For instance, Regnerus (2000) found that higher levels of per-pupil expenditures were related to higher educational expectations and that attending a school in a low-income neighborhood was related to lower educational expectations in a panel study of high school sophomores. These findings ought to be considered with caution since the study is cross-sectional in design, although the inclusion of a number of individual-, family-, and school-level variables that might be related to students' attendance at different types of schools helps increase our confidence in these findings to a degree.