Cronbach's Alpha
A statistical measure of internal consistency reliability. Values range 0-1; α=0.70-0.90 is ideal for personality tests. Indicates whether scale items correlate with each other.
Cronbach's alpha (Cronbach, 1951) is the most common reliability statistic in psychology. It measures: "To what extent do items measuring the same construct correlate with each other?"
Formula intuition: If you randomly split a test's items in half, do the two halves correlate? If yes, the test is internally consistent (high α). If the halves don't correlate, the test is measuring multiple things (low α).
Interpretation: α=0.70+ is considered acceptable; α=0.80+ is good; α=0.90+ is excellent (but may indicate redundancy). α<0.60 suggests items aren't measuring the same construct, the scale needs revision.
Limitations: Cronbach's alpha assumes items are unidimensional (measuring one thing). If a construct is genuinely multifaceted (Big Five Conscientiousness with multiple facets), a single alpha can underestimate consistency. Modern alternatives: McDonald's omega, factor-based indices.
Big Five scales typically report α=0.75-0.85 per dimension, indicating good internal consistency and justifying interpretation of the scores.
Source: Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297-334.