Level: Beginner
【Presenter】
BioStat Institute Co., Ltd. Representative Director Yukio Takahashi
[Presentation summary]
Finding the optimal model when analyzing multifactorial experimental data with the same number of repetitions is difficult because all factors, including interactions, are orthogonal to each other. You can get away with just pooling. However, if there are covariates that must be taken into account, it is not possible to create an analysis of variance table by decomposing the sum of squares. Miyagawa (2008), ``Science of Problem Discovery and Solution - Basics of SQC'' includes an example of analysis of three-factor experimental data including covariates, which mainly consists of a stratified scatter diagram overlaid with a regression line. The results and the results of regression analysis using dummy variables are shown. Miyagawa emphasizes that inference based on estimates of regression parameters is not easy, and that nothing beats careful stratification in a scatter plot. Therefore, after selecting variables including interactions using the ``Stepwise method'' in JMP's ``Model Fit'', we utilized the ``Prediction Profile'' and ``Interaction Profile'' using the ``Standard Least Squares Method'', and then used the ``Graph This paper presents a multifaceted and comprehensive problem-solving method that is connected to a grid-like stratified scatter diagram using "Builder". Furthermore, in addition to JMP plotting, we will also show how to create scatter plots needed for problem solving using Excel.
[Presenter profile]
I am a user since JMP4 and have been using SAS since before JMP. At the company I used to work for, I conducted statistical education using various types of JMP, and also carried out activities to disseminate JMP within the company. We have introduced this result at JMPer's meetings as appropriate. After retiring, I established a self-employed consulting company and continue to work there. Recently, I have felt that the ability to statistically analyze real data in many academic fields in Japan (excluding clinical statistics) is collectively declining. For me, experimental design was essential for learning statistical theory and putting it into practical use. This time's theme is the analysis of multifactorial experimental data that includes covariates, and we will focus on cases where there is an interaction between a covariate and one factor. Traditionally, in the analysis of experimental data that includes covariates, the aim is to statistically remove the influence of covariates and make comparisons between levels of other factors, assuming that there are no interactions. ．． Therefore, the situation is chaotic regarding analysis methods when there are interactions with covariates, and we feel the need to share analysis examples that utilize JMP's excellent analysis results in a variety of graphical displays.
