The proposal of a new index:: Double consistency index (PDC)

It has been empirically shown that the classification consistency indices based on the criterion-referenced context have been influenced by the position of cutoff score on the test distribution. In this paper, a new double consistency index free from the cutoff score has been proposed by combining some traditional methods; also a new formulae has been developed to calculate consistency index of a test. This new index has been based on splitting of test items as odd-even numbers, calculating the total score of each individual in each split, sorting these total scores, and then evaluating agreement between up-low 27% of the group in each half The index was basically underlined on a principle that if an individual is classified in up 27% of the group in the first split, s/he is also classified in up 27% of the group in the second split. The same rule is also valid for the low 27% of the group as an indicator of the consistency of the test. The index value ranges between 0.00 and 1.00. The formulae of the double consistency index is: P-DC = 1-[(f(UO) - f(UE)) + (f(LO) + f(LE))/NU+L] The calculation of the index is easier than the other traditional indices. It can be applied to both dichotomously and polytomously scored rests. It also provides information about ordinal consistency of a test as well as classification of the test. However, further studies should be conducted to test the distribution function of the new index.