Predicting performance by university students in introductory computing courses
暂无分享,去创建一个
Psychological testing as a method o f selection for computing personnel is no w the subject of a large literature [7]. Most of this is concerned with the use , validation and standardization of tests , or test batteries which may be used t o select potential computing personnel fro m a population of applicants for variou s positions in data and information processing departments. The validation o f this type of test, when used to selec t personnel for jobs requiring particula r skills or attributes, involves the demonstration by means of statistical test s that a relationship exists between tes t scores and some criterion of success o n the job. Sometimes in the process o f validation, correlations are made between test scores and class grades o f students in programming courses. Fo r instance, the IBM Aptitude Test fo r Programmer Personnel (ATPP) scores wer e compared with grades in programmin g courses for 610 students [3], and a considerable amount of research ha s since been done to relate results o n this, and other tests, to the subsequent job performances of the teste d population [7]. The use of such tests to predic t performance in computing courses in a college or university environment i s of obvious concern to faculty teachin g in these areas. Some work has bee n done in this field. SAVILLE [6] give s expectancy tables between the Compute r Programmer Aptitude Battery diagram-ming scores and examination performanc e in a special degree paper in compute r science. A correlation coefficient , r-.53, was computed and found significant at the .001 level. BATEMAN [1 ] uses a regression equation with 16 independent variables to predict the grade s of students in an introductory FORTRAN I V course. Three of the independent variable s are the IBM ATPP subtest scores ; ATPP1-the letter series test, ATPP2-th e figure series test and ATPP3-the arith-metical reasoning test. The equation ha s a coefficient of determination R 2 = .60, i .e. 60o of the variability of the students ' grades is explained by the variables included in the equation. . Use of this regression model to predict student grades in a subsequent class gives a highly significan t correlation coefficient (r = .70) betwee n actual and predicted grades. NEWSTED [5 ] derives two equations to predict grade s and abilities of …
[1] E. E. Grant,et al. Exploratory experimental studies comparing online and offline programming performance , 1968, CACM.
[2] Peter R. Newsted,et al. Grade and ability predictions in an introductory programming course , 1975, SGCS.