Likert Scale vs. Sama Radial Indicator (SRI) for Measuring Psychometric Variables

Abstract

Psychometrics is a discipline related to: Statistics, Data theory, Econometrics, Biometrics, Measurement theory and Mathematical psychology. Psychometric variables attempt to measure and quantify the mental processes that influence behavior of people. These variables are not directly observed, therefore inferred from other, observable variables referred as hidden variables or Latent Variables (LV). The LV's are measured by polytomus item scales, where an "item" is a question, and a "scale" is the resulting estimate of the LV. The Likert scale is the mostly applied polytomus item scale in measuring psychometric variables. It assigns "codes or ranks" (numerical values) to categories of a qualitative variable. Several items are develop to measure a LV and then the total of the item ranks is taken as the outcome of the random variable. Despite the usefulness of the Likert scales, studies shown several weaknesses and limitations of them. The limited number of responses of the scale induced the error in rating. If the number of responses are increased, then the meanings of the responses become unclear. For example, Likert 7-point scale including: "strongly agree, moderately agree, slightly agree, neutral, slightly disagree, disagree, and strongly disagree", has higher response options compared to a 5-point scale ("strongly agree, agree, neutral, disagree, and strongly disagree"), but hardly the respondent could understand the difference between moderately agree and slightly agree and vice versa. Hence the response would be a guess and the measurement error would be high. Also the variables measured by Likert scales are very unlikely to be normally distributed, hence parametric statistical methods are not applicable for data analysis. However, power of parametric statistical techniques are higher than the non-parametric counterparts, as misuse of Likert scale has become a common practice. Sama Radial Indicator (SRI) is developed to overcome the limitations of the Likert scale as well as other polytomus item scales. The development of SRI is based on the fundamental properties of a circle. The scale has only three anchors, but gives 180 response options, hence the error of the response will be less. SRI converts the psychometric ratings of respondents into a continuous scale, hence the LV's become normally distributed. As such, the decade old problems in measuring psychometric variables is solved.

Keywords: Qualitative variable, Latent variable, Likert scale