STATCOM at Iowa State

Download Report

Transcript STATCOM at Iowa State

StatCom at Iowa State
Luke Fostvedt
Iowa State University
Department of Statistics
JSM 2009 - Session 404
08-05-09
Outline





StatCom structure at Iowa State
Some Clients and Projects
Post-Election Audits
Benefits
Conclusions
StatCom Hierarchy
Student Involvement

Teams meet as necessary for the projects
◦ Meetings work around academic schedules

Every member of the team is involved
◦ Regardless of experience

Flexible work load
Volunteer Center of Story County




First Ever Client for Iowa State StatCom
About to Begin a new Project with VCSC
Began by developing a satisfaction survey for
organizations affected by VCSC
Next project is to develop a satisfaction survey for
the individual volunteers within each project
Post-Election Audits

Iowan’s for Voter Integrity

Explore Various Methodologies

Provides Exposure to Policy Making
◦ Transparency of the audit process
◦ Policy makers and election officials don’t often agree with
the statisticians (too much red tape)
Iowa Legislation

Proposed legislation similar to Minnesota
◦ Avoids Statistics and Computers at every step
◦ Proposed Procedure Does not Limit Risk

If a large margins of victory
◦ the method audits too many precincts (high cost)

If a small margin of victory
◦ the risk can be large (wrong candidate wins)
Risk Limiting audits

Method Based on Power and Margin of Victory
◦ Typically Power is 99% (pre-specified)
◦ Power must be high enough for voters to be confident
with the outcome

Efficiency
◦ Samples as few precincts as possible when outcome was
correct
◦ 99% is the probability of detecting a miscount if one exists
Simulations


Ran Simulations where the wrong candidate won
Tiered Method (Proposed in Iowa Legislation)
◦ Risk increases as Margin of Victory decreases
 0.5% margin of victory had roughly 80% power

Risk Limiting Method
◦ Simulations always found at least one miscount
◦ Miscounts were simulated both systematically and
randomly
Current and Future Objectives

Draft a White Paper describing our findings and
details a procedure for a post-election audit

Perform a Pilot Audit during the 2010 elections in
Iowa

Continue to consult with the legislators
Benefits

Organizations / Community
◦ Allows them to enjoy the benefits of the scientific process
◦ Decisions made from StatCom analyses directly benefit
the community
◦ The Organizations enjoy working with students
Benefits

Students
◦
◦
◦
◦
◦
Gain valuable consulting experience
Get exposure to statistical literature
Improve communication skills with non-statisticians
Get to work in a collaborative environment
Community Service Hours
Conclusion

StatCom can be an integral part of every Statistics
Program

Your Community Deserves it!