Joan E. Havens, Ph.D.
Christian Education Department
Columbia International University
Columbia, South Carolina 29230-3122
Keywords: Homeschooling, home schooling, home education, academic achievement
Some states attempt to regulate home schooling in a variety of ways. They may require that parents register their children with the public schools, or they may seek to impose criteria for instruction, such as textbook approval, home visits by school personnel, periodic testing, and parent certification (Lines, 1983; Roach, 1988). Although most states officially favor parent certification as a prerequisite for home instruction, they have not been able to legally restrict home schoolers solely on this basis (Lines, 1983; Whitehead & Bird, 1984).
The central issue seems to focus on whether teacher certification is necessary to ensure student learning. There appears to be little empirical evidence, however, that such a connection exists. Students taught by non-certified teachers in private schools often achieve higher scores on standardized tests than do public school students taught by certified instructors (Gorder, 1987; Harris, 1988; Sacken, 1988; Whitehead & Bird, 1984).
This raises the inevitable question regarding the academic achievement of home schooled children who are generally taught by non-certified parents. How do home schooled children score on the same standardized tests taken by public school children? Wartes found that home schooled children in Washington scored three to twelve percent above the national norms on the Stanford Achievement Test. He located the median score at the sixty-eighth percentile, in the top thirty-two percent of all students tested (Wartes, 1988). Rakestraw discovered that Alabama home schoolers placed at or above grade level in all subjects (Rakestraw, 1988). Frost and Morris found that children scored two to three grade levels higher on the Iowa Test of Basic Skills than public school children (Frost & Morris, 1988). A study by Ray (1990) revealed that on standardized tests home schooled children performed at or above the eightieth percentile on national norms in the areas of reading, listening, language, math, science, and social studies, while the national average in conventional schools was the fiftieth percentile. Ray reported that there was a minimal relationship between the educational level of parents and student achievement, and that the certification of parents had virtually no impact (Home School Legal Defense Association [HSLDA], 1990; Ray, 1990).
Home school educators have suggested that these positive results may be due to factors other than parent certification. The low teacher-student ratio, the close contact between parent and child, the individualized methodology, and the freedom of the student to interact independently and creatively with the curriculum may be important elements in encouraging academic achievement of home schooled children (HSLDA, 1990; Moore, 1988).
The purpose of this study was to explore four questions relating to home schooling. First, did home schooled children perform below, at, or above grade level on the subtests of the Stanford Achievement Test (SAT)? It was hypothesized that they would score at significantly higher grade levels than the national population. Second, did home schooled children make significantly higher grade equivalent scores on the second SAT, used as a post-test, than on the first SAT, used as a pre-test? It was assumed that the results from the second SAT, taken during the home schooling experience, would indicate that substantial academic progress had been made since the first SAT, taken prior to or at the beginning of the home schooling program. Third, did home schooled children whose parents were not certified teachers do as well on the SAT as those whose parents were certified? It was hypothesized that there would be no difference between the two groups. Finally, was there any difference in the children’s scores, based on their parent’s education level? It was expected that the educational level of the parent would have little or no effect on the children’s SAT scores. Other factors relating to the interactions among parent, child, curriculum, and methodology were assumed to be more efficacious than parent education level.
The population of this study included second through eighth grade Christian children in Texas who had been home schooled for at least one year and had taken the Stanford Achievement Test (SAT) two or more times. The children were located by contacting the leaders of each of the one hundred and nine support groups listed in the 1990 Handbook for Texas Home Schoolers (McCord, 1990) as well as six additional support groups formed after the publication of the Handbook. Since the criteria for participation in the study were so limiting, it was not possible to secure a sample of sufficient size by extracting a random sampling of the population. Therefore, all children who fit the criteria in families willing to participate were included in the study.
Of the 154 questionnaires mailed, 43 were returned, and 38 were usable. Five questionnaires were not valid since one described a first grade child, two listed scores for a test other than the SAT, and two included no scores at all. This 25% rate of return is not abnormally low for a mail survey (Knowles, Mayberry, & Ray, 1991). The return rate may be due to an optimistic response on the part of the support group leaders in requesting questionnaires, to parents’ realization that their children did not fit all the criteria, or to parental unwillingness to complete the questionnaire for some reason. The 38 usable questionnaires yielded 61 Christian home schooled children in the second through eighth grades who had taken the SAT twice. The results of this study can be generalized only to Christian home schooled children in grades 2 through 8 in Texas.
Participating families completed a questionnaire which requested two sets of scores on the SAT. The second SAT had been administered during conventional schooling or at the beginning of home instruction. Thirty-nine per cent of the students completed the post-test one year after the pre-test, 16% after two years, 25% after three years, and the remaining 30% from four to seven years after the pre-test. All students were tested at more advanced levels of the Stanford 7 Plus edition for the post-test.
The seventh edition of the SAT contained six levels of tests for grades one through nine (Primary Levels 1, 2, and 3; Intermediate Levels 1 and 2; and an Advanced Level). The subtests provided scores in word study skills, reading comprehension, vocabulary, listening comprehension, spelling, language, concepts of number, mathematics computation, mathematics applications, social science, science, using information, total reading, total listening, total language, total mathematics, basic battery total, and complete battery total (Gardner, Madden, Rudman, Karlsen, Merwin, Callis, & Collins, 1987). The Stanford 7 Plus norms (Gardner et al.) used in this study were updated in 1986 to reflect more current academic levels in the nation’s schools.
The objectives measured by the SAT were determined by surveying some 25 major textbook series used nationally, with the test items attempting to examine a common core of curriculum. The tests were reviewed by content experts, editors, measurement specialists, teachers, and an independent eight-member panel of minority educators who attempted to eliminate racial, cultural, ethnic, and sex bias (Gardner et al., 1987).
Kuder-Richardson 20 and alternate forms reliability coefficients were reported for the SAT. Of the 280 Kuder Richardson coefficients reported, 68% were above .90, and 97% were above .80. Concerning the 89 alternate forms of coefficients reported, 16% were above .90 and 81% were above .80. Therefore, the various composite and subtest scores were generally satisfactory in terms of reliability (Gardner et al., 1987).
The questionnaire also included questions regarding demographic information, such as educational background of the child and both parents, teacher certification, religious preference, family income, and ethnic background. Parents were requested to classify and rate their reasons for home schooling as well as the instructional methods most often employed in the home school. They were asked to describe the structure of their home school, such as the quantity of time spent in daily, weekly, and yearly instruction, and the child’s interest in self-initiated study and reading. They were also asked to determine the degree of rigidity or flexibility in their approach to utilizing the curriculum.
Leaders from the 115 support groups were contacted by telephone and letter and asked to submit the number of questionnaires needed, based on the stated criteria. They agreed to publish the request for participants in their support group newsletters. Questionnaires were mailed upon request and followed up by postcards and telephone calls. Support groups and questionnaires were assigned a code number to enable the researcher to record the responses and locate missing questionnaires. Home schooling families were identified only by code numbers in order to guarantee complete anonymity.
Returned questionnaires were categorized according to the grade level of the child, as well as the education level and certification of the mother. The two sets of SAT scores were recorded on a separate sheet for analysis. Demographic information and questions regarding the reasons for home schooling and the structure of the school were also recorded.
Of the 25 boys and 36 girls surveyed, all but 15 had received instruction in public, Christian, or private schools prior to home schooling. The average time spent in the home school was slightly over 4 years for both boys and girls. All families, except two, reported that the mother was the principal parent educator. The grandmother was the instructor in one family, while in another both mother and father shared equally in the teaching responsibilities. All the fathers were involved in the home school in some capacity, with more than one-third of them participating several times per week.
Nearly three-quarters of the fathers were professional people. Thirty-eight percent of the families earned between $18,000-$34,000 annually, and more than 43% earned over $50,000. Almost half of the mothers surveyed had received a college diploma, and six had completed work beyond the Bachelor’s degree. All but four of the mothers had attended college for varying durations of time, many having majored in subjects that they would later teach to their children in the home school. Eleven of the mothers were certified teachers and 27 were not.
Parents reported that they had decided to begin home schooling for a number of reasons. Nearly all believed that God had called them to home school, and most wanted to teach Christian principles and family values to their children. Other factors which influenced families to begin home instruction included building family unity and strength, avoiding negative peer pressure in the schools, developing a closer relationship with the children, and protecting them from humanistic education, drugs, alcohol, and premarital sex.
Most of the families surveyed attempted to follow the daily and yearly schedule of their local school systems. They averaged 5 hours of instruction per day and 187 days per year, suspending studies in the summer. The children averaged two hours per day in self-initiated study, and the majority said they greatly enjoyed reading. The most popular curricula were A Beka, Bob Jones, Advanced Training Institute of America (ATIA), and Saxon Mathematics. Mothers who were just beginning to home school tended to follow the suggested lesson plans quite closely, while more experienced parents often added their own creative ideas. Traditional teaching methods, such as silent reading, lecture and explanation, discussion, independent study, oral reading, use of workbooks, and question and answer, were most often used. Less frequently utilized methods included memorization, hands-on activities, visual aids, quizzes, critical thinking activities, games, projects, tests, library research, field trips, written reports, storytelling, art projects, creative writing, drama, puppets, and music.
One hundred-seventeen one-sample t-tests were used across each grade and SAT subtest to determine whether the home school sample differed from the known population. The means and standard deviations were calculated from the grade equivalent scores of the second SAT. Although it is preferable not to average grade equivalent scores to statistically compute gains, this procedure may be done in the absence of outliers which skew the results. There is value in using grade equivalent scores at the end of the distribution (Frisbie, 1992).
Multiple t-tests were used instead of ANOVA and multiple comparison procedures since the unit of interest was the individual comparison, not the entire experiment. The purpose of the statistical analysis was to locate differences where they existed. A comparisonwise procedure can detect differences between paired means more effectively than experimentwise procedures. Therefore, multiple t-tests were employed, in spite of the danger of committing a Type I error, a false positive. Moreover, by detecting differences between pairs, it also avoided the more crucial Type II error of overlooking differences which actually existed (Yount, 1985).
At the .05 level of significance it was determined that, with the exception of Total Listening in the second grade, all calculated values of t were greater than the appropriate critical values of t, thus indicating that home schooled children had scored significantly higher than the national norms on each subtest. They scored between 2 to 6 grade levels above the national norm on all subtests of the SAT with the most dramatic increases being found in grades 6 through 8.
Eighteen paired-samples t-tests were used to determine whether significant progress had been made between the first SAT, used as a pre-test, and the second SAT, used as a post-test. In the paired samples t-tests the mean differences were divided by the standard errors of difference which were computed from the differences between paired scores. In all cases the computed t-values were greater than the critical values on the t-distribution table for appropriate degrees of freedom at the.05 level of significance, indicating that children had made significant progress across the 18 subtests during the home school experience. (See Tables 1 and 2.)
Eighteen t-tests for independent samples were used to compare the SAT grade equivalent scores of children of non-certified parents with those of children of certified parents in order to determine whether teacher certification had any effect on children’s academic achievement. Before computing the value for t, it was necessary to calculate the means and standard deviations on all SAT subtests for non-certified and certified parents. The formula for independent t-tests merges the standard deviations of the two samples into one measure of variability. Since none of the computed t-values exceeded the critical values in the t-distribution table at the appropriate degrees of freedom at the .05 level of significance, it was concluded that teacher certification was not an influencing factor in any of the SAT subtests.
Eighteen one-way analysis of variance calculations were used to determine the effect of parent education levels across the 18 SAT subtests and the four populations of parents C those who had completed high school or less, two years of college, a college degree, and teacher certification. The means and F-ratios were calculated for each subtest of the second SAT. Since none of the F-ratios exceeded the critical values in the F-distribution table at the appropriate degrees of freedom of the numerator and of the denominator at the .05 level of significance, it was concluded that the level of the parent’s education was not a significant factor in the child’s academic progress.
Home schooled children in grades 2 through 8 scored significantly above grade level in all 18 subtests of the Stanford Achievement Test (SAT). The scores of the children surveyed ranged from 2 to 6 grade levels above the grade in which they would be located in school. The subtests included word study skills, reading comprehension, total reading, vocabulary, listening comprehension, total listening, spelling, language, total language, concepts of number, mathematics computation, mathematics applications, total mathematics, social science, science, using information, basic battery total, and complete battery total. The children consistently scored higher in areas related to languageand reading than in the mathematics group of subtests. An
Subtest N Means SD Significant
Word Study Skills 9 9.1 4.0 Yes
Reading Comprehension 14 9.1 3.8 Yes
Total Reading 14 8.9 3.3 Yes
Vocabulary 14 9.6 3.5 Yes
Listening Comprehension 10 8.6 3.6 Yes
Total Listening 10 9.3 3.6 Yes
Spelling 14 8.2 3.6 Yes
Language 11 8.3 3.2 Yes
Total Language 14 8.6 3.4 Yes
Concepts of Number 14 7.4 3.2 Yes
Mathematics Computation 14 7.2 2.8 Yes
Mathematics Applications 14 7.5 2.7 Yes
Total Mathematics 14 7.0 2.3 Yes
Social Studies 12 9.1 2.6 Yes
Science 11 9.0 3.5 Yes
Using Information 10 8.1 3.1 Yes
Basic Battery Total 13 7.7 2.2 Yes
Complete Battery Total 10 7.6 1.7 Yes
Table 1. Means and standard deviations for SAT subtests in grade 4.
Subtest N Means SD Significance
Word Study Skills 9 10.2 3.1 Yes
Reading Comprehension 12 9.3 3.2 Yes
Total Reading 12 9.6 3.1 Yes
Vocabulary 12 8.9 3.0 Yes
Listening Comprehension 8 7.9 3.4 Yes
Total Listening 10 8.3 3.4 Yes
Spelling 12 8.6 3.9 Yes
Language 8 10.9 2.3 Yes
Total Language 11 10.2 3.9 Yes
Concepts of Number 12 8.1 2.6 Yes
Mathematics Computation 12 8.1 3.4 Yes
Mathematics Applications 12 8.7 2.5 Yes
Total Mathematics 12 8.1 2.6 Yes
Social Studies 9 8.7 2.4 Yes
Science 9 9.0 3.2 Yes
Using Information 5 8.3 3.7 Yes
Basic Battery Total 8 8.2 3.0 Yes
Complete Battery Total 7 8.5 3.2 Yes
Table 2. Means and standard deviations for SAT subtests in grade 5.
exception to this conclusion was spelling. In grades 2, 3, 6, 7, and 8, spelling was among the lowest of the subtest scores. However, home schoolers still ranged from 2 to 6 grade levels above the national norms in all subjects. This range was somewhat higher than that found by Rakestraw (1988), as well as by Frost and Morris (1988) who reported that scores on the Iowa Test of Basic Skills ranged from grade level to three grades above the national norm. In the present study, the greatest contrast between grade equivalent scores and the national norm occurred among seventh grade children. They placed from 5 to 6 years above grade level in all subtests.
Children made significant academic progress between the first SAT, administered in conventional schools or home school, and the second SAT, administered during the home school experience. This suggests that children who were taught by parents at home were not academically disadvantaged, as some school boards have claimed (Lines, 1983). On the contrary, these children appeared to have made greater progress at home than in school. This may be partially due to a more progressive learning environment which allows time for free exploration and discussion, continuous and warm interaction between the child and the parent, and an individualized curricular approach which more adequately meets the child’s needs.
There were no significant differences in grade equivalent scores in any SAT subtest for children whose mothers were certified teachers and those who were not. It appeared that previous teacher training had no significant effect on the children’s academic progress. The finding concurs with those of Harris (1988) and Ray (1990; see also HSLDA, 1990). It does not support the assertion of school authorities that home school parents need to be certified in order to effectively teach their own children.
There were no significant differences in children’s SAT scores based on the education level of their parents. Mothers who had not completed high school were able to teach their children just as effectively as mothers with college degrees. This finding agrees to a great extent with Ray (1990) who reported only a slight or negligible correlation between parent education level and student achievement. He suggested that the home education environment may be more conducive to eliminating the effect of deficiencies in the parents’ educational background than is possible in other educational settings.
This study focused on the academic aspects of home schooling. It demonstrated that, for the sample selected, achievement was higher for home schooled children than for traditionally educated students. Although the sample size was comparatively small, the results obtained concurred with other studies of academic outcomes. It is possible that the children were naturally bright students who would have achieved just as much success in other educational settings. Perhaps the religious orientation of the families affected the children’s desire to succeed. Certainly, though, a key factor in the children’s academic achievements was the high interest and involvement of parents in the educational process. This is crucial to children’s success, not only in home schooling, but in all school environments.
As parents seek to raise and educate their children from a Christian perspective, they need to evaluate carefully all of the available options, including the neighborhood public school, Christian school, private school, and home school. It is hoped that this study has contributed to a greater understanding of the possibilities available through home schooling. Parents who are considering teaching their own children should not discount this option simply because they may not be certified teachers. No one understands the academic needs of children better than their own parents. Regardless of the educational setting, parents will always be their children’s most important teachers.
Frisbie, D. A. (1992). Book review of Understanding achievement tests: A guide for school administrators, by L. M. Rudner, J. C. Conoley, & B. S. Plake (Eds.). Journal of Educational Measurement, 29(3), 273-278.
Frost, E. A., & Morris, R. C. (1988). Does home schooling work? Some insights for academic success. Contemporary Education, 59, 223-227.
Gardner, E. F., Madden, R., Rudman, H. C., Karlsen, B., Merwin, J. C., Callis, R., & Collins, C. S. (1987). Stanford 7 plus multilevel norms booklet, national. San Antonio, TX: The Psychological Corporation, Harcourt Brace Jovanovich.
Gorder, C. (1987). Home schools: An alternative. Columbus, OH: Blue Bird Publishers.
Harris, G. (1988). The Christian home school. Brentwood, TN: Wolgemuth & Hyatt.
Home School Legal Defense Association (1990). Initial results from nationwide survey give high marks to home schooling. Home School Court Report, 1-8. (See Ray  for full report.)
Knowles, J. G., Mayberry, M., Ray, B. D. (1991). An assessment of home schools in Nevada, Oregon, Utah, and Washington: Implications for public education and a vehicle for informed policy decisions. United States Department of Education Grant #R117E90220; later published in Mayberry, Knowles, Ray, & Muchmore, (1995), Home Schooling: Parents as Educators, Thousand Oaks, CA: Corwin Press. (Available from the National Home Education Research InstituteCsee Announcements section in this periodical.)
Lines, P. (1983). Private education alternatives and state regulation. Journal of Law and Education, 12, 189-234.
McCord, K. & McCord, B. (1990). Handbook for Texas home schoolers. Austin, TX: H.O.P.E. for Texas.
Moore, R. S. & Moore, D. (1988). Home school burnout. Brentwood, TN: Wolgemuth & Hyatt.
Rakestraw, J. (1988). An analysis of home schooling for elementary school-age children in Alabama. Doctoral dissertation, University of Alabama, Tuscaloosa, AL.
Ray, B. D. (1990). A nationwide study of home education: Family characteristics, legal matters, and student achievement. (Available from the National Home Education Research Institute, c/o Western Baptist College, 5000 Deer Park Dr. S.E., Salem OR 97301.)
Roach, V. (1988). Home schooling in an era of educational reform. School Business Affairs, 54, 11-14.
Sacken, D. M. (1988). Regulating non-public education: A search for just law and policy. American Journal of Education, 96, 394-420.
Wartes, J. (1988). The Washington home school project: Quantitative measures for informing policy decisions. Education and Urban Society, 21, 42-48.
Whitehead, J. W. & Bird, W. R. (1984). Home education and constitutional liberties. Westchester, IL: Crossway Books.
Yount, W. R. (1985). A Monte Carlo analysis of experimentwise and comparisonwise type I error rate of six specified multiple comparison procedures when applied to small k’s and equal and unequal sample sizes. Doctoral dissertation, University of North Texas, Denton, TX.