In this paper, Anderson (2000) explores the main advantages and some possible drawback of computer-based test on language skills. It is limited to test language skills on a computer, like speaking and writing skills which can hardly be assessed. But also, we can see some obvious advantages in computer-based testing. First of all, its flexibility allows tests be freed from traditional paper forms and the formal test administration. Secondly, users do not need to be restricted in a large room with others but on their own, while testers will find it convenient to collect and mark the test data, and the results can also be issued directly. Nowadays, with the access to internet, computer-assisted test is becoming more applicable and flexible. Besides, thanks to the larger capacity of internet database, the testing items and results can be stored for future use, especially for the development of testing, with less worry about the test security. Instead of fixed-format tests, computer-adaptive test shows its advantage in providing test users an adapted difficulty level according to their ability level, which makes the tests efficiently reflect users’ real place with possible pedagogical implications on teaching and learning.
Since this kind of test becomes pretty common in current testing including TOEFL, we’ve already seen the claimed advantages and disadvantages and we still have some doubts about the reliability and validity of computer-assisted testing, like whether the users’ ability can be estimated through computer and how, what feedback or diagnosis can be given to users through computer according to their performance in the test, and how to judge the test method and delivery medium, etc..
Since this kind of test becomes pretty common in current testing including TOEFL, we’ve already seen the claimed advantages and disadvantages and we still have some doubts about the reliability and validity of computer-assisted testing, like whether the users’ ability can be estimated through computer and how, what feedback or diagnosis can be given to users through computer according to their performance in the test, and how to judge the test method and delivery medium, etc..