Abstract
In analytical methods, the presence of outliers may mislead the experimenters. The existing Cochran's tests apply to detect outliers when cent percent of observations in the data are determinate. But, when some or all observations are not precise, determined, and certain, the existing Cochran's tests cannot be applied for detecting outliers in the data. In this paper, we introduce Cochran's tests and modification of Cochran's tests based on Levene's test under the neutrosophic statistics. We present the design, implication, and decision criterion of both tests under an indeterminacy situation. The performances of both tests are evaluated using the power curves for various non-normal distributions. The comparison of both Cochran's tests is given and explained with the aid of a real example.