Before entering today's topic, let us first count the tests I took in order to enter an American university:
- Four SAT tests with essays
- Two SAT II tests
- Seven AP (Advanced Placement) tests
- Seven or Eight TOEFL tests (International Students ¯_(ツ)_/¯)
Simple math, I took four different kinds of tests for twenty times in total to get enrolled in a university in American. All of these lead me to reconsider the purposes of academic tests, which haven't occur to me before.
I define academic test (standardized test) to be:
a mapping from the performance of students into numerical scales, which allows the students as well as the teachers or university admission officers to quantify the knowledge and academic skills the students have.
One of the most frequent objections to academic testing is that the tests cannot usually perfectly represent the full potentials, knowledge, or skills a student has. One possible explanation to this is that there are not unanimous and clear definitions for terms like potentials, knowledge, and skills. In other words, the sources of the mapping are vague. Each of the students might have different understanding of the materials presented by the professors. Or we can say that the information the professors want the students to know cannot be copied into the minds of students due to the deficiencies origin in our information encoding and decoding system (e.g. written language, visuals, sounds, etc.). If I have an idea that I want to deliver to you, the idea will have to be encoded first (e.g. I described my idea in English), then be transmitted (e.g. Through the vibration of substances, or sound) to sensors you have (e.g. ears), and finally your brain will decode this piece of encoded message. However, we can see that during this process, there are various ways that the information could be twisted. For instance, I might not mention the assumptions I have to generate this idea (encoding); you miss hear the description (transmission); or you could understand the language I used (decoding). Therefore, I would say that it is hard to fully transmit a piece of information to another individual or intelligent beings, and the difficulty will rise along with the complexity of the information.
In conclusion, information will be skewed in transmission process when instructors teach students, which is the internal flaw in the test system,
and this will also happen when the students try to transmit the knowledge they have to text or even tests with few multiple choices questions.
Still, the academic/standardized tests do have certain merits, including but not limiting to 1) a unified numeric scale for judgement and 2) cost effective.
The unified numeric scale means that for a certain test in a field, the scores from every participant are on the same numeric scale, meaning that it is very easy for other people who are not familar with the academic field to evaluate the ability of test participants, which often times involved comparison within the test group. A common practice of this is the SAT and TOEFL for American university application process and the Gaokao in China. In this way, the academic institutes delegate the work of evaluating the incoming students to the test itself, which only need simply algebric comparsion. Considering the amount of students each year applying for universities and the lack of admission officers and their ability in judging the academic potentials of students, is an acceptable, but not optimal, solution. This also bring out the second advantage of testing that it is very cost effective compared with the optimal approach. In the optimal setting, to evaluate whether a student is suitable for a university is to put him into actual academic life so that we can know for sure whether he is suitable or not. However, neither the student nor the university itself have the time and money to support such evaluation, especially on the global scale. While on the other hand, testing can be done in just a few hours for each person in a parallel manner. In addition, as the first benefit points out, the numeric scale is very simple for comparing. Therefore the evaluation process is super quick and require minimum resource to complete.
Disclaimer: I might change my opinion on the matter addressed in this article depending on further information received and thus this article only represented my opinion to the issue at the time of publishing/modification.