Record: Rachana Bhatt, “Home is Where the School Is: The Impact of Homeschool Legislation on School Choice” in Journal of School Choice 8, no. 2 (2014): 192-212. [Abstract Here]
Summary: Bhatt, an economics professor at Georgia State University, here presents a sophisticated statistical model to try to determine the degree to which a State’s passage of an explicit law granting homeschooling rights to parents increases the tendency for parents to choose homeschooling.
Bhatt begins with a contextual orientation explaining her concept of “homeschool rights,” by which she means laws explicitly granting homeschooling parents the right to do it. She recognizes that states that do not have explicit homeschooling statutes nevertheless also grant parents this right implicitly through private school equivalence, but she nevertheless thinks that there is something important about these explicit laws.
Most of Bhatt’s data comes from the National Household Education Survey conducted by the National Center for Education Statistics (NCES), which every four years includes questions about homeschooling. I have summarized the most recent NCES data here.
Bhatt takes the NCES data, which includes zip code information, and combines it with information from a couple of other sources to create an impressive database of variables. She then subjects these variables to statistical analysis to try to tease out how much of an effect changes in a State’s homeschooling law might have on a parent’s decision to homeschool. To get there, she has to make a series of approximations. Most important is a binary variable she creates, giving a child a one “if, by the time a child is age three, his/her state of residence has enacted homeschool rights, and zero otherwise.” (p. 201) She then compares children who were three before a state passed such a law with children three or older after a state passed such a law, and with similar children in states that have no explicit law.
What does she find? That for very young children there is a modest difference but for older children there is none at all. Among the youngest cohort who live in states “where homeschool rights legislation was in place by the time they turned three,” there is a 1.4% difference in the number of children who homeschool compared to similar children in states without such a law.
Appraisal: I’d like to make two observations about this article. First, it is clear from her contextual piece at the beginning and her literature review that Bhatt has not read deeply in the homeschooling literature. Without going into too much detail let me briefly note that she misrepresents the history of compulsory schooling, misunderstands the nature of the critique many scholars have made of the studies of Ray and others about academic achievement, and, most importantly, misunderstands the significance of the various homeschool statutes that were passed in the 80s and 90s. I’ll return to that last one in a moment.
My second observation has to do with her actual quantitative study. While I am deeply impressed with Bhatt’s rigorous effort to construct quantitative data to try to answer her question, unfortunately for her it just doesn’t work. She sells her results as well as she can, claiming that she has uncovered something significant, but she really has found nothing at all. Her biggest finding is the 1.4% difference in young children living in states with explicit laws. First of all, even if her guesswork about how a law passed by the time a child is three might affect a parent’s decision making process is correct, 1.4% is just not very much of a difference. In her data, .6% of these extra homeschoolers are coming from public schools, and .8% from private. She notes in passing that such numbers “are not statistically different from zero” (p. 206), but that doesn’t stop her from concluding not only that an explicit law has a “large, positive impact on the likelihood of homeschooling” but that these extra homeschoolers are coming both from public and private schools (p. 209).
There are many ways to criticize Bhatt’s data and the conclusion she reaches, but let me do it the easy way, by using Bhatt’s own data. If you look at the data she presents for other age cohorts you find that a state’s homeschooling law makes 9-12 year-old children 1.8% more likely to attend private schools, with .9% of those children coming from homeschools and .9% coming from public. Of course that conclusion is absurd, but that she found an even stronger relationship here than she did for her own thesis illustrates that what we are talking about here is really just statistical white noise. A 1.4% difference between two groups in a statistical model so full of guesswork as Bhatt has created is meaningless.
The statistics are thus not convincing. But then neither is the entire premise. The premise requires us to assume that when parents are thinking about whether or not to choose homeschooling they are influenced by whether or not their state has an explicit homeschooling law. This is simply not true. Homeschooling rates vary widely by State, and this variance does not correlate at all with statute status. Texas and California, for example, have very high rates of homeschooling though they have no explicit statute (or in Bhatt’s odd terminology, no “homeschooling rights”), while many states that do have statutes have very low rates of homeschooling (many New England States, for example). Furthermore, there are vast differences between the kind of statutes States have, differences that I would hypothesize affect homeschooling rates more than simply whether or not a state has a law. The “law or no law” binary just doesn’t capture the legal situation at all.
To my mind, Bhatt’s article falls into that camp of experiments that ended up going nowhere. She clearly put in a lot of time compiling her statistics, and when the result she obtained was insignificant she couldn’t just admit that her hunch about the significance of explicit homeschooling laws had been wrong. So she wrote up her data and sold her little 1.4% difference as best as she could. I guess it worked out for her since the Journal of School Choice published it. They really shouldn’t have though, at least not in this form. A more accurate conclusion would have been that her original hypothesis proved to be false and that the mere existence of a state statute does not have a measurable impact on homeschooling rates.
Milton Gaither, Messiah College