Reading Prompt #12

Computer based testing is definitely going to be more and more common in the future, however for now, there are still some factors limit its development. The author mentioned its necessary psychometric assumptions, which I don’t really get the meaning, but I think it’s one of the reasons. In the CAT part, there’s a theory called Item Response Theory (IRT). This theory assumes that all the questions can be graded from easy to hard and the test takers have a particular amount of ability to be assessed on the language trait. Grading questions will take extra time to the testing system and it is not certain that the difficulty of the questions is appropriate. For example, the same question may be difficult for one student but be easy for another student. IRT has a model called unidimensionality which is commonly used. It states that all questions on the test assess the same construct. But it is hard to applied in the real world, questions are hard to be made for only one goal, such as fluency or accuracy assessing only in the speaking test. Usually they were mixed in one question, which may violate the assumption of unidimensionality. Another assumption is local independence. It states that each question is independent, if two questions relate to one same passage, then this assumption is violated. With these two violations, the test takers’ ability cannot accurately be assessed. That may cause unfairness. Then the controversy over an appropriate scoring algorithm and the impractical of large bank of test questions also limit the computer based test. I agree with the author’s points because besides all the reasons above, I think computer based test requires a lot of money to develop the hardware and the software, especially for those developing countries or areas.

For the Cummins&Davesne’s reading, I think it is more interesting for me than CBT. To assess a language learners’ progress, the best way is to compare their current level to former achievement. I still remember that several years ago, school used to post the students’ position of ranking among the whole class in the front wall of the classroom. This was thought to be a good way to show the students how much progress you’ve achieved, and stimulate them. However, it turned out that horizontal comparison only worked for some of the students, most of who keep in the top all the time, vertical comparison is better and the progress were no long be post to the whole class.