Construction risks such as cost and schedule overruns, technology failure, force majeure etc. are common in most construction projects. However, risks in secondary nature should not be ignored. For quality assurance (QA) of construction materials most state highway agencies and contractors use statistical methods which assume normal distribution of data. However, data analysis from several states identified highly skewed and kurtosis induced i.e. nonnormal data for asphalt content, material density and concrete compressive strength. High nonnormality can result falsely penalizing acceptable products, and rewarding bad products which can easily upset the relative profit margins of the contractor. A robust method named Box-Cox transformation with golden section search algorithm is developed that can correct nonnormality in such datasets.