Kiss 2013

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

This article was downloaded by: [University of Ulster Library]

On: 06 March 2015, At: 06:13


Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK

Mass Communication and


Society
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/hmcs20

Legislation by Agenda-Setting:
Assessing the Media's Role in
the Regulation of Bisphenol A
in the U.S. States
a
Simon J. Kiss
a
Journalism Program , Wilfrid Laurier University
Published online: 06 Sep 2013.

To cite this article: Simon J. Kiss (2013) Legislation by Agenda-Setting: Assessing the
Media's Role in the Regulation of Bisphenol A in the U.S. States, Mass Communication
and Society, 16:5, 687-712, DOI: 10.1080/15205436.2013.768345

To link to this article: http://dx.doi.org/10.1080/15205436.2013.768345

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the
information (the “Content”) contained in the publications on our platform.
However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness,
or suitability for any purpose of the Content. Any opinions and views
expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the
Content should not be relied upon and should be independently verified with
primary sources of information. Taylor and Francis shall not be liable for any
losses, actions, claims, proceedings, demands, costs, expenses, damages,
and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the
Content.
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions
Downloaded by [University of Ulster Library] at 06:13 06 March 2015
Mass Communication and Society, 16:687–712, 2013
Copyright # Mass Communication & Society Division
of the Association for Education in Journalism and Mass Communication
ISSN: 1520-5436 print=1532-7825 online
DOI: 10.1080/15205436.2013.768345

Legislation by Agenda-Setting:
Assessing the Media’s Role in
the Regulation of Bisphenol A in
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

the U.S. States


Simon J. Kiss
Journalism Program
Wilfrid Laurier University

Starting in 2008, debate about potential hazardous effects from exposure to


bisphenol A (BPA) migrated from the pages of scientific journals to the
U.S. media, regulatory authorities, and state legislatures. In the context of
deep scientific conflict about the existence of adverse health effects attributable
to BPA, this article asks why it was the case that some state legislatures con-
sidered or adopted legislative bans on products made from BPA, whereas
others did not. Drawing on existing theories of agenda-setting and policy
change via punctuated equilibrium as well as a well-defined methodology
(event history analysis), evidence of agenda-setting is presented. Particularly,
it is argued that routine and high-impact health coverage was significantly
related to the chance that a state legislature considered legislation banning
products made with BPA. This was indirectly, but importantly, related to
the actual adoption by state legislatures of legislative bans on products made
with BPA.

Simon J. Kiss (Ph.D., Queen’s University, 2008) is an Assistant Professor in the Journalism
program at Wilfrid Laurier University (Brantford campus). His research interests include the
media’s role in politics and the policy process with a focus on the media’s coverage of environ-
mental and scientific controversies.
Correspondence should be addressed to Simon J. Kiss, 73 George Street, Brantford,
Ontario, Canada, N3T 2Y3. E-mail: skiss@wlu.ca

687
688 KISS

INTRODUCTION AND THEORETICAL FRAMEWORK

Since 1997, some scientists and environmental groups have expressed


concerns about the consequences for human health posed by the common
chemical bisphenol A (BPA), which is widely used to harden plastics in bot-
tles and create effective sealants in jars. From 1997 to 2008, concerns about
this chemical were primarily restricted to the pages of scientific journals and
a small circle of regulatory authorities, scientific researchers, industry repre-
sentatives, and social movements. In 2008, however, the controversy spread
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

to European and North American media agendas with impressive speed and
force. In the wake of this explosion of media interest, nine American states
(Massachusetts, Vermont, Minnesota, Wisconsin, Washington, Maryland,
Maine, and New York), four national governments (Canada, Denmark,
the United States, and France), and the European Union have issued sub-
stantial regulatory prohibitions on the use, manufacture, sale, or import
of products made from the chemical, primarily polycarbonate baby bottles.
It is argued here that media coverage played an important role in shaping
the legislative process across the U.S. states. Although there is a
well-established literature that examines the diffusion of policies across
U.S. states, these studies have tended to ignore the effects of within-state
media coverage, emphasizing instead the demographic characteristics of
individual states (Berry, 1990), networks of professional policy entre-
preneurs (Mintrom, 1997), or, at best, levels of national media coverage
(Hays, 1997). However, there are good empirical and theoretical reasons
to think that within-state media coverage should play a role as well.
This article begins by presenting a theoretical framework for why we
would expect within-state media coverage to have shaped the legislative
process. Then it reviews the scientific debate about the potential hazards
from exposure to BPA with the goal of establishing the case that there is
no scientific consensus that humans are at risk of adverse health effects at
current levels of exposure. Then it introduces a statistical methodology that
is well suited to test the hypothesis that media coverage shaped the legislat-
ive process and details the control variables and data-collection procedures.
Last, it presents the results of two model fittings that establish that media
coverage—particularly critical, high-impact newspaper coverage—shaped
the diffusion of legislative bans on products made from BPA in U.S.
states by spurring the consideration of legislation in prolonged, multiyear
processes.
One finds some direct empirical evidence for the starting hypothesis in
Brewer and Ley (2011), who conducted a survey of residents of Milwaukee,
Wisconsin, in the wake of a high-profile advocacy campaign about BPA by
LEGISLATION BY AGENDA-SETTING 689

the Milwaukee Journal-Sentinel. The survey found that newspaper use was
significantly related to self-reported behaviors aimed at avoiding exposure
to BPA. Newspaper coverage, it seems, was able to concern citizens
sufficiently to change their behavior.
One finds theoretical justifications for this hypothesis in the voluminous
literature on agenda-setting. The power of the mass media to impact what
citizens deem to be important (the agenda-setting power) is very well
documented, particularly for issues that citizens do not experience in a direct
fashion, such as environmental issues (Behr & Iyengar, 1985; Erbring,
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Goldenberg, & Miller, 1980; McCombs & Shaw, 1972). Moreover, there is
equally strong evidence that newspaper coverage can influence what
politicians deem to be important issues. In his study of national-level
agenda-setting dynamics, Soroka (2002) distinguished between the media,
public, and the government’s agenda, arguing that the news media could
directly influence the policy agenda, because politicians use the news media
as a source of information of what constitutes an issue or problem. In
addition, Cook, Tyler, Goetz, and Gordon (1983) and Protess et al.
(1987) found evidence that increased coverage of an issue can make it more
salient for elites, even in the absence of effects on mass opinion. At the
municipal level, Mead (1994) found evidence that the local newspaper was
a major factor in pushing municipal government reform onto the policy
agenda over a prolonged period.
Other evidence suggests that this dynamic occurs specifically in the
domain of U.S. state legislatures as well. For example, Bybee and Coma-
dena (1984) found that newspapers were among the most frequently used
sources for information by legislators in Indiana. Herbst (2002) found that
legislative staff regularly relies on the news media as a source of information
and as a proxy measure of public opinion, suggesting an indirect route for
media influence. Another study of the Louisiana state assembly found that
nearly all legislators surveyed felt newspapers served as a source of infor-
mation for their legislative activity, regardless of whether they thought
newspapers provided balanced and accurate coverage (Kral, 2003, p. 46).
These trends are reflected in a quantitative analysis of the relationship
between the media agenda and the state policy agenda: Tan and Weaver
(2009) found moderate to high correlations between the number of stories
dedicated to any given issue in a state’s newspapers and the number of legis-
lative bills introduced in that session. Thus, media coverage can increase the
salience of any given issue for citizens, but it can also increase the salience of
any given issue for politicians. There is evidence of this at the state level.
But this body of evidence does not, in and of itself, explain policy change.
For that, it is necessary to adduce a theory of policy change via a process of
690 KISS

punctuated equilibrium developed by Baumgartner and Jones (2009).1 This


model suggests that specialists and insiders construct ‘‘policy images’’ out of
both factual and emotive elements that buttress a policy monopoly and
contribute to stability and stasis in any given policy field. This policy
image—a concept operationalized by measuring both the quality and quan-
tity of media coverage—is crucial to the stability of the policy monopoly
because it is almost the only contact that most citizens have with the
respective policy.
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Specialists, experts, and others with an economic interest in the nature of


public policy in a particular area are the dominant makers of public policy
in that area. Because the interested share preferences, or at least share under-
standings concerning the basic dimension of conflict, the paradox of voting
does not arise. The policy system is stable because those participating share
values. (Baumgartner & Jones, 2009, p. 19)

However, this stability—or equilibrium—is only ever partial. When that


policy image—the quality or the quantity of media coverage—changes
substantially, that policy monopoly breaks down.

In the process of agenda-setting, the degree of public indifference to given


problems changes dramatically. Since this is the structure on which policy sub-
systems are based, it should not be surprising if periods of agenda access are
followed by dramatic changes in policy outputs. Indeed, this is precisely
why policy entrepreneurs fight so doggedly either to push their issue toward
the public agenda or to ensure that it not arrive there. (p. 20)

Three studies of the role of media coverage in the fields of technology and
environmental health support the argument that changing levels and content
of media coverage can play an important role in shaping policies. Scheberle
(1994) found that a dramatic and sudden spike in national media interest
about the health effects of radon exposure was necessary to force formal
action by national legislators, whereas in the case of asbestos, a sustained
and constant period of media coverage forced the issue onto the formal
agenda. Nisbet and Lewenstein (2002) found that in the 1970s, newspapers

1
This model of policy change via punctuated equilibrium was developed primarily in con-
trast to previous models of policy change that emphasized stasis or incremental changes (e.g.,
Wildavsky, 1964). The relationship between the use of punctuated equilibrium in political
science and Wildavsky’s theory of budgetary incrementalism is analogous to the relationship
between punctuated equilibrium’s role in evolutionary biology, where it is contrasted to the pre-
vious theory of phyletic distinction, which posited steady, slow, and gradual evolution
(Eldredge & Gould, 1972).
LEGISLATION BY AGENDA-SETTING 691

covering biotechnology predominantly emphasized its contribution to


‘‘progress,’’ although there was also a cluster of stories raising questions
about the ethics and public accountability toward such technology. This
pattern contributed to a growing trend by local authorities to try to regulate
biotechnology (p. 378), as well as a reaction by scientists, seeking to form a
consensus around the importance of scientific autonomy in research. But in
the latter half of the 1990s, the policy image of biotechnology shifted again
in reaction to announcements about risks of gene therapy and possible eco-
logical consequences from genetically modified foods. This change in policy
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

image contributed to changes in policy at the national level including to the


repeated introduction of legislation that would ban human cloning. Last,
Pralle (2006) related a dramatic increase in Canadian coverage of domestic
pesticide use between 1999 to 2002 to an explosion in municipal bylaws
controlling pesticides from 2002 to 2005.
Thus, existing policies that seem to be stable and resistant to change are
not the way they are because of aggregated preferences. They are made that
way by those actors that have a compelling interest in the outcome of any
given regulatory processes. However, when the composition of those who
are aware of the regulatory process in any given policy field changes, that
is, when public attention is aroused by media coverage, the actors involved
begin to make different decisions. New majorities can emerge and seemingly
stable policy subsystems can change quite quickly.
Last, there also exists good evidence to suggest that the public carries an
intrinsic skepticism toward technological and scientific innovations in gen-
eral that could be triggered simply by an increased salience of a scientific
debate about the risks posed by BPA. For example, Mazur (1981) identified
moderate correlations between the level of coverage given toward fluori-
dation and nuclear power and opposition to these technologies. He postu-
lated that even balanced coverage of scientific disputes can persuade
voters to adopt a precautionary approach.

I have suggested that the quantity of coverage of a technical controversy can


have as much an effect on public attitudes as the semantic content of the stor-
ies that are presented. The public takes seriously any suggestion that a tech-
nology may be risky, particularly if the suggestion is repeated often enough.
(Mazur, 1981, p. 114)

This is supported by recent work on risk perception, which describes the


way in which risks perception is governed by affect, rather than reason. As a
result of evolutionary forces and brain structure, this research suggests that
there is an intrinsic bias to fear human-made risks more acutely than natural
risks (e.g., risks posed by extreme weather events; Ropeik, 2010).
692 KISS

The theoretical framework presented here can account for two major
characteristics of the diffusion of bans on products made with BPA. On
one hand, it can account for the expansion of the issue first from a small
community of scientists, environmental activists, chemical firms, and regu-
lators at the Food and Drug Administration (FDA) to state legislatures.
As social movements and scientists were concerned about regulatory
inaction, they looked to expand the venue in which this conflict was taking
place. In this they were aided by newspapers which capitalized on the
preexisting scientific conflict and made it a genuine public issue (on the
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

need for conflict to generate an issue, see Cobb & Elder, 1971). Second, it
can help explain why some states adopted legislative measures at different
points in time, namely, because levels of media coverage in each state
differed.
It is important to emphasize that nothing in this conceptual framework
precludes the news media as serving as an intervening variable between
important, politically relevant actions (e.g., lobbying by environmental
groups) and regulatory outcomes. However, it does imply—and supportive
evidence is provided next—that, at minimum, politically interested actors
must transform the policy image (i.e., change the quantity and quality of
news coverage) essentially to spur state legislators to action. This implies
raising the attention of disinterested, unknowing, or apathetic citizens via
news media coverage and changing the tone of media coverage, making it
less sympathetic to existing policy insiders. The theoretical framework
presented here conceptualizes high levels of critical media coverage as a
necessary, perhaps not a sufficient, tool on the road to policy adoption in
a U.S. state.
In the following analysis, the hypothesis is tested that media coverage
shapes policy change by examining the impact that raw levels of media
coverage, levels of critical health news, and levels of ‘‘high-impact’’ feature
stories had on the life of BPA as a political issue through the U.S. state
legislatures. However, first, the case is made that something other than a
scientific consensus drove policy.

SCIENTIFIC DEBATE OVER HAZARDS OF BPA

Concerns about BPA are part of a widespread debate over so-called endo-
crine disruptors, which are synthetic compounds that have been accused
of having adverse effects on animal and human health, particularly in
regards to human reproductive systems (Colborn, Dumanoski, & Myers,
1997). In 1997, Professor Frederick vom Saal and colleagues published
the results of a study that reported that exposure to very low doses of
LEGISLATION BY AGENDA-SETTING 693

TABLE 1
Exposure Estimates to Bisphenol A

x-fold <NOAEL
Age category x-fold<TDI(50 micro- (5,000 micro-
w per day Micrograms=Kg=B grams=kg=bw=day) grams=kg=bw=day)

0–4month-old infant 1.6 31 3100


6–12-month-old infant 0.8 62.5 6250
4–6-year-old child 1.2 41.6 4160
60 kg adult 0.37 135 13500
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Note. Source: European Commission. Scientific Committee on Food. (2002). bw ¼ body


weight.

BPA contributed to increased prostate glands in mice (Nagel et al., 1997),


launching a heated scientific debate which remains unresolved today.2
However, claims that BPA is hazardous to human health have been dis-
puted within scientific communities for several important reasons. Although
there is plenty of evidence to suggest that BPA can cause adverse effects at
levels far above what humans are exposed to, there are serious questions
about the reliability of evidence that suggests there are adverse effects at
environmentally relevant levels of exposure. Most regulatory agencies
accept a tolerable daily intake level for humans of 50 micrograms of BPA
per kilogram of body weight per day. This is derived from two multigenera-
tion, multidose studies commissioned and supervised by the European Food
Safety Authority (EFSA) and financed by the chemical industry (Tyl,
Myers, Marr, & Sloan, 2008; Tyl et al., 2002). Based on those studies, the
EFSA settled on a dose of 5 mg of BPA per kilogram of body weight per
day (kg=bw=day), below which there were no observable adverse effects.
Then, adopting a variation of the precautionary principle, it reduced that
limit by a factor of 100, resulting in a tolerable daily intake level of 50 micro-
grams=kg=bw=day. This level and these studies remain a benchmark in the
worldwide regulation of BPA. The U.S. FDA, Food Standards Australia
New Zealand, and the United Kingdom Food Services Authority all accept
this level, whereas Health Canada adopts a slightly more conservative and
precautionary approach, settling on a tolerable daily intake level of 30
micrograms=kg=bw=day.
It is evident from Table 1 that human exposure to BPA is far less than the
levels deemed by regulatory agencies as safe. Thus, virtually the entire
debate around the potential for effects hazardous to humans has centered

2
There are far more than 1,000 studies about BPA indexed in the PubMed database.
694 KISS

on the question of whether there are observable, adverse effects below these
tolerable daily intake levels. This ‘‘low-dose’’ hypothesis remains heatedly
controversial within toxicology and the scientific community remains
divided.3 Scientists are divided on fundamental questions of experimental
science: the validity of industry-funded science, the way in which causation
is assessed, the validity of animal-based findings for human risk assessment,
the meaning of null findings and the danger of publication bias. Nagel et al.
(1997) observed increased weights of mouse prostate glands at levels of 2
micrograms=kg=bw=day, and this finding serves as a bedrock for the scien-
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

tific argument for hazardous effects from contemporary exposure to BPA.


However, from the point of view of regulatory agencies, and most other tox-
icologists, the studies purporting to show hazardous effects at low levels
have been seen as flawed or not persuasive for several reasons. The EFSA’s
most recent assessment sums up the state of the science surrounding BPA
thusly:

EFSA was asked to evaluate a dietary developmental neurotoxicity study in


rats (Stump et al., 2010) and recent scientific literature (2007–2010) in terms
of relevance for the risk assessment of BPA. The impact of these studies on
the current TDI [Tolerable Daily Intake] of 0.05 mg BPA=kg=bw=day as set
by EFSA in 2006 was assessed. . . . Overall, based on this comprehensive evalu-
ation of recent toxicity data, the Panel on food contact materials, enzymes, fla-
vourings and processing aids (CEF) concluded that no new study could be
identified, which would call for a revision of the current TDI. . . . The Panel
noted that some studies conducted on developing animals have suggested
other BPA-related effects of possible toxicological relevance, in particular bio-
chemical changes in brain, immune-modulatory effects and enhanced suscep-
tibility to breast tumors. These studies had several shortcomings. At present
the relevance of these findings for human health cannot be assessed. . . . A min-
ority opinion is expressed by a Panel member and presented in an Annex to
this opinion. (European Food Safety Authority, 2010)

Even if the reader is not yet convinced that the scientific evidence under-
lying the claim hat contemporary exposure to BPA poses hazardous risks to
humans is inconclusive at best, then it is hoped that one will at least be per-
suaded that the rhetoric characterizing this position far outstrips the scien-
tific evidence. On one occasion, vom Saal argued that ‘‘the science is clear
and the findings are not just scary, they are horrific. When you feed a baby
out of a clear, hard plastic bottle, it’s like giving the baby a birth control

3
The European Food and Safety Authority convened a high-level scientific meeting in June
2012 to discuss the status of the scientific debate.
LEGISLATION BY AGENDA-SETTING 695

pill’’ (University of Missouri, Columbia, Division of Arts And Sciences.,


2005). On a later occasion, he claimed that ‘‘this is the global warming of
biology and human health’’ (as cited in Neimark, 2008). This alarmism
has been reflected in the news media. In 2007, the Globe and Mail published
a two-page feature on the scientific debate, featuring a demand by the execu-
tive director of Environmental Defence that Health Canada take steps to
ban the chemical before its risk assessment was complete and emptying
his house of polycarbonate bottles (Mittelstaedt, 2007). On December 8,
2010, the same newspaper published a story referring to a non-peer-reviewed
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

report from a coalition of American environmental groups that found traces


of BPA on receipts and cash bills. The paper darkly wrote in the first sen-
tence,‘‘There’s a new reason other than fear of germs to wash hands after
handling paper money: It contains traces of bisphenol A, the estrogen-like
chemical Health Canada has declared toxic’’ (Mittelstaedt, 2010).
There is a deep scientific conflict over whether BPA causes harmful
effects to humans at levels at which we are currently exposed. Regulatory
agencies have almost unanimously dismissed the existing concerns as unre-
liable and not proven. Therefore, something other than scientific consensus
has driven regulatory policy in U.S. states. The claim here is that media
coverage is partially, but importantly, responsible for shaping regulatory
outcomes in the U.S. states.

THE EVOLUTION OF BPA AS A POLITICAL ISSUE


IN THE UNITED STATES

Between 1997 and 2005, the debate about BPA was restricted almost entirely
to the scientific community, a few environmental organizations in the
United States and regulatory authorities in the National Toxicology Pro-
gram (NTP) and the FDA. Between those years, the number of
peer-reviewed studies increased substantially, with no corresponding
increase in regulatory activity. However, this started to change in 2005,
and the issue expanded out of this narrow scientific and regulatory circle.
First, state legislatures started to take up the issue in that year, with Califor-
nia debating legislation in 2005 and Minnesota and Maryland in 2006.
Second, the Wall Street Journal published a series of five news articles in
2005, one of which focused on the scientific debates around BPA. Third,
the NTP began a 3-year process of formally evaluating the risks to human
health posed by BPA.
Although the issue had migrated from the scientific to the regulatory
arena after 2005, the circle of actors remained small; it had no prominent
place on the public agenda. But the simmering conflicts between scientists,
696 KISS

regulators, and environmental groups made it possible for the conflict to


migrate to the news media. The year 2008 was a turning point in this regard.
First, the NTP issued a draft brief on its assessment of potential hazards
from BPA on April 14, 2008, and reported ‘‘some’’ concern about effects
from exposure to BPA on the development of fetuses and infants. This made
a minor impact on the media agenda; nine stories in this study’s data set
appeared following the announcement, including a front-page story in the
Washington Post. Second, and more dramatically, four days later Health
Canada published its own risk assessment of BPA, declaring it to be ‘‘toxic’’
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

according to the provisions of Canadian legislation and proposing contro-


versial risk management strategies, including that polycarbonate baby bot-
tles be banned. This made a much larger impact in American newspapers, in
part because Canada’s decision had a substantial impact on the market-
place. Wal-Mart indicated it was withdrawing bottles made from BPA from
sale in the United States, and Nalgene, one of the most popular makers of
polycarbonate plastic bottles ceased production of those products. Twenty-
two stories appeared in American newspapers in the two days following
Canada’s decision and the resulting changes by retailers. Third, the
Milwaukee Journal-Sentinel began an in-depth advocacy campaign in 2008,
publishing more than 40 stories over the course of the year. Later, in
September, a study published in the Journal of the American Medical Associ-
ation (Lang, Galloway, Scarlett, & Henley, 2008) attracted substantial
media attention (19 stories on September 17, 2008). This was a cross-
sectional survey of American blood samples that detected the presence of
BPA in 93% of the American population, with the highest concentrations
in those suffering from heart disease, diabetes, and liver problems. Although
its cross-sectional nature could not in any way assess causation, this subtlety
eluded most journalists. For example, the subsequent headline of the
Washington Post read, ‘‘Study Links Chemical BPA to Health Problems’’
(Layton, 2008). However, the increased media coverage in 2008 was not
equally high across all U.S. states; BPA entered the media agenda in parti-
cular states at particular points in time. These changes, it is argued here,
changed the calculations of legislators in individual states. In the absence
of FDA action on an issue that was suddenly high on the media’s agenda,
legislators in some states at particular points in time had strong political
incentives to act, whereas others did not.

METHODOLOGY AND DATA

One way to test formally whether within-state media coverage of BPA was
related to the legislative process is via event history analysis. This is a
LEGISLATION BY AGENDA-SETTING 697

well-established statistical method with roots in epidemiology, medicine,


public health, and engineering processes where the object of interest is the
start of some process until the onset of some other event (usually death,
failure of a machine process, or the onset of some condition; J. M.
Box-Steffensmeier & Jones, 2004, pp. 1–7). The technique has been adapted
for use in political science to map factors that influence the rate and adop-
tion of particular policies across the 50 American states (Berry, 1990). For
example, Haider-Markel (2001) examined the diffusion of bans on same-sex
marriage and Chamberlain and Haider-Markel (2005) examined the spread
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

of ‘‘lien’’ laws as a deterrent to hate crimes across the U.S. states. Hays
(1997) used event history analysis to argue that the level of media coverage
about living wills increased the probability that a state would adopt such
legislation in any given year. Similarly, Brace, Hall, and Langer (1998)
examined the factors that contributed to court cases being filed challenging
legislative restrictions on abortions following the Roe v. Wade decision.
These models are estimated to assess whether independent variables of
interest have a significant impact on what is known as the ‘‘hazard rate.’’
For discrete data (as is here the case), the hazard rate is simply a ratio of
the probability that a unit failed in any discrete period (in this case a legis-
lative year) to the probability that a unit would survive up to the same per-
iod. To express this colloquially, the rate of failure (dying) for humans in the
100th year of life is quite high; half of 100-year-olds might die in their 100th
year. But the probability of any given human surviving to 100 is quite low.
The hazard rate of failure (death) for humans in the 100th year of life is the
ratio of the failure rate in that year to the probability of survival to that
point. Of course, calculating the hazard rate of any given process is usually
only a starting point for any analysis. Usually, one is interested in assessing
whether any configuration of independent variables has any discernible
impact on the hazard rate.
In this case, we fit a Cox proportional hazards model to the data to exam-
ine the role that media coverage played in the spread of debate on policies
about BPA. It is important to emphasize that the Cox model is not a para-
metric model. That is to say, it makes no claims to estimate how long a pro-
cess might take until the event of interest happens; rather, it assesses the
change in hazard rates that independent variables can bring about. To
put this in the language of the current study: A Cox model of proportional
hazards could tell us the ratio of risk that Democratic states have in experi-
encing the event of interest (a ban on BPA) compared to Republican states,
but it could not tell us that the time to adoption of legislation for Demo-
cratic states was four years, whereas for Republican states it was six years.
It is worth pointing out that there are distinct legislative pathways
that are apparent in the diffusion of bans on products made with BPA
698 KISS

TABLE 2
Distribution of Legislative Processes

Counts

No events 14
Consideration without adoption 14
Debate in more than one session leading to adoption 8
Adoption in one session 1
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

(see Table 2). In 14 states, there were no legislative activities, whereas in


another 14, the state legislature considered legislative bans at least once,
without ever adopting a ban. Eight state legislatures considered a legislative
ban in one sitting, with the legislature later adopting a ban. Last, in one
state, the state legislature adopted a ban in one sitting without ever having
considered a ban in a previous sitting.
Two substantive questions arise from this observation. First, did news
media coverage contribute differently to legislatures considering (but not
adopting) legislative bans? One might hypothesize, for example, that high
levels of critical news media coverage might spur a legislature to consider
a legislative ban on products made on BPA but that other factors might
be more important in dictating whether the state legislature would actually
adopt that ban. For example, in the state of Oregon, the legislature has
repeatedly considered banning products made with BPA on several occa-
sions, but not adopted a ban, in part because of opposition by that state’s
fruit industry, which relies on the use of BPA to properly seal jars. Thus,
media coverage might be related to the chance that a legislature considered
legislation, but it might not be related to the risk that a legislature would
adopt a ban. Second, media coverage might play different roles depending
on the sequence of events. One could imagine a legislative process within
a state that is started by high levels of critical media coverage. For example,
one could imagine an environmental organization generating publicity,
spurring a legislator to action, introducing legislation banning products
made with BPA. However, one could also imagine an attentive legislator
taking notice of the scientific debate before any state newspaper and who,
acting out of sincere policy convictions, subsequently introduced legislation.
In the former case we would expect higher levels of media coverage prior to
the first legislative consideration, whereas in the latter, we would expect low
or no levels of media coverage.
Moreover, even if media coverage were not related to first legislative con-
siderations, it could also plausibly be the case that media coverage might
matter more in sustaining multiyear legislative attempts, rather than in initi-
ating them. Consider cases such as Illinois, where legislation banning
LEGISLATION BY AGENDA-SETTING 699

polycarbonate baby bottles was introduced in 2008, debated again in 2009,


reintroduced in 2010, and debated again in 2011, without ever being
adopted. Regardless of whether the news media played a role in initiating
that process, it might be the case that the process was sustained in later years
by high levels of media coverage to the issue. For example, one could
imagine an innovative legislator taking steps to introduce the legislation;
generating publicity; and, because of the attention it garnered, subsequently
reintroducing the bill the following year. Or one could imagine the same
legislator introducing the bill, failing to get publicity, and deciding not to
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

pursue the legislation in search of higher profile issues.


In the terminology of event history analysis, this complex political pro-
cess exhibits both repeated events (state legislatures can consider or adopt
legislation in several sittings sequentially) and competing risks (state legisla-
tures can do nothing, consider legislative bans, or actually adopt legis-
lation). Fortunately, event history analysis has developed sufficiently to
provide methodological tools to evaluate these types of questions (see, in
particular, Box-Steffensmeier & Zorn, 2002; Jones & Branton, 2005).4

Case Selection
In the case of the American states, 37 states are analyzed over the period
2005 to 2011. Thirteen states were eliminated because there was no suitable
daily newspaper within the state contained in the Lexis-Nexis database
(e.g., Delaware and Vermont), or there were no data on the environmental-
ism variable (e.g., Alaska and Hawaii). Then, state–year combinations were
created for each year in which the state legislature sat and could potentially
have passed legislation regarding BPA. Although most state legislatures
meet annually, some meet biannually; for those states, years where the
legislature did not sit were deleted.

Legislation Proposing to Ban Products Made with BPA


The dependent variable in the study is the rate at which states adopted or
considered legislation banning products made from BPA. Using data from
the National Council of State Legislatures, which tracks legislation on
environmental health initiatives, the progress of BPA legislation in the states
was gathered (National Council of State Legislatures, n.d.). The author
searched the database for bills responding to the search strings

4
For those interested in the technical details, the data set is doubled to account for compet-
ing risks. There each state–year combination is represented twice. To account for repeating
events, a conditional interevent counting process developed by Prentice, Williams, and Peterson
(1981) and advocated by Box-Steffensmeier and Zorn (2002) is implemented.
700 KISS

‘‘Bisphenol A,’’ ‘‘BPA,’’ and ‘‘Bisphenol-A’’ for legislation in the states


from 2005 to 2011. For each year, it was noted whether there was no legis-
lative activity, a consideration without an enactment, or an enactment of
legislation banning products made with BPA. Thus, there was no distinction
made between a session where a brand new bill was introduced or whether a
bill from a previous session was carried over. What mattered was whether
the legislature considered legislation without enacting it. Last, some bills
focused only on prohibiting products meant for children (baby bottles
and toys), and other bills focused on receipt paper, whereas other bills made
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

no distinction. For the purposes of this study, the only outcome of interest
was that there was a bill that considered banning products made from BPA
enacted or considered in any given session.

News Coverage
The primary independent variable of interest is within-state news coverage.
To gather this, a search was conducted in all daily newspapers in each state
contained in the Lexis-Nexis database. To be included in the data set, stories
had to be 150 words long and either mention the phrase ‘‘Bisphenol A’’
twice or mention the search string in the headline or the lead of the news
story. These measures were introduced to prevent frivolous and irrelevant
stories such as news digests or passing references to BPA from being
included. The stories selected represent a census of nontrivial newspaper
coverage in American daily newspapers indexed in Lexis-Nexis from the
period 2004 to 2010. The number of newspaper stories for each state pub-
lished in each period was normalized by dividing the number of stories by
the number of newspapers. Then an undergraduate student coded each
newspaper story along two variables, topic and tone.5

5
We achieved intercoder reliability in the following way. The author developed a coding
scheme provided to and discussed with the student. The initial pilot sample received low
reliability scores (0.46 for tone and 0.49 for topic, N ¼ 20, all statistics here are Cohen’s Kappa).
We discussed problems and revised the coding scheme, achieving much better results for topic
on the second round (0.86 for topic) but still low results for tone (0.49, both N ¼ 20). We revised
the coding scheme for tone again, providing clearer directions for distinguishing stories critical
of BPA from simply neutral accounts of developments in the story. Again, we achieved a low—
albeit improved—score for tone (0.52, N ¼ 20). It was apparent that all but two of the stories in
this subsample had been coded correctly. The author further clarified ways for the student to
distinguish ‘‘critical’’ from ‘‘neutral’’ stories, and a final sample was taken, which resulted in
identical coding. Cohen’s Kappa was 1 (N ¼ 25). In addition to the restrictions adopted at
the search string stage, the undergraduate coder also marked news stories that could only be
deemed as irrelevant. These were stories that were not caught by either of the restrictions just
noted but were still irrelevant to the debate at hand. An example of stories excluded under these
criteria was a story about BPA-free products as gift ideas.
LEGISLATION BY AGENDA-SETTING 701

TABLE 3
Number of News Stories by Tone and Year

2004 2005 2006 2007 2008 2009 2010

Critical of BPA 3 6 8 45 173 67 88


Neutral 1 1 3 22 46 11 16
Dismissive of concerns 0 0 0 4 16 3 0
about BPA
4 7 11 71 235 81 104
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Note. BPA ¼ Bisphenol A.

Most of the coverage was critical of BPA, but it involved a wide range of
topics (see Tables 3 and 4). Two clusters are evident: One cluster of news
stories included those that report on regulatory or political activity about
steps to regulate BPA. These types of news stories reported on legislation
moving through stages of debate, or on legislative initiatives by legislators
or the regulatory initiatives by the FDA. A second cluster focused on poss-
ible negative effects posed by BPA. These stories included reporting on the
publication of peer-reviewed studies about the potential negative effects,
activity by social movements to demonstrate or highlight negative effects,
or stories about commercial responses, including manufacturers offering
BPA-free products because of fears of adverse effects.
In the foregoing analysis, we test hypotheses about relationships between
three different types of news coverage and legislative outcomes. First, we
examine the relationship between absolute levels of news coverage about
BPA, with no regard for topic or tone, and legislative outcomes. We test this
hypothesis because of previous research findings previously noted that the
high prominence of news coverage, regardless of quality, can evoke
deep-seated popular fears about technological innovations (see Mazur,
1981). Second, the relationship between routine critical news stories about

TABLE 4
Topics of News Stories About Bisphenol A by Year

2004 2005 2006 2007 2008 2009 2010

Critical peer-reviewed studies 1 5 5 13 30 16


Social movement activity about negative effects 0 0 0 11 6 8 23
Industry activity responding to concerns 0 0 0 11 42 7 7
Industry activity denying concerns 0 0 2 0 4 6 0
Government activity regarding regulation 1 0 1 25 93 8 42
Political activity regarding regulation 0 1 2 4 28 32 15
Other 2 1 1 7 32 4 6
702 KISS

possible adverse health effects from exposure to BPA and legislative outcomes
is tested. This is examined for two reasons. First, according to Baumgartner
and Jones, the quality of news coverage is as important to the policy image as
quantity of coverage. It might be the case, therefore, that it is not absolute
levels of routine news coverage about BPA, which are linked to outcomes,
but levels of a particular type of news coverage. Given that there is a clear
cluster of news stories that deal with the health and safety of BPA, we assess
whether it is levels of routine, critical health coverage that impact the legislat-
ive outcome. Second, by restricting news coverage to encompass only news
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

stories that are critical in tone and deal only with adverse health effects, we
avoid a possible problem with endogeneity. Because many of the cases under
consideration involve multiyear legislative processes, if all news stories are
included, the risk exists that news stories reporting on a bill’s passage in 1 year
will be counted toward the legislative outcome the following year.
Last, there also exists the possibility that the legislative process could be
influenced by high-impact news stories rather than by a series of routinely
generated, low-impact news stories. Story length can communicate a great
deal of information about the salience of an issue, with longer stories com-
municating greater importance. For example, Pritchard (1986) found that
newspaper story length was inversely related to the chances that a pros-
ecutor would negotiate a plea bargain, whereas Peter (2003) found that
the number of stories, weighted by prominence, was importantly related
to the salience of European integration in a cross-national comparison.
Although the Lexis-Nexis database does not contain any consistent and
reliable information on the position of any story in the newspaper (e.g.,
front page, Life section, etc.), it does contain reliable information on the
word length of each story. Thus, in addition to the health news variable just
introduced, the content analysis coded a news story as a ‘‘high-impact’’
story if it was 1 standard deviation longer (359 words) than the average
word length (697 words). Thus, high-impact stories were longer than
1,056 words. To control for the number of newspapers that are included
in the data set, each variable is normalized by dividing by the number of
newspapers in the Lexis-Nexis database. Moreover, to avoid problems with
causal inference, all news stories are lagged by 1 year; that is to say, a story
that is published in the calendar year in 2008 counts only toward the out-
come of the 2009 legislative calendar (see Mills, 2011, p. 93).
Table 5 lists standard descriptive statistics for the continuous and discrete
variables in the analysis; it immediately reveals one challenge the distri-
bution of news stories presents. Namely, a significant portion of the cases
had no news stories about the issue, whereas only a small fraction had
any news coverage. As a consequence, hypotheses about relationships
between three news variables are tested first by converting news variables
LEGISLATION BY AGENDA-SETTING 703

TABLE 5
Descriptive Statistics

Variable Cases Null cases Missing values Min Max M SD

Professionalism 500 0 0 0.10 0.75 0.43 0.19


Environmentalism 500 0 0 0.07 0.63 0.21 0.12
Regional 500 468 0 0 0.50 0.02 0.08
All news 500 292 0 0 7.40 0.77 1.34
Critical health news 500 344 0 0 3.00 0.26 0.52
High impact health news 500 458 0 0 1.00 0.03 0.14
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Partisanship 500 (51)a


a
% Democrat.

into dichotomous variables. This does not allow for the testing of hypothe-
sized linear relationships between news and outcome, that is, as news cover-
age increases, the chance of legislative action increases. Rather, this allows
one to test whether any level of news coverage compared to no news cover-
age increased the chance of legislative action. Given the lopsided distri-
bution of cases where there was no coverage compared to where there
was coverage, this is an important assessment.

Environmental Sentiment
Although the issue of the potentially toxic effects of BPA on human health
may seem to be only tangentially related to traditional environmental issues
of ecosystem protection and species preservation, it remains the case that
BPA has manifested itself primarily as an environmental issue. In the United
States, the Environmental Working Group was active on the issue. It is
worth testing, therefore, whether the salience of environmentalism was also
an important variable that distinguished those jurisdictions that adopted a
ban from those that did not. Here, a measure of environmentalism is used
that was developed by Mazur and Welch (1999). They assigned a score to
each state (save Alaska and Hawaii) on an index derived from four mea-
sures: the size of the membership of three environmental organizations,
the rating of the state’s congressional delegation by the League of Conser-
vation Voters, the percentage of respondents saying the government spends
‘‘too little’’ on the environment over a period of 20 years, and a rating of
state policy on 50 different environmental policies.

Partisanship
Most legislatures that adopted bans on products made with BPA were con-
trolled by the Democratic party, moreover, environmental issues tend to be
704 KISS

promoted by those on the left of the political spectrum. Therefore, the


model includes a variable for the partisan composition of the legislature
for the relevant period, drawn from information contained in the U.S. cen-
sus (U.S. Census Bureau, 2011). This variable could take on a value of
‘‘Democrat’’ if both houses of the legislature were controlled by that party
in that year or ‘‘Non-Democrat’’ if there was any other configuration.6

Professionalism
A common finding in the literature on policy diffusion across the American
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

states is that states with professional legislators (e.g., a high level of compen-
sation with full-time staff support) tend to exhibit a greater willingness to
adopt policy innovations (Kousser & Cain, 2004, p. 168). Accordingly, this
model integrates the dominant variable for measuring the professionalism of
a legislature, a measure developed by Squire (2007) that accounts for salary,
staff support, and demands on the legislator’s time.

Regional Diffusion
One of the more robust findings in the literature on policy diffusion across
U.S. states is the impact of neighboring states on policy adoption (see
Mooney, 2001). The models introduced in the next section introduce a vari-
able that captures what proportion of any given state’s neighbors had
adopted a ban to control for this effect.

RESULTS AND ANALYSIS

As a first-pass analysis, the graphs in Figure 1 suggest important initial evi-


dence of a relationship between media coverage and the legislative process.
The first depicts the absolute, normalized number of stories per newspaper
prior to legislative sittings with various outcomes. The second graph shows
the levels of routine, critical health news coverage, and the last shows levels
of high impact, critical health coverage. In each graph, coverage is grouped
by different types of legislative events (an initial consideration, a later con-
sideration, a ban with no prior event or a ban with a prior event). For each
type of event, coverage is compared between states where the event did
occur to where it could have occurred but did not occur. The reader will
notice that for nearly every type of legislative outcome and for every type
of news coverage, coverage was higher prior to legislative sessions where
events occurred compared to legislative sessions where nothing occurred.

6
Nebraska was coded as ‘‘non-Democrat’’ because it has a non-partisan legislature.
LEGISLATION BY AGENDA-SETTING 705
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

FIGURE 1 Average number of three types of news stories—all news stories with no regard for
content, routine, critical health stories and high impact, critical health stories—by different legis-
lative outcome. Note. Values marked by gray bars are the normalized number of stories appear-
ing prior to legislative sessions where the legislature was at risk of experiencing that particular
legislative outcome, but it did not occur. Values marked by black bars are the normalized num-
ber of stories appearing prior to legislative sessions where the same outcome did occur.

This is highly suggestive of agenda-setting effects; however, these rela-


tionships do not control for any confounding variables. To do this, we fit
a multivariate Cox proportional hazards model as just discussed. First, we
examine the relationship between news coverage and first and second legis-
lative considerations. Results are reported in Table 6.7 In the first column,
no distinction is made between first and later events. The coefficients in this col-
umn could be considered to represent the ‘‘average’’ effect of the independent

7
The analysis that follows adopted the following modeling strategy. First, univariate models
were fit with each variable of interest just identified. Variables that had p values greater than .25
were excluded. Then a multivariate model was fit with the remaining variables. Last, models
were assessed for linearity, conformity with assumptions, and any interactions between
variables.
706 KISS

TABLE 6
Modeling the Repeating Events Nature of Legislative Considerations

Pooled First Second

Professionalism 0.29 (0.81) 1.97(2.41) 0.71 (0.70)


Environmentalism 2.97 (1.08) 4.75 (1.10)
Categorical critical health news 0.43 (0.46) 0.58 (0.77) 0.92 (0.75)
y
Categorical high impact health news 0.77 (0.29) 1.16 (0.62) 0.68 (0.26)
Categorical allnews 0.74 (0.52) 0.74 (0.50) 0.35 (1.09)
Partisanship 0.62 (0.58)
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Regional diffusion 3.18 (1.33)


N 250 199 51
Wald 31 on 5 df 53 on 6 df 13 on 5 df
p < .001 p < .001 p < .05
R2 0.09 0.14 0.12
Max 0.69 Max 0.56 Max 0.92

Note. Robust standard errors in parentheses.


y
p < .10.  p < .05.  p < .01.  p < .001.

variables on the hazard rate, without distinguishing the order of events in any
way. This initial fitting suggests positive relationship levels of high-impact
news coverage and environmentalism in state public opinion. On average, a
legislature was 2.1 times as likely to consider legislation when there was a
high-impact news story published prior to a legislative session compared to
legislative sessions where there had been no high-impact news story.
The next two columns report the coefficients for two separate models.
The first is fit only to cases at risk of experiencing the first occurrence of
a legislative consideration, whereas the second is fit only to cases at risk
of experiencing a second or later legislative consideration.
Disaggregating the legislative events by chronology reveals only a slightly
different pattern. On average, a state was 3.2 times more likely to consider legis-
lation for a first time when there was a high-impact story published in the pre-
vious calendar year, although this finding was only significant at the .1 level
(p ¼ .06). High-impact coverage was strongly and significantly related to the
chance a legislature would consider legislation a second time; states were 1.9
times more likely to consider a second piece of legislation following any high-
impact coverage compared to when there was no high-impact coverage. Neither
of the two other news variables were related to the legislative outcome.8

8
Following the comprehensive analysis, the exact models were fit above replacing the categ-
orical news variables with continuous variables. This did not significantly change the results; the
only major change was that high-impact news coverage did not have a significant, linear
relationship with the risk of a second consideration.
LEGISLATION BY AGENDA-SETTING 707

TABLE 7
Modeling Competing Risks of Legislation

Considerations Bans

Professionalism 0.29 (0.81)


Environmentalism 2.97 (1.08) 4.50 (2.93)
Categorical health news 0.43 (0.46)
Categorical high impact health news 0.77 (0.29)
Categorical all news 0.74 (0.52)
Partisanship 1.23 (1.08)
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

N 250 250
Wald 31 on 5 df 5 on 2 df
p ¼ .00 p ¼ .09
R2 0.09 0.02
Max 0.69 Max 0.18

Note. Robust standard errors in parentheses.


y
p < .10.  p < .05.  p < .01.  p < .001.

Next, we turn from considering distinct relationships dependent on


sequence to distinguishing distinct relationships based on outcome. Namely,
we consider that media coverage was a contributor to one of two possible,
competing events: that a legislature considered legislation banning products
made from BPA, and a legislature actually enacted such legislation. The
data are presented in Table 7. Again, high-impact news coverage was signifi-
cantly related to the risk that a legislature would consider legislation, but
there was no relationship between any news variable and the chance that
a legislature would actually adopt a ban on products made from BPA.
This is not to suggest that media coverage did not matter for state legis-
latures actually adopting bans, but it suggests that media coverage mattered
indirectly. Returning to Table 2, fully eight of the nine legislative prohibi-
tions on products made from BPA in the data set were adopted after the
state had previously considered such legislation. This suggests that even if
newspaper coverage was not directly related to the adoption of legislation,
it influenced the legislative process partially by initiating and certainly by
sustaining multiyear legislative processes. Although not all of these multi-
year processes culminated in the adoption of a legislative ban, there was
only one bill adopted without that state previously having considered such
legislation. In short, news coverage played an important role creating a
necessary, but not sufficient, condition for the adoption of legislation ban-
ning products made from BPA.
These are important findings for several reasons. First, they should be
interpreted as supporting Baumgartner and Jones’s theory of policy change
by punctuated equilibrium following changing policy images. Given the fact
708 KISS

that the only news variable to show an effect on the legislative process was
one that only counted news coverage of a particular subject matter reflects
the central tenet of their theory that quantity and quality are central compo-
nents of a policy image. This also suggests that Mazur’s hypothesis that raw
levels of media coverage—without regard for quality—might be sufficient to
evoke primal concerns about technology is not supported, at least in regard
for how media coverage is related to policy developments. In that arena,
quality of media coverage is as important as quantity of coverage. Going
further, these data also suggest that not all news stories are equal; the pres-
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

ence of high-impact critical health stories seems to matter more than the
presence of routine, shorter critical health stories. The literature on media
effects does not commonly account for the impact of news story length,
but this suggests that it should become more common. Electronic databases
usually contain information on word length and on position in the
newspaper; these could be integrated in future analyses of the impact of
newspaper coverage on agenda and policy change.
However, one dynamic of this issue, somewhat beyond the scope of the
event history analysis just presented, does not actually fit with predicted out-
comes of Baumgartner and Jones’s theory. In particular, policy change via
punctuated equilibrium should exhibit short and rapid bursts of policy
change followed by a return to stasis. The case of BPA regulation certainly
exhibits the short and rapid bursts of policy change but does not really fol-
low the return to stasis as posited. Rather, because the initial burst of inter-
est in 2008 has lingered, interest has declined, to be sure, but it would not be
fair to characterize the issue as having returned to stasis. Instead, the FDA
acceded to an industry request in 2012 to ban BPA from polycarbonate
baby bottles, while continuing to reiterate that it remains safe for human
exposure. The issue also periodically attracts significant media attention.
For example, in the fall of 2012, a new cross-sectional survey of American
children found a correlation between children’s obesity and blood levels
of BPA (Zhao et al., 2012).
Second, this case presents serious challenges for regulators in the tricky
field of the regulation of hazardous substances. Our societies do not handle
information about potentially hazardous substances very well; ‘‘chemophobia’’
is a very real problem. The distribution of topics in Table 4 shows that news-
papers in the United States tend to concentrate on straightforward reporting
of the regulatory process and allegations about potential adverse effects by
scientists and environmental groups. In this regard, newspaper coverage prob-
ably reflects the vox populi. In some ways this is laudable, but there is a cost
attached with this. Humans do not perceive risks terribly well, and excessive
media coverage can hamper more sober assessment of the risks, costs, and ben-
efits of any given technology. Certainly in the case of BPA it appears that media
LEGISLATION BY AGENDA-SETTING 709

coverage drove state legislators in many states to act in the absence of really
credible scientific evidence that a risk to human health existed. Media coverage
of potentially hazardous substances can seriously complicate the process of risk
assessment and management.

CONCLUSION

A review of the scientific evidence from industry, public researcher, and


regulatory bodies revealed that there simply was no scientific consensus
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

about the existence of threats to human health at current levels of exposure


to BPA. Instead, informed by findings from the mass media effects litera-
ture, the policy agenda-setting literature and previous research that empha-
sized how the increased salience of technological innovation can provoke
intrinsic fears, this article used event history analysis to model the diffusion
of both legislative considerations and legislative bans on products made
with BPA across the U.S. states. After coding newspaper coverage about
BPA according to topic, tone, and story length, it found that increased levels
of high-impact newspaper coverage about potential negative health hazards
was weakly related to the chance that a state would introduce legislation for
a first time but strongly and significantly related to the chance that a state
would introduce legislation banning products made with BPA a second
time. This is partial evidence that newspaper coverage played a role in initi-
ating multiyear legislative processes and strong evidence that it played a role
sustaining those processes. Although newspaper coverage was not directly
related to the chance that a state would adopt a ban, the fact that eight
out of nine legislative bans in the data set only occurred after legislation had
been previously considered suggests that the media played an important,
indirect role in creating conditions necessary, if not sufficient, for the adop-
tion of legislation. In sum, there is evidence that agenda-setting effects were
shaping the varied regulatory outcomes of BPA in the U.S. states between
2005 and 2011. Although this is mostly concurrent with Baumgartner and
Jones’s theory of policy change via punctuated equilibrium, the failure of
the BPA issue to quickly stabilize is not predicted by that theory. Last, these
results should be taken as evidence of the important role that media cover-
age and practices plays in the regulation of toxic substances, particularly in
the absence of scientific consensus.

ACKNOWLEDGMENTS

I gratefully acknowledge the support of Canada’s Social Sciences and


Humanities Research Council in the form of a postdoctoral fellowship
710 KISS

and Wilfrid Laurier University in the form of a Short-Term Research Grant.


I also acknowledge the invaluable support of Peter Buckles and Danielle
Crawford for providing research assistance. Last, I am very grateful to three
anonymous reviewers for helpful comments.

REFERENCES
Baumgartner, F. R., & Jones, B. D. (2009). Agendas and instability in American politics.
Chicago, IL: University of Chicago Press.
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Behr, R., & Iyengar, S. (1985). Television news, real-world cues, and changes in the public
agenda. Public Opinion Quarterly, 49, 38–57.
Berry, F. (1990). State lottery adoptions as policy innovations: An event history analysis. The
American Political Science Review, 84, 395–415.
Box-Steffensmeier, J. M., & Jones, B. S. (2004). Event history modeling. Cambridge, UK:
Cambridge University Press.
Box-Steffensmeier, J., & Zorn, C. (2002). Duration models for repeated events. The Journal of
Politics, 64, 1069–1094.
Brace, P., Hall, M., & Langer, L. (1998). Judicial choices and the politics of abortion: Institu-
tions, context, and the autonomy of courts. Albany Law Review, 62, 1265–1303.
Brewer, P. R., & Ley, B. L. (2011). Multiple exposures: Scientific controversy, the media, and
public responses to Bisphenol A. Science Communication, 33, 76–97.
Bybee, C., & Comadena, M. (1984). Information sources and state legislators: Decision-making
and dependency. Journal of Broadcasting & Electronic Media, 28, 333–340.
Chamberlain, R., & Haider-Markel, D. (2005). ‘‘Lien on me’’: State policy innovation in
response to paper terrorism. Political Research Quarterly, 58, 449–460.
Cobb, R. W., & Elder, C. D. (1971). Participation in American politics: The dynamics of
agenda-building. Baltimore, MD: Johns Hopkins University Press.
Colborn, T., Dumanoski, D., & Myers, J. P. (1997). Our stolen future: Are we threatening our
fertility, intelligence, and survival?: A scientific detective story: With a new epilogue by the
authors. New York, NY: Plume.
Cook, F., Tyler, T., Goetz, E., & Gordon, M. (1983). Media and agenda setting: Effects on
the public, interest group leaders, policy makers, and policy. Public Opinion Quarterly, 47,
16–35.
Eldredge, N., & Gould, S. (1972). Punctuated equilibria: An alternative to phyletic gradualism.
In T. Schopf (Ed.), Models in paleobiology (p. 115). San Francisco, CA: Freeman, Cooper.
Erbring, L., Goldenberg, E., & Miller, A. (1980). Front-page news and real-world cues: A new
look at agenda-setting by the media. American Journal of Political Science, 24, 16–49.
European Food Safety Authority. (2010). Scientific Opinion on Bisphenol A: evaluation of a
study investigating its neurodevelopmental toxicity, review of recent scientific literature on
its toxicity and advice on the Danish risk assessment of Bisphenol A. EFSA Journal, 8(9).
Haider-Markel, D. (2001, March). Policy diffusion as a geographical expansion of the scope of
political conflict: Same-sex marriage bans in the 1990s. State Politics & Policy Quarterly, 1,
5–26.
Hays, S. (1997). The role of agenda setting in policy innovation. American Politics Research, 25,
497.
Herbst, S. (2002). How state-level policy managers ‘‘read’’ public opinion. In J. Manza, F. L.
Cook & B. Page (Eds.), Navigating public opinion: Polls, policy, and the future of American
democracy (pp. 171–183). New York, NY: Oxford University Press.
LEGISLATION BY AGENDA-SETTING 711

Jones, B., & Branton, R. (2005). Beyond logit and probit: Cox duration models of single, repeat-
ing, and competing events for state policy adoption. State Politics & Policy Quarterly, 5, 420.
Kousser, T., & Cain, B. E. (2004). Adapting to term limits in California (Tech. Rep.).
Washington, DC.
Kral, A. (2003). Print media impact on state legislative policy agendas (Unpublished doctoral
dissertation). Louisiana State University, Baton Rouge.
Lang, I., Galloway, T., Scarlett, A., & Henley, W. (2008). Association of urinary bisphenol A
concentration with medical disorders and laboratory abnormalities in adults. Journal of the
American Medical Association, 301, 720.
Layton, L. (2008, January). Study links chemical BPA to health problems; findings reignite
debate over safety; scientists say more research is needed. The Washington Post, p. A03.
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

Mazur, A. (1981). Media coverage and public opinion on scientific controversies. Journal of
Communication, 31, 106–115.
Mazur, A., & Welch, E. (1999). The geography of American environmentalism. Environmental
Science & Policy, 2, 389–396.
McCombs, M., & Shaw, D. (1972). The agenda-setting function of mass media. Public Opinion
Quarterly, 36, 176.
Mead, T. (1994). The daily newspaper as political agenda setter: The Charlotte Observer and
metropolitan reform. State & Local Government Review.
Mills, M. (2011). Introducing survival and event history analysis. Thousand Oaks, CA: Sage.
Mintrom, M. (1997). Policy entrepreneurs and the diffusion of innovation. American Journal of
Political Science, 41, 738–770.
Mittelstaedt, M. (2007). ‘‘Inherently toxic’’ chemical faces its future. The Globe and Mail, p. A10.
Mittelstaedt, M. (2010, August). U.S. dollar bills contaminated, study finds. The Globe and
Mail. Available from http://www.theglobeandmail.com
Mooney, C. Z. (2001, March). Modeling regional effects on state policy diffusion. Political
Research Quarterly, 54, 103–124.
Nagel, S. C., vom Saal, F. S., Thayer, K. A., Dhar, M. G., Boechler, M., & Welshons, W. V.
(1997). Relative binding affinity-serum modified access (RBA-SMA) assay predicts the
relative in vivo bioactivity of the xenoestrogens bisphenol A and octylphenol. Environmental
Health Perspectives, 105(1), 70–76.
National Council of State Legislatures. (n.d.). Environmental health legislation database.
Retrieved from http://www.ncsl.org/issues-research/env-res/environmental-health-legislation-
databases-guide.aspx
Neimark, J. (2008, April). The dirty truth about plastic. Retrieved from http://discovermagazi-
ne.com/2008/may/18-the-dirty-truth-about-plastic
Nisbet, M., & Lewenstein, B. (2002, June). Biotechnology and the American media - The policy
process and the elite press, 1970 to 1999. Science Communication, 23, 359–391.
Peter, J. (2003, December). Country characteristics as contingent conditions of agenda setting:
The moderating influence of polarized elite opinion. Communication Research, 30, 683–712.
Pralle, S. (2006). The ‘‘mouse that roared’’: Agenda setting in Canadian pesticides politics.
Policy Studies Journal, 34.
Prentice, R., Williams, B., & Peterson, A. (1981). On the regression analysis of multivariate
failure time data. Biometrika, 68, 373–379.
Pritchard, D. (1986). Homicide and bargained justice: The agenda-setting effect of crime news
on prosecutors. Public Opinion Quarterly, 50, 143–159.
Protess, D. L., Cook, F. L., Curtin, T. R., Gordon, M. T., Lee, D. R., McCombs, M. E., &
Miller, P. (1987). The impact of investigative reporting on public opinion and policymaking
targeting toxic waste. Public Opinion Quarterly, 51, 166.
Ropeik, D. (2010). How risky is it, really?. New York, NY: McGraw-Hill Professional.
712 KISS

Scheberle, D. (1994). Radon and asbestos: A study of agenda setting and causal stories. Policy
Studies Journal, 22, 74–86.
Soroka, S. N. (2002). Agenda-setting dynamics in Canada. Vancouver, Canada: University of
British Columbia Press.
Squire, P. (2007). Measuring state legislative professionalism: The squire index revisited. State
Politics & Policy Quarterly, 7, 211.
Stump, D. G., Beck, M. J., Radovsky, A., Garman, R. H., Freshwater, L. L., Sheets, L. P., . . .
Hentges, S. G. (2010, May). Developmental neurotoxicity study of dietary bisphenol A in
Sprague-Dawley rats. Toxicological Sciences: An Official Journal of the Society of Toxicology,
115, 167–182.
Tan, Y., & Weaver, D. (2009). Local media, public opinion, and state legislative policies. The
Downloaded by [University of Ulster Library] at 06:13 06 March 2015

International Journal of Press=Politics, 14, 454.


Tyl, R., Myers, C., Marr, M., & Sloan, C. (2008). Two-generation reproductive toxicity study of
dietary bisphenol A (BPA) in CD-1 (Swiss) mice. Toxicological Sciences, 104, 362–384.
Tyl, R. W., Myers, C. B., Marr, M. C., Thomas, B. F., Keimowitz, A. R., Brine, D. R., . . .
Waechter, J. M. (2002, June). Three-generation reproductive toxicity study of dietary bisphe-
nol A in CD Sprague-Dawley rats. Toxicological Sciences: An Official Journal of the Society
of Toxicology, 68, 121–146.
University of Missouri, Columbia. Division of Arts, & Sciences. (2005, April). Evidence mounts
against chemical used widely in everyday plastic products. Retrieved from http://biology.missouri.
edu/news/evidence-mounts-against-chemical-used-widely-in-everyday-plastic-products/
U.S. Census Bureau. (2011). Composition of state legislatures by political party affiliation.
Washington, DC: Author.
Wildavsky, A. B. (1964). The politics of the budgetary process. Boston, MA: Little Brown & Co.
Zhao, H.-Y., Bi, Y.-F., Ma, L.-Y., Zhao, L., Wang, T.-G., Zhang, L.-Z., . . . Liu, J.-M. (2012,
September). The effects of bisphenol A (BPA) exposure on fat mass and serum leptin concen-
trations have no impact on bone mineral densities in non-obese premenopausal women.
Clinical Biochemistry, 45, 1602–1606.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy