Research Proposal.

The Social and Educational Outcomes of Homeschooling

Joseph Murphy

Frank W. Mayborn Chair, Vanderbilt University, Nashville, Tennessee, USA

In this article, we provide a comprehensive review and analysis of the outcomes of homeschooling in

America. We ground the work in an examination of the importance of homeschooling in society in

general and education in particular. We provide an analysis of the strengths and weaknesses of the

existing research base on homeschooling. With an eye on methodological weaknesses in the home-

school research, we compile data on what is known about the outcomes of this social movement

and educational reform. We document the impact of homeschooling on the social fabric of the nation

(e.g., families) and the institution of schooling (e.g., student learning outcomes).

In this article, we provide a comprehensive analysis of what is known about the impact of home-

schooling. We begin with a note to explain why the analysis of the impact of this social and

educational reform movement merits our attention. The first part of the article provides a portrait

of the quality of the existing research base. The final sections of the article distill what is known

about the impact of homeschooling across an array of outcomes: the social fabric of the nation,

schools, costs, families, and children. On the last topic, we explore what is known (and how well

it is known) regarding academic achievement, social development, and success after completing


On the one hand, homeschooling merits attention for what it conveys about the social fabric of

the nation. The study of homeschooling provides important insights on the conservative mosaic

that has been formed in the United States over the last 30 years. Such analysis reveals a good deal

about the tensions between individualism and community, as well as new ways to think about

these social constructs. The study of homeschooling allows us to peer more thoroughly into

the place of religion in the United States at the turn of the twenty-first century. Even more impor-

tant, an examination of homeschooling provides significant insights into the nature of American

families. Perhaps most centrally, the possibilities of movements that attempt to reverse the seg-

mentation of life in America are surfaced. We learn that homeschooling is both an animating force

for and an exemplar of efforts to provide an integrated frame for life in the postmodern world

(Collom and Mitchell 2005; Gaither 2008; Stephens 2001).

As with most social movements of significance, analysis of homeschooling reveals much

about the battle for the moral high ground in the country. As we attend to the history and devel-

opment of homeschooling itself, we learn as much about the ebb and flow of waves of influence

and the shifting pendulum in the area of social ideas and tastes in general (Murphy 2013). In a

Address correspondence to Joseph Murphy, Vanderbilt University, 210B Payne Hall, Box 414, 230 Appleton Place,

Nashville, TN 37023-5721. E-mail:

Sociological Spectrum, 34: 244–272, 2014

Copyright # Taylor & Francis Group, LLC ISSN: 0273-2173 print/1521-0707 online

DOI: 10.1080/02732173.2014.895640



similar vein, homeschooling reveals how history is both a product of as well as a platform for

powerful figures to push and pull ideas onto society’s central stage (Gaither 2008). Considerable

insights about the legal workings of the nation are also exposed. Dynamics about social organiz-

ing become visible, especially in getting society to embrace ideas that once seemed anathema.

By studying homeschooling, we accumulate a good deal of knowledge about the evolving role

of government in the affairs of its owners. We track an evolution from government as the

unquestioned mechanism to produce a better society to government as a self-forged and confin-

ing manacle that also has the potential to hinder improvement. In short, a study of homeschool-

ing exposes dynamics that transcend the content of the topic at hand (Cibulka 1991; Gaither

2008; Van Galen 1991).

At the same time, homeschooling merits examination for what it reveals about education and

schooling in America (Ray 2000a). Ongoing and dilemma-based questions about governance

and control play out across its pages in new ways. Much can be learned here about possible

forms of schooling in the twenty-first century, new conceptions and models that were unimagin-

able to the previous generation. Issues of funding and costs lurk in the background of home-

schooling, but nonetheless offer important insights into financing the nation’s most costly and

critical state and local service—and other services as well (Ray and Weller 2003). Labor issues are prevalent in the homeschooling literature but usually cast obliquely. Even so, it is difficult to

overlook the implications for the traditional and deeply rooted notions of civil service in the

nation, especially in light of the prevalence of parallel trends in education (e.g., vouchers, tax

credits, privately managed charter schools) and the larger society. A careful study of home-

schooling produces considerable wisdom on the role of markets and profits in the education

sector (Murphy 1996, 1999). Much can also be gleaned about the linkages between schooling

and social justice by examining homeschooling (Apple 2005; Lubiensky 2000).

We also need to study homeschooling because it is the most robust form of educational

reform in the United States today. The growth of homeschooling in the U.S. has been nothing

short of remarkable, even using the most conservative estimates available. Only 10,000 to

15,000 children were being homeschooled in the 1970s. By 2010, somewhere in the neighbor-

hood of 2 million students were part of this group. Scholars confirm that homeschool enrollment

is now about one-fifth the size of private school enrollment (Belfield 2004b; Isenberg 2007). We

find that almost twice as many youngsters are schooled at home as are educated in conservative

Christian schools (Glanzer 2008). And more children are taught at home (2,000,000) than attend

charter schools (1,500,000) and receive vouchers combined (Apple 2007; Belfield 2004b).

When life cycle numbers are compiled, we discover that 6 to 12 percent of all students will have

been educated at home at some time in their K-12 educational career (Houston 1999; Isenberg



One of the most stark conclusions one draws when interrogating the scholarly literature on the

impacts of homeschooling is just how thin the empirical knowledge base is on this social phenom-

enon and educational movement. To be sure, the literature exposes a good deal of ideological

bantering as well as some solid conceptual modeling. Homeschoolers have provided a trace of

good reports on how to engage the work of educating a child at home, and fine collections of




resources to assist in those efforts. But the research cupboard is not well stocked, especially in the

domain of outcomes. This assessment first surfaced as the homeschool movement reached early

adolescency. It was revealed that the entire domain was largely uncharted in a scientific sense (see

Delahooke 1986; Gladin 1987; Groover and Endsley 1988; Knowles 1989; Mayberry 1989;

Schemmer 1985; Taylor 1986a; Williams et al. 1984; Wartes 1987; Wright 1988). More troub-

ling, this same conclusion was consistently reached by scholars, analysts, and policy makers

throughout the 1990s as well (see Dalaimo 1996; Duvall et al. 1997; Hertzel 1997; Houston

1999; Kelley 1991; Knowles et al. 1992; Luebke 1999; Mirochnik and McIntire 1991; Rudner

1999; Van Galen 1991). Even more disheartening, the next generation of reviewers has uncovered

little evidence that the limited empirical evidence deficiency was addressed with much sense of

robustness during the first decade of the twenty-first century (see Cogan 2010; Green and Hoover-

Dempsey 2007; Ice and Hoover-Dempsey 2011; Isenberg 2007; Kunzman 2005, 2009a; Muntes

2006; Nemer 2002; Taylor-Hough 2010).

An Assessment of Research Evidence

As Medlin (2000:118) informs us, the summative narrative of research in homeschooling

parallels the chronicle found in many new domains of study:

no guiding theory, inadequate experimental design, poorly defined research questions, untried and

weak measures, unorthodox treatment and presentation of data, and conclusions based on subjective

judgments. Even a cursory look at the research reveals that many studies are qualitative descriptions of

so few participants that the results cannot be generalized. Many are surveys that rely exclusively on

parental reports but offer no idea of how reliable those reports may be. Many test only home-schooled

children without comparing them to children attending conventional schools, making it very difficult

to know what the results might mean. Further, all home school research is correlational (because

researchers have no way to control the type of schooling children experience), samples are usually

self-selected (because researchers cannot require home schooling families to participate), and how-

ever carefully researchers try to match their home-schooled and traditionally schooled groups, there

are probably still important differences between the two.

Reinforcing many of the points and adding some additional insights, Belfield (2004a:10)

exposes two major problems that plague research that compares the impact of homeschooling

against other types of schooling, especially public schooling.

The first is the common concern over the endogeneity of school choice, that is different types of

families choose the type of school that their children attend, and little can be inferred about the impacts

of schools for students who do not attend them. The second is the need to distinguish the absolute

performance of home-schoolers from the treatment effect of home-schooling. Given the above median

resources of many home-schooling families, academic performance should be even if home-schooling

itself is not differentially effective. Full controls for family background are needed, however, to

identify a treatment effect.

Blok (2004) also reminds us that the body of empirical work on the impacts of homeschooling is

rather thin. Analysts also have pointed out that most of the research that has been undertaken has

been conducted by investigators with a good deal of interest in shaping results into positive stor-

ies, i.e., by advocacy groups whose defined mission is to promote the cause of homeschooling




(Houston and Toma 2003; Kunzman 2005; Reich 2005). The conclusion at present is that

research on the impacts of homeschooling leaves a good deal to be desired. Most troubling is that

we know almost nothing about the causal links in the homeschool theory of action and their

connections to various outcomes (Ray 2009b).

Much of what we do know about homeschooling is anecdotal in nature (Houston 1999). Stories

of individual children who have demonstrated remarkable achievements in academic competitions

of varied sorts or in higher education are especially prevalent. On the other side of the ledger,

negative stories are sometimes spotlighted to confirm the dangers of homeschooling. Many of

these stories are proof of the impact of homeschooling; however, stories and anecdotes, personal

experiences, and folklore lack the authority of scientific evidence (Stevens 2001; Taylor 1986b).

As we attempt to move beyond anecdotes and stories, we find that rigorous empirical research

on the effects of homeschooling remains scarce (Houston 1999; Stevens 2001). We learn that

studies on homeschooling effects suffer from major, interconnected problems that significantly

limit the degree of certainty we can draw from research reports. Problems with samples and

controls are particularly troublesome. On the first issue, analysts have routinely urged caution

in accepting findings at face value because of the nonrepresentativeness of the samples employed

in almost all homeschool research (Kaseman and Kaseman 1999; Kunzman 2005; Stevens 2001).

Weak sampling frames mean first that samples are almost never drawn on a representative group

of homeschoolers (Houston 1999; Kaseman and Kaseman 1999). Studies employ highly selective

samples (Bauman 2002; Winstanley 2009). Self-selection and sampling bias have been and con-

tinue to be the norm in the study of homeschool effects (Lines 2000a, 2000b; Ray and Wartes

1991). Nonrepresentativeness means, of course, that findings cannot be generalized to the home-

school population. Even in these studies with nonrepresentative samples, return rates often fall

below acceptable standards (Dahlquist et al. 2006; Stevens 2001). We rarely see efforts to exam-

ine nonresponders (Wright 1988). (For good treatments of these problems in context of Rudner’s

[1999] classic study, see Kaseman and Kaseman 1999, and Welner and Welner 1999.)

Concomitantly, researchers are quick to point out that the claimed benefits of homeschooling

rest on shaky ground because studies rarely control for other explanatory variables in the causal

effects equation (Blok 2004; Ray 2000b). Random assignment or other less powerful methods

that can help eliminate alternative explanations for effects are rarely used (Collom 2005;

Kunzman 2005). In particular, analysts decry the absence of controls for socioeconomic variables

such as income, occupation, and education (Dahlquist et al. 2006), previous achievement (Ice et al.

2011), marital status (Burns 1999), and parental support and commitment (Barwegen et al. 2004;

Hertzel 1997). Because these conditions are linked to student learning, they need to be accounted

for in homeschool effects research. Without appropriate controls, it is impossible to establish

whether outcomes are the result of the treatment (i.e., homeschooling) or other factors (e.g.,

family income) (Belfield 2005). Or as Lines (1995:3) nicely penned it, without controls research

does not allow us to ‘‘determine whether the same children would perform better or worse in a public classroom or in a home-schooling arrangement’’ (emphasis in original).

The question aptly raised is whether any cause and effect relationship exists vis-à-vis home school

education. . . . To date, no controlled studies exist that shed significant light on the important question. . . . Until some type of study is conducted, using control and experimental groups, the question likely will be left for speculative—rather than concrete—answers. (Wilhelm and Firman 2009:310–311)




Other difficulties are visible in the portfolio of research on homeschool outcomes. For

example, in studies of effects on homeschool youngsters comparisons to national norms are

traditional. While not without informative power, this strategy leaves a good deal to be desired

(Welner and Welner 1999). We also very rarely hear from children in homeschool families about

their perceptions of homeschool work and their assessments of outcomes (Mayberry et al. 1995;

Schemmer 1985). Testing conditions in some homes are problematic (Basham et al. 2007).

Practical problems arise in getting information to establish effects (Belfield 2005). For example,

almost all of the data available because children are enrolled in public schools are missing for

homeschool children (Isenberg 2007). There are very few longitudinal studies.

There are significant practical and methodological difficulties in counting these children

(Lines 1991; Belfield 2004a). As Reich (2005) reports, in many places, registration of home-

schooled children (a prerequisite to being counted) is not required. In other places, homeschool

families simply neglect to register when they are required to do so (Bates 1991; Kleist-Tesch

1998; Lines 1999). In still other cases, parents are so opposed to governmental oversight of their

families they refuse to participate in data collection activities, both census and research efforts

(Bates 1991; Collom 2005; Kunzman 2005). Others fail to register because they lack trust in

government agencies (Wartes 1988). Still others fall under provisions that do not require

religious-based homeschoolers to register (Lines 1999). Some families operate as ‘‘private

schools’’ thus again eliminating the registration requirement (Aurini and Davies 2005; Lines

1999). Finally, problems stemming from the nature of the intervention itself—a small population, definitional issues, geographical dispersion, and decentralization to hundreds of thousands of

sites (homes)—make counting (and studying) homeschoolers a difficult task (Belfield 2004a; Collom and Mitchell 2005). All of these conditions lead to production of less-than-satisfying

estimates, ones that are biased downward.

Counting problems also arise from the methods used to arrive at estimates (Mirochnik and

McIntire 1991). For example, researchers often turn to lists of families who have joined home-

school associations to draw estimates. At other times, they rely upon lists of those who purchase

materials from homeschool curriculum providers. However, since some families do not join sup-

port groups and=or purchase from homeschool providers these sources are likely to undercount homeschoolers (Lines 1999). It is also important to remember that there is a lack of uniformity

among states in how and when they collect data on this population (Lines 1999).

More recently, researchers have employed household surveys to arrive at the number of

homeschoolers in the United States. While this approach overcomes many of the problems

inherent in the previously discussed methods, the procedure is not free of problems (Bielick

et al. 2001). In particular, because they are often such a small percentage of school-age children,

very few of them are likely to be included in national household surveys (Wenger and Hodari


Perhaps the most puzzling finding in the area of homeschooling effects is that almost every

potential domain of impact that defines the intervention from parents’ perspectives is ignored

while researchers chase down data on whether homeschool children can answer two or three

more questions correctly on standardized tests than their public school peers. At the macro level, this is the case because the impacts of homeschooling as a broad social movement are generally

not investigated. A fair amount of conceptual work on the issue of the impact of homeschooling

on the social fabric of the nation is available, especially by those who foresee potential negative

consequences (e.g., Apple 2007; Lubienski 2000; Reich 2005). However, with the exception of




work from scholars such as Gaither (2008) and Stephens (2001) there is scant guidance in the

literature about how to think about operationalizing and measuring societal impacts.

At the mid level absence of attention to core ou