Comparative Extension Projects: Pennsylvania

1
Comparative Extension Projects Pennsylvania
Parallels to US Project

• Project Funding
• The General Assembly funded the project based on
  a proposal by Temple University to build a
  database similar to the national Policy Agendas
  model for the years 1979-2006.
• Leaders of all four caucuses have supported the
  project and appoint representatives to a steering
  committee. The three year budget is 630,000
  virtually all of which pays for student
  researchers.

• PA (1979-2006)
• Legislative Hearings
• Acts, Bills, Resolutions
• Executive Orders
• Governors Budget Addresses
• PA Supreme Court Decisions
• State Finances (Census)
• Governors News Digests
• Governing Magazine
• State Polls

• US (1947-2004)
• Congressional Hearings
• Public Laws (and Bills)
• Executive Orders
• State of the Union Addresses
• US Supreme Court Decisions
• Federal Budgets
• New York Times
• Congressional Quarterly
• Gallup Polls

Coding by Policy Impact The US and PA projects
code data by policy impact, not governmental or
legal structure. A bill giving veterans a tax
break is coded 1609, veterans issues, not 107,
state taxation, even if it amends the state tax
code, is considered by the legislative tax
committees, and will be administered by the
Revenue Department. In determining a code,
students are trained to ask, Who is this bill
trying to help, not what government agency or
legislative committee will oversee this policy.
But the PA database has filters to further
identify important cross-cutting structural and
constituency dimensions.

• Computer-Assisted Coding Plans
• The project is preparing a stratified sample of
  bills to train the computer to code remaining
  legislation.
• Our working assumption is that computer coding
  will reduce researcher hours by 50.
• Researchers will reexamine records whose coding
  accuracy is rated low-probability by the computer
  and revise codes if necessary.
• We will then submit news stories and other
  records for computerized coding.

• Differences in the Datasets
• The text and history of US statutes is not
  directly accessible through the national database
  website.
• The text and history of all acts, bills,
  resolutions, and legislatively authorized studies
  will be accessible online through the PA
  database quantitative analysis of bill histories
  will be available (bills passed that passed one
  house, died in conference, etc.)
• PA records are not as complete as US records.
  Many committee hearings probably have been lost.
• The national project abstracts a sample of New
  York Times stories, many of which are not about
  public policy. The database can answer the
  question, How much attention was the news media
  (New York Times) paying to various issues from
  one year to another?
• The PA database abstracts a sample of stories
  from newspapers, wire services, and television
  and radio states across the state that were
  included in the daily news digests published by
  the governors (or House) press offices. The
  database can answer the question, How much
  attention were policymakers paying to media
  accounts of various issues.

• Issues in Computer-Assisted Coding
• Although the PA Constitution requires the short
  title of all bills to reflect their content, PA
  titles appear to be less specific than titles of
  congressional legislation. We may ask the
  computer to test bill titles against bill
  content.
• Our news abstracts have been created by students
  working on separate campuses and are probably
  less consistent in use of language than the New
  York Times abstracts in the national database.
• Because funded by the General Assembly, our
  project has less tolerance for random coding
  error the computer will not have the last word.

Project DirectionEach university team is led by
a faculty member who provides oversight. Each
team includes a graduate research manager (GRM)
who provides day-to-day direction to student
researchers, is responsible for quality control,
and maintains communication with the Temple
staff, and 4 to 6 researchers who abstract and
code records. In addition to researchers,
Temples team includes computer science faculty
and students who are building the website as a
class project.

• Processing Records
• Student researchers abstract bills, resolutions,
  hearings, and news articles. Students do not
  apply policy codes to records for which they
  created the abstract.
• Using the policy codebook but working separately,
  two different students then code each record.
  The goal is to achieve 90 inter-coder
  reliability for major topics and 75 for minor
  topics.
• The graduate research manager checks student work
  and resolves differences in coding. The GRM
  meets with researchers each week to go over their
  work and discuss hard cases.

• University AssignmentsAlthough the decentralized
  organization of the project presents challenges,
  it also has advantages. Governors news digests
  are not centrally archived
• Governor Dick Thornburghs news digests (1979-86)
  are at the University of Pittsburgh.
• Governor Robert P. Caseys news digests (1987-94)
  are at Penn States University Park Campus.
• News digests for the period 1995-2006 are
  archived in the Capitol complex in Harrisburg.

• Changes to the National Codebook
• 13 Substantive Changes (2 Major Topic and 11
  Subtopic Codes)
• 20 Subtopic Additions
• 1 Major Topic Code Eliminated (Culture and
  Entertainment)
• 44 Unused Federal Sub-codes

• Benefits to State Government
• Policy-makers and aides can more efficiently
  research recurring issues and previously-tried
  solutions, avoiding need to reinvent the wheel.
  It supplements existing information-retrieval
  systems.
• Integration of government records, news accounts,
  and opinion data should facilitate fuller
  insights into the underlying causes and politics
  of issues.
• Comparability with national database should
  provide deeper understanding of the
  inter-relationship of federal and state policies.
• Policy-makers and aides can more efficiently
  research recurring issues and previously-tried
  solutions, avoiding need to reinvent the wheel.
  It supplements existing information-retrieval
  systems.
• Integration of government records, news accounts,
  and opinion data should facilitate fuller
  insights into the underlying causes and politics
  of issues.
• Comparability with national database should
  provide deeper understanding of the
  inter-relationship of federal and state policies.

• Participating Universities
• Temple University College of Liberal Arts
  (project HQ).
• Penn State University, University Park
• The Graduate School of Public and International
  Affairs, University of Pittsburgh
• The Heinz School of Public Policy and Management,
  Carnegie Mellon University
• Penn State University, Harrisburg
• The Fels Institute of Government, University of
  Pennsylvania.

Status of Record Processing

• Further InformationOur web site is at
  www.temple.edu/papolicy. Although datasets are
  not yet available, you can see
• Our code book, forms used for record processing,
  and update guidance we have sent to researchers.
• A progress report that integrates data sets for
  1999 and 2000 as an example of final output.