252anovaex2 11/15/06 MINITAB EXAMPLE Two-way ANOVA Explanation: The data set has already been prepared and stored as 2f2.mtb. The first column represents costs. The second column tells to what fuel the cost refers and the third column tells to what insulation the cost refers. For example, one of the values for fuel 2 and insulation 1 is 120. The data set is retrieved and three ‘table’ commands are issued. The first table tells us how many measurements there are in each cell, each row and each column. The second table tells the actual measurements in each row and column. The third table gives cell, row, column and overall means. Finally the ‘twoway’ command is issued and an ANOVA table is produced. There are three null hypotheses: (i) equal row means, (ii) equal column means and (iii) no interaction. Minitab Output: ————— 11/15/2006 4:44:20 PM ———————————————————— Welcome to Minitab, press F1 for help. MTB > WOpen "C:\Documents and Settings\RBOVE\My Documents\Minitab\2F2.MTW". Retrieving worksheet from file: 'C:\Documents and Settings\RBOVE\My Documents\Minitab\2F2.MTW' Worksheet was saved on Wed Nov 15 2006 Results for: 2F2.MTW MTB > print c1-c3 Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 COST 89 101 87 87 120 110 98 104 100 98 86 96 FUEL 1 1 1 1 2 2 2 2 3 3 3 3 INS 1 1 2 2 1 1 2 2 1 1 2 2 MTB > table fuel ins Tabulated statistics: FUEL, INS Rows: FUEL 1 2 3 All Columns: INS 1 2 All 2 2 2 6 2 2 2 6 4 4 4 12 Cell Contents: Count 1 252anovaex2 11/15/06 MTB > table fuel ins; SUBC> data cost. Tabulated statistics: FUEL, INS Rows: FUEL Columns: INS 1 2 1 89 101 87 87 2 120 110 98 104 3 100 98 86 96 Cell Contents: COST : DATA MTB > table fuel ins ; SUBC> mean cost. Tabulated statistics: FUEL, INS Rows: FUEL 1 2 3 All Columns: INS 1 2 All 95 115 99 103 87 101 91 93 91 108 95 98 Cell Contents: COST : Mean MTB > Twoway c1 c2 c3; SUBC> GBoxplot; SUBC> GNormalplot; SUBC> NoDGraphs. Two-way ANOVA: COST versus FUEL, INS Source DF SS MS FUEL 2 632 316 INS 1 300 300 Interaction 2 24 12 Error 6 192 32 Total 11 1148 S = 5.657 R-Sq = 83.28% F 9.88 9.38 0.38 P 0.013 0.022 0.702 R-Sq(adj) = 69.34% 2,6 and F 1,6 that could be used for significance tests. Minitab Comment: You should be able to find the values of F.05 .05 takes a p-value approach, which means that it computes pvalue P F 2,6 9.88 0.13 and pvalue P F 1,6 9.38 0.022 . Since the first two are below our significance level of 5%, we reject the null hypothesis of equal fuel (row) means and the null hypothesis of equal insulation (column) means. However, the p-value for interaction is well above 5%, so we cannot reject the null hypothesis of no interaction. The two diagrams that were printed out follow on the next page. The first diagram shows the range of costs for each combination of fuel and insulation and how the change in insulation affects the costs for each fuel type. The second diagram is a Normal plot of the residuals. To do this the computer calculates an expected value for each number based on the row, column and cell means. The residual is the difference between the expected and observed value. The fact that the points on the plot show no curvature seems to indicate that the underlying data is Normally distributed. 2 252anovaex2 11/15/06 Boxplot of COST by FUEL, INS Normplot of Residuals for COST MTB > 3