Toward Citizen Understanding and Use of Statistical Data

1
Toward Citizen Understanding and Use of
Statistical Data

• Gary Marchionini
• University of North Carolina at Chapel Hill
• march_at_ils.unc.edu www.ils.unc.edu/march
• Digital Government Workshop
• May 17, 2000 Los Angeles

2
The General Problem Citizen Access

• Statistics are used for decision making in an
  information society
• Statistical literacy of population is poor
• Federal agency mission to leverage technology to
  improve services
• Statistical tables are a neglected medium in
  Internet environment

3
Challenges

• Human
• Diverse user population
• Internet usage/expectations change
• Tasks are emergent
• Government
• Cooperation and interoperation
• Systems
• Information structure and flow
• Technical developments
• A bad interface is worse than no access

4
Project Research Goals

• Investigate why and how people seek and use
  statistical tables
• Develop and test prototype interfaces that aid
  finding, understanding, and using tables
• Develop strategies of transferring research
  results to government agency production systems

5
The Team

• NSF Digital Government Initiative Funding for 2
  year project
• Partners include
• PIs Syracuse U (Carol Hert, Liz Liddy), UMD (Ben
  Shneiderman), UNC (Gary Marchionini)
• Statistical Agencies including BLS, Census, EIA,
  NCHS, EPA, BJS

6
Empirically Defined Table Interface Grammar
7
People and Statistics

• Past work
• Task and user type taxonomy
• Current work
• Focus groups interviews
• User-system interactions
• Planned work
• Workshops with various citizen groups (such as
  seniors)
• Eye tracking study

8
Results to Date

• People want a known value/set or to explore (and
  this may cross tables)
• People attempt a variety of comparisons
• Expert users want data raw or interactive
  tables
• The provision of context for a table is critical
• definitions of terms
• measures of variance, etc. for each cell
• other metadata
• Substantial home use (gt 1/2 work, 1/3 home)
• Novices do not have agency views of data

9
Results (cont)

• Geography an important entry point
• Citizen barriers, level of
• knowledge of survey methodology
• knowledge of the domain
• IT literacy/experience
• Data barriers
• granularity (month/year, state/substate, etc.)
• people want to print tables

10
Find a Table

• Search grammar for retrieval
• Understand what tables are available through
  words
• Dynamic query overviews of data
• Understand what tables are available through
  context

11
Search Grammar

• Sublanguage Analysis
• Develop based on sample of users natural
  language queries (EAG)
• Basic elements
• Location (where?)
• Condition (what?)
• Population of Interest (whom?)
• Time (when?)
• Percentage/Number/Rate (how much?)
• Others

12
Sample Query and Tags

• What percentage of people living in Alabama did
  not have jobs in the summer of 1999?
• WhatWP percentageNN ofIN peopleNN livingVBG
  inIN ltLOCgt AlabamaNP lt/LOCgt didVBP notRB
  haveVB jobsNNS inIN theDT summerNN ofIN
  1999CD ? 
• WhatWP ltHOW MUCHgt percentageNN lt/HOW MUCHgt
  ofIN ltWHOgt peopleNN lt/WHOgt living VBG inIN
  ltSTATEgtAlabamaNP lt/STATEgt didVBD ltCONgt notRB
  haveVB jobsNNS lt/CONgt inIN theDT ltPTIMgt
  summerNN ofIN 1999CD lt/PTIMgt ??

13
Table Elements

• Title Geographic Location (e.g. state)
• Row Condition and Population (e.g.
  Unemployment Non-farm)
• Column Time (e.g. month)
• Cell Quantity (e.g. rate, number)
• WHERE Alabama
• WHAT not have jobs unemployment
• WHOM people Labor Force
• WHEN summer of 1999 June 1999 July 1999
  August 1999
• HOW MUCH what percentage _________

14
Overviews and Previews

• Explore what data is available with direct
  manipulation
• Overview data collections
• Preview partitioning results
• Highly dependent on metadata as partitioning
  attributes

15
Relation Browser
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
Read and Use a Table

• Understand through metadata and annotations
• On-demand, cascading elements
• Manipulate data in basic ways
• Alternative views
• Subtable definition and combination
• Save/print

22
Interface Grammar

• Objects
• cell (the atomic unit, may have multiple layers)
• row (has a header, may have multiple layers)
• column (has a header, may have multiple layers)
• header (modifies a cell, row)
• subtable
• table
• survey
• agency

23
Interface Grammar

• Actions
• View
• scroll/pan, zoom, mouseover, jump
• Change View
• resize, reorder, graph/visualize
• Define Table
• selection from menu, drag and drop, partition,
  find
• Calculate
• Help
• tell story, tutorial, demo, reference
• Print, Save

24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
(No Transcript)
29
Manipulation Challenges

• While necessary manipulation functionality (from
  the users perspective) is reasonably well known,
  expectations will evolve as people use the data.
• Knowing which information to provide at which
  points via which mechanisms (e.g., mouseovers,
  clicks) requires additional investigation.
• Customizable, alternative interfaces without
  overloading user.

30
Architecture
31
Next Steps

• Continue user studies (extend to testing
  interface features)
• Define metadata layers and build into prototypes
  for on-demand elaboration
• Explore implementation paths
• Link to ontology work
• Link to graphic work
• Tiny browser

32

• http//istweb.syr.edu/tables/
• http//ils.unc.edu/march/CUU/tables.pdf