![[TOP]](/icons/top.gif){kind=icon}
by
Patricia M. Lines
Web Edition, Spring 1999
(previously released as a paper in Spring 1998)
This paper is intended to promote the exchange of ideas among researchers and policymakers. The views expressed in it are part of ongoing research and analysis and do not necessarily reflect the position of the U.S. Department of Education.This technical paper provides an estimate of the number of children in homeschooling. For general background on homeschooling, the reader should consult the earlier working paper: P. Lines, "Homeschooling: An Overview for Education Policymakers," revised March 1997; a version of this paper will appear in Private Education and Educational Choice, edited by James G. Cibulka (Greenwood Press, forthcoming). A single copy of this paper may be obtained by contacting Dorothy Yates at Dorothy_Yates@ed.gov or by phone at (202) 219-2079.
Return to SAI Publications
Many persons read earlier drafts of this paper and provided valuable advice. Hopefully, the author improved the paper on following their suggestions, but if she did not, the fault is hers, not theirs. They include Judith Anderson, Senior Research Associate, National Institute on Student Achievement, Curriculum, and Assessment at the U.S. Education Department (ED); Stephen P. Broughman, Statistician, National Center for Education Statistics (NCES), ED; Michelle Doyle, Director, Office of Non-Public Education, ED; Patrick Farenga, Publisher of Growing Without Schooling; Jack Klenk, Special Advisor, Office of Non-Public Education; Chris Klicka, Executive Director, Home School Legal Defense Association; Brian D. Ray, President, National Home Education Research Institute (NHERI); Thomas D. Snyder, Senior Research Analyst, NCES, ED; Lisa Towne, Presidential Management Intern, Planning and Evaluation Service, ED; the many individuals in State Education Agencies (SEAs) who helped by examining the paper, particularly for insights into their states; and the homeschooling leaders in various states who were willing to provide insights into the practices in their states. They are too many to mention, but the debt is greatest to these last two groups.
Homeschooling is the education of school-aged children under their parents' general monitoring, and it replaces full-time attendance at a campus school. Some homeschooling children enroll part time at a campus-based school, or share instruction with other families, but most of their educational program is under the direct oversight of parents. While many activities take place in the home, parents often draw on their community, neighboring institutions, and travel opportunities to complete the program. The definition used for this paper includes families who self-identify as homeschoolers, even if they utilize part-time school enrollment.
Homeschooling has more than doubled possibly tripled – in the 5 years between the 1990-91 school year and the 1995-96 school year. By the 1995-96 school year, from 1 to 2 percent of the total school-aged population were in homeschooling. Within the private education world, it has become a major sector, where it represents approximately 10 percent of the privately-schooled population. In some states homeschooling may exceed 20 percent of the privately-schooled population.1
Given the evidence provided here, the total number of homeschoolers in the 1990-91 school year seems to have been between 250,000 to 350,000 children nationwide; and around 700,000 to 750,000 in 1995-96. Based on limited evidence from four states, the number is still growing; the rate was between 7 and 15 percent from the 1995-96 school year to the 1996-97 school year.2 Assuming the larger growth rate, which is more consistent with past growth,3 the number could have reached about 1,000,000 children by the 1997-98 school year.
Growth has persisted over three decades. Earlier estimates, based on different methodologies, suggested 60,000 to 125,000 school-aged children for the fall of 1983; and 122,000 to 244,000 for fall of 1985; between 150,000 to 300,000 for fall of 1988; and between 250,000 to 350,000 for fall of 1990. A retroactive estimate done in 1988 suggested 10,000 to 15,000 children received their education at home in the late 1970s and early 1980s, close to an estimate made at the time by an early leader of the homeschooling movement, educator and author, John Holt.
The data collected by State Education Agencies (SEAs) show a steadily growing homeschooling movement. Thirty-two states and the District of Columbia have released data on documented children in homeschooling for both years examined in this study. These data account for over 76,000 documented children in 1990-91 and almost 214,000 in 1995-96. This represents a fraction of the total, as it includes only children in families that have filed papers indicating that they are in homeschooling. Many families do not file papers, although it is required. Nor does the documented number include children enrolled in distance learning programs unless their parents have also filed papers with state or local officials indicating that theirs is a homeschooling program. Appendix A presents these data for each state.
States usually collect these data from local districts, although some collect it directly. Often the local education agency (LEA) collects the data in late fall or early winter and forwards it to the SEA by spring. Some states continuously monitor the number, and adjust it when families return to school, and publish only the year-end number. There is neither uniformity among states in procedures, nor in the dates for collection.4
Based on discussions with state officials and homeschooling leaders familiar with the state, and on a nationwide survey (discussed more fully below), it is possible to identify three states where both homeschoolers and the SEA believe that filing rates approached 100 percent by 1995-96: Maine, New Mexico, and Wisconsin. In 1995-96, the total number of documented homeschooling children in the three states was 24,369 or 1.52 percent of the school-aged population in these states. The distribution within each of these states was approximately the same percentage.
Even here, the official filings make for a conservative statement of the number of homeschoolers in these states. There will be a small number of homeschoolers in these states who do not file papers, but we have no basis for estimating their numbers. In Maine, in particular, some homeschooling families operate as unapproved private schools and do not file papers with the state or local government.5 If these states represent the country as a whole, there would be over 250,000 children in homeschooling, nationwide, in 1990-91; and almost 750,000 by 1995-96.
Except in states with near 100 percent filing, the state's documented data do not account for all homeschooled children. Many families do not file any paperwork with state or local officials because their homeschooling child is not old enough to come under the state's compulsory education law, and the state's law does not require filing. Some families are homeschooling under a state constitutional or statutory provision that excuses religious-based homeschoolers from filing requirements. In other states, some families may choose to follow the state's laws for private schools, and in a few of these states, these family-based schools do not file paperwork. In addition, the state data may be missing reports from some districts or be missing late-filed data.
It is possible to estimate the non-filers, based on a survey by Brian Ray, of the National Home Education Research Institute (NHERI). In 1990 and 1995 Ray surveyed a sample of the membership of the Home School Legal Defense Association (HSLDA), which had commissioned the survey. In both survey years about 63 percent6 of respondents nationwide reported that they had filed "any kind of" paperwork with the state or local district.
As filing rates vary considerably from state to state, and depend on the state legal and policy environment, it is necessary to consider each state separately. Appendix B provides an estimated percentage of filers by state, based on the survey, and used with appendix A, provides the basis for the calculations given here. These appendices allow those knowledgeable about their state to make individual state-level estimates. Those using appendix B to estimate the number for a single state must have a good sense of whether the survey results for non-filers represents all homeschoolers in the state.
For most states it is necessary to estimate non-filers to get an accurate picture of homeschooling numbers. For a national estimate, one can minimize bias due to the survey by using data only from states with high filing rates. There are 14 states where 80 percent or more of the parents responding to the 1995 Ray survey said that they filed some kind of paperwork: Arkansas, Georgia, Maine, Maryland, Minnesota, Montana, Nebraska, New Hampshire, New York, North Carolina, South Dakota, Pennsylvania, Wisconsin, and Wyoming.7
The analysis of this group must omit two states where the state-level data is incomplete: Pennsylvania and New York. In Pennsylvania, most children under age 8 are missing from the state data reports.8 As the age of compulsory education in that state does not begin until age 8, parents may file papers on older children (and are treated as filers based on the survey), but would not provide information on their children under age 8.9 In New York state, the SEA believes it is missing data on documented children for New York City for 1995-96; it received no data at all from New York City in 1990-91.10
The remaining 12 states accounted for 88,581 documented children in 1995-96. These documented homeschoolers are 1.2 percent of the total school-aged population in these states. The weighted filing rate in the Ray survey indicates 87.7 percent of homeschoolers filed papers in these states in 1995-96. If the survey is an accurate indicator of the percentage of filers, we can estimate about 101,000 homeschooling children, or 1.4 percent of the total school-aged population in these states. This would be consistent with a national homeschooling population of around 691,000 children. The calculations used to arrive at this conclusion appear in appendix C, and illustrate the method for calculating all subsequent estimates of homeschooling numbers. For both 1990-9111 and 1995-96, the percentage estimated for these states is slightly lower, but close to the percentage computed for the three states where the filing rate was assumed to be 100 percent.
The range of estimated percentages of homeschoolers in these states varies. The most popular homeschooling states in this group appear to be South Dakota and Montana, where an estimated 2.1 percent of the population were in homeschooling in 1995-96. Arkansas, Georgia, Maryland, and North Carolina have the smallest estimated percentage of homeschoolers, at 1.3 percent. Minnesota had 1.4 percent; Maine, Nebraska and Wisconsin had 1.5 percent; and New Hampshire and Wyoming both had an estimated 1.6 percent of the population in homeschooling.
A third estimate uses all 19 states that in 1990 met the criterion that there be at least 12 respondents to the Ray survey for that state.12 In this estimate, potential bias due to selection of states decreases, while bias in the estimate of non-filers based on the survey increases. Of the total in the 19 states, an estimated .42 percent were in homeschooling in 1990-91. For all 28 states that in 1995-96 met the criterion, an estimated 1.26 percent were in homeschooling. If typical of all states, this suggests about 211,000 homeschooling children nationally for 1990-91; and about 611,000 in 1995-96. However, these estimates include New York and Pennsylvania, which are missing children for whom parents filed papers. If we exclude these two states, the remaining 17 states had an estimated 106,350 children in homeschooling, or 1.43 percent of the school-aged population in 1995-96. This is consistent with 701,000 children in homeschooling nationwide in that year.
As noted, the accuracy of the estimated number of non-filers depends on the representativeness of the Ray survey.13 In 1990, Ray drew a random sample from the membership of HSLDA. In 1995, he repeated this step, and to broaden the base of his survey, he sent questionnaires to other homeschooling organizations with a request that the organization distribute the surveys to a sample of membership, following the enclosed instructions for drawing the sample. The response rate in 1990 was good — at approximately 70 percent. In 1995, he could not find out how many questionnaires were mailed by cooperating organizations, or even the number of cooperating organizations. As a result, he could not compute a response rate. He believes the response rate was around 30 percent.14 It seems likely that the reduction in response rate was due to a large nonresponse from the organizations outside the HSLDA network, and that his survey continues to reflect HSLDA membership or those sympathetic to the HSLDA principles.
It is hazardous to generalize to an unknown population, based on a sample from only a portion of that population. At least, one must understand the origins and limitations of the sample, and proceed with caution. HSLDA supports Christian homeschooling, and members pay an annual fee of $100 for prepaid legal services for assistance in the event that they face problems with their homeschooling program. HSLDA does not accept members who utlize public school programs for homeschoolers. This sets its members apart from other homeschoolers in both predictable and unpredictable ways.
An important source of underestimation lies in the structure of the survey itself. While it elicits separate data for each homeschooled child, there is only one question about filing papers with state or local officials. In some families, parents may file papers for one child, and so answer "yes," but for children above or below the age of compulsory education, the answer would have been "no," had there been a question for each child. Again, the Ray survey will overstate filing rates as a result. In a state such as Pennsylvania, where the state has data on only a handful of children under the age of 8, this can produce a large underestimate of the actual number of children in homeschooling.
Another issue involves whether nonresponders on any survey will differ from responders in their compliance with state laws requiring filing of paperwork. It seems reasonable that nonresponders may be less organized, less punctual, or more cautious about revealing their homeschooling status. Thus, it seems likely that families that respond to a survey include a disproportionately high number that would respond to filing requirements.
Another difference might arise if HSLDA members have different filing rates than other homeschoolers. In some states, such as New York, where large numbers of families who homeschool for philosophical reasons, and who, also for philosophical reasons reject standardized testing, the HSLDA members may not be representative, and actual filing rates will be lower than assumed for purposes of this paper.
All things considered, it seems likely that the estimate of non-filers is low. For that reason, this paper relies on the survey in limited ways. It is used only to help identify states with high filing rates; and to estimate non-filers only in such states. By limiting the use of the survey in this way, a relatively small percentage of the total is based on an estimate of non-filers. The examination of the three states believed to have near 100 percent filing rates does not rely on the survey at all, and is probably a better indicator of the experience nationwide than the subsequent two estimates.
Loose corroboration of the estimates can be made in a variety of ways. For example, one could test the evidence by attempting to predict the homeschooling population in a state, based on the estimates, and compare this with what is known about the state. California is a special case, because of its size. California was not included in the estimates provided above because the legal context is ambiguous, and the interpretation of the Ray survey is hazardous, given the special circumstances of California.15 Instead, incidence of homeschooling in California, for purpose of the national estimate above, is inferred from the average in other states.
This inference can be tested. In California, with almost 6 million school-aged children, one would expect 90,000 children in homeschooling, if the 1.5 percent average is an accurate predictor. If one tallies up the known number of children in homeschooling in California, one can account for almost 62,000 children.16 This includes children in private schools with less than five children – generally assumed to be homeschools. HSLDA in fact encourages its members to file as private schools in California. It also includes children enrolled in independent study in a public program.17 The SEA regards this as something other than homeschooling, while the families and the local public schools appear to regard it as homeschooling with curricular assistance from a public school teacher.18 Finally, it includes children enrolled in public school home study programs. As the SEA regards this last option as the only legal option for homeschooling under state law,19 many California homeschoolers may prefer to remain underground. If just one-third or more are underground, then the estimates based on experience in states with good filing rates does help predict, but only loosely, the incidence in California.20
Other studies also help corroborate the estimate offered here. In 1990, I triangulated the approach and made three independent estimates. The high estimate was based on curricular sales, and the low estimate based on information from state homeschooling associations.21 The estimate based on state data, with its middle position made it attractive as a single estimate. Since then, moreover, more states have been able to provide data.
In a separate study, Ray recently estimated numbers based on the sale of complete, graded curricular packages by four popular providers of such packages. Based on the percentage of survey respondents that said that they used such a package (4.88 percent),22 he computed 1.2 million for 1996, but he also estimated that this contained a large margin of error.23 Some of the difference will be due to growth from 1995-96 (the date for the estimate here) to 1996-97, the date for which Ray made his estimate. The remainder of the difference will be due to bias in the samples informing both our estimates. The bias in the estimate offered in this paper has been discussed above. For an estimate based on curricular packages, there is a possibility of severe upward bias if any of the four curriculum providers rounded the number upwards when providing data on their sales.24 There could also be an upward bias if many families accelerate a child through more than one package a year. It is our mutual impression that the estimate presented here is low, and Ray's is high.25
An alternate way to assess the extent to which Americans have turned to homeschooling would be a household survey. The United States Census and the U.S. Department of Education's National Center on Education Statistics (NCES), have begun to include questions about homeschooling in household surveys. The 1994 Current Population Survey, for example, included a series of questions about homeschooling. Analysis of these data is expected in about a year.26 When complete, the analysis will serve as a useful check on the analysis presented here.
What to make of this growth? If one is considering the size of the constituency, one might note that the total number of individuals with some homeschooling experience will be much greater than the number homeschooling in any single year. Some observers think that the turnover rate is about 2 years per child.27 True, many families homeschool their children for their entire school-aged years, but many others try it for a shorter time. If the average turnover rate is 2 years, then the number of children with some homeschooling experience, by age 18, would be around 6 to 12 percent of the population.
The general size and direction of the homeschooling movement is beyond doubt. Policymakers needing more precise knowledge, such as planning for teacher hiring, should consult the estimates only with great care. Only those states with high filing rates can make accurate calculations based on their documented data. On the other hand, the estimates should help policy decisions that need a general, broad picture. For example, a district interested in starting a public school program for homeschoolers can be reassured that there is probably a constituency for the program in the district. As a second example, districts interested in including homeschoolers in planning can make some rough estimates of the number in their district.
1 This seems likely in the western states, where the private school population usually is less than the national average and homeshooling is relatively popular.
2 Nebraska, South Carolina, Wyoming, and Pennsylvania sent the author their most recent data as it became available. Growth in Pennsylvania and South Carolina increased about 15 percent from one year to the next. It was closer to 7 percent in Nebraska and Wyoming. The data are provided in appendix A, third column.
3 The growth rate from the 1990-91 to the 1995-96 school years appears to be between 20 and 25 percent per year.
4 One reader of an early draft of this paper suggested the state number is inflated because students drop out of homeschooling and go back to school during the year. This may be true, but it seems that, given the growth curve for homeschooling, it is more likely that others will begin homeschooling after the count has been made for the year.
5 Interview with Edwin N. Kastuck, Maine Department of Education, May 14, 1998.
7 Only states with 12 or more respondents on the survey are included.
8 The state data show for 1995-96, 104 children age 5; 339 at age 6; 981 at age 7; and 1,768 at age 8 in homeschooling. The jump from 7 to 8 suggests that most parents are not providing information on their children below age 8.
9 E-mail correspondence with Susan Richman, a Pennsylvania homeschooling leader, April 16, 1998.
12 See appendix B for the number of respondents for each state.
23 HSLDA, Home Education Across the United States: Family Characteristics, Student Achievement, and Longitudinal Traits (NHERI, 1997). See http://www.hslda.org for an abridged copy of this report.
25 Personal communication, November 12, 1997.
-###-
Appendix A: State Education Agencies Summary of Available Data on Homeschooling