Addressing Study Error in the National Evaluation of Upward Bound – Do the Conclusions Change?

A number of sampling design and non-sampling error issues combined together to confound some of the major conclusions that have thus far been published concerning Upward Bound (UB). This report examines the Upward Bound National Evaluation random assignment study that followed a multi-grade cohort from 1992-93 to 2003-04, based on four basic assumptions of random assignment studies:

• The sample is representative of the population for which the study is intended to generalize;
• The treatment and control group are equivalent on factors related to outcomes;
• The treatment and control group are treated equally except for the treatment; and
• The treatment and control group are mutually exclusive with regard to the treatment.

Our major finding is that when study error issues are addressed, the Upward Bound program demonstrated statistically significant and substantive positive impacts on the major goals of the program, post-secondary entrance, application for financial aid; and attainment of post-secondary credentials.