Validity and Reliability of Survey Scales

The objective of this paper is to evaluate Likert and non-Likert scales for quantitative survey. The data used in the evaluation of the scale is the scale components. The scales used for the evaluation include the following types: (0,1,2,3), (1,2,3,4,5, (1,2,3,4,5,6,7), and (1,2,3,4,5,6,7,8,9,10). These scales are categorized into two types, namely Likert and non-Likert. The scale (0,1,2,3) is classified as non-Likert; the remaining scales are Likert scales. The efficacy of various scales is evaluated on the basis of fitness. We defined fitness as the ratio between shape and scale of the scaled obtained through the QQ plot linear equation. We found that scale (0,1,2,3) is the most effective scale type for quantitative response choice. The efficacy of the scale was measured by the absolute error of the scale’s fitness CDF. The absolute error of the CDF of the fitness were 0.14, 0.22, 0.25 and 0.26 for the following types: (0,1,2,3), (1,2,3,4,5, (1,2,3,4,5,6,7), and (1,2,3,4,5,6,7,8,9,10), respectively. The results of GOF under the likelihood ratio test, Wald statistic and Langrangian multiplier shows that the non-Likert scale (0,1,2,3) has the best fit in the probability space of the unit circle: 0.71, 0.68, and 0.70, respectively. Response in a form of (0,1,2,3) is the best form of response scale for quantitative survey. This finding is a contribution to the field because the common use of the Likert scale has made findings and conclusion in many cases in social science research lacking validity due to low accuracy.