'Difficult exam questions' heat up forums: Exam question making process needs to be professionalized

According to Mr. NTD, the current test-making process depends too much on personal feelings and experience. Although they are highly qualified teachers, many people in the test-making team lack expertise in testing, and do not even know anything about basic statistical methods to assess the difficulty of a question or the effectiveness of interference options. This turns the test-making process into a manual, unscientific process; causing controversy about the level of difficulty, ease, shortness, length, and potential risks.

Candidates for high school graduation exams Math and English complained that it was difficult. Photo: Nhu Y

The consequence of this is “perceptual” difficulty: A question is considered “difficult” or “easy” based on the subjective perception of the test taker, or experience from previous years. However, this perception can be completely different from reality when applied to tens of thousands of candidates. Without indicators such as the difficulty index (p-value) calculated from real data, we cannot know exactly how “difficult” or “easy” a question is compared to the overall ability of the candidates.

Distractor analysis is useless. Without statistical analysis of the effectiveness of distractor analysis, incorrect choices can become too obvious or, conversely, too deceptive to reflect true ability. This makes the question lose its discriminatory power, turning it into a chance question or a simple test of memory.

The question lacks discrimination. Without this index, we cannot know whether a question is actually doing a good job of filtering. A question that is too easy for anyone to do, or too difficult for anyone to do, is a poor discriminator.

When setting questions based on personal experience, test makers may unintentionally include questions that are biased towards a certain aspect of knowledge that they are familiar with, or use expressions that are only accessible to a certain group of candidates. This undermines the fairness of the exam and is highly risky.

The solution to this problem, according to Mr. NTD, requires a revolution in the approach to test making, towards professionalization and scientific application. This includes in-depth training in testing; building a standardized question bank; professional test making process; and the participation of statistical experts.

Moving away from emotions and experiences to a professional test-making process based on statistical data not only improves the quality of test questions but also strengthens society's trust in the fairness and transparency of entrance exams.

"Dissecting" the 4 causes

Dr. Tran Nam Ha, an examination expert, analyzed the reasons why the exam questions for some subjects of the 2025 high school graduation exam are "absolutely" difficult.

According to Dr. Nam Ha, the first reason is the mismatch between the exam objectives and the exam structure. The number of questions at the application and high application levels accounts for a large proportion, while the knowledge and understanding questions, which are the “support” for average students, have decreased sharply. This bias shows that the exam is aiming for classification instead of graduation.

More importantly, the way questions are asked and materials are selected in many academic subjects that are deep, lengthy, and unfamiliar - especially in Literature and English - causing students to spend a lot of time reading and understanding, leaving them with insufficient energy to solve the core requirements.

The second is the error in understanding and applying the test matrix. The test matrix, which is supposed to be a tool to ensure balance in content and cognitive level, has been misunderstood and applied incorrectly. Many units consider the matrix to be just a table of question number distribution, from which they use random drawing software while ignoring the detailed specifications for each question.

Lacking control over the competencies to be assessed, the difficulty and the discrimination, the test becomes a “lottery product” instead of a scientific assessment tool. More worryingly, the confusion between the matrix and the test specifications has caused the test construction process to be completely detached from the requirements of the 2018education program.

Third, another core weakness is the absence of a national standardized question bank. Current question-setting still relies heavily on expert experience without empirical data on difficulty or discrimination. This makes the quality of questions dependent on personal feelings and experience, leading to strong fluctuations between years and between exam codes.

As a result, average high school students, the main target of the graduation exam, no longer have the opportunity to access the exam questions fairly.

The final cause is inconsistency in policy communication and technical procedures. In February, representatives of the Ministry of Education and Training announced the construction of a question bank for the exam according to the new program.

But just a few months later, at a press conference (after the exam), a leader stated that the first year “no test bank was used.” Inconsistency in policy communication not only undermines social trust but also misdirects teachers and students’ preparation.

Nghiem Hue

Source: https://tienphong.vn/de-thi-kho-nong-bong-cac-dien-dan-quy-trinh-ra-de-thi-can-duoc-chuyen-nghiep-hoa-post1756576.tpo