MFIN 604 (Statistics) Workshop on ANOVA and Nonparametric Statistics: May 9, 2014

Mahmut Parlar

 

Software : You will need to download the free Excel add-in MegaStat 10.2 to solve the numerical examples. Here is the link to download the add-in.

 

Notes : Workshop notes (original); mostly handwritten (but with some typos). The actual handwritten notes from the workshop are here.

 

Normality of population data

 

Many data sets of populations approximate normal distributions. For example, the GMAT scores of 2003 admissions are approximately normal. This is also true, in general, for exam scores, IQ scores, etc.

 

However, in many cases population data do not exhibit normality. Here is an example: 2013 McMaster salaries. When this is the case, we need to use nonparametric statistics.

 

Analysis of Variance (ANOVA)

 

Hypothesis testing for a single population (review) : GMAT-1.xlsx 

 

Hypothesis testing for two populations (review) : GMAT-2.xlsx

 

ANOVA for multiple populations : GMAT-3.xlsx

Nonparametric Statistics

Sign test (single population) : GMAT-1-Sign.xlsx

Mann-Whitney test (two populations) :  GMAT-2-MannWhitney.xlsx

 

Kruskal-Wallis test (multiple populations) : GMAT-3-KruskallWallis.xlsx