Crocker, Douglas C. (1986, ASQC) Ohio State University, Columbus, OH
This session will introduce some general principles involved in regression analysis (RA) and its use in quality control. The material will be presented from the QC practitioner's viewpoint. This will not be just a statistical treatment of the subject which might be found in a textbook, but a comprehensive approach to the subject including the specification of goals and the related steps taken in the RA modelling process. Model assessment, data assessment, use of the regression model for prediction and control, and calculating "tolerance limits" for future observations, all part of this process, will be briefly examined. The related inverse estimation or "calibration" problem will be treated in detail, including methods for controlling precision. This involves examining the relative impact on precision of sample size, in-use replication size, and original data-collection "design". Multiple linear regression (MLR) application to quality control will be introduced with the modelling of attributes and categorical structures. Understanding of significance testing and confidence intervals will be assumed.