The previous discussion of homeschooling rates in 2003 (see table 2) demonstrated a number of bivariate relationships between homeschooling and student, family, or household characteristics. For example, White students were more likely to be homeschooled than Black or Hispanic students, and students in households with three or more children were more likely to be homeschooled than were students with fewer siblings. Bivariate relationships are important because they describe homeschooling rates among different segments of the population.
It is also worthwhile, however, to use multivariate analysis to test relationships among several variables simultaneously. Multivariate analysis can provide answers to questions about homeschooling, like, “How is a child’s race/ethnicity associated with homeschooling, holding other factors, such as income, constant?” Table 7 presents results from a multivariate logistic regression analysis in the form of odds ratios (beta coefficients and standard errors are available in table A7). Reference categories, which are listed first for each variable in table 7, have an odds ratio of one. For each variable, if a category has an odds ratio greater than one, then students in that category are more likely to be homeschooled than students in the reference category. For example, a category with an odds ratio of 2 means that the odds of a student in that category being homeschooled are twice that of the student in the reference category. If a category has an odds ratio that is less than one, then students in that category have lower odds of being homeschooled than students in the reference category. Asterisks (*) are used to denote findings that are statistically significant at the level of 0.05.
A number of the bivariate findings discussed previously persisted when using multivariate analysis. After controlling for the student, family, and household characteristics listed in table 7, the following relationships were evident: White students were about 4 times more likely to be homeschooled than were Hispanic students; students in households with three or more children were about 2 times more likely to be homeschooled than were students with no siblings; and students in two-parent households where one parent was in the labor force were about 5 times more likely to be homeschooled than were students in two-parent households where both parents were in the labor force.
Three bivariate associations discussed earlier were not detected after controlling for the other factors listed in table 7. Differences between White and Black students, students in two parent households and students in one-parent households, and parent educational attainment found in the bivariate tests, were not detected in the multivariate analysis.