AGENDA
Exam 3 review
REVIEW
Validation
Replication analysis
Experimental Design
Output analysis of two systems
Output analysis of more than two systems
Non-terminating system analysis
Modeling pertaining to the above subjects

VALIDATION
Face
Statistical
Data collection issues
Modeling issues
Process
Check for normality with chisquared test
F-test
Independent-T, equal variance
Smith-Satterwaithe unequal variance
Non-parametric test

REPLICATION ANALYSIS
Relative precision approach
Procedure

EXPERIMENTAL DESIGN
Factors
Levels
2^k designs
TWO SYSTEMS
Very simple simulation studies
Base and one alternative
Two different candidate systems
Welch CI approach...

WELCH CI APPROACH 
Assumes worst case with respect to variance
Uses CI based on adjusting degrees of freedom
If the CI covers 0 then there is no statistically significant difference

MORE THAN TWO SYSTEMS
One Way Analysis of Variance...
Duncan Multiple Ranges Test ...

ANOVA
Calculate
Sums of squares
Mean sums of squares
F statistic
Interpret results 

DUNCANMULTIPLE RANGE TEST
Calculate least significant range values
Determine ranges with no statistical difference
Graphically represent results
Be able to provide recommendations based on availability of economic data

NON-TERMINATING ANALYSIS
One long replication
Remove initial transient
Determine batch size to remove autocorrelation
Rerun the replication
Divide into batches
Use same techniques for
Replication analysis
Comparison

MODELING
Use of the arrivals element for validation
Use of the tallies element for validation
Use of the outputs element for
Replication analysis
Comparison of means

FORMAT
Multiple choice
Analysis