Share Share
Text size Increase font size Decrease font size

iconEncyclopedia

The SAGE Encyclopedia of Social Science Research Methods

Michael S. Lewis-Beck & Alan Bryman & Tim Futing Liao

Pub. date: 2004 | DOI: http://dx.doi.org/10.4135/9781412950589

Print ISBN: 9780761923633 | Online ISBN: 9781412950589

Share Share
Text size Increase font size Decrease font size

Encyclopedia Entry

Split-Half Reliability

Stuart J. McKelvie

In classical test score theory, X = T + e (where X = observed score, T = true score, and e = error). Some error is constant, reducing test validity, and some is variable, leading to inconsistency of scores. Reliability is the degree to which the test is free from variable error. Formally, given that σ 2 x = σ 2 1 + σ 2 1 , reliability is the ratio of true score variance to error score variance: Because they are necessarily a sample from the domain being measured, test items constitute one source of variable error. It can be estimated by comparing scores from two equivalent (alternate) forms of the test given at different times, but performance on the second test may be affected by the first one. Split-half reliability requires only one test administration, so participants are unaware of any comparison. It is obtained by splitting the ...

Users without a subscription are not able to see the full content on this page. Please, subscribe or login to access all Methods content.

Click here to see full text

Articles in Google Scholar by