Who chooses non‐public schools for their children?

21
This article was downloaded by: [University of North Texas] On: 21 November 2014, At: 15:14 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK Educational Studies Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ceds20 Who chooses nonpublic schools for their children? Philip Q. Yang a & Nihan Kayaardi a a Texas Woman's University , USA Published online: 07 Oct 2010. To cite this article: Philip Q. Yang & Nihan Kayaardi (2004) Who chooses nonpublic schools for their children?, Educational Studies, 30:3, 231-249, DOI: 10.1080/0305569042000224198 To link to this article: http://dx.doi.org/10.1080/0305569042000224198 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms- and-conditions

Transcript of Who chooses non‐public schools for their children?

Page 1: Who chooses non‐public schools for their children?

This article was downloaded by: [University of North Texas]On: 21 November 2014, At: 15:14Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Educational StudiesPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/ceds20

Who chooses non‐public schools fortheir children?Philip Q. Yang a & Nihan Kayaardi aa Texas Woman's University , USAPublished online: 07 Oct 2010.

To cite this article: Philip Q. Yang & Nihan Kayaardi (2004) Who chooses non‐public schools fortheir children?, Educational Studies, 30:3, 231-249, DOI: 10.1080/0305569042000224198

To link to this article: http://dx.doi.org/10.1080/0305569042000224198

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to orarising out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Page 2: Who chooses non‐public schools for their children?

Educational Studies, Vol. 30, No. 3, September 2004

Who chooses non-public schools for theirchildren?Philip Q. Yang* & Nihan KayaardiTexas Woman’s University, USA

Using the pooled 1998–2000 GSS data, this study examines what kinds of parents tend to selectnon-public schools for their children, a question that is fundamental but lacks direct, adequateanswers in the literature. The results of logistic regression analysis show that religion, socio-econ-omic status, age, nativity, number of children and region play significant roles in parental choiceof religious schools, but race, gender, urban residence and family composition make no difference.Parental socio-economic status is a key factor in determining their choice of non-religious privateschools. Currently, no significant differences in demographic, religious, socio-economic andfamily-structure characteristics can be detected with regard to which parents are more likely tohome-school their children. Implications of the findings are considered.

Keywords: School choice; Determinants; Non-public schools; Private school; Religiousschool; Home school

School choice is probably the most important movement in US education reformstoday. The support for wider parental choice of schools has gained momentum inrecent years (Goldhaber, 1999). Concomitantly with this movement, a growingbody of the literature on school choice has emerged. This vast literature coversseveral major strands. One emphasizes the policy perspective of school choice andcompares a variety of school choice programmes (e.g. Godwin et al., 1998;Schneider et al., 1998; Weiss, 1998). A second strand concerns the relationshipbetween school choice and parental school involvement (Bauch, 1988, 1991;Cooper, 1991; Bauch & Goldring, 1995). Another large strand of research focuseson the outcomes of school choice in terms of student achievement, school improve-ment, equity and community (Coleman et al., 1982; Goldberger & Cain, 1982; Cain& Goldberger, 1983; Noell & Myers, 1983; Coleman & Hoffer, 1987; Goldhaber,1996, 1999). Still another thread of research deals with the determinants of schoolchoice, which include both descriptive analysis of reasons for choosing a particulartype of school (e.g. Maddaus, 1990; Bauch, 1991; Crawford & Freeman, 1996) andexplanatory analyses of school choice determination (e.g. West & Palsson, 1988;Hamilton & Macauley, 1991; Lankford & Wyckoff, 1992). Our study tackles thislast aspect of school choice.

*Corresponding author: Department of Sociology and Social Work, Texas Woman’s University,Denton, Texas, 76204-5887, USA. Email: [email protected]

ISSN 0305-5698 (print)/ISSN 1465-3500 (online)/04/030231-19 2004 Taylor & Francis LtdDOI: 10.1080/0305569042000224198

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 3: Who chooses non‐public schools for their children?

232 P. Q. Yang & N. Kayaardi

Most American students historically attended public schools, and a largemajority of them still do today. Within the confines of the public system, twoalternatives are magnet schools with subject-specialized programmes and chapterschools, a hybrid between a public school system and private choice. Outside thedomain of the public system, a main alternative is non-religious private orindependent school (hereafter private school); other alternatives of non-publicschools include religious schools and, most recently, home school. Who sendstheir children to non-public schools? This is a basic question to educators,researchers and policy-makers, and the main question of this research. An insightinto this question can help us understand factors that influence parental schoolchoice and address the issue of social equity in school choice. Since familybackground characteristics affect student performance, an answer to this questioncan also shed light on educational outcomes of school choice. However, surpris-ingly few studies have directly answered this basic question. Even among a fewstudies that directly or indirectly address this question, there are limitations. Forexample, the studies of West & Palsson (1988) and Hamilton and Macauley(1991) are aggregate-level analyses of private/public school choice, and thus theirresults may not be applicable to parents’ probability of sending children to privateschools at the individual level. Lankford and Wyckoff (1992) focused on whochose religious schools versus public schools, using the 1980 Census PUMS datafrom New York State merging with New York school data rather than nationaldata. Coleman and his associates (Coleman et al., 1982; Coleman & Hoffer,1987) did use the national data—the General Social Survey data gathered inspring 1980—and examined the differences among Catholic schools, other pri-vate schools and public schools. However, their emphasis was on the variation ineducational outcomes; they did not directly address the question of who choosesnon-public schools for their children; and their results are largely based oncross-tabulations without much control of relevant variables. In addition, with thedemographic shifts and transformations in US class structure that occurred in thepast two decades, there could be changes in parental characteristics that influencetheir school choice. At the present, there is little national-level information thatcan distinguish which parents choose religious schools in general (not just Cath-olic schools) over public schools and which select non-religious private schoolsagainst public schools. Furthermore, basically no information exists with regardto what kinds of parents home-school their children because of the recency ofhome-schooling.

In this paper, we examine who is more or less likely to send their children tononpublic schools, including religious schools, private schools and home schools.We are particularly interested in determining whether there are any patterns ofparental characteristics that are associated with their choices of non-public schools.We perform our analysis using the unique 1998–2000 General Social Surveys (GSS)data and logistic regression. In the following, we first develop a conceptual frame-work for analysing parental school choice; we then describe our data and method;and empirical findings are then presented. This paper concludes with a summary ofour findings and a discussion of their implications.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 4: Who chooses non‐public schools for their children?

Who chooses non-public schools? 233

Theoretical framework and hypotheses

We believe that which types of parents choose non-public schools for their childrenvary by four types of parental characteristics: religion, socio-economic status, familystructure and demographic characteristics. In this section, we discuss how thesefactors influence parental choice of nonpublic schools, review pertinent findingsfrom prior studies and propose hypotheses testable using the GSS data.

Religion

Obviously, parents’ religion is an important consideration in the selection of areligious school for their children. This is so because religious parents normallyprefer to raise their children in a way consistent with their personal religiouspreference. Religious schools fulfil this function of instilling students with particularreligious beliefs. Available empirical evidence shows that parental religious prefer-ence does have a positive effect on the selection of Catholic or private school. Forinstance, Coleman et al. (1982) found that a family’s Catholic religion increases theprobability of children’s enrolment in Catholic schools. Lankford and Wyckoff(1992) also reported that Catholicism is positively associated with enrolment inreligious elementary and secondary schools. Based on the above theorization andempirical evidence, we hypothesize that parents with religious backgrounds are morelikely than non-religious parents to send their children to religious schools in lieu ofpublic schools. The effect of parental religion on the selection of non-religiousnon-public schools should be much weaker.

Socio-economic status

The existing literature has established socio-economic status as an important factorin determining parental choice of schools for their children. Parents’ education andfamily income are two indicators often cited in the literature on school choice.Education represents an orientation towards education and a value placed oneducation (Coleman & Hoffer, 1987). That is, parents with higher educationalattainment better understand the importance of education, what different kinds ofschools offer and what they want their children to acquire, and therefore are in abetter position to make an informed decision. Empirical findings consistently showa positive relationship between parents’ educational attainment and the likelihood ofsending children to either religious or private school (Coleman et al., 1982; Noell,1982; Noell & Myers, 1983; Coleman & Hoffer, 1987; Long & Toma, 1988;Lankford & Wyckoff, 1992). Family income is an indicator of resources. Simply,higher family income increases one’s ability to afford tuition in private and religiousschools, which are believed to offer a better education. Schneider et al. (1996)concluded that, with more resources, high-income families are more likely to sendtheir children to non-public schools than low-income families, who simply cannotafford otherwise. Coleman and his associates (Coleman et al., 1982; Coleman &Hoffer, 1987) also found that the higher the family income, the higher the private

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 5: Who chooses non‐public schools for their children?

234 P. Q. Yang & N. Kayaardi

school enrolment. Another indicator of socio-economic status is occupational pres-tige score, which primarily reflects orientation but is associated with resources aswell. Occupational prestige score should be positively related to the selection ofnon-public schools. In addition, socio-economic status may be measured by asubjective measure for social class. We expect a positive relationship between socialclass and the likelihood of choosing non-public schools. We predict that, in general,the higher the socio-economic status, the more likely the parents are to send theirchildren to non-public schools.

Family structure

Family structure is pertinent to parental school choice because it affects familyresources and environment for education. Family structure includes family compo-sition and number of children in the family. Family composition influences parentalschool choice because an intact family provides a better environment and moreresources (e.g. money, time) for children’s education, and wellbeing in general, thana single parent family. Hence, parents of an intact family are more likely than thosein a single-parent family to select non-public schools, who are presumed to havebetter-quality education. Available evidence shows that two-parent families areindeed more likely to send children to private schools, including Catholic schoolsand other private schools (Coleman & Hoffer, 1987). On the other hand, morechildren dilute family resources. Families with fewer children are less likely to sendtheir children to public schools because they are more involved in their children’seducational attainment (Martinez et al., 1996). Thus, we expect that parents withmore children will be less likely to choose non-public schools for their children.

Demographic characteristics

Past research has detected the impacts of certain demographic variables on parentalschool choice decisions, although empirical results are sometimes contradictory andinconclusive. For example, previous studies have revealed significant racial differ-ences in parental school choice, yet the effect of race is mixed. Coleman and hisassociates (Coleman et al., 1982; Coleman & Hoffer, 1987) found that blacks andHispanics are under-represented in private schools compared to non-Hispanicwhites; however, when religion is statistically controlled, black children are as likelyas white children to be in Catholic schools. Similarly, Noell and Myers (1983)reported that non-whites are less likely to register children in religious schools thanwhites, but there is no racial difference in choosing non-religious schools. Long &Toma (1988) found that whites are more likely to attend both religious schools andprivate schools. Since racial minorities on average possess less resources than whites,we expect that all else being equal, blacks and other racial minorities will be lesslikely to send children to non-public schools than whites. Lankford and Wyckoff(1992) found insignificant gender differences in parental selection of religiouselementary and secondary school. Nevertheless, Schneider et al. (1996) argued thatif the female parent is dominant in the decision, the family is more likely to send its

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 6: Who chooses non‐public schools for their children?

Who chooses non-public schools? 235

children to public schools. This can be attributed to the lower income associatedwith female-headed households in the USA. Hence, male parents might be morelikely to favour sending their children to non-public schools than their femalecounterparts. Age may be positively associated with probability of choosing non-public schools because older people are more likely than younger ones to prefer thetradition of attending religious and private schools to new options. Immigrantparents should be less likely to send their children to non-public schools because oftheir lack of resources relative to their native-born counterparts. Urban residence isfound positively associated with the likelihood of choosing non-public schools(Lankford & Wyckoff, 1992). Hence, parents living in urban areas should be morelikely to register their children in non-public schools than rural parents. Coleman etal. (1982) also found that people in the North-east had a higher probability ofenrolment in Catholic schools. Thus, regional variation in parental school choiceshould be tested. Given the historical tradition and a high concentration of Catholicsin the North-east, we expect that parents in the North-east should be more likely tosend their children to non-public schools than those in the West, and there could bedifferences between those in the Mid-west and South and those in the West as well.Since we use data of 1998 and 2000, survey year should be controlled to see if thereis any significant difference in parental school choice between the two years. In brief,demographic variables ought to be tested for their possible effects, or at least becontrolled in order to yield more accurate estimates for the other three categories offactors.

Data and method

The data for this analysis come from the 1998–2000 GSS data (NORC, 2001). TheGSS is a nationally representative sample of the US adult population. In the latest1998 and 2000 surveys, the GSS asked respondents a series of questions aboutwhich types of schools they sent their children to. This provides a unique oppor-tunity to study what types of parents choose which types of schools for theirchildren. We pooled the 1998 and 2000 surveys in order to increase the sample sizefor more reliable results and to assess the change over time. We restrict our analysisto the respondents with children who provided valid responses to the surveyquestions on school choice. This yields a sample size of 2416. The sample sizes oflogistic regression models vary because of missing values for certain independentvariables.

Our main interest lies in what kinds of parents send their children to schools otherthan public schools attended by most children in the USA—the standard option. Weconstructed several dependent variables (see Table 1): (1) religious school is adichotomous variable, coded 1 for sending child to a religious school and coded 0for sending child to public school; (2) private school is also a dichotomous variable,with 1 indicating sending child to a non-religious private school and 0 indicatingsending child to public school; and (3) home school is a dichotomous variable, with1 for keeping child at home school and 0 for sending child to public school.

Table 1 also summarizes the measurements of parental characteristics (the inde-

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 7: Who chooses non‐public schools for their children?

236 P. Q. Yang & N. Kayaardi

Table 1. Descriptive statistics of variables used in the analysis

Variable MeanMeasurement Standard deviation

Dependent variables0.110 0.32Religious school 1 � religious school, 0 � public

schoolPrivate school 1 � private school, 0 � public 0.033 0.18

school0.16Home school 0.0281 � home school, 0 � public

schoolPredictors

15.48Age 51.61Years0.59Female 0.491 � female, 0 � male

0.47Survey year 0.671 � 2000, 0 � 1998Race

0.38Black 0.171 � black, 0 � else0.05Other race 0.221 � other race, 0 � else

0.29Nativity 0.911 � native born, 0 � elseUrban 0.881 � urban, 0 � rural 0.32Region

0.20North-east 0.401 � North-east, 0 � else0.43Mid-west 0.251 � Mid-west, 0 � else

0.38South 0.481 � South, 0 � else0.35Christian 0.851 � Christian, 0 � else

12.94Education 2.94Years of schooling13.73Occupational prestige score 43.73Points (0–100)

Family income 15.6123-point ordinal scale 5.360.67Class 2.494-point ordinal scale

Intact family 0.561 � two parents, 0 � else 0.50Number of children 2.65Number 1.48

Notes:For race, the reference category is whites (mean � 0.78, SD � 0.42). For region, the omittedcategory is the West (mean � 0.18, SD � 0.38).

pendent variables) used in the analysis. Since the description is straightforward, onlya brief explanation is necessary. We include four categories of predictor variables.We use a dummy variable to represent religion: Christian. We use four indicators tomeasure socio-economic status: education, occupational prestige score, family in-come and a 4-point subjective social class scale. To measure family structure, we usea dummy variable indicating the presence of both parents (i.e. an intact family) anda continuous variable for number of children. Finally, demographic variables includedummy variables for survey year, gender, race, nativity, urban residence and region,as well as a continuous variable for age.

It should be noted that there are limitations with the GSS data. Many otherfactors could potentially influence parental choice of non-public schools, such asschool characteristics (e.g. the quality, curriculum, resources, discipline, extra-cur-riculum activities, teaching staff and student backgrounds of schools), child’s desire,the legal and regulatory environment of public education and household residential

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 8: Who chooses non‐public schools for their children?

Who chooses non-public schools? 237

Figure 1. Percentage distribution of parental choice of school by type of school, 1998–2000

choice. The GSS data lack information on these potential determinants of parentalschool choice. Nonetheless, the large, nationally representative sample, the latestinformation, and many available parental characteristics in the GSS provide aunique opportunity to gain insight into the determinants of parental school choice.

We use logistic regression to test the determinants of parental school choice.Logistic regression is most appropriate because the dependent variables are alldichotomous. For each of the dependent variables, we first test a baseline model thatincludes the demographic variables largely as controls, and we then gradually add tothe demographic model the religion variable, the socio-economic status variablesand, finally, the family structure variables. This strategy allows us to determinewhich factors influence a dependent variable and how the effect of a predictorchanges when new variables are included.

Results

Descriptive analysis

Figure 1 confirms that the majority of the parents sent their children to publicschool. However, a total of about 17% of the parents sent their children tonon-public schools, including 10.7% to religious schools, 2.9% to private schools,2.4% to home schools and 1.4% to other schools.

The means and standard deviations of the variables used in the analysis are alsoreported in Table I. The average age of the respondents was 51.6 years. The meanof a dummy variable can be interpreted as percentage. Women made up 59% of thesample. More respondents were surveyed in 2000 than in 1998. Whites (the

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 9: Who chooses non‐public schools for their children?

238 P. Q. Yang & N. Kayaardi

reference category) accounted for 78% of the sample, blacks 17% and other races5%. An overwhelming majority of the respondents (91%) were born in the USA.Most (88%) lived in urban areas. There were more respondents residing in theSouth than in any other regions. About 85% reported being Christian. A typicalrespondent finished a bit more than high school and had an occupational prestigescore of 43.7 points out of 100 points. A score of 15.6 for the family income variableindicates that the average family income of the respondents was between $30000and $34999. A typical respondent self-proclaimed to be middle class. More thanhalf of the respondents had an intact family, and they had an average of 2.7 children.We now turn to multivariate analyses.

Parental choice of religious school

Table 2 reports the estimates of four nested logistic regression models predictingparental choice of religious schools versus public schools. Model 1 includes only thedemographic variables as predictors. Model 2 fits the data better than Model 1,because by adding religion, the model �2 increases by 13.2 ( � 43.3 � 30.1), whichis statistically significant at beyond the 0.001 level with a difference of 1 degree offreedom ( � 11 � 10). This suggests that religion has a great predictive power of theparental choice of religious school against public school. As indicated by the model�2 change, Model 3 is a better fitting model than Model 2. The difference betweenthe two model �2s is 34, which is highly significant at beyond the 0.001 level witha difference of 4 degrees of freedom. This result means that socio-economic statusis important in predicting parental choice of religious school. Finally, adding the twofamily structure variables also significantly improves the model fit of Model 4 overModel 3. The difference between the two model �2s is 6.4, which is significant at the0.05 level with a difference of 2 degrees of freedom. Hence, Model 4 is the bestfitting model, on which our interpretations of the logistic regression estimates willfocus.

As shown in Model 4, age has a significant positive effect on the dependentvariable. This suggests that, as expected, older parents are more likely to send theirchildren to religious schools than younger parents. Nativity is also a significantpredictor. The significant, negative coefficient suggests that US-born parents are lesslikely than their foreign-born counterparts to send children to religious schools. Thisresult is at odds with our hypothesis. One plausible explanation for this unexpectedresult is that many immigrant parents are not pleased with the quality of US publicschools, but cannot afford the more expensive non-religious private schools; hence,they select religious schools as a compromise. Discipline and safety may also pullthem towards religious schools. Compared to those in the West, respondents livingin the North-east and Mid-west are more likely to send their children to religiousschools, but the difference between those in the South and those in the West is notstatistically significant. These results are generally very consistent across the fourmodels.

However, several demographic variables consistently show no effect on the depen-dent variable. Gender makes no difference in parental choice of religious school.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 10: Who chooses non‐public schools for their children?

Who chooses non-public schools? 239

Tab

le2.

Log

isti

cre

gres

sion

esti

mat

esfo

rde

term

inan

tsof

pare

ntal

choi

ceof

relig

ious

scho

olve

rsus

publ

icsc

hool

s

Pre

dict

ors

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Bex

p(B

)B

exp(

B)

Bex

p(B

)B

exp(

B)

Dem

ogra

phic

char

acte

rist

ics

�0.

213

0.80

8�

0.21

00.

810

Yea

r(2

000

�1)

�0.

216

0.80

6�

0.22

10.

802

(0.1

35)

(0.1

35)

(0.1

52)

(0.1

53)

�0.

115

0.89

2�

0.15

40.

858

0.13

01.

139

0.16

7F

emal

e1.

181

(0.1

31)

(0.1

32)

(0.1

49)

(0.1

51)

0.00

8*1.

008

0.00

71.

007

0.01

6**

Age

1.01

60.

013*

1.01

3(0

.004

)(0

.004

)(0

.005

)(0

.006

)R

ace

(whi

te�

ref.

)�

0.26

40.

768

�0.

275

0.75

9B

lack

0.04

71.

048

�0.

039

0.96

2(0

.191

)(0

.191

)(0

.224

)(0

.229

)�

0.11

80.

888

Oth

er�

0.11

40.

892

�0.

148

1.15

90.

143

1.15

4(0

.313

)(0

.313

)(0

.332

)(0

.332

)�

0.44

2*0.

643

�0.

468*

0.62

6�

0.56

9*N

ativ

e-bo

rn0.

566

�0.

574*

0.56

3(0

.217

)(0

.219

)(0

.245

)(0

.246

)U

rban

0.63

1*1.

880

0.66

1**

1.93

60.

425

1.52

90.

425

1.53

0(0

.251

)(0

.252

)(0

.265

)(0

.267

)R

egio

n(W

est�

ref.

)N

orth

-eas

t0.

538*

1.71

20.

536*

1.70

90.

627*

*1.

873

0.60

8*1.

837

(0.2

17)

(0.2

18)

(0.2

42)

(0.2

43)

Mid

-wes

t0.

508*

1.66

10.

491*

1.63

30.

584*

1.79

20.

542*

1.72

0(0

.214

)(0

.215

)(0

.231

)(0

.233

)0.

262

1.30

00.

241

1.27

3S

outh

0.36

21.

437

0.37

21.

451

(0.2

09)

(0.2

10)

(0.2

26)

(0.2

27)

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 11: Who chooses non‐public schools for their children?

240 P. Q. Yang & N. KayaardiT

able

2.—

cont

inue

d

Pre

dict

ors

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Bex

p(B

)B

exp(

B)

Bex

p(B

)B

exp(

B)

Rel

igio

nC

hris

tian

0.77

7***

2.17

50.

915*

**2.

498

0.88

0***

2.41

1(0

.238

)(0

.274

)(0

.275

)S

ocio

-eco

nom

icst

atus

Edu

cati

on0.

065*

1.06

80.

068*

1.07

0(0

.032

)(0

.033

)O

ccup

atio

nal

pres

tige

0.01

21.

012

0.01

3*1.

013

(0.0

06)

(0.0

06)

Fam

ilyin

com

e0.

062*

**1.

064

0.06

9***

1.07

1(0

.019

)(0

.021

)�

0.10

80.

898

�0.

103

0.90

2C

lass

iden

tific

atio

n(0

.132

)(0

.132

)F

amily

stru

ctur

eIn

tact

fam

ily�

0.08

20.

921

(0.1

76)

Num

ber

ofch

ildre

n0.

140*

*1.

150

(0.0

53)

�2.

776*

**0.

062

Con

stan

t�

3.36

3***

0.03

5�

6.13

3***

0.00

2�

6.44

9***

0.00

2(0

.427

)(0

.468

)(0

.715

)(0

.731

)�

2lo

glik

elih

ood

1670

.116

56.0

1325

.113

13.8

30.1

43.3

77.3

83.7

Mod

el�2

1011

df15

172,

416

N2,

412

1,99

91,

996

Not

es:

The

Bis

the

logi

stic

regr

essi

onco

effic

ient

;ex

p(B

)or

odds

rati

ois

the

anti

log

ofB

;an

dst

anda

rder

rors

are

inpa

rent

hese

s.*p

�0.

05;

**p

�0.

01;

***p

�0.

001.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 12: Who chooses non‐public schools for their children?

Who chooses non-public schools? 241

Race has no significant effect despite the expected negative coefficient for the blackdummy variable. There is no significant difference between respondents in 2000 andthose in 1998 in parental choice of religious schools. Urban residence has asignificant effect on the dependent variable in Models 1 and 2. Nevertheless,controlling for the socio-economic variables and the family structure variablesrenders the effect of urban residence insignificant, although the positive relationshipremains.

As anticipated, parental religion has a substantial effect on parental choice ofreligious school. The odds ratio (2.411) indicates that parents who are Christian areabout 2.4 times as likely as those who are non-Christian to choose religious schoolsfor their children. This is not surprising because parents want their children tofollow their religious beliefs.

All of the three objective measures of socio-economic status have a positive effecton parental choice of religious school. The higher the levels of parental educationalattainment, occupational status and family income, the more likely the parents areto send their children to religious school. These results are consistent with ourhypothesis. On the other hand, the subjective measure of social class does not havea significant effect on parental choice of religious school.

Finally, the effects of the family structure variables contradict our hypotheses.Whether intact family exists does not influence parental choice of religious school.This suggests that the presence of both parents may not be important in parentalselection of a religious school. The significant positive relationship between numberof children and parental choice of religious school also runs counter to our predic-tion that more children reduce family resources and therefore lower the likelihood ofsending children to non-public schools. This result suggests that the importance ofreligious education and discipline in religious schools may overshadow their cost,which is relatively low compared to private schools. Parents with more children mayprefer to send them to the same school to maintain consistency in education, to keeptheir children company and to save time and expense on transportation.

Parental choice of private school

We performed the same logistic regression analysis of the parental choice of privateschool versus public school. The results are reported in Table 3. Comparing themodel fit statistics at the bottom, it is clear that Model 2 fits the data better thanModel 1, and Model 3 is the best fitting model (the difference between Models 2and 3 in model �2s is 57.8, which is highly significant at beyond the 0.001 level witha difference of 4 degrees of freedom). However, according to the parsimonyprinciple, Model 4 does not improve the model fit over Model 3 since the increasein model �2s is 1.5, which is not significant at the 0.05 level with a difference of 2degrees of freedom. Hence, we focus our interpretation on Model 3.

Only two variables show significant effects on the parental choice of privateschool. Respondents in the South are more likely to send their children to privateschools than those in the West. Parental educational level significantly increases thelikelihood of sending children to a private school versus to a public school. This

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 13: Who chooses non‐public schools for their children?

242 P. Q. Yang & N. Kayaardi

Tab

le3.

Log

isti

cre

gres

sion

esti

mat

esfo

rde

term

inan

tsof

pare

ntal

choi

ceof

priv

ate

scho

ols

vers

uspu

blic

scho

ols

Pre

dict

ors

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Bex

p(B

)B

exp(

B)

Bex

p(B

)B

exp(

B)

Dem

ogra

phic

char

acte

rist

ics

�0.

175

0.84

0�

0.19

60.

822

Yea

r(2

000

�1)

�0.

386

0.68

0�

0.39

80.

671

(0.2

52)

(0.2

53)

(0.2

79)

(0.2

80)

�0.

010

0.99

10.

077

1.08

00.

458

1.58

00.

467

Fem

ale

1.59

5(0

.246

)(0

.249

)(0

.286

)(0

.291

)0.

008

1.00

80.

010

1.01

00.

019

Age

1.02

00.

017

1.01

7(0

.008

)(0

.008

)(0

.011

)(0

.011

)R

ace

(whi

te�

ref.

)�

0.55

90.

572

�0.

512

0.59

9B

lack

0.52

81.

695

0.47

61.

610

(0.3

70)

(0.3

72)

(0.4

11)

(0.4

15)

�6.

039

0.00

2O

ther

�6.

055

0.00

2�

5.75

60.

003

�5.

774

0.00

3(9

.459

)(9

.372

)(9

.500

)(9

.580

)�

0.16

50.

848

�0.

134

0.87

5�

0.31

5N

ativ

e-bo

rn0.

730

�0.

339

0.71

3(0

.447

)(0

.448

)(0

.489

)(0

.489

)U

rban

1.01

52.

759

0.96

22.

616

0.50

71.

660

0.47

41.

606

(0.5

26)

(0.5

27)

(0.5

47)

(0.5

50)

Reg

ion

(Wes

t�re

f.)

Nor

th-e

ast

0.12

51.

133

0.13

71.

147

0.44

11.

555

0.41

41.

513

(0.3

82)

(0.3

83)

(0.4

40)

(0.4

40)

Mid

-wes

t�

0.91

30.

401

�0.

868

0.42

0�

0.95

90.

383

�1.

022

0.36

0(0

.478

)(0

.479

)(0

.601

)(0

.603

)0.

367

1.44

40.

422

1.52

4S

outh

0.83

5*2.

304

0.80

8*2.

242

(0.3

38)

(0.3

39)

(0.3

73)

(0.3

74)

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 14: Who chooses non‐public schools for their children?

Who chooses non-public schools? 243

Rel

igio

n�

0.83

7**

0.43

3C

hris

tian

�0.

592

0.55

3�

0.60

60.

545

(0.2

80)

(0.3

27)

(0.3

26)

Soc

io-e

cono

mic

stat

usE

duca

tion

0.21

7***

1.24

30.

218*

**1.

244

(0.0

64)

(0.0

65)

Occ

upat

iona

lpr

esti

ge0.

016

1.01

60.

016

1.01

7(0

.012

)(0

.012

)0.

079

Fam

ilyin

com

e1.

082

0.09

5*1.

099

(0.0

41)

(0.0

45)

Cla

ssid

enti

ficat

ion

0.11

61.

123

0.11

61.

123

(0.2

57)

(0.2

58)

Fam

ilyst

ruct

ure

Inta

ctfa

mily

�0.

273

0.76

1(0

.330

)N

umbe

rof

child

ren

0.11

01.

116

(0.1

12)

�4.

360*

**0.

013

�3.

882*

**0.

021

�9.

970*

**0.

000

�10

.142

***

Con

stan

t0.

000

(0.8

26)

(0.8

46)

(1.3

67)

(1.3

83)

�2

log

likel

ihoo

d60

7.2

599.

044

8.4

446.

827

.835

.893

.695

.1M

odel

�2

1011

df15

172,

216

N2,

212

1,83

71,

835

Not

es:

The

Bis

the

logi

stic

regr

essi

onco

effic

ient

;ex

p(B

)or

odds

rati

ois

the

anti

log

ofB

;an

dst

anda

rder

rors

are

inpa

rent

hese

s.*p

�0.

05;

**p

�0.

01;

***p

�0.

001.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 15: Who chooses non‐public schools for their children?

244 P. Q. Yang & N. Kayaardi

supports our hypothesis. Family income is positively associated with the dependentvariable, but it is only marginally significant at the 0.055 level in Model 3 (it attainsstatistical significance in Model 4). These results suggest that parents’ education andpossibly family income are important in parental choice of private schools.

Religion has a significant effect on parental choice of private school in Model 2,but it lost statistical significance after holding the socio-economic status variablesconstant. This suggests that socio-economic status intervenes in the relationshipbetween religion and school choice and overshadows religion in parental choice ofprivate school.

All other predictors fail to show significant effects on the dependent variable. Nosignificant gender, racial, age, nativity, urban residence and year effects can bedetected. Family structure has no impact on the dependent variable either. Theseresults reinforce the conclusion that it is primarily parents’ socio-economic statusthat makes a difference in the selection of private school.

Parental choice of home school

We replicated a similar logistic regression analysis for parental choice of home schoolversus public school. Table 4 shows the results. The best fitting model is Model 1in terms of the parsimony principle. Only one predictor variable, namely survey year,has a significant effect on the dependent variable. The negative coefficient indicatesthat, holding other factors constant, parents in 2000 were less likely to send childrento home school than parents in 1998. This effect is highly significant and consistentacross all four models. This result may indicate a temporary fluctuation or thebeginning of a cooling-off period in home-schooling.

All other demographic, religious, socio-economic status and family structurevariables do not reach statistical significance, probably due to a small number ofhome-schooling cases in the sample. The weak yet consistent effects of somevariables may give us some hints about what may happen to who chooses home-schooling in the future. Blacks and other racial minorities are less likely than whitesto home-school their children, but the differences are not statistically significant atthe 0.05 level. Also insignificant at the 0.05 level, women are more likely to favourhome-schooling than men. In addition, native-born parents, urban parents, Chris-tian parents and parents in intact families are more likely to home-school theirchildren than their respective counterparts. Home-schooling is greater in the North-east, South and Mid-west than in the West. However, all of the foregoing differencesare not large enough to reach the 5% level at the present time. More time and dataare needed to further detect discernible patterns of who is more likely to choosehome-schooling.

Conclusion

One of the major findings of this research is that religion is not the only factor thatinfluences parental choice of religious schools. In fact, multiple factors are at work.Parents who are Christian, who are older, who are foreign-born, who have a higher

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 16: Who chooses non‐public schools for their children?

Who chooses non-public schools? 245

Tab

le4.

Log

isti

cre

gres

sion

esti

mat

esfo

rde

term

inan

tsof

pare

ntal

choi

ceof

hom

esc

hool

vers

uspu

blic

scho

ols

Pre

dict

ors

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Bex

p(B

)B

exp(

B)

Bex

p(B

)B

exp(

B)

Dem

ogra

phic

char

acte

rist

ics

�1.

236*

**0.

291

�1.

223*

**0.

294

Yea

r(2

000–

1)�

1.39

8***

0.24

7�

1.37

0***

0.25

4(0

.269

)(0

.270

)(0

.315

)(0

.316

)0.

310

1.36

40.

271

1.31

10.

339

1.40

30.

383

Fem

ale

1.46

6(0

.274

)(0

.275

)(0

.316

)(0

.318

)�

0.01

40.

987

�0.

015

0.98

6�

0.00

3A

ge0.

997

0.00

01.

000

(0.0

09)

(0.0

09)

(0.0

01)

(0.0

12)

Rac

e(w

hite

�re

f.)

�0.

861

0.42

3�

0.87

4*0.

417

Bla

ck�

0.47

20.

623

�0.

397

0.67

2(0

.445

)(0

.446

)(0

.508

)(0

.509

)�

1.50

90.

221

Oth

er�

1.50

40.

222

�1.

041

0.35

3�

0.99

90.

368

(1.0

39)

(1.0

39)

(1.0

58)

(1.0

58)

0.25

31.

288

0.24

91.

283

0.14

8N

ativ

e-bo

rn1.

159

0.15

71.

170

(0.5

46)

(0.5

47)

(0.6

40)

(0.6

39)

Urb

an0.

847

2.33

30.

875

2.39

90.

783

2.18

80.

880

2.41

1(0

.531

)(0

.532

)(0

.620

)(0

.623

)R

egio

n(W

est�

ref.

)N

orth

-eas

t�

0.07

60.

927

�0.

072

0.93

00.

209

1.23

20.

267

1.30

6(0

.419

)(0

.420

)(0

.481

)(0

.485

)M

id-w

est

�0.

047

0.95

4�

0.05

70.

944

0.02

11.

021

0.08

51.

089

(0.4

02)

(0.4

03)

(0.4

57)

(0.4

61)

�0.

014

0.98

6�

0.04

20.

958

Sou

th0.

103

1.10

90.

129

1.13

7(0

.375

)(0

.377

)(0

.421

)(0

.423

)

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 17: Who chooses non‐public schools for their children?

246 P. Q. Yang & N. KayaardiT

able

4.—

cont

inue

d

Pre

dict

ors

Mod

el1

Mod

el2

Mod

el3

Mod

el4

Bex

p(B

)B

exp(

B)

Bex

p(B

)B

exp(

B)

Rel

igio

nC

hris

tian

0.69

32.

000.

947

2.57

70.

910

2.48

6(0

.477

)(0

.607

)(0

.610

)S

ocio

-eco

nom

icst

atus

Edu

cati

on0.

087

1.09

10.

092

1.09

6(0

.067

)(0

.067

)O

ccup

atio

nal

pres

tige

0.01

51.

015

0.01

51.

015

(0.0

14)

(0.0

14)

Fam

ilyin

com

e0.

011

1.01

1�

0.02

10.

979

(0.0

38)

(0.0

42)

0.02

81.

028

0.00

4C

lass

1.00

4(0

.277

)(0

.277

)F

amily

stru

ctur

eIn

tact

fam

ily0.

637

1.89

0(0

.400

)N

umbe

rof

child

ren

�0.

078

0.92

5(0

.127

)�

3.22

6***

0.04

0C

onst

ant

�3.

773*

**0.

023

�6.

721*

**0.

001

�6.

759*

**0.

001

(0.8

86)

(0.9

69)

(1.4

86)

(1.5

31)

�2

log

likel

ihoo

d52

3.9

521.

139

6.7

393.

833

.936

.539

.842

.6M

odel

�2

1011

df15

172,

205

N2,

201

1,82

21,

820

Not

es:

The

Bis

the

logi

stic

regr

essi

onco

effic

ient

;ex

p(B

)or

odds

rati

ois

the

anti

log

ofB

;an

dst

anda

rder

rors

are

inpa

rent

hese

s.*p

�0.

05**

p�

0.01

***p

�0.

001.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 18: Who chooses non‐public schools for their children?

Who chooses non-public schools? 247

socio-economic status and who have more children are more likely than theirrespective counterparts to send their children to religious schools against publicschools. Parents in the North-east and Mid-west appear to show a greater tendencyto select a religious school than those in the West. Nevertheless, there are nosignificant racial, gender, urban/rural and family-type differences in parental choiceof religious schools.

Another important finding is that, consistent with prior studies, parental socio-economic status is a key factor that determines the choice of non-religious privateschools. Since students in private schools are selective towards the higher end of thesocio-economic spectrum, one can expect that their school performance is onaverage better than that of students in public schools because of the former’s higherstarting-point and more resources. However, their higher level of academic perform-ance may not necessarily be the outcome of private school education (Goldhaber,1997).

One of the key concerns about school choice is that it will increase racial and classsegregation in school (Henig, 1994; Smith & Meier, 1995; Ascher et al., 1996). Wefound significant class differences but insignificant racial differentials in parentalschool choice. Our findings suggest that the concern of increased class segregationas a result of school choice may be warranted but the concern of increased racialsegregation is not.

We found no significant differences in demographic, religious, socio-economicand family-structure characteristics with regard to which parents are more likely tohome-school their children. Our finding is, by and large, tentative, and may be inpart a result of a small number of home-schooling parents in our sample. We didfind consistently, however, that all else being equal, parents in the year 2000 wereless likely to send their children to home schools than their 1998 counterparts. Aswe know, the estimated number of home-schoolers in the USA increased rapidly inthe 1990s from 250 000 to 350 000 children in the 1990/91 school year to possiblyone million children in the 1997/98 school year (Lines, 1999). With the 2000 dataincluded, our result could be a signal of fluctuating or cooling down in home-school-ing. As a relatively new phenomenon, home-schooling calls for more research in thefuture.

Note

An earlier version of this paper was presented at the Ninety-eighth Annual Meeting of theAmerican Sociological Association, Atlanta, USA, August 2003. We thank the panel participantsfor their constructive comments.

References

Ascher, C., Fruchter, N. & Berne, R. (1996) Hard lessons: public schools and privatization (NewYork, Twentieth Century Fund Press).

Bauch, P. (1988) Is parental involvement different in private schools? Educational Horizons, 66,78–82.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 19: Who chooses non‐public schools for their children?

248 P. Q. Yang & N. Kayaardi

Bauch, P. (1991) Linking parents’ reasons for choice and involvement in inner-city Catholic highschools, International Journal of Educational Research, 15, 311–322.

Bauch, P. & Goldring, E. (1995) Parental involvement and school responsiveness: facilitating thehome–school connection in schools of choice, Educational Evaluation and Policy Analysis, 17,1–21.

Cain, G. & Goldberger, A. (1983) Public and private schools revisited, Sociology of Education, 56,208–218.

Coleman, J. & Hoffer, T. (1987) Public and private high schools (New York, Basic Books).Coleman, J., Hoffer, T. & Kilgore, S. (1982) High school achievement: public, catholic, and private

schools compared (New York, Basic Books).Cooper, B. (1991) Parent choice and school involvement: perspectives and dilemmas in the

United States and Great Britain, International Journal of Educational Research, 15, 235–250.Crawford, J. & Freeman, S. (1996) Why parents choose private schooling: implications for public

school programs and information campaigns, ERS Spectrum, 14, 9–16.Godwin, K., Kemerer, F., Martinez, V. & Ruderman, R. (1998) Liberal equity in education: a

comparison of school choice options, Social Science Quarterly, 79, 502–522.Goldberger, A. & Cain, G. (1982) The causal analysis of cognitive outcomes in the Coleman,

Hoffer and Kilgore report, Sociology of Education, 55, 103–122.Goldhaber, D. (1996) Public and private high schools: is school choice an answer to the

productivity problem? Economics of Education Review, 15, 93–109.Goldhaber, D. (1997) School choice as education reform, Phi Delta Kappan, 79, 143–147.Goldhaber, D. (1999) School choice: an examination of the empirical evidence on achievement,

parental decision making, and equity, Education Researcher, 28, 16–25.Hamilton, B. & Macauley, M. (1991) Determinants and consequences of the private–public

school choice, Journal of Urban Economics, 29, 282–294.Henig, J. (1994) Rethinking school choice: limits of the market metaphor (Princeton, NJ, Princeton

University Press).Lankford, H. & Wyckoff, J. (1992) Primary and secondary school choice among public and

religious alternatives, Economics of Education Review, 11, 317–337.Lines, P. (1999) Homeschoolers: estimating numbers and growth (Washington, DC, National

Institute on Student Achievement, Curriculum, and Assessment, Office of EducationalResearch and Improvement, US Department of Education).

Long, J. & Toma, E. (1988) The determinants of private school attendance, 1970–1980, EconomicStatistics, 70, 351–356.

Maddaus, J. (1990) Parent choice of school: what parents want and do, Review of EducationResearch, 16, 267–296.

Martinez, V., Godwin, K. & Kemerer, F. (1996) Public school choice in San Antonio: whochooses and with what effects?, in: B. Fuller and R. F. Elmore (Eds) Who chooses? who loses?:culture, institutions, and the unequal effects of school choice (New York, Teachers CollegePress), 50–69.

National Opinion Research Center (NORC) (2001) General social surveys, 1972–2000: cumulativecodebook (Chicago, IL, University of Chicago, NORC).

Noell, J. (1982) Public and Catholic schools: a reanalysis of ‘public and private schools’, Sociologyof Education, 55, 123–132.

Noell, J. & Myers, D. (1983) The demand for private schooling: an empirical analysis (WorkingPaper) (Washington, DC, Office of Planning, Budget, and Evaluation, US Department ofEducation).

Schneider, M., Marschall, M., Teske P. & Rosch, C. (1998) School choice and cultural wars inthe classroom: what different parents seek from education, Social Science Quarterly, 79,489–501.

Schneider, B., Schiller, K. & Coleman, J. (1996) Public school choice: some evidence from thenational education longitudinal study of 1988, Education Evaluation and Policy Analysis, 18,19–29.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 20: Who chooses non‐public schools for their children?

Who chooses non-public schools? 249

Smith, K. & Meier, K. (1995) The case against school choice: politics markets, and fools (Armonk,NY, ME Sharpe).

Weiss, J. (1998) Policy theories of school choice, Social Science Quarterly, 79, 523–532.West, E. & Palsson, H. (1988) Parental choice of school characteristics: estimation using

state-wide data, Economic Inquiry, 26, 725–740.

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14

Page 21: Who chooses non‐public schools for their children?

Dow

nloa

ded

by [

Uni

vers

ity o

f N

orth

Tex

as]

at 1

5:14

21

Nov

embe

r 20

14