【正文】 The National Graduate School of Quality Management ? 132 Bowling Example (continued) ?The more the lines diverge from being parallel, the more the interaction. ?We see that the strongest interaction (still not significant) is between the lane and the ball. ?We know from our earlier analysis that none of。The National Graduate School of Quality Management ? 130 Bowling Example (continued) ?Select InteractionPlot and then Setup….. 169。The National Graduate School of Quality Management ? 128 Bowling Example (continued) ?The magnitude of the vertical displacement indicates the strength of the main effect for that factor. Here we see that the wristband has dramatically more effect than any other factor. We know from our earlier plots that the wristband is the only statistically significant effect 95% confidence. ?This plot also shows you the direction of the main effects. We clearly see that the “with” condition is related to the higher level of performance. 169。The National Graduate School of Quality Management ? 126 Bowling Example (continued) ?Select Main Effects Plot and then Setup… 169。The National Graduate School of Quality Management ? 124 Bowling Example (continued) Note that only one effect has a significance greater than 95%. All the remaining factors and interactions are not statistically significant. 169。The National Graduate School of Quality Management ? 122 Bowling Example (continued) Note that Selected Terms has all of the available choices already selected. We need do nothing further. Select OK. Then, at the main screen select Graphs 169。The National Graduate School of Quality Management ? 120 Bowling Example (continued) Go to ?Stat…. ?DOE… ?Factorial... ?Analyze Factorial Design... 169。The National Graduate School of Quality Management ? 118 Bowling Example (continued) Remove the option to Randomize Runs…. Now, select OK and go back to the main screen. Once at the main screen select OK... 169。The National Graduate School of Quality Management ? 116 Bowling Example (continued) Select your design…. We will be using the Full (Factorial) and again we can see that it will require 8 runs… Now, select OK and go back to the main screen. Once at the main screen select Factors... 169。The National Graduate School of Quality Management ? 114 First Create an Experimental Design... Select 2 Level Factorial design with 3 factors Then go to Display Available Designs…. 169。The National Graduate School of Quality Management ? 112 DESIGN OF EXPERIMENTS (DOE) FUNDAMENTALS 169。The National Graduate School of Quality Management ? 110 Main Effects Select ?C4 Response ?C5 Factors ?OK 169。The National Graduate School of Quality Management ? 108 Note that the P value is less than .05 that means that we reject the null hypothesis Analyzing the results…. 169。The National Graduate School of Quality Management ? 106 ?Select ? C4 Responses ? C5 Factors ?Then select Graphs…. Set up the analysis…. 169。The National Graduate School of Quality Management ? 104 Let’s set up the analysis ?Load the file Anova … ?Stack the data in C4 and place the subscripts in C5 169。The National Graduate School of Quality Management ? 102 Note that you will have the critical population and test statistics displayed in the session window. ?Minitab builds the table for you. Note that our original data is presented and directly below, Minitab calculates the expected values. ?Here, Minitab calculates the Chi Square statistic for each data point and totals the result. The calculated Chi Square statistic for this problem is . Performing the Analysis…. 169。The National Graduate School of Quality Management ? 100 Let’s set up a contingency table…. ?Contingency tables are found under Stat…. Tables… Chi Square Test…. 169。The National Graduate School of Quality Management ? 98 CONTINGENCY TABLES (CHI SQUARE) 169。The National Graduate School of Quality Management ? 96 Paired Comparison ?Select the Boxplot for our graphical output.. ?Then select OK.. 169。The National Graduate School of Quality Management ? 94 Paired Comparison ?Go to Stat…. ?Basic Statistics… ? Paired t….. 169。The National Graduate School of Quality Management ? 92 Paired Comparisons ? In paired parisons we are trying to “pair” observations or treatments. An example would be to test automatic blood pressure cuffs and a nurse measuring the blood pressure on the same patient using a manual instrument. ? It can also be used in measurement system studies to determine if operators are getting the same mean value across the same set of samples. ? Let’s look at an example: 169。The National Graduate School of Quality Management ? 90 ?In the session window we have each population’s statistics calculated for us.. ?Note that here we have a P value of .922. We therefore find that the data does not support the conclusion that there is a significant difference between the means of the two populations... Interpreting the results…. 169。The National Graduate School of Quality Management ? 88 ?Since we already have our data stacked, we will load C4 for our samples and C3 for our subscripts. Setting up the test…. ?Since we have already tested for equal variances, we can check off this box… ?Now select Graphs…. 169。The National Graduate School of Quality Management ? 86 ?Here, we see the 95% confidence intervals for the two populations. Since they overlap, we know that we will fail to reject the null hypothesis. ?The F test results are shown here. We can see from the PValue of .263 that again we would fail to reject the null hypothesi