A novel method to characterize immunity pass/no pass test results

In most cases the results of susceptibility tests based upon pass/fail criteria are purely qualitative; because of this reason it is impossible to define the susceptibility safety margin defined as the difference between the 50% probability of malfunction and the probability of the most critical test conditions. With the help of Logistic Regression one can give a quantitative interpretation also to qualitative test results and at the same time to provide the safety margin. Logistic Regression, which is a statistical analysis tool used for epidemiological studies, can be successfully exploited to interpret susceptibility test results with the threefold aim of understanding which are the most significant causes affecting the Equipment under Test (EUT), of providing a technique to define a global figure of safety margin when many test parameters are involved and of predicting the EUT immunity performance in specific electromagnetic environmental conditions different from the ones adopted in the EMC test lab. The background of Logistic Regression is briefly summarized and then a graphical interpretation of the results is illustrated. Ordered Logistic Regression where the output is an ordinal (ordered category) variable is also briefly described; it can be used in those cases where the output may have several possible outcomes.