Full Text

Change of PD DC C in Various n, k and Percent Manipulations and Its Comparison with ρ, Split-Half Coefficient and Various Indices

Adnan Erkuş
2005; 20(55):115-126

        [Turkish Summary]        [Turkish PDF]    [English PDF]    [Mail to Author]
Recently, a new validity classification has been proposed by Murphy and Davidshofer (2001) and Erkuş (2002 and 2003): Validity of measurement and validity for desicions. In this study, it has been examined how PDC which is proposed as the index of classification and ordinal validity changes depending on various n, k, and % manipulations and its relation with ρ, split-half coefficient and the classification consistency indices (κ and Po). Analyses on obtained data collected from 161 individuals by using Hopelessness Scale and Submissive Acts Scale have shown that 27%-cutoff area is the best percent; the value of index decreases as n decreases (when k is constant) for dichotomous data, unlike for polytomous data; when n is constant k decreases, PDC value is opposite to n manipulation. When PDC is corrected according to Spearman-Brown formulea (for 2*%46-cutoff area) and is compared with ρ and split-half coefficient, it has been observed that PDC has values between two conventional coefficients. Morever, when PDC is compared with κ and Po which are the classification consistency indices, the results have pointed out that it has values between κ and Po. As a result, it is suggested that PDC can be used as both classification and ordinal validity index; 46%-Spearman-Brown correction for PDC is appropriate; at the same time, PDC includes information in relation to reliability of the scales. But, it needs further studies especially on its distribution function.

Keywords: PDC index, classification and ordinal validity, validity of measurement, validity for decisions