Currently, almost every institution of higher learning has computers for English teaching, and quite a few universities or colleges have implemented computer scoring of students' test papers, but most of the English teaching units have only utilized the most basic data of computer scoring - the scores of every student in a particular test, beyond which there is nothing about further analyses and studies of the data. In this way, using these data to guide classroom teaching has not been fully exploited, resulting in the waste of these precious resources. The author of this paper, by combining the testing theory with computer programming, according to each student's or all the students' different responses in a class to each item of a particular set of test paper, has analyzed, from different angles, the guiding role of test item's facility value, discrimination index etc. in classroom teaching.