split-half correlation

n. The calculation of a product-moment correlation between a set of scores on any two halves of a test so as to estimate the test's internal consistency reliability. This procedure has been supplanted by using Kuder-Richardson formula 20 for true-false or right-wrong test formats and by Cronbach's alpha coefficient for multiple-choice tests, both of which are the average of all possible split-half reliability coefficients corrected for attenuation.