252chisqx2 10/27/06 Computations for the First Chi-squared example. ————— 10/27/2006 5:10:10 PM ———————————————————— Welcome to Minitab, press F1 for help. MTB > PRINT C11-C14 Data Display Row 1 2 3 O1 10 5 15 O2 15 10 15 O3 15 15 10 O4 10 10 20 This is the original O table. Top line is labels. This is the same data as you were given for the first problem that I did in class. MTB > chisq c11-c14 Chi-Square Test: O1, O2, O3, O4 Expected counts are printed below observed counts Chi-Square contributions are printed below expected counts O1 10 10.00 0.000 O2 15 13.33 0.208 O3 15 13.33 0.208 O4 10 13.33 0.833 Total 50 2 5 8.00 1.125 10 10.67 0.042 15 10.67 1.760 10 10.67 0.042 40 3 15 12.00 0.750 15 16.00 0.063 10 16.00 2.250 20 16.00 1.000 60 Total 30 40 40 40 150 1 These are the column labels. This is the first line of the O This is the first line of the E This is (E – O) squared, divided by E. Chi-Sq = 8.281, DF = 6, P-Value = 0.218 All that you have here is the value of chisquared that we computed in class, the degrees of freedom that we computed in class and a p-value which is too high for us to reject the null hypothesis of independence or homogeneity. MTB MTB MTB MTB MTB MTB MTB I am now setting up the problem as we did it in class. The original data stays in columns 11-13. C14 is the sum of the rows. C15 is the proportion in each row. By multiplying the column totals by the row proportions I got E. > > > > > > > let c15 = c11+c12+c13+c14 let c16 = c15/sum(c15) let c17 = c16*sum(O1) let c18 = c16*sum(O2) let c19 = c16*sum(O3) let c20 = c16*sum(O4) print c11-c20 Data Display Row 1 2 3 O1 10 5 15 O2 15 10 15 O3 15 15 10 O4 10 10 20 sumO 50 40 60 pr 0.333333 0.266667 0.400000 E1 10 8 12 E2 13.3333 10.6667 16.0000 E3 13.3333 10.6667 16.0000 E4 13.3333 10.6667 16.0000 252chisqx2 10/27/06 MTB MTB MTB MTB MTB > > > > > Stack c11-c14 c1. stack c17-c20 c2 let ‘O-E’ = C1 – C2' let chisq1 = 'O-E' * 'O-E' let chisq1 = chisq1/E To do the chi-squared computations I stacked the O columns in C1 and the E columns in C2. O – E was computed. The last column was computed by squaring (O – E) and then dividing by E. MTB > sum chisq1 Sum of chisq1 This is the value of chi-squared we got. Sum of chisq1 = 8.28125 MTB > print c1-c4 This is the table we had in class. Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 MTB MTB MTB MTB O 10 5 15 15 10 15 15 15 10 10 10 20 > > > > E 10.0000 8.0000 12.0000 13.3333 10.6667 16.0000 13.3333 10.6667 16.0000 13.3333 10.6667 16.0000 let let let let O-E 0.00000 -3.00000 3.00000 1.66667 -0.66667 -1.00000 1.66667 4.33333 -6.00000 -3.33333 -0.66667 4.00000 chisq1 0.00000 1.12500 0.75000 0.20833 0.04167 0.06250 0.20833 1.76042 2.25000 0.83333 0.04167 1.00000 c7 = c11/sum(c11) c8 = c12/sum(c12) c9 = c13/sum(c13) c10 = c14/sum(c14) To demonstrate the computations in Chisqnote I made C7 – C11 into proportions of their column sum. These were stacked in C6. This is O in proportions. MTB > stack c7 c8 c9 c10 c6 MTB > print c7 - c10 Data Display Row 1 2 3 C7 0.333333 0.166667 0.500000 C8 0.375 0.250 0.375 C9 0.375 0.375 0.250 C10 0.25 0.25 0.50 MTB > stack c16 c16 c16 c16 c7 MTB > print c6 c7 Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 C6 0.333333 0.166667 0.500000 0.375000 0.250000 0.375000 0.375000 0.375000 0.250000 0.250000 0.250000 0.500000 C7 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 To get E as proportions I stacked the row proportions in C7. 252chisqx2 10/27/06 MTB > let c8 = c6 MTB > let c9 = c7 To protect my work, I copied it into C8 and C9. MTB > print c8 c9 c10 Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 pbars 0.333333 0.166667 0.500000 0.375000 0.250000 0.375000 0.375000 0.375000 0.250000 0.250000 0.250000 0.500000 p0s 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 csums 30 30 30 40 40 40 40 40 40 40 40 40 I am now computing 2 MTB > let c5 = c8 - c9 ij pij pi0 2 pi0 nj . This is the method described in chisquarednote. MTB > print c5 c8 c9 Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 C5 0.000000 -0.100000 0.100000 0.041667 -0.016667 -0.025000 0.041667 0.108333 -0.150000 -0.083333 -0.016667 0.100000 C5 is the difference between expected and observed. pbars 0.333333 0.166667 0.500000 0.375000 0.250000 0.375000 0.375000 0.375000 0.250000 0.250000 0.250000 0.500000 MTB > let c5 = c5*c5 MTB > let c5 = c5*c10 MTB > let c5 = c5/c9 MTB > sum c4 Sum of chisq1 Sum of chisq1 = 8.28125 MTB > sum c5 Sum of chisq2 Sum of chisq2 = 8.53125 p0s 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 To get chi-squared, square C5. Multiply result by co column sums. 252chisqx2 10/27/06 MTB > print c4 c5 c8 c9 c10 Data Display Row 1 2 3 4 5 6 7 8 9 10 11 12 chisq1 0.00000 1.12500 0.75000 0.20833 0.04167 0.06250 0.20833 1.76042 2.25000 0.83333 0.04167 1.00000 chisq2 0.00000 1.12500 1.00000 0.20833 0.04167 0.06250 0.20833 1.76042 2.25000 0.83333 0.04167 1.00000 pbars 0.333333 0.166667 0.500000 0.375000 0.250000 0.375000 0.375000 0.375000 0.250000 0.250000 0.250000 0.500000 p0s 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 0.333333 0.266667 0.400000 csums 30 30 40 40 40 40 40 40 40 40 40 40 Results for: ChiEx1a.MTW MTB > WSave "C:\Documents and Settings\RBOVE\My Documents\Minitab\ChiEx1a.MTW"; SUBC> Replace. Saving file as: 'C:\Documents and Settings\RBOVE\My Documents\Minitab\ChiEx1a.MTW' MTB >