[PLEASE READ FIRST]
This app does not judge anyone's chances at any school, and it holds no list of
school or university scores. It works out a T-score from the numbers you type.
The "top X%" figure is a guide that assumes the marks follow a normal
distribution.
──────────
🔢 Standard Score Calculator — a T-score is not the height of a mark
Eighty marks. On its own that settles nothing.
Mean 60, standard deviation 15 → T-score 63.3
Mean 70, standard deviation 5 → T-score 70
The same 80, two different answers. A T-score reads where a mark sits inside
its group, not how high it is. That is why this app asks for three numbers.
■ A test taken by two people always gives 60 and 40
As long as the standard deviation is the population one — half the gap (÷ n).
A hundred and a zero, fifty-one and fifty: the same answer. The mean sits
halfway between the marks, so the higher is one deviation above and the lower
one below. With the sample deviation (n−1, Excel's STDEV) the same two come out
at 57.1 and 42.9.
Nothing shows more quickly that a T-score reads position and nothing else.
■ One Japanese word, two numbers
z-score (mark − mean) ÷ SD 0 at the mean, 1 one deviation above
T-score 50 + 10 × z 50 at the mean, 60 one deviation above
T-score (50 + 10z) is used in Japan for exam scores, and in psychological
testing elsewhere. The Japanese word for "standard score" covers both numbers,
while in English it normally means the z-score — easy to mix up: one deviation
above the mean is "1" on one scale, "60" on the other.
This app never asks you to pick one. It shows both at once.
■ Three topics
▪ Find the T-score … mark, mean and SD give the T-score and the z-score
▪ Mark needed … target T-score, mean and SD give the mark and the gap
▪ Top % … a T-score gives the top per cent and the rank out of 100
■ No ceiling and no floor
A T-score can go above 100 and below 0. If one person in a group of n stands
right out on their own, and the deviation is the population one (dividing by n),
the highest they can reach is 50 + 10√(n−1).
2 people 60
10 people 80
100 people 149.5
1000 people 366.1
With the sample deviation (n−1) the highest out of a hundred is 149.0 instead.
If one person out of a hundred gets full marks and the other ninety-nine score
the same, that person is at 149.5 and the rest at 49.
That is why the mark for a target T-score can come out above full marks; this
app does not know what they are.
■ Top per cent
T-score 30 97.72%
T-score 40 84.13%
T-score 50 50%
T-score 60 15.87%
T-score 70 2.28%
T-score 80 0.13%
An eleven-row chart in steps of five sits on the same screen, with the rank out
of a hundred beside it. These figures assume a normal distribution, which real
marks never quite follow, so treat them as a rough guide — furthest off when
the group is small.
■ When the standard deviation is zero
If everyone scored the same there is no T-score: you would be dividing by zero.
Type a zero and the app stops and says why.
The other way round: the smaller the deviation, the further one mark moves it.
With a mean of 60 and a deviation of 2, 62 is already a T-score of 60.
■ What this app does not do
▪ It does not estimate anyone's chances at a school
▪ It holds no list of school or university scores
▪ It keeps no record of results over time
▪ It does not work out a mean and SD from a list of marks
▪ It does not distinguish sample and population deviation; it uses what you type
▪ It draws no bell curve. Numbers only
No account is needed.
The calculation happens on your device, and nothing you type is sent out.
■ What the screen holds
▪ One to three input fields. No settings screen
▪ Common values are one tap away
▪ The last five entries stay in the history; results copy out in one tap
▪ A quick chart on the top-per-cent topic, and six short notes
▪ Japanese and English, chosen by your device language
──────────
For when a result lands and you want to know where the number came from.
NISAKE GAME
Mark, mean and SD in: T-score and z-score out. A T-score of 60 is top 15.87%