Date of Award


Degree Name

Educational Leadership


College of Education and Professional Development

Type of Degree


Document Type


First Advisor

Ernest R. Goeres

Second Advisor

William G. Monahan

Third Advisor

Lawrence D. Romboski

Fourth Advisor

Edwin R. Smith

Fifth Advisor

John O. Andes


The problem was to determine the feasibility of predicting a student's success specific college, on the basis of data available at the time of the student's admission to the college. In addition to considering the development of a regression equation which could be applied to the data of any undergraduate entering the college, the feasibility of developing a statistically significant prediction equation at the school and major level was also examined.

The study was carried out in the following manner. Stepwise multiple regression was used on the admissions data of the graduating class of the 1978-1979 academic year. The cumulative grade point average at the time of graduation was used as the dependent variable and sixteen independent variables were considered in the development of the equation. This prediction equation was then developed a second time thru stepwise multiple regression allowing only the first six independent variables which were added in the first procedure to be considered in the second equation.

This six variable prediction equation was then used to predict the cumulative grade point average for the entire graduating class of 1979-1980 using admissions data for the six independent variables• Since the final grade point average was already available for this class, the predicted values were compared to the actual values.

This comparison was made by performing a correlation analysis on the two sets of values, the predicted and observed grade point averages. The correlation analysis indicated a high level of correlation between the two sets of values. The correlation coefficient was .67 significance at the .001 level.

The procedures outlined above which were used on the data for the entire graduating classes of the two periods were then repeated for certain subgroups of the 1978-1979 graduating class and then tested upon the corresponding subgroups of the 1979-1980 graduating class. These subgroups consisted of students in each school of the college and students in certain majors where the number of cases was high enough (greater than 24).

Based upon the results obtained, the following conclusion appears warranted• There was a significant correlation between actual grade point averages and predicted grade point averages using equations developed for the entire graduating class and also for subgroups consisting of students belonging to the same school of the college and even for subgroups consisting of students in the same major. It therefore appears that the development of prediction equations at the specific college tested in the study is feasible.


The University of West Virginia College of Graduate Studies became the WV Graduate College in 1992 and was subsequently merged with Marshall University in 1997.


Universities and colleges – Admission.

College students – Rating of – Mathematical models.