Politics and Policy Discretion: How Local Contextual Characteristics Influence the

Implementation of Statewide Ballot Initiatives*

Garrick L. Percival
Department of Political Science
University of California, Riverside
email: ## email not listed ##

*Prepared for delivery at the annual meeting of the American Political Science Association,
Chicago, IL September 2-5, 2004

Abstract
Much of the scholarship on initiatives emphasizes the link between voting behavior and
initiative election outcomes or the strategies employed during the policy enactment phase. Little
research has considered implementation or whether initiatives ultimately achieve their intended
goals. Although the initiative process provides a direct link between the people and policy, it
cannot be assumed that upon voter approval, an initiative will simply be implemented in a
meaningful and uniform manner within the scope of the initiative’s original intent. Using a case
study of California’s Substance Abuse and Crime Prevention Act (SACPA/Proposition 36) I ask
how local politics and other local contextual factors influence the implementation of a statewide
voter initiative. Using aggregate data drawn from each of California’s 58 counties, results
indicate that politics at the county-level, measured via counties’ general ideological dispositions
and more specific policy preferences toward drug abuse and drug offenders affect how they
implement the initiative. Contrary to popular conceptions, results here confirm that initiatives are
not uniformly implemented, but in fact, will be manifested in quite different ways due to local
contextual differences.
Introduction
The initiative process is an institutional tool that citizens and policy entrepreneurs can use
to promote policy change that is too risky or otherwise unacceptable to elected politicians
(Gerber, 1999; Mintrom, 2000). Increasingly over the past two decades voters residing in states
with the initiative have mandated policy changes in areas as diverse as environmental protection,
insurance reform, criminal sentencing laws, and affirmation programs among others. Given
voters’ propensity to impose drastic policy change via the ballot initiative , it is not surprising
that a considerable degree of scholarly attention has been paid to better understanding the
process. However, much of the scholarship on initiatives emphasizes the link between voting
behavior and initiative election outcomes or the strategies employed during the policy enactment
phase (Bowler and Donovan, 1998). Little research has considered implementation or whether
initiatives ultimately achieve their intended goals. Unlike the traditional legislative process,
where voters’ influence on public policy is indirect, direct democracy offers voters opportunities
to bypass the traditional legislative process to directly affect public policy. Because of this, it
might be assumed that upon voter approval, an initiative will simply be implemented in a
meaningful and uniform manner within the scope of the initiative’s original intent.
In contrast to much of the previous work on the initiative process, this paper provides a
relatively unique examination of the fate of an initiative after its enactment through an analysis
of the implementation of California’s Substance Abuse and Crime Prevention Act
(SACPA/Proposition 36). The initiative delegates large implementation responsibilities to local
county governments operating in widely different contextual environments. Because of this, I

1
argue that the implementation of the initiative should not be expected to be uniform but vary as a
function of local political preferences and other local contextual characteristics. Results
demonstrate, in fact, that a statewide ballot measure can be manifested in quite different ways
due to local contextual differences.

Voter Imposed Authority Migrations and the Added Importance of Local Politics
One unmistakable characteristic of direct democracy is that it continues to transform the
institutional contexts of states’ governments. A useful distinction can be made between those
initiatives that transfer policy making authority from local to state government to those that
transfer authority from state to local governments. Perhaps this distinction is best exemplified in
the state of California. The landmark anti tax initiative, Proposition 13, is a good example of an
initiative that transfers significant policymaking authority away from local governments and
instills power in the hands of state agencies and officials. More recently however, voters have
transferred significant programmatic responsibilities (and discretion) to local governments
operating in diverse contextual environments through the enactment of initiatives like
Propositions 36.1 Whether policy making authority is transferred from local to state officials or
vice versa is likely to impact the extent to which local political forces impact the implementation
of a voter initiative. Importantly for my purpose here, the impact of local politics is most likely
to be felt when initiatives transfer implementation responsibilities to local governments operating
in diverse political environments.
The transfer of policymaking authority to local governments via the initiative process has
important implications to the study of direct democracy. For initiatives that place large
implementation responsibilities upon local governments, to what extent do policy outputs vary
by locality, and to what degree do local politics and other local contextual characteristics explain
such variation? On a similar note, is compliance less likely when implementing responsibilities
are largely transferred to local governments, where local political forces could potentially shape
and reshape policy outputs and outcomes? Or put another way, to what degree do political forces
operating at the local level impact whether local governments comply with an initiative’s
original intent?
I present below a discussion as to why a case study of a California initiative is best suited
to answering these important questions. This is followed by a brief background of Proposition
36, a review of the literature highlighting the influence of contextual characteristics on policy
implementation, a presentation of the measures and methods used, formal hypotheses, and
empirical findings. I conclude with a discussion concerning the implications of this research to
the study of direct democracy.

Why a Case Study?

Nicholson-Crotty and Meier (2002) argue that case studies, such as that employed here,
are most advantageous when unique characteristics of a state make it an ideal venue in which to
test a theory. Because of California’s wide use of the initiative process, its diverse citizenry, and
widely divergent local political orientations, the state provides a unique “laboratory” to examine
in detail the expectation that local politics encourage or discourage the implementation of a
citizen’s initiative. Single-state analyses also allow researchers to incorporate more contextual
detail or nuance into a study (Nicholson-Crotty and Meier, 2002). As will be detailed below,
contextual factors like local political dispositions are influential in determining policy outputs
and outcomes, but are often vaguely measured in multi-state studies. A research design focused

2
on one state and one initiative, allows for more explicit measures of important theoretical
concepts. For example, focusing solely on the state of California lends me the use of the
California Field Poll, a multi-year survey enterprise that provides an excellent methodological
tool to create more reliable and valid measures of local political dispositions that are a central
focus of this research. Then, for both theoretical and measurement reasons, California offers a
unique perspective of the implementation of a voter initiative not available in multi-state
analyses.

California’s Proposition 36: A Dramatic Shift in the State’s Drug Policy

Proposition 36 is an example of a statewide initiative that places specific programmatic
responsibilities on county governments operating in widely different contextual environments.
Approved by 61% of California voters, Proposition 36 was enacted in November 2000.
Supported by the Drug Policy Alliance and other public health related agencies, but opposed by
the California Republican Party and the California District Attorney’s Association, Proposition
36 imposes a dramatic shift in California’s fight against illegal drug use and drug-related crime
from a more punitive approach to a more rehabilitative one, as the initiative mandates that non-
violent drug offenders be placed into drug treatment programs rather than sentenced to
incarceration (Riley et al., 2001).
Since the early 1980s, California’s approach to fighting drug-related crime has followed a
deterrence and incapacitation theory, which promotes increased arrests, stricter probation and
parole monitoring, mandatory sentences, and higher rates of incarceration to dissuade street
crime by removing offenders from the larger community (Macallier et al., 2000; Maxwell, 1999;
Tonry, 1999). These policies have proven to be popular politically as they allow state and local
politicians to portray themselves as “tough on crime and drugs” (Beckett, 1999, Tonry, 1999).
Given the political and electoral benefits of the “get tough” approach, California has experienced
a 25-fold increase in the number of drug offenders sentenced to state prison during the past two
decades. By the year 2000, California’s rate of incarceration for drug offenders led the nation,
with the state incarcerating 130 per 100,000 population (Males, Macallair, and Jamison, 2002)
To break this punitive trend the initiative appropriates to county governments using a
standardized formula that takes into account county size and number of drug arrests, $60 million
in fiscal year 2000/2001 for initial start-up costs, and $120 million for each of five subsequent
fiscal years concluding with fiscal year 2005/2006 (Ford and Smith, 2001). These funds are to
be primarily used for drug treatment but can also be used to defer probation and court costs
accrued as a result of SACPA clients.
Implementation requires cooperation between county and state institutions including the
California Department of Alcohol and Drug Programs (DADP), county public health and
criminal justice agencies, and community treatment providers. A significant degree of discretion
is given to counties to determine their respective treatment regimens and number of treatment
facilities that provide services to Proposition 36 clients (Ford and Smith, 2001). Once determined
eligible for SACPA treatment services by counties’ criminal justice officials, drug offenders are
to be placed into outpatient or residential treatment programs.
Implementing Citizen Initiatives and Other Public Policies
The study of policy implementation is far from reaching a consensus on what variables
serve as good predictors of policy outputs and policy compliance (deLeon and deLeon, 2002).
Scholars have more recently treated contextual variables as pivotal factors that help explain the
implementation process. A central goal of contextual studies is to advance our understanding of

3
the extent to which contextual variables—or aggregate measures of characteristics found within
specific political locales (such as a state, county, or city) influence policy choices made by
individuals operating in those environments (Books and Pysby, 1991). The nexus between
context and public policy implementation is most commonly illustrated in research of federal
policy implementation in the decentralized American federalism structure. Research shows that
in a decentralized institutional structure, policy is not implemented according to hierarchical
Weberian principles, where policy enacted by national politicians is simply implemented by
state and local officials in ways that mirror the original intent of national policy makers. Instead,
states’ contexts matters, and policy outputs reflect political forces found at the sub-national level
(Pressman and Wildavsky, 1973; Scheberle, 1997).
Although California’s counties are the specific contextual unit of interest here, prior
research brings attention to several contextual variables relevant to policy implementation and
policy outputs at the local level (Goggin et al. 1990; Mazmanian and Sabatier, 1983). These
variables can be separated into four primary dimensions: political factors, program budget
allocations, policy needs, and socioeconomic characteristics. These variables and their effects on
policy outputs are explained in greater detail below.

Political Factors
Prior research, most commonly found in the study of comparative state politics has found
clear associations between state policy decisions and general political orientations like political
ideology and more specific public demands (Erikson, Wright, and McIver, 1993; Hill and
Hinton-Andersson, 1995; Hill and Leighley, 1996). Political ideology influences individuals’
views about the proper scope of government in public affairs and works to constrain policy
preferences and policy-making decisions. Erikson, Wright and McIver (1993), construct
aggregate measures of state-level political ideology and find convincing evidence that states with
more liberal publics produce more liberal policy outputs. Rather than focusing on a particular
states’ ideological make-up my focus here is on the impact that ideological variation within a
particular state (measured at the county level) has on the implementation of a statewide ballot
initiative. Counties’ ideological dispositions may affect implementation in a couple of ways.
First, the institutional structure of California’s county governments is likely to strengthen the
connection between counties’ ideological dispositions and policy outputs. County boards of
supervisors, district attorneys, and superior court judges—all key implementing actors--are
elected officials operating in diverse political environments. Because they are elected rather than
appointed, it is likely these officials are likely to curtail the implementation of the initiative to
reflect the ideological preferences of their constituents which then works to increase their own
chances of reelection. Second, street-level bureaucrats in more conservative/liberal counties may
be more likely to be ideological conservatives/liberals, and therefore work to implement the
initiative in ways that closely reflect their ideological preferences (see Lipsky, 1980 for a
discussion of the significance of street-level bureaucrats on policy implementation).
In addition the influence of general ideological orientations, researchers have
increasingly found that more specific policy attitudes regarding abortion rights, environmental
protection, and the death penalty (among others) influence policy outputs across the states
(Arceneaux, 2002; Brace, Sims-Butler, Arceneaux, and Johnson, 2002; Norrander, 2001). In
short, policy outputs tend to be reflective of a given locality’s general ideological dispositions
and more specific policy preferences.

4
Program Budget Allocations
The degree to which public programs are funded can have a large impact on public
programs’ ultimate success or failure (Goggin et al., 1990). Including measures of public
programs’ budget allocations as a contextual variable to be considered in the local policy making
process can serve two purposes. Budget allocations can function as measures of policy outputs,
or as O’toole (1986) has shown, predictors of other outputs associated with a specific program At
the local level of government program budgets may influence implementing agencies’ capacity
to carry out implementation tasks thereby altering the amount of benefits and services the public
receives. The target of spending also matters. Research on the Resource Conservation and
Recovery Act found that states with better environmental protection records were not necessarily
the wealthiest, but those that targeted a larger proportion of their budgets toward the specific
purpose of clean up (Goggin et al, 1990). Overall, the extent to which programs are funded, and
the target of that spending is expected constrain the policy choices of implementing officials.

Community Needs
In its simplest form, the needs based approach to explaining policy outputs suggests that
local governments will adopt or implement policies in response to an objective need for a policy.
Where a problem is more severe, it might be expected that the need to fix the problem is more
urgent (Thompson and Scicchitano, 1985). Because local government is associated with close
contact between the people and policy makers, it is often believed that local government can best
match the response of government to the needs of the community. Indeed, a small body of
research shows clear links between need and adoption of economic development incentives,
groundwater protection programs, and drug treatment programs (Rubin and Rubin, 1987; Lester
and Kepter, 1984; Meier, 1994).

Socioeconomic Characteristics
Socioeconomic characteristics refer to the constraints stemming from the resources and
capacities within a given locale. Localities with higher socioeconomic status tend to have higher
tax revenues and therefore have a greater capacity to increase valued public services or
regulatory oversight (Treadway, 1985). At the state level for example, fiscal capacity is a strong
predictor of policy outputs (Dye, 1966; 1979).
Taken together, it is expected that these contextual variables structured around four
primary dimensions: political characteristics (e.g. general ideological dispositions and specific
policy preferences), program budget allocations, community needs, and socioeconomic
conditions constrain the choices (and thus help direct) the policy decisions of county
implementing officials. Constraint placed on the decisional choices made in the implementation
process should affect policy outputs that ensue.

Implementing Citizen Initiatives

The political science profession is just beginning to tackle the question of what happens
to initiatives after their enactment. With the recent budget crisis in California, journalistic
accounts have frequently highlighted that once enacted, initiatives produce many unintended
negative consequences and are largely to blame for the decline in valued public services as voter
imposed mandates severely constrain the policy choices that legislators have to work with
(Schrag, 1998). Gerber et al., (2001) and Bali (2003) are two of the few systematic studies that
focus on the fate of initiatives after their enactment. Using case studies of several California

5
initiatives, Gerber et al., (2001) explain why the implementation of initiatives often fails. Using
a principle-agent model, they argue that initiatives do not enforce themselves, but rather
implementation often requires the help of multiple government officials and agencies operating
at both the state and sub-state levels. Because of this, opportunities become available for
members of state legislatures, localized bureaucrats, or specialized interest groups to implement
policies in ways that meet their own preferences rather than the general public. Studying the
implementation of California’s Proposition 227, Bali (2003) finds political, institutional, and
socio-demographic characteristics within school districts affects compliance with that initiative.
The work of Gerber et al. (2001) and Bali (2003) provides enlightening analysis of how
the decentralized structure of the implementation process influences policy outcomes of several
important California initiatives. This paper, however, measures the extent to which a set of more
explicit county contextual variables affects policy outputs of an initiative that places major
programmatic responsibilities on a critical component of the state's governing apparatus--local
government.

Understanding SACPA Implementation in County Context

Motivated by the literature highlighting the importance of contextual characteristics on
policy implementation and policy outputs, I test the influence of counties’ political orientations,
in addition to other local contextual characteristics on four important policy outputs associated
with Proposition 36—the proportion of SACPA expenditures targeted toward drug treatment,
quality of treatment services that SACPA clients receive, the percentage of clients who
successfully complete their treatment regimens, and the number of persons in each county
incarcerated for low-level drug offenses during the first two years of implementation (2001-
2002).
Recall that counties have significant discretion when determining how to target SACPA
allocations—funds may be targeted toward drug treatment purposes or criminal justice activities
such as probation monitoring and court costs. Providing a measure of the amount of
expenditures counties target toward drug treatment is important insofar as it serves as an
indicator of policy variation between counties, but more importantly, may serve as a predictor of
counties capacity to provide high quality drug treatment and overall treatment success.
Moreover, measuring treatment quality and the proportion of clients within each county that
successfully complete treatment is important insofar as they help gauge counties’ abilities to
achieve the initiative's ultimate goal of statewide reduction in drug addiction and drug-related
incarceration and crime. Because the initiative provides county governments with considerable
discretion in determining their respective treatment regimens, it is expected that the quality of
treatment services, and the proportion of clients who successfully complete treatment will vary
as well, as differing county contextual characteristics are expected to influence counties’
response to the initiative and their abilities to meet the treatment needs of SACPA clients.
Recording the number of persons incarcerated for low-level drug possession tests the
influence that local contextual factors such as counties’ more specific political preferences
toward drug offenders has on counties’ compliance with the initiative’s original intent. Because
it is difficult to measure counties’ compliance when outputs are expected to vary in areas like
treatment quality, a better and more interesting test is found by analyzing specific provisions of
the initiative where counties appear to have little or no discretion in its implementation and
where little variation in outputs might be expected. One such provision is that which determines
eligibility standards for SACPA clients. Recall that as a result of Proposition 36, “non-violent”

6
drug offenders charged with simple drug possession are to be made eligible for treatment rather
than be incarcerated. Similar to the intent of California’s 1994 3-strikes initiative, Proposition 36
is structured to standardize the behavior of the state’s criminal justice system as a whole.
However, this may be difficult given California does not have one criminal justice system but 58
—one in each county (Riley at al., 2001)
It is well documented in the political science literature that institutional rules affect policy
outcomes by limiting the “nature and scope of political actions and choices (Bickers and
Williams, 2001, pp 41).” Initiatives, and the rules connected to them are no exception and
should be expected to influence the choices implementing officials make. Relative to the
considerable discretion given to counties in determining their treatment regimens, Proposition 36
appears to provide little or no discretion to county judges and prosecutors when determining
eligibility requirements for SACPA services. Yet, a closer look at the text of the initiative shows
that criminal justice officials do have a sliver of discretion when sentencing potential clients.
This discretion provides a window of opportunity for criminal justice agents operating in
political environments opposed to the initiative to circumvent the initiative’s original intent by
charging potential SACPA clients with offenses that make them ineligible for the program and
allow offenders to be incarcerated under preexisting law (Riley et al., 2001). 2

Hypotheses
The literature presented above suggests that localized politics and general policy attitudes
like political ideology affects policy implementation and outputs (Erikson, Wright, and McIver,
1993). Given this, it is expected that county ideology dispositions will have both a direct and
indirect impact on the implementation of the initiative. The first hypothesis tests the expectation
that more liberal counties will target a greater percentage of SACPA funds toward treatment
during implementation than conservative counties. This follows that traditionally ideological
liberals are more likely to believe that solving drug abuse and drug-related crime is best done via
the treatment process, whereas ideological conservatives have tended to support tougher punitive
measures (Beckett, 1999). Following this H1 posits:
Ideologically liberal counties are more likely to target a greater proportion of SACPA funds
toward drug treatment than are conservative counties.
In turn, the target of SACPA funding is also expected to impact counties capacity to
provide higher quality treatment services in addition to influencing the rate of treatment success
rates across counties. It is expected that counties that expend a greater proportion of SACPA
funds toward treatment will be more likely to provide more high quality treatment services and
be more likely to have a greater proportion of clients that successfully complete their treatment
regimens. This flows from the assumption that counties that spend more SACPA funds on
treatment will also have a greater capacity to provide higher quality treatment services to a larger
number of clients. In addition, greater spending on drug treatment likely increases counties’
abilities to curtail treatment regimens to meet individual needs of clients which improve the
likelihood of successful treatment outcomes (Gerstein et al., 1994).
The quality of treatment services, measured here, makes a distinction between the types
of services SACPA clients receive. Higher quality services are considered those treatment
programs that are residential in nature, where clients receive more intensive treatment, are longer
in duration, and programs that require clients to remain in residence during the full course of
treatment. Research on the effectiveness of different modalities of drug treatment finds
residential treatment and length of treatment both strong predictors of decreased drug usage

7
(Gerstein et al., 1994). Drug treatment success includes clients who are reported by their drug
treatment providers to have successfully met all the obligations of their treatment program as
outlined by a court mandate. Formally hypothesis H 2 posits:

Those counties that target a greater proportion of SACPA funds toward treatment are more
likely to provide more high quality treatment services by sending a higher proportion of SACPA
clients to residential programs during SACPA implementation.

H 3 posits:

Those counties that target a greater proportion of SACPA funds toward treatment are more
likely have higher a proportion of clients who successfully complete treatment during SACPA
implementation.

Given the nature of the expected relationships posited in hypotheses 1-3, it is expected
that counties’ ideological dispositions will have an indirect impact on treatment quality and
treatment success during implementation. Formally, hypothesis H 4 posits:

More liberal counties are more likely to provide more high quality treatment services and have
higher rates of treatment success as a result of targeting a greater proportion of SACPA
expenditures toward drug treatment.

Finally, the fifth hypothesis aims to test the extent to which counties’ more specific
political preferences toward drug offenders influences counties’ compliance with the initiative’s
mandate that individuals charged with simple drug possession or drug use be diverted away from
incarceration and instead placed into treatment. Given this, H 5 posits:

Counties that have traditionally taken a more punitive approach to drug users and drug abuse
before the enactment of Proposition 36 are more likely to incarcerate individuals for low-level
drug possession during the implementation of Proposition 36.

A priori, where there is little policy discretion given to county governments in areas such
as determining SACPA treatment eligibility, it might be expected that the influence of local
political factors on implementation would be attenuated. If however, results support this
hypothesis, and counties are finding ways to implement SACPA in ways that closely matches
their own political preferences, this would further exemplify the importance of local politics on
the implementation of the initiative. Empirical evidence showing some counties incarcerating
more individuals for low-level drug offenses than others, a behavior that subsequently works to
disqualify more people for SACPA services, would provide evidence that local political forces
can move policy outputs away from the initiative’s original intent of treatment, even under
conditions where it is least expected.

8
Modeling Proposition 36 Policy Outputs

Dependent Variables
The first dependent variable labeled SACPA drug treatment expenditures measures the
percentage of SACPA funds targeted toward treatment during implementation. The expenditure
data used here is gathered from Ford and Smith (2001, 2002) and measures the total amount of
SACPA funds targeted toward treatment between 2001-2002. Calculations are based on 100%
of useable funds. Larger numbers represent greater spending toward drug treatment. As will be
illustrated in the causal diagram below, the drug treatment expenditures variable also serves as
an independent predictor of service quality and treatment success.
The second dependent variable, labeled treatment quality measures within each county
the percentage of SACPA clients referred to residential treatment programs during
implementation. This variable serves as a measure of service quality where residential services
are considered higher quality services than less intensive outpatient programs. Data is collected
from the California Department of Alcohol and Drug Treatment Programs.
An additional variable labeled drug treatment success is also included and measures the
percentage of SACPA clients who satisfactorily completed their mandated treatment
requirements as reported by clients’ treatment providers. Data are gathered from the California
Department of Alcohol and Drug Programs.
Finally, a fourth dependent variable, labeled drug incarcerations measures the total
number of persons incarcerated (per 1,000 persons) in each county for low-level drug possession
during the first two years of SACPA implementation. Data are gathered from California
Department of Correction’s Data Analysis Unit. The low-level drug possession measure is the
principal controlling offense and refers to individuals charged with possession of small quantities
of drugs for personal use. Incarcerations for low-level drug offenses is of most interest here
because the initiative mandates that low-level, non-violent drug users are to be placed into
treatment, while high-level drug offenses are not eligible for SACPA services.

Independent Variables
Several contextual variables are included as predictors of the dependent variables.
Among political contextual factors is a measure of county-level ideology created from pooling
California Field Poll surveys between 1990-1999. (See the Appendix for measures of reliability
and stability and further discussion of the methodology used). The ideology variable provides a
general measure of counties’ political orientations and policy preferences and is used to predict
SACPA drug treatment expenditures, treatment quality and success, and incarcerations of low-
level drug offenders. This latter expectation is motivated by the assumption that more
conservative counties will incarcerate a greater number of low-level drug offenders during
SACPA implementation considering that traditionally, ideological conservatives tend to be
“tougher on drugs” (Beckett, 1999).
Three additional political variables are included as predictors and serve as more specific
measures of counties’ drug policy preferences. The first measure, what I call “tough on drugs” is
included as a predictor of incarceration rates for low-level drug offenses during SACPA
implementation. This measure is calculated by averaging the total number of incarcerations due
to all drug offenses in each county between 1996 and 1999. Drug offenses in this measure
include those incarcerations for drug sales, manufacturing, possession in quantities large enough
to presume intent to sell, and low-level drug possession. 3 The second is an averaged measure of

9
counties’ drug treatment expenditures between 1995-1999. This serves as a predictor of SACPA
drug treatment expenditures where it is expected those counties that expended greater funds
toward treatment before the enactment of the initiative will seek to maintain this policy trend by
targeting a greater proportion of SACPA funds toward treatment during implementation. The
third is an averaged measure of counties’ probation expenditures between 1996-1999. 4 It is
expected counties appropriating greater funds toward probation prior to the initiatives enactment
will continue this policy trend by targeting less SACPA money toward treatment and more
toward criminal justice related activities.
Counties’ drug problem severity is included as a measure of counties’ policy needs and is
used to predict drug treatment expenditures, treatment quality, in addition to serving as a control
variable predicting drug incarcerations. The degree of drug problem severity is measured using
the average number of deaths caused by drug overdose in each county between 1999-2001. Data
are collected from the Center for Health Statistics, California Department of Health Services. Its
use as a control for predicting drug incarcerations follows an alternative expectation that some
counties where the drug problem is more severe may take a more punitive approach to
decreasing drug use and drug-related crime by incarcerating more individuals for low-level
possession.
Although the political contextual variables are central to testing the hypotheses, several
additional variables are included as controls to complete the regression models presented below.
The total number of residential treatment facilities in 2002 is added to predict treatment quality
across the counties. This serves as a proxy measure of residential treatment capacity within each
county. This is an important measure insofar as the residential treatment capacity within each
county likely constrains the choices that implementing officials make when determining the
quality of treatment SACPA clients receive. For example, counties that have fewer/greater
numbers of residential facilities capable of providing those services might be expected to refer a
lower/higher percentage of SACPA clients to residential programs.
Counties’ socioeconomic status is included as a factor score combining measures of
median income and educational attainment. 5 Counties with higher socioeconomic characteristics
are expected to be less likely to incarcerate low-level drug offenders. A race variable, measuring
the number of racial minorities (Blacks and Hispanics) per 1,000 persons in each county is
included as a predictor of drug incarcerations. Counties where a greater number of racial
minorities reside might be expected to incarcerate more low-level drug offenders as prior
research has shown racial minorities are more likely to be incarcerated for drug offenses than are
whites (Meier, 1990). Finally, two individual level measures including SACPA clients’ drug
addiction severity and levels of educational attainment are included to predict treatment success
rates. The severity of SACPA clients’ drug addiction is measured by the percentage of SACPA
clients in each county who self-reported daily drug use before entering treatment. It is expected
that counties where clients have less severe addiction problems will have higher overall
treatment success rates. Clients educational attainment is measured by calculating the percentage
of SACPA clients in each county who have at least “some” college education or have completed
a college degree. It is expected that those counties where SACPA clients have higher
educational levels will have greater rates of treatment success.

Results
Table 1 shows results of regressing the percentage of SACPA funds targeted toward drug
treatment on counties’ general ideological preferences and more specific drug policy preferences.

10
As expected in the first hypothesis, counties general ideological orientations influence the extent
to which counties target SACPA funds for drug treatment services rather than criminal justice
related activities. During the first year of implementation, counties ideological dispositions did
not significantly impact expenditure choices, however during the second year of implementation
there are substantial differences between more liberal and conservative counties and how they
choose to target SACPA funds. Ceteris paribus, more ideological liberal counties expend a
greater proportion of funds toward treatment than do conservative counties. This follows
traditional ideological beliefs among liberals that fighting drug abuse and drug related crime is
best done via the treatment process. As will be shown in greater detail below, greater
expenditures toward drug treatment-related services among liberal counties provides an
important foundation for higher treatment service quality and treatment success.

[Place Table 1 About Here]

Results in Table 1 also show that counties’ specific policy preferences toward drug abuse,
in this case measured via counties’ prior drug treatment expenditure levels, explains policy
choices during the implementation of SACPA above and beyond that of counties’ general
ideological dispositions. Those counties’ that chose to appropriate a greater amount of public
funds toward drug treatment and rehabilitation before the enactment of Proposition 36 are more
likely to continue that policy trend during the implementation of the initiative by targeting a
greater proportion of SACPA funds toward treatment.
Finding variation between counties in terms of how they choose to target SACPA funds
is intriguing but perhaps not that surprising. Because counties have considerable discretion in
choosing how they target SACPA funds, significant variation should be expected given that local
elected officials and other implementing agents operating in diverse political contexts are likely
to curtail spending outputs in ways that match local political preferences. Given this, it is perhaps
more interesting to test the extent to which variation in SACPA spending impacts more
“substantive” outputs associated with Proposition 36 such as treatment service quality and drug
treatment success rates.
Table 2 provides results of testing the extent to which the target of SACPA funding
impacts the quality of treatment services. It is important to note that the SACPA drug treatment
expenditures variable that was used as a dependent variable in Table 1 is now used a predictor of
treatment quality during the second year of implementation. The positive and statistically
significant association between the SACPA drug treatment expenditures variable and the
treatment quality measures shows support for the second hypothesis. After controlling for other
confounding factors such as counties policy needs and residential treatment capacity, results
indicate that those counties that target a greater amount of SACPA funds toward treatment have
a greater capacity to fund higher quality treatment programs and thus are able to send a greater
proportion of SACPA clients to higher quality residential treatment programs.

[Place Table 2 About Here]

One major criticism of the implementation of Proposition 36 has been the relatively high
proportion of clients who do not successfully complete their treatment regimens. Indeed,
averaging drug treatment success rates across all 58 counties during the first year of
implementation shows that only 26% of clients successfully completed their treatment programs.

11
This improved slightly during the second year, as 32% of clients successfully completed
treatment. Although it is out of the scope of this particular paper to understand all the factors
behind the relatively low overall treatment success rates, it is important nonetheless to attempt to
explain some of the variation in success rates across the counties.
Results presented in Table 3 shows even after controlling for clients educational
attainment and drug addiction severity, SACPA drug treatment expenditures have a significant
influence on counties’ treatment success rates, indicating support for the third hypothesis.
Counties appropriating greater funds toward treatment, on average, have a greater proportion of
SACPA clients who successfully complete their treatment programs. Reviewing the aggregate
data used here, it is difficult to determine exactly how increased expenditures improve treatment
success rates. However, initial evidence from a statewide survey of implementing officials (the
survey is currently in the “field”) suggests that higher levels of treatment funding allows counties
to make better assessments of clients’ treatment needs, provides drug treatment professionals
more opportunities to curtail treatment programs to the specific needs of clients, and in some
cases, provides for transportation services to and from treatment facilities.

[Place Table 3 About Here]

The importance of counties’ ideological dispositions on the quality of SACPA treatment

services and overall treatment success is perhaps best understood by viewing the causal model in
Figure 1 which shows the indirect effects of county ideology on treatment quality and treatment
success during the second year of implementation. As shown, county ideology influences the
target of SACPA allocations, which in turn directly affects the treatment quality and the
treatment success rates of SACPA clients. Translating abstract standardized coefficients into
more meaningful units, this model suggests a one standard deviation (in this example moving a
standard deviation in the liberal direction) in a county’s ideology score produces a 2% increase
in SACPA funds used for drug treatment related services. Ceteris paribus, liberal San Francisco
County (ideology score=-25.21) which is over 3 standard deviations below the ideology mean
(that is, moving in the liberal direction), would increase the proportion of SACPA funds used for
treatment by approximately 8% more than Kern County (ideology score=36.62) which is over
one standard deviation above the mean.

A one standard deviation increase (S.D.=10%) in SACPA funds targeted toward drug
treatment produces a 2.6% increase in the number of clients who receive higher quality treatment
services. As predicted in hypothesis four, counties’ ideological dispositions have an indirect
impact on both the quality of services SACPA clients receive and treatment success rates where
liberal counties are more likely, on average, to provide higher quality services and have greater
treatment success than more conservative counties.

[Place Figure 1 About Here]

Recall the fifth hypothesis posits that counties that were “tough on drugs” before the
enactment of Proposition 36 will be more likely to incarcerate individuals for low-level drug
possession during the implementation of the initiative, effectively making those incarcerated
ineligible for SACPA services.

12
 Indeed, the total number of individuals incarcerated for low-level drug possession in
California has dramatically decreased since the implementation of Proposition 36. Data
collected from California Department of Correction’s Data Analysis Unit indicates that in 2000,
the last year prior to the implementation of the initiative, 12,156 individuals were incarcerated
for low-level drug possession. By the end of the second year of SACPA implementation, this
number had nearly been cut in half with 6,439 individuals incarcerated. Although these numbers
are intriguing, they do not provide an answer to an important question—are county governments
uniformly complying with the initiatives’ original intent that low-level drug offenders receive
treatment and avoid incarceration? To help answer this question, and to test the fifth hypothesis,
I turn attention again to analysis of county-level data.
I begin testing hypothesis H 5 by placing counties into 1 of 3 categories--low, moderate,
or high—with their placement dependent upon their respective average number of incarcerations
due to drug related offenses between 1996 and 1999. Those with a high number of
incarcerations due to drug related offenses are considered “tough on drugs” (and placed into the
“high” category), with those incarcerating the fewest, considered the least “tough” (and placed in
the “low” category). 6 Counties that are considered the toughest toward drug offenders before the
enactment of Proposition 36 incarcerate greater numbers of individuals for low-level drug
possession during the 2001 and 2002 SACPA implementation period than do counties that are
considered least tough on drug offenders. On average, counties considered traditionally toughest
on drugs are 4.1 times more likely to incarcerate individuals for low-level drug possession than
are counties that have been traditionally most lenient on drugs during the second year of SACPA
implementation. This represents only a slight decrease from 2000, when counties considered
toughest on drugs were 4.5 times (incarcerating 45 per 100,000 persons) more likely to
incarcerate for low-level drug possession than the most lenient counties (incarcerating 10 per
100,000). Although the absolute number of individuals incarcerated for low-level drug
possession has decreased across all counties during SACPA implementation, between the years
of 2000 and 2002 there has been only a slight change in the rate of incarceration among counties
that are considered toughest on drugs relative to those in considered most lenient. Overall, the
results suggest that even with a specific initiative mandate where, a priori, little variation might
be expected between the counties, politics at the local level, in this case measured via counties’
specific policy preferences toward drug offenders has a significant impact on the implementation
of the initiative. At least in some counties, these political forces “pull” policy outputs away from
the original intent of policy formulators and in doing show signs that some are working to
circumvent the original intent of the initiative.
A multivariate analysis shows similar results. Table 4 provides results of regressing the
number of individuals incarcerated for low-level drug possession during the first two years of
SACPA implementation on the tough on drugs measure, county ideology, drug problem severity,
socioeconomic status, and the racial minority variables included as controls. 7 The standardized
beta coefficients in Table 4 indicate that the tough on drugs measure is the strongest predictor of
incarcerations in both 2001 and 2002 with the full models explaining 65% and 66% of the
variance in the dependent variable in each respective year. Controlling for other confounding
variables the results remain substantively the same--those counties toughest on drugs incarcerate
more individuals for low-level drug possession during the implementation of Proposition 36.

[Place Table 4 About Here]

13
 In sum, it appears that as intended, Proposition 36 is working to reduce the number of
individuals incarcerated for low-level drug offenses—the number of those statewide incarcerated
for these offenses has been nearly cut in half during the first two years of SACPA
implementation. These results only tell part of the story however. How counties implement
Proposition 36, and more specifically how counties deal with low-level drug users is not
uniform. This variation in policy outputs is a function of local political forces as counties
implement Proposition 36 in ways that closely matches their policy preferences before the
initiative was even enacted.

Discussion
This research has provided a unique look at a statewide voter initiative that has imposed
specific programmatic responsibilities to local governments. The results confirm that contrary to
popular expectations, voter initiatives are not uniformly implemented but rather outputs are a
function of local political preferences and other local contextual characteristics.
The empirical results presented here provide explicit measures of several contextual
factors that influence the implementation of Proposition 36. Using the California Field Poll to
create a measure of county ideology, I find that both counties’ general ideological dispositions
and more specific policy preferences matter in shaping the implementation of initiatives that are
exogenously introduced. Counties’ ideological dispositions are found to have an indirect
influence on the quality of treatment services SACPA clients receive in addition to clients’
treatment success rates. Ceteris paribus, more liberal counties expend a greater proportion of
SACPA funds toward drug treatment which provides SACPA implementing officials operating
in those environments a greater opportunity to refer more SACPA clients to higher quality
residential treatment services in addition to improving clients’ overall treatment success rates.
Counties specific policy preferences toward drug offenders also matters, where counties
considered tough/less tough on drug offenders implement the initiative in ways that closely
match these preferences during the first two years of SACPA implementation--counties
considered tough on drugs are more likely to incarcerate low-level drug offenders than those that
are more lenient. Thus, even in the relatively early stages of Proposition 36 implementation,
local political forces are in some counties moving outputs away from the initial intent of policy
formulators.
How does this study of Proposition 36 help our understanding of the fate of statewide
initiatives after enactment? The results indicate that when large implementation responsibilities
are transferred to local governments via the initiative process, policy outputs should be expected
to vary given the opportunities provided to local governments operating in widely different
political environments to shape policy.
On one hand, it can be argued that variation in outputs like treatment expenditures and
quality of treatment is not surprising, and is in fact beneficial—local governments are merely
producing outputs that are more closely responsive to the state’s diverse citizenry. On the other
hand, when considering policy outputs (e.g. incarcerations) that provide a better measure of
compliance, and where policy formulators appeared to expect more uniformity across counties,
these findings are in some respects, rather troubling. During the initiative’s campaign,
proponents repeatedly sold the initiative as a vehicle to transform the state’s approach to drug
use and drug related crime and not a policy that would force some counties to comply with the
initiative’ intent but not others. With respect to determining sentencing and eligibility for drug
treatment, Proposition 36 attempts to standardize across the state those who qualify for

14
treatment, and in doing so, tries to take discretion away from counties in how they deal with non-
violent offenders after they are arrested. Whereas prior to the enactment of the initiative,
counties could choose among alternatives of incarceration or treatment, Proposition 36 tries to
eliminate this latter choice by mandating treatment for nonviolent offenders. Although the
results here show fewer offenders are being incarcerated for low–level drug possession, the fact
that some counties appear to be finding small, but significant areas of discretion where they are
able to curtail policy outputs to match their policy preferences is likely disconcerting to some.
Importantly, the findings here suggest that when the implementation of initiatives is placed in the
hands of local officials operating in differing political environments, uniformity in compliance
may be difficult to achieve.
Future research on this initiative should consider more closely, measures of policy
outcomes or policy change. Studying this initiative over a longer period of time would provide
researchers a great opportunity to analyze its substantive impact on drug addiction rates across
counties, growth in treatment capacities, and the impact of treatment on drug-related crime. This
particular research will hopefully encourage others interested in direct democracy and policy
studies to pay closer attention to the policy outputs and outcomes of voter initiatives. Because the
impact of policy ultimately reflects the implementation of policy, not merely its enactment,
measuring the actual implementation of initiatives is just as important as understanding why
voters choose to support or oppose initiatives.
Appendix
Erikson, Wright, and McIver (1993) significantly advanced the understanding of state-
level public opinion by creating political ideology and partisanship measures by pooling 1976-
1988 nationally sampled CBS/New York Time’s polls and disaggregating them to the state level.
These measures were shown to be reliable, stable, and valid.
Here, I use similar methodology to create measures of county-level ideology by
disaggregating to the county-level statewide Field Poll surveys conducted between 1990-1999.
The Field Poll uses samples of the California telephone household population drawn from
random digit dial (RDD) samples of Survey Sampling Incorporated. In each survey, the primary
sampling unit is California counties where samples are systematically stratified to all counties in
proportion to each county’s share of telephone households in the survey area. Further sampling
information can be referenced from the California Field Poll Code Books 1990-1999.
A total of 51,930 individual respondents were gathered from a total of 48 Field Poll
surveys. Respondents were asked to place themselves along a 3-point political ideology
continuum. Specifically, respondents were asked, “do you consider yourself to be politically
conservative, liberal, middle-of-the road, or don’t you think of yourself in this way.”
Conservatives were coded 100, middle-of-the-road 0, and liberals –100. Fortunately, the Field
Poll asks each respondent his or her county of residence allowing me to link each response to a
given county. Individual responses were then reaggregated to create ideological scores for 52 of
California’s 58 counties. The number of cases in each county ranged from 13,873 in Los Angeles
county to 17 in Alpine county (mean=659.01). Six small rural counties including, Alpine
(n=17), Colusa (n=36), Mono (n=33), Modoc (n=33), Sierra (n=18), and Trinity (n=30) are not
used in the analysis because of low sample size. Ideology scores ranged from the most
conservative Madera county (41.26) to the most liberal San Francisco county (-25.35) with a
mean=21.88.

Auditing the County-Level Measure of Ideology

15
 Individual responses are treated here as aggregate data, and therefore it is not appropriate
to use standard measures of individual-level reliability like Cronbach’s alpha (Brace et al., 2002).
Because of this, Jones and Norrander (1996) recommend testing reliability analysis on the basis
of aggregate units, and not individuals. To first test the reliability of the ideology measure I use
the O’Brien coefficient (see O’Brien, 1990). Presented by Jones and Norrander (1996), the
O’Brien reliability test seeks to compare the within-unit variance to the across-unit variance
while taking into account sample size (Norrander, 2001: 113). 8 An O’Brien reliability
coefficient that exceeds .70 is considered to be highly reliable, and values between .60 and .70
are considered to be moderately reliable. The O’Brien coefficient for the county-ideology
measure is .96.
An additional test of reliability is the split-half approach used by Erikson, Wright, and
McIver (1993). The split-half approach involves splitting the Field Poll sample into two subsets
by assigning odd-year surveys to one subset and even years to the other. Mean scores for county
ideology were calculated for each subset and correlated using Pearson’s r coefficients. The
Spearman-Brown prophesy formula was then used to assess the reliability of each measure.
Reliability scores of .70 and above are considered reliable and those between .60 and .70 are
considered moderately reliable, and those below .60 are considered unreliable (Jones and
Norrander, 1996). The Spearman-Brown formula:

Where r 12 = the Pearson’s r correlation between the split-halves. The Spearman-Brown

coefficient for the ideology measure equals .60.
To test the stability, the Field Poll sample was divided into “early” and “late” subsets.
The early subset included survey years between 1990-1995 and the late subset between 1996-
1999. Mean scores for county ideology were calculated and correlated. The Spearman-Brown
coefficient for the ideology measures equals .62.
In sum, the assessment of the reliability of the ideology and partisanship measures is
mixed. The O’Brien measure is highly reliable although the Spearman-Brown coefficients using
the split-half approach are at the low end of scores considered to be “moderately” reliable. I
have chosen to use the ideology measure here, but at the same time make note of its possible
deficiencies.

Validity
To test the validity of the Field Poll sample a series of demographic characteristics were
derived from the Field Poll and correlated with county demographic characteristics collected by
the U.S. Census (see Brace et al., 2002). Results presented in Table 5 show that county samples
obtained from the Field Poll are remarkably representative. Specifically, I find a strong
correlation between the educational attainment of the sample and educational attainment reported
by the U.S. Census in 1990 and 2000. A similarly strong relationship is found among between
the income of Field Poll respondents and U.S. Census statistics. Racial characteristics of
respondents, although showing a slightly weaker correlation to U.S. Census figures than do the
education and income figures, are moderately strong nonetheless. The strong correlations for
education and income, and the moderately strong correlations for the race variables suggest that
the Field Poll samples reflect county populations.

16
References
Arceneaux, Kevin. 2002. “The ‘Gender Gap’ in State Legislative Representation: New Data to
Tackle and Old Question.” Political Research Quarterly 54:143-160.
Bali, Valentina A. 2003. “Implementing Popular Initiatives: What Matters for Compliance”?
The Journal of Politics 65: 1130-1146.
Beckett, Katherine. 1997. Making Crime Pay: Law and Order in Contemporary American
Politics. New York, NY: Oxford University Press.
Books, John W., and Charles L. Pysby. 1991. Political Behavior and the Local Context. New
York, NY: Praeger.
Bowler, Shaun, and Todd Donovan. 1998. Demanding Choices. Ann Arbor, MI: University of
Michigan Press.
Brace, Paul, Kellie Sims-Butler, Kevin Arceneaux, and Martin Johnson. 2002. “Public Opinion
in the American States: New Perspectives Using National Survey Data.” American
Journal of Political Science 46: 173-186.
deLeon, Peter, and Linda deLeon. 2002. “What Ever Happened to Policy Implementation? An
Alternative Approach.” Journal of Public Administration Research and Theory 4: 467-
492.
Dye, Thomas R. 1966. Politics, Economics, and Public Policy: Outcomes in the American
States. Chicago, IL: Rand McNally.
Dye, Thomas R. 1979. "Politics and Economics: The Development of the Literature on Policy
Determination.” Policy Studies Journal (June) 652-662.
Erikson, Robert S., Gerald C. Wright, and John P. McIver. 1993. Statehouse Democracy. New
York, NY: Cambridge University Press.
Feiock, Richard C., and Jonathan West. 1993. “Testing Competing Explanations for Local
Policy Adoption: Municipal Solid Waste Recycling Programs.” Political Research
Quarterly 46, 2: 399-419.
Ford, A., and Micky J.W. Smith. 2001. "Substance Abuse and Crime Prevention Act of 2000:
Analysis of Plans from the 58 Counties." Report prepared by Health Systems Research
Inc. for the Center for Substance Abuse and Mental Health Services Administration
Gerber, Elisabeth R. 1999. The Populist Paradox: Interest Group Influence and the Promise of
Direct Democracy. Princeton, NJ: Princeton University Press.
Gerber, Elisabeth R., Arthur Lupia, Matthew McCubbins, and D. Roderick Kiewiet. 2001.
Stealing the Initiative: How State Government Responds to Direct Democracy. Saddle
River, NJ: Prentice-Hall.
Gerstein, D.R., R.A. Johnson, H.J. Harwood, D. Fountain, N. Suter, and K. Malloy. 1994.
Evaluating Recovery Services: The California Drug and Alcohol Treatment Assessment.
Chicago, IL: National Opinion Research Center; and Fairfax, VA: Lewin.
Goggin, Malcolm, Ann O’M. Bowman, James P. Lester, and Laurence J. O’Toole Jr. 1990.
Implementation Theory and Practice: Toward a Third Generation. Glenview, IL: Scott,
Foresman/Little Brown.
Hill, Kim Quaile, and Angela Hinton-Andersson. 1995. “Pathways of Representation: A Causal
Analysis of Public Opinion-Policy Linkages.” American Journal of Political Science
39:924-35.
Hill, Kim Quaile, and Jan E. Leighley. 1996. “Political Parties and Class Mobilization in
Contemporary United States Elections.” American Journal of Political Science 40:787-
804.

17
Jones, Bradford S., and Barbara Norrander. 1996. “The Reliability of Aggregated Public Opinion
Measures.” American Journal of Political Science 40: 295-309.
Lester, James P., and Patrick Keptner. 1984. “State Budgetary Commitments to Environmental
Quality Under Austerity.” In Western Public Lands, eds. John G. Francis and Richard
Ganzel. Totawa, NJ: Rowman and Allanheld.
Lester, James P., and Ann O’M. Bowman. 1989. “Implementing Environmental Policy in a
Federal System: A Test of the Sabatier-Mazmanian Model.” Polity 21, 4: 732-753.
Lipsky, Michael. 1980. Street-level Bureaucracy: Dilemmas of the Individual in Public
Services. New York: Russell Sage Foundation.
Macallier, Daniel, Michael Males, Cheryl Rios, and Deborah Vargas. 2000. “Drug Use and
Justice: An Examination of California Drug Enforcement Policy.” Report prepared for
Center on Juvenile and Criminal Justice.
Males, Mike, Daniel Macallier, and Ross Jamison. 2002. “Drug Use and Justice 2002: An
Examination of California Drug Policy Enforcement. Report prepared for Center of
Juvenile and Criminal Justice.
Mazmanian, Daniel, and Paul Sabatier. 1983. Implementation and Public Policy. Chicago, IL:
Scott Foresman and Company.
Maxwell, Sheila R. 1999. "Conservative Sanctioning and Correctional Innovations in the United
States: An Examination of Recent Trends." International Journal of the Sociology
of Law 27: 401-412.
Meier, Kenneth J. 1994. The Politics of Sin: Drugs, Alcohol and Public Policy. Armonk, NY:
M.E. Sharpe, Inc.
Mintrom, Michael. 2000. Policy Entrepreneurs and School Choice. Washington D.C.:
Georgetown University Press
O’Brien, Robert M. 1990. “Estimating the Reliability of Aggregate-Level Variables Based on
Individual-Level Characteristics.” Sociological Methods and Research 18: 473-504.
Nicholson-Crotty, Sean, and Kenneth J. Meier. 2002. “In Defense of Single-State Studies.”
State Politics and Policy Quarterly 2, 4: 411-422.
Norrander, Barbara. 2001. “Measuring Public Opinion with the Senate National Election Study.”
State Politics and Policy Quarterly 1: 111-125.
Pressman, Jeffrey L., and Aaron Wildavsky. 1973. Implementation. Berkeley, CA: University of
California Press.
Riley, Jack. K., Pat Ebener, James Chiesa, Susan Turner, and Jeanne Ringel. 2001. “Drug
Offenders and the Criminal Justice System: Will Proposition 36 Treat or Create
Problems?” RAND Corp.
Rubin, Irene, and Harold Rubin. 1987. “Economic Development Incentives: The Poor (Cities)
Pay More.” Urban Affairs Quarterly 23:39-57.
Scheberle, Denise. 1997. Federalism and Environmental policy. Washington D.C.: Georgetown
University Press.
Sharpe, Elaine. 1999. The Sometime Connection: Public Opinion and Social Policy. Albany,
NY: State University of New York Press.
Thompson, Frank, and Michael J. Scicchitano. 1985. “State Implementation and Federal
Enforcement Priorities.” Administration and Society 19: 95-124.
Tonry, Michael. 1999. "Why Are U.S. Incarceration Rates So High?" Crime and Delinquency
45: 419-437.
Treadway, Jack M. (1985). Public Policymaking in the American States. New York: Praeger

18
TABLE 1. SACPA Drug Treatment Expenditures Regressed on Selected Predictors

Predictors % SACPA Funds % SACPA Used % SACPA Used

Used Toward Drug Toward Drug Toward Drug
Treatment (2001) Treatment (2002) Treatment Avg 01-
02

b Beta b Beta b Beta

Ideology -.101 -.152 -.127 -.205** -1.37 -.221**

(.089) (.079)

Drug Treatment .00009 .432*** .00009 .432*** .00009 .470***

Exp. 1995-1999 (.000) (.000) (.000)

Probation Exp. -.076 -.145 .0016 .982 -.003 -.064

1996-1999 (.068) (.000) (.062)

Drug Problem -18.60 -.085 -30.72 .138 -29.40 -.144

Severity (30.32) (30.05) (27.66)

(Constant) 90.26 (4.52) 88.78** 90.35

(4.37) (4.12)

R2 = .23 F =3.68** R2 = .22 F =3.52** R2 = .26 F = 4.46**

Note: Reported above are unstandardized and standardized OLS regression coefficients with
standard errors in parentheses **p<.01, ***p<.001 (one-tail test). N=58

TABLE 2. SACPA Treatment Quality Regressed on Selected Predictors

Predictors Treatment Quality 2002

b Beta
Ideology .0004 .053
(.001)

Drug Problem Severity -.360 -.110

(.370)

Residential Treatment 4.46 .592***

Capacity 2002 (.865)

19
 SACPA Drug Treatment .002 .182*
Expenditures (.002)

(Constant) -.138
(.138)
R2 = .38 F=8.10***

Note: Reported above are unstandardized and standardized OLS regression coefficients with
standard errors in parentheses *p<.05, **p<.01, ***p<.001 (one-tail test). N=58

TABLE 3. Percentage of Successful Treatment Outcomes Regressed on Selected Predictors

Predictors % Successful Treatment % Successful Treatment

Outcomes 2001 Outcomes 2002

b Beta b Beta
SACPA Treatment .003 .537*** --
Expenditures 2001 (.001)
SACPA Treatment -- .00005 .
Expenditures 2002 629***
(.000)
SACPA Client .818 .293** -.031 -.014
Education Attainment (.312) (.262)
SACPA Client -.046 -.044 -.003 -.035
Addiction Severity (.120) (.112)
(Constant) 14.20 27.78
(7.16) (6.20)

R 2 = .45 R 2 = .41
F=13.27*** F=11.47***

Note: Reported above are unstandardized and standardized OLS regression coefficients with
standard errors in parentheses. *p<.05, **p<.01, ***p<.001 (one-tail test). N=58

TABLE 4. Low-Level Drug Incarcerations Regressed on Selected Predictors

Predictors Low- Drug Low-Level Drug Low-Level Drug

Level Incarcerations 2001 Incarcerations 2002 Incarcerations Avg.
2001-2002
Tough on Drugs .758*** .736*** .781***
(.060) (.048) (.048)
Drug Severity .081 -.078 .118
(.411) (.330) (.329)
Racial Minority .130** .041 .095

20
 (.000) (.000) (.000)
Ideology .099 (.002) .13* .11
(.011) (.001)
Socioeconomic Status -.089 .001 .051
(.022) (.017) (.017)

(Constant) (.070)* (.056)

(.056)*

R 2 =.65, F=17.72*** R 2 =.66, F=17.95*** R 2 =.71, F=22.34***

Note: Reported above are OLS standardized beta coefficients with standard errors in
parentheses. *p<.10, **p<.05, ***p<.01. (one-tail test) N=58

TABLE 5. Representativeness of Field Poll County Samples

2000 1990 Average

1990/2000
Education .91** .91** .91**
Income .90** .92** .92**
White .75** .76** .76**
Black .87** .88** .88**
Asian .86** .87** .87**
Democratic Party .76** .78** .77**
Registration
Note: ** p<.01 Education is measured by the percentage of county residents who have earned a
bachelors degree or higher. Income is measured by correlating the percentage of Field Poll
respondents who mentioned their total household income was between $20,000-$40,000 dollars
and the median household income of the respondent’s county reported by the U.S. Census.
Racial characteristics are based on sample estimates drawn from self-reported information from
the Field Poll and are correlated with U.S. Census data. Democratic Party registration is based
on the percentage of Field Poll respondents who identified themselves as members of the
Democratic Party and correlated with voter registration data housed by the California Secretary
of State.

21
FIGURE 1. The Indirect Impact of Local Political Variables on the Quality of SACPA
Treatment Services and Treatment Success Rates

County Idiology Succes Drug Treatment Rate

-.205* .629***

.432***
% SACPA Funds
Spent on Treatment
(2002) .182*

Drug Policy Preferences Treatment Quality

(Drug Treatment Expenditures 1996-
1999)

Reported above are standardized OLS regression coefficients drawn from results in Tables 1-3.

Notes
1 California’s Children and Families First Initiative (Proposition 10) enacted in 1998, is another
recent example of an initiative placing specific programmatic responsibilities on county
governments. The initiative mandates newly established county-level commissions to fund
community education, childcare, healthcare and other social services for children five years of
age and under.
2 During the 2000 campaign proponents of the initiative claimed criminal justice agencies might
try to charge potential and/or existing clients with offenses that make them ineligible for
services. There are several ways criminal justice agents can disqualify individuals for treatment
and sentence him or her to incarceration. For example, if a probationer participating in SACPA
treatment is arrested for simple drug possession, a judge can revoke probation and treatment if
the state has proven “by a preponderance of the evidence that the defendant poses a danger to the
safety of others.” [Sec. 1210.1(e)(3)] Likewise, if an individual violates probation during
treatment and is “unamenable to drug treatment provided and all other forms of drug treatment”--
incarceration is also permitted [Sec. 1210.1(c)(2).]. Both of these provisions permit judges and
prosecutors operating in environments with more punitive views toward drugs to circumvent the
overall intent of the act through broad interpretations (Riley et al., 2001).
In addition, there are opportunities for county prosecutors to divert potential SACPA
clients to incarceration before they even begin the treatment process. Under the initiative, it is

22
possible to divert to incarceration a defendant who is charged with simple-drug possession and
who is otherwise eligible for probation and treatment by charging him with co-occuring offenses
other than the one related to drugs [Sec. 1210.1(b)(2)]
3 Ideally, it would be better to add a length of sentence component to the tough on drugs
measure. Unfortunately, 1993 was the last year the Administrative Office of the Courts collected
statewide data for drug offenders sentenced to prison. Because data have not been collected for
over 11 years at the time of this writing, I have chosen not include this data in the measure.
4 Drug treatment and probation expenditures are gathered from the California Association of
County Governments.
5 Because of multicollinearity concerns between the education (percentage of county residents
who have earned a bachelor’s degree or higher) and county median income variables, these
measures are collapsed to form a “socioeconomic” factor variable which explains 90.2% of the
common variance.
6 Counties are ranked based on the average number of incarcerations per 1,000 persons between
1996 and 1999. Numbers are calculated based on incarcerations for high-level drug offenses
(drug sales, manufacturing, or possession of quantities large enough to presume intent to sell,
and simple drug possessions). Counties in the bottom one-third of rankings are placed in the
“low” category, middle-third “medium” category, and top-third, “high” category.
7 As noted above, several scholars have found that aggregate measures of specific policy
attitudes like the death penalty and abortion explain specific policy outputs above and beyond the
impact of general ideological orientations (Brace et al., 2002; Norrander, 2001). Moreover, in the
context of initiative implementation, Gerber et al., (2001) argue that the political sanctions for
not complying with the intent of an initiative becomes more costly where public support is
higher. Under conditions where public opinion is more supportive of a specific initiative, their
research suggests policy outputs should be more responsive to public preferences. Based on this
research I included in other models not shown here the percentage of voters in each county who
voted “yes” on Proposition 36 as an additional measure of specific attitudes toward the initiative.
The expectation is that those counties with greater support for the initiative will provide higher
quality services and be less likely to incarcerate individuals for low-level drug possession. In all
models the measure failed to reach statistical significance. I believe this is partly due to
multicollinearity between the ideology and the percentage voting “yes” on the initiative where
Pearson’s r correlation equals .71 (p<.01). This may also be explained by the low salience of the
issue (see Sharpe, 1999). In a Field Poll taken only 2 weeks before the election, just over 50% of
those polled said they had never heard of Proposition 36. Because the salience of the initiative is
low, linkages between specific public opinion and outputs may be attenuated.
8 O’Brien’s (1990) generalizability coefficient for the R:A design contemplates the mean square,
an estimate of the population variance between aggregate units, MS(a), and the mean square for
individual-level scores within the aggregated units, MS(r:a), using the formula

23