High stakes tests attempt to determine the level of proficiency in the basic skill areas and if the student could be successful in the next course. Existing state standards help to ensure that students have mastered the necessary skills to perform well in future courses. Students today are spending more time than ever preparing for high stakes tests---presumably a side effect of the importance placed on tests and testing today (Jones, Jones, Hardin, Chapman, Yarbrough & Davis, 1999). North Carolina End-of-Course mathematics teachers have helped to standardize these achievement tests by evaluating classroom performance with the students’ actual performance on these tests (North Carolina Department of Pubic Instruction, 2005a). Many studies explain the correlation between gender, ethnicity and SES of the family (Holman, Gonzalez & McNeil, 1993; Kohr, Masters, Coldiron, Blust & Skiffington, 1989; Okpala, Okpala & Smith, 2001; Willie, 2001), but there is a gap in the research which explores the relationship of these variables to the possible significant difference between teacher-predicted grades and actual student performance.

This study analyzed Algebra II EOC scores, Algebra II course grades and teacher predicted course grades for six high schools in an urban North Carolina school district. It also analyzes the perception gap which exists between teacher predictions and actual student performance in mathematics. One-way ANOVA analyses and independent samples t-tests were computed to analyze the mean difference relationships between the variables and influences of gender, ethnicity and socioeconomic status. The quantitative findings of this study tell us that ethnicity, gender and SES are highly predictive of student performance in mathematics. The data show that teachers are accurate predictors of student achievement and add some validation to the significance and value of teacher grading and assessment in present and future academic achievement.