Abstract: The uncertainty analysis of the monitoring data in metro construction is significant for the decision-making and management of engineering. Statistical inference theory is employed frequently to analyze uncertainty. Three key issues must be ascertained in order to employ the theory scientifically. (1) What is the principle of monitoring data collection to ensure the statistical scientificity of monitoring data? (2) What are normally the statistical properties of monitoring data with statistical scientificity in practical cases? (3) How to test the statistical scientificity and properties of monitoring data in practical construction? The study on these issues is implemented by employing statistical theory and statistical test methods. Some conclusions achieved are that: (1) All the monitoring points selected to collect data must have similar engineering conditions to ensure that the data collected can comply with the demands of a random sample and statistical scientificity. (2) The monitoring data with statistical scientificity should follow Gaussian distributions because of the central limit theorem, therefore the normality assumption of practical analysis has a rationale. (3) The practical analysis should verify the statistical scientificity by employing run tests, and verify the normality of the monitoring data by probability plots and χ2 tests. These can ensure the scientificity and reliability of statistical inference by combining statistical theory and tests.

Key words: statistical scientificity, normality assumption, run test, probability plot, χ² test