The proportion of each ethnic group in the two datasets is similar, but the NLTS-2
sample has a slightly higher percentage of Caucasian students and much lower proportion
of Asian American students. This sample also has far more low-income students, a
slightly higher number of upper-income students, and a much smaller proportion of
middle-income students than the ELS sample. The distribution of income in the full
sample of students with any form of disability in the NLTS-2 has a similar distribution to
the ELS dataset, but the NLTS-2 sample still has a slightly larger proportion of lowincome
families. This suggests that the differences in income of the two samples are due
to patterns of diagnosis rather than sampling problems in the dataset. This is also
confirmed by Ong-Dean’s (2009) historical analysis of patterns of diagnosis in that lowincome
students are more likely to be diagnosed instead of addressing other educational
86
issues such as being an English as a second language learner, having behavior problems,
or needing remediation. High-income students are more likely to be diagnosed with a
learning disability as a means of parents trying to secure resources for their students
(Ong-Dean, 2009).