Reliability and Validity in Assessment — Previous Year Questions
📊 Topic Overview
| Metric | Value |
|---|---|
| Total Questions | 7 (3.1%) |
| Priority | 🟡 MEDIUM — Section B and C |
🎯 Must-Know Concepts
- Definition of reliability
- Types: test-retest, split-half, inter-rater, parallel forms
- Definition of validity
- Types: content, criterion (predictive/concurrent), construct, face
- Relationship between reliability and validity
- Standard error of measurement
📝 Master Answer — Section B (250 words / 6 marks)
Q. Elucidate the concepts of reliability and validity. (6 marks)
Also: "Define Reliability and Validity as measurement concepts." (2021-Jun B9); "Discuss the concept of validity and its relevance in personality testing." (2014-Jun)
Reliability
Reliability refers to the consistency of a measurement instrument — the extent to which it produces the same results under the same conditions. A reliable test is free from random measurement error.
Types of Reliability:
- Test-retest reliability: Administering the same test to the same group on two occasions and correlating the scores. Indicates temporal stability.
- Split-half reliability: Splitting test items into two halves and correlating the scores. Measures internal consistency.
- Inter-rater reliability: Degree of agreement between two or more scorers. Critical for projective techniques and behavioural observation.
- Parallel forms reliability: Correlation between two equivalent versions of the test. Eliminates practice effects.
Reliability is expressed as a correlation coefficient (0–1). A coefficient above 0.80 is generally acceptable for personality measures.
Validity
Validity refers to the accuracy of a measurement instrument — whether it measures what it is intended to measure. Validity is always specific to a purpose.
Types of Validity:
- Content validity: Items represent the full domain of the construct (judged by expert review)
- Criterion validity:
- Predictive: Scores predict a future criterion (e.g., EPQ predicts antisocial behaviour)
- Concurrent: Scores correlate with a current criterion measure
- Construct validity: The test measures the theoretical construct — demonstrated through convergent validity (correlates with related measures) and discriminant validity (does not correlate with unrelated measures)
- Face validity: The test appears to measure what it should — important for respondent acceptance
Reliability vs Validity
A test can be reliable without being valid (consistent but measuring the wrong thing), but a valid test must be reliable. Reliability is a necessary but not sufficient condition for validity.
💡 Exam Tips
- The reliability vs validity distinction is the most important conceptual point.
- For Section C short notes on "Validity": define + list 3 types with one-line explanations.
- Always state: "Reliability is necessary but not sufficient for validity."