You scored 80 on math and 75 on English — yet the English deviation score came out higher, because the deviation score measures relative position within the group, not raw points. This article works through the formula with substituted examples, finds the mean and standard deviation from raw scores, and explains the z-score connection.
The definition
The deviation score standardizes a raw score so that the mean becomes 50 and one standard deviation becomes 10 units:
Deviation score = 50 + 10 × (x − μ) ÷ σ （μ is the mean, σ the standard deviation, x your score）
Two parts: (x − μ) ÷ σ counts how many standard deviations your score sits from the mean (the z-score), and 50 + 10 × rescales that onto a ruler centered at 50. So 50 is the mean, 60 is one standard deviation above, 40 one below.
When the mean and standard deviation are known
Example: scoring 80 on a test with mean 60, σ 10
Deviation score = 50 + 10 × (80 − 60) ÷ 10 = 50 + 20 = 70
Three steps: subtract the mean (80 − 60 = 20), divide by σ (20 ÷ 10 = 2), then calculate 50 + 10 × 2 = 70.
Scoring exactly the mean
Deviation score = 50 + 10 × (60 − 60) ÷ 10 = 50 + 0 = 50
Whenever x equals μ, the deviation score is exactly 50.
Finding the mean and standard deviation yourself
When an exam report does not list them, compute both from the scores.
Mean μ = (sum of all scores) ÷ (number of test takers). For five scores 40, 50, 60, 70, 80: μ = 300 ÷ 5 = 60.
Standard deviation σ is the square root of the average of squared deviations from the mean:
- Deviations: −20, −10, 0, 10, 20
- Squared: 400, 100, 0, 100, 400
- Average: 1000 ÷ 5 = 200
- Square root: σ = √200 ≈ 14.14
The OpenStax statistics text (§2.7) introduces the standard deviation this same way. Pasting the scores into the statistics calculator produces both values instantly. When an exam report lists them, use those — they cover the full population of takers, while hand-computed values are sample estimates.
The z-score connection
The z-score expresses the same distance in standard-deviation units:
z = (x − μ) ÷ σ ⇔ Deviation score = 50 + 10 × z
For the earlier example: z = 2 and deviation score = 70. A z-score lives in a world with mean 0 and σ 1; the deviation score re-centers it at 50 and scales by 10. International statistics use z-scores, Japanese exams use deviation scores — same idea, different ruler. The sign helps too: z = +2 is two σ above the mean, z = −1.5 is 1.5 below (deviation scores 70 and 35).
Rough guide under a normal distribution
| Deviation score | Approximate position (assuming normality) |
|---|---|
| Below 40 | Bottom ~16% |
| 40–60 | The middle ~68% |
| Above 60 | Top ~16% |
| Above 70 | Top ~2% |
The 68-16-2 pattern comes from the empirical rule (68-95-99.7): about 68% of values lie within ±1σ and 95% within ±2σ. A deviation score of 60 sits at mean + 1σ and 70 at mean + 2σ; real exams only approximate this.
Why the same 80 gives different deviation scores
| Test | μ and σ | Deviation score for 80 |
|---|---|---|
| A (easy) | μ75, σ5 | 60 |
| B (typical) | μ60, σ10 | 70 |
| C (hard) | μ40, σ15 | 76.67 |
On the hard test the same 80 sits more than two σ above the mean; on the easy test it is barely above average. Raw scores measure absolute performance; deviation scores measure relative performance.
Three common mistakes: reading a deviation score of 70 as "70 points" (it means mean + 2σ, not a raw score); computing with σ = 0 (the division is undefined — everyone scoring identically has no deviation score); and comparing deviation scores across different populations directly (a 60 in one mock exam and a 60 in another reflect slightly different positions when the test-taker pools differ).
Verify with the tool
Tools Hub’s Deviation Score Calculator takes the mean, standard deviation and your score, and returns the deviation score with the substituted formula shown.
How to use it (3 steps)
Enter the three values
Type the mean μ, standard deviation σ and your score x from the exam report. Preset buttons (60±10, 80 points, etc.) are included.
Read the deviation score
The result appears in large type — 50 means the mean, 60 means mean + 1σ.
Check the working
The Working panel shows 50 + 10 × (x − μ) ÷ σ with your numbers substituted, so each hand-calculation step can be compared.
The tool featured in this article
Deviation Score Calculator
Deviation score from mean, standard deviation and score with substituted working. Zero-σ errors handled — free and browser-based.
To reproduce the example, enter 60, 10 and 80: the tool shows 70 with the substituted working. Without listed values, first paste the scores into the statistics calculator, then feed its output here.
Summary
- Deviation score = 50 + 10 × (x − μ) ÷ σ; 50 is the mean and each 10 is one standard deviation
- Without published values, compute the mean and standard deviation from the raw scores first (the statistics calculator does it instantly)
- z = (x − μ) ÷ σ; the deviation score is 50 + 10 × z — same idea, different scale
- Above 60 ≈ top 16%, above 70 ≈ top 2% under normality (empirical rule)
- The same raw score yields different deviation scores on different tests — relative, not absolute
FAQ
What level is a deviation score of 50?
Exactly the mean. About half the group scored above and half below.
Can deviation scores be compared across subjects?
Yes. Because every subject is standardized to mean 50 and σ 10, all subjects share the same scale — that is how an English 75 can outweigh a math 80.
Does the deviation score change with the population?
Yes. A school test with 50 takers and a national mock exam with tens of thousands have different means and spreads, so the same raw score gives different deviation scores. Larger populations produce more stable values.
How do deviation scores and z-scores relate?
They are the same measurement on different rulers: z has mean 0 and σ 1, the deviation score has mean 50 and σ 10. Convert with deviation score = 50 + 10 × z.
References
- OpenStax, Introductory Statistics 2e, Section 2.7 “Measures of the Spread of the Data” (standard deviation, values a number of standard deviations from the mean): openstax.org
- OpenStax, Introductory Statistics 2e, Section 6.1 “The Standard Normal Distribution” (z-scores and the 68-95-99.7 empirical rule): openstax.org
