Statistics Glossary → Split-half reliability and Guttman's λ
Split-half reliability and Guttman's λconcept
A method that estimates the reliability of a whole scale from the relationship between the total scores of its two halves; Guttman's λ coefficients are generalized lower bounds built on the same idea.
The items are split into two groups (e.g. first and second half, or odd and even items) and the correlation r between the two half scores is computed. This correlation is the reliability of a half-length test; for the full test it is stepped up with the Spearman-Brown formula: ρ = 2r / (1 + r). Guttman's split-half coefficient uses the variances of the halves instead and is less optimistic than Spearman-Brown when the halves have unequal variances.
The result depends on how the items are split; different splits give different values. Guttman's λ1-λ6 are lower bounds to the true reliability: λ3 equals Cronbach's alpha, λ2 is always greater than or equal to it, λ4 is the value for a particular split into halves, and λ6 is based on how well each item is explained by the others. None of them shows that the scale is unidimensional.
An everyday example
A 20-item maths test is given to 100 students. The scores on the first 10 items and the last 10 items correlate at r = 0.70. With the Spearman-Brown formula, the reliability of the full test is estimated as 2 × 0.70 / 1.70 ≈ 0.82.
In MerQur
In Cronbach's Alpha and Reliability Analysis, 'Additional output: Other reliability models: split-half + Guttman λ1-λ6' and 'Split-half: items in part 1' (0 = automatic, the first ⌈k/2⌉ items).
Analyses where this concept appears
Cronbach's Alpha and Reliability Analysis
Related concepts
Spearman-Brown formula (single and average measures)McDonald's ω (omega) reliability coefficient