Politics of the environment

MISSING IMAGE

Material Information

Title:
Politics of the environment an analysis of state regulations and special interest behavior
Physical Description:
vi, 114 leaves : ; 29 cm.
Language:
English
Creator:
Davis, Mary Elizabeth
Publication Date:

Subjects

Subjects / Keywords:
Economics thesis, Ph.D   ( lcsh )
Dissertations, Academic -- Economics -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 2003.
Bibliography:
Includes bibliographical references.
Statement of Responsibility:
by Mary Elizabeth Davis.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 030441457
oclc - 83024631
System ID:
AA00013580:00001


This item is only available as the following downloads:


Full Text









THE POLITICS OF THE ENVIRONMENT: AN ANALYSIS OF STATE
REGULATIONS AND SPECIAL INTEREST BEHAVIOR












By

MARY ELIZABETH DAVIS







A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2003













ACKNOWLEDGEMENTS

This dissertation would not have been possible without the help of my committee, Jon

Hamilton, David Denslow, Donna Lee, and my chair, Larry Kenny. I am especially

grateful to Dr. Kenny for his encouragement and direction, and for the many hours spent

providing guidance on the work in progress. Special thanks go to my family and friends

as well, for their unconditional love and support during my graduate studies.














TABLE OF CONTENTS

page

ACKNOWLEDGEMENTS ............................................................. ii

A B ST R A C T ............................................... .................................... v

CHAPTER

1 INTRODUCTION ...................................................................... 1

State Environmental Regulations................................................... 1
Special Interest Participation...................................................... 3
Plan of the Dissertation ............................................................ 4

2 THE POLITICS OF STATE ENVIRONMENTAL STANDARDS.............. 5

Introduction......................... ................................................. 5
Literature Review ...................................................................... 5
Theoretical Model................................. ................. 9
Conclusions .............................................. .... ................... 11

3 THE POLITICS OF STATE WATER QUALITY STANDARDS............. 12

Introduction..................................................... ..................... 12
Water Quality Standards........................................................... 13
Database of Toxic Metals.......................................................... 16
Empirical Design............................................... ...................... 18
Results......................... ........................................................ 26
Conclusions....... ................ ...................................... 35
N otes............... .................. ............ .................. 36

4 THE POLITICS OF STATE AIR QUALITY STANDARDS.................. 38

Introduction........ ............. ............ ...................... 38
National Ambient Air Quality Standards........................................ 39
Empirical Design............................... ...... .....................44
Results................. ................................ ................. 54
Conclusions ........................... ..................................... 60
N otes............................................................... .................. 6 3



iii






iv



5 THE POLITICS OF SPECIAL INTEREST VOTER SCORECARDS

Introduction ........................................................ .................. 64
Theory and Implications........................................................... 67
League of Conservation Voters.................................................... 72
Christian Coalition........ .... .............. ...... ............ 82
R esults............................................................... ................. 96
Conclusions......................................................... ................. 103

6 SUMMARY AND CONCLUSIONS............................................... 106
State Environmental Regulations................................................. 106
Special Interest Participation...................................................... 109


REFERENCE LIST.......... ......................... ................... 11

BIOGRAPHICAL SKETCH............................................................ 114













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

THE POLITICS OF THE ENVIRONMENT: AN ANALYSIS OF STATE
REGULATIONS AND SPECIAL INTEREST BEHAVIOR

By

Mary Elizabeth Davis

August 2003

Chair: Lawrence W. Kenny
Major Department: Economics

Why do some states pass strict environmental regulations, while others are content

with the baseline standards required by the federal government? This dissertation seeks

to answer that question by looking at the costs and benefits to a state from developing

strong environmental standards. This work outlines the state environmental choice as a

tradeoff between the desires of consumers (who want better environmental quality) and

of producers (who want less restrictive environmental standards). A rational state

legislator maximizes her chances of being re-elected by balancing these two competing

forces when setting environmental policy. This dissertation provides empirical evidence

that the differences in state standards for sulfur dioxide and toxic metals are a function of

per capital income levels, special interest participation, strength of polluting industries,

and natural differences in climate and location that inflate the cost of compliance.

Also explored in Chapter 5 of the dissertation is the degree to which special interest

voter scorecards are representative of actual candidate positions on certain issues. An





vi

analysis of a series of voter guides from the League of Conservation Voters and the

Christian Coalition provides evidence that these organizations slant the reported positions

of candidates by including non-issue partisan votes in the calculation of their scorecards.

Including these peripheral votes allows the groups to increase the scores of their favored

party members and decrease the scores of the opposing party.













CHAPTER 1
INTRODUCTION

State Environmental Regulations

The legislation of the 1970's, which includes the Clean Water Act and the Clean Air

Act, provided the foundation for modem day environmental politics. Before this time,

the protection of environmental resources had been primarily the responsibility of states.

However, public concern over the lack of state responsiveness to environmental

problems, along with a growing awareness of the externalities between states, led to

federal intervention in this area.

In order to ensure citizen access to basic health and environmental protections, the

legislation minimized state sovereignty over environmental policy. In particular, the

Clean Water Act seeks to make all surface waters within the United States "fishable and

swimmable." Likewise, the goal of the Clean Air Act is to limit the amount of airborne

pollutants everywhere in the United States.

Although these laws are designed to provide equal protection to all Americans, they

also recognize the important role of states in this process. Pollution control problems

often require a detailed understanding of local industries, geography, and housing

patterns. For this reason, states are delegated some authority under the programs to

personalize their environmental agenda. For example, the Clean Water Act provides

flexibility to states in setting standards for toxic metals, while reserving the right to reject

regulations that do not comply with the national goals. The Clean Air Act allows states








to have stronger pollution controls but prohibits them from having weaker standards than

those required nationwide.

The flexibility that these two laws provide to states poses a number of interesting

empirical questions that I seek to answer in the following chapters. Do states select the

stringent environmental standards favored by consumers or the more relaxed standards

preferred by industrial polluters? Do states take advantage of favorable climate

conditions by setting stricter environmental standards? Do relatively poorer states react

differently to changes in income levels when setting an environmental agenda?

The likelihood a state will adopt a strong pollution control program is analyzed within

Peltzman's (1976) theoretical model of group conflict. In the context of this work, a

rational legislator will maximize political support by weighing the gain in consumer votes

(from better environmental quality) against the loss in industry votes (from increased

restrictions on firm activities). Also taken into account are the natural peculiarities across

states that affect the cost of compliance with environmental regulations, such as

geographic and climate conditions.

The existence of an inverted-U shaped curve linking state environmental standards

and income is also investigated. This application of the inverted-U hypothesis asserts

that the evolution of environmental regulations is essentially different at low and high

levels of income. More specifically, income growth at lower levels will precipitate a

weakening of environmental standards. However, beyond a certain point further

increases in income will be associated with a strengthening of these same standards.

The results presented in this dissertation provide a valuable contribution to the

literature addressing the inverted-U relationship between income and environmental








quality. Although the results of previous studies could have simply reflected lax

enforcement of existing regulations, evidence provided in this work implies that low-

income states actually choose weaker environmental standards.

Special Interest Participation

The final issue that will be addressed in this dissertation (Chapter 5) is the degree to

which special interest scorecard rankings provide reliable estimates of consumer demand

for social policies. These special interest voter guides rank a politician based on her

behavior on a selection of "issue-specific" roll call votes. Although there has been no

rigorous analysis of voter scorecards, economists often use these ratings in empirical

work to proxy for consumer or legislator preferences.

The current literature does not address either the incentives that special interest groups

have to manipulate these scores, or the extent to which this manipulation favors a certain

political party. The main contribution of this chapter is to extend the literature by

examining whether the inclusion of peripheral votes in issue-specific scorecards can be

shown to provide evidence supporting the hypothesis of interest group bias. The

incentive to misreport candidate behavior in favor of one party is examined within a

standard voting model that addresses the decision by a group member on whether or not

to vote in the current election.

This chapter examines two voting guides for partisan bias, specifically a series of

congressional scorecards (1997-2001) for the League of Conservation Voters

(environmental scorecard) and the Christian Coalition (religious scorecard). Once the

voting guides are revised to exclude non-issue votes and to correct for methodological








inconsistencies, legislators fare better on opposing indices and worse on supporting

indices.

The results from this chapter are used to construct a revised unbiased measure of the

environmental liberalism of elected officials. This variable, composed of a four-year

state average of the revised LCV scores, is used as an explanatory variable in the earlier

two chapters. Although this measure provides only limited success in explaining the

variation in state environmental policies, it nonetheless outperformed all other alternative

measures that were constructed for this purpose.

Plan of the Dissertation

A common theoretical model and literature review is outlined in Chapter 2 that will

provide the foundation for the empirical investigations of Chapters 3 and 4. More

specifically, Chapter 3 tests the Peltzman model with a newly assembled dataset of state

toxic metal water quality standards, and Chapter 4 follows with a similar examination of

sulfur dioxide air quality regulations. Chapter 5 develops an unbiased estimate of

environmental liberalism of state elected officials that is used as an explanatory variable

in the earlier chapters. This section is self-contained with its own literature review and

theoretical component. Finally, Chapter 6 presents concluding remarks and summarizes

the contributions and results of the dissertation.











' Other variables included a ten-year average of party identification of elected officials (state and federal),
ADA scores, and original LCV scores.














CHAPTER 2
THE POLITICS OF STATE ENVIRONMENTAL STANDARDS

Introduction

This chapter provides the foundation for the analysis of state air and water quality

standards. A review of the current literature on state environmental policy behavior is

outlined, with specific references to the contribution of this work. Also provided in this

chapter is the theoretical basis for the investigation, an application of Peltzman's (1976)

model of legislative decision-making.

Literature Review

State Environmental Policies

The variance across states in environmental quality is intimately linked to the policy

measures states employ to control these externalities. However, a study of the

determinants of strong pollution control programs has been sidelined by the easier task of

directly analyzing environmental quality outcomes across states. An understanding of the

divergence among states in their willingness to adopt restrictive environmental policies

provides an essential piece to the environmental puzzle, and neglecting this aspect will

most certainly leave unanswered questions regarding state-level differences in

environmental quality. Only two studies in the literature have provided a detailed focus

on the environmental policy differences across states, and a discussion of their

contributions is useful.








Lowry (1992) examines the divergence in state environmental programs by attempting

to explain a number of general measures of state regulatory efforts. For air pollution

control programs, the dependent variables include a ranking of State Implementation

Plans (SIP's) as strong or weak, state air expenditures per capital, a ranking of

enforcement programs from one to ten based on monitoring and inspections data, and

finally the extent to which the state performs acid rain research. He compiles similar

aggregate rankings of water quality programs to use as the dependent variable in those

analyses.

However, none of these variables represent the actual standards in place within the

states, and for that reason do not provide a clear and unbiased view of the differences in

state responsiveness to environmental concerns. For example, the SIP's represent only a

plan of action, while the level of state expenditures does not account for either how the

money is spent or the degree to which this spending is efficient. This dissertation

specifically addresses the problem by using state environmental standards as the

dependent variable to be explained in the model.

Lowry has limited success in characterizing the variation in environmental policies

across states with his chosen model.2 However, he finds general support for the theory

that states respond to pollution problems by setting more restrictive policies. More

specifically, as the importance of polluting industries increases within a state, so does the

likelihood that the state will respond with stricter environmental controls. These results

are based on a specification that aggregates the industrial strength variable to include


2 Independent variables in Lowry's model for air and water programs include: percent of state population
living in excess SO2 areas, manufacturing and utility sectors, personal income per capital, voter turnout and
party competition, federal subsidies, EPA sanctions, and percent of state waters fishable or swimmable.








both the manufacturing and utility sectors as a percent of gross state product. However,

aggregating the variable in this way ignores potential differences in pollution intensity

and market structure across polluting industries.

For example, public utilities represent well over half of the sulfur dioxide emissions in

the U.S., with private manufacturing firms coming in a distant second. This difference is

compounded by the fact that public utilities represent a regulated natural monopoly in

most markets, and do not face the same competitive forces as private industries. These

two differences could potentially impact the development of state policy, and should be

accounted for in the model. This paper addresses this concern by separating the variable

for industrial strength into private manufacturing firms and public utilities.

Ringquist (1993) provides a second analysis of state environmental policy

responsiveness. In this work, he summarizes various economic and political models that

are designed to account for cross-state variation in environmental policy outputs, and

develops a model that seeks to integrate these approaches. The dependent variable he

uses to describe the variance in state policies for air and water pollution control programs

ranks states from weakest (one) to strongest (ten or thirteen, depending on the program).3

Among other characteristics, this ranking takes into consideration state differences in

enforcement mechanisms, EPA sanctions, and environmental budget expenditures.

Ringquist finds similar evidence to suggest that states respond to an increase in the

prevalence of polluting industries by setting more restrictive policies, although his

specification excludes public utilities entirely. Other empirical results from the model

include the observation that state dependence on fossil fuels decreases the level of state


3 Ringquist's ranking is taken from the State of States: 1987, published by the Fund for Renewable Energy
and the Environment (FREE).








responsiveness to environmental concerns, while an increase in the level of

environmental activism (special interest participation in environmental organizations)

increases the likelihood states will develop strict environmental programs.

This dissertation extends the analyses of Ringquist and Lowry by utilizing the actual

standards as the dependent variable, instead of the expenditures and measures of the

effectiveness of pollution control programs examined in their works. Since both the

causes and effects vary with the pollutant studied, it is reasonable to assume that a

ranking system ignoring these differences will be less successful on a case-by-case basis.

Moreover, aggregating the dependent variable in this way makes it harder to test the

theory, since it commingles the effects of regulations, enforcement, and initial conditions.

Environmental Kuznets Curve Application

The existing literature concerning the relationship between environmental quality and

income is extensive. The controversial inverted-U hypothesis, also known as the

Environmental Kuznet's Curve (EKC), posits that environmental degradation initially

increases with economic development, but beyond a certain threshold level, increases in

income are associated with better environmental quality.

For example, increases in income for a poor economy results primarily from

industrialization, which leads to higher pollution levels. However, as an economy

expands and grows richer, a structural change within that economy towards the service

sector and away from heavy industry occurs. Changing voter preferences for

environmental quality enhances this effect the income elasticity of demand for

environmental quality is relatively small at low income levels, but becomes larger as

income increases.








Previous research concerning the identification of an inverted-U shaped curve linking

income and environmental quality has produced conflicting results. [See, for example,

Selden and Song (1994), Holtz-Eakin and Selden (1995), Kaufmann et al. (1998), Torras

and Boyce (1998), List and Gallet (1999), and Dasgupta et al. (2002)]. The majority of

these works have focused on estimating this relationship by means of a cross-country

reduced form analysis. The most influential of these, a work by Grossman and Krueger

(1995), examines this relationship with panel data from the Global Environmental

Monitoring System (GEMS). Their research provides evidence to support the inverted-U

hypothesis, and they suggest that the critical threshold level occurs at a GDP per capital of

less than $8,000 (in 1990 dollars) for the majority of pollutants Greece and Portugal are

among countries at this income level.

Grossman and Krueger also emphasize that "a review of the available evidence on

instances of pollution abatement suggests that the strongest link between income and

pollution in fact is via an induced policy response" (371-372). However, surprisingly

little research has been done to define this relationship. The contribution of this

dissertation is to directly examine the impact of income on environmental regulations. If

the results in the EKC literature reflect differences in environmental standards, then these

standards would follow the same income path as environmental quality standards

initially deteriorate as income grows, followed by a strengthening of these same

standards at higher income levels.

Theoretical Model

In a 1976 article, Peltzman outlined a theoretical model in which he proposes that a

legislator will choose her policy stance in order to maximize electoral support. In the








context of this paper, a legislator will receive the greatest number of votes by taking into

account both consumer and industrial concerns when setting environmental standards.

Algebraically, a legislator will maximize the net majority voting for her (M) by

optimizing the following equation:

M=W*F-L*A

where W = # of winners from industry due to looser environmental standards
= # of potential voters in industry and stockholders

L = # of losers from looser environmental standards
= # of consumers with no industrial ties

F = net probability that winner votes for legislator
= probability of favorable vote -probability of unfavorable vote

A = net probability that loser votes against legislator

For example, weak environmental standards will gain support from industry at the

expense of consumer votes. Since a legislator wants to maximize the votes she receives,

the first order conditions specify that this be achieved where the marginal gain of

industrial votes is just offset by the marginal loss of consumer votes. In other words,

support from industry and opposition from consumers will be equated at the margin.

This paper seeks to test the Peltzman theory by developing an empirical model that

accounts for these conflicting interests. More specifically, greater levels of consumer

opposition should result in more restrictive environmental policies, while the opposite

effect will occur when large industrial interests are present in a state. A legislator also

would be expected to take into account natural variations in the cost of compliance when

setting an environmental agenda.








Conclusions

The following chapters build upon the existing literature of state environmental policy

responsiveness to more effectively characterize the determinants of strong state pollution

control programs. More specifically, Chapter 3 tests a newly developed database of state

water quality standards for toxic metals, and Chapter 4 follows with an analysis of

ambient air quality standards for sulfur dioxide. The theoretical foundation for both of

these empirical investigations is an application of the Peltzman model of legislative

decision-making.

The following chapters provide two major contributions to the economics literature.

The first is to develop a set of dependent variables that represent the actual standards in

place, and not the commonly used ranking systems of state policies. This specification of

the dependent variable provides additional insight into the determination of individual

policies, and to the broader implications that these policy results may have for similar

environmental regulations. The other significant contribution of this work is to extend

the inverted-U hypothesis to define the relationship between income and environmental

standards. More specifically, I propose that standards initially deteriorate as income

grows, followed by a strengthening of these same standards at higher income levels.













CHAPTER 3
THE POLITICS OF STATE WATER QUALITY STANDARDS

Introduction

Anthropogenic contamination of surface waters poses significant health risks to

Americans, as 160 million receive their drinking water from this source daily. In 1972,

Congress addressed the issue with the Clean Water Act (CWA), which outlined a clear

objective "to restore and maintain the chemical, physical, and biological integrity of the

nation's waters." It defined the "fishable and swimmable" goals of the Act, which

"provide for the protection and propagation of fish, shellfish, and wildlife, and recreation

in and on the water" (Environmental Protection Agency, National Water Quality

Inventory 3). Under this legislation, states are granted the autonomy to design and

implement their own system of water quality standards, offering states a true choice

between environmental quality and growth.

Since the Clean Water Act is one of the largest federal programs ever to delegate

primary responsibility to states, a better understanding of how states have set their

standards in this large program is especially important. This chapter performs an

empirical analysis of a newly compiled dataset to both identify key variables in the

decision-making process and test the hypothesis that states trade off idiosyncratic benefits

and costs in setting environmental policy. This is the first work to compile the data

necessary for this task, and consequently the first to attempt an explanation of the

variation among states for this aspect of environmental policy.








The theoretical framework for this chapter is an adaptation of Peltzman's model on

legislator vote-maximization, which guides the selection of variables used in the

empirical analysis. Specifically, this chapter identifies the agriculture industry as a

motivator for weak state water quality standards, while heavy industry does not play an

integral role in the process. Certain state-specific environmental characteristics also

prove significant, as they reflect the added costs of implementing stricter standards.

The existence of an inverted-U shaped curve linking state environmental standards

and income is also investigated. This application of the inverted-U hypothesis asserts

that the evolution of environmental regulations is essentially different at low and high

levels of income. More specifically, income growth at lower levels will precipitate a

weakening of environmental standards. However, beyond a certain point further

increases in income will be associated with a strengthening of these same standards.

Water Quality Standards

Water quality standards are laws or regulations imposed by states to accomplish the

goals of the CWA, which are 1) maintain and restore the integrity of the Nation's waters,

2) protect aquatic life and provide recreation in and on the water, 3) prohibit harmful

amounts of toxic pollutants from entering the waters, and 4) eliminate the discharge of

pollutants to navigable waters.

These water quality standards are to apply to all surface water within the state and

generally consist of three elements. The first is the designated beneficial uses of water

bodies within the state. Typical beneficial uses include public water supply, propagation

of fish and wildlife, agricultural and industrial uses, and recreation. The second element

is the antidegradation policy, which ensures that waters that are already meeting the








minimum requirements will not be degraded below their current levels. The third

element of water quality standards is the development of a list of numeric criteria that are

necessary in order to protect the beneficial uses that have been designated within the

state. It is this third category that will be examined here because it allows for a

quantitative comparison of standards across states.

National Toxics Rule

The CWA requires the Environmental Protection Agency to periodically develop and

publish revised numeric criteria for water quality. The National Toxics Rule establishes

the water quality criteria that the EPA believes to reflect the most current scientific

knowledge about the effects of toxic pollutants. It also establishes the maximum

acceptable concentration levels that will generally be safe for human and aquatic life

protection.

These recommendations do not reflect local considerations, such as natural variations

in climate conditions and location or the economic impacts of meeting the proposed

standards. For this reason, they are intended only to provide guidance for states in

adopting their own set of standards.

The current National Toxics Rules for the metals studied in this work are listed in

Table 3-1, along with the number of states that have adopted these recommended

standards. Also provided in the table are summary statistics for the state metal standards,

including the mean and standard deviations, and high and low values. With the exception

of acute chromium III and chronic zinc, the average state standard for each of the toxic

metals is above (weaker than) the EPA recommendation. However, most of the state

averages are very close to the federal guidelines.









Table 3-1: Summary of Acute Toxic Metal Standards and National Toxics Rule*
Acute Obs. Mean Standard Min. Max. National # of
Dev. Value Value Toxics states
Rule at NTR
Copper 45 17.43 4.03 6.25 30.21 13 0
Chromium 40 1724.98 135.22 918.65 1803.8 1804 0
III
Chromium 43 17.28 7.20 15 60 16 18
VI
Lead 44 83.67 11.86 36.66 122.78 82 7
Mercury 43 2.39 0.741 1.65 6.5 1.65 4
Nickel 45 1,267.34 739.94 5.02 3,822.55 470 0
Silver 38 4.56 4.29 1.18 30 4 4
Zinc 44 123.44 33.68 22.49 277.73 120 5

Table 3-2: Summary of Chronic Toxic Metal Standards and National Toxics Rule
Chronic Obs. Mean Standard Min. Max. National # of
Dev. Value Value Toxics states
Rule at NTR
Copper 48 11.72 2.69 6.25 20 9 0
Chromium III 41 176.97 52.84 86.05 210.47 86 0
Chromium VI 44 11.97 4.65 10 40 11 20
Lead 46 5.75 7.28 3 36.66 3.2 8
Nickel 46 142.54 80.34 5.02 424.95 52 0
Zinc 46 112.24 31.64 22.49 250.05 120 7
*All standards are measured in micrograms per liter

Monitoring and Assessment

Once these state water quality standards are set and approved by the EPA, they

constitute the benchmark by which a state monitors and evaluates the health of its

waters.4 States are required to submit a triennial assessment of the quality of their waters

to Congress, and a state with harsher standards will have a tougher benchmark by which

the success of its environmental progress is judged.

Furthermore, states use the water quality standards to establish point source discharge

limits under the National Pollution Discharge Elimination System (NPDES). This is

important because it sets a limit on the amount of pollution that industrial sources in the


4 See Note 1 for further discussion of EPA approval process.








region are permitted to release into a particular water system. A state with stricter water

quality standards will issue fewer pollution rights in the form of discharge permits, and

this can have critical economic implications for the industries involved.

Database of Toxic Metals

I compiled a dataset from the recently published EPA website that contains

information on state water quality standards effective as of May 30, 2000. In order to

construct the database, I had to search through a summary of each state's standards, some

of which were quite lengthy, for the relevant data on water quality. This took

considerable time, and great care was taken to accurately catalogue the set of standards

for each state.

I initially intended to include in the database regulations governing levels of fecal

coliform, ammonia, nitrogen, and phosphorus. Although important indicators of water

quality, these measures were impossible to compile because they simply were not

reported by a majority of states. Another problem associated with collecting these

specific indicators stems from the inconsistencies in how the states that reported these

standards described them.

Therefore, I decided to use the state standards in place for toxic metals as the

dependent variables for analysis. Metals accumulate in the water both naturally and from

anthropogenic sources such as mining, agriculture, and industry. Without adequate

abatement and treatment, high levels of these toxic metals are ultimately destined for the

public water supply. Some metals can also bioaccumulate5 in the fish and shellfish

consumed by humans, causing further detrimental effects. In general, high-level


Bioaccumulation: the process by which a compound is taken up by, and accumulated in the tissues of an
aquatic organism from the environment, both from water and through food.









exposure to these metals can pose serious health risks, including an increase in the

incidence of cancer in human and animal populations.

The standards for toxic metals are separated into acute6 and chronic7 measures. The

acute criteria reflect short-term exposure, while the chronic criteria represent continued

exposure to toxic levels over a period of time. The chronic criteria are stricter than their

acute counterparts because of increased exposure time, and both levels are included in the

current analysis.

Another characteristic of the database is that it is constructed for the category of

freshwater aquatic life. I decided to use aquatic life as the benchmark variable because

there did not seem to be much variation in the standards protecting human health, perhaps

due to more stringent regulation on the part of the federal government. Also, I applied

freshwater standards instead of those available for saltwater; using saltwater standards

would have severely limited the number of observations in an already small sample base

of 50 by eliminating inland states. Criteria for groundwater standards were rarely

available in the database and for that reason were excluded as well.8

Water quality standards for arsenic, cadmium, and selenium were collected but are

excluded from the current analysis. This is due to the lack of variation in these standards;

most states chose the same guideline (often the federal guideline), or the state standards

were clustered at two values. Also, most states did not report the chronic values for silver

or mercury, and for this reason they are only represented in their acute form. Thus, the


6 Acute: involving a stimulus severe enough to rapidly induce a response; a response measuring death
observed within 96 hours or less is typically considered acute.

7 Chronic: involving a stimulus that lingers or continues for a relatively long period of time; the
measurement of a chronic effect can be reduced growth, reduced reproduction, or death.

8 See Note 2 for technical discussion of database.








metals chosen for analysis in the this study copper, chromium III and VI, lead, mercury,

nickel, silver, and zinc all have substantial variation in their standards, permitting the

use of a common statistical technique.

There are missing data points where states did not report standards. The number of

observations for each toxic criterion ranges from 38 to 48, and the average number of

observations per toxic metal is 44. For example, Massachusetts is not represented in the

sample because its criteria are site-specific and no distinct statewide standard is listed for

either the chronic or acute levels. Florida and North Carolina only listed chronic level

standards, so these states are not represented in the acute regressions.

The distributions of state standards are highly non-normal, i.e., some states have much

stricter or looser standards than others and they represent outliers in the dataset. For

example, Hawaii had standards that were much tougher than any other state, while Iowa,

Missouri, Nebraska, and Louisiana, were among the weakest regarding water quality

regulation. In order to ensure that extreme states such as Hawaii did not bias the

outcome, I did specification checks by running the regression with and without the outlier

states. The regressions performed better with these observations included, so there was

no justification for deleting them. Therefore, all of the states that have published data

available are included in the sample.

Empirical Design

The Peltzman model of legislative decision-making is employed in describing the

variation among state environmental standards. In the context of this work, a state

legislator will balance the loss of industry votes due to stricter standards against the gain

in consumer votes from enhanced environmental quality. This being the case, each









state's standard-setting behavior will reflect its own political, economic, and social cost-

benefit structure. The set of independent variables that are used to test this theoretical

model are described below, and summary statistics are provided in Table 3-3. Since the

numeric standards are measured directly, lower values of the dependent variable

correspond to stricter water quality standards.

Table 3-3: Summary of Independent Variables
Mean Standard Min. Max.
Dev. Value Value
Sierra Club 0.203 0.120 0.043 0.547
LCV Score 42.39 26.00 5 98
Medinc 28,948 5,558 20,136 41,721
Industry 5.80 2.66 1.13 12.91
Agriculture 41.06 25.07 0.2 92.5
Precipitation 35.62 14.16 9.06 57.77
Temperature 52.39 8.62 40.1 77.2
Coastal 0.56 0.50 0 1-

Consumer Influence

I measure consumer influence in the empirical model by including a set of

independent variables that capture the demand for environmental quality among the state

electorate. The variables that I identify for these purposes are participation in the Sierra

Club, the level of environmentalism of elected officials (LCV scores), and median

income.

Sierra Club. The percent of the state population belonging to the Sierra Club is used

to measure the demand for environmental quality among state consumers. As the largest

grassroots conservation organization (over 700,000 members nationwide), the Sierra

Club variable should provide an adequate proxy of consumer tastes for the environment.

Following the Peltzman model, a state with a larger percentage of consumer participation








in the Sierra Club should have stricter regulatory standards at the state government level.

Therefore, the predicted sign for Sierra Club is negative.

LCV Score. The voting behavior of elected officials on environmental issues is a

second measure of consumer preferences in this model. The propensity to elect

environment-friendly officials is an important indicator of overall demand for these

goods. This variable is measured as the average rating on a series of recent League of

Conservation Voters congressional scorecards for members of the U.S. House of

Representatives (1997-2001), revised to exclude potential partisan bias by the

organization [see Chapter 5 for a detailed explanation of the revised index]. The LCV

scorecard rates congressional voting behavior on select environmental votes, where a

100% score corresponds to perfect agreement with the LCV agenda.

The expected sign for this variable is negative. The higher the adjusted state LCV

scores, the more likely consumers are concerned with environmental issues (as evidenced

by their voting behavior), and thus the stricter should be the state standards.

Environmental Kuznets Curve. The inclusion of median income levels as an

indicator of increased consumer opposition to relaxed environmental standards is less

straightforward. An extensive amount of research has provided evidence that median

income and environmental quality are better represented by a quadratic form (inverted-U

hypothesis). This theory argues that environmental quality suffers as income rises

initially, but beyond a certain threshold income level, economic growth is associated with

greater levels of environmental quality.

The empirical literature does not extend this relationship between median income and

environmental quality to the actual standards in place to achieve these results, which is a








focus of the current paper. The hypothesized signs for the coefficients on median income

and median income-squared are positive and negative, respectively, i.e. toxic metal

standards are weakened at low levels of income growth, and strengthened at high levels

of income growth.

A final consideration is whether the regression suffers from multicollinearity among

the three consumer variables. Although a significant amount of correlation does exist

between Sierra Club, liberalism, and median income (between 0.55 and 0.59),

specification checks did not provide evidence that this correlation was having a

noticeable effect on the outcome. Therefore, for simplicity, only the specifications

reporting all three of the consumer variables will be reported.

Polluter Influence

In the Peltzman model, industrial concerns that are affected by an increase in the

stringency of environmental regulations will lobby legislators against such policies. The

state officials will weigh the gain in industry votes from decreasing environmental

regulations against the loss in consumer votes that results from lower levels of

environmental quality. For that reason, states with greater interests in heavily polluting

industries are more likely to set weaker air quality regulations.

However, some research has linked higher levels of industrialization to more liberal

environmental policies [See, for example, Lowry (1992) and Ringquist (1993)]. The

hypothesis behind this assumption is that states with large polluting industries find it

necessary to develop reactionary environmental policies to control already existing

pollution problems within the state. In the context of this model, the value to consumers

of reducing pollution levels is higher in relatively dirtier states, which will boost overall








consumer opposition to relaxed environmental standards that favor industry.

Subsequently, a vote-maximizing legislator would set more restrictive environmental

policies in these states. Two different specifications for industry (specific and general)

were developed to test these competing theories.

Specification 1. For each dependent variable, I identified major polluting industries

(responsible for over 90% of the toxic releases to land and water from 1987 to 1993)

from the list of sources provided by the EPA. I then mapped these polluting industries

into categories that could be used as the independent variables with an index of Census

data this mapping is provided in Table 3-4. The polluters of each toxic metal are added

together to obtain a single composite measure of the percentage of the labor force in the

major polluting industries specific to each toxic criterion.

Also included in this specification are primary mining activities, as this can result in

elevated pollution levels as well. To test the possibility that the mining industry is a

major player in the development of state standards, a dummy variable specific to each

metal is included to identify whether a state is a producer of that metal (from primary

mining activities, secondary extraction as a byproduct, or through recycling methods).

The hypothesized sign on these variables is positive if industrial concerns are weighed

more heavily than consumer concerns a higher level of each activity within a state

would increase the political cost of abatement, leading to weaker standards. On the other

hand, a negative sign on the industry and mining coefficients would imply a more

responsive role for the state government with regard to increasing pollution problems.









Table 3-4: Matching of Independent Variables with Major Polluting Industries
Toxic Metal Major Polluting Industries* Independent Variables**
Copper Primary copper smelting Nonferrous foundries
Other nonferrous smelting "
Plastic materials Plastic materials
Steel blast furnaces Blast furnaces
Chromium III Industrial organic Industrial organic chemicals
and VI Steel blast furnaces Blast furnaces
Electrometallurgy Electrometallurgy
Lead Steel blast furnaces Blast furnaces
Lead smelting, refining Electrometallurgy
Iron foundries "
Copper smelting
China plumbing fixtures Vitreous plumbing fixtures
Mercury Chemical and allied products Chemical and allied products
Paper mills Paper mills
Nickel Primary nonferrous metals Primary nonferrous metals
Steel blast furnaces Blast furnaces
Ind. organic and inorganic chemicals Industrial inorganic/organic
Petroleum refining chemicals
Petroleum refining
Silver Metal plating Misc. fabricated metal products
Photographic processing No IV found
Zinc Fertilizers Agricultural chemicals
Nonferrous smelting Nonferrous foundries
Chemical and allied prods Chemical and allied prods
*Major polluting industries as reported by the Toxics Release Inventory, 1987-1993
**All independent variables are measured as the percentage of the state labor force in the polluting industry

Specification 2. A more general measure of polluting industries is tested in this

model as well. This second industry variable is defined as the percent of the 1998

workforce in the top seven polluting industries. Unlike the first specification, this

variable does not differ with the dependent variable, but is a composite measure of the

overall importance of toxic industries in setting standards. The major polluters are

identified in this specification as the motivators for laxer environmental policy.

Agriculture

Agricultural operations are another major polluter of surface waters. Toxic metal

pollution from these sources generally results from the run-off of agricultural chemicals








and pesticides, as well as livestock and slaughtering facilities. The variable agriculture is

defined as the percentage of state acreage used for agricultural purposes9 and attempts to

capture these effects.

In terms of the Peltzman model, a state with a large agricultural base will see greater

benefits from looser standards, as more of their state economy is dependent on the

industry. A legislator will maximize votes in these states by adopting weaker standards,

and the expected sign on the variable is positive. However, if state officials respond to

the intensity of pollution from these sources by setting stricter environmental standards,

then the sign on the coefficient would be negative.

Geographic Characteristics

The natural variations in geographic location and climate conditions will affect the

cost of compliance with environmental regulations. The following explanatory variables

are designed to characterize these differences across states.

Precipitation. The first of these geographic variables is precipitation. In order to

capture the effects of rainfall on environmental compliance costs, I constructed a 50-year

average of state annual precipitation levels. Precipitation levels are important in

understanding the natural chemistry of the water, because rainfall will affect the rate and

flow of pollution deposits. Higher precipitation levels can lead to excess pollution

overflow from agricultural, urban, and industrial run-off, which increases the cost of

compliance with stricter standards. For this reason, I hypothesize that precipitation will

have a positive sign; more specifically, as the cost of compliance goes up, industrial

opposition to stricter standards increases.



9 A farm would be included in this acreage if it produced or sold at least $ 1,000 in agricultural products.








The second geographic variable I include to identify natural peculiarities among states

is temperature, and consists of a 50-year average of the daily mean temperature. The rate

of dissolution of toxic pollutants is highly dependent on water temperature: all else

equal, the pollutant is taken out of the water column much more quickly in a warm water

environment. Since an increase in average temperature leads to a drop in the cost of

compliance (and a decrease in industrial opposition), a legislator will be more likely to

adopt stricter standards. For this reason, the hypothesized sign on the temperature

variable is negative.

Finally, a dummy variable is included that indicates whether or not the state borders

an ocean or the Great Lakes. Research shows that pollution is less severe in coastal

cities, possibly due to the dispersal from offshore winds, which limits the deposit of

airborne pollutants in the waterways, and to the smaller average inflow of pollution from

neighboring cities. Since the political cost of abatement is smaller for coastal states, I

expect the sign on coastal to be negative.

Statistical Issues and Specification

There was concern that states adopting the national guidelines did so in order to

expend the least amount of effort complying with the federal mandate. If this were the

case, the inclusion of these states as observations in the study of state standard setting

behavior could bias the regression results. However, specification checks omitting these

states did not improve on the overall results.

I also investigated whether a two-step regression procedure would be appropriate in

these circumstances. The first step consists of explaining the decision of whether or not

to adhere to the federal guideline. The second stage addresses the variation in standards








for states that set their own levels (taking into account the likelihood of adopting the

guideline). I experienced little success in explaining which states adopted the federal

standard, and it thus appears that no systematic bias is associated with the adoption of the

federal guideline.

This being the case, two specifications are developed to test the Peltzman model in

this policy area. The first specification develops a pollutant specific explanatory variable

for industry and mining. These independent variables directly measure the prevalence of

the industries responsible for the majority of each specific pollutant.

Specifically, I estimate:

y, = Po + p1(Sierra Club)i + 12 (LCV score), + 33(Medinc)i + P4(Medinc2)i + P5
(Metal-specific industries)i + 36(Mining)i + p7(Agriculture)i + 38(Precipitation), +
39(Temperature)i + p0o(Coastal) + E,

where ei is an iid and normally distributed error term.

The second specification tests a more general measure of industry that does not vary

across the toxic metals. This independent variable is instead a composite measure of the

major industries responsible for all toxic metal pollution.

Specifically, I estimate:

yi = 3Po+ 1P (Sierra Club), + 02(LCV score)i + 33(Medinc)i + P4(Medinc2)i + P5
(Heavy industry), + p6(Agriculture)i + p7(Precipitation)i + p8(Temperature)i + P9
(Coastal)i + ei

where ei is an iid and normally distributed error term.

Results

The results from the two specifications developed to explain the variation across states

in toxic metal water quality standards are provided in Tables 3-5 through 3-9. The

specifications are separated according to the acute and chronic values of the dependent








variable. The first two tables report the specification that separates the explanatory

variables for heavy industry and mining into pollutant-specific sources. Tables 3-8 and

3-9 provide the results from the general measure of toxic metal polluting industries that

does not vary across pollutants.

In order to test the significance of the coefficients across equations (i.e. to assess

overall significance), I performed an inverse chi-square test, also known as a Fisher or

Pearson Px test. [For example, see Dewey, Husted, and Kenny (2000)]. Based on

evidence from the regressions, it can be shown that E-21ogepi, i=1, 2,...,k, has a (k)2

distribution with 16 degrees of freedom for the acute regressions and 12 degrees of

freedom for the chronic regressions, where pi is the probability given for each coefficient.

A similar Px test was performed across the regressions to test the inverted-U hypothesis

of median income. These results are reported in Table 3-7.

I used this test for the acute and chronic levels of the metals, assuming independence

across regressions within these separate values. For example, the regression for chronic

copper can reasonably be assumed to be independent of chronic zinc, but that it is not

necessarily independent of acute copper. For this reason, separate P; tests are performed

on the chronic and acute values.

Finally the results from one-tailed t tests are reported for Sierra Club, medinc,

medinc2, LCV score, precipitation, temperature, and coastal, since the null hypothesis on

these variables makes specific sign predictions. The results from two-tailed t tests are

provided for the other explanatory variables (including agriculture and the aggregate and

general specifications for heavy industry).








For simplicity, the regressions in which there are no significant coefficients are

excluded from the results tables. This includes one regression for nickel and chromium

III, and two regressions for lead. However, the statistical values from these metals are

included in the overall calculation of the Px test results.

Metal-Specific Results

Tables 3-5 and 3-6 separate the metal-specific results into the acute and chronic

values. Copper and zinc are the best performing dependent variables, with significant

coefficients on a number of the independent variables. Moreover, the average F

probabilities for these two metals is 0.057 and 0.028, along with an average R2 that is

relatively high compared to the other regressions, at 0.35 and 0.41, respectively. Lead

provides similar results for only the acute standards, while the other metals offer less

convincing evidence in support of the chosen model. The individual explanatory

variables used to explain state water quality standards in this specification provide

varying levels of support for the theory.

Agriculture is significantly positive in half of the regressions, and the Px tests show the

farming coefficients to be significantly positive across the equations. A one standard

deviation rise in the variable agriculture causes between a 0.37 and 0.56 standard

deviation weakening in the state standards. The positive and significant coefficients on

agriculture support the theory that legislators take into account the strength of farming

interests when setting environmental policy. In particular, a more dominant agriculture

industry will decrease the likelihood that a legislator will set strict state standards for

water quality.










Table 3-5: Acute Regression Results (Metal-specific industries)
Variable Copper Chromium Chromium Lead Mercury Silver Zinc
III VI
Sierra Club -4.98 -340.94- 1.07 -12.38 0.335 -1.97 17.81
(-0.67) (-1.37) (0.08) (-0.60) (0.19) (-0.23) (0.31)
LCV score 0.012 1.42 0,025 -0.080 0.335 0.026 -0.262
(0.38) (1.21) (0.42) (-0.88) (0.19) (0.59) (-1.07)
Medinc 0.0018' -0.031 0.002 0.007" -0.00004 0.0005 0.015o
(1.57) (-0.71) (0.73) (2.09) (-0.15) (0.30) (1.63)
Medinc2 -0.00000003** 0.0000004 -0.000000024 -0 0000001** 0.0000000006 -0.000000006 -0.0000002*
(-1.70) (0.61) (-0.68) (-2.16) (0.15) (-0.25) (-1.57)
Metal- -2.08 -32.32 -0.758 2.96 0.018 606 7.08
specific (-1.16) (-0.53) (-0.26) (0.34) (0.06) (1.20) (0.71)
industries
Mining 0.906 -198.56 -3.19 2.09 0.192 0.724 -3.71
(0.76) (-1.34) (-0.40) (0,74) (0.56) (0.49) (-0 50)
Agriculture 0.059* 0.937 0,141** 0.030 0.012* 0.075* 0.720***
(1.84) (0.88) (2 53) (0.35) (1.91) (1.69) (3.03)
Precipitation 0.152*** -5.35 0.149 0.214 0.006 0.041 1.09*
(2.29) (-2.07) (1.22) (1.27) (0.46) (0.44) (2.23)
Temperature -0 1055 3.65 0.006 -0.696*** -0.001 -0.080 -1.31"*
(-1.38) (1.02) (0.04) (-3.06) (-0.07) (-0.79) (-2.09)
Coastal -1.64 168.77 -4.30* 6.42 -0.192 -0.648 -19.17*
(-1.00) (2.56) (-1 31) (1.36) (-0.55) (-0.28) (-1.48)
Constant -8.22 2158.36 -19.34 12.05 2.12 -6.48 -98.12
(-0.43) (2.96) (-0.51) (0.23) (0.51) (-0.25) (-0.65)
Turning $29,290 N/A $34,110 $30,610 N/A $38,490 $33,315
Point
for Medinc
# of obs 45 40 43 44 43 38 44
Prob F 00780 0.2172 0.4791 0.0569 0.8645 0.6707 0.0309
R-squared 0.3599 0.3301 0.2345 0.3855 0.1394 0.2180 0.4173
Root MSE 3.6645 128.34 7.213 10.615 0.78799 4.438 29.343
T statistics in parenthesis: significantn t at the 1% level, significant at the 5% level, *significant at the 10% level, for a one-tailed
test (except for Heavy industry, Mining, and Farms, where a two-tailed test is used); Nickel regression nol reported (no significant
varables)

The geographic variables also perform relatively well in explaining the variation in

state toxic metal standards. Precipitation is significantly positive in five of the

regressions, where a one standard deviation rise in precipitation levels leads to an average

0.41 standard deviation increase in the dependent variable. These results are further

supported by the Px tests, in which precipitation is significantly positive across the

regressions. Temperature is significantly negative in four of the regressions, but the Px

tests support these results only for the acute standards. Moreover, a one standard

deviation rise in temperature levels prompts between a 0.22 and a 0.51 standard deviation

strengthening in the state standards.










Table 3-6: Chronic Regression Results (Metal-specific industries)
Variable Copper Chromium III Chromium VI Nickel Zinc
Sierra Club -6.40* -10.85 0.462 -129.45 27.27
(-1.33) (-0.11) (0.05) (-0.87) (0.51)
LCV score 0.024 0.106 0.013 0.362 -0.273
(1.17) (0.22) (0.32) (0.57) (-1.19)
Medinc 0.0006 -0.014 0.001 0.020 0.011*
(0.83) (-0.79) (0.73) (0.84) (1.35)
Medinc2 -0.00000001 0.0000002 -0.00000002 -0.0000003 -0.0000002
(-0.89) (0.66) (-0.68) (-0.90) (-1.25)
Metal-specific -1.41 -61.37*** -0.480 -15.17 5.64
industries (-1.24) (-2.61) (-0.26) (-0.53) (0.63)
Mining 0.40 -24.90 -1.62 -35.08 -5.89
(0.50) (-0.42) (-0.31) (-0.59) (-0.84)
Agriculture 0.041 -0.349 0.084** 1.02 0.701***
(1.90) (-0.81) (2.29) (1.59) (3.14)
Precipitation 0.066* -0.725 0.088 1.73* 0.963**
(1.53) (-0.70) (1.12) (1.43) (2.17)
Temperature -0.061 0.980 -0.010 -1.11 -1.22**
(-1.27) (0.72) (-0.11) (-0.71) (-2,19)
Coastal -1.40 26.32 -2.70 -45.97* -20.25**
(-1.27) (0.98) (-1.25) (-1.33) (-1.67)
Constant 363 412.12 -9,99 -149.52 -55.45
(0.30) (1.47) (-0.42) (-0.37) (-0.40)
Turning Point $29,710 N/A $34,300 $30,310 $34,600
for Medinc
# of obs 48 41 44 46 46
Prob F 0.0847 0.3858 0.5827 0.3040 0.0256
R-squared 0.3338 0.2703 0.2056 0.2608 0.4096
Root MSE 2.4702 52,123 4.7337 78.325 27.563

for a one-tailed test (except for Heavy industry, Mining, and Farms, where a two-tailed test is used); Lead regression
not reported (no significant variables)

Table 3-7: Pearson Px Test Statistics
Metal-specific industries Aggregate industries
Variable Acute Chronic Acute Chronic
( 2,ON) (2'(12)) (X2(16) 2( 12a)
Sierra Club 19.04 14.02 18.26 14.71
LCV Score 10.95 8.01 12.08 6.98
Medinc High values High values High values High values
26.13* 17.10 32.10"** 18.56*
Low values Low values Low values Low values
27.47** 15.56 34.03*** 20.44*
Metal-specific Neg: 13.23 Neg: 19.90* N/A N/A
industries Pos: 13.89 Pos: 7.03
Mining Neg: 14.11 Neg: 11.45 N/A N/A
Pos: 13.17 Pos: 8.68
Heavy industry N/A N/A Neg: 22.66 Neg: 19.94*
Pos: 6.65 Pos: 4.59
Agriculture Neg: 1.92 Neg: 3.62 Neg: 2.40 Neg: 3.98
Pos: 50.16*** Pos: 38.82*** Pos: 50.11"** Pos: 38.82***
Precipitation 34.52*** 23.27** 47.55*** 24.78**
Temperature 34.11*** 17.50 35.50*** 17.33
Coastal 21.47 20.23* 23105 20.96*
Low and high levels of median income are the approximate min and max values for the variable
medinc ($20,000 and $40,000, respectively).








Finally, the coastal dummy provides significant and correctly signed coefficients in

four of the sixteen possible regressions, and is significant across the regressions for both

the acute and chronic standards. A one standard deviation rise in coastal leads to an

average decrease of 0.45 standard deviations in the dependent variable. These overall

results for the geographic variables support the theory that location and climate

conditions that inflate the cost of compliance with environmental regulations will

increase industrial opposition to stricter standards. This will cause a vote-maximizing

legislator to set weaker overall standards.

The inverted-U hypothesis is supported in only the acute regressions, where three of

the eight possible equations provide confirmation of the theory. The Px tests for the acute

regressions confirm this relationship with significantly positive coefficients at low

income levels and significantly negative coefficients at high income levels. Also, the

average peak of median income (beyond which further increases in income will be

associated with a strengthening of water quality standards) is $31,000 for those

regressions where this relationship holds. This is near the national median household

income, which is approximately $29,000. States with a median income level near this

threshold level include Georgia, Indiana, Pennsylvania, Wisconsin, and Utah.

These results suggest that increases in median income will affect the state

environmental regime differently in relatively poorer or richer states. For states with

income levels below the national average, increases in income levels will further weaken

water quality standards, while those above the national average will choose to tighten

standards as income rises. These results provide an interesting addition to previous








findings in the empirical literature, extending beyond the simple explanation linking GDP

and environmental quality to the actual standards in place that regulate these changes.

The mining and LCV variables do not provide significant results, while the variables

for Sierra Club membership and metal-specific industries are significant in only one of

the regressions. With the exception of the significantly negative coefficients across the

chronic regressions for industry, the P% tests provide no further support for the theory.

Aggregate Results

The results are similar for the aggregate specifications for industry, with coefficient

signs and significance levels that are generally consistent with the earlier results. Copper

and zinc are still the best performing dependent variables, with average F probabilities of

0.043 and 0.024, along with relatively high R2 values of 0.364 and 0.402. Nickel

performs well for the chronic standard, while lead offers similar evidence for the acute

standards.

Agriculture is significantly positive in over half of the regressions, and is supported by

both Pxtests at the 1% significance level. In these regressions, a one standard deviation

rise in agriculture prompts between a 0.32 and 0.52 standard deviation weakening of the

state standards.

Precipitation is significantly positive in nine of the regressions as well, and is

supported across the equations by the Px tests. A one standard deviation rise in

precipitation levels leads to an average 0.39 standard deviation weakening of the

dependent variable. Temperature is significantly negative in five of the regressions, but

the PA tests support these results for only the acute standards. A one standard deviation










rise in temperature levels prompts between a 0.22 and 0.54 standard deviation

strengthening in the state standards.

Table 3-8: Acute Regression Results (Ag rebate industries)
Variable Copper Chromium VI Lead Mercury Nickel Silver Zinc
Sierra Club -4.66 0.133 -14.25 -0.297 -1175.44 -1.19 5.12
(-0.66) (0.01) (-0.70) (-0.18) (-0.3) .13) (0.09)
LCV score -0.0009 0.025 -0.097 0.005 3.72 0.014 -0.230

.,Medi1, ,- *-. **'* i I "

(-2.18) (-0.82) (-2.25) (-0.52) (-1.22) (-0.67) (-1.71)
Heavy industry -0.400 -0.397 -0.247 -0.066 -62.21 0.091 -0.635
(-1.55) (-0.77) (-0.26) (-1.16) (-1.18) (0.19) (-0.29)
Agriculture 0.052* 0.139" 0.010 0.011" 7.65 0.071' 0.696***
(1.91) (2.55) (0.13) (1.88) (1.34) (1.83) (2.94)
Precipitation 0.154"** 0.184* 0.253* 0.011 20.87** 0.080 1.20**
(2.53) (1.48) (1.32) (0.82) (1.67) (0.90) (2.39)
Temperature -0.114* -0.14 -0.740*** -0.002 -12.86 -0.048 -1.16"*
(-1.54) (-0.10) (-3,40) (-0.09) (-0.87) (-0.48) (-1.93)
Coastal -1.87 -4.32* 5.89 -0.159 -323.78 -1.75 -17.43'
S (-1.22) (-1.34) (1.26) (-0.46) (-1.02) (-0.80) (-1.34)
Constant -13.82 -23.13 6.81 0.107 -2339.77 -17.02 -128.38
( -0.72) (-0.60) ...1: ....., ........ ... ... ..... :,
Turning Point $29,310 $33,220 1 ,. .a. i' 1. S,," *: :.,
for Medc
#ofobs 44 42 43 42 44 37 43
Prob F 0.0303 0.3693 0.0416 0.6893 0.3441 0.7574 0.0246
R-squared 0.3904 0.2416 0.3825 0.1679 0.2366 0.1747 0.4090
Root MSE 3.576 7.1728 10.639 0.77463 735.18 4.5581 29.516
T statistics in parenthe,. L-. i .*li 'l i .111 1 i,, i I .rl ... ', r.,- 111 I ,-i I-'ra e-tailed
test(except for Heavyi..i. -, .r..... .. -Wia .:r, .. .P.,,. rr. ill.- i .. r :.j1 ..- .. ...'. ..i i.-,>rl i

Finally, the coastal dummy provides significant and correctly signed coefficients in

five of the sixteen possible regressions, but is significant across the regressions for only

the chronic standards. A one standard deviation rise in coastal causes between a 0.52 and

0.60 standard deviation decline in the dependent variable.

Heavy industry is significant in only one of the regressions (the chronic value of

chromium III). However, the Px tests show that this variable is significantly negative

across the chronic regressions at the 10% level, supporting the hypothesis that states take

a responsive role to pollution problems by setting stricter standards. Sierra Club

membership is significant and correctly signed only for chronic copper, while LCV score

is not significant in any of the equations.









Table 3-9: Chronic Regression Results (Aggregate industries)
Variable Copper Chromium III Chromium VI Nickel Zinc
SierraClub -6.10* 1.49 -0.174 -134.24 24.13
(-1.30) (0.01) (-0.02) (-0.90) (0.45)
LCV score 0.019 0.114 0.013 0.376 -0.239
(0.94) (0.22) (0.34) (0.59) (-1.05)
Medinc 0.0008 -0.014 0.001 0.031 0.011
(1.07) (-0.74) (0.81) (1.26) (1.25)
Medinc -0.00000001 0.0000002 -0.00000002 -0.0000005* -0.0000002
(-1.12) (0.62) (-0.76) (-1.31) (-1.16)
Heavy industry -0.225 -6,88* -0.234 -3.92 0.965
(-1.37) (-1.84) (-0.74) (-0.73) (0.50)
Agriculture 0.039** -0.407 0.083** 1.10** 0.684***
(2.13) (-0.91) (2.33) (1.76) (3.06)
Precipitation 0.066** -0.628 0.106* 2.11** 0.887**
(1.67) (-0.58) (1.34) (1.67) (1.96)
Temperature -0.069* 0.708 -0.026 -1.21 -1.01**
(-1.43) (0.51) (-0.28) (-0.79) (-1.84)
Coastal -1.52* 20.37 -2.68 -47.44* -19.10*
(-1.44) (0.74) (-1.27) (-1.40) (-1.57)
Constant 2.14 443.73 -10.98 -305.38 -60.24
(0.18) (1.48) (-0.46) (-0.77) (-0.43)
Turning Point $29,940 N/A $33,720 $30,410 $34,000
for Medinc
# ofobs 47 40 43 45 45
Prob F 0.0551 0.6529 0.4670 0.2090 0.0230
R-squared 0.3378 0.1857 0.2125 0.2703 0.3956
Root MSE 2.4625 54.798 4.7067 77.776 27.85
T statistics in parenthesis. ***significant at the 1% level, ** significant at the 5% level, *significant at the 10% level,
I I. ..l : ...I i '..'i -. where a two-tailed test is used); Lead regression not reported


The unusual divergence in the results for agriculture and heavy industries in both

specifications (only agriculture provides consistently significant coefficients) can be

interpreted in a number of ways. First of all, industrial concerns may be more likely to

lobby for looser standards on either the federal or local levels, and hence do not target

state level standards as a major lobbying effort. For example, on the local level, a

polluting firm can lobby for increased discharge rights through the number of permits it

receives. It can also target the specific water body of concern by lobbying for the

establishment of a designated use that allows for greater dumping levels, i.e., have the

river declared for industrial uses, instead of the stricter requirements set forth in uses

designed for human recreation or fishing.








Moreover, industry may be more easily organized at the federal level than agriculture,

where it is difficult to lobby across states lines because of increased organizational costs.

Therefore, for agricultural interests, state level lobbying efforts offer a more reasonable

alternative to influencing policy objectives. These organizational costs of lobbying may

be smaller for heavy industry, and there would be a general incentive to lobby at the

federal level for companies with facilities spanning more than one state.

Finally, the inverted-U hypothesis is more strongly supported in the acute regressions,

where three of the eight possible equations provide confirmation of the quadratic

relationship. However, the Px tests for the both the acute and chronic values support this

hypothesis with significantly negative coefficients at high income levels and significantly

positive coefficients at low income levels. Also, the average peak of median income

(beyond which further increases in income will be associated with a strengthening of

water quality standards) is $30,700 for those regressions where this relationship holds,

only $300 below the average peak of the earlier specification.

Conclusions

Using Peltzman's model as the theoretical framework, this paper explains the

variation in water quality standards across states. The results from the two specifications

that are developed for these purposes show that states take into account not only the

agriculture industry, but also income levels and geographical considerations, when setting

water quality standards. Toxic metal standards for water quality are weakest in

predominantly agricultural states, while heavy industry does not appear to have an

equivalent effect on standards.








There is also limited confirmation of an inverted-U shaped curve with respect to

environmental standards and income levels. These results verify that increases in median

income will have different effects on environmental standards in relatively poorer or

richer states. In a state with median income levels below the national average, increases

in income will lead to weaker water quality standards. States with above average levels

of median income will select stricter standards as income grows. The findings in the

previous literature establishing a link between environmental quality and income levels

could have reflected lax enforcement of existing regulations. However, this work implies

that low-income states actually choose weaker standards, instead of simply turning a

blind eye to feeble enforcement efforts.

Notes

1. States are required by the CWA to review their standards every three years and
have them re-approved by the EPA before implementation a process that
eliminates the possibility that a state could enact unacceptably low standards.
Consequently, there is a fuzzy lower limit under which the EPA will reject weak
standards, as they do on a regular basis. However, the rejection of state proposals
is not an exact science, as the EPA uses a different measuring stick for each state
that takes into account regional-specific considerations, along with scientific
evidence available at the time of approval. Unfortunately, the EPA does not
maintain data on which state plans are initially rejected, or a past record of state
water quality standards. At present, the only data available for analysis are the
current standards in place for each state.

2. Most of the toxic criteria are calculated with respect to a hardness factor
(expressed in milligrams per liter as calcium carbonate, CaCO(3)). The hardness
factors were given at varying levels (50, 100, 200), and I used 100 ml/I for each
state in order to make comparison feasible. Also, the metal criteria were listed in
either their dissolved or total recoverable form, followed by a conversion factor
specific to each metal that should be used to calculate one given the other. I
standardized the state criteria by documenting only the total recoverable form, and
converting from the dissolved form if that were the only one listed by the state.
To illustrate this, assume that a state lists the acute standard for copper using a
hardness factor of 50 in the calculation. In order to facilitate comparison across
states, I would have to recalculate the standard using the given equationto with a

'0 An equation for calculating the acute copper standard is: [e^(1.8190(In(hardness))+3.688)].









hardness factor of 100. Also, if the standard were listed in its dissolved form, I
would multiply this value by a conversion factor" in order to transform it to the
total recoverable form.


' The conversion factor for the acute copper standard is .960.













CHAPTER 4
THE POLITICS OF STATE AIR QUALITY STANDARDS

Introduction

The Clean Air Act of 1970 outlined regulations for six air pollutants, also known as

the National Ambient Air Quality Standards (NAAQS). Although states are allowed

flexibility in designing an individualized program to meet these federal requirements,

they are restricted from setting standards weaker than NAAQS. The Clean Air Act does,

however, provide states the ability to set standards stronger than these national levels. In

particular, sulfur dioxide regulations vary significantly across states over 20% currently

impose greater restrictions for this aspect of air quality. The variability in SO2 standards

facilitates an empirical investigation into the characteristics of strong air pollution control

programs, and will provide the focus of the current analysis.

The theoretical foundation of this work is Peltzman's (1976) legislator vote-

maximization model. In the context of this paper, state officials behave rationally by

setting environmental policy to maximize political support (and increase the probability

of re-election). More specifically, legislators face a trade-off between the desires of

consumers for better environmental quality, and that of polluters who favor less stringent

regulations.

This chapter tests the Peltzman model by developing a number of specifications to

characterize the variation in sulfur dioxide standards across states, controlling for both

the strength of industry and consumer groups, as well as geographical differences that

affect the cost of compliance. The results from this investigation strongly suggest that








the decision to adopt SO2 standards stricter than NAAQS is responsive to these forces,

supporting the theory that legislators trade-off consumer and producer interests when

setting environmental policy.

The existence of an inverted-U shaped curve linking state environmental standards

and income is also investigated. This application of the inverted-U hypothesis asserts

that the evolution of environmental regulations is essentially different at low and high

levels of income. More specifically, income growth at lower levels will precipitate a

weakening of environmental standards. However, beyond a certain point further

increases in income will be associated with a strengthening of these same standards.

National Ambient Air Quality Standards

The Clean Air Act of 1970 required the newly formed Environmental Protection

Agency to establish national standards for a set of principal air pollutants. These new

regulations, known as the National Ambient Air Quality Standards (NAAQS),

specifically target the reduction of six air pollutants, namely sulfur dioxide (SO2),

nitrogen dioxide i '.i, particulate matter (PMio and PM2.5), carbon monoxide (CO),

ozone (03), and lead (Pb).

Two different types of standards were defined for NAAQS purposes, and these

regulations are outlined in Table 4-1. A stricter primary standard was developed for

the protection of human populations, including that of sensitive sub-populations such as

children, the elderly, and asthmatics. Another secondary standard targeted the

protection of public welfare, such as decreased visibility, damage to animals, crops, and

property. However, not all pollutants are regulated by both primary and secondary

standards. Their application to a specific pollutant depends on both its chemical








composition and potential negative effects. For example, the primary and secondary

standards are equivalent for nitrogen dioxide, particulate matter, ozone, and lead, while

carbon monoxide is regulated by a primary standard alone. Furthermore, sulfur dioxide

is the only criteria pollutant that has distinct primary and secondary standards.

Table 4-1: National Ambient Air Quality Standards
Pollutant Standard Value' Standard Type
Sulfur Dioxide (SO2)
Annual Arithmetic Mean 0.030 ppm Primary
24-hour Average 0.14 ppm Primary
3-hour Average 0.50 ppm Secondary

Nitrogen Dioxide (NO2)
Annual Arithmetic Mean 100 Pg/m3 Primary & Secondary

Particulate Matter
(PM 10)2
Annual Arithmetic Mean 50 pg/m3 Primary & Secondary
24-hour Average 150 pg/m3 Primary & Secondary

(PM 2.5)3
Annual Arithmetic Mean 15 pg/m3 Primary & Secondary
24-hour Average 65 pg/m3 Primary & Secondary

Carbon Monoxide (CO)
8-hour Average 9 ppm (10 mg/m 3) Primary
1-hour Average 35 ppm (40 mg/m3) Primary

Ozone (03)
1-hour Average 0.12 ppm Primary & Secondary
8-hour Average 0.08 ppm Primary & Secondary

Lead (Pb)
Quarterly Average 1.5 pg/m3 Primary & Secondary
Parenthetical value is an approximately equivalent concentration
2 Particles with diameters of 10 micrometers or less
Particles with diameters of 2.5 micrometers or less
Definitions:
ppm = parts per million by volume
mg/m3 = milligrams per cubic meter of air
pg/m3 = micrograms per cubic meter of air








NAAQS is also distinguished by the time periods over which the ambient air

conditions are monitored, such as yearly, daily, and hourly averages. To illustrate, the

three-hour average for SO2 (0.50 parts per million) is allowed to peak at 17 times the

annual mean (0.03 parts per million). This variability in the standards across time periods

is intended to account for temporary spikes in emissions that may occur during the year.

Although states are prohibited from pursuing policies that preclude compliance with

existing federal standards, they are allowed the flexibility to develop regulations stricter

than NAAQS. Tables 4-2 through 4-6 lists the states that currently have regulations

stricter than those mandated by the federal government under the NAAQS program.

With the exception of PM 2.5 and lead, all of the criteria pollutants are subject to some

variation across the states. Between one and four states impose stricter standards for

NO2, PM10, CO, and 03.

Table 4-2: Sulfur Dioxide (SO2) Standards
State Annual Arithmetic 24-hour Average 3-hour Average
Mean (Primary) (Primary) (Secondary)
NAAQS 0.03ppm 0.14ppm 0.5ppm
California Same 0.04ppm Same
Colorado Same Same 0.27ppm
Florida 0.02ppm 0.1 ppm Same
Maine 0.022ppm 0.088ppm 0.440ppm
Minnesota* Same Same 0.35ppm
Montana 0.02ppm 0.10ppm Same
New Mexico** 0.02ppm 0.10ppm Same
North Dakota 0.023ppm 0.099ppm 0.273ppm
Oregon 0.02ppm 0.10ppm 0.050ppm
Washington 0.02ppm 0.1 ppm 0.4ppm
Wyoming 0.02ppm 0.10ppm Same
*Applies to Air Quality Control Regions 127, 129, 130, and 132; NAAQS apply to all other regions
**Standards apply except within 3.5 miles of the Chino Mines Company smelter furnace stack at Hurley,
where NAAQS apply








Table 4-3: Nitrogen Dioxide (NO2) Standards
State Annual Arithmetic Mean
(Primary and Secondary)
NAAQS 100mp3g/m3
Hawaii 70pg/m3

Table 4-4: Particulate (PM 10)
State Annual Arithmetic Mean 24-Hour Average
(Primary and Secondary) (Primary and Secondary)
NAAQS 50pg/m3l 150p.g/m3
California 30rg/m3 50ig/m3
Maine 40g/m3 Same

Table 4-5: Carbon Monoxide (CO) Standards
State 8-hour Average 1-hour Average
(Primary) (Primary)
NAAQS 9ppm (10mg/m3) 35ppm (40mg/m3)
California Same 20ppm
Hawaii 5mg/m3 10mg/m3
Minnesota Same 30ppm
Montana Same 23ppm

Table 4-6: Ozone Standards (03)
State 8-hour Average 1-hour Average
(Primary and Secondary) (Primary and Secondary)
NAAQS 0.08ppm 0.12ppm
California Same 0.09ppm
Montana Same 0.10ppm
Nevada Same 0.10ppm*
*Applies to Lake Tahoe Basin (#90), NAAQS apply to all other regions

Sulfur dioxide standards offer by far the greatest amount of variation, as eleven states

currently implement standards stricter than the federal regulations. The increased

variability in SO2 standards facilitates an empirical examination of the factors that

determine whether a state adopts stricter air quality standards, and for that reason will be

the focus of the current analysis.

Sulfur dioxide. Sulfur dioxide is an atmospheric pollutant that results from the

burning of fuels that contain sulfur, a common ingredient in raw materials such as crude

oil, coal, and metallic ores. The major sources of SO2 emissions are pictured in









Figure 4-1. According to the EPA, 67% of SO2 emissions come from electric utilities

that bum coal and petroleum (a total of 13 million tons per year). Other fuel combustion

sources include industries such as petroleum refineries, cement manufacturing, and metal

processing facilities. These industries bum either coal or oil to process heat, or derive

their products from raw materials containing sulfur. Non-road engines and vehicles,

including locomotives, large ships, and diesel equipment, contribute 5% of the total

sulfur dioxide emissions.

Figure 4-1

Non-Roial
Engies &
Vehicles Metal
l 5% Processmqi
Other 3%
70h
Fuel
Combustion
Industrial
& Other
18% |





Fuel Combustion
Electrical Utilities
67%

Available from the US EPA Office of Air Quality Planning & Standards, November 2000

The effects of SO2 pollution are considered to be a serious threat to public health, with

the most vulnerable segments of the population being the elderly, children, and

asthmatics. Among other exposure related effects, it can contribute to respiratory illness,

aggravate existing heart and lung diseases, and lead to skin irritation and inflammation.

The environmental consequences of SO2 pollution are also numerous, the most notorious

of which being acid rain. Acid rain is responsible for damaging forests and crops,

decreasing soil fertility, and acidifying lakes and streams to the extent that they are no








longer suitable for aquatic life. Acid rain has also been blamed for damaging buildings

and outdoor structures, and has decreased visibility in many areas.

Empirical Design

The Peltzman model of legislative decision-making is employed in describing the

variation among state ambient air quality standards for sulfur dioxide. In the context of

this work, a state legislator will balance the loss of industry votes due to stricter standards

against the gain in consumer votes from enhanced environmental quality. This being the

case, each state's environmental standard-setting behavior will reflect its own political,

economic, and social cost-benefit structure.

Sulfur dioxide ambient air quality standards will be used as the dependent variable in

the analysis, and will take on a value of one if the state sets any standard stricter than

NAAQS, and zero otherwise. This specification will be discussed in detail, along with a

second specification that uses the actual SO2 standards as the dependent variable. The set

of independent variables chosen to characterize this variation are described below, and

summary statistics are available in Table 4-7.

Consumer Influence

I measure consumer influence in the empirical model by including a set of

independent variables that capture the demand for environmental quality among the state

electorate. The variables that I identify for these purposes are participation in the Sierra

Club, the level of environmentalism of elected officials (LCV scores), and median

income.

Sierra Club. The percent of the state population belonging to the Sierra Club is used

to measure the demand for environmental quality among state consumers. As the largest








grassroots conservation organization (over 700,000 members nationwide), the Sierra

Club variable should provide an adequate proxy of consumer tastes for the environment.

Following the Peltzman model, a state with a larger percentage of consumers

participating in the Sierra Club should select stricter regulatory standards. Therefore, the

predicted sign for Sierra Club is positive.

Table 4-7: Summary of Variables
Mean Stand. Dev. Min. Value Max. Value
SO2 Standard 0.229 0.425 0 1
Annual SO2 Standard 0.16 0.370 0 1
24-hour SO2 Standard 0.18 0.388 0 1
3-hour SO2 Standard 0.12 0.329 0 1
Sierra Club 0.202 0.121 0.043 0.547
LCV Score 42.88 25.68 5 98
Median income 40,940 6,204 29,696 55,146
Coal-burning utilities 0.004 0.007 0 0.040
Coal-burning industries 0.825 0.497 0.236 2.74
Distance to coal quality 1046.4 669.73 0 2115.2
Temperature 51.99 7.65 40.1 70.6
Coastal 0.563 0.501 0 1
PSD parkland 0.644 1.42 0 6.48

LCV Score. The voting behavior of elected officials on environmental issues is a

second measure of consumer preferences in this model. The propensity to elect

environment-friendly officials is an important indicator of overall demand for these

goods. This variable is measured as the average rating on a series of recent League of

Conservation Voters scorecards for members of the U.S. House of Representatives

(1997-2001), revised to exclude potential partisan bias by the organization [see Chapter 5

for a detailed explanation of the revised index]. The LCV scorecard rates congressional

voting behavior on select environmental votes, where a 100% score corresponds to

perfect agreement with the LCV agenda.








The expected sign for this variable is positive. The higher the adjusted state LCV

scores, the more likely consumers are concerned with environmental issues (as evidenced

by their voting behavior), and thus the stricter should be the state standards.

Environmental Kuznets Curve. The inclusion of median income levels as an

indicator of increased consumer opposition to relaxed environmental standards is less

straightforward. An extensive amount of research has provided evidence that median

income and environmental quality are better represented by a quadratic form (inverted-U

hypothesis). This theory argues that environmental quality suffers at low levels of

income growth, but beyond a certain threshold income level, economic growth is

associated with greater levels of environmental quality.

The empirical literature does not extend this relationship between median income and

environmental quality to the actual standards in place to achieve these results, which is a

focus of the current paper. The hypothesized signs for the coefficients on median income

and median income-squared are negative and positive, respectively, i.e. SO2 standards are

weakened at low levels of income growth, and strengthened at high levels of income

growth. In order to test for a simpler monotonic effect of income on environmental

standards, the results from an alternative specification excluding median income-squared

from the equation will also be reported.

A final consideration is whether the regression suffers from multicollinearity among

the three consumer variables. Although a significant amount of correlation does exist

between Sierra Club, liberalism, and median income (between 0.55 and 0.59),

specification checks did not provide evidence that this correlation was having a large








effect on the outcome. However, two additional sets of results are provided that

individually exclude Sierra Club and liberalism to address these concerns.

Polluter Influence

In the Peltzman model, industrial concerns that are affected by an increase in the

stringency of environmental regulations will lobby legislators against such policies. The

state officials will weigh the gain in industry votes from decreasing environmental

regulations against the loss in consumer votes that results from lower levels of

environmental quality. For that reason, states with greater interests in heavily polluting

industries are more likely to set weaker air quality regulations.

However, some research has linked higher levels of industrialization to more liberal

environmental policies. [See, for example, Lowry (1992), Ringquist (1993)]. The

hypothesis behind this assumption is that states with large polluting industries find it

necessary to develop reactionary environmental policies to control already existing

pollution problems within the state. In the context of this model, the value to consumers

of reducing pollution levels is higher in relatively dirtier states, which will boost overall

consumer opposition to relaxed environmental standards that favor industry.

Subsequently, a vote-maximizing legislator would set more restrictive environmental

policies in these states. The contrasting theories regarding the regulatory response to

industrial strength will be tested by three explanatory variables that measure the

importance of SO2 polluting industries.

Coal-burning electric utilities. The EPA identifies coal-burning electric utility

plants as the source of 67% of sulfur dioxide emissions within the United States (see

Figure 4-1). Since these generating plants contribute so heavily to SO2 pollution








nationwide, the first independent variable seeks to capture their prevalence across states.

The variable coal-burning utilities measures the strength of this industry within a state,

and is defined as the per capital energy produced from coal (in million kilowatt-hours).

The coefficient will take on a negative sign if industrial interests are weighed more

heavily than consumer concerns. However, a positive sign on this coefficient is more

likely for two reasons. First, since coal-burning energy producers are responsible for the

majority of sulfur dioxide emissions, consumer opposition and responsive state policies

are likely to be more prevalent in this industry than in others.

Moreover, public utilities represent government sanctioned natural monopolies. Due

to this lack of competition in the market for energy, these producers have less to fear

from strict environmental policies. They are not threatened by either the possibility of

going out of business or the loss of potential customers, and simply pass on the higher

compliance costs to its consumers.

Heavy industry. Although greater than half of sulfur dioxide emissions come from

coal-burning utilities, it is important to control for other contributors to SO2 pollution as

well (see Figure 4-1). According to the EPA, these industries include the manufacture of

petroleum and coal products (NAICS 3273), cement and concrete products (NAICS 331),

and primary metals (NAICS 324). The second independent variable that is constructed to

measure the effects of industrial pollution is the percent of the state labor force working

in these three industries.

The hypothesized sign on coal-burning industries is positive if consumer concerns are

weighed more heavily than industrial interests. However, a negative sign on this

coefficient is more likely since stricter environmental standards will result in lower








profits and possible plant closures for local private industries. Also, these private firms

provide a credible threat of relocation, increasing the likelihood that a legislator will

pursue policies that favor industry.

Distance to quality coal. It is also important to take into account differences in the

production processes across firms that affect the severity of SO2 emissions. The largest

factor responsible for differences in sulfur emissions is the quality of coal used (high or

low sulfur coal). Firms that utilize coal sources low in sulfur will release significantly

fewer emissions than firms that bum high sulfur coal. In order to adequately control for

this effect, the average quality of coal available to firms within a state must be taken into

account.

Distance to the closest state mining low sulfur coal (measured in miles between state

capitals) is used to proxy for the availability of clean coal inputs to state industries. A

low sulfur coal source is one that produces coal with a sulfur percent by weight no greater

than 0.75% (Wyoming, Montana, Utah, Colorado, Arizona, North Dakota, and New

Mexico are included in this variable as low sulfur coal sources). The cost of complying

with air quality standards is lower for states with better access to low sulfur coal sources.

As a result, opposition from industry to strict standards should be lower than in states that

are farther from away these sources. The hypothesized sign for this variable is negative,

i.e. as the distance to a low sulfur coal source increases, the cost of complying with strict

regulations goes up and decreases the likelihood a state will adopt a stricter SO2 standard.

Geographic Characteristics

The natural variations across states in geographic location and climate conditions will

affect the cost of compliance with environmental regulations. For this reason, it is








necessary to control for these inherent differences in order to avoid an omitted variable

bias in the regression outcomes.

Temperature. The first geographic variable identifies natural peculiarities among

states in temperature. Sulfur dioxide emissions react with other atmospheric chemicals to

form sulfuric acid, a key component of acid rain. The speed and intensity of these

chemical reactions is dependent in part upon the availability of heat and sunlight, and

temperature is included to proxy for these effects.

Higher temperature levels are associated with an increase in the cost of compliance for

polluting industries. A vote-maximizing legislator will account for these greater

abatement costs to industry by setting less stringent regulations, and the hypothesized

sign on this coefficient is negative.

Coastal location. A dummy variable is included to determine whether the state

borders an ocean or the Great Lakes. Research shows that pollution is less severe in

coastal cities, possibly due to the dispersal from offshore winds and to the smaller

average inflow of pollution from neighboring cities. Since the cost of compliance for

polluters is smaller in coastal states, industrial opposition should be less severe. For this

reason, I expect the sign on coastal to be positive, i.e. coastal states are more likely to

implement stricter SO2 standards because the cost of doing so is less compared to non-

coastal states.

PSD parklands. Another geographic issue deals with the location of Class I

Prevention of Significant Deterioration Areas.12 Since states with Class I regions are

already subject to harsher federal regulations regarding the degradation of air quality, it is


12 See Note I for detailed explanation of PSD policies.








possible that the overall state regulations are a function of the existing PSD program

within the state, i.e. stricter standards already apply within the majority of the state.

In order to control for this possibility, I constructed an estimate of PSD regions as the

percent of state acreage belonging to the mandatory Class I areas including national

parks, national wilderness areas, national memorial parks, and international parks. States

have the authority to classify other lands as Class I as well, so this measure

underestimates the total Class I regions within a state.

However, since other Class I regions are not required by federal mandate, the effect is

better captured by these exogenous Class I areas. A state with a larger percentage of land

in national parklands will be more likely to impose stricter state air quality standards

because more stringent regulations already apply widely within the state. Therefore, the

predicted sign for PSD parkland is positive.

Excluded geographic measures. Two excluded variables measuring climate

differences across states are precipitation and wind. The first of these, precipitation, is

the medium through which sulfur dioxide returns to the land in the form of acid rain -

higher precipitation levels will therefore lead to a greater incidence of acid rain within a

state. A correlation matrix showed that precipitation was highly correlated with the coal

quality measure (0.80), and specification checks provided evidence that this collinearity

was affecting the results (precipitation was significant only when coal quality was

excluded from the regression). Precipitation was also correlated with the coastal dummy

(0.50), which exacerbated the existing multicollinearity problem. For these reasons, this

variable was dropped from the regression.








Another geographic variable was constructed that measured wind, an important factor

in the dispersal and deposit of air pollutants such as sulfur dioxide. However, this

variable was difficult to define since direction and speed varied within the state, and data

are only available for a limited number of these sites. I constructed a measure of the

average annual wind speed, but this variable did not improve the explanatory power of

the model. The likely reason for the failed results is that the variable constructed did not

adequately measure the effects of wind on SO2 deposit and dispersal.

Statistical Issues and Specification

The dichotomous dependent variable that has been developed to test the current model

would generally lead to the application of a probit (or logit) empirical design. However,

since some of the independent variables in this model so closely predict the outcome, use

of these techniques becomes problematic. The probit has difficulty determining the

impact of an explanatory variable in the model when it almost perfectly explains which

states adopt the stricter standards. [For greater discussion, see Kenny and Lotfinia

(2002)]. For that reason, a linear probability model will be employed to test the

theoretical model in this paper.

Table 4-8 provides the results from four specifications that are developed to test an

aggregate measure of state sulfur dioxide standards. In these specifications, a state is

considered to have an SO2 standard stricter than NAAQS as long as one of the three

standards for sulfur dioxide (annual arithmetic mean, 24-hour average, or 3-hour average)

exceeds the federal guidelines. The first specification includes all of the explanatory

variables introduced in the preceding sections (except for precipitation and wind) as

possible determinants in the adoption of strict SO2 standards. Alternative specifications








exclude the variables for LCV score, Sierra Club membership, and median income-

squared, respectively. The analysis of these additional specifications is intended to test

the robustness of the model and the extent to which multicollinearity is affecting the

outcome.

Specifically, I estimate the following linear probability model:

yi = Po+ P3(Sierra Club)i + P2(LCV score),+ P3Medinci + p4(Medinc)2, + 35(Coal-
burning utilities)i + 36(Coal-burning industries)i +37(Distance to coal quality)i +
Pg(Temperature)i + 39(Coastal)i + 30o(PSD parklands) + e,

where e is an iid and normally distributed error term and the dependent variable has the

following form:

y, = 0 if state adopts NAAQS for SO2
y, = 1 if state imposes any standard stricter than NAAQS for SO2

The results from an alternative specification of the dependent variable are reported in

Table 4-9, which separates it according to each time-dependent standard (annual mean,

24-hour average, and 3-hour average). In these specifications, the dependent variable

takes on the actual value of the state SO2 standard, in contrast to the dichotomous

dependent variable of the previous regressions. Since smaller numeric standards translate

into stricter regulations, the hypothesized signs on each of the independent variables will

be reversed, i.e. smaller values of the dependent variable correspond to tougher standards.

For simplicity, only the results for the initial specification that includes all explanatory

variables (specification 1 of Table 4-8) will be reported for the disaggregated dependent

variable.

The results from one-tailed t tests are reported for Sierra Club, medinc, medinc2, LCV

score, coastal, PSDparkland, temperature, and distance to quality coal, since the null

hypothesis on these variables makes specific sign predictions. The results from two-








tailed t tests are provided for the other explanatory variables (including coal-burning

utilities and coal-burning industries). Also, the study is limited to the contiguous United

States because of the difficulty of measuring proximity to low sulfur coal for Alaska and

Hawaii.

Results

Aggregate SO2 Specifications

Table 4-8 reports the empirical results for the four specifications aggregating the

dependent variable. In these regressions, the dependent variable takes on a value of one

if the state sets an SO2 standard stricter than NAAQS, and zero otherwise. All of the four

specifications perform well in explaining the variation in state sulfur dioxide standards -

the average R2 is 0.68 and the average adjusted R2 is 0.60. Moreover, the overall fit of

each of the equations is highly significant, as evidenced by their F-probabilities of 0.000.

The first specification that includes all of the independent variables has the highest R2

value (0.70), while the second specification excluding environmental liberalism produces

the highest adjusted R2 (0.63).

Consumer influence. The set of variables that measure consumer influence on

legislative decision-making provide results of varying success. Sierra Club is the best

performer of these variables and is significant and correctly signed in each of the

specifications. A one standard deviation rise in the percent of the state population

belonging to the Sierra Club results in an average 0.23 increase in the likelihood that the

state will adopt a stricter SO02 standard.








This provides strong evidence to support the hypothesis that as the percentage of

citizens belonging to the Sierra Club rises (reflecting a greater demand for environmental

quality), the more heavily consumer interests are weighed relative to polluter interests.

Table 4-8: Regression Results Aggregating the SO2 Standard
Variable Specification 1 Specification 2 Specification 3 Specification 4
Sierra Club 1.85*** 1.91*** N/A 1.82***
(3.44) (3.96) (3.36)
LCV score 0.0008 N/A 0.0049** 0.0012
(0.27)) (1.67) (0.41)
Median income -0.00014** -0.00014** -0.00012* -0.0000358***
(-1.76) (-1.82) (-1.32) (-4.08)
(Median income)2 0.0000000012* 0.0000000012* 0.0000000011 N/A
(1.31) (1.36) (1.03)
Coal-burning 12.37 12.43* 3.52 13.53*
utilities (1.65) (1.68) (0.44) (1.80)
Coal-burning -0.172** -0.174** -0.19* -0.178**
industries (-2.02) (-2.08) (-1.98) (-2.08)
Distance to -0.0002*** -0.0002*** -0.0003*** -0.0002***
quality coal (-2.74) (-3.15) (-3.10) (-2.65)
Temperature -0.014** -0.014** -0.018** -0.012*
(-1.86) (-2.19) (-2.20) (-1.61)
Coastal 0.384*** 0.393*** 0.301*** 0.363***
(3.39) (3.66) (2.39) (3.21)
PSD parkland 0.087** 0.086*** 0.145*** 0.078**
(2.39) (2.41) (3.99) (2.16)
Constant 4.18** 4.27** 4.11** 1.96***
(2.34) (2.46) (2.03) (3.49)
Turning Point $56,980 $56,910 $54,770 N/A
For Medinc
# of obs 48 48 48 48
Prob F 0.0000 0.000 0.0000 0.0000
RI 0.7033 0.7028 0.6085 0.6896
Adj. R2 0.6232 0.6324 0.5158 0.6161
Root MSE 0.26074 0.25754 0.29556 0.26318
T-statistics in parenthesis
***significant at 1% level, **significant at 5% level, *significant at 10% level, for a one-tailed test (except
for Coal-burning utilities and Coal-burning industries, where a two-tailed test is used)

The variable measuring average LCV scores of elected officials is significant and

correctly signed in only one of the three possible regressions, specifically the regression

which excludes Sierra Club. For this specification, a one standard deviation rise in the








revised LCV score prompts a 0.13 increase in the likelihood of adopting a stricter SO2

standard. Although this variable is not significant in the other two regressions, it is

important to note that each coefficient does have the expected positive sign. The weak

results in the specifications that include Sierra Club as an independent variable may be

due to the correlation between these variables (0.55).

The inverted-U hypothesis is tested in the first three specifications, and produces

significant results in two of these. Median income is significant and correctly signed in

all four of the specifications, including the regression that omits its squared value.

Median income-squared produces similar results, and is significant and correctly signed

in each specification except for the one excluding the Sierra Club variable. These results

support the hypothesized quadratic relationship between income and standards (standards

are worsened at low levels of income growth and strengthened at higher levels).

However, the average peak value of median income (beyond which standards would

begin to increase in stringency) is around $56,000, nearly $1,000 above the maximum

value for state median income. This high value suggests that, although a quadratic

relationship exists, the potential for pollution rises more rapidly than the demand for a

clean environment at these income levels.

Polluter influence. The independent variables measuring the importance of industrial

concerns in legislative decision-making display interesting results as well. The variable

that represents the prevalence of coal-burning utilities within a state is significant in two

of the specifications, and marginally significant in a third (.107). A one standard

deviation rise in the amount of per capital energy produced from coal results in a 0.9

increase in the likelihood a state will adopt a stricter SO2 standard. In interpreting these








results, it is important to note that all of the coefficients are consistently signed (positive).

This supports the hypothesis that greater amounts of polluting energy generation will spur

consumer opposition to the extent that reactionary policies are developed to address the

pollution.

The variable that measures the industrial strength of other pollution sources displays

the opposite results. The coefficients on the coal-burning industry variables are

consistently negative in sign, and are significant in all four of the specifications. A one

standard deviation rise in the percent of the state labor force belonging to polluting

industries results in a 0.9 decline in the probability a state will adopt a stricter SO2

standard. This suggests that the alternative scenario is at work for private polluting firms,

where the industrial lobby is sufficiently strong to curb further increases in environmental

regulations beyond the federal requirements.

Distance to high quality (low sulfur content) coal is significantly negative in all of the

regressions. A one standard deviation rise in the distance to a high quality coal source

will lead to an average decrease of 0.15 in the likelihood that a state will adopt a sulfur

dioxide standard stricter than NAAQS. These results provide evidence to support the

hypothesis that states farther away from a high quality coal source are less likely to set

stricter SO2 standards. Better access to high quality coal will decrease industrial

opposition to stricter environmental policies, as the cost of complying with these

regulations goes down.

Geographic characteristics. The variables added to control for geographical

differences across states also performed well in each of the specifications. Temperature

is significantly negative in all four of the regressions. A one standard deviation rise in








temperature results in an average drop in the dependent variable by 0.12. The

consistently negative coefficients support the hypothesis that industrial opposition to

stricter environmental standards increases as unfavorable climate conditions inflate the

cost of compliance.

The coastal dummy is significantly positive in all of the specifications, consistent with

the theory that the cost of complying with stricter regulations is lower in coastal states

relative to their inland neighbors. Moreover, the probability of adopting stricter SO2

standards is 0.36 higher in coastal states.

The prevalence of mandatory Class I PSD regions, measured as the percent of the state

that is taken up by national parklands, has consistently positive and highly significant

coefficients in all of the specifications. A one standard deviation rise in PSD parklands is

estimated to increase the probability by 0.14 that a state will enact stricter regulations

than those mandated under NAAQS. These results support the hypothesis that a state

with a greater percentage of Class I regions is more likely to set stricter SO2 standards, as

more stringent PSD regulations already apply within the state.

Results of Three Separate SO2 Standards

The results from the alternative specification of the dependent variable, separating it

according to the three distinct standards for SO2 (annual mean, 24-hour average, and 3-

hour average) are available for review in Table 4-9. The predicted signs for the

coefficients are now reversed since larger values of the dependent variable correspond to

weaker standards.








Table 4-9: Regression Results for Three Separate SO2 Standards
Variable Annual Standard 24-hour Standard 3-hour Standard
Sierra Club -0.010* -0.069*** -0.365***
(-1.60) (-2.70) (-2.57)
LCV score 0.00000312 -0.0000762 -0.00098*
(0.10) (-0.57) (-1.31)
Median income 0.000000995 0.00000787** -0.00000438
(1.13) (2.12) (-0.21)
(Median income)2 -0.0000000000078 -0.000000000072** 0.00000000011
(-0.75) (-1.64) (0.44)
Coal-burning -0.106 -0.182 -2.75
utilities (-1.26) (-0.51) (-1.39)
Coal-burning 0.001 0.006 0.014
industries (1.42) (1.42) (0.61)
Distance to 0.000000762 0.00000565* 0.00007***
quality coal (0.75) (1.31) (2.95)
Temperature 0.00014** 0.0049* -0.00032
(1.71) (1.42) (-0.17)
Coastal -0.0027** -0.013*** -0.067**
(-2.08) (-2.40) (-2.24)
PSD parkland -0.00074** -0.009*** 0.019
(-1.81) (-4.99) (2.00)
Constant -0.004 -0.074 0.557
(-0.19) (-0.88) (1.18)
Turning Point $63,950 $54,960 $20,660
for Medinc
# of states deviating 8 9 6
from NAAQS
#of obs 48 48 48
Prob F 0.0044 0.0000 0.0095
R2 0.4460 0.7191 0.4364
Adj. R 0.3216 0.6431 0.2841
Root MSE 0.00293 0.01235 0.06896
T-statistics in parenthesis
***significant at 1% level, **significant at 5% level, *significant at 10% level, for a one-tailed test (except
for Coal-burning utilities and Coal-burning industries, where a two-tailed test is used)

These regressions are less successful overall than the linear probability regressions

just discussed, but still provide some interesting results. The 24-hour standard achieves

the best results, with an R2 of 0.72 and an adjusted R2 of 0.64. However, this is not

surprising since it also possesses the greatest degree of variation among states (a total of

nine deviate from the federal standard).








Sierra Club membership and the coastal dummy are the best performing explanatory

variables, as they provide significant and correctly signed coefficients in each of the

regressions. PSD parklands, temperature, and distance to high quality coal provide

similar confirmation of the theory in two out of the three specifications. The average

adjusted LCV score of elected officials is significant and correctly signed only for the 3-

hour standard. Finally, the inverted-U hypothesis is supported in the 24-hour standard,

but not for the other two specifications.

It is not surprising that the results are not as conclusive as those offered by the

aggregate specifications, as separating the dependent variable in this way limits the

number of states that deviate from the national standards, and therefore provides less

variation to be explained in the model. However, these results appear to generally

confirm those of the earlier specification, as the signs on these variables are consistent

with the previous results. In particular, the 24-hour standard provides conclusive results

to support the Peltzman model, similar to those offered by the aggregate specifications of

the dependent variable.

Conclusions

The results from an empirical analysis of state-level sulfur dioxide standards

provide evidence to support the Peltzman model of legislator vote maximization. The

decision to adopt SO2 standards stricter than NAAQS is proven to be highly responsive to

environmental organizations and industrial interests, as well as other geographic factors

that affect the cost of compliance. These results are consistent with the idea that

legislators trade-off the interests of consumer groups and industrial polluters when setting

an environmental agenda.








The first question posed in this paper is the following: Do legislators select the

standards favored by consumer groups, or the more relaxed standards preferred by the

industrial polluters? The most active explanatory variable measuring this trade-off is

Sierra Club membership, where a one standard deviation rise in the percent of the

population belonging to the organization results in an average increase of 0.23 in the

likelihood a state will adopt a stricter SO2 standard. Although the adjusted LCV scores of

elected officials offers less convincing results, it is significantly positive in a third of the

regressions, and is correctly signed in all but one of these.

The industrial variables, including energy generation from coal sources, labor force

participation in major polluting firms, and availability of high quality coal, provide a

more complicated view of legislative decision-making. First, stricter sulfur dioxide

standards are set for states that rely more heavily on coal-burning electricity generation,

i.e., consumer concerns are more salient to legislators than are the interests of these

polluting utilities. Since these energy-generating firms represent government sanctioned

natural monopolies, they will produce energy regardless of environmental regulations.

For this reason, they are less likely to feel threatened and lobby against potential

increases in environmental stringency.

The variable that measures industrial strength of other primary pollution sources

displays the opposite effect. Although the coefficients for coal burning industries are

significant only in the aggregate specifications, the signs on all of the coefficients suggest

that private industrial forces lobby effectively against further restrictions on firm

activities. Since these interests do not have the monopoly power that energy producers








possess, they are more threatened by state regulations that raise their production costs

relative to competitors' costs in other states.

Finally, the distance to a high quality coal source is highly significant in all but one of

the specifications, and is consistently signed in all of them. This supports the hypothesis

that as the distance to a low sulfur coal source increases, the cost of complying with SO2

regulations goes up as well. This will lead to greater opposition from industry to stricter

environmental standards, making a legislator less likely to adopt the stricter regulations.

Do relatively poorer states react differently to changes in income levels when setting

an environmental agenda? Limited confirmation of the inverted-U hypothesis is provided

by the current analysis, as all but one of the specifications have coefficients with the

predicted signs, and half of these coefficients are significant. The average peak in this

relationship (after which further increases in income will result in stricter environmental

regulations) is above the maximum observation for median income, suggesting that the

potential for pollution rises faster than the demand for environmental quality at current

U.S. income levels.

Do states take advantage of favorable location and climate conditions by setting

stricter standards? This paper provides evidence that states set stricter standards when

geographic characteristics are beneficial to pollution control and weaker standards when

these same conditions increase the cost of compliance. The most important explanatory

variable controlling for natural variations across states is the coastal dummy, where the

probability of adopting stricter SO2 standards is 0.36 higher in coastal states.

Evidence also shows that higher temperature levels, which inflate the cost of

complying with strict SO2 regulations, are associated with less restrictive policies.








Finally, a state with a higher percentage of land belonging to Class I mandatory

Prevention of Significant Deterioration regions are more likely to set stricter standards, as

more stringent federal regulations already apply within the majority of the state.

In conclusion, the considerable degree of variance in states adopting stricter sulfur

dioxide air quality standards provides an excellent laboratory for the study of

comparative state environmental politics. The decision by a state to enact strict or weak

environmental standards appears to follow the Peltzman model, as the strength of

consumer and producer groups, as well as the natural differences in the cost of

compliance across states, all have some effect on the outcome of sulfur dioxide standards.

Notes

1. The Prevention of Significant Deterioration regulations for attainment regions are
grouped into three classes, which differ in the amount of polluting growth that is
allowed to occur in the area. The mandatory Class I regions include national
parks, wilderness areas, national memorial parks, and international parks. The
Class I areas allow only a minimal amount of deterioration and therefore provide
little room for industrial growth, while more development and some degradation
of the existing air quality is permitted in Class II areas. Class III areas are
afforded the greatest amount of polluting growth, and ambient air conditions in
these regions are allowed to degrade down to the NAAQS levels. Most PSD
regions, except for the mandatory Class I areas cited above, can be reclassified
with EPA approval. Nonattainment Area (NAA) provisions are much stricter than
the PSD regulations, as priority is given to improve degraded ambient air quality,
not just to maintain existing quality levels.














CHAPTER 5
THE POLITICS OF SPECIAL INTEREST VOTER SCORECARDS

Introduction

Recent media attention has sparked debate over the use of scorecards by special

interest groups to "expose" the voting records of political candidates. These voter guides

typically assign a percentage score to elected officials based on their voting records on a

set of pre-determined issues of importance to the interest group. The scores range from

zero to 100, with a score of 100 signifying perfect agreement between the desires of the

group and the politician's voting record.

The tax-exempt status of many of these special interest groups legally precludes their

involvement in partisan politics. The Federal Election Campaign Act requires groups

that actively participate in elections to register as political committees, subject to both

taxation and federal disclosure laws. In order for these "public interest" organizations to

maintain their tax exemption, involvement in campaigns and partisan politics cannot be

their primary objective. Under the FEC rules, voter guides are perfectly legal as long the

main purpose is to educate voters, and not to advocate on behalf of a particular political

party.

Although these groups defend the voting indices as neutral issue-based evaluations, an

examination of the votes included in the 1997, 1998, 2000, and 2001 scorecards for the

Christian Coalition (CC) and the League of Conservation Voters (LCV) provides

evidence to the contrary. The data suggest that both groups manipulate voting records in








order to display contrast between the two parties; i.e., partisan non-issue votes are

included that inflate the scores of the party that supports the group, while decreasing the

scores of the other party. In addition to loading the scorecards with non-issue votes, the

Christian Coalition further influences the final scores by deviating from the common

practice of counting missing votes (absences) as a vote against the organization. The

methodology applied by the Christian Coalition varies across years, but overwhelmingly

favors Republican Party members.

These two groups were chosen as the focus of the current analysis for a number of

reasons. First, scrutiny by the media and other sources has specifically targeted the

scorecards for the League of Conservation Voters and the Christian Coalition as

promoting disguised partisanship. [See, for example, Simpson (1996), Hunt (1996,

1998), Cathey (2002), and Strassel (2002)]. A preliminary content analysis of the votes

included in the scorecards appears to confirm these accusations.

Also, the separation of issue and non-issue votes for the Christian Coalition and the

League of Conservation Voters is facilitated by their emphasis on relatively narrow

interests, specifically that of religious values and the environment. Many other special

interest scorecards are not so narrowly defined, such as the AFL-CIO (pro-labor), the

Chamber of Commerce (pro-business), or the ACLU (pro-civil liberties). For these

indices, it is more difficult to determine whether the included votes are relevant or

peripheral to the goals of the organization.

Finally, a content analysis of other narrow issue groups such as the National

Organization of Women, Planned Parenthood, the National Right to Life Committee, and

the National Education Association, among others, did not provide similar evidence of








partisan bias. This would appear to suggest that the media has correctly identified the

scorecards that suffer from some partisan bias.

The sample of scorecards to be analyzed includes those for 1997, 1998, 2000, and

2001. Although, LCV scorecards from 1979 to present are publicly available online, the

Christian Coalition is not as amiable in providing archived data only the current

scorecard (2001) is available through their website. I was able to locate the additional

years (1997, 1998, and 2000) with the help of Americans United for Separation of

Church and State, a group opposing the Coalition platform. In order to facilitate a

comparison of the LCV and the Christian Coalition, I analyze only the four years for

which I have data from both organizations.

The results show that once these scorecards are revised to exclude non-issue votes and

to correct for methodological inconsistencies, legislators perform better on opposing

indices and worse on supporting indices. That is, Democrats fare better on Christian

Coalition scorecards and worse on LCV scorecards, and vice versa for Republicans.

Furthermore, the correlation between the scorecards and a simple liberal-conservative

index, such as that published by the ADA, decreases as peripheral votes are excluded, as

does the correlation between the scores and party affiliation. This diminished correlation

with party and liberalism offers empirical support for the hypothesis that the unadjusted

scorecards are reflecting some partisanship and not merely congressional voting in an

environmental or religious dimension.








Theory and Implications

Literature Review

The potential for bias within special interest scorecard ratings has been examined in

the political science literature, including research by Fowler (1982). In her analysis, she

focuses on the selection of issues that are rated by groups as the root cause of bias in the

indices, and specifically examines the selection process of several major interest groups

in order to test the accuracy and consistency of their scorecards. She concludes that

group emphasis on a specific set of salient votes has the tendency to bias the scorecards

toward only a few issues, presenting a polarized and misleading view of congressional

politics. A later work by Snyder (1992) reinforces this potential for bias by pointing out

that the special interest ratings emphasize close roll call votes (at the expense of other

less partisan issue votes), which leads to an exaggeration in the degree of extremism of

the rated legislators.

Another line of research addresses the variability of issue selection over time, and the

degree to which this affects intertemporal and interchamber comparisons of interest

group ratings. For example, the issues selected by the ADA to measure liberalism in

1980 are not the same types of issues selected for these purposes in 1990 or 2000. This

variability across time periods in the ADA selection process will make accurate and

unbiased comparisons of these scores impossible. A number of works have tried to

circumvent this problem by creating measures to eliminate the time-dependent selection

bias. [See, for example, Groseclose, Levitt, and Syder (1999); Shipan and Lowry

(2001)].








However, the current literature does not address either the incentive that special

interest groups have to manipulate these scores or the degree to which this manipulation

favors a certain political party. The main contribution of this paper is to extend the

literature by examining whether the inclusion of peripheral votes in issue-specific

scorecards can be shown to provide evidence supporting the hypothesis of interest group

bias. The incentive to misreport candidate behavior in favor of the group's preferred

party is examined within a standard voting model that addresses the decision by a group

member of whether or not to vote in the current election.

Theoretical Model

The theoretical model proposed here draws from a number of works in the economics

and political science literature, namely Riker and Ordeshook (1968), Filer and Kenny

(1980), Uhlaner (1989), and Morton (1991). An individual will vote as long as the

expected benefit received from doing so is greater than the costs associated with it.

Formally, an individual will vote only if:

(1) Ap*B+D>C

where Ap is the probability of affecting the outcome, B is the benefit to the voter if the

preferred official is elected rather than the opponent, D is a consumptive benefit or taste

for voting (a feeling of civic duty would be included here), and C is the utility cost of

voting. An increase in the perceived benefit from voting (LHS of equation) will draw

greater political activism. Similarly, a fall in the cost of voting will lead to a rise in

participation.

In addition to spurring donations and campaign assistance, voter scorecards are

designed to increase voter turnout in favor of preferred candidates. For example, by








prepackaging and identifying the candidates that best represent the group ideology, they

effectively decrease the need for information gathering on the part of the voter (lowering

the cost of voting). Following the model above, a decrease in the cost of acquiring

political information should lead to an increase in the voter turnout among group

members, since these are the voters most likely to receive and care about the scorecards.

This in turn benefits the group by increasing the probability that its preferred candidates

win.

Another means by which special interest groups attempt to achieve a favorable

outcome is through changes in the perceived benefit of voting. A leader can appeal to the

group on ideological grounds, raising B in equation (1) and increasing the likelihood

group members will vote for the supported candidate. For example, the LCV could

solicit electoral support from members through their published voter guides by painting a

picture of looming environmental catastrophe; similarly, the Christian Coalition could

appeal to church members on a moral basis. However the group targets its members,

appeals such as these create a sense of loyalty and urgency for the cause.

Also, as evidenced in the growing literature on group voting, the probability of

affecting the outcome (Ap) increases with group participation. As long as the group is

sufficiently large relative to the electorate, coordinated group action will raise the

probability of winning. Uhlaner pointed to the strength of the union vote in the 1982

election as supportive of the group model, and noted "leaders can use the group's

communications resources to mobilize members to vote" (392). Voting aids, such as

scorecards, are an integral part of this process.








Bayesian updating

It is also useful to analyze changes to the perceived benefit (or utility of voting) in a

Bayesian framework where voters continually update their beliefs with new information.

[See, for example, Husted, Kenny, and Morton (1995)]. This updated information is

generally associated with reductions in voter error in rating a candidate on ideological

grounds. In the context of this paper, the new information provided by the voter

scorecards is mixed with prior beliefs about a candidate's position, typically improving

the group member's ability to evaluate candidate behavior.

Assume that an LCV group member at election time has an expectation of a

Democratic candidate's policy position (LCVED) equal to a weighted average of her prior

beliefs (LCVPri'rD) and those reported in the scorecards (LCVportedD):

LCVED= a(LCVReprtedD) + (-ao)(LCVPrioro), where a>0O.

The member's final estimate on the position of candidates (and therefore the voting

decision) will depend not only on the prior and reported estimates, but also on the degree

to which she regards the special interest rating as reliable (a). As such, if the group

member places greater emphasis on the LCV ratings relative to her own prior knowledge

about a candidate, she will modify her estimate more.

The group member's voting decision hinges on the perceived stakes in the outcome,

and is represented by the difference in the expected utility she receives if the favored

candidate wins versus the utility she receives if the opposed candidate wins. For LCV

group members, the absolute difference in utility (B) can be represented by I LCVED-

LCVER I. A larger value of B reflects higher stakes in the outcome, and will increase the

likelihood that a group member will vote. Thus, information that widens the distance








between the two perceived platforms will lead to greater participation by group members

and increase the probability that the group-favored candidate wins the election.

Social welfare implications

Do these scorecards make society better off by reducing voter error? If the reported

LCV scores are a close approximation of the actual behavior of a candidate on

environmental issues (LCVReprted D= LCVAItualD), then voter expectation becomes closer

to the actual candidate position. Therefore, if the scorecards are truthful representations

of candidate behavior, the new information will reduce voter error on judging the actual

positions of candidates.

However, if the scorecard reports are not representative of actual candidate behavior

(LCVReportedD LCVActualD), then the situation is more complicated. Voter error may rise

only if the reported LCV score moves the voter in a direction opposite the true position or

if the updated estimate overstates the actual position of a candidate. For example, prior

LCV scores, actual LCV candidate positioning, and group reported LCV scores are

represented on the lines below. Case 1 shows that voter error will increase if the reported

LCV and the actual LCV scores are on opposite sides of a voter's prior beliefs. This will

cause the group member to update the estimate of a candidate's position in the wrong

direction, leading to a greater error in judgment.

CASE 1

LCVReported LCVprior LCVActual


However, in Case 2, voter error will decrease relative to the prior belief as long as the

updated estimate remains to the right of the actual candidate position. Only if the new

information moves a voter to the left of the actual candidate position will voter error








begin to rise. However, the new error may be smaller than the initial error, depending on

the magnitude of the position shift.

CASE 2

LCVReported LCVActual LCVprior

Dimensionality of issue-specific scorecards

A final theoretical consideration that will be addressed in this paper is whether the

special interest voter scorecards provide a multi-dimensional evaluation of political

behavior. It is assumed that elected officials can be easily rated along a single dimension

by liberal-conservative indices. Special interest groups argue that their scorecards

provide further information beyond these ideological ratings, i.e. that candidate behavior

and activity can also be measured on a multi-dimensional issue-based spectrum.

However, the inclusion of non-issue liberal-conservative votes narrows the

dimensionality of the different assessments of candidate behavior. Their similarity to

existing liberal-conservative indices increases and they provide less novel information to

the voter.

League of Conservation Voters

The primary mission of the League of Conservation Voters, founded in 1970, is to

represent the environmental movement by exposing the voting records of anti-

environmental candidates. In pursuit of their goal to create an environmentally conscious

political machine, the LCV publishes an annual scorecard that evaluates candidates on a

series of "environmental" roll call votes. Four recent LCV scorecards for the U.S.

Congress are examined to test whether the group unfairly favors Democrats.








In order to address the question of partisan bias, a revised index is constructed for

each environmental scorecard based on the results from both content and factor analyses.

If the scores are biased by the inclusion of peripheral votes, deleting them from the index

should have a positive effect on Republican scores and a negative effect on Democrat

scores. The extent to which these scores change should provide a measurement of the

degree to which the LCV scorecards are biased in favor of Democratic candidates.

Also, the correlation between the scores and other liberal-conservative indices such as

that produced by the Americans for Democratic Action (ADA) should decrease with the

deletion of superfluous liberal-conservative votes, as should the correlation between the

scores and party affiliation. This would support the hypothesis that a pure environmental

index provides information in addition to the already existing liberal-conservative

indices. The more partisan bias that exists within the LCV score, the more likely that it

will produce results similar to a one-dimensional ideological rating.

The 2001 LCV Votes

The fourteen congressional votes included in the 2001 LCV scorecard represent

votes from the first half of the 107'h Congress and a summary of each environmental

issue is provided in Table 5-1. This scorecard published votes on a wide range of

environmental topics such as energy efficiency, land conservation, and program budgets.

However, the index also rated congressmen on controversial non-environmental

initiatives such as abortion and trade. In an attempt to reduce the partisan bias in the

scorecard, the revised LCV index excludes these two non-issue votes from the total score

calculation.









Table 5-1: 2001 LCV Scorecard Votes
Bill Name Environmental Issue
1. Arctic Drilling 18/1/01 (Roll call vote #316, Amendment to Energy bill limiting the size
approved 228-201) NO is pro-LCV vote of drilling in the Arctic Refuge to 2,000
acres
2. Arctic Drilling II 8/1/01 (Roll call vote #317, Amendment to strike the Arctic drilling
rejected 206-223) YES is pro-LCV vote provision from the Energy bill
3. Hardrock Mining 6/21/01 (Roll call vote #182, Amendment to block efforts to weaken
approved 216-194) YES is pro-LCV vote newly issued environmental regulations for
the mining industry
4. Monuments Drilling 6/21/01 (Roll call vote Amendment to ban energy exploration on
#180, approved 242-173) YES is pro-LCV vote national monuments
5. Gulf Drilling 6/21/01 (Roll call vote #181, Amendment to delay oil and gas leasing off
approved 247-164) YES is pro-LCV vote the Florida coastline
6. Great Lakes Drilling 6/28/01 (Roll call vote Amendment to postpone new oil and gas
#203, approved 265-157) YES is pro-LCV vote drilling in the Great Lakes region
7. Farm Conservation 10/4/01 (Roll call vote #366, Amendment to Farm bill providing $5.4
rejected 200-226) YES is pro-LCV vote billion a year to land conservation programs
8. Arsenic 7/27/01 (Roll call vote #288, approved Amendment to EPA funding bill prohibiting
218-189) YES is pro-LCV vote the EPA from delaying or weakening the
new arsenic standard
9. EPA Enforcement 7/27/01 (Roll call vote #289, Amendment to restore EPA enforcement
rejected 182-214) YES is pro-LCV vote funding
10. Fuel Economy 8/1/01 (Roll call vote #311, Amendment to increase fuel economy
rejected 160-269) YES is pro-LCV vote standards for light trucks and SUV's
11. National Energy Policy 8/2/01 (Roll call vote House Energy bill which included key
#320, approved 240-189) NO is pro-LCV vote points from the Bush energy plan
12. Energy Efficiency 6/21/01 (Roll call vote #178, Amendment to increase funding for energy
rejected 153-262) YES is pro-LCV vote conservation programs
*13. International Family Planning 5/16/01 (Roll Amendment to remove language reversing
call vote #115, approved 218-210) NO is pro-LCV restrictions on funding foreign organizations
vote that provide abortion services
14. Fast Track Trade Authority 12/6/01 (Roll call Fast Track Authority bill granting the
vote #481, approved 215-214) NO is pro-LCV vote president the ability to directly negotiate
trade agreements
*Votes excluded from revised LCV scorecard (No factor analysis revised scorecard available)

Content analysis. Voting records regarding the use of public lands and resources

constitute 43% of the overall LCV score. The first of these issues include two

amendments to the energy bill opening the Artic refuge to oil and gas exploration. The

first amendment was a Republican proposal to implement a 2,000-acre limit on the area

open for development (vote 1). Environmentalists saw this proposal as deceptive, as

certain exemptions to the "limitation" would allow environmental damage equal to that of








the initial legislation. The second amendment was a bipartisan proposal to strike the

drilling provision entirely from the energy bill and continue the ban on Arctic Drilling

(vote 2).

A third vote concerns the attempt by the Interior Secretary to roll back newly imposed

environmental regulations on the mining industry (vote 3). Environmentalists supported

these updated standards as a significant improvement to the previous industry regulations,

which provided better clean up, allowed the Bureau of Land Management to deny permits

on the basis of potential environmental effects, and required mining companies to pay

cleanup costs.

Another LCV vote (vote 4) pertains to energy exploration at national monument sites.

This amendment to the 2002 Interior Appropriations bill proposed prohibiting the leasing

of any national monument land for energy exploration purposes, including the twenty-

two controversial new monuments created by the Clinton administration.

The fifth and sixth votes included in the congressional ranking concern oil and gas

leasing programs in the Gulf of Mexico (vote 5) and the Great Lakes region (vote 6).

Environmentalists fought hard to postpone drilling, which would irreparably damage

these coastal environments.

An amendment to the farm bill is included in the scorecard as well. This initiative

would have increased funding to $5.4 billion a year for a program offering financial

incentives to farmers that engage in land preservation efforts (vote 7).

The vote on an amendment to the EPA funding bill intended to safeguard stricter

arsenic standards is also included in the scorecard. The Bush administration hoped to

rescind the more stringent 10ppb standard instituted by the Clinton regime to the previous








standard set in 1945 of 50ppb (vote 8), arguing that the new rule was not based on sound

science.

A vote concerning the funding of EPA enforcement efforts is included in the LCV

rating. The proposed amendment to restore EPA enforcement funding (vote 9) was a

reaction to the Bush administration's efforts to cut the program by $25 million and

redistribute the money to state agencies in the form of grants. Environmentalists objected

to the decrease in funding, claiming that it would limit the ability of the EPA to oversee

important environmental laws.

Votes on the issues of energy use and global warming also appear on the scorecard.

The first of these was a failed attempt to increase fuel economy standards for light trucks

and SUV's. The proposed amendment to the energy bill (vote 10) would have closed the

"light truck loophole" by requiring these vehicles to match the current 27.5 miles per

gallon standard for regular cars by the year 2007.

Also making its way into the scorecard is President Bush's highly criticized national

energy policy (vote 11), which was seen as promoting fossil fuel development at the

expense of cleaner energy sources. The vote on H.R. 4 is included in the LCV scorecard,

since it contained key features of the Bush energy plan. Also, an amendment to the

Interior Appropriations bill that would have shifted funding from fossil fuel development

to energy conservation programs (vote 12) is included in the LCV score as well.

Promoting energy efficiency and renewable energy sources is at the heart of the debate on

global warming.

A content analysis of the 2001 LCV index suggests that the last two votes on Family

Planning and Fast Track Authority should be excluded from the revised LCV scorecard.








Not only are these two votes largely peripheral to the environmentalist platform, but their

polarization along party lines (nearly 90% of Republicans and Democrats voted with their

party on these issues) will exacerbate liberal bias in the final scoring. For these reasons,

their exclusion is necessary in order to get an accurate view of legislator responsiveness

to environmental concerns.

The first of these excluded votes concerns a motion to strike an amendment

overturning the Bush administration's restrictions on international family planning

organizations (vote 13). The policy banning the use of U.S. funds to support foreign

organizations that provided or supported legal abortion services prompted a hot partisan

debate, pitting conservatives and liberals against each other over the issue of abortion

rights. Although the environmental effects of overpopulation are a serious concern, the

major focus of this piece of legislation was abortion rights. Therefore, this vote is

considered peripheral to the environmental cause and should be excluded from a non-

partisan environmental rating.

Another quasi-environmental vote that will be deleted is presidential Fast Track

Authority (vote 14). The Fast Track Authority bill enabled the President to directly

negotiate trade agreements without amendment by Congress, which was allowed only an

up-down vote on the agreement. Environmentalists felt that such broad authority did not

provide adequate environmental safeguards. Although certain aspects of this vote were

seen as disagreeable to environmentalists, voting procedures are not central to the

environmentalist platform, and this vote is therefore excluded from the revised scorecard.

These last two votes were subjectively chosen as inconsistent with the major objectives








of environmentalism, and excluded from the index to provide a more accurate

examination of legislator policy towards environmental issues.

Factor Analysis. A more systematic method of eliminating unrelated scores, such as

factor analysis, is also helpful in testing for partisanship. The results from a factor

analysis of the 2001 LCV scorecard votes do not separate out into recognizable different

dimensions (such as environmental versus ideological). Unfortunately, this lack of

consistency between the factor analysis and the subjective exclusions described above

will make conclusions regarding the liberal bias of the 2001 LCV scores more difficult,

as there is no rigorous method applied to exclude the non-issue votes.

Therefore, the revised 2001 LCV scorecard consists of the core environmental issues

(one through twelve), which include land management and conservation, environmental

standards and EPA funding, and energy and global warming issues. The omitted votes on

international family planning and trade policy are largely non-environmental party-line

issues. If the addition of peripheral votes creates bias within the index, then the revised

scores should provide a better view of a candidate's environmental agenda.

Other LCV Scorecards

A summary of the votes included in the 1997, 1998, and 2000 LCV scorecards is

provided in Tables 5-2 through 5-4. The 1998 scorecard is not revised to reduce partisan

bias, as both content and factor analyses of these issues suggest that all of the included

roll call votes are primary environmental issues. However, this is not the case with the









Table 5-2: 1997 LCV Scorecard Votes
Bill Name Environmental Issue
1. Endangered Species Act Flood Waivers 5/7/97 Amendment to narrow the proposed
(Roll call vote #108, approved 227-196) YES is pro- exemption to the Endangered Species Act
LCV vote for flood damage relief purposes
2. Logging Roads Subsidies 7/10/97 (Roll call vote Amendment to reduce proposed cuts in
#262, approved 211-209) NO is pro-LCVvote Forest Service subsidies to timber
companies for new logging roads
3. Property Rights 10/22/97 (Roll call vote #519, Bill to weaken existing regulations
approved 248-178) NO is the pro-LCV vote regarding land use protections by allowing
developers to sue in federal court
4. Grazing I 10/30/97 (Roll call vote #549, "Forage Improvement Act" bill to revise
approved 242-182) NO is the pro-LCV vote federal grazing policies, including the
increase of grazing fees
5. Grazing II 10/30/97 (Roll call vote #546, rejected Amendment that would increase grazing
205-219) YES is pro-LCV vote fees on federal lands to equal the
appropriate state grazing fee
6. National Wildlife Refuges 9/23/97 (Roll call vote Bill to establish fish and wildlife
#424, approved 419-1) YES is pro-LCV vote conservation as the basic mission for all
national wildlife refuges
7. National Monuments 10/7/97 (Roll call vote Bill to weaken presidential authority over
#495, approved 229-197) NO is pro-LCV vote the designation of natural monument sites
8 World Heritage Sites and Biosphere Reserves Bill to restrict U.S. participation in
10/8/97 (Roll call vote #504, approved 236-191) NO UNESCO World Heritage and Biosphere
is pro-LCV vote programs
9. Sugar Subsidy 7/24/97 (Roll call vote #312, Amendment to restrict USDA loans to sugar
rejected 175-253) YES is pro-LCV vote producers
10. Animas-La Plata Irrigation Project 7/25/97 Substitute amendment to limit funding of
(Roll call vote #328, approved 223-201) NO is pro- the Animas-La Plata Irrigation project under
LCV vote certain criteria
11. Clean Coal Technology Program 7/11/97 (Roll Amendment to cut $292 million in funding
call vote #264, rejected 173-243) YES is pro-LCV from the "clean coal" program
vote
12. Texas Low-Level Radioactive Waste Disposal Bill to approve the transport and disposal of
Compact 10/7/97 (Roll call vote #497, approved low-level radioactive wastes from Vermont
309-107) NO is pro-LCV vote and Maine to a facility in west Texas
13. Nevada Nuclear Waste Dump 10/30/97 (Roll Bill to allow an interim nuclear waste dump
call vote #557, approved 307-120) NO is pro-LCV to be situated near the proposed permanent
vote repository at Yucca Mountain
14. Air Quality Standards 7/97 (197 sponsors) NO Sponsorship of a bill to roll back new EPA
is pro-LCV vote standards for ozone and particulate matter
*15. International Family Planning 1 2/13/97 (Roll Resolution to release blocked funding to
call vote #22, approved 220-209) YES is pro-LCV international family planning organizations
vote
*16. International Family Planning II 9/4/97 (Roll Substitute amendment that would
call vote #326, rejected 210-218) YES is pro-LCV distinguish between international family
vote planning organizations that use funds to
prevent or promote abortion
*Votes excluded from revised LCV scorecard (identical to factor analysis revised scorecard)









Table 5-3: 1998 LCV Scorecard Votes
Bill Name Environmental Issue
1. Land Use Protections 3/12/98 (Roll call vote #52, Bill to allow polluters to challenge long-
approved 230-180) NO is pro-LCV vote settled federal environmental safeguards in
appellate courts
2. Logging in National Forests 3/27/98 (Roll call Bill to allow the Forest Service to increase
vote #80, rejected 181-201) NO is pro-LCV vote commercial logging within national forests
for "recovery" purposes
3. Roadless Areas in Forests 3/27/98 (Roll call vote An amendment to exempt roadless areas of
#79, approved 200-187) YES is pro-LCV vote national forests from development and
"recovery" projects
4. Alaska Logging Roads 7/23/98 (Roll call vote Amendment to prohibit the use of funds to
#329, rejected 186-237) YES is pro-LCV vote construct new roads in the Tongass National
Forest
5. Alaska Wildlife Area Road 8/18/98 (Roll call Amendment to prevent easement for a
vote #, rejected 176-249) YES is pro-LCV vote commercial road through the Chugach
National Forest
6. Gulf of Mexico Fisheries Management 8/5/98 Substitute amendment to grant state
(Roll call vote #395, rejected 141-283) NO is pro- authority over Gulf fishing waters within
LCV vote three to nine miles from shore, nullifying
federal bycatch standards for these areas
7. Anti-Environment Riders 1 5/19/98 (Roll call Amendment to create a new point of order
vote #157, rejected 190-221) YES is pro-LCV vote against bills that weaken or roll back
environmental regulations
8. Anti-Environment Riders II 7/23/98 (Roll call Amendment to override all anti-environment
vote #334, rejected 176-243) YES is pro-LCV vote riders attached to EPA spending bill
9. Health and Safety Protections 5/19/98 (Roll call Bill to establish new point of order against
vote #160, approved 279-132) NO is pro-LCV vote environmental legislation imposing private
sector costs of more than $100 million
10. Energy Efficiency Program Funding 7/21/98 Amendment to reduce funding for energy
(Roll call vote #313, rejected 212-213) YES is pro- efficiency programs by $25 million
LCV vote
11. Global Warming Gag Rule 7/23/98 (Roll call Amendment to override language
vote #332, approved 226-198) YES is pro-LCV vote prohibiting educational activities regarding
global warming before Kyoto treaty was
approved by the Senate
12. Environmental Reporting and Information Bill to waive civil penalties for first-time
3/26/98 (Roll call vote #74, approved 267-140) NO violations of reporting requirements
is pro-LCV vote mandated by certain environmental
regulations
13. Tropical Forest Conservation 3/19/98 (Roll call Bill authorizing $325 million over three
vote #63, approved 356-61) YES is pro-LCV vote years to restructure debt in developing
countries in exchange for land conservation
efforts
*No revised LCV scorecard for 1998









Table 5-4: 2000 LCV Scorecard Votes
Bill Name Environmental Issue
1. Land Conservation Funding 5/11/00 (Roll call Bill that would permanently fund the Land
vote #179, approved 315-102) YES is pro-LCV vote and Water Conservation Fund
2. National Monuments 6/15/00 (Roll call vote Substitute amendment to maintain language
#280, rejected 187-234) NO is pro-LCV vote prohibiting the use of funds for national
monuments created since 1999
3. Utah Wilderness 6/7/00 (Roll call vote #240, Substitute amendment that would authorize
rejected 210-214) NO is pro-LCV vote the Bureau of Land Management to decide
whether off-road vehicles would be allowed
on certain Utah wilderness lands
4. Columbia Basin Land Management 6/15/00 (Roll Substitute amendment to maintain language
call vote #279, rejected 206-221) NO is pro-LCV requiring that the Colombia Basin plan not
vote adversely impact small businesses
5. Timber Sale Subsidies 6/14/00 (Roll call vote Amendment to divert funds from the
#277, rejected 173-249) YES is pro-LCV vote subsidization of timber sales to fish and
wildlife protection programs
6. Wild Predator Control 7/11/00 (Roll call vote Amendment to prevent federal funding of
#382, rejected 190-228) YES is pro-LCV vote lethal predator control programs
7. Clean Water 6/21/00 (Roll call vote #304, Amendment to remove provisions from a
rejected 208-216) YES is pro-LCV vote spending bill that would prohibit the EPA
from enforcing the current arsenic standard
8. Air Right to Know 6/21/00 (Roll call 305, Amendment to prohibit the EPA from
approved 225-199) NO is pro-LCV vote identifying areas that failed to meet a newly
developed ozone standard
9. Superfund Exemption 9/26/00 (Roll call vote Bill to lessen small businesses liability for
#494, approved 253-161) NO is pro-LCV vote toxic wastes and Superfund sites
10. Nuclear Waste 3/22/00 (Roll call vote #63, Bill to allow transport of nuclear waste to
approved 253-167) NO is pro-LCV vote Yucca Mountain before completion of the
permanent facility
11. Delaware River Dredging 6/27/00 (Roll call vote Amendment to restrict funding for the
#338, rejected 176-249) YES is pro-LCV vote Delaware River dredging project
12. Property Rights 3/16/00 (Roll call vote #55, Bill to allow developers the right to sue
approved 226-182) NO is pro-LCV vote directly in federal court, bypassing local
planning officials and land use procedures
13. Global Climate Change 6/26/00 (Roll call vote Amendment to approve funding of already
#323, approved 217-181) YES is pro-LCV vote exiting global warming programs
14. International Family Planning 7/13/00 (Roll Motion to strike restrictions on funding of
call vote #396, rejected 206-221) YES is pro-LCV international family planning organizations
vote that provide abortion services
*Vote excluded from revised LCV scorecard (No factor analysis revised scorecard available)

1997 and 2000 scorecards, which both include at least one vote on international family

planning, discussed earlier in detail. A revised scorecard is constructed for both years to

exclude this peripheral issue (two deletions out of 16 for the 1997 scorecard and one


exclusion out of 14 for the 2000 scorecard).









A factor analysis of the roll call votes from these two years was also conducted in

order to more rigorously test for a separate non-environmental dimension. A factor

analysis of the 1997 roll call votes supports the deletion of family planning from the

scorecard when three factors are defined (these results are available below in Table 5-5).

However, a factor analysis of the votes included in the 2000 scorecard does not provide

similar confirmation of the subjective revisions.

Table 5-5: 1997 LCV Votes-VARIMAX Rotated Common Factor (3)
Variables Factor 1 Factor 2 Factor 3 Uniqueness
Flood waivers 0.77235 0.13809 0.33376 0.27301
Logging Roads 0.63516 0.35295 0.34939 0.34993
Property Rights 0.73928 0.13472 0.37473 0.29489
Grazing I 0.81589 0.10876 0.28077 0.24367
Grazing II 0.72485 0.32207 0.24748 0.30962
National Wildlife Refuges 0.00158 -0.18441 0.08198 0.95927
National Monuments 0.84014 -0.02694 0.32120 0.19027
World Heritage and Biosphere Reserves 0.77546 -0.05375 0.42770 0.21285
Sugar Subsidy 0.08156 0.51586 0.06395 0.72315
Irrigation Project 0.39286 0.43102 0.19584 0.62153
Clean Coal Technology 0.05504 0.48832 0.10232 0.74804
Texas Waste Disposal 0.44346 0.19188 0.14619 0.74516
Air Quality 0.45012 0.30552 0.37599 0.56268
International Family Planning I 0.41986 0.05571 0.85063 0.09703
International Family Planning II 0.40570 0.07414 0.85691 0.09561

Therefore, the LCV scores for 1997, 2000, and 2001 are revised to exclude partisan

bias by dropping the non-environmental votes on family planning and fast track based on

a content analysis of the indices. A factor analysis confirms the deletion of the two

family planning votes in the 1997 scorecard, but does not offer similar evidence for the

2001 or 2000 scorecard years.

Christian Coalition

"How would Jesus vote?" This is presumably the question that the Christian

Coalition, founded in 1989 by Robert Reed and evangelist Pat Robertson, seeks to answer

with its guides. Although the self-purported goal of the Coalition is to support candidates








with a "moral" or "pro-family" agenda, it is not difficult to see that this often corresponds

to the Republican agenda. In fact, the group has openly supported Republican platforms

and initiatives, and has defended charges from the Federal Election Commission for

violating numerous election laws in support of the GOP. In order to account for potential

Republican bias in the Christian Coalition voter guides, a series of revised scorecards are

constructed that exclude peripheral non-issue votes such as taxation and campaign

finance reform.

The 2001 CC Votes

The twelve congressional votes included in the 2001 Christian Coalition scorecard

represent votes from the 107th Congress, and a summary of each is provided in Table 5-6.

Two revised indices are constructed for this scorecard in order to correct for partisan bias

in the Christian Coalition scoring. The first index is based on a subjective content

analysis of the votes included in the CC scorecard, while the second utilizes factor

analysis to separate the votes into distinguishable categories.

Content analysis. Abortion and other related issues are a topic of great importance to

the religious right and are represented four times in the scorecard (votes 1-4), 33% of the

overall score. The first of these votes is an amendment that removes language reversing

President Bush's restrictions on funding international family planning organizations that

provide abortion services, counseling or advocacy (also 2001 LCV vote 13). A second

bill made it a criminal offense to transport a minor over state lines without parental

consent in order to obtain an abortion. The last two votes deal specifically with the legal

status of the unborn fetus, making it a criminal offense to harm a fetus during the









commission of a violent crime, or to engage in human cloning experiments for any

reason.

Table 5-6: 2001 Christian Coalition Scorecard Votes
Bill Name Moral Issue
1. Human Cloning 7/31/01 (Roll call vote Amendment to Title 18 of the U.S. Code
#304, approved 265-162) YES is pro-CC vote prohibiting human cloning
2. Abortion Restrictions 4/17/02 (Roll call Amendment to Title 18 of the U.S. Code
vote # 97, approved 260-161) YES is pro-CC prohibiting the transport of minors across state
vote lines to circumvent laws requiring the
involvement of parents in abortion decisions.
3. International Family Planning 5/16/01 (Roll Amendment to remove language reversing
call vote # 115, approved 218-210) YES is pro- restrictions on funding foreign organizations
CC vote that provide abortion services
4. Unborn Victims of Violence 4/26/01 (Roll Amendment to Title 18 of the U.S. Code and
call vote #89, approved 252-172) YES is pro- the Uniform Code of Military Justice
CC vote protecting the unborn fetus from assault and
murder
5. Domestic Partners Benefits 9/25/01 (Roll Amendment to prohibit the funding of the
call vote #352, rejected 194-226) YES is pro- District of Columbia Domestic Partnership Act
CC vote
6. School Vouchers 5/23/01 (Roll call vote Amendment to Education bill providing federal
#135, rejected 155-273) YES is pro-CC vote funding for certain students to attend private
(including religious) schools
7. Faith-Based Community Solutions 7/19/01 Bill providing incentives for charitable
(Roll call vote #254, approved 233-198) YES is contributions and extending federal funding to
pro-CC vote faith-based community organizations
8. Marriage Penalty and Family Tax Relief Amendment to the Internal Revenue Code
3/29/01 (Roll call vote #75, approved 282-144) reducing taxes for married couples and
YES is pro-CC vote increasing tax credits for children
*9. Income Tax Reduction 3/8/01 (Roll call Amendment to the Internal Revenue Code
vote #45, approved 230-198) YES is pro-CC reducing individual income tax rates
vote
10. Death Tax Elimination 4/4/01 (Roll call Amendment to the Internal Revenue Code
vote #84, approved 274-154) YES is pro-CC phasing out estate and gift taxes
vote
*11. Campaign Finance Reform 12/13/02 Amendment to Campaign Finance Bill
(Roll call vote #22, rejected 188-237) YES is providing First Amendment protection,
pro-CC vote including the right to free speech
* 12. Campaign Finance Reform II 2/14/02 Bill banning all soft money contributions and
(Roll call vote #34, approved 240-189) NO is imposing restrictions on issue advocacy
pro-CC vote communications
*Votes excluded from revised CC scorecard (additional votes 6-8 excluded from factor analysis revised
scorecard)

One education initiative (vote 6) was included in the scorecard tally dealing with the

issue of school vouchers. This failed amendment to an education bill would have








provided federal funding for certain students to attend private (including religious)

schools of their choice and was widely popular with religious organizations nationwide.

Social policy supported by the Christian Coalition surfaces twice in the scorecard,

making up 17% of the overall score (votes 5 and 7). The first of these includes an

amendment that was rejected seeking to prohibit funding of the District of Colombia

Domestic Partnership Act, which would extend health benefits to the unmarried domestic

partners of Washington DC employees. The second concerns a vote promoting religious

organizations, allowing them to compete equally with other non-governmental groups for

federal funds in providing social services. It also aids fundraising efforts by providing

$13.3 billion in tax breaks for charitable giving over 10 years.

Taxes are represented almost as much as abortion in the scorecard, making up 25% of

the total score (votes 8-10). A tax issue that is somewhat related to the Christian cause is

a vote to reduce the marriage penalty. The bill allows married couples to claim a

standard deduction twice that of single filers and it increases the threshold for low-

income couples qualifying for the earned-income tax credit. The law also doubles the tax

credit for children younger than 17 to $1,000. The religious right supported this law

because it offers financial incentives in favor of "traditional" family values.

However, two non-issue tax votes are also included in the scorecard namely, income

and death taxes. These proposed amendments to the IRS Code, the first reducing

individual income tax rates and the latter phasing out estate and gift taxes, share little

commonality with the moral issues generally supported by the religious right. Therefore,

while the marriage and family tax bill is retained in the revised index, the income and

death tax votes are omitted as peripheral partisan issues.








The last issue targeted by the scorecard is that of campaign finance reform (votes 11

and 12). The final vote on this legislation (vote 11) bans soft money contributions to

national political parties, while permitting up to $10,000 in soft money contributions to

state and local parties to aid voter participation drives. The legislation also stops issue ads

from targeting specific candidates within 30 days of the primary or 60 days of the general

election. Vote 12 is a proposed amendment to the campaign law that would guarantee

First Amendment protection, such as the right to free speech, under the new legislation.

The Christian Coalition has come out against reforming the electoral process, as it will

limit their ability to support preferred candidates. However, while this issue makes up

17% of the overall score, it does not provide any information regarding the candidate's

views on traditional family values or morality. This vote is instead a partisan issue

concerning the political power of well-funded organizations.

In conclusion, based on a subjective analysis of the CC scorecard content, a revised

index for the Christian Coalition is constructed without the votes concerning Campaign

Finance Reform (11 and 12) and the Income and Death Tax (9 and 10). This leaves the

index with eight votes, including the issues of cloning and abortion, education, social

policy, marriage incentives, and support for religious organizations. Each vote now

constitutes nearly 13% of the total score, instead of the 8% provided by the previous

index. The scorecard is re-constructed in this way in an attempt to eliminate the partisan

bias that exists within the original index.

Factor analysis. A factor analysis of the Christian Coalition votes is utilized in order

to apply more systematic criteria for eliminating unrelated votes. The results provided in

Table 5-7 separate the Coalition votes into two distinct dimensions money and social









issues. Specifically, factor one includes all abortion issues along with the gay rights

initiative, while the second factor includes all the tax and campaign finance reforms, as

well as school vouchers and funding for religious organizations.

Table 5-7: 2001 Christian Coalition Votes-VARIMAX Rotated Common Factor (2)
Variables Factor 1 Factor 2 Uniqueness
Human Cloning Ban 0.79569 0.37721 0.22458
Abortion Restrictions 0.79818 0.39015 0.21069
Foreign Aid/Abortion 0.77217 0.44667 0.20424
Unborn Victims of Violence 0.82125 0.38493 0.17738
Domestic Partners Benefits 0.62709 0.56812 0.28400
School Vouchers 0.39290 0.64720 0.42676
Funding for Religious Organizations 0.46140 0.79597 0.15354
Marriage Tax 0.39262 0.68176 0.38105
Income Tax 0.41042 0.83721 0.13063
Death Tax 0.38986 0.70460 0.35155
Campaign Finance Reform 0.47793 0.68662 0.30013
Campaign Finance Reform 0.47237 0.74374 0.22373

The division of the Christian Coalition votes by factor analysis goes beyond the

subjective method described above, as the earlier revision includes the fiscal initiatives

that can be reasonably linked to Christian values. Since the factor analysis provides a

more rigorous determination of the extent of partisan bias in the Christian Coalition

scores, both revised indices will be employed to test for Republican favor by the

organization. More specifically, factor analysis suggests deleting all but the abortion and

gay rights issues (votes 1-5). The corresponding qualitative assessment of the scorecard

is less restrictive as it also includes the votes regarding the marriage tax, school vouchers,

and religious agency funding (votes 5-8).

Christian Coalition Methodology

A final issue of concern regarding the Christian Coalition voter guides is the

methodology applied in calculating the scores. The 2001 index used a consistent

technique to report votes for and against the designated platform, i.e., a positive score for








a favorable vote and a negative score for an unfavorable vote. However, when a

congressman was absent (and therefore did not vote on this issue) varying methods were

employed in calculating that missing vote. The common methodology for treating

missing votes is to count them negatively in the overall score, as is done in the ADA and

LCV indices.

For example, Representative Cubin (R) from Wyoming voted on only eight of the

twelve issues (positive on all), but received a 100% score from the Christian Coalition.

Representative Traficant (D) from Ohio voted on nine out of twelve votes (positive on all

but one), but received a 67% on the CC rating. In the case of Representative Cubin, the

missing votes either counted positively or were excluded from the scoring, whereas with

Representative Traficant, his absence most certainly counted against him. If his score

had been calculated in the same manner as his Republican colleague, he would have

received either an 89% or a 92% (depending on whether you dropped the missing votes

or counted them as positives).

Table 5-8 provides a look at the scores for congressmen who were absent for at least

one of the 2001 Christian Coalition votes. Three different possible reporting methods are

provided for calculating the score: the missing vote is calculated as a vote against the

platform (Neg Absent Vote), calculated as a vote in favor of the platform (Pos Absent

Vote), or excluded from the final score calculation (Exclude Absent Vote). The scores

under those procedures are listed along with the actual score given to the representative

by the Christian Coalition.










Table 5-8: Christian Coalition Methodology in Reporting Absent Votes in 2001
Party Neg Absent Vote Pos Absent Vote Exclude Absent Vote CC score
Stark (CA) D 0 8 0 0
Becerra (CA) D 0 8 0 0
Roybal-Allard (CA) D 0 8 0 0
Meek (FL) D 0 8 0 0
Visclosky (IN) D 0 8 0 0
Rothman (NJ) D 0 8 0 0
Ackerman (NY) D 0 8 0 0
Meeks (NY) D 0 8 0 0
Owens (NY) D 0 8 0 0
Velazquez (NY) D 0 8 0 0
Serrano (NY) D 0 8 0 0
Kennedy (RI) D 0 8 0 0
Baldwin (WI) D 0 8 0 0
Hastings (FL) D 0 17 0 0
Rush (IL) D 0 17 0 0
McKinney (GA) D 8 17 9 8
Dingell (MI) D 8 17 9 8
Towns (NY) D 8 17 9 8
Rangel (NY) D 8 17 9 8
Engel (NY) D 8 17 9 8
Lampson (TX) D 8 17 9 8
Clybum (SC) D 17 25 18 17
Tanner (TN) D 42 50 45 42
John (LA) D 58 67 64 58
Peterson (MN) D 58 67 64 58
Skelton (MO) D 58 67 64 58
Clement (TN) D 58 67 64 58
Traficant (OH) D 67 92 89 67
Lipinski (IL) D 75 83 82 75
Shows (MS) D 83 92 91 83
Leach (IA) R 42 50 45 42
Forbes (VA)* R 50 83 75 100
Roukema R 58 67 64 58
LaTourette (OH) R 67 75 73 67
Ros-Lehtinen (FL)* R 67 83 80 58
Cubin (WY) R 67 100 100 100
Bereuter (NE) R 75 83 82 75
Wilson (SC)* R 81 100 100 75
Smith (NJ) R 83 92 91 83
Dunn (WA) R 83 92 91 92
Riley (AL) R 83 100 100 100
Brady (TX) R 83 100 100 100
Hefley (CO) R 92 100 100 100
Latham (IA) R 92 100 100 100
Cooksey (LA) R 92 100 100 100









Table 5-8 contd.: Christian Coalition Methodology in Reporting Absent Votes in 2001
Party Neg Absent Vote Pos Absent Vote Exclude Absent Vote CC score
Ballenger (NC) R 92 100 100 100
Thornberry (TX) R 92 100 100 100
*None of the three possible scoring methods correspond to score assigned by the Christian Coalition
Definitions: Neg Absent Vote: CC score if absentee votes are calculated as negative votes
Pos Absent Vote: CC score if absentee votes are calculated as positive votes
Exclude Absent Vote: CC score if absentee votes are excluded from the score calculation
CC Score: Actual score reported by the Christian Coalition

Table 5-9 shows these results separated out by party. If the absent congressman was a

Democrat, the missing value was always counted as a negative vote.3 The scores of

thirty Democratic congressmen were affected in this way. In contrast, when the absent

official was a Republican, which occurred in the case of fourteen congressmen, the

missing votes were counted as a positive vote 64% of the time14 (a vote in favor of the

Christian Coalition). Deviation from the common methodology of counting missing

votes as a negative resulted in a net gain of 116 points for 9 Republicans, an average of

13 points per representative, while Democrat scores were unaffected. 5

Table 5-9: Absentee calculations by the Christian Coalition in 2001

Counted as a positive vote Counted as a negative vote
Democrat 0 30
Republican 9 5

The revised Christian Coalition indices take into account this methodological

inconsistency by re-counting each absence as a negative vote, regardless of party

affiliation. These methodological changes, along with the exclusion of partisan non-issue




13 If the Democrat had a prior score of zero, it is possible that the missing vote was actually omitted from
the final score calculation.
4 If the Republican had a prior score of 100, it is possible that the missing vote was actually omitted from
the final score calculation.

" The results exclude three Republican representatives whose scores are miscalculated by the Christian
Coalition.








votes from the final score (by content and factor analyses), should provide a more

accurate measure of a representative's standing on "moral" or "Christian" issues.

Other Christian Coalition Scorecards

A similar examination by content and factor analyses is performed on earlier

scorecards to revise them for partisan bias as well. A description of the 1997 votes is

provided in Table 5-10. For this CC scorecard, a content analysis suggests the omission

of two votes regarding taxes and term limits (votes 8-9). A factor analysis of the 1997

votes provides confirmation of these findings when three factors are defined, the two

peripheral votes load onto a single factor (see Table 5-11). Therefore, only one revised

index is constructed for 1997 that represents both the content and factor analyses.

A content analysis of the 1998 scorecard (Table 5-12) suggests only the deletion of a

tax limitation amendment (vote 12) as peripheral to Christian values. However, a factor

analysis, the results from which are available in Table 5-13, further refines the 1998

scorecard into social and fiscal issues, similar to that provided by the 2001 scorecard

analyses. This revision maintains only the four abortion votes and the needle exchange

program as central issues, while omitting taxation and funding for education, the arts, and

legal services for the poor (votes 6-11).

The revision of the 2000 Christian Coalition scorecard by content analysis (Table 5-

14) excludes five non-issue votes, including four tax and campaign finance reform

initiatives (votes 12-15). Also excluded is a vote concerning a national missile defense

system (vote 11 oddly enough, the Christian Coalition supported missile proliferation).

A factor analysis of these same votes does not provide independent confirmation of the

subjective omissions, as the votes for this scorecard do not load well onto distinguishable









factors. For this reason, only one revised index is constructed for the 2000 scorecard that

separates the votes by content analysis.

Table 5-10: 1997 Christian Coalition Scorecard Votes
Bill Name Moral Issue
1. Partial Birth Abortions 3/20/97 (Roll call vote Bill to prohibit abortion of a fetus as it is
#65, approved 295-136) YES is pro-CC vote coming through the birth canal
2. International Family Planning 2/13/97 (Roll call Resolution to send additional foreign aid to
vote #22, approved 220-209) NO is pro-CC vote overseas organizations that promote or
perform abortions
4. Abortions in Military Hospitals 6/19/97 (Roll call Amendment to repeal the current law which
vote #217, rejected 196-224) NO is pro-CC vote prohibits U.S. military medical facilities
from performing abortions
3. Ten Commandments Display 3/5/97 (Roll call Motion to express congressional support for
vote #31, approved 295-125) YES is pro-CC vote public display of the Ten Commandments in
government buildings
5. Violent Juvenile Crime 5/8/97 (Roll call vote Legislation authorizing $1.5 billion in
#118, approved 286-132) YES is pro-CC vote federal bonuses for states and local
authorities to fight juvenile crime
6. National Endowment for the Arts Funding Resolution allowing for the elimination of
7/10/97 (Roll call vote #259, approved 217-216) taxpayer funding for the National
YES is pro-CC vote Endowment for the Arts
7. Revoking Most Favored Nation status for China Resolution disapproving renewal of Most
6/24/97 (Roll call vote #231, rejected 173-259) YES Favored Nation (MFN) status to China
is pro-CC vote
*8. Term Limits for Congress 2/12/97 (Roll call Joint resolution to impose a 12-year lifetime
vote #21, rejected 217-211) YES is pro-CC vote limit on congressional service in both the
House and the Senate
*9. Tax Limitation 4/15/97 (Roll call vote # 78, Constitutional amendment which would
rejected 233-190) YES is pro-CC vote require a two-thirds majority vote in both
the House and the Senate in order to raise
taxes
*Votes excluded from revised CC scorecard (identical to factor analysis revised scorecard)

Table 5-11: 1997 Christian Coalition Votes-VARIMAX Rotated Common Factor (3)
Variables Factor 1 Factor 2 Factor 3 Uniqueness
Partial Birth Abortions 0.57805 -0.26534 -0.59490 0.24155
International Family Planning 0.84780 -0.26653 -0.17763 0.17863
Abortions in Military Hospitals 0.84683 -0.21716 -0.26391 0.16607
Ten Commandments Display 0.49005 -0.33683 -0.57019 0.32128
Violent Juvenile Crime 0.32938 -0.47480 -0.48805 0.42788
NEA Funding 0.60637 -0.55425 -0.22739 0.27342
MFN Status for China 0.10585 0.08795 0.22311 0.93128
Term Limits 0.35810 -0.49660 -0.33191 0.51500
Tax Breaks 0.53448 -0.63350 -0.27983 0.23471









Table 5-12: 1998 Christian Coalition Scorecard Votes
Bill Name Moral Issue
1. Partial Birth Abortion 10/8/97 (Roll call vote Motion to agree to the Senate language to
#500, approved 295-133) YES is pro-CC vote prohibit the abortion of a fetus as it is
coming through the birth canal
2. International Family Planning 9/4/97 (Roll call Amendment to allow organizations that
vote #362, rejected 210-218) NO is pro-CC vote promote or perform abortions to remain
eligible for U.S. international family
planning funds
3. Parental Notification for Title X Family Planning Substitute amendment denying parents the
Clinics 9/9/97 (Roll call vote #378, approved 220- right to be notified when minor children
201) NO is pro-CC vote were provided contraception and abortion
referrals through federal Title X family
planning clinics
4. Abortions in Military Hospitals 6/19/97 (Roll call Amendment to repeal the current law which
vote #217, rejected 196-224) NO is pro-CC vote prohibits U.S. military medical facilities
from performing abortions
5. Needle Exchange Programs 9/11/97 (Roll call Amendment to prohibit the use of federal
vote #391, approved 266-158) YES is pro-CC vote taxpayer funds to carry out or promote any
program that distributes needles for illegal
drug use
6. Opportunity Scholarships for D.C. Students FY 98 D.C. Appropriations bill which
10/9/97 (Roll call vote #513, approved 203-202) included a scholarship program allowing
YES is pro-CC vote 2000 eligible low-income students to attend
alternative public, private, or parochial
schools
7. H.E.L.P. Scholarships 11/4/97 (Roll call vote Bill which would allow states to use federal
#569, rejected 191-228) YES is pro-CC vote education funds to provide scholarships to
low-income families to send their children
to a school of their choice
8. Education Savings Accounts (IRAs) 10/23/97 Bill which would allow tax breaks for
(Roll call vote #524, approved 230-198) YES is pro- parents who save money for education
CC vote expenses (K-12 for public, private, or home
school)
9. Prohibit Funding of Federal Tests 2/5/98 (Roll Bill prohibiting the use of taxpayer funds for
call vote #9, approved 242-174) YES is pro-CC vote any federally sponsored national tests for
elementary or secondary education without
first receiving specific and explicit consent
from Congress
10. National Endowment for the Arts Funding Resolution allowing for the elimination of
7/10/97 (Roll call vote #259, approved 217-216) taxpayer funding for the National
YES is pro-CC vote Endowment for the Arts
11. Legal Services Corporation Funding 9/25/97 Amendment to increase taxpayer funding
(approved 246-176, approved 246-176) NO is pro- for the federally funded Legal Services
CC vote Corporation
*12. Tax Limitation 4/15/97 (Roll call vote # 78, Constitutional amendment which would
rejected 233-190) YES is pro-CC vote require a two-thirds majority vote in both
the House and the Senate in order to raise
taxes
*Vote excluded from revised CC scorecard (additional votes 6-11 excluded by factor analysis)









Table 5-13: 1998 Christian Coalition Votes-VARIMAX Rotated Common Factor (2)
Variables Factor I Factor 2 Uniqueness
Partial Birth Abortion 0.42679 0.65823 0.38459
International Family Planning 0.39093 0.86329 0.10191
Parental Notification 0.45139 0.76843 0.20576
Abortions in Military Hospitals 0.41411 0.82705 0.14450
Needle Exchange Programs 0.46600 0.6604 0.34720
Opportunity Scholarships 0.85792 0.39658 0.10670
Education IRAs 0.83940 0.38333 0.14847
Federal Tests 0.75875 0.45209 0.21992
NEA Funding 0.84929 0.40648 0.11348
Legal Services Corporation 0.67516 0.46492 0.32801
Tax Breaks 0.69017 0.44072 0.32942

Table 5-14: 2000 Christian Coalition Scorecard Votes_
Bill Name Moral Issue
1. Abortion Restrictions 6/30/99 (Roll call #261, Bill that would make it a federal crime for
approved 270-159) Yes is pro-CC vote anyone other than a parent to transport a minor
across state lines to seek an abortion
2. International Family Planning 7/29/99 (Roll call vote Amendment to bar U.S. population control funds
#349, approved 228-200) Yes is pro-CC vote to foreign organizations that perform abortions
3. Unborn Victims of Violence 9/30/99 (Roll call vote Bill making it a criminal offense to injure or kill
#465, approved 254-172) Yes is pro-CC vote a fetus during the commission of a violent crime
4. Needle Exchange Programs 7/29/99 (Roll call vote Amendment to prohibit D.C. from the use of
#344, approved 241-187) Yes is pro-CC vote federal, local or other funds for a needle
exchange program
5. Traditional Family Adoptions 7/29/99 (Roll call vote Amendment to bar joint adoptions in D.C. by
#346, rejected 213-215) Yes is pro-CC vote any couple not related by blood or marriage
6. National Endowment for the Arts Funding 7/14/99 Amendment to reduce funding for the NEA by
(Roll call vote #287, rejected 124-300) Yes is pro-CC vote $2.1 million
7. Casino Gambling 7/14/99 (Roll call vote #289, Amendment to prohibit funding for casino-style
rejected 205-217) Yes is pro-CC vote gambling on Indian lands except through a
tribal-state compact
8. Religious Discrimination in Public Schools 6/17/99 Amendment to prohibit the Office of Juvenile
(Roll call vote #223, rejected 210-216) Yes is pro-CC vote Justice from discriminating against the religious
beliefs of program participants
9. Religious Liberty Protection 7/15/99 (Roll call vote Bill to prohibit governmental interference with
#299, approved 306-118) Yes is pro-CC vote individual religious practices unless the it can
prove "compelling state interest"
*10. Straight A's Education Reform 10/21/99 (Roll call Bill establishing a pilot program allowing 10
vote #532, approved 213-208) Yes is pro-CC vote states to develop student performance goals
*11. Anti-Missile Defense 3/18/99 (Roll call vote #59, Bill to declare that it is U.S. policy to deploy a
approved 317-105) Yes is pro-CC vote national missile defense system
* 12. Tax Limitation 4/15/99 (Roll call vote #90, rejected Joint resolution to propose a constitutional
229-199) Yes is pro-CC vote amendment requiring a two-thirds majority vote
in order to increase taxes
*13. Tax Cut Conference Report 8/5/99 (Roll call vote Adoption of the conference report on the bill to
#379, approved 221-206) Yes is pro-CC vote reduce taxes by $792 billion
*14. Campaign Finance Reform 1 9/14/99 (Roll call vote Bill banning all contributions of soft money and
#422, approved 252-177) No is pro-CC vote imposing restrictions on issue advocacy
communications
*15. Campaign Finance Reform II 9/14/99 (Roll call vote Amendment to exempt voter guides from "issue
#413, rejected 189-238) Yes is pro-CC vote advocacy" regulations
*Votes excluded from revised CC scorecard (No factor analysis revised scorecard available)




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E86IVNCTD_NO6KGQ INGEST_TIME 2013-03-12T13:52:30Z PACKAGE AA00013580_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E1B8XVOZP_8O4UP2 INGEST_TIME 2013-04-08T21:38:05Z PACKAGE AA00013580_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES