Generation of an interval metric scale to measure attitude

Abstract

This article discusses issues of scales in measuring attitude, demonstrates how a metric scale can be generated based on three main features, and presents results from a repeated measurement survey to verify the generated scale. The design of the generated metric scale is introduced and named Ruler and Option (RO). The population for repeated measurement survey was 1,870 bachelor students from a public university. Two sets of questionnaire (identical items), one with 7-point Likert scale and another with RO scale, were distributed to a sample of 595 bachelor students chosen using stratified random sampling method. Data were analyzed descriptively using SPSS version 20 and structural equation modeling using AMOS version 21. Results showed that data from RO scale performed better than data from 7-point Likert scale in terms of number of items per construct, factor loadings, squared multiple correlations, higher ratio of degrees of freedom to number of parameters, and higher reliability coefficients. In terms of validity coefficients, measurement models from both data sets attained almost the same level of discriminant and construct validity. Further studies are recommended to elicit the strength and weakness of RO scale to identify the situations where it is most suitable.