ASSURE: Adopting Statistical Significance for Understanding Research and Engineering

Fausto Galetto, Independent Scholar, Past Lecturer at Politecnico di Torino, Turin, Italy
2021 Journal of Engineering and Applied Sciences Technology  
We start with the question "Is Statistical Significance outdated?" The question is originated by a set of papers suggesting leaving out the use of Statistical Significance; the cause of that idea depends on the fact that many researchers identify the "bad p-values" with the concept of Statistical Significance.We will consider the various concepts involved, we will show the idea of Confidence Interval (with a larger view that in the Statistics and Probability books), we will give examples
more » ... to Control Chart with Non_Normal distributed data [and the wrong T Charts, very much considered in medical settings, using Minitab, SPSS, SAS, ...]; we will suggest to abandon the p-values by showing that they discard the degrees of freedom used to compute them, when one wants to pool the results of various samples.Many Statisticians, Certified Master Black Belts, practitioners, workers, students, all over the world, are learning wrong methods and will take wrong decision.We suggest the form of Confidence Interval to be CI(H0, n, g, 1-CL, Distribution), where H0=Null Hypothesis, n=the physical sample size, g=the number of the random variables that provided the collected data (form which we get the Degrees of Freedom), CL=the Confidence Level (used to compute the CI, 1-CL=α) and Distribution=the type of the distribution of the random variables that provided the collected data (e.g., Normal Exponential, Poisson, Inverse Normal, Weibull, ...)
doi:10.47363/jeast/2021(3)118 fatcat:fkcchyrshbcg3lkg7sb237jpra