In the research reported in this dissertation, I explored the issue of why the lowest performing eighth-grade students in some states out perform their peers in other states on the 2005 National Assessment of Educational Progress (NAEP). Specifically, I consider the lowest quartile of students in five of the top performing states and five of the lowest performing states on the 2005 assessment and provide characteristics of these groups along with information that could explain some of the differences between the groups.

NAEP data are based on a stratified, multi-stage sample design and thus special statistical techniques must be used even to compute basic statistical metrics. The data consist of responses to student, teacher, and school questionnaires administered with each assessment. Data also include responses to the mathematics items themselves. With regards to this last data type, I used item-level analysis to examine the degree to which certain item types (e.g., procedural algebra problems versus pattern-finding algebra problems) favor one group versus the other.

A primary result of my analysis is a description (demographics, school environment, etc.) of some of the lowest performing students with regards to the NAEP exam. The two groups were shown to be different from each other as well as different from the national student population. More specifically, I found differences between the two groups with regard to, among other things: student demographics, calculator use, technology use, teacher professional development, and the utilization of federal assistance programs. I argue that by considering the use of federal assistance programs, in conjunction with other variables, a key factor with regard to achievement on the NAEP is not only the socioeconomic status of students, but also the degree to which these students are located in poorer communities. Finally, I found little difference with regard to performance on the various item types except in the area of probability and statistics.