Measuring Media Bias

1
Measuring Media Bias

• Michael Cardwell
• George Mason University
• March 1, 2005

2
Overview

• These are headlines from March 10 and 11 2004
  regarding the case of Susan Lindaurer who was
  accused of illegally taking payments to lobby for
  Iraq. All are true, yet they shed different
  light on the situation
• Washington Post Md. Activist Charged With
  Working for Iraq Indictment Says Ex-Press Aide
  Had Intelligence Links to Hussein
• NY Times Former U.S. Aide Accused of Working
  With Iraq
• Seattle Post-Intelligencer Accused spy is cousin
  of Bush staffer
• Wiz Bang Blog Democratic Staffer Indicted For
  Spying For Iraq
• Is this media Bias at work?

3
Overview

• Headlines regarding a single event can be
  factually true yet lead readers down different
  ideological and political paths
• Is the news media biased? Conventional wisdom
  and scholarly evidence often suggests it is, but
  compared to what?
• Lott and Hasslett (2004) Disparate reporting on
  similar events.
• Groseclose and Milyo (2003) Reporting ideology
  does not reflect political ideology

4
Theory

• Media bias matters because we require political
  information to make political decisions.
• In light of this requirement, systematic bias
  from journalists is suggests we receive
  misleading information with which to make
  decisions

5
Theory

• A public interest approach (a la the Society of
  Professional Journalists) suggests an ethical
  solution. Journalists morally bound not to
  distort or misrepresent facts
• A public choice approach.
• (Need not be mutually exclusive to public
  interest approach).
• Examine institutional constraints and
  opportunities for biased journalism

6
Theory

• Bovitz et al (2002) demonstrate conditions under
  which leading can occur
• Leading media presents an ideologically
  distorted report that shifts reader opinion
• Bias a report that is ideologically at odds with
  a given point of reference.

7
Theory

• Groseclose and Milyo demonstrate method of
  observing bias using ADA Ratings
• ADA Ratings are 0-100 scores based on correct
  answers (5 points each) to 20 legislative
  questions.
• Identify ADA Ratings for Congress
• Identify ADA Ratings for media outlets
• Observe media outlets are well to the left of
  Congress
• Hence, it is at least feasible that leading
  occurs if bias is observed. If no bias is
  observed at all, leading is either not happening
  or VERY efficient.

8
Methodology

• Starting point is Groseclose and Milyo approach
  to measuring bias, but with significant changes
• Essential features
• Measures bias as difference between Newspaper ADA
  rating (based on editorial position) and House of
  Representatives ADA ratings
• In Aggregate for all major newspapers
• For Designated Market Areas (DMAs)
• New York
• Washington DC
• For 2001-2002 (107th Congress)

9
Methodology

• Starting point is Groseclose and Milyo approach
  to measuring bias, but with significant changes
• ADA ratings calculated directly from newspaper
  editorials
• Bias, if it exists, is likely to reflect official
  editorial positions
• Easier to do, generally less interpretive than
  passive analysis

10
Methodology

• Intuition is that market area is important. New
  York Times largely sold in New York.
• Market area, not national aggregate, is proper
  area for determining bias.
• DMAs are industry standard geographic
  categorization for newspaper markets.
• Circulation of major papers used to develop
  weighted average news outlet ADA score.
• Population and Congressional ADA ratings used to
  develop weighted average political outlet ADA
  score.

11
Results National ADA
12
Results District ADA
13
Results DMA ADA
14
Results Aggregate ADA
15
Results - Analysis

• In the measured markets, observed newspaper
  ideology is similar to observed political
  ideology.
• Market sample, however, is limited and liberal.
• Does not consider competition from other media
  types (e.g. radio, internet)

16
Conclusions Consumers Lead the Market

• Lack of observed bias implies leading by media is
  either very effective or not very effective at
  all
• I suggest the latter is the case
• Perfect leading by numerous independent media
  outlets, simultaneously resulting in an aggregate
  similar to the political market outcome seems
  unlikely
• An unbiased aggregate ideology that emerges
  because consumers of political information desire
  it and also vote appears more likely
• Hence, the market responds to a desire for
  ideology, it rather than impose it.

17
Conclusions A Happy Result?

• Is it an unsatisfying result to conclude
• Consumers demand political information tailored
  to ideological whim, rather than purely
  objective?
• Unbiased results exist in the aggregate, but all
  individuals and firms may be biased?
• Perhaps, but it appears to be an example of
  spontaneous order, but we also desire individuals
  who can view opposing perspectives
  sympathetically
• Several limits exist which limit the extent to
  which ideology can dominate

18
Conclusions Consumer limits

• Tullock Consumers of media buy a package of
  goods
• Desire information
• But also entertainment
• Purpose is not just to vote, but to converse
• Suggests consumers will forego some ideological
  purity to gain
• Non political information (e.g. Sports Page)
• Opposing political information (know your enemy)

19
Conclusions Market Forces

• Producers are not oblivious to consumer needs.
  They may trade-off on the ideology-content access
  as well.
• Observation Market concentration may promote
  moderation (note moderation is not necessarily
  good)
• Unlike Congressmen, Newspapers can avoid the
  really hot topics

20
Conclusions Issue Avoidance and Contestability
21
Conclusions Issue Avoidance and Contestability

• Newspapers do not take an overt stand on issues
  that may cost them profits.
• And finally, interpretations of facts can be
  inherently contestable and open to interpretation
  as in opening example.
• These factors, along with consumer desire for
  some opposition view, seem to set limits to the
  extent to which papers are slaves to consumer
  ideology and the fact widespread bias is not
  observed implies neither is the opposite the case.