PSLE Old Scoring System: How T-Scores Worked and Why MOE Replaced Them

Quick Answer: Before 2021, PSLE results were reported as T-scores — a normalised score that compared each student's performance in each subject against the cohort average. T-scores were calculated using a statistical formula that considered both the student's raw mark and the cohort's mean and standard deviation for that subject. The four subject T-scores were added to produce an aggregate score, typically ranging from about 80 to 300, which determined secondary school posting.

The T-score system was in place for decades and became synonymous with the PSLE itself. Many Singaporean parents today sat their own PSLE under the T-score system, which is why questions like "what PSLE score did you get" still evoke aggregate numbers like 230 or 250 — figures that made sense under T-scores but have no equivalent under the current AL system. Understanding the old system matters not for nostalgia but because the differences between T-scores and AL bands explain why MOE made the change and what the new system is designed to fix.

How the T-Score System Actually Worked

A PSLE T-score for a single subject was not the student's raw mark. It was a transformed score calculated using the formula: T = 50 + 10 × (raw mark − cohort mean) ÷ cohort standard deviation.

This formula placed the average student at a T-score of 50 for each subject, with most students falling between 30 and 70 per subject. A student whose raw mark was exactly at the cohort average received a T-score of 50 in that subject. A student who scored one standard deviation above the cohort mean received a T-score of 60. The total aggregate T-score was the sum of four subject T-scores, amplified to a 200–300 scale for easier reading, though the underlying range was roughly 80 to 300.

For parents, the practical difficulty of the T-score system was that no one could predict a child's aggregate from raw marks alone. A child who scored 85/100 in English on a school paper could not say whether that was a "good" PSLE score, because the T-score depended on how everyone else did. This opacity created anxiety: parents and students chased every possible mark because they could not know which mark might tip the cohort ranking in their favour or against them.

Why the T-Score System Was Replaced

MOE identified several structural problems with T-scores. First, the system encouraged excessive competition over marginal mark differences. A student scoring 89 in a subject could receive a meaningfully different T-score from a student scoring 88, even though the one-mark gap reflected little genuine difference in ability. Second, T-scores communicated a false precision — an aggregate of 231 was treated as qualitatively different from 230 — when the underlying measurement was not that precise. Third, the system labelled students with a single number that followed them through posting, streaming, and, in some cases, their own self-perception as a learner.

The replacement — the Achievement Level (AL) system — addresses each of these problems. AL bands group raw marks into eight broad categories, so a difference of two or three marks within the same band has no effect on the reported AL. Posting bands are now wider, so students with AL totals of 4–6, 7–8, and so on have access to the same set of schools, reducing the pressure to chase a specific aggregate. And the shift to Full Subject-Based Banding means the AL subject scores inform subject-level placement rather than a single stream assignment, so a student who is strong in English and weaker in Mathematics can access different levels in each subject from Secondary 1.

T-Score vs AL Score: A Side-by-Side Comparison

What the Old System Got Right — and Why Some Parents Miss It

The T-score system had one genuine strength: fine differentiation at the top. For highly competitive secondary schools where demand far exceeds places, T-scores allowed posting decisions to separate students who were very close in ability. Under the AL system, more students can tie at the same total score, and ties are resolved by citizenship, choice order, and computerised balloting. A parent of a high-achieving child might feel the AL system introduces more uncertainty at the top end, because an AL total of 6 no longer distinguishes between a student who scored 90 in every subject and one who scored 89 — both receive AL 1 in each subject and the same AL total of 4.

MOE's view is that this is a feature, not a bug. Students with raw marks in the 90–100 range in the same subject are similarly capable, and the secondary school they attend should not depend on which of them scored 90 versus 91. The trade-off is real, however: families at the top end now face ballot-based posting more often than under T-scores. Whether this trade-off is acceptable depends on what one believes the PSLE should optimise for — maximum differentiation for the purpose of school allocation, or reduced competition for the purpose of healthier learning. MOE chose the latter.

FAQ

How were PSLE T-scores calculated exactly?

The formula was T = 50 + 10 × (raw mark − cohort mean) ÷ cohort standard deviation. This produced a score centred at 50 per subject. The four subject T-scores were summed and then transformed to a total on a scale where 200 was roughly the overall average, though the exact mapping varied by cohort. A student who scored at the cohort mean in every subject received an aggregate around 200.

Why did a one-mark difference sometimes cause a big T-score change?

Because T-scores were based on standard deviations, a one-mark raw difference near the cohort mean had a small T-score effect, but a one-mark raw difference at the extreme ends of the distribution — where the cohort was sparse — could have a larger T-score effect. This was a mathematical property of the normal distribution, not an intentional design feature, and it is one reason MOE considered the system unnecessarily volatile.

Can I still compare old PSLE scores with new PSLE scores?

No direct conversion exists between T-score aggregates and AL totals because they measure fundamentally different things — one is relative to the cohort, the other is criterion-referenced. A child who would have scored 240 under the T-score system cannot be said to have an AL total of X under the new system, since the AL total depends only on whether raw marks fall within specific bands, not on cohort performance.

Why did MOE choose 2021 as the year to change the scoring system?

MOE announced the change in 2016 to give schools, parents, and students a five-year lead time. The 2021 cohort was the first to enter Primary 1 after the announcement, meaning their entire primary school journey — from P1 onward — took place with the knowledge that AL bands, not T-scores, would be used for PSLE reporting. This avoided the confusion of a mid-stream change.

Is the new AL system easier or harder than the old T-score system?

The examination standard itself did not change. What changed is how the score is calculated and reported. A student who would have scored in the top band under T-scores will score AL 1 under the new system. The difference is that fine variations within the top raw-mark range no longer affect the reported result, which reduces the pressure to push from 90 to 95 in a subject where the child is already strong.

Summary

The PSLE T-score system calculated a relative score per subject based on cohort performance, producing a continuous aggregate that differentiated students down to the decimal point. While it served Singapore's education system for decades, it created pressure to chase marginal marks, produced scores that were opaque to parents and students, and labelled children with a single number. The AL system that replaced it in 2021 trades fine differentiation for broader bands that reduce competition, increase predictability, and align with a secondary school system where students take subjects at different levels rather than being locked into a single stream. For parents who sat the old PSLE themselves, understanding this shift is key to supporting their children under a system that looks similar on the surface but operates on a fundamentally different logic.