Measures split-half reliability by estimating the average correlation that would be obtained by considering every possible way to split the test in half; can be used when there are more than two response alternatives:__________