Statistical Analysis Professor Lynne Stokes Department of Statistical Science Lecture 6QF Multivariate Normal Distribution, Chi-square Distribution of Quadratic Forms, Testing the Significance of Factor Effectrs
Quadratic Forms Distributional properties of q depend on both the properties of the known matrix A and the distribution of the random vector x.
Multivariate Normal Distribution
Properties of the Covariance Matrix Nonsingular Symmetric Positive Definite Positive (Semi-) Definite Matrices Similar Definitions: Negative (Semi-) Definite, Indefinite
Distribution of Quadratic Forms in Normal Random Variables
Trace of a Square Matrix Definition Properties Cyclic Permutations li = eigenvalues of A Symmetric Idempotent Matrix li = 1 or 0
Sample Variance Probability Distribution
Total Sum of Squares Quadratic Form Show Degrees of Freedom n -1 = ar - 1 = rank(AT) = tr(AT)
Main Effect Sum of Squares Quadratic Form Show Degrees of Freedom a -1 = rank(AA) = tr(AA)
Main Effect Sum of Squares Probability Distribution Show
Main Effect Noncentrality Parameter
Error Sum of Squares Quadratic Form Degrees of Freedom Show n - a = rank(AE) = tr(AE)
Error Sum of Squares Probability Distribution Show
Pairwise Independence of Quadratic forms Independence of the Main Effect and Error Sums of Squares
Statistical Tests for (Fixed) Main Effects and Interactions : Balanced Complete Factorials Single-Factor Experiment Response Distribution y ~ N(m1 + XAa , s2I)
Statistical Tests for (Fixed) Main Effects and Interactions : Balanced Complete Factorials Distributional Properties SSA & SSE are statistically independent
Statistical Tests for (Fixed) Main Effects and Interactions : Balanced Complete Factorials iff a1 = ... = aa = 0
Testing Factor Effects Single-Factor Model yij = m + ai + eij i = 1, ..., a; j = 1, ..., r Equivalent Simultaneous Test for Main Effects H0: a1 = a2 = ... = aa vs. Ha: ai aj for some (i,j)
Test Statistic
Assignment Verify the ‘Show’ Results