Predicting success in the Pre-International Baccalaureate program

The purpose of this study was to develop a model for predicting student success in the Preliminary International Baccalaureate (Pre-IB) course of study at Atlantic and Suncoast Community High Schools in Palm Beach County, Florida. The objective was to examine the relationship of the five predictor variables, previous year's grade point average (INIGPA), scaled scores from the Reading Comprehension (SSREAD) and Total Mathematics (SSMATH) portions of the Stanford Achievement Test, recommendation ratings (RECOM) and the score received on a writing sample (WRITING), with the criterion, which was the student grade point average at the completion of one year in the program (FINALGPA). The study involved 216 students, 135 ninth graders and 81 tenth graders. All data were analyzed separately for each grade level. The multiple regression analyses determined the unique contribution of each independent variable to predicting the criterion. Each grade level sample was divided into subgroups based on race (black and white students) and gender. The results of this study for the ninth grade sample indicated that INIGPA provided the most statistically significant relationship (p <.0001) to FINALGPA. Similar results occurred for the subgroups male, female and white students. Ninth grade black student results indicated a statistically significant relationship of INIGPA (p <.05) with FINALGPA. For the total ninth grade sample, as well as the subgroups white and female students, SSMATH (p <.01) was also a valid predictor. SSMATH was not a valid predictor for ninth grade males or black students. For the tenth grade sample INIGPA was the most valid predictor (p <.0001) of student success. RECOM (p <.05) was also a significant contributor to the variance in the criterion measure. INIGPA (p <.0001) had the greatest statistical significance in its relationship to the criterion for the subgroups male, female and white students. The subgroup, male students, had two other statistically significant predictors, RECOM (p<.01) and SSREAD (p <.05). The tenth grade black student sample was too small to perform multiple regression analyses.