Why is it considered desirable to convert raw scores to standard scores ? Below is given the distribution of test scores of 62 children :
Test Score : 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3
f : 1 | 4 | 6 | 10 | 8 | 13 | 18 | 2
Obtain standard scores with mean 50 and standard deviation 10.
10 marks · directive: Obtain
The core of a good answer
Raw scores from different tests have different means and spreads, so they cannot be compared or added; standard scores place them on a common scale. Here N = 62, mean = 355/62 = 5·726 and SD = 1·724, so T = 50 + 10(X − 5·726)/1·724 gives T = 74·8, 69·0, 63·2, 57·4, 51·6, 45·8, 40·0 and 34·2 for raw scores 10, 9, 8, 7, 6, 5, 4 and 3.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
Give reasons for standardisation, compute the mean and standard deviation of the frequency distribution, then the z score and the linear transformation to a scale with mean 50 and SD 10 (T scores) for each raw score.
Answer outline
Introduction
Standard score z = (X − X̄)/σ has mean 0 and SD 1; the transformed score T = 50 + 10z has mean 50 and SD 10 and avoids negatives and decimals.
Step 1: Why standard scores
- Raw scores depend on the difficulty and length of a test; a 7 on one test is not comparable with a 7 on another.
- Standard scores express each score as distance from the mean in SD units, showing relative position in the group.
- Scores on different tests can be compared, averaged or added with equal weight.
- Under a normal distribution, a standard score translates into a percentile rank.
- The 50, 10 scale removes negative values and most decimals, easing reporting.
Step 2: Mean
- ΣfX = 10 + 36 + 48 + 70 + 48 + 65 + 72 + 6 = 355; N = 62.
- X̄ = 355/62 = 5·7258
Step 3: Standard deviation
- ΣfX² = 100 + 324 + 384 + 490 + 288 + 325 + 288 + 18 = 2,217.
- σ² = 2,217/62 − 5·7258² = 35·7581 − 32·7848 = 2·9732; σ = 1·7243 (divisor N; with N − 1, s = 1·7384).
Step 4: Standard scores T = 50 + 10(X − 5·7258)/1·7243
- X = 10: z = 2·479, T = 74·79
- X = 9: z = 1·899, T = 68·99
- X = 8: z = 1·319, T = 63·19
- X = 7: z = 0·739, T = 57·39
- X = 6: z = 0·159, T = 51·59
- X = 5: z = −0·421, T = 45·79
- X = 4: z = −1·001, T = 39·99
- X = 3: z = −1·581, T = 34·19
Conclusion
Each one-point rise in raw score adds 10/1·7243 = 5·80 points on the new scale; the transformed distribution has mean 50 and SD 10 (check: Σf·T/62 = 50). Values computed in Python.
Examples and data to cite
- T scores (mean 50, SD 10) are used in educational assessment.
Facts to remember
- z = (X − X̄)/σ; mean 0, SD 1.
- Linear standard score: T = 50 + 10z.
- Linear transformation does not change the shape of the distribution.
- σ = √(ΣfX²/N − X̄²).