Analysis of Experimental Design and Data Variability

This document is a technical report that discusses the analysis of experimental design with a focus on understanding data variability. It outlines traditional analysis methods that emphasize hypothesis testing of means, referencing Fisher’s fundamentals of design, which include randomization, replication, and local control of error. The report contrasts these traditional methods with modern analysis techniques that utilize ancillary data to remove variability in the response of interest, emphasizing that remaining variability is deterministic rather than random. The document includes examples of critical temperature analysis in jet turbines and degradation analysis in electric batteries, detailing the methodologies used to strip out variability and predict future degradation. It also discusses model performance measures and the importance of understanding sources of variability in data analysis. The report concludes with comments on the necessity of proper data analysis in scientific and engineering contexts.