Issue link: https://iconnect007.uberflip.com/i/1530610
68 PCB007 MAGAZINE I DECEMBER 2024 e mean values for opens for the current and new process are 23.8 and 9.1, respectively. e two-tailed p-value is significantly lower than the alpha value of 0.05. With an alpha value of 0.05, there is only a 5% chance of rejecting the null hypothesis when it is, in fact, true (the population means are equal). e process engi- neer concludes that the means are statistically significant; the results in the data are not likely explainable by chance alone. e new process has statistically proven to improve inner layer yields. Next, the process engineer analyzes the vari- ance using the two-sample F-test. e pro- cess engineer loads the data into ToolPak for analysis (Figure 3), and the output is shown in Table 9. A Worked Example A young engineer is undertaking an inner layer improvement project. e goal is to achieve world-class quality, "on target with minimal variation," by reducing opens. e engineer designs an experiment: the current process and a new process (Table 6). Table 6: Inner layer experiment e process engineer then collects data for each process over 12 consecutive days. ey process 100 inner layers through each process and record the total count of opens for that day (Table 7). Note that the same job num- ber is used for each process to minimize noise (uncontrollable factors). Table 7: The count of the inner layer opens e next step is the analysis. e process engineer wants to evaluate both the mean and standard deviation. e first step is the two- sample t-Test. Using the ToolPak, the process engineer loads the data for analysis (Figure 2). e output is shown in Table 8. Figure 2: Two-sample t-Test assuming unequal variances. Table 8. Two-sample t-Test output