HOMESCHOOLING IN THE UNITED STATES:

REVELATION OR REVOLUTION?

Jessica S. Howell
California State University, Sacramento

Michelle E. Sheran
University of North Carolina at Greensboro

September 2008

Abstract. The number of homeschooled children grew by 7-15 percent each year during the 1990s with
the result that today as many as 2 million children are currently being educated at home in the United
States. Despite this dramatic increase, homeschooling has received surprisingly little attention from
economists. In this paper, we briefly discuss homeschooling and the state regulations that govern this
educational alternative. We further discuss the available data on homeschooling and utilize the National
Household Education Survey data to examine the determinants of a family’s decision to homeschool.
Preliminary results reveal that maternal employment, parental education, race, and the number and age of
siblings play significant roles in a family’s decision to homeschool. We also take a first step toward
understanding the potential impact of homeschooling on public school resources by analyzing the
decision to homeschool part-time versus full-time. We find that part-time homeschoolers more closely
resemble non-homeschooling families in several important ways and that part-time homeschooling is
much more popular in Western states than elsewhere in the U.S. Finally, we utilize data from the General
Social Survey to examine how attitudinal variables (e.g., political affiliation, participation in organized
religion, views on homosexuality) impact the homeschooling decision.

JEL Classification: I21, J22

1. INTRODUCTION

Recent legislated education reform known as the No Child Left Behind Act of 2001 was

designed to support accountability in public schools and expand school choice for those students

in public schools that fail to meet established academic standards. As many school districts

discover that they lack the funding necessary to meet their state’s accountability standards, our

attention is drawn to the increasing number of parents who are taking the education of their

children into their own hands. State-level data indicate that the number of homeschooled

children grew by 7-15 percent each year during the 1990s resulting in an estimate of as many as

2 million children currently being educated at home in the United States (Lines (1999), Ray

(2003)). This estimate of the number of homeschoolers amounts to approximately four percent

of all school-aged children and is nearly three times as large as the number of kids currently

enrolled in charter schools in the U.S.1

Homeschooling has received little attention from economists.2 The incredible growth in

homeschooling is reason enough to take a closer look at this often-ignored schooling alternative,

but there are additional factors which indicate that further research on homeschooling is

warranted. First, there is evidence that the growth in homeschooling will continue. In a 2003

survey of adults who were themselves homeschooled, 82 percent revealed that they would

homeschool their own children and, of those surveyed, 75 percent were already homeschooling

their own kids (Ray, 2004). Second, the notion that homeschooling families pay taxes to support

public schools but do not utilize any school resources is somewhat of a myth. Survey data

indicate that approximately 20-25 percent of homeschooling families have access to curriculum,

books, and other materials provided by their local public school, as well as the opportunity to

1
attend classes and participate in extracurricular activities.3 Third, there are a variety of public

finance issues surrounding increasingly popular partnerships between home educators and

traditional public schools. Some public schools collect small amounts of state public education

funding for providing ancillary services to homeschoolers (like testing) while others collect as

much as they are typically allotted by the state for a full-time-equivalent student. Many public

schools (especially charters) have created ‘independent study programs’ that are tailored

specifically to the homeschooling population. Participation in these programs technically

requires a student to be enrolled, which enables the school to count that student as full-time for

the purpose of determining the school’s receipt of state funding, but all or nearly all education

still occurs in the home.4 Finally, homeschooling represents a more dramatic departure from the

traditional education system than other innovations like charter schools and voucher programs.

If we believe that charters and vouchers have influenced the public provision of education in this

country, then we might logically expect homeschooling to have an even more dramatic influence

on the traditional education system.

This paper examines the determinants of a family’s decision to homeschool using several

datasets that contain information on this decision that is rarely included in nationally

representative survey data. The results indicate that maternal employment, parental education,

race, number and age of siblings, and a variety of attitudinal variables play significant roles in a

family’s choice to homeschool. We also take a first step toward understanding the potential

impact of homeschooling on public school resources by analyzing the decision to homeschool

1
The Center for Education Reform currently estimates that nearly 1.2 million students are enrolled in charter
schools (Source: http://www.edreform.com/index.cfm?fuseAction=stateStatChart&psectionid=15&cSectionID=44).
2
Notable exceptions include Belfield (2004), Houston and Toma (2003), and Isenberg (2002).
3
Parent Survey of the National Household Education Surveys Program (National Center for Education Statistics,
U.S. Department of Education, 1999).
4
See Lines (2000) for an in-depth study of public school programs geared toward homeschoolers in Washington.

2
part-time versus full-time. The paper is organized as follows. In the next section, we briefly

discuss what it means to home school and the state regulations that govern this educational

alternative. In Section 3, we discuss the datasets utilized to examine the decision to homeschool

and present our findings on the determinants of the homeschooling choice. Section 4 concludes

by discussing the implications of our findings for education policy.

2. HOMESCHOOLING LEGISLATION

Homeschooling is identified where a child is being educated at home and where any

schooling outside of the home does not exceed 25 hours per week. This definition allows

children who attend a traditional school three days out of the week to be classified as

homeschooled. Note also that the definition of homeschooling does not require that the child’s

parent be the individual providing the instruction, allowing for the possibility that the teacher

could be a guardian, another child’s parent, or a credentialed individual unrelated to the child

such as a tutor.

Most homeschooling rights were established in individual states during the 1980s, and

home-based education was practiced legally in all states by 1994. There are a variety of ways to

legally homeschool and, since these rules vary slightly from state to state, we discuss the state of

California as an example.

There are three ways for parents to legally homeschool a child in California: (1) establish

their own private school by filing an affidavit with the California Department of Education, (2)

enroll the child in an independent study program that is run through a public/charter school or

existing private school, or (3) tutor the child by obtaining teaching credentials in all

3
subjects/levels taught or by hiring an individual who is appropriately credentialed in the state.5

The only additional requirement, which is common to all states, is that homeschooling families

keep attendance records so that they are able to prove that they are in compliance with

compulsory attendance laws. In California, there are no other requirements regarding student

progress (e.g., standardized testing or receipt of a GED).

California is considered a low-regulation state when it comes to homeschooling rules.

The primary distinction between states with no regulation and those with low regulation is the

process of notifying the proper authorities of the intention to homeschool. In a ‘no regulation’

state, like Texas, parents are able to withdraw a student from the public school system and

simply write a letter to the principal indicating that the child will be homeschooled. Low

regulation states, like California, have a more formal notification process that involves filing an

annual certificate of enrollment with the state department of education States with moderate

regulation add to this notification procedure a requirement to administer standardized tests or

have the student’s academic progress otherwise evaluated by a professional. There is additional

variation across ‘moderate regulation’ states in the frequency with which students are to be

tested. In North Carolina, for example, homeschooled students are required to be tested in

grammar, reading, spelling, and math every year. By contrast, homeschooled students in

neighboring Tennessee are required only to take a standardized achievement test in grades 5, 7,

and 9. Finally, high regulation states have additional requirements that range from state approval

of the curriculum (i.e., Pennsylvania) to home visits (i.e., New York).

5
State-specific rules for legally homeschooling are discussed in detail by each state’s homeschooling association. A

4
3. HOMESCHOOLING DATA

The small number of quantitative studies on the topic of homeschooling is likely due to

the dearth of data on this segment of the population. The lack of data might be explained in

several ways. First, state- or district-level data requires a certain amount of coordination

between school districts and state and local education agencies. In practice, many school

districts have no systematic mechanism for tracking homeschooled students over time because

state law does not require them to do so. Any data that are available through these channels are

aggregate measures of the number of homeschoolers in a district or state and, therefore, do not

permit analyses based on observed individual choices.6 Second, anecdotal evidence suggests that

homeschooling families as a group might be less tolerant than non-homeschooling families of

questions and surveys that they view as intrusive. A 1996 survey of homeschooling families

administered by National Home Education Research Institute (NHERI) had only a 28% response

rate (Ray (1997)) despite the fact that NHERI and the survey were endorsed by the national

homeschooling advocacy organization, the Home School Legal Defense Association (HSLDA).

There are several nationally-representative datasets that include questions on all

schooling choices made by households, including homeschooling. We utilize two of these

datasets in our analysis, the National Household Education Survey (NHES) and the General

Social Survey (GSS). Both sources are used because they collect different types of information

on family characteristics and social views. Each is discussed below, along with the results of our

analyses.

complete list of these state associations (with links) is available at www.teachinghome.com/states/states.cfm.

5
3.1 HOMESCHOOLING IN THE NATIONAL HOUSEHOLD EDUCATION SURVEY (NHES)

The Parent and Family Involvement portion of the National Household Education Survey

(NHES) began asking parents about homeschooling beginning in 1996. The survey asks parents

why they chose to homeschool as well as information about the availability and utilization of

public school resources. This survey was repeated in 1999 and 2003 (although it is not a panel).

Table 1 compares the characteristics of homeschooling families over these three years (1996,

1999, and 2003). Approximately 250-300 homeschoolers were identified in each survey year,

corresponding to 1.4 percent of the 1996 sample, 1.7 percent of the 1999 sample, and 2.0 percent

of the 2003 sample.

Note that variable definitions and questions were constant over all three years with only a

few exceptions, so it is possible to inspect Table 1 for trends in the attributes of homeschoolers.

Black and Hispanic families gained representation in the sample of homeschoolers between 1996

and 1999, but this increase in diversity diminished somewhat between 1999 and 2003. Other

sources indicate that minority representation among homeschoolers has continued to rise over

time, particularly in areas with large minority populations and underperforming schools (Marech,

2006). The left tail of the age distribution of homeschooled children appears to have changed

over this period, a trend that is also somewhat evident among non-homeschooling families.

Another trend apparent in Table 1 that is mirrored among non-homeschoolers is the increase in

disabilities that impede a child’s learning. The question underlying this variable asks whether or

not the child has a disability that affects his/her ability to learn, so the increase might simply

reflect higher rates of diagnosis of learning disabilities among all children over this time period.

Table 1 also indicates that between 1996 and 2003 the percentage of homeschooling moms that

worked part-time fell and the percentage that were not in the labor force at all increased.

6
Houston and Toma (2003) and Isenberg (2002) are examples of these kinds of studies.

6
Homeschooling parents were also more likely to have at least a college education in 2003

compared to 1996. Finally, there appears to be a trend toward increased representation in the

Southern states and decreased representation in the Western states.

In Table 2, we pool the three years of NHES cross-sectional data discussed above and

examine how homeschooling families compare to their non-homeschooling counterparts.7 The

sample of homeschoolers looks similar to the sample of non-homeschoolers with regard to

gender and the distribution of children’s ages, however, that is where the similarities end.

Homeschoolers are significantly less racially diverse and more likely to be two-parent families

with more children than non-homeschoolers. Not surprisingly, 54.5 percent of mothers in

homeschooling families are not in the labor force, compared with only 22.8 percent of mothers in

non-homeschooling families. The lower mean income among homeschooling families likely

reflects this smaller labor force participation, especially since similar percentages of

homeschoolers and non-homeschoolers own their homes. Those parents who choose

homeschooling have more education on average and are more likely to live in Southern and

Western states compared to those who choose public and private school. Finally, a higher

proportion of homeschoolers than non-homeschoolers live in rural areas.8

Parents surveyed in the NHES were also asked about their reasons for homeschooling

their children. Parents were provided with a list of potential reasons and asked to check all

reasons that applied. Table 3 identifies the proportion of parents who checked various reasons.

In all three years of the NHES survey, concerns about the other schooling alternatives available

7
Tables 2a, 2b, and 2c compare homeschoolers to non-homeschoolers by year, rather than in the pooled cross
section.
8
The urbanicity definitions changed slightly in 1999, affecting the ‘urban’ and ‘suburban’ categories. The
definition of ‘rural’ was maintained over all three sample years.

7
were cited by the most parents.9 Three-quarters of parents voiced concerns about their children’s

schools, where these concerns included objecting to what was taught in schools, feeling that their

children were not being challenged in school or that parents could provide a better education for

their children at home, as well as fears about the school learning environment. The next most

popular reason for homeschooling, cited by 47 percent of families in the pooled cross section,

was to provide religious or moral instruction. Table 3 also indicates that religion dramatically

gained in popularity as a reason for homeschooling over the three survey years.

There is a fair amount of overlap in these two most popular reasons because parents were

permitted to cite multiple reasons. In 1996, 23 percent of parents checked both ‘school reasons’

and ‘religious reasons’ for homeschooling. In 1999 and 2003, the percentages citing both

reasons were 29 and 69 percent, respectively. The 2003 survey also asks a related follow-up

question in which parents were asked to choose one of the reasons that they cited as the most

important reason. Of the parents citing ‘religious reasons’ as one of the reasons for

homeschooling, 42 percent chose religion as the most important reason. Similarly, of the parents

citing ‘school reasons’ as one of the reasons for homeschooling, 47 percent chose school as the

most important reason.

3.2 DETERMINANTS OF THE DECISION TO HOMESCHOOL USING THE NHES

Next we explore the determinants of the decision to homeschool. Define y it* as the

propensity of family i to homeschool in year t and Xit as a vector of demographic characteristics

describing family i in year t. The homeschooling decision can then be described by the

relationship y it* = X it β + ε it , where ε it is a randomly distributed error. Because we cannot

9
The list of possible reasons was not constant over the three survey years, so questions were categorized into the

8
observe y it* and instead observe the binary outcome, y it = 1( y it* > 0) , the probability that family i

chooses to homeschool in year t is given by

( )
Pr y it* > 0 = Pr ( y it = 1) = Pr ( X it β + ε it > 0) . We assume that ε is distributed iid logistic and

estimate the equation y it = X it β + ε it with logistic regression.

Table 4 presents the estimated marginal effect of various family characteristics on the

probability of choosing to homeschool. The marginal effects are based on estimates from a

logistic regression, which are not reported. The marginal effects in Table 4 are interpreted as the

change in the probability of homeschooling associated with a discrete change in a binary variable

or a one standard deviation change in a continuous variable. For example, the first cell in Table

4 indicates that the probability that a white family homeschools in 1996 is 0.6 percentage points

larger than a non-white family after controlling for differences in all other demographic

characteristics. The marginal effect of being white diminishes in 1999 and 2003, although the

latter is not statistically significant, reflecting the increased racial diversity among

homeschooling families that was evident in Table 1. Table 4 also reveals that the probability of

homeschooling is 1.5 percentage points higher in families in which the mother is not in the labor

force and 0.9 percentage points higher in families in which the mother works part-time compared

to families in which the mother works full-time.10 Results further indicate that the probability of

homeschooling is increasing in parental education. Lastly, while the probability of

homeschooling is increasing in the number of siblings in a family, it is decreasing in the number

of children under the age of 6, perhaps indicating that young children are more time intensive for

parents.

five groups presented in Table 3.

10
Maternal labor supply decisions are endogenously determined, but typical corrections for this endogeneity are
limited by data availability.

9
3.3 FULL-TIME VERSUS PART-TIME HOMESCHOOLING USING THE NHES

In the previous section, we analyzed the extensive margin decision made by parents

regarding the type of schooling they choose for their children. Next we examine a decision on

the intensive margin; whether families who opt to homeschool do so on a full-time or part-time

basis. The following analysis of the determinants of the full-time vs. part-time homeschooling

decision is a first step in understanding the potential impact of homeschooling on public school

resources and illuminating discussions about collaborative programs between home educators

and traditional schools.

The 1999 and 2003 NHES surveys inquire about whether homeschooled children receive

all of their instruction at home or if they get some instruction at a traditional school. Of the

sample of 557 homeschoolers in these years, approximately 21 percent choose to homeschool

part-time. Part-time homeschooling refers to attendance at a traditional school for any positive

number of hours. In this sample, 75 percent of part-time homeschooled children attend a

traditional school for ten or fewer hours each week. Summary statistics for both full-time and

part-time homeschoolers are provided in Table 5.

According to Table 5, part-time homeschoolers are more racially diverse than their full-

time counterparts, with even greater minority representation than the sample of non-

homeschoolers in Table 2. Part-time homeschoolers also look more like non-homeschoolers

with regard to family size and, to some extent, family type. Part-time homeschooling families

are also more likely than full-time homeschooling families to have a mother in the labor force

and a child with a disability. Families who homeschool full-time are more likely to have

college-educated parents and live in both rural areas and the Southern states. In contrast, part-

time homeschooling appears to be very popular in the West. This last result is consistent with

10
data from the Center for Education Reform (2004) that indicates that Arizona, California, and

Oregon have charter school legislation that provides considerable latitude to schools. This

makes charter schools a popular option in those states (approximately one third of all enrolled

charter school students in the U.S. reside in these three Western states), including popularity

within the homeschooling community for the independent study programs widely available

through charters.

Using the logit model introduced in the previous section, we now explore the

determinants of the decision to homeschool full-time versus part-time. Table 6 presents the

estimated marginal effect of various family characteristics on the probability of choosing to

homeschool full-time. The patterns that were evident in Table 5 also appear in the results in

Table 6. For example, the probability that a white family chooses to homeschool full-time is

10.3 percentage points higher than a non-white family. A one-standard deviation increase in the

number of siblings increases the probability of full-time homeschooling by 2.9 percentage points.

The largest marginal effect is associated with living in a Western state, which decreases the

probability of homeschooling full-time relative to part-time by 11.4 percentage points. Based on

the year dummy for 2003, there doesn’t appear to be a statistically significant increase in full-

time versus part-time homeschooling between 1999 and 2003. Finally, this model does a good

job of predicting the proportion of families who will choose to homeschool full-time. The

predicted proportion, 81.9 percent is only slightly higher than the 79.2 percent who do

homeschool full-time in the NHES sample.

We also examined parents’ stated reasons for choosing to homeschool with regard to the

full-time decision. The only noticeable difference is in the prevalence of citing religion as a

factor or as the most important factor. Among full-time homeschoolers, religious reasons were

11
cited by 56.2 percent, compared to 39.7 percent of part-time homeschooling families.

Additionally, religion was cited as the most important factor by 31.4 percent of full-time families

and only 22.4 percent of part-time families. Despite the fact that these differences are

statistically significant, adding a dummy variable for citing religious reasons to the logistic

regression discussed above yielded a positive but insignificant coefficient and left the other

marginal effects essentially unchanged.

3.4 HOMESCHOOLING IN THE GENERAL SOCIAL SURVEY (GSS)

We next analyze the decision to homeschool using the General Social Survey, which is

administered by the National Opinion Research Center (NORC) at the University of Chicago.

Questions on homeschooling on the 1998 and 2000 surveys allow us to explore how attitudinal

variables (e.g., political affiliation, participation in organized religion, views on homosexuality,

abortion, gun control, and the quality of political and religious leadership) vary over school

choices. Table 7 compares various characteristics of homeschooling families to those of families

who choose to send their children to one of the following three alternatives: public school, non-

religious private school, and religious private school.

A number of interesting differences emerge. Parents who homeschool are more likely

married than parents who send their children to public school and, on average, are younger than

parents who send their children to private school. Moreover, homeschooling families are more

likely white than public school families. On average, homeschooling families report lower

incomes than private school families. Table 7 further demonstrates that parents who homeschool

are, on average, more educated than parents who send their children to public school but less

educated than parents who send their children to private school. We also find that compared to

12
families that send their children to private schools, homeschooling families are more likely to

live in the Midwest and less likely to live in the South. Lastly, the average homeschooling family

has more household members compared to the average public school family.

The contemporary image of homeschooling often depicts such families as deeply

religious, socially conservative, and anti-government. Because the GSS asks a large number of

attitudinal questions, we are able to explore these stereotypes in Tables 8-12. Table 8 presents

various measures of religiosity for families that homeschool compared to families that send their

children to public school, secular private school (hereafter simply called private school), and

private religious school. The table demonstrates that compared to public and private school

families, homeschoolers are less likely Protestant, more likely Catholic, and less likely no

religion. Compared to families that send their children to private religious schools,

homeschoolers are more likely Protestant, less likely Catholic, and more likely no religion. Table

8 further reveals that homeschoolers are least likely to never attend religious services and most

likely to attend more than once a week. Interestingly, homeschoolers are more likely than public

school families but less likely than private and private religious school families to self report

strong religious beliefs. Lastly, they are most likely to believe that the Bible is the actual word of

God.

We next explore how the political beliefs of families with children older than 5 years of

age vary across school choices. As shown in Table 9, parents who homeschool are less likely

Democrat and more likely Republican compared to all other groups. Based on self-reporting,

they are also more conservative than parents who send their children to public and private school

but more liberal than parents who send their children to private religious schools.

13
 Table 10 reveals how attitudes on government spending vary across the four possible

school choices. Not surprisingly, the table demonstrates that homeschoolers are generally

against big government. They are most likely to think that they pay too much in federal income

taxes, and they are least likely to think that the government spends too little on national defense

and crime. Homeschooling families are also less likely than public and private schoolers to think

that the government spends too little on welfare.

The result for education is confounded. Homeschoolers may value education more than

most and, thus, think that the government should spend more on education. Additionally, they

may homeschool because they think the public school system is inadequate and, thus, think that

that the government should spend more on education. On the other hand, they may think the

government should spend less overall, including on education. The data show that

homeschoolers are less likely than private school families but more likely than public and private

religious school families to think that the government spends too little on education.

Table 11 demonstrates that homeschoolers have more conservative views on a variety of

social issues compared to private school families. However, we observe mixed results when

comparing homeschoolers to public and private religious school families. Results do indicate,

though, that the families that homeschool are substantially more likely to be against sex

education in public schools, more likely to be against premarital sex, and more likely to support

harsher divorce laws. Given that these views are family related, we conclude that families that

homeschool have more conservative family values.

Finally, Table 12 shows that homeschoolers have more conservative views on parenting

and think that parents should have authority over their children. Homeschoolers are more likely

than both private school families and private religious school families to be in favor of spanking

14
as a form of discipline. Homeschoolers are also more likely than public and private schoolers to

be against making birth control available to children. In addition, families that homeschool are

most likely to think that working mothers are inferior to stay-at-home mothers. Homeschooling

families are also most likely against sex education in public schools and are most likely to think

that the most desirable quality in a child is that he or she obeys the parents.

3.5 DETERMINANTS OF THE DECISION TO HOMESCHOOL USING THE GSS

We next employ the GSS data to explore the determinants of the decision to homeschool

using a logit model. Results are similar to those presented in Table 4 using the NHES data. Table

13 shows that the probability that a white family homeschools is 0.7 percentage points larger

than a non-white family after controlling for differences in the other demographic characteristics.

Results further indicate that the probability of homeschooling is 0.8 percentage points higher for

mothers who have at least a college education. Lastly, while the probability of homeschooling is

increasing in the number of children in a family, it is decreasing in the number of children under

the age of 6. This result perhaps reflects that young children are more time intensive for parents.

Table 14 explores how the decision to homeschool varies across attitudinal variables. The

results indicate that the probability of homeschooling increases by a relatively large 1.8

percentage points if a parent believes that sex education should not be taught in public schools.

The table also shows that married parents have a higher probability of homeschooling than their

single counterparts. Results further reveal that the probability of homeschooling is increasing in

the frequency with which a family attends religious services. Lastly, we find that parents who

believe that they pay too much in federal income taxes are more likely to homeschool.

15
4. DISCUSSION AND CONCLUSION

Our analysis of the decision to homeschool using two nationally representative datasets

bears out some stereotypical images of homeschooling families, but also reveals a few interesting

surprises. Examining attitudinal information in the General Social Survey shows that

homeschooling families generally have more conservative views of religion, family values, and

political ideology and affiliation. Analyzing homeschooling families over time using the

National Household Education Survey, however, reveals interesting changes in various

demographic characteristics of homeschoolers. In particular, the pool of homeschooling families

has become more racially diverse and differently concentrated across regions of the U.S.

Our examination of the decision to homeschool part-time versus full-time yields some

evidence that part-time homeschooling has gained popularity in the same areas of the country

where the availability and regulatory latitude of charter schools has also grown. Additionally,

part-time homeschooling appears to be more widespread in urban areas, where there are

presumably plenty of schooling options available to families. Thus, another stereotype of

homeschoolers as rural families with few school alternatives geographically nearby appears not

to hold among those who choose to homeschool part-time.

Although data on the homeschooling population is quite scarce, the determinants of the

decision to homeschool documented here indicate that the families choosing this schooling

alternative perhaps deserve more attention than their small numbers have historically garnered.

In this day of increased school accountability and parental knowledge of numerous measures of

school quality, continued growth in the homeschooling population and in the charter school

movement may result in dramatic changes to the pool of students who choose a traditional public

school. The effects of such a shift have implications for school funding, peer effects, and

16
numerous other topics of interest to educational researchers, practitioners, and public policy

makers. Our hope is that more and better data are collected on this growing population and the

novel ways in which these families interact with new schooling alternatives that are now widely

available across the United States.

17
REFERENCES
Bauman, Kurt J. (2002). “Home Schooling in the United States: Trends and Characteristics.”
Education Policy Analysis Archives. 10(26). (http://epaa.asu.edu/epaa/v10n26.html)

Belfield, Clive R. (2004). “Home-Schooling in the US.” National Center for the Study of
Privatization in Education, Occasional Paper No. 88.

Center for Education Reform. (2004). Charter School Laws Across the States: Ranking and
Scorecard, eighth ed. Washington, DC: Center for Education Reform.

Houston, Robert G. and Eugeneia F. Toma (2003). “Home Schooling: An Alternative School
Choice.” Southern Economic Journal. 69(4): 920-935.

Isenberg, Eric (2002). “Home Schooling: School Choice and Women’s Time Use.”
Unpublished Ph.D. dissertation, Washington University, Department of Economics.

Lines, Patricia M. (1999). “Homeschoolers: Estimating Numbers and Growth.” Web Edition.
Washington DC: National Institute on Student Achievement, Curriculum, and Assessment,
Office of Educational Research and Improvement, U.S. Department of Education.

Lines, Patricia M. (2000). “When Home Schoolers Go to School: A Partnership Between

Families and Schools.” Peabody Journal of Education. 75(1&2), 159-186.

Marech, Rona. (2006). “Homeschooling Draws More Blacks.” The Baltimore Sun, January 23,
1.A.

Ray, Brian D. (1997). Strengths of Their Own. Oregon: National Home Education Research
Institute Publications.

Ray, Brian D. (2003). “Facts on Homeschooling.” National Home Education Research Institute
website (http://www.nheri.org/modules.php?name=NHERI_Research).

Ray, Brian D. (2004). Home Educated and Now Adults: Their Community and Civic
Involvement, Views About Homeschooling, and Other Traits. Oregon: National Home
Education Research Institute Publications.

18
 Table 1: Characteristics of Homeschoolers in the NHES

1996 1999 2003

Male 47.0 49.8 51.4
Female 53.0 50.2 48.6
White 83.0 72.1 77.9
Black 2.4 8.0 7.2
Hispanic 9.7 13.9 8.4
Other race/ethnicity 4.9 6.0 6.4
Child’s age: 5 and under 11.7 14.6 7.2
Child’s age: 6 - 7 9.7 11.3 13.3
Child’s age: 8 - 10 21.9 21.9 18.1
Child’s age: 11 - 14 32.4 28.2 30.9
Child’s age: 15 - 18 24.3 23.9 30.5
Number of siblings: 0 17.0 20.3 12.9
Number of siblings: 1 32.0 31.6 34.9
Number of siblings: 2 27.5 24.6 23.3
Number of siblings: 3 13.4 16.6 16.9
Number of siblings: 4 2.8 3.3 6.0
Number of siblings: 5+ 7.3 3.6 6.0
Family type: 2 parents 83.0 81.4 83.9
Family type: 1 parent 14.2 16.6 14.9
Family type: other 2.8 2.0 1.2
Mom works full-time 17.8 20.3 18.1
Mom works part-time 25.9 23.6 16.5
Mom not in LF 51.4 52.5 59.8
Mom looking for job or no mom in HH 4.9 3.6 5.6
English mom's 1st language 93.5 92.7 91.2
Child disability impedes learning 2.0 10.6 11.1
Owns home 68.4 72.8 75.5
Income $41,437 $48,081 $42,533
Parental Education: Less than HS 5.7 1.3 2.0
Parental Education: HS graduate 21.1 16.6 22.1
Parental Education: Some college 35.6 38.5 28.5
Parental Education: College 22.7 23.9 24.5
Parental Education: Advanced degree 15.0 19.6 22.9
Northeast 10.9 8.6 13.2
South 35.6 43.2 45.0
Midwest 21.5 18.9 19.3
West 32.0 29.2 22.5
Urban 23.9 55.8 57.8
Suburban 49.8 16.3 19.3
Rural 26.3 27.9 22.9
N 247 301 249
Percent of Sample 1.4% 1.7% 2.0%

19
 Table 2: NHES Pooled Cross Section (1996, 1999, 2003)

Homeschoolers Non-Homeschoolers
Male 49.4 51.2
Female 50.6 48.8
White 77.3 62.4
Black 6.0 14.5
Hispanic 10.9 17.5
Other race/ethnicity 5.8 5.6
Child’s age: 5 and under 11.4 7.1
Child’s age: 6 – 7 11.4 15.2
Child’s age: 8 – 10 20.7 23.5
Child’s age: 11 – 14 30.4 31.4
Child’s age: 15 – 18 26.1 22.9
Number of siblings: 0 16.9 20.5
Number of siblings: 1 32.7 43.9
Number of siblings: 2 25.1 24.4
Number of siblings: 3 15.7 7.9
Number of siblings: 4 4.0 2.3
Number of siblings: 5+ 5.6 1.0
Family type: 2 parents 82.7 70.8
Family type: 1 parent 15.3 26.0
Family type: other 2.0 3.2
Mom works full-time 18.8 49.4
Mom works part-time 22.1 19.8
Mom not in LF 54.5 22.8
Mom looking for job or no mom in HH 4.6 8.0
English mom's 1st language 92.5 79.9
Owns home 72.3 70.1
Income $44,529 $46,091
Parental Education: Less than HS 2.9 7.8
Parental Education: HS graduate 19.7 26.5
Parental Education: Some college 34.5 30.8
Parental Education: College 23.7 18.0
Parental Education: Advanced degree 19.2 17.0
Northeast 10.8 17.4
South 41.4 37.0
Midwest 19.8 21.2
West 28.0 24.4
Urban 46.6 54.0
Suburban 27.6 25.3
Rural 25.8 20.7
N 797 46754

20
 Table 2a: Summary Statistics, 1996 NHES

Homeschoolers Non-Homeschoolers
Male 47.0 51.6
Female 53.0 48.4
White 83.0 66.1
Black 2.4 14.2
Hispanic 9.7 13.9
Other race/ethnicity 4.9 5.7
Child’s age: 5 and under 11.7 7.5
Child’s age: 6 – 7 9.7 14.5
Child’s age: 8 – 10 21.9 23.8
Child’s age: 11 – 14 32.4 32.1
Child’s age: 15 – 18 24.3 22.0
Number of siblings: 0 17.0 21.6
Number of siblings: 1 32.0 43.0
Number of siblings: 2 27.5 24.3
Number of siblings: 3 13.4 7.6
Number of siblings: 4 2.8 2.4
Number of siblings: 5+ 7.3 1.1
Family type: 2 parents 83.0 72.2
Family type: 1 parent 14.2 24.6
Family type: other 2.8 3.2
Mom works full-time 17.8 50.1
Mom works part-time 25.9 20.2
Mom not in LF 51.4 22.3
Mom looking for job or no mom in HH 4.9 7.4
Child disability impedes learning 2.0 4.5
English mom's 1st language 93.5 84.1
Owns home 68.4 69.4
Income $41,437 $45,319
Parental Education: Less than HS 5.7 7.7
Parental Education: HS graduate 21.1 28.8
Parental Education: Some college 35.6 30.6
Parental Education: College 22.7 17.0
Parental Education: Advanced degree 15.0 15.8
Northeast 10.9 17.6
South 35.6 36.2
Midwest 21.5 21.9
West 32.0 24.2
Urban 23.9 28.2
Suburban 49.8 47.0
Rural 26.3 24.8
N 247 17442

21
 Table 2b: Summary Statistics, 1999 NHES

Homeschoolers Non-Homeschoolers
Male 49.8 51.0
Female 50.2 49.0
White 72.1 60.7
Black 8.0 15.6
Hispanic 13.9 18.5
Other race/ethnicity 6.0 5.2
Child’s age: 5 and under 14.6 7.9
Child’s age: 6 – 7 11.3 16.0
Child’s age: 8 – 10 21.9 23.5
Child’s age: 11 – 14 28.2 30.1
Child’s age: 15 – 18 23.9 22.6
Number of siblings: 0 20.3 21.8
Number of siblings: 1 31.6 43.4
Number of siblings: 2 24.6 24.0
Number of siblings: 3 16.6 7.6
Number of siblings: 4 3.3 2.2
Number of siblings: 5+ 3.6 1.0
Family type: 2 parents 81.4 68.4
Family type: 1 parent 16.6 28.3
Family type: other 2.0 3.3
Mom works full-time 20.3 49.9
Mom works part-time 23.6 19.4
Mom not in LF 52.5 22.7
Mom looking for job or no mom in HH 3.6 8.0
Child disability impedes learning 10.6 8.4
English mom's 1st language 92.7 78.7
Owns home 72.8 68.4
Income $48,081 $48,962
Parental Education: Less than HS 1.3 8.3
Parental Education: HS graduate 16.6 26.0
Parental Education: Some college 38.5 31.0
Parental Education: College 23.9 17.3
Parental Education: Advanced degree 19.6 17.4
Northeast 8.6 17.2
South 43.2 38.8
Midwest 18.9 20.1
West 29.2 23.9
Urban 55.8 66.8
Suburban 16.3 13.0
Rural 27.9 20.3
N 301 17380

22
 Table 2c: Summary Statistics, 2003 NHES

Homeschoolers Non-Homeschoolers
Male 51.4 51.2
Female 48.6 48.8
White 77.9 59.6
Black 7.2 13.3
Hispanic 8.4 21.1
Other race/ethnicity 6.4 6.0
Child’s age: 5 and under 7.2 2.3
Child’s age: 6 - 7 13.3 15.1
Child’s age: 8 - 10 18.1 22.8
Child’s age: 11 - 14 30.9 32.3
Child’s age: 15 - 18 30.5 24.6
Number of siblings: 0 12.9 17.0
Number of siblings: 1 34.9 45.9
Number of siblings: 2 23.3 25.3
Number of siblings: 3 16.9 8.6
Number of siblings: 4 6.0 2.2
Number of siblings: 5+ 6.0 1.0
Family type: 2 parents 83.9 72.5
Family type: 1 parent 14.9 24.5
Family type: other 1.2 3.0
Mom works full-time 18.1 47.6
Mom works part-time 16.5 19.6
Mom not in LF 59.8 23.8
Mom looking for job or no mom in HH 5.6 9.0
Child disability impedes learning 11.1 9.9
English mom's 1st language 91.2 75.5
Owns home 75.5 73.6
Income $42,533 $40,809
Parental Education: Less than HS 2.0 7.3
Parental Education: HS graduate 22.1 23.6
Parental Education: Some college 28.5 30.8
Parental Education: College 24.5 20.3
Parental Education: Advanced degree 22.9 18.1
Northeast 13.2 17.4
South 45.0 35.4
Midwest 19.3 21.9
West 22.5 25.4
Urban 57.8 73.3
Suburban 19.3 11.4
Rural 22.9 15.3
N 249 11932

23
 Table 3: Stated Reasons for Homeschooling in the NHES*

Proportion citing: 1996 1999 2003 Pooled

Available schooling alternatives 70.5 65.1 92.8 75.4

Religious or moral instruction 32.8 37.5 71.5 46.7

Child has disability 8.9 8.0 31.7 15.7

Child is temporarily ill 3.6 3.0 2.8 3.1

Other reasons 32.0 44.2 18.5 32.4

* Columns do not sum to 100% because parents were encouraged to check all that applied

24
Table 4: Determinants of the Decision to Homeschool

Marginal Effect
White 0.006 ***
White * 1999 -0.004 *
White * 2003 0.001
Parental education 0.006 ***
Family income ('000s) 0.000 ***
# of siblings 0.002 **
# in HH under age 6 -0.001 ***
Child disability impedes learning 0.001
Mom works PT 0.009 ***
Mom not in LF 0.015 ***
Mom's 1st language is English 0.010 ***
Family lives in suburbs 0.003 **
Family lives in rural area 0.003 **
Family lives in South 0.006 ***
Family lives in Midwest 0.004 **
Family lives in West 0.008 ***
1999 Year dummy 0.007 **
2003 Year dummy 0.006 ***

Predicted choice probability 0.010

Actual choice probability 0.017
N 42,418
*** p<.001, ** p<.01, * p<.05
Means and standard deviations of continuous variables:
Income - 45.673 (30.366), # siblings - 1.28 (1.07), # under age 6 - .576 (.788)

25
 Table 5: Full-Time vs. Part-Time Homeschooling in the NHES

FT Homeschoolers PT Homeschoolers
Male 51.3 46.6
Female 48.7 53.4
White 78.7 56.9
Black 5.0 17.2
Hispanic 9.5 19.8
Other race/ethnicity 6.8 6.0
Child’s age: 5 and under 12.7 11.2
Child’s age: 6 – 7 11.8 12.9
Child’s age: 8 – 10 21.1 15.5
Child’s age: 11 – 14 29.0 29.3
Child’s age: 15 – 18 25.4 31.0
Number of siblings: 0 15.0 23.3
Number of siblings: 1 32.4 36.2
Number of siblings: 2 24.5 23.3
Number of siblings: 3 17.7 12.9
Number of siblings: 4 5.2 1.7
Number of siblings: 5+ 5.2 2.6
Family type: 2 parents 83.7 77.6
Family type: 1 parent 15.4 18.1
Family type: other 0.9 4.3
Mom works full-time 17.5 25.9
Mom works part-time 18.8 25.0
Mom not in LF 60.3 40.5
Mom looking for job or no mom in HH 3.4 8.6
Child disability impedes learning 23.1 30.2
English mom's 1st language 93.4 85.3
Owns home 74.6 70.7
Income $45,898 $46,510
Parental Education: Less than HS 1.8 1.7
Parental Education: HS graduate 18.1 23.3
Parental Education: Some college 36.1 25.9
Parental Education: College 22.7 28.5
Parental Education: Advanced degree 21.3 20.7
Northeast 10.7 10.3
South 47.4 32.8
Midwest 19.7 15.5
West 22.2 41.4
Urban 53.5 71.6
Suburban 19.1 11.2
Rural 27.4 17.2
N 441 116
Percent of Sample 79.2% 20.8%

26
Table 6: Determinants of the Decision to Homeschool Full-Time

Marginal Effect
White 0.103 **
# of siblings 0.029 *
Child disability impedes learning -0.054
Mom not in LF 0.085 *
Family lives in suburbs 0.097 *
Family lives in rural area 0.077 *
Family lives in West -0.114 **
2003 Year dummy 0.010

Predicted choice probability 0.819

Actual choice probability 0.792
N 557
*** p<.001, ** p<.01, * p<.05
Means and standard deviations of continuous variables: # siblings - 1.75 (1.36)

27
 Table 7: Characteristics of Homeschoolers in the GSS

Home Public Private Private

School School School Religious
Respondent is married 0.72 0.55*** 0.68 0.62
Age of respondent 49.25 51.40 54.40** 53.47**
Race of respondent
White 0.89 0.77** 0.85 0.80
Black 0.10 0.18 0.15 0.15
Other race 0.02 0.05 0.00 0.05
Family income 51179.25 44713.90 79337.50*** 56048.64
Mother’s education 13.52 12.72** 15.19*** 13.66
Father’s education 13.93 13.07* 16.40*** 13.88
SMSA 0.79 0.71 0.88 0.80
Northeast 0.16 0.15 0.12 0.18
Midwest 0.25 0.25 0.09** 0.28
South 0.36 0.39 0.51* 0.33
West 0.20 0.18 0.18 0.13
# Household members
Total 3.08 2.73* 2.76 2.76
Under 6 yrs 0.23 0.18 0.13 0.16
Btwn 6 & 12 yrs 0.65 0.45** 0.48 0.33***
Between 13 & 17 yrs 0.32 0.28 0.23 0.33
Read paper > once wk 0.39 0.43 0.50 0.46
Hours of tv per week 3.10 3.07 2.61 2.61
Sample size 61 2139 68 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

28
 Table 8: Religion in the GSS

Home Public Private Private

School School School Religious

Religion
Christian 1.64 1.68 0.00 3.08
Protestant 55.74 62.37 66.18 42.69*
Catholic 34.43 20.57*** 7.35*** 46.15*
Jewish 0.00 1.92 8.82** 1.15
No religion 8.20 10.94 16.18 5.00
Other Religion 0.00 2.34 1.47 1.92
Sample Size 61 2139 68 260

Frequency of Religious Service Attendance

Never 6.67 19.58** 17.91* 8.63
Twice per year or less 23.33 39.45** 26.86 17.65
Once per week or more 46.67 25.71*** 31.34* 51.37
More than once a week 25.00 8.03*** 13.43* 14.12**
Sample Size 60 2089 67 255

Strength of Religious Beliefs

Self report strong belief 44.26 35.30 47.06 55.77
Bible is actual word God 40.98 32.02 20.59** 32.69
Sample Size 61 2139 68 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

29
 Table 9: Politics in the GSS

Home Public Private Private

School School School Religious
Democrat 31.15 46.75** 41.18 45.38**
Republican 52.46 32.96*** 39.71 39.62*
Independent 14.75 18.33 16.18 13.85
Liberal 24.59 23.38 33.82 17.31
Moderate 29.51 38.62 26.47 31.54
Conservative 44.26 31.42** 38.24 45.00
Sample size 61 2139 68 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

Table 10: Views on Government Spending in the GSS

Home Public Private Private

School School School Religious
Pay too high taxes 54.10 42.12 30.88*** 45.77*
Govt spends too little on
Education 37.70 34.13 50.00 32.31
Crime 24.59 29.73 33.82 32.31
Defense 6.56 11.27 11.76 13.85
Welfare 6.56 9.12 11.76 6.54
Sample Size 61 2139 68 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

30
 Table 11: Views on Social Issues in the GSS

Home Public Private Private

School School School Religious
For capital punishment 63.93 64.24 58.82 61.15
Against gun permits 11.48 11.59 7.35 14.62
For tough on crime courts 73.77 72.00 61.76 74.23
Against public sex educn 24.59 9.30*** 7.35*** 12.69**
Against ban on school prayer 37.70 41.80 27.94 45.38
Against abortion 42.62 45.40 27.94* 46.15
For increase in divorce laws 47.54 34.13** 33.82 38.46
Against premarital sex 37.70 27.21* 20.59** 32.69
Against homosexuality 42.62 43.06 32.35 45.77
Gun owner 24.59 24.26 16.18 23.85
Sample Size 61 2139 73 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

Table 12: Views on Parenting in the GSS

Home Public Private Private

School School School Religious
Against teenage sex 60.66 60.87 58.82 61.92
For spanking as discipline 52.46 50.68 29.41*** 44.62
Against birth control for kid 39.34 29.50* 27.94 38.46
Most desirable kid quality
Obey parents 19.67 14.73 14.71 13.85
Be popular 1.64 0.61 0.00 0.38
Help others 1.64 8.93** 5.88 10.38**
Think for self 31.15 29.92 47.06* 31.92
Work hard 8.20 10.80 11.76 12.31
Stay at home moms are best 29.51 25.95 23.53 22.69
Against public sex educn 24.59 9.30*** 7.35*** 12.69**
Sample size 61 2139 68 260
*, **, *** Statistically different from homeschoolers at the 10%, 5%, and 1% levels.

31
Table 13: Determinants of the Decision to Homeschool in the GSS

Marginal Effect Standard Error

White 0.007* 0.004
Mom has at least college educ 0.008 0.006
Income (1000s) -0.000 0.000
# in HH under age 18 0.004*** 0.001
# in HH under age 6 -0.003 0.004
Parent is immigrant -0.006 0.005
Family lives in SMSA 0.005 0.004
Family lives in South 0.000 0.005
Family lives in Midwest -0.002 0.005
Family lives in West 0.002 0.006
* Significant at the 10% level ** Significant at the 5% level *** Significant at the 1% level

Table 14: Determinants of the Decision to Homeschool in the GSS

Marginal Effect Standard Error

Republican 0.005 0.004
Against sex education in public schools 0.018** 0.009
Married 0.007** 0.003
Thinks pay too high income taxes 0.006* 0.003
Frequency of church attendance 0.001** 0.001
Thinks working moms inferior -0.000 0.004
* Significant at the 10% level ** Significant at the 5% level *** Significant at the 1% level

32