This preview shows pages 1–15. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: McGill University Advanced Business Statistics MGSC272 Fall10 Analysis of Variance Read: Business Statistics (A Second Course) , 2 nd Custom Edition for McGill University Chapter 11 ANOVA ANOVA is a statistical test of significance for the equality of several (2 or more) population (treatment) means. Assumptions: 1. All populations are normally distributed. 2. All populations have the same variance: 1 2 = 2 2 = 3 2 = ........ = k 2 = 2 . 3. Independent random samples. Notation Y ij = the i th observation from the j th treatment. = the mean of the j th sample. T j = the total of the j th sample. n j = the size of the j th sample. n T = n j = the total number of observations. p = the number of treatments. = the overall (grand) mean. j = the mean of the j th treatment. = the common standard deviation for all treatments. j Y Y ANOVA Table Source of Sum of Variation Squares df Mean Square F* Treatment SSTR p  1 MSTR = SSTR/(p  1) MSTR/MSE Error SSE n T p MSE = SSE/(n T p) Total SSTO n T 1 ANOVA Test of Hypothesis H : 1 = 2 = 3 = ........ = p ( All the population means are equal). H 1 : Not all j are equal (At least one of the means is different). TS: F* = MSTR/MSE (or MSB/MSW). CV: F( ; p  1; n T p ). DR: Conclude H if F * CV and Conclude H 1 if F * > CV. The Formulas 2 SSTO = ( ) ij Y Y ( 29 2 SSTR j j n Y Y = 2 SSE = ( ) = SSTO  SSTR ij j Y Y Partition Theorem SSTO = SSTR + SSE df TO = df TR + df E Oneway ANOVA A Completely Randomized Design A completely randomized design to compare p treatment means is one in which the treatments are randomly assigned to the experimental units, or in which independent random samples are drawn from each of the p populations. EXAMPLE A student society has undertaken a survey of the cost of a night out in three different cities: Montreal, Toronto and Ottawa. The results of the survey are as follows (in dollars): Montreal Toronto Ottawa 16 19 23 17 26 19 22 24 24 14 24 19 27 20 Summary Statistics Sample sizes : n 1 = 6 n 2 = 5 n 3 = 3 Means: 1 18 Y = 2 24 Y = 3 22 Y = n T = 14; Y = 294 / 14 = 21 Calculating Sums of Squares SSTR = = 102. SSE = = (16  18) 2 + (17  18) 2 + (22  18) 2 + (14  18) 2 + (19  18) 2 + (20  18) 2 + (19  24) 2 + (26  24) 2 + (24  24) 2 + (24  24) 2 + (27  24) 2 + (23  22) 2 + (19  22) 2 + (24  22) 2 = 94. SSTO = SSTR + SSE = 102 + 94 = 196. 2 2 2 2 ( ) 6 (18 21) 5 (24 21) 3 (22 21) j j n Y Y = + + 2 ( ) ij j Y Y ANOVA Table Source of Variation Sum of Squares df Mean Square F* Treatment 102 2 51.00 5.97 Error 94 11 8.56 Total 196 13 Note: E(MSE) = 2 MSE s 2 . The test of hypothesis H : 1 = 2 = 3 (All three population means are equal)....
View
Full
Document
This note was uploaded on 02/28/2012 for the course MANAGEMENT MGSC 272 taught by Professor Smith during the Spring '12 term at McGill.
 Spring '12
 smith

Click to edit the document details