Converting university admission scores in 2025: Why is it becoming a chaotic matrix?

(Dan Tri) - Using an inadequate benchmark as the basis for converting all other admission methods is like building a house on sand.

The 2025 admissions season is underway, marking the peak of “multi-method admissions”. Candidates can use high school graduation exam scores, academic records, results of competency-thinking assessment tests, or international certificates...

However, after all, an undeniable fact is: most of these methods, whether directly or indirectly, are referenced to the combined score of the high school graduation exam. However, this reference system is unstable, not standardized and not sufficiently reliable in terms of science .

The story of difficult exam questions that are not suitable for teaching and learning practices is a topic that has received much debate after the 2025 high school graduation exam (Illustration: Bao Quyen).

The reference system seems diverse but is actually single-line.

On the surface, the 2025 university entrance exam gives the impression of a diverse and flexible system. But in reality, these methods do not operate independently, but are all implicitly converted to the same standard axis: the 30-point scale of the high school graduation exam.

From school report cards (10-point scale), ability assessment tests (1,200-point or 150-point scale...), thinking assessment (100-point scale), to foreign language certificates such as IELTS or SAT, ACT, ACT... All are interpolated or percentile-ized to be compatible with high school exam scores.

This single-line reference has lost the independent evaluation value of each form. The system, which was designed to increase flexibility, is being constrained by a common conversion standard.

High school graduation exam questions have not been standardized - the benchmark is "drifting"

One of the prerequisites of any conversion is that the reference system must be stable and standardized. However, the current high school graduation exam has not yet reached that level.

According to the theory of assessment measurement, standardized tests must go through pre-testing steps, adjusting difficulty, ensuring classification ability and stability between years.

However, the high school graduation exam was announced as the "first time" in recent years that it did not use a question bank, the matrix was randomly generated, and the question level could be changed and the exam questions were created using expert methods.

This leads to the 2025 high school graduation exam in some subjects having a very large difference in difficulty, such as in math, only 12% of candidates achieved 7 points or more, and in English, more than 15%. Meanwhile, in other subjects, this rate is quite high, such as physics, more than 53%, chemistry, more than 33%...

Lack of cross-sectional data - cannot establish reliable correlation

To convert between forms, cross-sectional data is needed - that is, groups of candidates participating in multiple methods simultaneously to establish correlations. But in reality, the number of candidates taking both the competency assessment and high school graduation exams is very small, not enough to create reliable conversion matrices. The fact that candidates only take one or two combinations makes the conversion model even more inaccurate.

In addition, technical data such as regional score distribution, question discrimination coefficient, or difficulty of each exam are not published. Therefore, all conversion models are just subjective estimates, lacking scientific verification.

GPA - correlation data not strong enough to convert

Academic transcripts are currently a popular admission method, but many studies show that the correlation coefficient between academic transcript scores and high school graduation exam scores is only 0.4-0.6, showing that the level of reflection of actual capacity is still limited.

However, if we only rely on that to evaluate the transcript, it is one-sided, because the high school exam itself has not been standardized in terms of content and difficulty.

The high school transcript admission method is used by many universities (Photo: Lan Phuong).

It is also undeniable that current academic transcript data still has some shortcomings. The situation of “excellent score inflation” in some schools has lost its classification, making high scores so common that they no longer have any distinguishing meaning.

More seriously, some training institutions also convert transcripts to a 30-point scale or add priority equivalent to test scores, which inadvertently equates transcripts with standardized test results.

There is no unified conversion standard - different places have different models.

Currently, there are many different conversions in use, some schools use percentiles, others use Z-score normalization, and some convert to the average score of the admitted group. Each method has its pros and cons and depends on the input data set.

Some schools this year have not even converted scores even though they enroll students using many high school graduation exam combinations. This leads to the same candidate being evaluated differently depending on the admission location and conversion method.

As a result, each school and each industry has its own way of converting, leading to a situation where everyone does their own thing. In particular, when using the same results of their own exams, each school has a different way of converting scores and the conversion rates are also not the same. Parents and candidates cannot fully understand the specific principles to make a fair comparison between the admission forms.

When the foundation is shaky

Any conversion system is only meaningful when the reference system is reliable, stable and standardized. Using an inadequate benchmark as the basis for conversion for all other selection methods is like building a house on sand.

Parents and candidates listen to 2025 university admissions consultation at Ho Chi Minh City University of Economics and Finance (Photo: Phong Doan).

The predicted consequence is that admissions may be unfair. Candidates who study regularly and consistently may be disadvantaged by those who “know the test well,” and those with outstanding thinking abilities may be undervalued if they take the wrong exam.

Recruitment is no longer about finding truly talented people, but has become a game of guessing the correct conversion formula, an approach that is not truly scientific and perhaps unfair in recruitment.

To achieve a fair and transparent admission system, it is necessary to start with standardizing the test questions, publicizing technical data and building a scientifically based national reference system. Only then will the admission methods be truly independent and still be able to be compared reliably.

Score conversion can be useful, but only if the assessment platform is standardized and the data system is reliable. To improve substantially, we must start from the root: standardize the test, make the data transparent, limit the combination of subjects to the subject, and have consistent direction from the management level.

Phan Anh

Expert in educational assessment and testing

Source: https://dantri.com.vn/giao-duc/quy-doi-diem-xet-tuyen-dai-hoc-2025-vi-sao-dang-tro-thanh-ma-tran-hon-loan-20250802222232417.htm