Record: Sharon Green-Hennessy, “Homeschooled Adolescents in the United States: Developmental Outcomes” in Journal of Adolescence 37, no. 4 (June 2014): 441-449 [Abstract here]
Summary: Green-Hennessy is a psychology professor at Loyola Maryland. After beginning with a very strong lit review, she describes the methodology of the data set she’ll be using in this study, the National Survey on Drug Use and Health (NSDUH). It is a yearly, nationally representative survey of U.S. household residents age 12 and over. Subjects are interviewed by trained professionals and paid $30 for their trouble, which results in very high response rates (between 69 and 77% during the years Green-Hennessy uses). Green-Hennessy combined the data on children aged 12 to 17 for the years 2002-2011, which gave her 182,351 subjects overall. The demographics of this massive sample reflects the nation at large quite well. Since one of the questions asked on the survey was type of schooling, Green-Hennessy was able to use this data to determine to what degree homeschooling prevents or exacerbates behaviors known to put adolescents at risk for drug use.
Of the 182,351 adolescents surveyed by NSDUH between 2002 and 2011, only 1094, or .6% reported being homeschooled. Since the survey instrument also presents self-reported religious attendance and commitment levels, she was able to subdivide this group into a highly religious cohort (N=658) and a weaker religious cohort (N=436). She did the same for the much larger sample of adolescents attending institutional schools, and then compared all four groups to one another on various measures, controlling all the while for demographic variables like SES, parent marital status, race, and so forth through regression analysis. Here are the results:
Demographically speaking, the religious homeschoolers were more likely to have multiple siblings than either less religious homeschoolers or adolescents attending institutional schools. They were also more likely to be white, female, from two-parent families, and middle class. That’s why logistic regression had to be used.
When such controls were put in place religious homeschoolers were still far less likely than the other groups to report having been arrested or booked (1.6% compared to 3.3% of religious conventionally schooled and 9.7% of less religious homeschooled).
Religious homeschoolers were also far less likely to report a substance disorder (3.0% for religious home vs. 6.1% of religious conventionally schooled and 15.8% of less religious home). Relatedly, a very high percentage of less religious homeschoolers said their parents wouldn’t really disapprove if they used illegal substances (35%), while much smaller percentages of both very religious homeschoolers (6.5%) and religious conventionally schooled (8.5%) said their parents wouldn’t care.
Another common factor that often is associated with negative life outcomes is falling behind grade level. On this variable family income was the dominant predictor, but even when controlling for SES homeschooling still had an impact. Less religious homeschoolers were three times more likely than conventionally schooled adolescents to report being behind their expected grade level for their age. Surprisingly to Green-Hennessy, even highly religious homeschoolers were twice as likely to report being behind grade level. She concludes, “having stronger religious ties appeared to have only had a modest buffering effect on self-reported grade level among homeschoolers.” (p. 445)
Finally, Green-Hennessy looks at extracurricular activities. Less religious homeschoolers were 2.5 times more likely than less religious conventionally schooled adolescents to not participate in any extracurriculars. Contrary to her expectation going into the study, she found that highly religious homeschoolers were 60% less likely to be isolated than highly religious conventionally schooled adolescents. While religious homeschoolers are clearly connected, for a large percentage of them that connection is exclusively church-related (20.7%). Another notable discovery was that a full 55% of religious homeschoolers and 51% of less religious homeschoolers reported involvement in some sort of school-based activity.
In her discussion of this data Green-Hennessy notes that intense religion seems to serve as a powerful buffer for adolescent homeschoolers against substance abuse and delinquent behavior and an encouragement toward extra-curricular activity, at least at church. Contrary to what she expected and to what is often reported in the glowing studies of homeschooler academic achievement funded by advocacy organizations, homeschooled adolescents, whether religious or not, were more likely to be significantly behind their schooled peers in grade level. She calls for more “nationally representative studies using objective assessments of academic outcomes” to see if what the NSDUH data suggests her is really true (p. 446). Finally, she points out that non-religious homeschoolers seem to be legitimately at risk as a group for a number of negative outcomes. They are far more likely than all other groups to experiment with drugs and alcohol, to have run-ins with law enforcement, to fall behind grade level, and to be isolated from others. She wonders if the lax parenting strategy often associated with “unschooling” might be to blame, or if perhaps the data simply reflects the situation of troubled teens leaving school for homeschooling. “If the latter is true,” she notes, “such children may well be struggling academically before homeschooling commences.” (p. 446)
Appraisal: First of all, it needs to be said that this is a marvelous study to have. Green-Hennessy has tapped a previously unused database to give us quantitative information about homeschooling that is not based on convenience samples. Moreover, the data she has uncovered is fascinating because it both reinforces and challenges what other researchers have found. Like Martin-Chang and Hanna she finds that breaking the U.S. homeschooling population into religious and less-religious, or in the venerable categories of pioneer homeschool researcher Jane Van Galen, “ideologues” and “pedagogues,” continues to be a valuable exercise. U. S. homeschoolers still tend to fall on one side or the other of the religious and structured vs. less religious and less structured continuum. Furthermore, as Martin-Chang also found, the less religious and less structured approach seems to be the less effective strategy.
Second, the real bombshell revelation of this article is clearly the data on homeschoolers reporting being behind grade level. Her results here reinforce what another randomized sample, the Cardus Education Survey, also found, which was that homeschooled young adults are not doing as well as their demographically equivalent schooled peers in terms of their academic careers. When this Cardus data first dropped its homeschooling findings were derided by advocates of homeschooling, for the conclusions seemed to contradict the very positive results of decades of advocacy-funded research. This study joins Cardus and the results reported by Hill and Den Dulk in another study based on high quality randomized sampling, all of which suggest that the results of homeschooling might not be especially rosy for everyone. On the other hand, and also like Cardus, Green-Hennessy’s data doesn’t tell us how long these at-risk students have been homeschooling, and her speculation that her data could reflect a trend toward struggling students dropping out of institutional schools to try homeschooling as an alternative may very well be the case. It needs also be kept in mind that we’re talking here about only a small fraction of the overall sample. Green-Hennessy didn’t provide a percentage for religious homeschoolers, but she did say that only 13.6% of less religious homeschoolers reported being 2 or more grades behind their age cohort. While this percentage is 2.5 times higher than the national average, it still means that 86% reported being only a year behind, at, or above grade level. The percentage of religious homeschoolers reporting the same would have been even higher.
Finally, some might wonder at the small percentage of the overall sample who reported homeschooling. The National Center For Education Statistics’ most recent data has 3.4% of the school-age population homeschooling, while this NSDUH data has only .6%. Green-Hennessy herself speculated that it this small number could be partly because her data begins in 2002, when a smaller percentage of children homeschooled (2.2% in 2003, according to the NCES), and partly because recent structural changes (especially the growth of online schooling) have blurred the distinction between home and institutional schooling. Without discounting those possibilities let me also suggest that the most important factor is probably that by age 12-17 most homeschooled children have returned to school. While there has been a trend toward increased support for homeschooling in the higher grades, it remains the case that homeschooling is still a lot more common among young children than among adolescents. The National Survey of Children’s Health used by Cordner begins at age six and thus had a 2.4% homeschooling percentage in 2007.
Milton Gaither, Messiah College