Handout for ANOVA

Use Data Example628Anova.SPSS

Test the mean of pension (pension1) classified by education (edu1)

1. Test normality

Kolmogorov – Smirnov (n > 50)

Shapiro – Wilk (n <=50)

The result show that pension is normality (cannot reject null hypothesis).

2. Test Variance

Equal variance  ANOVA

At least one pair unequal  Welch Test or Brown’s Test

Test of Homogeneity of Variances

pension1

Levene Statistic / df1 / df2 / Sig.
4.926 / 8 / 781 / .000

Reject Null hypothesis (equality of variance)  Unequal  Brown or Welch

Robust Tests of Equality of Means

pension1

Statistic(a) / df1 / df2 / Sig.
Welch / 21.182 / 8 / 38.749 / .000
Brown-Forsythe / 20.319 / 8 / 294.010 / .000

a Asymptotically F distributed.

If we assume that the variances are equal (but we should not in this case)  use ANOVA

ANOVA

pension1

Sum of Squares / df / Mean Square / F / Sig.
Between Groups / 9454730421.224 / 8 / 1181841302.653 / 10.703 / .000
Within Groups / 86240428570.918 / 781 / 110423083.958
Total / 95695158992.141 / 789

Null hypothesis is rejected  unequal mean

Test More than one factor (marital status)

Tests of Between-Subjects Effects

Dependent Variable: pension1

Source / Type III Sum of Squares / df / Mean Square / F / Sig.
Corrected Model / 12526232560.329(a) / 31 / 404072018.075 / 3.683 / .000
Intercept / 24653833541.307 / 1 / 24653833541.307 / 224.695 / .000
edu1 / 3298753459.617 / 8 / 412344182.452 / 3.758 / .000
marital1 / 765407863.027 / 3 / 255135954.342 / 2.325 / .074
edu1 * marital1 / 1170006896.356 / 20 / 58500344.818 / .533 / .953
Error / 83168926431.813 / 758 / 109721538.828
Total / 458514256487.802 / 790
Corrected Total / 95695158992.141 / 789

a R Squared = .131 (Adjusted R Squared = .095)