Psychology: Reliability and Validity

Study Notes

Greater number of items in a test enhances its reliability.
A test can be reliable but may not be valid; however, it cannot be valid unless it is reliable.
Reliability refers to consistent test results from the same test-taker over time.

Test-Retest Reliability: Evaluates stability by testing at two different times.
Parallel Forms Reliability: Compares two different forms of a test measuring the same attributes.
Internal Consistency Reliability: Assesses if items within a test measure the same construct.
- Split-Half Reliability: Divides the test into halves and compares scores.
- Cronbach's Alpha: Measures reliability for tests with multiple-choice answers (polytomous).
Inter-Rater Reliability: Evaluates consistency among different evaluators using Kappa statistics.

Validity determines if a test measures what it intends to measure (e.g., intelligence tests measure intelligence).

Criterion Validity: Assesses performance against a criterion (e.g., predictive validity using GPA to evaluate entrance exams).
Content Validity: Judged by experts on how well the test measures content.
Construct Validity: Measures abstract variables based on conceptual frameworks.
Convergent Validity: Tests relationships with similar constructs (e.g., love).
Divergent Validity: No correlation with unrelated variables.
Face Validity: Assesses the external appearance of the test.

History: External events affecting test-takers before taking the test.
Maturation: Internal changes over time impacting results, particularly in longitudinal studies.
Testing Effects: Impact of prior testing on scores (practice effect).
Instrumentation: Variability in the administration and materials used in tests.
Statistical Regression: Tendency for extreme scores to move towards the mean.
Selection: Lack of random assignment can lead to biased results.
Subject Mortality: Loss of participants from a study can skew results.

Random error is unpredictable, while systematic errors can be addressed.
Classical Test Score Theory addresses measurement errors and aims for standardized testing.

Reliability estimates in clinical settings should be ≥ 0.90, while those in industrial settings should be ≥ 0.70.
Hypotheses:
- Alternative Hypothesis indicates significant differences.
- Null Hypothesis indicates no significant differences.
Item Analysis: Evaluates the effectiveness of test items based on difficulty and discriminability.

Phenomenology: Studies lived experiences with minimal manipulation.
Case Studies: Provide detailed accounts of individual or group experiences.
Field Studies: Observe behaviors in natural settings.
Archival Studies: Analyze existing records for new insights.
Qualitative Research: Focuses on observations and narratives rather than quantitative data.

Single Blind Experiments: Subjects are unaware of group assignments to reduce bias.
Double Blind Experiments: Neither subjects nor experimenters know group assignments to avoid bias.
ANOVA: Used to determine significant differences among three or more groups.
Norm-Referenced Testing: Compares individual scores to a broader population.
Classified Tests:
- Class A: General quizzes.
- Class B: Personality and intelligence assessments.
- Class C: Individualized intelligence and projective tests.