Two new reports were released in March 1988 from the Washington Homeschool Research Project. The first, Report from the 1987 Washington homeschool testing, contains a detailed tabulation of the scores of 873 home schoolers on the Stanford Achievement Test. The second, The relationship of selected input variables to academic achievement among Washington’s homeschoolers, is based upon a sampling of 470 and presents an analysis of test scores as a function of parent education level, degree of structure in the home schooling, number of years home schooled, and other variables.
Sampling and Instrument
One aspect of Washington’s law is the requirement for an annual assessment of home schoolers by a qualified person. A number of people around the state have gone into the business of offering testing services in order to meet this demand. The research method utilized here simply involves tapping into the test scores of those home schoolers utilizing these services. Each of six testing services forwarded a set of scores for each home schooler who was tested; thus this sampling represents a 100% reporting of scores for these six organizations.
All six testing services utilized the Stanford Achievement Test series. The series consists of the Stanford Early School Achievement Test, 2nd edition for grade K, the Stanford Achievement Test, 7th edition for grades 1 to 8, and the Test of Academic Skills, 2nd edition for grades 9 to 12. 1986 norms were used.
Parents of home educated children utilizing these services were also asked to fill out a questionnaire dealing with various aspects of the family or their home schooling. Results from these questionnaires are another important component of this research.
Results and Discussion
Ten findings from the research follow. The findings are related to the achievement scores and the selected variables and their relationship to achievement scores. Percentiles, normal curve equivalents, t-tests, analysis of variance, and simple linear regression were used to analyze the data. The alpha level was set at .01.
In the 1987 sampling of 873 students, the median scores were in the range of the 65 to 66%ile range. The highest scores were in the area of science (70%ile) and in the verbal areas of listening (71%ile), vocabulary (79%ile), and word reading (76%ile). The lowest scores were in math computation (42%ile). Math application scores, on the other hand, were notably stronger (65%ile).
It is apparent that this sampling of home schoolers, as a group, did well. Fears that home schooling children are at an academic disadvantage compared to conventionally educated students are not confirmed.
Selected Variables and Achievement
A statistically significant positive relationship merged between parent education level and test scores, but its magnitude was weak. Parent education level is not a good predictor of test scores. Interestingly, in this sample, children of parents who have only a twelfth grade education are, as a group, scoring slightly above the national norm.
Because of the weak relationship shown here, policy decisions that would limit access to home schooling based upon some arbitrary level of parent education are likely to run the risk of limiting a significant number of home schoolers who are, in fact, doing fine. Thus, such policy decisions are not supported by this evidence.
Due to the small number of home school students who have contact with a certified teacher appearing in the sample, no comparison between those who have teacher contact to those who do not have teacher contact was possible. However, examination of the high extreme in teacher contact (i.e., home school children of certified teachers) produced mixed results. Seventeen percent of the parents in the sample are teachers. In general, children of teachers did outscore children of non-teachers. However, this result did not hold when one considers students who have been home schooled two years or longer. It also did not hold when non-teacher parents having a roughly equivalent level of education to ta teacher were compared. At the opposite extreme on a teacher contact continuum, children who had no teacher contact at all scored, as a group, at the 70%ile on national norms.
This latter finding suggests that contact with a certified teacher is not a necessary component of academic success. Policy decisions that would, as a general matter, require contact with a certified teacher as a condition to home school are not supported by this data.
Within this sample, home educators rated the level of “structure” in their style of home schooling as slightly toward the structured side of middle on a very unstructured to very structured continuum. The number of hours per week of “structured schooling” increases somewhat with student age and averages 16 hours/week. The test data suggests that there is virtually no relationship between level of structure and academic outcomes. The level of structure has no value in predicting test scores. Policy decisions that might impose a curriculum for the sake of its structure and/or minimum hours per week of formal schooling are not supported by these data.
No relationship was found between academic outcomes and the number of consecutive years the student had been home schooled.
A regression analysis of the data suggested no relationship between academic outcomes and the grade level of the home school student within the K to 9 grade range. The analysis was restricted to K to 9 because of the small sample size at grades 10 to 12. However, the 28 students appearing in the sample in grades 10 to 12 appear to be doing well; their mean score was at the 72%ile on national norms.
There was no relationship between family income and academic outcomes of home school students.
This sampling, using a secular measure of academic outcomes (Stanford Achievement Test), provides no evidence supporting a relationship between the degree of religious content in the home schooling and achievement test scores in general, or in social science or science scores in particular.
Students in this sample who will discontinue home schooling scored, as a group, in the 65 to 70%ile range on national norms. One would expect that they will not have academic difficulties in adjusting to a more conventional school setting. The small sampling, however, cautions against generalization of this finding at this time.
Among home schoolers who have previously been in conventional education, the slight majority were in the public schools (54% from public schools and 46% from private schools). Interestingly, when one considers that only 8 to 9% of Washington’s school age children attend private schools, it is apparent that, on a proportional basis, formerly private school students are considerably over represented among home schoolers.
The report goes on to provide findings on eight additional topics. They are (1) people who have taken a class in home based education, (2) the relationship between religious content and amount of structure, (3) the relationship between parent education level and family income, (4) the age distribution of home schoolers, (5) the number of handicapped home school students, (6) the proportion of two-parent families, (7) scores on parent administered versus non-parent administered tests, and (8) differences in test scores between those who did not respond to the questionnaire versus those who did respond.
The preceding findings are highlights of those resulting from the research project. It is prudent to point out that the findings should be taken as the first word rather than the last word on these topics. Replication studies would be valuable.
The Washington Homeschool Research Project was created as a cooperative and volunteer effort on the part of 18 individuals (including home schoolers and several public school educators) to gather objective information about Washington’s home schoolers.
Several detailed reports are available from the author at 16109 N.E. 169 Place, Woodinville WA 98072. Write for information.