Construction of Multiple Linear Regression Prediction Model of PRETCO-A Scores and Its Positive Backwash Effect on Teaching

Abstract

PRETCO-A is a standardized English proficiency test set up to evaluate the English application ability of the students in higher vocational college. In order to improve the passing rate, a multiple linear regression prediction model is constructed in this paper. A significance test was first performed on the regression model and the regression coefficient to verify a high correlation among the variables. The confirmed model was then put into application to predict the students’ scores and identify the students who may fail the exam, leading to targeted tutoring assistance given to those students in advance. Finally, 60 students with predicted scores lower than 60 points were selected as research samples, and randomly divided into the control group and the experimental group, 30 students in each group. Finally, the experimental group students were given 40 teaching hours of precision assistance and targeted training, while the control group did not engage in any teaching intervention. The experimental results indicate that the pass rate of experimental group is 20% higher than the control group, which means the backwash effect of the test prediction is positive. The prediction model is proved to be scientific and reliable for teaching.