You did WHAT with only ten samples

							You did WHAT with only ten
samples?

   Barry D. Nussbaum




         OEI National Meeting   November 30, 2005   1
Statistical Support Services

 Statistical Support Services (S3) has provided
 statistical analysis at a variety of levels throughout
 EPA
 The inquiries fit in certain categories depending on
 the question and the internal statistical analysis
 capabilities of the inquiring organization
 S3 is developing a framework that efficiently and
 effectively addresses the most common types of
 inquiries


             OEI National Meeting   November 30, 2005     2
Not a Common Inquiry

 MIT Student tries to strike it rich
 Should have contacted OEI first




            Photo courtesy http://philip.greenspun.com


            OEI National Meeting     November 30, 2005   3
Two Most Common Inquiry Types

 1. Sample Size
 2. Questionnaire Review




          OEI National Meeting   November 30, 2005   4
Avoid early worries about sample size

There are bigger concerns
Incrementally, this may not be the biggest
cost




          OEI National Meeting   November 30, 2005   5
Nussbaum’s Bias

 It’s the Question, Stupid !!

 It’s the Stupid Question !!




            OEI National Meeting   November 30, 2005   6
Non-Sampling Errors

 Convenience Sample
 Purposeful Sample
 Decision to revise after viewing early results
   Different sample
   Different sample size
 Rejection of unrepresentative observations
 Substitution of neighboring sampling units
 Response and Non-response errors

            OEI National Meeting   November 30, 2005   7
          Inquiry Examples

Hydrocarbon Refrigerants: S3 helped defend
the EPA decision
Oversight on Contract Payments: Advised
OCFO on how to stratify samples to gain
more information
Results of Region VII Survey: EPA
commuters and public transit



         OEI National Meeting   November 30, 2005   8
A Real Example that Went to Court

 Recalling 208,000 vehicles
   1975 Model year Chryslers with 360 and 400 cu.
   in. displacement
   Nationwide
   Carbon Monoxide
   Convince a lay judge




            OEI National Meeting   November 30, 2005   9
Sampling Vehicles in Use

 What is the population?
 What is the sampling frame?
 How do you know vehicles are properly
 maintained and use?
 What is the incentive to owners?
 How do you get a random sample?
 How do you get a representative sample?


          OEI National Meeting   November 30, 2005   10
And how many do you sample?

 10
 Ten
 Diez
 Zehn
 Dix




        OEI National Meeting   November 30, 2005   11
…And this we gave to the President




         OEI National Meeting   November 30, 2005   12
Collaborative Effort

 S3 methods focus on collaboration
   Common understanding of the objective
   Understanding the problem
   Provide cost effective support
   Focus on informing the real decision
 S3 works with you, not for you!




            OEI National Meeting   November 30, 2005   13