Laurence
Lynn, Jr., Carolyn Heinrich, and Carolyn Hill
As government programs and policies are decentralized and brought
closer to the people they are intended to serve, policymakers and public
managers are under increasing pressure to ensure performance and accountability
across diverse service units and locations. Performance across sites
and programs can vary for a number of reasons - the characteristics
or needs of the people served, skills of service providers, local site
management, the interpretation of policy directives, the extent of
system-wide coordination, and other structural characteristics. The
question then becomes: how can these programs and policies be organized
and managed to better achieve their purpose?
In a recent Harris School working paper, Studying
Governance and Public Management: Challenges and Prospects,
Harris School Professor Laurence Lynn, Jr., Carolyn Heinrich, assistant
professor of public policy analysis, University of North Carolina
at Chapel Hill; and Carolyn Hill, a Harris School doctoral candidate,
offer a first step in framing a theory-based research model that
produces a fuller and more rigorous understanding of governance.*
The authors argue for the importance of looking
at the big picture when analyzing why certain organizations or policies
are more or less effective, why outcomes can differ across organizations,
and why some organizations or policies are more successful than others.
The "logic
of governance," they argue, is a resultant of numerous interrelated
factors, and looking too narrowly not only is likely to produce
a partial accounting of reasons for outcomes, but also may lead to
a biased partial account. The authors therefore propose a model that
considers a broader, and more dynamic, view of governance regimes.
The model they set forth captures possible associations between various
independent and dependent variables of interest, and encourages researchers
to locate particular theories and models within a more general framework
of possible theories and models.
The Need for a New Model
It is inherently difficult to pinpoint why some policies or programs
are more effective than others. Many factors can be responsible for
the outcomes, and often the multiple factors work together in combination
to effect change. How to tease out those factors that are responsible
for the outcomes has been a goal of research in the public management
field. The approaches taken have varied. Some researchers, for example,
begin by measuring the effects of specific programs or policies on
individual outcomes. From this starting point, researchers can look
at how the structural and managerial context of programs and policies
influence outcomes. For example, centralized administration and goal-oriented
management may produce better program outcomes than decentralized,
broadly focused administration. Further, the choice of a unit of analysis
also affects research findings. Using individual outcomes as a unit
of analysis may produce different estimates of program effects than
using outcomes aggregated by organizational unit or program.
The complexity of governance often forces researchers to simplify
assumptions, use methods that are less suited to the task, or measure
crudely what we know is complex. The model set forth by the authors
encourages researchers to consider the broadest feasible context for
their models and analyses when drawing conclusions from necessarily
incomplete data and information.
Underlying the model is a hierarchy of relationships that ultimately
influences outcomes. This hierarchy consists of relationships between
citizen preferences, political interests, the structure and management
of organizations, the core focus of public agencies, client outcomes,
stakeholder assessments, and, coming full circle, back again to public
and political concerns. Ultimately, this schematic shows how the values
and interests of citizens, policymakers, organizations, and clients
are linked in a dynamic process.
The fact that outcomes can be affected by
a dynamic interaction across individuals, systems, and political
interests calls for theoretical and statistical models that can capture
the complexity. These models must specify and identify causal relationships
between governance and performance. A common approach when studying
governance is to employ the ordinary least squares (OLS) technique.
Unfortunately, OLS models tend to mask the effects of interaction
among variables at various levels of governance. Multilevel modeling,
in contrast, can better investigate hierarchical relationships and
the influences that policy, administrative, and structural variables
might have at the client or constituent levels. Multilevel modeling
allows researchers to examine how factors or variables at one level
of administration might interact with variables at another level.
In multilevel models, the assumption of independence of observations
in the OLS approach is dropped, and relationships in the data are
allowed to vary (for recent research using multilevel modeling, see
Bryk & Raudenbush, 1992; Roderick, Jacob & Bryk,
2000; Heinrich & Lynn, 1999).
Data limitations, however, often dictate method. Recent advances in
administrative data show promise for the study of governance, especially
in the ability to link multiple sets of administrative data from different
departments or agencies. Combining multiple data sources (e.g., survey
data with administrative data; qualitative data with administrative
data) further enhances research.
The Reduced Form Model
Within a governance framework, investigators can explore the determinants
of policy and program impacts without becoming distracted by the alleged
split between policy-level (top-down) and street-level (bottom-up)
explanations of outcomes. The authors propose the following model to
incorporate the multiple levels of variables.
O = f (E, C, T, S, M)
where:
O = outputs/outcomes (individual level and /or organizational outputs or
outcomes)
E = environmental factors
C= client characteristics
T = treatments (primary work, core processes or technology)
S = structures
M = managerial roles and actions.
In other words, the outcomes are a result of the interaction between
environmental factors, client characteristics, treatments, structures,
and managerial roles and actions. Environmental factors could include
political structures, degree of competition, the larger economy, characteristics
of the target population, and legal practices. Client characteristics
are those attributes and behaviors of the clients that may affect outcomes.
Treatments can include organizational mission or objectives, eligibility
criteria for the target population, program scope, and intensity of
services. Structures can include organization type, level of integration,
degree of centralization, administrative rules, budget, contractual
arrangements, and institutional culture or values. Finally, managerial
roles and actions can include leadership practices, staff management,
professionalism, and accountability mechanisms.
This reduced form model is not, in itself, a theory; rather, it suggests
possible associations between various variables. Research that is framed
and interpreted through a logic of governance, as proposed by the authors,
can produce enduring knowledge about how, why, and with what consequences
public-sector activity is and can be structured and managed.
References
Bryk, Anthony, & Stephen Raudenbush. (1992). Hierarchical linear
models: Applications and data analysis methods. London: Sage.
Heinrich, Carolyn, and Laurence E. Lynn, Jr. (1999). Means and ends:
A comparative study of empirical methods for investigating governance
and performance. Working Paper #108. Northwestern University/University
of Chicago: Joint Center for Poverty Research
Roderick, Melissa, Brian Jacob, & Anthony
Bryk. (2000). Evaluating Chicago's efforts to end social promotion.
In Laurence E. Lynn, Jr., and Carolyn J. Heinrich (Eds.), Governance and Performance: Models,
Methods, and Results. Washington, D.C.: Georgetown University Press.
* The working paper was
later published as "Studying Governance and Public Management: Challenges
and Prospects. Journal of Public Administration Research and Theory,
vol. 10, no. 2 (2000): 233-61.
Research Summaries
are designed to help broaden the dissemination of current policy-relevant
research. These Summaries are funded by the Irving B. Harris Graduate
School of Public Policy Studies at the University of Chicago.
For more information, contact Jamie Rosman at HarrisSchool@uchicago.edu or
(773) 702.2287.
