This post reviews Eric J. Isenberg, “What Have We Learned About Homeschooling?” in Peabody Journal of Education 82 (2007): 387-409.
Isenberg, affiliated with Mathematica Policy Research, Inc., here tries to synthesize all of the best research on several topics related to homeschooling. After a brief historical orientation he begins by bemoaning the paucity of available data to do rigorous quantitative study of homeschooling. The best available sources are the data collected by some state education departments and the massive cross-sectional National Household Education Survey (NHES). Isenberg uses them and some other less reliable data to summarize what can be known about “how many, why, and how parents homeschool their children.”
About half of the states keep data on their homeschooling populations, but these are of limited reliability since the collection process used by several states is “haphazard,” and in many cases the data does not include homeschooled children who fall under alternative legal categories (e.g. enrolling as a private school) or who have gone underground. As such these records, while helpful, are not the best source for making generalizations.
The best source is the NHES data, and Isenberg spends a good deal of time here explaining the methodology of the phone surveys NHES conducts and describing its homeschooling component. He then proceeds to offer the best answers he can to the various questions he poses by comparing the NHES data to other, less complete data sets.
First, how many children are being homeschooled? While the numbers fluctuate and are only estimates, Isenberg is confident that as of 2006 there were “at least 1 million” homeschoolers. All his data sets show a similar growth pattern as well: high rates of growth from the mid 1980s to a peak in 2003, and either stagnation or even slight decline since then.
How do families homeschool? Isenberg finds that around 55% of homeschooling parents send at least one of their children to a conventional school, though families “who self-report homeschooling for religious reasons are much more likely to homeschool all their children.” About 21% of homeschooled children attend some sort of school part-time. Isenberg finds high rates of home-school attrition: 37% of homeschooled children go back to school after the first year, though the rate is lower for religiously-motivated families. Even among them, however, only 48% are still homeschooling their child after six years (compared with 15% of “secular homeschoolers.”)
Why do parents choose homeschooling? This question is hard for Isenberg to answer given his quantitative approach. His data is limited to self reporting by parents choosing among pre-fabricated options and inferences he can draw from multiple regression analysis of variables like household income, educational attainment, number of children, quality of local schools (based on test scores) and the like. Despite these limits, Isenberg estimates that the most common motivator for homeschooling is to provide a better education for the child, while a “significant minority” of homeschoolers do so for primarily religious reasons, and around 14% do so because of a child with special needs.
One particularly interesting finding from the multiple regression analyses Isenberg performed was that very religious families are less likely to homeschool when they live in school districts with heavy concentrations of Evangelical protestants, either because local public schools are more likely to reflect protestant values in such communities or because the high density allows for a flourishing religious private school scene.
Another factor his analysis reveals to be an important variable in rate of homeschooling is school quality. For example, “A decrease in math scores from 1 standard deviation above the mean to 1 standard deviation below the mean increases homeschooling by 29%” using one data set, and by 20% using another.
Isenberg finds that several family variables affect homeschooling rates. The more children a family has, the more likely it is to homeschool. Rates also go up if more adults (like grandparents) live in the household and can lend a hand. The wealthier a family, the less likely it is to homeschool, because (speculates Isenberg) such families are more likely to be able to buy a nice home in a desirable school district or to send their children to private schools. Perhaps most interesting of all, Isenberg finds that the more educated the mother, the more likely she is to homeschool her children in sixth grade or below, but after that even well-educated mothers are no more likely than others to teach their children at home.
Isenberg wraps up with some observations about children themselves. Older children taught at home are about twice as likely to have special needs as are younger homeschoolers. Boys and girls are homeschooled at about the same rate.
And that’s it. Anyone looking for the most reliable data we’ve got on homeschooling would do well to obtain a copy of this paper. Its summary and bibliography take us to the limit of what quantitative methods can deliver. What they deliver is important and helpful, but, as Isenberg would be the first to admit, it is by no means a complete picture. Quantitative methods, even when the data is more robust than this, does a great job of giving us raw counts and illuminating connections between such variables as social class, race, region, religious affiliation, and educational choice. But it is ham-fisted when trying to probe below the surface to such things as motivation and attitudes, for these things are difficult to reduce to survey-ready questions. For some of the questions Isenberg touches on, a review of some of the rich qualitative research being done on homeschooling families would have enhanced his conclusions considerably. But be that as it may, we should all be thankful for the rigorous statistical genius on display in this fine report.