Maryland 2002 Election Usability

Benjamin B. BedersonComputer Science Department

Human-Computer Interaction LabUniversity of Maryland, College Park

Reporting results from work withPaul Herrnson

Owen Abbe Dept. of Government and Politics

www.cs.umd.edu/~bederson/voting

University of Maryland, Human-Computer Interaction Laboratory

2002 Election Exit PollMontgomery & Prince George’s

Counties

Administered broad exit poll questionnaire 1,276 respondents 22 precincts in two counties response rate was 74.6%

Summary: Majority like new system But significant minority have concerns

University of Maryland, Human-Computer Interaction Laboratory

Diebold AccuVote-TSDeployed at 2002 General Election

University of Maryland, Human-Computer Interaction Laboratory

University of Maryland, Human-Computer Interaction Laboratory

University of Maryland, Human-Computer Interaction Laboratory

University of Maryland, Human-Computer Interaction Laboratory

University of Maryland, Human-Computer Interaction Laboratory

University of Maryland, Human-Computer Interaction Laboratory

Voter Acceptance

The voting system was easy 94% I was comfortable using the system 93% Characters on the screen were easy

to read 94% Terminology on screen was precise 93% Correcting my mistakes was easy91% I am confident that my vote was

accurately recorded 90%

University of Maryland, Human-Computer Interaction Laboratory

Voter Trust

Previously, voters had used punch cards or mechanical lever systems.

I trust the previous voting machine71%

I trust the touch screen voting machine 91%

University of Maryland, Human-Computer Interaction Laboratory

Problems Using the System

Asked for help using the machine 9% Received help using the machine 17% Experienced technical problems 3%

Election officials are pro-active Most technical problems are with cards Navigation was troublesome – sometimes

jumping multiple screens Couldn’t change language after selection Ballot review with scrollbar was difficult

University of Maryland, Human-Computer Interaction Laboratory

Usability and Assistanceby Computer Use

Usability and Assistance

0%

20%

40%

60%

80%

100%

120%V

otin

gsy

stem

easy

to u

se

Term

inol

ogy

was

pre

cise

Ask

ed fo

rhe

lp u

sing

mac

hine

Got

hel

pus

ing

mac

hine

Frequency of Computer Use

Perc

enta

ge R

epor

ting

Frequence of ComputerUse <= 1/month

Frequence of ComputerUse 2/month - 2/week

Frequence of ComputerUse >= 3/week

University of Maryland, Human-Computer Interaction Laboratory

Assistance by Education

Assistance by Education

0%

5%

10%

15%

20%

25%

30%

35%

Asked for help Got help

Perc

ent r

ecei

ved

No college

Some college to BS

Graduate school

University of Maryland, Human-Computer Interaction Laboratory

Assistance and Trustby Race

Assistance and Trust by Race

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Asked for help Got help Trust machine

Perc

ent r

ecei

ved

Black

White

Other

University of Maryland, Human-Computer Interaction Laboratory

Assistance and Trust by Sex

Assistance and Trust by Sex

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Asked for help Received help Trust machine

Percent received

Female

Male

University of Maryland, Human-Computer Interaction Laboratory

Assistance by Age

Assistance by Age

0%

5%

10%

15%

20%

25%

30%

Asked for help Received help

Percent received

18 to 24

23 to 34

35 to 49

50 to 64

65 or older

University of Maryland, Human-Computer Interaction Laboratory

Information Visualization

Visualization helps users see patterns and detect outliers in large data sets

A ballot is a large dataset Most DREs show less than 4 races per screen How do voters understand how they voted?

Show more than fits on the screen by: Good, dense information design Overview+detail Abstracted representations Simple navigation mechanisms

University of Maryland, Human-Computer Interaction Laboratory

Navigating Large Spaces

Imagine driving from NY to CA with only street maps.

You need abstracted overview maps – that show states and highways.

We have the same problem with voting systems: How do you get an overview of the state

of your ballot?

University of Maryland, Human-Computer Interaction Laboratory

A Motivating Example

Zoomable User Interface (ZUI) Single screen interface Overview + Detail Natural navigation and

progress indication

University of Maryland, Human-Computer Interaction Laboratory

Conclusion

Studies leave us optimistic, but concerned With elections called by 1%, leaving 10%

unconfident voters is a problem The requirements of DREs are unique, but

the design issues aren’t Typical of public access information systems Need closer work with HCI professionals Need qualitative and quantitative user studies Need further field studies

www.cs.umd.edu/~bederson/voting