Journal article Open Access
The objective of this paper is to introduce a new sample size calculation method based on the type of response scale used surveys. The current literature on sample size calculation focuses data attributes and distribution. There is no prior research using response scale as the basis for minimum sample size calculation. This paper fills that gap in the literature. We introduced a new minimum sample size calculation method called n* (n-Star) by using the Monte Carlo iteration as the basis to find asymptotic normality in the survey response scale. This new method allows us to achieve up to 95% accuracy in the sample-population inference. The data used in this study came from the numerical elements of the survey scales. Three Likert and one non-Likert scales were used to determine minimum sample size. Through Monte Carlo simulation and NK landscape optimization, we found that minimum sample size according to survey scales in all cases is n* = 31.61±2.33 (p < 0.05). We combined four scales to test for validity and reliable of the new sample size. Validity was tested by NK landscape optimization method resulted in error of F(z*) = 0.001 compared to the theoretical value for the center of the distribution curve at F(z) = 0.00. Reliability was tested by using Weibull system analysis method. It was found that the system drift tendency is L = 0.00 and system reliability R = 1.00.
ARTICLE 5, Vol 3, No 3, Minimum Sample Size Method Based on Survey Scales.pdf