Living in higher risk neighborhoods was associated with being African American, with being a single-parent household, with firm-responsive parenting, and with participating in activities with the child. Firm-responsiveness was associated with being African American and with family activities. Family activity participation was positively related to family supportiveness. Finally, African American children were more likely to live in single-parent households, and they rated their families as more involved with them in activities and as more supportive than did Caucasian children.

The possibility that there were preexisting differences between children whose parents selected school choice was tested by comparing the mathematics grades, reading grades, Wisconsin Reading Test reading comprehension scores, conduct grades, and teacher rating of school adjustment of children who remained in their third-grade assigned schools (n = 87) for the course of the study with those whose parents moved them from assigned to choice schools (n = 15) after third grade. Comparison of least square means was used to test for differences due to the unbalanced design. No differences in reading grades (stable M = 1.66, SD = 0.06; moved M = 1.78, SD = 0.15), math grades (stable M = 1.74, SD = 0.06; moved M = 2.00, SD = 0. 14), conduct grades (stable M = 2.10, SD = 0.02; moved M = 2.15, SD = 0.05), teacher rating of work habits (stable M = 3.64, SD = 0.06; moved M = 3.53, SD = 0.15), or the Wisconsin Reading Test were obtained (stable M = 32.35, SD = 0.80; moved M = 31.66, SD = 1.91).

The relationship between school choice and children's mathematics achievement, reading achievement, and school orientation was tested, controlling for child gender and those family factors identified in the above analyses as distinguishing between families in choice and assigned schools. Simultaneous multiple regression analyses were used to determine the unique effects of school choice on children's academic achievement and school orientation, controlling for child's gender and race, per capita income, neighborhood risk, family structure, firm-but-responsive patenting strategies, shared family activities, family supportiveness, and parental involvement in schooling. Table 5 displays the results of these analyses.

Controlling for family characteristics, children who attended choice schools had higher mathematics achievement and stronger school orientations. Children did not differ on reading achievement between school types.

Inspection of the betas reveals that greater involvement in family activities was a negative predictor of reading and math achievement and unrelated to school orientation. The strongest positive predictor of reading and math achievement and of children's school orientation was parental involvement in schooling.

The results of parent and teacher ratings of the homeschool relationship are presented in Table 6. Parents and

teachers both had positive ratings of their relationship with each other. These ratings did not differ across choice conditions. At the same time, parents reported that teachers in choice schools used more parent involvement practices, but teacher reports of parent involvement practices were similar in both types of schools. No difference was found on the parent ratings of school quality between choice conditions.

The current study underscores the importance of considering family characteristics in studies of school choice. Significant differences were found between those families who utilized school choice options and those families whose children attended assigned schools. Specifically, in the current study, those families who selected their children's school were more likely to be African American, have less income, and live in higher risk neighborhoods. These findings are consistent with other reports that have found that racial minority and low-income parents are more likely to utilize school choice ( Bauch, 1988; Coleman et al., 1991;

Witte, Bailey, & Thom, 1992). Within the current sample, "skimming" of the more prosperous families from the assigned schools did not appear to be occurring. We also did not observe significant differences between those families who reported selecting neighborhoods for their schools and other families who used assigned schools. In future work that focuses on middle-income as well as low-income families, it will be important to determine if families with greater financial resources are more likely to select neighborhoods based on their schools.

In the current study, several psychological differences between families associated with school choice were found. Parents who chose schools were higher on firm-butresponsive parenting strategies than those using assigned schools. Other research has reported that firm-but-responsive parents are active in meeting their children's needs by considering their interests and establishing a stimulating environment ( Baumrind, 1989). Choosing a school seems to be another way of expressing this characteristic.

Yet another psychological difference was that parents who exercised school choice were reported by their children as engaging in more activities with them and as being more emotionally supportive. Greater time spent with children and more supportiveness may indicate that this group of parents was more committed to their parenting roles, which, in turn, led them to select specific schools for their children. An interesting question to explore in future research would be whether parents who select schools are more knowledgeable about their children and more committed to their roles, prior to exercising school choice. In contrast to our prediction, participation in family-based activities with the child was negatively related to the children's academic achievement. Clark ( 1983) offers one explanation of this finding. He argues that parental involvement must be focused on academic work in order to benefit the child in school.

It is notable that parents in the current study did differ in terms of their involvement in their children's schooling. According to the teachers, and contrary to our predictions, choice parents were less likely to be directly involved in their children's schooling. Instead, parents were more involved in assigned schools that were typically located in their own neighborhoods. This finding supports the claims made by critics of school choice that low-income parents will be at a disadvantage in being involved in their chil

dren's schools when the schools are not in the neighborhood due to limited access to transportation ( Harris, 1983; Riddle & Steadman, 1990). Another possible reason may be that parents who have selected the school feel more confident with the school practices and feel less need to be directly involved.

In the current study, we controlled for these demographic and psychological differences when examining associations between school choice and children's development. Otherwise, obtained associations with children's academic achievement might be explained by the demographic and psychological factors. Small (but statistically significant) differences in children's mathematics achievement and school orientation were associated with school choice, when per capita income, race, neighborhood risk, firm-responsive parenting, family-based activities, and parental involvement in school activities were controlled.

An important issue that we were not able to address in the current study is why school choice was related to better mathematics achievement and school orientation. One possibility is that the quality of choice schools is higher in ways that were not indicated by the global measures of school effectiveness. A second explanation for the beneficial effect on school orientation associated with school choice is that the action of selecting the school represents a psychological investment that then contributes to improved school performance. Parents' interpretations of teachers as practicing more parent-involvement strategies in choice schools (although teachers reported similar practices in both school types) suggested some psychological investment in the choice schools.

It must be underscored, however, that the obtained effects associated with school choice were quite small. A more powerful predictor of academic achievement and school orientation was parental involvement in their children's schooling. This involvement included homework supervision, school visits, and communication with the teacher. Given the strength of this proximal process compared to the modest effect of school choice, policies and programs to increase direct parental involvement in schooling may be of greater benefit in terms of children's school outcomes.

There are several limitations of our findings, which should be considered. First, children in our sample had attended choice schools for varying amounts of time. Some had always attended choice schools, whereas others began attending during fifth grade. We were unable to obtain reliable estimates of children's tenure in choice schools so were unable to control this potentially important variable. In addition, we do not know why parents selected the particular schools, but this would be an important question for future research. Our measures of school quality also were limited. A final limitation is that we examined choice only within a public school framework. Proponents of private school subsidization have argued that this form of school choice can contribute positively to student achievement ( Coleman, Hoffer, & Kilgore, 1982). It will be important for future research to examine this form of school choice more closely, especially considering family selection effects that may contribute to differences in child academic achievement.

In summary, we found that school choice was associated with differences in family characteristics. School choice was associated with small but statistically significant differences in mathematics achievement and school orientation, when these family demographic and psychological background factors were controlled. Parent's involvement in their children's schooling was found to have a more powerful effect than school choice on both mathematics achievement and school orientation. In light of these findings, future studies are needed to determine whether policy and practice may be best directed at engaging parents in direct and active involvement with their children's schooling rather than in instituting school choice.

Bauch, P. A. ( 1988 ). "Is parent involvement different in private schools?" Educational Horizons, 66, 78-82.

Baumrind, D. ( 1989 ). Rearing competent children. In W. Damon (Ed.), Child development today and tomorrow (pp. 349-377). San Francisco: Jossey-Bass.

Becker, H., & Epstein, J. ( 1982 ). "Parent involvement: A survey of teacher practices". The Elementary School Journal, 83, 85-102.

Bempechat, J. ( 1990 ). The role of parent involvement in children's academic achievement: A review of the literature. Trends and Issues No. 14. New York: Columbia University, Teacher's College, Institute for Urban and Minority Education. (ERIC Document Reproduction Service No. ED 322 285)

Blank, R. ( 1984 ). "The effects of magnet schools on the quality of education in urban school districts". Phi Delta Kappan, 66, 270272.

Brown, B., & Clasen, D. ( 1985 ). Self-reported behavior index. Unpublished manuscript, University of Wisconsin-Madison Department of Educational Psychology Spencer Peer Group Study.

Capell, F. ( 1978 ). A study in alternatives in American education (Vol. 6). Santa Monica, CA: Rand.

Carnegie Foundation for the Advancement of Teaching. ( 1992 ). School choice. Princeton, NJ: Author.

Clark, R. ( 1983 ). Family life and school achievement. Chicago: University of Chicago Press.

Clune, W. H. ( 1990 ). Educational governance and student achievement. In W. H. Clune & J. F. Witte (Eds.), Choice and control in American education: Vol. 2. The practice of choice, decentralization and school restructuring (pp. 391-423). New York: Falmer.

Coleman, J. ( 1992 ). "Some points on choice in education". Sociology of Education, 65, 260-262.

Coleman, J., Hoffer, T., & Kilgore, S. ( 1982 ). High school achievement: Public, Catholic, and private schools compared. New York: Basic Books.

Coleman, J., Schiller, K., & Schneider, B. ( 1991 ). Parental involvement in school choice. Unpublished manuscript, University of Chicago.

Cookson, P. ( 1994 ). School choice: Struggle for the soul of American education. New Haven, CT: Yale University Press.

Darling-Hammond, L., & Kirby, S. N. ( 1985 ). Tuition tax deductions and parent school choice: A case study of Minnesota (Report No. R-3294-NIE). Santa Monica, CA: Rand. (ERIC Document Reproduction Service No. ED 273 047)