a .. :x......n.4..;..u. .1
a

a... . .. ,. . . . . . , . v. u a ”h.
i . . . . . x . . _ ,1.£3.:¢M
.. . . . . . . . LS. .. .3,
. . . .....u... raga“?

1.31: f. . . . , . . 5F Ir...“ r
__ ..M_....,E . , ...

A

”an, . ##s...

. km" "
Ts . I . . . . . _.
ynr v.17 otv an. kw”. ,.
. $3"an ”a .w‘ ,. .

. ,. .5.

_ . .. 25.5....
.9 .. . :.
4.....- . . PL“.
:2. a I;
cf... .
m.» (In.
. ‘£..vb.vr.o
Hui-s ~vVP.-
: .
2

1% N . , , x26. ...\: a.

.12. 1,.
rgxrl. an...
vs: 3

k .. ‘
tutti.
1.57:. .i
. L . . , . ,
(“W N a: . 1.. .5»
t...” 2 NH. ., 7t . , . . .
V ’7‘. Low.

1
1»
.1. .,:.1u.:
:t‘gl...) i...
)5...- .h

. ‘1.?..a.\.
E. A y?
.9 .

121.. .u.

.. 2.1153. .
.Lr... . .
(flu.-

..i 1......» o...
. . £3..?£.
X 9.37.»...
)3:

Li

... . P ~
1.1:...7.
r. iyrc

.7.» s .

 

, mix—5 L
. n

17

\. .Vx.
5.1. :

 

 

.th Phuwnl.‘

 

THFSlS
2:

.“l I

’1'. \

This is to certify that the
dissertation entitled

PERCEIVH) FAIRIESS IN NATIRAL RESMCE DECISIM "AKIN:
INFUBKIES All) museums

presented by

Patrick Do] i ndo Sui th

has been accepted towards fulﬁllment
of the requirements for

Rh!) degree in Jinan——

 

   

Major professor

’7 /-
Date , ngo

MSU is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

 

 

 

LBIHARY
Michigan State

University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested. ’

 

DATE DUE

DATE DUE

DATE DUE

 

MAY [13 2092

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 chlRC/DateDuepGS—MS

 

PERCEIVED FAIRNESS IN NATURAL RESOURCE DECISION MAKING:
INFLUENCES AND CONSEQUENCES

By

Patrick Delinde Smith

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Forestry

2000

3“]
5L;

COTS

Inak

mea

cord

ABSTRACT

PERCEIVED F AIRNESS IN NATURAL RESOURCE DECISION MAKING:
INFLUENCES AND CONSEQUENCES

By

Patrick Delinde Smith

How can trust between citizens and decision makers be increased? Over the last
century, decision makers have tried a wide variety of participation techniques in an
attempt to increase citizen trust in their agencies and achieve better resource
management. In this study, several cases of natural resource decision making in
Michigan were compared in order to determine what aspects of the decision making
context affect citizen evaluations of the fairness of the process and outcome of decision
making. Psychological theories of procedural and distributive justice were used to create
measures of fairness. Contextual factors tested in the study included the agency
conducting the participation, the characteristics of the citizens involved, and the nature of
the decision making situation in terms of power distribution, intensity of conﬂict, and
participation technique used. Predictions were made as to how citizen evaluations would
inﬂuence the important consequence for the agency of trust in decision makers. Data to
test the hypotheses were collected through 1550 mailed surveys to citizens and 455
surveys to agency decision makers. Descriptive, correlational and regression results
largely conﬁrmed the predicted relationships, showing that perceived fairness is affected
by the decision making context. These results provide clear directions for decision
makers and natural resource agencies that wish to build effective collaborative

partnerships with citizens.

ACKNOWLEDGEMENTS

Many individuals were involved in the successful completion of this dissertation.
Staff in the Michigan Department of Natural Resources Forest Management Division and
the Huron-Manistee National Forest helped me compile the mailing lists and critique the
questionnaire. Fellow students at MSU, particularly Marcello Wiechetek and Christina
Kakoyannis, helped me think through my research and make sure the survey was actually
sent on time. My committee, Larry Leefers, Dennis Propst, and Bernard F inifter were
generous with their time and advice. In particular, my advisor, Maureen McDonough,
was always a source of wisdom, guiding me through the shark-infested dissertation
depths to make sure I emerged with only minor cuts and bruises. Of course, all the
guidance in the world does not sufﬁce if there are no data. I am indebted to the many
citizens of Michigan who willingly (or with a little prodding) shared their experiences
and opinions about natural resource decision making.

Finally, a dissertation depends on social support. Stephanie F ozzi not only
assisted at crucial times with the actual survey, but was also an enchanting companion
who made sure I took time off to play. Finally, thanks to my parents, Richard and Marcia

Smith, and my sister, Dana, who made me what I am today... a Ph.D.!

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES ........................................................................................................... ix
CHAPTER 1: INTRODUCTION ....................................................................................... 1
CHAPTER H: A BRIEF HISTORY OF PUBLIC PARTICIPATION IN NATURAL
RESOURCE DECISION MAKING ................................................................................... 7
American culture and public participation: Justice for all ............................................ 7
The age of professionals: 1900-1960 .............................................................................. 9
The age of participation legislation: 1960-1990 .......................................................... 1 1
The age of citizen-professional partnerships: 1990-2000 ............................................ 15
CHAPTER IH: JUSTICE THEORIES AND PUBLIC PARTICIPATION ..................... 19
Theories of justice ......................................................................................................... 19
Justice and public participation .................................................................................... 23
CHAPTER IV: A CONTEXTUAL THEORY OF PERCEIVED F AIRNESS IN
NATURAL RESOURCE DECISION MAKING ............................................................ 27
A theoretical framework for perceived fairness ............................................................ 28
Fairness and the decision making context .................................................................... 30
Agency context factors ............................................................................................. 3O
Situation context factors ........................................................................................... 33
Citizen context factors .............................................................................................. 36
Elaboration of the framework ....................................................................................... 37
Research questions ........................................................................................................ 39
CHAPTER V: RESEARCH METHODS 1: CASES, SAMPLING, INSTRUMENTS,
SURVEY ADMINISTRATION, NON-RESPONSE BIAS AND ANALYSIS .............. 42
The cases ....................................................................................................................... 42
Sampling frame ............................................................................................................. 43
The instruments ............................................................................................................. 46
Context variables ...................................................................................................... 48
Evaluation of experience variables ........................................................................... 52
Evaluation consequence variable .............................................................................. 53
Survey administration and response rates .................................................................... 53
Non—response bias ......................................................................................................... 58
Coding and data entry .................................................................................................. 66
Analysis ......................................................................................................................... 68
CHAPTER VI: RESEARCH METHODS 2: COMPUTING THE VARIABLES AND
CHECKING RELIABILITY AND VALIDITY .............................................................. 73
Reliability and validity of researcher assigned context variables ................................ 75
Factor analysis and reliability of agency survey context variables ............................. 76
Validity of agency survey context variables .................................................................. 83
Factor analysis and reliability of citizen survey context variables .............................. 87
Validity of citizen survey context variables .................................................................. 9O
Factor analysis and reliability of citizen evaluation variables .................................... 92
Validity of citizen fairness evaluation variables ........................................................... 94
Factor analysis, reliability, and validity of the evaluation consequence variable ....... 95

iv

Summary ....................................................................................................................... 96

CHAPTER VII: RESULTS AND DISCUSSION: COMPARING AGENCIES ............. 99
Cross-agency variation on agency variables ................................................................ 99
Cross agency variation on situation variables ........................................................... 107
Cross agency variation on citizen characteristics variables ...................................... 115
Cross agency variation on fairness evaluation and consequence variables .............. 118
Testing the theoretical framework of perceived fairness ............................................ 120
Summary ..................................................................................................................... 128

CHAPTER VIII: RESULTS AND DISCUSSION: CITIZEN AND SITUATION

INFLUENCES ON FAIRNESS AND TRUST .............................................................. 129
Citizen evaluations of fairness and trust in decision makers ...................................... 129
Participation technique and evaluation variables ...................................................... 130
Conﬂict and evaluation variables ............................................................................... 133
Equal power and evaluation variables ....................................................................... 133
Prior relationships and evaluation variables ............................................................. 134
Citizen characteristics and evaluation variables ........................................................ 135
Summary ..................................................................................................................... 137

CHAPTER IX: RESULTS AND DISCUSSION: THE COMBINED INFLUENCE OF
AGENCY, SITUATION, AND CITIZEN FACTORS ON FAIRNESS AND TRUST 139

Analysis overview ........................................................................................................ 141
Testing the mediation of fair process .......................................................................... 143
The relative inﬂuences of context variables on fair process ....................................... 148
Interactions in the inﬂuences of context variables on fair process ............................ 151
Diﬂ'erences between agency means on fair process ................................................... 153
Testing the mediation of fair outcome ........................................................................ 156
The relative inﬂuences of context variables on fair outcome ..................................... 159
Interactions in the inﬂuences of context variables on fair outcome ........................... 161
Diﬂerences between agency means on fair outcome .................................................. 164
Summary ..................................................................................................................... 165
CHAPTER X: SUMMARY AND CONCLUSION ....................................................... 169
Summary ..................................................................................................................... 169
Study limitations ......................................................................................................... 1 82
Future research ........................................................................................................... 183
Management implications ........................................................................................... 185
APPENDIX A: CITIZEN QUESTIONNAIRE .............................................................. 192
APPENDIX B: AGENCY QUESTIONNAIRE ............................................................. 200
APPENDIX C: SHORT FORM (PHONE SURVEY) QUESTIONNAIRES ................ 208
REFERENCES ............................................................................................................... 214

 

 

 

 

 

la
in-

Ta]

Tat

in-r

Tab

Tabi
Tab

Tab.

103:

LIST OF TABLES

Table 1. Variables and how they were measured. ........................................................... 49
Table 2. Response rates to the mailed and phone follow—up surveys of citizens ............. 54
Table 3. Response rates to the mailed and phone follow-up surveys of decision makers. 56
Table 4. Citizen reasons for not returning survey. ........................................................... 58

Table 5. Comparison of variable means between respondents (resp) and non-respondents
(n-resp) to assess citizen non-response bias ...................................................................... 60

Table 6. Agency reasons for not returning survey. .......................................................... 63

Table 7. Comparison of variable means between respondents (resp) and non-respondents
(n-resp) to assess agency non-response bias. .................................................................... 64

Table 8. Items used to measure citizen knowledge and citizen self-interest. .................. 76

Table 9. Factor structure for perceptions of the beliefs of most other employees on

bureaucracy and expertise variables. ................................................................................ 78
Table 10. Factor structure for fairness importance variables ........................................... 80
Table 11. Factor structure for fairness performance variables ........................................ 81

Table 12. Citizen level correlations among agency variables with sample sizes from 282
to 325. ............................................................................................................................... 85

Table 13. Agency level correlations among agency variables with a sample size of 5. .. 86

Table 14. Factor structure of fair process variables. ........................................................ 93
Table 15. Citizen level correlations among citizen evaluation variables ......................... 95
Table 16. A summary of variables and their validity ....................................................... 97

Table 17. Means, standard deviations and signiﬁcant differences between agencies on
citizen knowledge and citizen self-interest variables ...................................................... 102

Table 18. Means, standard deviations and signiﬁcant differences between agencies on
fairness importance variables .......................................................................................... 102

vi

 

 

 

 

 

Table 19. .\
rm pet

Inc; ‘11 I

us’l’ﬁ UV'

WWII}

Levi
055151011 IT;
Ts: H \
1.1316 bb- o
-' I

. . «h ‘
Cﬁilil‘i Lin;
1's: 1* i
1451’: -.‘. .

ClilZCﬁ CV2

Table 24.
ci‘aluatior

Table 25:
fairness e‘

Table 26;
fairness e

Table 2.“.

ClilZ'L’n C0

Table 28.
makers Vi

Table 39
ms in d

l'ma‘n‘. ~-

mt:

Table 19. Means, standard deviations and signiﬁcant differences between agencies on
fairness performance variables. ...................................................................................... 104

Table 20. Participation techniques used by the agencies. .............................................. 109

Table 21. Means, standard deviations and signiﬁcant differences between agencies on
decision making situation variables. ............................................................................... 113

Table 22. Means, standard deviations and signiﬁcant differences between agencies on
citizen characteristics variables ....................................................................................... 117

Table 23. Means, standard deviations and signiﬁcant differences between agencies on
citizen evaluation and trust variables. ............................................................................. 119

Table 24. Anova analysis of differences between government levels in fairness
evaluations. ..................................................................................................................... 121

Table 25: Agency level Pearson correlations and p-value signiﬁcance levels of citizen
fairness evaluations with agency culture and resources (N = 5 agencies). ..................... 123

Table 26: Agency level Pearson correlations and p-value signiﬁcance levels of citizen
fairness evaluations with situation and citizen variables (N = 5 agencies) ..................... 126

Table 27. Correlations between citizen evaluations, situation context variables, and
citizen context variables. ................................................................................................. 132

Table 28: Model summary of stepwise ANCOVA for dependent variable trust in decision
makers with total fair process evaluation as one of the independent variables .............. 145

Table 29: Standardized beta coefﬁcients of stepwise ANCOVA for dependent variable
trust in decision makers with total fair process evaluation as one of the independent
variables .......................................................................................................................... 145

Table 30: Model summary of stepwise AN COVA for dependent variable total fair
process evaluation ........................................................................................................... 146

Table 31: Standardized beta coefﬁcients of stepwise ANCOVA for dependent variable
total fair process evaluation ............................................................................................ 146

Table 32: Model summary of stepwise AN COVA for dependent variable trust in decision
makers with fair outcome evaluation as one of the independent variables ..................... 157

Table 33: Standardized beta coefﬁcients of stepwise AN COVA for dependent variable

trust in decision makers with fair outcome evaluation as one of the independent variables
......................................................................................................................................... 157

vii

 

 

exalt-:3:-

Table:

fair our

Table 34: Model summary of stepwise ANCOVA for dependent variable fair outcome
evaluation ........................................................................................................................ 1 5 8

Table 35: Standardized beta coefﬁcients of stepwise ANCOVA for dependent variable
fair outcome evaluation ................................................................................................... 158

viii

LIST OF FIGURES
Figure 1. Overview of the theoretical framework of inﬂuences and consequences of
perceived fairness .............................................................................................................. 29

Figure 2. Predicted directions of inﬂuence between variables in the theoretical framework
of perceived fairness. ........................................................................................................ 38

Figure 3: The interaction of power equality with education on total fair process
evaluation. ....................................................................................................................... 1 52

Figure 4: The interaction of age with participation technique on fair outcome evaluation
......................................................................................................................................... 162

Figure 5: The interaction of gender with participation technique on fair outcome
evaluation ........................................................................................................................ 163

ix

CHAPTER 1 : INTRODUCTION

I felt much better about it then, because I felt that someone cared enough to
give me the information. Even after the fact it still made [me] feel good...
If these are the kinds of people who are watching the store, then I have a
greater sense of trust in them running the operation. It’s almost as if I have
a personal relationship with that faceless thing called the DNR.

The above quote, from a focus group participant recalling an interaction with a
forester from the Michigan Department of Natural Resources (Smith, McDonough, &
Mang, 1999), highlights how public participation can lead to greater trust in natural
resource managers. Relationships of trust and respect between citizens and agency
employees are crucial to the relationship-based collaborative management regimes
currently being promoted in the United States.

The use of collaboration reﬂects recent efforts to make government agencies more
responsive and accountable to citizens. This is being done to counterbalance the previous
emphasis on neutrality in government. Over the course of US. history, governments
have emphasized either neutrality or accountability in search of a balance between the
two that maximizes justice (Knott & Miller, 1987). Neutrality and accountability in
decision making can be interpreted as contrasting aspects of justice. Neutrality focuses
on giving all citizens equal, unbiased treatment. In the US. the movement for making
government more professional assumed neutrality would be achieved through the use of
accurate information and expert judgement (Box, 1998). Accountability emphasizes
making sure that government respects the rights of all citizens to participate in decision

making and to have an opportunity to inﬂuence the outcomes. Accountability is

embodied in the practice of democratic governance and citizen participation.

 

 

 

 

\h
f?

Principles of neutrality and accountability have been emphasized to varying
degrees over the 20th century. At the beginning of the century, governments in the US.
focused on neutrality. People largely trusted the professionals who cared for their lands.
As a result, public participation in resource management was rarely practiced (Kaufman,
1960). Management was oriented towards commodity production and economic
development. Direct involvement of citizens was kept to a minimum. Decision makers
were experts who decided the goals and presumed to know how to reach them.

In the 1960’s and 1970’s, citizens became increasingly dissatisﬁed with
governmental policies and demanded a voice in decision making. Numerous mandates
for involving citizens and giving them opportunities to voice their concerns were given to
agencies (Taylor, 1984). In addition to using participation forums such as public
hearings, citizens and environmental organizations sued agencies in court for not
fulﬁlling their mandates to consider citizen concerns. As a result, resource managers
began to give greater consideration to preservation and environmental concerns (Jones &
Mohai, 1995).

In the 19705 and 19805, in response to the growing power of environmental
groups, traditional resource constituencies like timber companies, oil producers and land
developers fought back. Natural resource decision making became a process dominated
by conﬂict (Wondolleck, 1988). In addition, many resource managers resisted the move
towards accountability. Public participation was generally implemented in a formally
correct way that missed the spirit of the new laws, and simply led to ratiﬁcation of
agency decisions (F ortrnan & Fairfax, 1991; Lawrence et al., 1997). The greater

involvement of the courts led to a focus on avoiding lawsuits. This meant that

 

C0

pu

rel;

mar
inCl’I
115 a,

gore

participation processes emphasized formal procedures instead of considering constituent
concems (Taylor, 1984). Participation as tokenism also existed in antipoverty, urban
renewal, and model cities projects (Arnstein, 1977).

Although this situation continued into the 1990’s in many areas, the real need for
collaboration between agencies and citizens continued to grow. Natural resource
managers began attempting ecosystem management, an approach which is holistic and
focuses on the connections between resources (COS, 1999). Ecosystem management
also focuses on natural resources at the landscape level, necessitating coordination across
ownerships (Salwasser, 1994). Cross-boundary management often meant voluntary
commitments ﬁ'om citizen landowners to manage their lands consistently with adjacent
public lands. Achieving cooperative efforts between citizens and agencies required they
trust each other. As agencies facilitated the way forward, they needed to have positive
relationships with citizens.

In order to continue the movement towards collaborative management, resource
managers need to understand why some decision making causes citizens to have
increased trust in the agency, support for its decisions, and improved relationships with
its employees, while other decision making does not. The brief review of the history of
government in the US. suggests that justice issues of neutrality and accountability are
very important. In fact, research on procedural and distributive justice shows that
fairness of the decision making process and fairness of the decision outcome oﬁen
increase trust in authorities and support for their decisions. For example, Tyler and
Degoey (1995) found that residents supported authorities and complied with their rules

during a water shortage in San Francisco. This trust and willingness to undergo personal

 

 

 

jcszi

(Lint

deprivation for the good of the larger community was largely based on perceptions that
the decision making process was fair. Lauber and Knuth (1997, 1998) documented how
perceptions of fair decision making processes in a moose reintroduction decision were
related to satisfaction with the process as well as with the state agency making the
decisions. The consequence of increased support for authorities based on perceptions of
justice has also been demonstrated in many contexts including law, business, and politics
(Lind & Tyler, 1988).

A noteworthy feature of justice theory is the extent to which good participation
processes can fulﬁll many of the important principles of procedural justice. This was
conﬁrmed in a recent study of the Northern Lower Michigan Ecosystem Management
Project (Smith & McDonough, 2001). When asked how they judged the fairness of
natural resource decision making, citizens often focused on characteristics of the process
like adequate participation opportunities, early notiﬁcation, broad representation of
affected groups, and serious consideration of citizen concerns. This is not surprising
because these characteristics all relate to accountability and neutrality. Broad
representation is necessary to make sure that the information collected represents all
views equally, allowing decision makers to try reaching decisions which are unbiased.
Adequate opportunities to participate and early notiﬁcation also help to make sure that
there is equal access to decision makers, increasing neutrality. However, effective
participation is also related to accountability because it makes sure that citizens can
express their thoughts to decision makers and potentially inﬂuence the decision
outcomes. If citizens are able to inﬂuence outcomes this shows the decision makers are

being held accountable to meet the needs of the citizenry. By following effective

 

participation
making Pm:
lites:

win: are the l
question is irr
' fluence ho.“
theory and {he
pote ti 1 mil-u
characteristics
agency place i.
fair experience
making situatn
and citizens ha
process and on
oithe personal
WOT Panicipalt
faintess. In son
he decisions at

areWinnie} inf

participation practices, there is a greater chance that citizens will evaluate the decision
making process and outcome as fair. Greater trust in the agency should be the result.

If research suggests that perceived fairness can increase trust in decision makers,
what are the factors which, in turn, may increase perceived fairness? Answering this
question is important because the circumstances surrounding decision making may
inﬂuence how citizens perceive the decision making. Reviews of social psychological
theory and the history of natural resource management in the US. suggest a number of
potential inﬂuences on fairness. The ﬁrst group of possible inﬂuences concerns
characteristics of the agency making the decisions. For example, if the employees in an
agency place importance on achieving fairness, it is more likely that citizens will have
fair experiences. The second group of inﬂuences on fairness are aspects of the decision
making situation. For example, when the issues being decided are highly controversial
and citizens have intense conﬂict with each other, it is likely that perceived fairness of the
process and outcome will be reduced. The ﬁnal group of inﬂuences on fairness consists
of the personal characteristics of citizens. For example, someone who has a great deal of
prior participatory experience may be less critical of the agency and perceive greater
fairness. In summary, the agency making the decisions, aspects of the situation in which
the decisions are being made, and characteristics of the citizens judging the decision all
are potential inﬂuences on perceived fairness, and thus on trust in decision makers.

The purpose of this study is to understand the role of fairness in inﬂuencing
citizen trust in a decision making context. The ﬁrst objective is to conﬁrm the inﬂuence
of perceived fairness on the consequence of citizen trust in decision makers in natural

resource contexts. In other words, how big an impact does perceived fairness have on

 

 

trust and support fc
is achiei ed. In 0th!
perceived faimcss'.‘
the relationships of

The stud} b
public participation
resource decision it
psychological theor

uses the history in (

 

immeri'orl' relating
fainess to the cons
outlined through a .~
to administer and a:
validity oithe ram
ipresents an lllle‘.

“155 analysis of co.

hone research. and

 

 

trust and support for decisions? The second objective is to better understand how fairness
is achieved. In other words, what aspects of the decision making context inﬂuence
perceived fairness? These objectives will help address the problem of how to improve
the relationships of citizens with the employees and agencies managing natural resources.
The study begins by examining the history of natural resource management and
public participation in the US. A case is made for the importance of justice in natural
resource decision making. Chapter 3 then gives a more detailed accounting of
psychological theories of justice and how these relate to public participation. Chapter 4
uses the history in Chapter 2 and the theory in Chapter 3 to construct a theoretical
ﬁamework relating the decision making context to perceived fairness and perceived
fairness to the consequence of trust in decision makers. Testing of the framework is
outlined through a set of research questions. Chapters 5 and 6 present the methods used
to administer and analyze the surveys, compute the variables, and assess reliability and
validity of the variables. Chapter 7 examines the data at the level of the agency, Chapter
8 presents an individual citizen level analysis, and Chapter 9 combines them in a step-
wise analysis of covariance. Chapter 10 concludes with the summary, study limitations,

future research, and management implications.

 

 

 

 

 

 

 

CHAPTER ll:

\thy is p
LS. yet so pro‘c
in the history of
nation strugglec
accountability.
legislative mar
chapter. nature
the lens of jus
US, a case is

better relatior

American a.
The
because it t
Charactenz
Was the \t'
gorem in.
”will et
gum 10 a

CHAPTER II: A BRIEF HISTORY OF PUBLIC PARTICIPATION IN NATURAL
RESOURCE DECISION MAKING
Why is public participation in natural resource management so important in the

US, yet so problematic and difﬁcult to achieve? The answer to this question lies partly
in the history of the country and its search for justice. Over the years, citizens of the
nation struggled to achieve justice through the goals of governmental neutrality and
accountability. This has resulted in legacies of agency culture, citizen conﬂict, and
legislative mandate which both facilitate and hinder neutrality and accountability. In this
chapter, natural resource decision making in the 20th century is brieﬂy reviewed through
the lens of justice. Starting with an overview of the cultural foundations of fairness in the
US, a case is built for the argument that justice in decision making holds the key to

better relationships between citizens and their governments.

American culture and public participation: Justice for all

The importance of public participation is deeply rooted in the history of the US.
because it taps into the cultural value of justice. The US. is a nation of immigrants,
characterized by a frontier spirit of independence. A deﬁning moment of cultural identity
was the War of Independence against England in which the settlers fought for the right to
govern themselves in a democracy. They fought to create a society in which “all men are
created equal” and established an elaborate system of courts to ensure that justice was
given to all. To this day, the neutrality of the court system in the US. remains a powerful

belief (Ewick & Silbey, 1998).

 

 

The (lCSli'.

toxerrntent oper:

 

 

 

males decisions r
interests get is ha:
often maintained
being “captured"
is concerned it till
Governments uh
violate their delt
is designed to cor
mechanisms fort
elections. public
QQV ﬁtment do
‘3“: \0 \' all; :
citizens,
The
as the gov
(ii the C80
Dermal bu
pain C0m:
Iii/755,57]; [f

honest, CORGI:

The desire for justice has created a philosophical tension in which US.
government operates. On the one hand, justice requires a neutral government which
makes decisions without bias. Overtly political decisions in which people with one set of
interests get what they want violate the sense of fairness. In practice, governments have
often maintained a somewhat distant relationship with citizens to avoid the appearance of
being “captured” by special interests (Knott & Miller, 1987). On the other hand, justice
is concerned with the protection of basic human rights such as freedom of speech.
Governments which become authoritarian and disconnected from their citizens can
violate their basic rights and treat citizens unjustly. The democratic form of government
is designed to counteract the tendency towards authoritarianism by providing
mechanisms for citizens to control and regulate their own government. In addition to
elections, public participation has become a way for ordinary citizens to inﬂuence
government decisions which affect their lives. Thus governments concerned with justice
have to walk a ﬁne line between maintaining neutrality and being accountable to their
citizens.

The history of natural resource management in the US. shows pendulum swings
as the government alternately emphasized neutrality or accountability. In the beginning
of the century, widespread perceptions of political party patronage led to calls for a
neutral bureaucracy in which employees were hired based on their expertise, not their
party connections. In the 19605 and 19703, citizens felt the government was not
respecting their basic rights and had gotten out of control. In response to widespread

protest, congress mandated citizen participation in many government agencies. The end

 

 

oitiecentur} h.

neutrality in the

The age ofprot}
The us
making got err;
patronage syste
goi'ernrnent tor
appornted boar
lhlS. local gm:
Counctl~pr0fcs
DEpression er;
created and 51;
“IE philosoph
Society
Dunn;
ullt‘epsn} -ba<
‘ROblnson, 1g
foregim' CTEt
Ulllzianan Use
SUCl] a; limbe

Europe deSl'ﬂ
M An‘

of the century has found agencies struggling with how to integrate accountability and

neutrality in their decision making.

The age of professionals: 1900-1960

The ﬁrst half of the twentieth century was characterized by an emphasis on
making government more professional. This was in response to a party and political
patronage system which had led to widespread corruption and inefﬁciencies in
government towards the end of the nineteenth century (Box, 1998). Large numbers of
appointed boards led to fragmentation of power and lack of coordination. In response to
this, local governments moved to a strong mayoral model and then ultimately to the
council—professional manager structure. At the federal level, especially during the
Depression era, government grew larger and more centralized. New agencies were
created and staffed with a professional civil service dedicated to efﬁciency and neutrality.
The philosophy advocated attaining a community-wide vision, the greater good for
society.

During this period the Forest Service was created as were the ﬁrst professional
university-based schools for training the foresters needed to staff the new agency
(Robinson, 1975). Gifford Pinchot, the ﬁrst chief of the Forest Service, proclaimed the
foresters’ creed to be “the greatest good for the greatest number.” Emphasizing the
utilitarian use of the national forest lands for water production and consumptive products
such as timber and grazing, he brought scientiﬁc forest management principles from
Europe designed to achieve a sustained yield of products over the long term. This federal

model of forest management was also adopted by many states. Foresters at the state level

 

 

 

 

 

 

 

had sin

ratersl

lrrrr'.)
hut.»
1

Pt at
v'bll
, .
Toe:

Willi

SCSI

“*-

(It?

had similar training and similar objectives of sustained production of timber and
watershed protection.

The role of citizens in this new forestry was small. Public participation was
limited to public relations and federal foresters were frequently transferred to prevent the
development of potentially collusive relationships with local people (Kaufman, 1960).
Foresters were portrayed as professionals working for the broader good of society, in
contrast to publics who were motivated by narrow self-interest. Thus, agencies did not
see much point in involving citizens in natural resource decision making.

These beliefs were also present at the local level, although in a more complex
way. Zoning and planning became widespread practices for governing the use of urban
land resources in the face of industrialization and the resulting rapid development of
urban areas. Although zoning efforts were sometimes used to keep ethnic and racial
groups separate, they still revolved around the use of the land resource (Molotch, 1976).
Once the profession of urban planning was created, planners worked with city
government to contain development in an orderly fashion. This was often done in the
service of what was termed the “growth machine,” a group of business elites who stood
to proﬁt from land development.

Citizens could and did become involved when the desires of the growth machine
conﬂicted with their own needs for neighborhood amenity maintenance (Steele, 1987).
In contrast to federal and state forest management, which often occurred in remote
locations, urban land management was done in citizens’ back yards. From the beginning,
citizens had motivation to demand voice in local planning and zoning. However, at this

early period, the balance of power oﬁen lay in the professionals’ hands.

10

 

 

 

 

 

 

 

the 7-
un ll

dent

no.

citiz

At all three levels of government, the professional management philosophy
profoundly affected the relationship between citizens and managers. As Box explains:
Many practitioners have been trained to maintain a separation between themselves
and the public, because they are the experts and citizens are the “customers,” people
who know little about public services except the end products they receive and the
fees or taxes they pay. (1998, p 145)
The concept of citizen as customer, someone who lacks knowledge about how to manage
the resource, represented a profound change from earlier models of J effersonian
democracy in which citizens were assruned to be able to participate directly in
governance (Pateman, 1970). J effersonian democracy had emphasized direct
involvement partly in order to achieve the aspect of justice related to the protection of
citizen rights and the assurance of government accountability. The newer government
philosophies stressed instead the aspect of justice concerned with neutrality. At the end
of the 19505 and beginning of the 19605 citizens demanded a return to a more
Jeffersonian approach. People began to question the ability of government to protect
their rights and demanded a more accountable, less authoritarian government. They

wanted a greater role than that of passive “customer.”

The age of participation legislation: [960-1990

The 1960’s and 1970’s were a period of social upheaval in the United States,
marked by the civil rights movement, the environmental movement, and anti-Vietnam
War protests. Publics demanded a greater voice in their government and Congress

responded to these concerns with sweeping new legislation that mandated citizen

ll

participation at local and federal levels. During the late 19705 and 19805, agencies and
natural resource managers struggled to learn how to implement these laws.

At the federal forest management level, both the National Environmental Policy
Act of 1970 (NEPA) and the National Forest Management Act of 1976 (NF MA)
mandated a planning process which included public involvement (F rome, 1984). These
Acts had to be followed by national forest planners, but did not apply to state level
managers, setting up a potential difference in the amount of citizen involvement.

During this period, national forest management goals were dramatically reshaped
by the new inﬂuence of environmental groups. Through lawsuits based on NEPA and
NFMA as well as other new laws like the Endangered Species Act of 1973, production of
traditional products like timber declined, while other uses such as recreation and wildland
preservation increased (Jones & Mohai, 1995). This introduced a new intensity and level
of conﬂict between different user groups that often played itself out in public
participation arenas (Culhane, 1981).

The new requirements for participation were implemented by the Forest Service
through standardized techniques based on the requirements of NEPA (USFS, 1977). The
ﬁrst step was scoping, in which all the affected constituents, particularly people living
near to the management area, were told of the impending decision and given an
opportunity to share their opinions. Then a set of management alternatives was internally
developed and made available for comment during a deﬁned public comment period.
Public comments were collected through written comments as well as in formal public
hearings. Collecting and using this new inﬂux of information was a challenge for

professionals accustomed to making decisions autonomously.

12

Local level decision making was also affected. In 1966, the Demonstration Cities
and Metropolitan Act provided funding for community programs with “widespread
citizen participation”. The Housing and Development Act of 1974 required that cities
“provide citizens with an adequate opportunity to participate in the development of an
application” (Pollack, 1984). Because much local level urban planning centered on
funding from these Acts, local governments were forced to implement citizen
participation strategies. The new federal funding requirements necessitated that
municipalities conduct comprehensive citizen participation when doing planning
(Pollack, 1984). This created an environment in which citizen concerns had to be taken
more seriously. Another change was that planners were pressured to preserve open space
and to consider issues like water quality when making decisions (Steele, 1987).

The new-found importance of citizen participation had an effect on public-
manager relationships at all levels of government. Many managers saw the newly
empowered publics as troublemakers who upset the status-quo system and challenged
their professional authority. Other managers identiﬁed with the new democratic ideals
and saw citizen involvement as a way to do things better. A regional forester looked
back on that era and surmnarized the varying attitudes towards the new role of citizens:

Some administrators in the past have felt public involvement was cumbersome and
a stumbling block in the way of efﬁcient management. That I can’t accept. It has
proven a great beneﬁt to the national forests of the northern region. Participating
individuals and groups help Forest Service personnel make better decisions and
conﬁrm the democratic approach to public land management. National forests

don’t belong to just those working in the Forest Service- they are part of our
national heritage (Frome, 1984, p. 96).

A5 a result of the varying views of professionals, it is not surprising that

implementation, as well as citizen satisfaction, was mixed (F rome, 1984). In a review of

13

 

 

 

 

 

 

 

 

loud I
tanzcu

oi can;

participation in National Forest planning processes during the 19805, the Forest Service
found that 95 percent of respondents indicated there was compliance with the public
participation section of the regulations (Russell, et al., 1990). However, only 3 percent
of citizen respondents felt their comments had inﬂuenced planning decisions. This
suggests that citizen involvement may have been viewed as input to be considered by
decision makers, but was not important or valid enough to affect actual planning
decisions. Fortman and Fairfax (1991) stated that public involvement in forest planning
was often used to ratify agency decisions. Ample evidence of participation as tokenism
also existed at local levels in antipoverty, urban renewal and model cities projects
(Amstein, 1977).

In addition, public participation was primarily one-way communication (Blahna
& Yonts-Shephard, 1989) and often culturally inappropriate (McDonough, 1991, 1994).
For example, most opportunities for public input into forest planning were structured as
large public meetings advertised through standard media. The use of these techniques, in
addition to excluding non-traditional users, tended to intensify rather than resolve the
underlying conﬂicts between user groups (Wondolleck, 1988).

In summary, from the beginning of the century through the 19805, natural
resource decision making went through a dramatic transformation that left its mark on
agencies and citizens. The early desire for a more neutral government led to agency
policies and the promotion of a cultural of professionalism which served to prevent
relationships between citizens and resource managers. As the resource use preferences of
citizens changed, agencies did not adjust, leading to citizen demands for greater

accountability. Although legislation followed which required citizen participation,

14

 

 

 

 

 

 

The

roll

C01

alt

ho

sit

or

"(J
H

"‘1

'1.

r.)

'7.)

agencies’ cultural and structural factors created in the service of neutrality became
barriers to agency accountability. Participation was implemented as required, but in a
way that did not always lead to accountability. However, the changes had begun, and as
the century came to a close, recognition of the need for active citizen involvement in
natural resource management led to the use of more accountable and creative decision

making approaches.

The age of citizen-professional partnerships: 1 990-2000

The present era has seen the wider adoption, at least in principle, of a more active
role for citizens. The Forest Service has experimented with varying techniques of
conﬂict resolution and agency-citizen partnering. Some states have also tried this,
although Michigan has focused on traditional techniques like advisory boards, open
houses, and public meetings (McDonough & Thorbum, 1997). Local level planning also
still largely relies on traditional public hearings and written comments, although
particular localities have experimented with community Visioning processes. Within the
profession of local managers, there is a growing movement towards what Kemmis (1990)
has termed “barn-raising.” According to this philosophy, citizens are encouraged to take
responsibility for their communities. One piece of evidence for the changing role of
citizens in local planning is the commitment of Public Management, a leading
practitioner journal aimed at local government, to publishing more on empowering
citizens (Benest, 1996).

The Forest Service recently adopted this philosophy of cooperative management

with the second commissioning of the Committee of Scientists. The committee prepared

15

 

 

 

 

 

 

 

 

 

l99‘h

lt’US

rm‘
inf»

CSCE

C031

a set of recommendations, many of which may be codiﬁed into new regulations (COS,
1999). The report suggested that planning needs to be adaptive, collaborative, and
focused on building stewardship capacities in communities. One of the key organizing
principles of this new approach was ecosystem management. At both the federal and
state level, the older ideas of multiple use are being augmented with new goals that
emphasize maintaining ecosystem processes and functions. Because humans are also
seen as part of the ecosystem, their participation in management decision making and
implementation on private lands remains essential (Smith & McDonough, 1999).

These proposals for planning are hardly new ideas. For the last decade some
forest managers have tried new methods of planning and public participation. The
Hoosier National Forest has been involved in conﬂict for many years about recreational
use of the Deam Wilderness (Slover, 1996). Horse riders and non-riders were involved in
escalating conﬂict over trail use and managers’ traditional approach of gathering
comments and proposing solutions was not working. A new approach of collaborative
planning was tried in which these groups and many others were brought together to
focus, not on their differences, but rather on their shared desire for good management of
an area special to them. Sinnon, Shands and Ligget (1993) have described this type of
process as creating a community of interest. Numerous workshops, ﬁeld trips and
discussions resulted in a comprehensive plan to which all agreed. Volunteers from the
groups are now helping make the plans a reality (Slover, 1996).

Shannon (1987) reported that a new responsive management style is emerging
within the Forest Service. This approach requires a balance between participatory

planning and technical competence. Public participation involves bringing in parties and

16

 

 

 

 

creati r
with he :

relation“:

0; .
conlllkl

g.
ﬁﬁ9.‘~ .
ckiuAILt:

creating new relationships between them. The old model can be described as a wheel
with the agency as a hub and each party communicating directly to it. The new set of
relationships resembles a wheel with no hub and all the parties communicating directly
with each other in a web of spokes. The agency takes the role of strategic thinker,
facilitator, and knowledge resource (Shannon, 1992). The ability of citizens to understand
and deliberate on technical issues is respected (Doble & Richardson, 1992) and the
decision making process is no longer a black box.

Another trend utilizing more collaborative approaches is that of alternative
conﬂict resolution. Codiﬁed in the Alternative Dispute Resolution Act of 1990 as an
appropriate response to management conﬂicts, altemative conﬂict resolution focuses on
using a variety of creative approaches outside of the courtroom to ﬁnd solutions to
conﬂicts between agencies and citizens as well as among citizens. Success often hinges
on the ability of conﬂicting parties to understand opposing interests and so be willing to
ﬁnd compromises and creative solutions that address the underlying cause of the conﬂict.
Although not practiced much by the Forest Service (Schumaker, O’Laughlin, &
Freemuth, 1997) it has been used in a few cases (Wondolleck, 1988) and shows promise.
These practices have also been extensively developed by the Army Corps of Engineers
(Priscoli, 1990) and include negotiation, facilitated collaborative problem solving,
mediation, conciliation, mini-trial, and arbitration.

In summary, the history of public participation in natural resource management
shows an evolution in the roles and relationships of managers and citizens driven by the
conﬂicting demands for neutrality and accountability. The ﬁrst half of the century saw

the rise of professional resource managers who were generally trusted by citizens to

17

 

 

 

 

C4;

'YJ

provide for the greater good of the citizenry. The agencies emphasized neutrality in
decision making and kept an image of being distant from citizens. During the turbulent
period of the 1960’s and 1970’s, citizens became disillusioned with professional
management, began to mistrust government, and demanded and received greater direct
participation in governmental decision making. The 19805 were a time when agencies
grappled with and institutionalized citizen participation. The twin justice goals of
neutrality and accountability were difﬁcult to achieve simultaneously. For this reason
participation often did not have effects on management decisions, frustrating citizen
participants and failing to restore citizen trust in government. In the 19905, new
approaches have been tried which attempt to reconcile professionalism and citizen
involvement by emphasizing collaboration. The hope is that by increasing agency
responsiveness the mistrust between managers and publics which arose in the 1960’s and
1970’s will be reduced.

One of the keys to understanding this history of citizen involvement in natural
resource management is the concept of justice. Justice has been described in rather
simple terms as pertaining to neutrality and accountability, an approach consistent with
the political science literature. However, justice has also been explored from a
psychological perspective. Social psychologists have conducted detailed investigations
of why justice is important to people and how they assess it. Theories about the fairness
of the process used to reach decisions as well as the decision outcomes have been
developed and extensively tested. The next chapter outlines the psychological

perspective on justice.

18

CHAPTER III: JUSTICE THEORIES AND PUBLIC PARTICIPATION

How do citizens evaluate decision making processes and outcomes? The
historical review in the previous chapter suggests that citizens in the United States expect
their governments to be neutral and accountable when they make decisions. Neutrality
and accountability are aspects of justice. It is no surprise that researchers and theorists
have developed psychological theories explaining how and why people often search for
fairness in decision outcomes as well as processes (Lind & Tyler, 1988). The theories
propose extensive lists of fairness principles which derive from deep-seated needs for
dignity and social status.

In this chapter, psychological theories of justice are reviewed in order to
understand more clearly how people form fairness perceptions, and why these
perceptions are so important. These theories are then related to existing public
participation conceptualizations. The claim is made that justice theory is a
comprehensive approach which subsumes many of the key decision making prescriptions

found in writings about public participation.

Theories of justice

How do people evaluate the fairness of a decision? Social psychological theories
and research suggest that both the decision outcome and the process by which that
decision is reached affect perceptions of fairness (Hegtvedt & Markovsky, 1995).
Concerns related to decision outcomes have been termed distributive justice because they
relate to how beneﬁts and costs are distributed among people. An equivalent term would

be outcome fairness. In this discussion, justice and fairness are used interchangeably.

l9

Concerns related to decision making processes have historically been termed procedural
justice, interactional justice, and procedural fairness. The literature is mixed as to
whether or not there is need for a distinction between procedural and interactional justice
(F olger, 1993; Lind & Tyler, 1988). For the purpose of providing an overview of the
ﬁeld, all three terms are taken as equivalent and used interchangeably.

Another key aspect of j ustice theories is that fairness is conceptualized as being
perceptual and subjective. In other words, fairness is in the eye of the beholder.
However, this does not mean that a person will always see something different from
another person when they experience the same event. They may come to largely similar
conclusions about the fairness of a situation because they share a common perspective,
background and expectations. In fact, justice research suggests that there is a shared set
of principles people use when judging the fairness of the process and outcome of
decisions.

Early research on justice focused on decision outcomes. Three principles of
distributive justice emerged: equity, equality and need (Hegtvedt, 1992). Equity, which
received most of the research attention, is based on the concept that everyone should get
rewards in proportion to their efforts or costs (Homans, 1961; Walster et al., 1973). In
contrast, equality requires that everyone beneﬁt equally regardless of costs or efforts.
Finally, need requires that people receive beneﬁts according to their needs, either because
they are deﬁcient relative to others or because they have a need for greater resources than
others.

In 1975, Thibaut and Walker published a classic book which established a new

focus for justice research: procedural justice. Procedural justice focuses on the fairness

20

of the process by which decisions are reached. Thibaut and Walker’s theory and research
emphasized direct participation in the decision and the opportunity to voice one’s opinion
as key principles by which people judge the fairness of a process. Soon after, Leventhal
(1976; Leventhal et. al., 1980) added to this by proposing six principles for procedural
fairness: l) consistency over persons and across time, 2) suppression of personal self-
interest (bias), 3) use of accurate information, 4) modiﬁability of decisions, 5)
representativeness of the concerns of all recipients, and 6) adherence to prevailing ethical
and moral standards (ethicality).

Lind and Tyler (1988) developed a model which seeks to explain the underlying
processes implied by these procedural justice principles. Their group value model
suggests that people are concerned about their social standing in society and these
principles give them information about how others evaluate them. Their model posits
three basic components of procedural justice judgments: 1) neutrality, 2) trust in the
benevolent intentions of decision makers, and 3) status recognition. Neutrality is the
belief that decisions are based on a complete and accurate assessment of the facts and
implies that over time everyone will receive fair outcomes, even though this may not be
so in the short term. Intentions are important because, to the extent that authorities are
benevolent, they can be trusted to make decisions that are fair. Status recognition
involves treatment that enhances the person’s sense of dignity and reﬂects that he/she has
good standing in the group.

The group value model can incorporate many of Thibaut and Walker’s and
Leventhal’s principles. Lind et al. (1997) found that much of the variance in procedural

fairness attributed to voice was explained by group value variables, lending support to the

21

 

hypothesis that being able to express oneself gives one a sense of dignity. Leventhal’s
principles of consistency across persons and time, bias suppression, accuracy,
correctability, and representativeness all may contribute to the neutrality of the decision
process because they help to insure that accurate and complete information is used.

These principles also communicate status information. For example, a strong bias against
a person suggests that the person has no status and that the authority does not have
benevolent intentions.

Folger (1993) developed an even more basic explanation for justice judgments.
He integrated distributive and procedural judgments by positing that people are primarily
concerned about being treated with dignity. Outcomes which meet distributive fairness
principles or processes which meet procedural principles suggest that the people are
being treated as ends in themselves, rather than as a means to something else. F olger
suggests that seeing oneself as an end is central to a feeling of dignity.

F olger’s attempt to integrate procedural and distributive concerns reﬂects a long-
standing observation that the two interact. It has been shown repeatedly that when
outcomes are favorable or fair, procedural justice has small or no effect on participant
reactions. However, if outcomes are unfair, procedural justice works to make participant
reactions more positive (Brockner & Siegel, 1996). An opposite phenomenon has also
been demonstrated. When procedures are unfair, fair outcomes can make participant
reactions more positive (Van den Bos et al., 1997).

Fairness heuristic theory (Lind et al., 1993) can be used to explain the interactions
between procedural and distributive justice. Initial faimess impressions set the stage for

interpreting and evaluating subsequent experiences. They do so by leading to judgments

22

of the benevolent intentions, neutrality, and respect for a person’s status by the authority.
These concerns could be summarized as the “trustworthiness of the authority.”
Procedural justice works by providing assurance that long-term outcomes will be just.
Participants can trust that the process will eventually beneﬁt them. High procedural
faimess and high distributive fairness also show that authorities can be trusted to continue
treating participants with dignity. This interpretation is supported by numerous studies
showing how trust leads to greater procedural justice (e. g. Tyler, 1994) and how trust in
authorities is a consequence of procedural justice (e. g. Tyler, 1990; Tyler & Degoey,
1995).

In summary, justice theories have proposed a wide set of fairness principles.
Outcome fairness principles include equity, equality and need. Procedural fairness
includes the importance of voice, broad representation, lack of bias, use of accurate
information, and direct control over decision processes and outcomes. These principles
are deeply important to people because they communicate the level of respect and social
standing they are accorded by decision makers. This in turn affects their sense of self-
worth and dignity. The deep-seated nature of these principles explains why fairness

judgements can have large impacts on support for decisions and trust in decision makers.

Justice and public participation

As the preceding section explains, justice theories suggest a set of principles,
grounded in basic human motivations, that can be used to evaluate a decision making
process and outcome. However, justice theories have only recently been applied to

natural resource decision making. In contrast, an extensive body of writing related to

23

citizen involvement in natural resource decision making has developed over several
decades. These writings have also proposed various criteria for evaluating how citizens
are involved in decision making. The inﬂuence of these writings on natural resource
decision making practice suggests that they be compared with justice theories to
determine commonalities and differences.

Classically, citizen participation in decision making has been thought of in terms
of power and in terms of democratic philosophy. Arnstein’s (1977) “ladder of citizen
participation” outlined 8 rungs of citizen power, ranging from manipulation to
consultation to citizen control. Another way to analyze these rungs is that, as power
increases, citizens have more voice and more participation in decisions - the two
procedural justice concerns documented by Thibaut and Walker (1975). In contrast to the
power approach, Laird (1993) derives evaluative principles from democratic philosophies
of direct participation and pluralism. He suggests that participants need to learn and
become informed, participation should be as broad as possible, and decision making
power should be shared. These last two are essentially identical with Leventhal et al.’s
(1980) fairness principles of representativeness and participation in decisions.

Another approach that is more explicitly congruent with justice theory is based on
J urgen Habermas’ linguistic/philosophical theories for a fair and competent ideal speech
situation (Renn et al., 1995). A fair situation requires that anyone interested must be able
to attend, initiate discourse, participate in discussion, and inﬂuence the collective
consensus decision. A competent situation requires access to information and its

interpretations, and the use of the best possible procedures for selecting knowledge.

24

As an exploration of Habermas’ theory, Tuler and Webler (1999) and Webler and
Tuler (forthcoming) conducted a qualitative study of participants’ evaluations of the
decision making process in a forest advisory council. Their results support Habermas’
theory and the social psychological theory of justice presented here, suggesting that the
theories are very similar. For example, ability to attend is related to the
representativeness of participants involved. Ability to initiate discourse and to participate
in discussion is another way to express the concept of voice. Inﬂuencing the collective
decision relates to participation in the decision by having control over it. They also found
some evaluative criteria not present in Habermas’ theory, but which are suggestive of
justice theories. For example, respect, openmindedness, honesty, understanding,
listening, and trust are suggestive of Lind and Tyler’s (1988) group value theory because
they are signs of good group standing and benevolent intentions.

In summary, it appears that many justice principles are implied by recent writing
about public participation. An advantage of justice theory is that it has a strong empirical
base. It already includes some of the principles which Webler and Tuler (forthcoming)
demonstrated empirically, but could not ﬁt neatly into Habermas’ theory. Justice theory
also adds new principles like ethicality and neutrality, making it a comprehensive
framework within which public participation ﬁts as a tool to help achieve perceived
fairness. Lawrence et al.’s (1997) call for a focus on procedural justice when doing
natural resource public participation supports this approach.

The purpose of this study is to examine justice theory in a natural resource
decision making context to determine how the context shapes fairness judgements and

how those judgments inﬂuence citizen trust in decision makers. In the next chapter a

25

theoretical framework is proposed which argues that fairness increases citizen trust in the
decision makers. The framework also identiﬁes how fairness judgements may be
affected by the decision making context. The history of public participation reviewed in
Chapter 2 is used to create a set of factors which together deﬁne the decision making
context. Social psychological literature in areas like conﬂict resolution and
organizational culture is used to more clearly specify the nature of the bivariate
relationships between each context factor and fairness. Finally, a set of research

questions designed to test the theoretical framework is presented.

26

' I. .i .
1 ill “Mr P slung!“-
\(' ml) «imam .

 

 

 

 

{IIZQ/

 
 
  
  
 
  
  

“him (
be lﬂiohe
dEIEMIT): “f7,

be 11335th a}

and crux \O

l ‘ ‘ h],
/
/ I
I" c'.
lr/l’lﬂj/r ,

all j! ./
’1‘! 'rﬁ'r
‘ ‘ l i "1‘ A
l ”"1”!!de

rp lilli

 

/
,r

CHAPTER IV: A CONTEXTUAL THEORY OF PERCEIVED FAIRNESS IN
NATURAL RESOURCE DECISION MAKING

The importance of fairness and the well-developed nature of j ustice theory
suggest that fairness principles can be used to measure citizen evaluations of natural
resource decisions and decision making. Given that citizens will perceive different levels
of fairness, what are the reasons? Many decision makers would like to know why certain
decisions and decision making processes lead to positive citizen responses and others do
not. One way to identify possible reasons is by examining the historical decision making
record. Chapter 2 laid out a history of public participation in natural resource decision
making. Close examination of this history allows the development of a set of factors
which combine to describe the variety of decision making contexts in which citizens can
be involved. Fairness principles can then be used to compare the various contexts to
determine which are associated with higher levels of perceived fairness. Conclusions can
be reached about which contextual factors have the greatest impact on perceived fairness
and trust in decision makers.

In this chapter a theoretical framework of the inﬂuences on and consequences of
perceived fairness in natural resource decision making is proposed. The inﬂuences on
fairness are a set of context factors derived from the brief historical review in Chapter 2.
The consequence of fairness is trust in decision makers. Once the theory has been
explained, a set of speciﬁc research questions guiding the analysis is presented. The
questions are designed to guide the testing of the contextual theory of fairness proposed

in this chapter.

27

 

 

 

 

'0
it?

decision ,
"513' man.

1213’) ”II/Warmer c‘.

brows the process‘
mwswv; ws‘xw
‘g\\\\\'~§&\\'&3
“wages.

Th1

/ /
1'1 ,.
“Minds. 1

Mm: .

,1. . it ,
ii‘t‘ ." '
“Whiz will.“

“ll

 

A theoretical framework for percei ved fairness

The centerpiece of this study is a theoretical framework which speciﬁes how
context factors affect fairness and how fairness affects trust in decision makers. The
theoretical framework has three main sections (Figure 1). The ﬁrst part of the framework
consists of the contextual variables derived from the review of the history of public
participation in natural resource management. These factors, grouped into those related
to the agency, the situation, and the citizens, all hold the potential to cause citizens to
evaluate their participation experience as being more or less fair.

The second part of the framework consists of the ways citizens evaluate a
decision making process. Justice theory suggests that fair outcomes and fair process are
very important to citizens. Although process and outcome judgements are different, they
may inﬂuence each other, particularly if the person only knows the outcome or only
knows the process (Van den Bos et al., 1997). As a result, process and outcome fairness
are usually positively related. The framework proposes that the decision making context
will have a large inﬂuence on citizen evaluations of the fairness of their decision making
experiences.

The third component of the framework is the consequence of the citizen
evaluations. Decision makers would like to be trusted by citizens. Citizen trust is
manifested in citizen support for decisions, citizen satisfaction with the agency making
the decisions, and good relationships between citizens and decision makers. The
literature review suggests that trust of decision makers increases when citizens evaluate

decision making processes and decision outcomes as being fair.

28

MS

 

.ﬁl’ltrl /
iii’t-l”*"’“7""l
tr till limit

I

, lid; 4
lbw/till.” ill lrlm,‘ , ,

 
  

-i .,.- ..
mil their halter

Decision Making Context:
Inﬂuences on Citizen
Evaluations

Agency

 

 

Agency culture
Level of government
Agency resources

 

 

Situation

 

 

Participation technique
Amount of conﬂict

Equal power distribution
Prior respectful relationshJirs

 

 

Citizen

 

 

Prior level of involvement in
decision making

Age

Gender

Education

 

 

Citizen Consequence

 

 

Evaluation of of Citizen
Experience Evaluations
Fair process Trust in

—’ decision
Fair outcome makers

 

 

 

 

 

 

Figure 1. Overview of the theoretical framework of inﬂuences and consequences of

perceived fairness.

The framework assigns a central role to perceived fairness by hypothesizing that

the inﬂuences of context factors on trust are mediated by fairness. Mediation means that

the inﬂuence of one variable on another variable is transmitted through an intervening

variable. The mediating role of fairness in this model is based on the idea that the context

factors are related to a speciﬁc experience. Citizens then evaluate that speciﬁc

experience in terms of fairness. The fairness of that particular experience may then

change their beliefs about the decision makers.

29

p

mt'Oli‘c’

oqerate “71h”

$S\§ ‘mﬁ s

\‘m§_\§
‘ \

\

\wséi“

\‘ “

lit?”
Mil ll-

if
to
"Which-til

    
 

in - . H
. arr-'ririith-

4m

Fairness and the decision making context

The framework described above and depicted in Figure 1 includes a broad array
of context factors which are predicted to affect fairness. In this section the inclusion of
each context variable is justiﬁed in terms of the history of natural resource management
presented in Chapter 2 and social psychological literature. The context factors can be
divided into three main types: the agency making the decisions, the situation in which

decisions are made, and the citizens evaluating the decisions.

Agency context factors

The agency making the decisions has the primary control over the process used to
involve citizens. The agency also hires certain types of employees and rewards certain
types of employee behaviors and attitudes. Agencies create their own policies, have to
operate within legislative boundaries, are given authority only over certain types of
issues, and deal with a particular set of constituencies. Different natural resource
agencies would vary on these and other dimensions. In this study, the three particular
agency dimensions explored are the level of government, the agency culture, and the
resources available for public participation

Level of government: Natural resource agencies can exist at local, state, or
federal levels. Level of government may be important because the geographic scope of
affected constituencies differs dramatically among various levels. Federal agencies are
accountable to the nation as a whole and have to balance local, state, and federal
constituencies. State agencies are accountable to the state and its population. Local

government is only directly accountable to local people. This suggests that notiﬁcation

3O

 

next
theprer

mow/Ma

772: fiﬁ
sass“ Q\\.\\\‘
‘ m\\\\“ <:

651$

"t’ljlt‘lolii ,4
'{L/l l (N

l

   

lira! .ﬁlriﬂlﬂﬂ-
vmrr'ﬁ/ZMJ/v ,
/m

\WKT I A I

“a“

\\\
has
“\K r

and adequate involvement should be easier at the local level. Decisions may also be
more closely tailored to citizen desires because there may be less diversity of opinion. It
may also be more convenient and less intimidating for citizens to be involved at a local
level. Because notiﬁcation and involvement are related to the fairness principle of
representation, and because tailoring decisions to citizen needs is related to principles of
inﬂuence and fair outcomes, perceived fairness may be higher at local levels.

However, local planning commission decisions affect neighborhood
characteristics which may invoke deeper commitments from citizens than would issues of
forest management in a distant forest. The decision of whether or not to build a Walmart
next door has a huge impact on the day-to-day lives of nearby residents. In contrast to
the previous arguments, this suggest that citizens at the state and federal level may be less
critical and have higher fairness perceptions.

The ﬁnal aspect of level of government is the legislative environment. The
history outlines that federal regulations required both local and federal agencies to
involve citizens. However, these regulations did not affect states. Thus one would
expect less effort to be responsive to citizens at the state level and therefore more
negative citizen evaluations. In conclusion, although it is difﬁcult to predict which level
of government will been seen as the most fair, there are likely to be some effects of level
of government.

Agency culture: The second agency factor, agency culture, is the set of beliefs,
attitudes, and behaviors of the people in the agency making the decisions (I-Iofstede,
1998). In preliminary exploratory conversations for this study, agency personnel

suggested that the comprehensiveness of public participation varied widely based on the

31

personal beliefs of the person in charge. This could potentially be extended to the agency
as a whole. Agencies may differ in terms of how decisions are made, and therefore in
terms of citizen-perceived fairness, because their membership collectively has different
beliefs than the membership of other organizations.

Kweit and Kweit (1981) specify the aspects of bureaucratic culture that may have
an impact on public participation and thus on the fairness of decision making. These
include an emphasis on expertise, regularity (use of rules, hierarchical authority,
consistency), efﬁciency, and organizational self—maintenance. Authors writing on public
participation have also long suggested that beliefs about the importance of expertise may
interfere with performance of public participation by downgrading the value of citizen
comments and enhancing the importance “of the preferences of “neutral agency
professionals”. These considerations suggest that agencies with cultures emphasizing
neutrality at the expense of accountability to citizens will be judged by citizens to be less
fair. In addition, agencies may have cultural beliefs about the importance of fairness in
decision making and this may also affect citizen evaluations. Speciﬁcally, when an
agency’s culture emphasizes expertise, downgrades citizen knowledge and motivations,
and places less importance on fairness, citizens will evaluate their experience as having
an unfair process and outcome.

Amount of participation resources: The ﬁnal agency factor is the presence of
resources for participation. The amount of personnel, money, and training available for
public participation could limit the quality and quantity of citizen involvement. Thus,
greater amounts of staff, money, and training resources for public participation should

cause citizens to evaluate their experience as having fair process and outcome.

32

Situation context factors

The second set of contextual factors emerging from the historical review
describes the situation in which the decision is made. How did people interact at the
particular meeting evaluated by the citizen? Was there a great deal of conﬂict? Were
there large differences in ﬁnancial resources or expertise among the citizens? What kind
of participation technique was used? Within any particular agency, there may be a wide
variety of decision making episodes and differences among them would impact citizen
experiences.

Participation technique: The ﬁrst situational factor emerging from the historical
review is the type of participation technique used. In the historical review, a picture is
painth of participation in which more formal, less interactive forums, such as letters and
public hearings, were replaced more recently by discussion oriented, collaborative
efforts. While this general trend does exist, the reality is that discussion-based advisory
groups have been used for many years and less interactive forms like letters are still used
effectively. Because a given agency often uses several techniques, citizens may be
exposed to a variety of involvement methods.

It is important to evaluate participation technique because many practitioners
want to know if certain types of techniques work better. It is critical to ﬁnd out if there is
any kind of relationship between technique and perceived fairness. One of the key
ﬁndings from previous focus group research (Smith & McDonough, 2001) was that
principles of fairness were equally important across all techniques. However, citizens

and agency employees preferred methods which involved repeated personal interactions.

33

Techniques commonly used in natural resource decision making include written letters of
comment, one-on-one interactions, formal public hearings, discussion oriented meetings,
and on-going advisory boards. These can be arranged in a hierarchy from no or low
levels of discussion and personal interaction to high levels of repeated interactions over
many years. The ﬁnding that citizens prefer discussion based methods suggests that
participation techniques having more personal interaction and discussion will lead to
citizens evaluating their experience as having a fairer process and outcome.

Amount of conﬂict: The situation may also may also vary in terms of the amount
of conﬂict. During the 20th century, conﬂict over resource use was a driving force in the
institutionalization of systematic public participation. The need to successfully resolve
that conﬂict was one reason new collaborative decision making techniques were
developed (Syme & Eaton, 1989). Because of differences in the participants and the
issues, the intensity of conﬂict may vary. A conﬂict in which people hold onto their
positions deeply, express strong emotion, and hold positions which are highly
incompatible with those of other people is harder to resolve (Floyd, Germain, & Horst,
1996). It should also be harder to achieve an outcome all participants see as being fair
because their positions are more divergent and are held onto more strongly. Identifying
compromise settlements which everyone can accept as fair would be harder. People
intensely involved in an issue may also have more trouble seriously considering the
views of others, threatening the sense of procedural justice. This suggests that decision
contexts characterized by a high level of conﬂict will lead to citizens evaluating their

experience as having an unfair process and outcome.

34

Pripr respectf_ul relationships: It seems reasonable to suggest that factors which
have been shown to affect the resolution of conﬂict may also affect perceptions of
fairness by participants. A factor which makes conﬂict resolution easier is the presence
of prior respectful relationships among participants. Pruitt (1998) reports the results of
many studies on social dilemmas which demonstrate that parties who have prior positive
relationships are more likely to successfully resolve the dilemma with cooperative
strategies. The use of cooperative strategies should lead to careﬁrl consideration of each
party’s interests and thus procedural justice. It might also lead to more fair outcomes as
participants c00perate in their search for solutions that affect everyone equally. In
addition, as justice research has shown, fairness is symbolic of respect and social
standing. Thus if people have a good relationship, they are more likely to feel respect
towards others and treat them in ways which demonstrate that respect, in other words,
fairly. This suggests that a situation in which many citizens know each other and know
the decision makers may be evaluated as having fair process and outcome.

Eg'pr ggpal power ar_nong citizens: The fourth aspect of the situation which may
affect fairness evaluations is the extent of power equality among citizens. Power
inequalities among citizens are a direct threat to the fairness principle of neutrality
because they could lead to suspicions that people are not treated equally. They may also
lead to the impression that one’s comments are not being seriously considered, violating
the principle of inﬂuence. Finally, power inequalities may lead to the judgment that
decision outcomes favor one person over another simply because of the person’s greater
resources, threatening the sense of outcome fairness. These predictions are consistent

with an experimental study which found that procedural justice was most important to

35

people when they were in an unequal power situation (Barrett-Howard & Tyler, 1986).
The authors inferred that the experimental participants were conﬁdent that equal power
leads to fair outcomes, but in an unequal power situation they needed additional
assurances that the decision making process was a fair one and did not reﬂect any bias
towards those with more power. It can be predicted that decision contexts in which
power is more equally shared among participants will lead to citizens evaluating their

experience as having a fairer process and outcome.

Citizen context factors

The ﬁnal set of contextual factors describe the characteristics of the citizens
making the judgements. It is important to remember that fairness judgements are
ultimately individual perceptions. It may very well be that individual characteristics like
gender, education, and prior experiences with citizen involvement will impact the
experiences the citizen has and the way those experiences are interpreted.

Age, gender, education: Previous research found that women and persons with
less formal education felt less control over the political arena and participated less in
governmental decision making (Smith & Propst, 2001). If one feels less control, this
may translate to mean that one feels decision makers are not responsive. Because
responsiveness is an aspect of the justice principle of inﬂuence, women and less educated
people may perceive a more unfair process. A lack of responsiveness may also lead to
decision outcomes which do not treat the citizen equally, violating outcome fairness.
This suggests that demographic attributes like gender, age, and education may explain

variation in perceptions of outcome and process fairness.

36

Prior involvement in decision making: Another important participant

characteristic is the general level of involvement in the political process. General level of
involvement in politics is important because it is correlated with one’s general attitudes
towards government and therefore towards judgments of speciﬁc experiences. People
who are more involved are more likely to believe government is trustworthy (F inkel,
1985). A lack of trust can also lead to disengagement from the political process. A
citizen having greater trust in the government is less likely to interpret an unsatisfactory
outcome as being caused by lack of fairness. It can be predicted that people with limited
prior participation experience will evaluate their experience as having an unfair process

and outcome.

Elaboration of the framework

All of the predicted relationships among variables in the preceding sections are
part of the theoretical framework. In Figure 2, the predicted directions of relationships
between variables in the framework are speciﬁed. As explained above, it was possible to
classify most characteristics of the context as either increasing or decreasing perceptions
of fairness. One exception was for level of government, in which some aspects of the
level of government predicted greater fairness for local levels, and others predicted
greater fairness for state or federal levels. The other exception was age because the
literature did not suggest age-related trends in fairness perceptions.

Although not predicted by the framework, and therefore not speciﬁed in Figure 2,
context factors may also affect each other. For example, it may be that when power

equality is low, conﬂict strongly reduces fairness. However, when power equality is

37

Decision Making Context:
Inﬂuences on Citizen
Evaluations

Agency

 

 

- Agency culture which believes in
fairness and citizen participation

- Level of government

- Greater perceived agency
resources for participation

 

 

Situation

 

 

- Participation technique which
emphasizes discussion

° Greater level of conﬂict

- Equal power distribution

0 Prior respectful relationships
between citizens and between
citizens and decision makers

  
  

 

 

Citizen

 

 

 

- High level of involvement in
decision making

° Older

- Female

0 Higher education

 

 

 

Citizen Consequence
Evaluation of of Citizen
Experience Evaluations

 

 

 

 

 

Fair process Trust in
+ —> decision
Fair outcome makers

 

 

 

1‘ .1
u,

 

 

+ positive inﬂuence
- negative inﬂuence
+ ? unknown inﬂuence

 

 

 

Figure 2. Predicted directions of inﬂuence between variables in the theoretical framework

of perceived fairness.

high, conﬂict may reduce fairness to a lesser degree. This might occur because when

there are strong differences of opinion and power is very unequal, the decisions may be in

favor of the powerful. But when power is equal, conﬂict leads to more equitable

outcomes and people perceive greater fairness. While this framework does not try to

predict interactions, it acknowledges that they may exist. In the results chapters of this

study, once the simpler bivariate relationships presented above are tested, multivariate

approaches are used to explore complex patterns of inﬂuence between context variables.

38

Research questions

Up to this point, the theoretical framework has been discussed without explicit
reference to whether the relationships among variables are psychologically based and
therefore can only be observed at the individual level, or if they also have a social
component and exist at the agency level. For example, do the data support the prediction
that agencies with a higher percentage of female citizen participants receive lower
fairness evaluations? If connections among variables can be drawn at the agency level,
there may be agency level phenomenon driving the relationships. For example, the
agency may have policies which affect how women are treated, causing an observable
relationship between gender and perceived fairness at the agency level. In addition to, or
instead of, agency pattems there may be observable relationships between gender and
perceived fairness at the individual level. In other words, women may have experiences
or perspectives which lead them to be more critical no matter which agency is making the
decision. Finally, levels can be combined to see if there are patterns at the agency level
after controlling for any individual level relationships among variables. The set of
research questions are organized to examine the framework at different possible levels.

The ﬁrst questions examine the agency level.

1. How and why do agencies differ in terms of context, fairness, and trust variables?
2. At the agency level, are the directions of inﬂuence among context, fairness, and

consequence variables consistent with the theoretical framework?

Answering question 1 by looking at differences among agencies allows

assessment of the impact of agency histories, legislative environments, and particular

39

circumstances on context variables. It is also helpful to ﬁrst identify if agencies do in
fact differ before examining if those differences can be explained by the theoretical
framework. Question 2 examines the patterns of agency differences to see if they
conform to the predictions of the theoretical framework. In addition to the agency level,

the framework may also apply at the individual level.

3. At the citizen level, do context factors related to the situation and the citizens
inﬂuence perceived fairness in the predicted directions?

4. At the citizen level, does perceived fairness increase trust in the decision makers?

Research questions 3 and 4 focus on the situation and citizen variables because
these were measured at the individual level. Question 3 examines the relationships
between context and fairness variables to see if they match the predictions in Figure 2.
Question 4 concerns the relationship between fairness and trust in decision makers that is
also speciﬁed in Figure 2. The ﬁnal research questions address the combination of

individual and agency levels.

5. Does fairness mediate the effect of context factors on trust?

6. Which context factors explain the most variation in perceived fairness?

7. Does the level of one context factor affect how other context factors inﬂuence
fairness?

8. Can the differences between agencies in terms of fairness perceptions be explained by
individual and situation factors, or are is there agency level variation in fairness that

can be explained by agency level factors?

40

Questions 5 to 8 address the theoretical framework in a manner most closely
resembling the complex real world by combining variables across both the agency and
individual level. Question 5 concerns the central idea of the framework, which is the
mediation role of fairness. Mediation means that the effect of one variable on another is
transmitted through an intervening, or mediating, variable. In the proposed framework
this means that context affects fairness and fairness affects consequences like trust in
decision makers. Question 6 elaborates the framework by identifying the context factors
which cause the largest changes in fairness judgements. Knowing this may help
managers develop strategies to achieve greater fairness. Question 7 further elaborates the
framework by looking at whether the value of one variable causes a change in the effect
of another variable on fairness. This means that although a variable may not have a
signiﬁcant inﬂuence by itself, it does have inﬂuence in combination with another factor.
Finally, question 8 seeks to separate out the individual and agency level effects on
fairness. Can differences among agencies be explained by the situation and citizens
involved, or do agencies differ in ways that can only be explained by agency level
variables like employee culture?

Answering these research questions is very important because they address the
problem of how to build relationships of trust between citizens and natural resource
decision makers. Better relationships are necessary for the citizen-agency partnerships
needed to address increasingly complex management issues. The ﬁrst step to answering
the questions is to collect the needed data. The next two chapters explain the methods

used to measure the variables and test their relationships.

41

CHAPTER V: RESEARCH METHODS 1: CASES, SAMPLING, INSTRUMENTS,
SURVEY ADMINISTRATION, NON-RESPONSE BIAS AND ANALYSIS

The previous chapter outlined a theoretical framework specifying how contextual
factors may affect citizen perceptions of procedural and distributive justice, and
ultimately, citizen trust in decision makers. In this study, the relationships in the
framework were tested with data collected through questionnaires mailed to citizens and
decision makers involved with a variety of agencies. This chapter ﬁrst describes how the
cases were chosen, indicating how many respondents were in each case. Then the
creation of the mailing lists and the resulting sampling frame is described. The
instruments and variable operationalizations are given, followed by survey administration

procedures, response rates, estimation of non-response bias, data entry and analysis.

77w cases

The basic methodology involved comparing cases. A comparative approach was
used because variation in context factors was needed to test the theory. An experimental
design would have required manipulation of the context, which is impractical when one is
studying real decision making. A random population survey also would not have worked
because only a small proportion of citizens have actually participated in natural resource
decision making. By contrast, choosing speciﬁc cases because they differ on variables
important to the study allowed a more targeted, smaller sample. The comparative case
approach also facilitated consideration of qualitative differences among agencies, such as
their historical circumstances, that might explain differences in citizen evaluations.

In this study, each case was a different agency. Speciﬁc agencies were chosen for

a variety of reasons. The ﬁrst was that they were related to natural resource management

42

and preferably to forest management. The second was that they had ongoing citizen
participation programs with enough citizens involved in the past two years (at least 100)
for statistical testing. Sufﬁcient documentation had to be available to allow the creation
of a mailing list of citizen participants that included close to 100% of the citizens
involved with the agency. Because the funding was from a Michigan Agricultural
Experiment Station grant, agencies had to be in Michigan. Finally, because one of the
hypotheses was that level of government would affect citizen experiences, the set of
agencies had to include federal, state, and local level agencies.

According to these criteria, the Huron—Manistee National Forest of the USDA
Forest Service was chosen to represent the federal level, the Forest Management Division
of the Michigan Department of Natural Resources was chosen for the state level, and
Delta, Delhi, and Monitor Township planning commissions were chosen for the local
level. Although the planning commissions did not deal directly with forest management,
they were chosen because they were partly rural townships experiencing rapid
development. These commissions were often grappling with the decision of converting

farmlands and forests into new developments — a natural resource management issue.

Sampling frame

For each of the agencies above, written questionnaires were mailed to all citizens
recently involved for whom an address could be found. In addition, a written
questionnaire was also mailed to all employees of the Huron-Manistee National Forest
and the MDNR Forest Management Division, as well as the township planning
commissioners. All citizens and employees received questionnaires to make sure that

there were enough respondents from each agency (at least 50) for accurate statistics.

43

The initial mailing to citizens involved with the Huron-Manistee National Forest
was sent to everyone on a Huron-Manistee Forest Plan Revision mailing list. Although
the list was being used for mailings related to revising the Forest’s plan, it had been built
up over several years and included anyone who had contact with the agency for any
reason. It included individuals who had sent in letters, gone to hearings or meetings, or
requested information. It also included members of a long-running advisory group, the
Friends of the Forest. There were also many citizens, such as local government ofﬁcials,
who had been put on the list by the Forest Service as a matter of policy, not because those
individuals had ever had contact with the Huron-Manistee agency.

There were 780 people on the Huron-Manistee National Forest mailing list, which
was reduced to 715 after correcting for envelopes returned because of a bad address or
because the person was deceased. The initial questionnaire about the Forest Plan
Revision asked if the respondent had ever been to a Friends of the Forest meeting and if
he/she would be willing to return a follow-up about the Friends of the Forest. Out of 94
people indicating attendance, 56 agreed to do a follow-up and were also sent the Friends
of the Forest questionnaire.

The mailing to citizens involved with the Michigan DNR was done from a list
compiled by the researcher. MDNR management units throughout the state were
contacted to obtain names and addresses of individuals who had participated in any way
with the Forest Management Division in the past 5 years. Most records came from
meeting sign-up sheets that were sometimes illegible and often had minimal addresses.
Using Internet searches, complete addresses were obtained for most of the names, leading

to a mailing of 245 questionnaires. Correcting for bad addresses and deceased

44

respondents led to 215 deliverable questionnaires. In addition, a list from the MDNR of
people who had attended the Pigeon River Advisory Council was obtained. The list had
50 people and this was adjusted to 47 deliverable questionnaires.

The mailing to citizens involved with planning commissions was done from lists
compiled by the researcher. Planning ofﬁces for Delhi, Monitor, and Delta townships
were visited. Names and addresses were gathered from meeting minutes and sign-up
sheets going back a period of two years. The Internet was used to ﬁnd complete
addresses where necessary, leading to a mailing of 253 questionnaires to Delhi, 142 to
Monitor, and 184 to Delta. Correcting for bad addresses and deceased respondents led to
220 deliverable questionnaires for Delhi, 122 for Monitor, and 172 for Delta.

Mailings to decision makers were based on organizational directories provided to
the researcher. All 314 MDNR Forest Management Division employees and 139 Huron-
Manistee National Forest employees received the questionnaire. A ﬁlter question at the
beginning allowed them to return the survey unanswered if they had no interactions with
citizens during the course of their job. Adjusting for this ﬁltering led to 297 MDNR and
126 Huron-Manistee employees who were potential respondents. Surveys were also
mailed to all current and recent (within the past two years) planning commissioners.
Recent commissioners were included because many of the citizen respondents were
involved up to two years earlier so they would have interacted with those commissioners.
A total of 14 were sent to Delhi, 7 to Monitor, and 11 to Delta. All of these addresses

were correct.

45

 

The instruments

There were two instruments: one for citizens and one for decision makers.
Instruments were pre-tested by administering them to people similar to the intended
respondents of the ﬁnal version. The citizen pre-test instrument was mailed to a list of
100 citizens taken from sign-up sheets of the Meridian Township Planning Commission.
After a reminder postcard, half were returned. The agency instrument was distributed to
20 headquarters employees of the Michigan Department of Environmental Quality
(MDEQ). The WEQ was chosen because it was recently separated from the MDNR
and so would have similar culture and types of employees. Twelve were returned.
Patterns of missing data, correlations among items, and respondent comments in the
margin were used to rewrite several questions.

On the citizen questionnaire pre-test, a long list of items about different topics
was combined together under the general heading , “Your opinions about the Planning
Commission process.” All items had a ﬁve point disagree-agree response scale. In the
revision, this list was broken up into different topics (e. g. decision making process,
conﬂict, power) sometimes with different response scales. This was done to make it
easier for the respondents to deal with the long length. On the employee survey, the
needed revisions were fewer. The main comment from many pretest respondents was
that it was too long. For this reason, some items were deleted from the section on
bureaucratic culture. Deleted items were ones which respondents thought were unclear
or redundant.

Once pre-testing was completed, ﬁve versions of the citizen instrument were

created with slight modiﬁcations to make them agency speciﬁc. In addition, a special

46

version was created for the Friends of the Forest of the Huron-Manistee. This was a
follow-up so it did not need to contain the demographics and background information
already asked on the ﬁrst Huron-Manistee questionnaire. A special version was also
created for the Pigeon River Council Advisory Board of the Michigan DNR. This was
done because many of the council members may have been involved with the DNR in
other ways and so would have received the general DNR questionnaire. A special Pigeon
River questionnaire made sure that they also commented about the Pigeon River Council
separately. Both of these special versions were created because they pertained to unique,
long running participation efforts conducted by the agencies which deserved special
attention. However, all seven versions were kept as similar as possible so answers could
be directly compared. Because there were several versions mailed out, and because some
citizens were involved with more than one agency, some people received multiple
questionnaires. In the analysis these were treated as coming from separate respondents.
Five versions of the decision maker instrument were also made, one for each agency. An
example of a citizen and a decision maker instrument are in Appendices A and B.

On each questionnaire, citizen respondents were asked to focus their answers to
the questions on their most recent participation experience with the agency. Because the
questions were quite detailed, they required a clear memory of the experience and the
most recent experience was likely to be remembered most accurately. In addition, by
choosing an experience based on recency, possible biases resulting from remembering
particularly bad or good experiences would be avoided. The intention was that across the

many respondents who had participated at different times and in different ways, the

47

proportion of good to bad experiences obtained would be representative of the agency’s

overall performance.

Context variables

The variables measured in the questionnaires are listed in Table 1. Most variables
were calculated by adding up each respondent’s scores on a number of items and then
dividing by the number of items. The resulting average score contained less random
variation and was thus a more reliable measure of the underlying construct.

The ﬁrst set of context variables measured agency culture and were measured on
the agency questionnaires. Five scales were created for agency culture. The ﬁrst three
scales were based on theoretical writing by Kweit and Kweit (1981). The ﬁrst scale
assessed employee perceptions of the beliefs of most others in their unit about the level of
citizen knowledge of decision making topics. The second measured respondent
perceptions of the beliefs of most others in their unit about whether citizens were
motivated by self-interest and had short-term views. The third scale sought to measure
the relative importance accorded to expertise versus citizen participation. All these items
had to be newly created because there was no previous quantitative research on the tOpic.
Respondents were asked about the beliefs of most others in their unit because this was an
attempt to measure agency culture which is shared beliefs.

The fourth and ﬁfth agency culture scales dealt with justice. Process and outcome
fairness items were copied from the citizen instrument for use on the agency survey.
Employees were ﬁrst asked to rate how important each item was to themselves and then

were asked to rate “how often you and other employees in your unit actually achieve

48

Table 1. Variables and how they were measured.

 

 

 

 

 

 

L Data source LOperationalization
Context variables - agchfactors
Agency culture Agency survey
Beliefs about citizens
- citizen knowledge 2 items
- citizen self-interest/short-tenn view 2 items
Importance of expertise versus participation 11 items
Importance of fairness principles 21 items
Performance of fairness principles 2] items
Level of Evemment Assigned Federal, state, local
Amount of agency participation resources Agency survey 3 items

 

 

 

Context variables — situation factors

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Participation technique Citizen survey 1 item
Amount of conﬂict Citizen survey 4 items
Prior equal power distribution Citizen survey 5 items
Prior respectful relationships Citizen survey 6 items
Context variables — citizen factors

Level of respondent involvement in decision Citizen survey

making

- Total involvement across all agencies 11 items“
- Involvement in the agency being evaluated 6 items“
Respondent age, sex, education Citizen survey 1 item each
Evaluation of experience variables

F air process Citizen survey 16 items
Fair outcome 5 items
Consequence of evaluation variables

Trust in decision makers I Citizen survey I 6 items

 

Chapter VI lists the questionnaire items and measurement scales used for each variable.
Variable values are from item score averages except * which were assessed as the

percentage of total items checked.

49

 

 

them.” These two measures reﬂect a distinction between cultural values and practices.
Hofstede (1998) found that these two ways of measuring culture achieved very different
results. Values tended to be individually held and so did not predict organizational
variables, but practices were very culturally consistent across organizational members.

The next context variables were level of government of the agency and agency
resources for participation. Level of government was assigned by the researcher.
Agency resources were assessed on the agency survey by asking employees how much
money, staff, and training were available for public participation. Assessing resources in
this perceptual way was consistent with the other perceptual measures on the surveys, but
it may not be entirely consistent with the actual resources available.

Another context variable was participation technique used. On the citizen survey,
participants were asked to focus their answers on a particular participation experience.
This was done because the theoretical framework speciﬁed that evaluations of a
particular experience affected global levels of trust in decision makers. Testing the
theory required that questions be answered in terms of a speciﬁc experience. For the
speciﬁc experience, citizens were asked how they had participated. Possible ways to
participate included receiving mail, sending written comment, having a one-on-one
interaction, attending a hearing, attending a meeting, and participating in a number of
agency speciﬁc opportunities. These possibilities were classiﬁed into six techniques
ordered from least to most interpersonal interaction: mail/non-speciﬁed, written, one-on-
one, hearing, meeting, and advisory board.

Respondents had the option of saying they had participated in a particular

planning project and these projects had to be coded into techniques. The speciﬁc projects

50

for the MDNR included the Presque Isle Management Plan, the Menominee River
Management Plan, and the Lake Superior Forest Pilot Project. These projects consisted
of three to six meetings and were classiﬁed as “meeting”. The MDNR also allowed
people to participate in open houses and compartment reviews. This opportunity was
classiﬁed as a hearing because it was a one-time event and concerned discussion of very
speciﬁc decisions, not general policy. The MDNR also had two advisory boards (Pere
Marquette Friends of the Forest, Pigeon River Country Advisory Council) and the Huron-
Manistee had one advisory board (the Friends of the Forest). Finally, in addition to their
regular hearings, Monitor township held a Visioning workshop which was classiﬁed as a
meeting because it was very discussion-based and focused on more general, policy-level
issues.

The next variable was the amount of conﬂict. Although the importance of this
variable is well supported in the literature, there were no operationalizations that could be
applied to the wide variety of cases included in this study. Therefore a new set of items
was created that measured both the intensity of emotion and the incompatibility of
interests.

Other context variables were the existence of prior equal power among citizens
and prior relationships of respect among citizens and between citizens and decision
makers. New scales were created for both variables. Relationships items asked how
many people were present in the participation experience and with how many of them
they had prior respectful relationships. Power equality was measured with an item about
overall power distribution supplemented by items based on theories of capital (Coleman,

1988) which assessed access to knowledge (human capital) and ﬁnancial resources

51

(ﬁnancial capital). Distribution of these types of capital is important because they are
sources of power.

The ﬁnal context variables concerned the characteristics of the citizens
participating. There were two measures of citizen involvement in governmental decision
making. Previous research (Smith & Propst, 2001) developed a general scale for
measuring total participation in political and natural resource decisions. This was
replicated here to allow comparison across cases. Another scale measured total
participation within each agency compared to the possible involvement in the agency.
This gave a percentage score of agency involvement which could be compared across
cases. Finally, standard demographic items on age, gender, race/ethnicity, and education

were asked (Smith & Propst, 2001).

Evaluation of experience variables

Most of the fair process items were modiﬁed versions of those in Lauber and
Knuth (1998). Other standard fair process items modiﬁed ﬁ'om Tyler (1994) were
politeness and honesty of authorities, consideration of citizen comments, and overall
fairness of the process. New items on inﬂuence over the outcomes as well as over the
process were created for the fair process scale because they reﬂected principles of control
(Thibaut & Walker, 1975). Additional procedural justice items, based on ﬁndings from
previous focus group research (Smith & McDonough, 2001), included convenience of
attendance, involvement of local people, notiﬁcation, and question answering.

Fair outcome items were modiﬁed from Tyler (1994). New items about fair

outcome concerns of equity, equality and need were also created.

52

Evaluation consequence variable

A scale measuring citizen trust in decision makers was created by combining four
types of items. Citizen trust in the agency’s decision making and citizen satisfaction with
the agency’s job performance were assessed with questions modiﬁed from Tyler and
Degoey (1995). Citizen support for the decisions made in the particular experience being
evaluated was measured with items modiﬁed from Tyler (1994). An item assessing the
improvement in citizen — decision maker relationships was created for this survey. These

four types of items were used in combination to measure citizen trust in decision makers.

Survey administration and response rates

Both the citizen and decision maker instruments were administered in the same
way. All surveys were stamped on the back with a number and then sent with a cover
letter. The stamped number was used to track who had and who had not responded in
order to facilitate reminders. Only the researcher had access to this information.
Respondents were told that following the completion of data analysis, the connection
between the list of names and survey numbers would be deleted.

Questionnaires were designed so they could be folded in half and mailed back
postage prepaid without an envelope. After two weeks, reminder postcards were sent to
people who had not returned the survey. After another two weeks, a second copy of the
survey was sent to those who had still not responded. Starting two weeks aﬁer that,
approximately 20 % of the nonrespondents for each mailing were randomly chosen and
an attempt was made to contact them by phone. If a phone number could be obtained for
either home or work, phone contact was attempted repeatedly (sometimes up to 10 times)

for a month. The number of successful contacts is given in Tables 2 and 3.

53

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2 NF MN 08 m. F. m 5 3- mm ONN =38: 2:235 53:
N 0.0 N m 328% -8: Eco
w w N.No 4 N N 4m macs; 53o
F F Ndo F F m N 88-8-28 EEG
F F 58 F m N 53E» EEO
N N FF on: @9- m Fm N ta 8.5— =oym “752
a F F mm who who N mm mm mm m FN 258.5 «ZOE
N v NNN F F F m 8 850898: as: mzoz
80.35 8an “mono-F
meme F N m 2% 889a 83 2292
.83 F832
o.om F F m 83m 852852 MZQE
N8 F N 53 22 258; M752
amouom 23 MO m—ucoim
F F 082 F F N N usages: 28 mzoz
3032 Eons-ages
F F 02 F F F mF \ was: :80 2292
v v .92 m mm N FN 2828-28 M752
F F 53 F m m 5E:9 «202
F F F 98 5mm F M F on om mpg "22:
ov vm MK Qmm Ndv NF oF F 09 va m FN 938.5 mar.— “22:
mm 3t 3: m. 8 mm 3. 35888: as: Em
F N FE N 8F 8 mm 8:85 mzzm
m m msm m 3 FF mm 28-8-28 ”F223
F m Qwv VF m F mm 52:3 "F22:
racism 88sec E828 229:8 BEBE v m N F “8sz 9333.
28.5 0:95 0:93 .x. .x. Eek F258 8 8538 u oEEoz—ov u :38. o=3=58 FE: tag-:5

 

 

 

 

 

 

 

 

 

85qu .3 9823 Erase—.8 0:23 98 Foo—8E 05 8 mean 8.898% .N aim-F.

54

.8 Eu :8 >258. 88—880 m E 98¢.
8 822: :8th 0:? 3:82:88: .8 FF 2:: 288 88 80:88 823. 2:8 838m 8 2828 2:88am?— mo n n A228 2823
88:: an 888: 28888.. .8 u H 88:8 32a
”88:88 283 2:8 288 823 :8 2:28:88: .8 n n 3528 wreck
”cm-$24: >283. 22885 I 4+m+N+c .2: n 8888 x
$833 2883880 FF E8..- \$+m+N+: .52 u F8538: o\e 3qu
88288: d. 23.888 23 8E8 8:2: :33 u 82:8 8833.... 8388288 FF BEN
8:88:88 2: 8 9:83:38 83 88:88:
2: 82:82 2.: .8 282 E 282882 2: 823: 85 888 28 Be: E 8:2: 225 :2: 833m 2: 82» 8:28 8:: 2:-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

55 mo F on F ms} F.mm mN mmN mmN mom Dam F 12.-HOB
w oF mF who nm N 0N mm vm NF :8th 5:835. 58:
F N 0.0 F N F F 88:88-8: 580
o N mNm F N wN N 8:8: 25
F F 0...: F N F 28-:0-28 SEQ
5mm F N v ~8st «:89
5 OF N F m.wm w. Fm N m F FN m F NN F 238.5 @5833. 8:82
N v ed 0 F 8588-8: 8:82
F F New F N m m 8:283 wEaEmE 8::02
m v New F m m m mctmo: 8::o2
ode F F m 28-8-28 8:82
F F 0.00 F F F 8st? 8:82
888 8880 8888 22980 8:88: V m N F 8—8:: 2638
88:: 28:: 285: Ax. X 88:. F288 8 8:88: a 28838: a .88- 85:88 :8 8?:5

 

8.2.882 .N 22;

55

 

 

 

 

  

 

 

 

 

 

 

 

 

 

II 2 Cu I 12 dd t C «xvi» till L-Hn Q8 K ..?I..
I Nutrmzm-leUSWSC-J .JQFCUZS 0.0.2:..30 3025.5.- V M- l-N ~ F F528:- mkutzm
020:; 026—7— DCCSQ .X. c: 129-- Fu—Zzi a0.- ..UFCZEE xx O~£=LO>207 xx 229k-
.mLUv—sz nunvmthU—v-‘C ﬁxﬁOer-um. Q:r>3c:awu nun-nuznh 7:: 205:2: 9.: a:

sat-veal

«ﬂu-:35 392:..vas2rdz .M- U~n~=r~

  

.8 E: :5 >955 cos—9:8 : :m 38

o: 98me :02? 0:3 35338: No u 2:: 0:23 :96 $5528 >023 2:8 833: 8 360%: £52258 .3 u .I. A925. 28:
6:23 3 8:82 352038 No x n 8828 393

629:on 203 2:8 0:2: :03? :8 $5283: No a u 358:: 285%

é+m+~+c >333 22:58: . v+m+~+z .2: u makes x

mmzotsm 033033: No 3: :38. >v+m+N+$ *2: .I. 3553 X 383
308:8 :oNEo o: @5865 89: E: 393380 558.: 25:: 3:2: :32 u 32:5 .3322: m3:.§>.¢mh\o u BER

 

56

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5N om Nv 9mm m. Fm m F on 3% VF F mmv A<HOH

m m v Qt. m. Fm N F N v FF 55:35. «:99
F F F 063 5mm m F N N 35:32. 3:52
v m o Q8 v. E N N F m E 9:252. 23:
n m NF NFm mdm N mF Fv mF ONF hZ—ZE
NF n F m F odw wk» 5 nm mm mm nmN 1752
52% 38:8 E803: 82:88 BEBE v m N : 3:3: @323

0:25 0:93 0:95 o\.. o\° 30H :58: :o BEBE 3: 03803—0: x 30H zogm

 

 

£83.: 538: Ho 9323 9-30:8 28:: :5 3:2: 05 9 33: SEX—mom .m 03.;

\hhen co
survey and then
then agreed to h
those people. T}
who either ans“
response to the p

A seriou:
required accurai.
Whose involvem ‘ '
they had trouble
citizen questionrf
NOIably‘ cases \\
questionnaires, J
IaCked informarzr

thal the Sample 0

 

 

When contacted by phone, the person was asked why they had not retumed the
survey and then was asked to answer some of the survey questions over the phone. Some
then agreed to ﬁll out and return the completed survey. Another copy was mailed to
those people. The “phone survey” columns in Tables 2 and 3 show the number of people
who either answered the short form over the phone or returned a questionnaire in
response to the phone call.

A serious difﬁculty with these questionnaires was that they were very detailed and
required accurate memory of events. People who had not been involved recently or
whose involvement was minimal reported in written comments or over the phone that
they had trouble answering some of the questions. Thus only 54% of the total returned
citizen questionnaires were complete. Table 2 gives the percent complete for each case.
Notably, cases which were mail/unspecified had the lowest percentages of complete
questionnaires. This is probably because these people had less involvement so they
lacked information to answer some of the questions. When these cases are removed so
that the sample only represents those people who were actively involved, the total
percentage of complete questionnaires rises to 66%, a much more acceptable level. This
level is similar to the proportion of the agency questionnaires that were complete (86% of
returned questionnaires) (Table 3).

Percentages of complete questionnaires were based on answering all questions on
the survey, so they are conservative. Because not all variables were used in all analyses,

incomplete questionnaires were included in the data set. This meant that sample size for

57

each anal

Sample 5

hon-res;
C
non-resp
surrey. '
with the .
The only
making.

atleeted.

Table 4,

Reason
\

T00 busy
Intended
NWT rec
"\lready l'
SW9)" to
Lost. thr:
Mel De
oesrjOn S
ThOursvht S

Nor imi’rei

iSCOUTaar
Satisﬁed;
Responds“

each analysis varied and was usually much closer to the total number of surveys returned.

Sample sizes are reported with the results of each analysis.

Non-response bias

Citizen survey: Non-response bias was estimated in several ways. First, when
non-respondents were reached by phone they were asked why they had not returned the
survey. The most commonly cited reasons were that they were too busy, not involved
with the agency, and lacked the information needed to answer the questions (Table 4).
The only bias these results suggest is that non-respondents were less involved in decision

making. Any variables correlated with involvement levels would probably also be

affected.

Table 4. Citizen reasons for not returning survey.

 

Reason # of mentions
Circumstantial/survey

Too busy 24
Intended to, but haven't yet 11
Never received survey 11
Already returned survey 10
Survey too long and detailed 6
Lost / threw out survey 5
Survey not a priority 5
Questions too general / inappropriate 4
Thought survey was for someone else 4

Involvement level

Not involved with agency 28

Lacked the information to answer questions 13
Attitude towards agency

Not interested in agency 7

Discouraged with agency 5

Satisﬁed with agency 2

 

Respondents often mentioned more than one reason.

58

Directly testing whether or not respondents and non-respondents answered
questions differently required doing t-tests of their answers to certain survey questions.
When non-respondents were called, an attempt was made to get them to answer a shorter
version of the survey over the phone or have them commit to ﬁlling out and returning the
complete mailed survey. Although not everybody followed through on their promises
(suggesting that everyone should be strongly encouraged to answer the phone questions)
a signiﬁcant amount of data was collected on the major variables in the study.

The short form (Appendix C) included the demographic items, general attitudes
towards the agency, extent of involvement in the case, and extent of involvement in
general politics and environmental decision making. A few additional items about
speciﬁc participation experiences were asked if the person indicated that they had been
actively involved in ways other than receiving mailings. The items asked over the phone
were chosen because 1) they would be easy for all respondents to answer, and 2) they had
some theoretical relationship with other unasked questions. This was done to avoid the
common mistake of measuring just a few simple demographics, ﬁnding no difference,
and then assuming there are no differences in the key variables of interest (Rylander,
Propst, & McMurtry, 1995).

Group comparisons of respondents (survey rounds I, 2, and 3) with non-
respondents (survey round 4 and phone survey) showed a few signiﬁcant differences
(Table 5). Conﬁrming the qualitative results, non-respondents were signiﬁcantly less
involved with the agency they were evaluating. Closer examination of the data showed
signiﬁcantly lower involvement in almost all ways (i.e. written comments, one-on-one,

hearings, and meetings). However, non-respondents’ total involvement with politics and

59

li~I-1.I 1| «1|
2.: ¢ .55: -2 K3: .549? : Q3:

tag: :.;....-:.i::.. v .v. $.35

'71 ﬂ. ' '4 .40 I . 1.1 a A 4 1 . a .
I. .9. 9.20979. :3: 203:”. 7707?: C. AQrUerv m.—.U_:=X_v.rtr::: 3:: :39: 5:01:3Lmbg =00>30£ v2.5.2 072.45..

Eon 5 2m moocoeotmc Emu—mama
dozmo wEon “can >995 a E 2.8 8

0:23 2: 53 30:83 @2033“ 556 0:3 89: 2a macoccoameéoz .m 38 .N ._ 823, Sow teamwoewwm 8m mos—g Eowcoamom

 

 

vomd cmd mm.m no 5n 553:3—
mg... 8... a... M: Em A case n _ .22: u o C 5m
32 3:. w? 8 an own .0» 2 2a new Ba :2 B 505 €235 «3.
Bad 3.5m awn no can 9.32: 5128—. .5555.._>=o “:3 85.2. 3.35» a. «58235 .55. .x.
wmnd wad cod om _ mm AmcoEEo Eon 33% $55 53, $8qu 2: Eamon $593§$ 8:280
coed Sud- mmd- 3 00m GENE”. $88.“ 3:33 @225me 83 6303 3:559 .339—
Quantum—xv

ﬁnd owd- mud- ov omc £5 E 3:82 566% a 63m 8 Evans .8230 3?ch 9 51 C team—3. .83qu
:8... was mad- mm m G 82? Ecoﬁoa >8 5:5 325.80 203 3:82 32203 BC 2:32.: but
mood wad- omd- mm wwm C32: 203 £56826 .3 $8830 of» 2:825 huh
Cod nmd wmd :V 03 C5 203 $8663 838 8 wow: 8538:. 2:» 839:— ham
comd Nw.~ .vo.m am .26 2.55.03 statute.—
cccd 3.3 N~.- ﬂ. own $86: 6-3 >o=0u< E 32:29:.— ..e .961— .x.
83 2d cod we NP $823195»: 5 as;
mégﬁ :88 :85 2 Z mBESA

as: 5.3-: an? $3-: Q8»

.35 omcommorcoc :oNEo $33 2 Emorcv $=3=oa$$=o= can Emu: mucous—0&8 5053 £82: 2353 .«o :OmtquoU .m 2an

6O

 

 

 

 

 

natural} resou1
respondents.
respondems.
population}
“ho did n01
female and
9% esﬁma‘
may not
0f their d:
OUICOme '
Yesmnde

“on-TESF

eXCept‘
Saliem‘
“me;-
qui‘szii
iittle i
nhQSe

undo

natural resource decision making in general was not signiﬁcantly different from
respondents. The percentage of people who were female was significantly lower for non-
respondents. This particular result can be confirmed for the entire non-respondent
population because the surveys were numbered and could be matched up to names to see
who did not respond. This analysis showed that the respondent population was 21%
female and the total non-respondent population was 17% female, which is higher than the
9% estimate obtained from round 4 and phone survey respondents. Thus gender bias is
likely not a concern. Finally, non-respondent evaluations of the fairness of the outcome
of their decision making experience were mixed (Table 5). Of the two items measuring
outcome fairness, one showed significantly more positive evaluations for non-
respondents, but the other showed no signiﬁcant difference between respondents and
non-respondents.

These results suggest that non-respondents are basically the same as respondents
except that they are less involved. Less involvement means the survey will be less
salient, usually one of the strongest indicators of whether a survey is returned. This
survey was also very detailed, requiring a certain level of involvement to answer all the
questions. In fact, the issue of how much detail to ask and whether or not those who had
little involvement should be ﬁltered out a was difficult design decision. Ultimately, even
those with little involvement were included so as to ﬁnd out what they did know. This
undoubtedly lowered the response rate.

However, for the present study, the differences cited above may not be relevant.
The heart of the analysis presented in Chapters 8 and 9 rests on correlations among

variables, not their absolute values. Are there any differences between respondents and

61

non-resp
represent
survey a:
responde
responde
oftt'hieh

same dir

they eta"
ICSponde
and the c

A
Phone ar.
reaSons V
also state

responds

non-respondents in terms of relationships among variables? The non-respondents can be
represented by those people who either answered questions over the phone or sent in a
survey after being called. Bivariate correlations among the variables using the non-
respondent data suggest there were minimal differences between respondents and non-
respondents. Twenty signiﬁcant correlations were found in the non-respondent data, all
of which were also found in the larger respondent data set and all of which were in the
same direction as those in the respondent data set.

In conclusion, although non-respondents tended to be less involved in the agency
they evaluated, the patterns of how they answered questions were the same as those of
respondents. For the present study there is no evidence of meaningful non-response bias
and the conclusions reached apply to everyone who was on the original mailing lists.

Agency survey: Non-respondents to the agency survey were also contacted by
phone and asked why they had not returned the survey. The most commonly cited
reasons were that they were too busy or not interested in ﬁlling out the survey. Some
also stated that they did not have contact with citizens as part of their jobs (Table 6).
These results suggest that non-respondents were not systematically different from
respondents on the variables measured.

Non-respondents were also asked if they would answer a short version of the
survey over the phone or send in the complete questionnaire. The short version
(Appendix D) had questions asking about their experiences with citizen participation,
their beliefs about bureaucratic and expertise-based values, and their age, gender, race,

education, professional association memberships, and tenure in the organization.

62

Table 6. Agency reasons for not returning survey.

Reason # of mentions

 

Not interested / survey not a priority
Too busy

Intended to, but haven't yet

Never received survey

Questions too vague

A-bUIUIﬂOO

Do not have contact with citizens

 

Respondents oﬁen mentioned more than one reason.

When respondents and non-respondents were compared, there were only two
signiﬁcant differences, both showing that non-respondents had worked a fewer number of
years in their agency (Table 7). Another non-signiﬁcant trend conﬁrmed the qualitative
result that non-respondents had fewer experiences with citizen participation.

How might these differences between respondents and non-respondents affect the
present study? In this study, only the agency culture variables are used. Examination of
correlations from the respondent data-set on cultural variables suggested that employees
with less participatory experience place less importance on fair process ( r = 0.166, p =
.005). However, correlations also showed that employees who had shorter work tenures
in the agency tended to believe citizens possessed the knowledge and abilities to
participate ( r = .119, p = .041). They also believed that citizens are not self-interested
and are not focused on the short-term ( r = -.157, p = .007). These results contradict each
other. The participatory experience bias suggests non-respondents have a culture which
downgrades participation whereas the tenure bias suggests non-respondents feel citizens
can and should participate. If the data were adjusted for both tenure and participation

experience, the net effect on agency culture would probably be close to zero, suggesting

63

.23 5 2m moocohobﬁ Emu—“Ema
66:8 wﬁon Sun 523 a 5 25m do

6:23 2: 696 28526 @8033“ SEE 0:3 805 98 352038-52 .m 28 .N 4 8.33 Bot uoﬁwouwwa 2a 829» “cowcoamom

 

 

:3. a? 3.2 2 8m aaﬁaaoueeazn 5 2:5...
:8... 3.2 2.2 cm 2% 5:32.65 a. 2:59
R2 2.0 3o 3 8m c3505 u _ 5305-8: u e V $65 3838...... 5 niece—52
Sod 3o 56 NN EN C3808 u _ 5255-8: n o v Ea a 3:225:
$3 9% Se. 3 won 838:3.
33 2.0 one 2 RM A vase u _ .22: n o V 5m
22 can 8e 8 o: am“ new 2 2... 2x Ba :2 3 sea £225 on...
32. 3o NS mm 8m $3.8 «3:25
Boo m2- Ed- 2 2% e23 3.22.33:
83 38 mm? mm ME” 8512.3 8:33.25. .x.

m3~u>iQ sows :63: 2 2 NNQELGA

>83 98%: 5.3 3.3-: $3

.85 omcoamocéoc xocowm 888 2 aware—v mEoEOQmBéoc use 38: 8:26:88“ 56253 memo:— oEmtg mo acmEQEoU .N. 633.

64

that adjustment is not needed and that non-response bias in terms of agency culture is
minimal. Given the high overall response rate to the agency surveys of 72%, non-
response bias is not a problem.

In summary, non-respondents to the citizen survey tended to be less involved than
respondents. Non-respondents to the agency survey tended to have less tenure in the
agency and less participation experience. However, when examining bivariate
relationships among variables for both agency and citizen data, there were no meaningful
differences between respondents and non-respondents. This lack of non-respondent bias
on the key research issues, while comforting, does not address the issue of biases
resulting from systematically excluding people from the sample frame.

Frame errors arise from the fact that the mailing lists only contained people who
had participated with the agency. People who were strongly disenchanted with
government in general or the agency in particular would not even bother trying to
participate. In addition, people with no strong interest in natural resource issues, those
generally apathetic about participation in government, and those simply too busy also
would not be on the list. Evidence for this bias comes from comparison of the sample in
this study with that of another study. Smith and Propst (2001) had citizens who were
members of community groups ﬁll out a questionnaire which included the same items
about participation in politics and natural resource decision making that were used in the
present study to measure general involvement in natural resource decision making.
Averaging across the eleven items, the Smith and Propst study found an involvement

level of 27.6% compared to this study which had 39% involvement. This supports the

65

conclusion that there is frame bias in terms of excluding those who are less involved in
politics and government.

In terms of demographics, the sampling frame does not represent the state of
Michigan. The sample was 80% male, 61% over the age of 50 years old, 55% possessing
a four-year college degree or higher, and 93% white. There were only 1% Asian-
American, 0.2% African American, 0.5% Hispanic, and 3% Native American
respondents to the citizen survey. It is also clear that minority groups, particularly
African Americans and Hispanics, were not represented on the lists. How might the
white, elderly, highly-educated male sampling frame error lead to bias in the research
results? The problem is that the questions on the survey asked respondents to evaluate
and describe their participation experience. If they never participated then they simply
would not be able to answer. Mailing the survey to a random sample of Michigan

residents, and thus using a less biased frame, would not have worked.

Coding and data entry

Data from the surveys were entered into MS Excel and then imported to SPSS for
data analysis. Responses to questions asking for frequency of participation were entered
in terms of the number of times a person had done something. If they wrote “10+” then it
was entered as 10. This introduces only minimal inaccuracy. On the citizen survey, when
citizens were asked to choose the technique they used to participate in their most recent
participation experience, they sometimes indicated more than one technique. Since only
one value could be entered, the rest of the answers were checked for clues as to which to

enter. For example, if they wrote a description of when, where and why the experience

66

 

 

 

mmﬂmn
were involve:
most agency-
mten com
was entered
and so wen
Dé’pamner
WBT‘),

mailings '

511g gestir

hm a\‘9 9 a

Vail

g.
{/7

occurred, this sometimes gave a clue. Also a later question asking how many people
were involved was also helpful. If nothing could be deduced, then the technique with the
most agency-citizen contact was chosen. For example, if they checked “submitted
written comment”, “talked one-on-one”, and “attended an open house,” the open house
was entered as their participation experience. Some people did not specify a technique
and so were put in the “non-speciﬁed” categories. On the surveys related to the Michigan
Department of Natural Resources (MDNR) and Huron-Manistee National Forest
(HMNF), respondents were also allowed to check that they would base their answers on
mailings they received. However, many of them wrote comments in the margin
suggesting their answers related to experiences other than mailings. For this reason the
“mail” and the “non-speciﬁed” cases were combined.

Answers to questions with an agree-disagree scale were entered as the following
values: Strongly Disagree = -2, Disagree = -1, Neither Agree Nor Disagree = 0, Agree =
1, Strongly Agree = 2. Answers to questions on a true-false scale were entered as:
Deﬁnitely False = -2, More False Than True = -1, More True Than False = 1, Deﬁnitely
True = 2. Answers to the importance scale questions were entered as: Not at all
Important = 0, Somewhat Important = 1, Important = 2, Essential = 3. For frequency
scales, answers were: Never = 0, Seldom = 1, Sometimes = 2, Often = 3, Always = 4.
Extent of impact scales were: Zero = 0, Small = 1, Medium = 2, Large = 3, Dominating =
4. Amount available scale was: Almost None = 0, A little = 1, Some =2, A Lot =3.

On all questions with a Don’t Know option, Don’t Know was entered as a missing
value for that question. The presence of this option meant that respondents could

distinguish between being neutral on the issue (i.e. Neither Agree Nor Disagree) and not

67

knowing the answer. The presence of Don’t Know was crucial because some questions
were very speciﬁc and a respondent may not have remembered the details or their
experience was too limited for them to ﬁnd out. Sometimes it was clear that they had
used the Neither Agree Nor Disagree category as equivalent to Don’t Know. This was
apparent when they marked all questions on a page this way. In this case, the answers
were entered as missing values.

In the demographic section, people sometimes checked multiple racial/ethnic
categories. This was almost always a combination of Caucasian and another group and
the person was assigned to the minority group. If two or more minority groups were
chosen, the “Other” category was used. A few people, obviously bothered by the
question, wrote in “American” or “Born in the U.S.A.”. These were entered as “Other,”
although they were probably Caucasian.

Respondents often wrote comments in the margin or in the open section at the
end. Some also called with comments or questions. A list of categories of comments
was created. When a respondent’s comment ﬁt a particular category, the respondent’s
questionnaire ID was added to the list under that category. If the comment related to a
particular question, the question number was also entered. This allowed easy retrieval of
qualitative data on topics related to the study. Questions that may have been confusing

to respondents could also be identiﬁed.

Analysis

There were four basic components of the analysis. The ﬁrst involved factor

analysis and the validity and reliability checks needed for creation of the variables from

68

the items in the questionnaires. The second part used one-way analysis of variance and
post-hoe tests to identify differences among agencies in mean values of the variables in
the study. The third part involved examining the bivariate relationships among variables
proposed in the theoretical framework. Finally, the fourth part used stepwise analysis of
covariance to combine the individual and agency levels of analysis with multiple
variables.

Reliability checks involved factor analysis of items which had been designed to
measure the same construct. For example, the 16 fair process items on the citizen survey
were factor analyzed to see if they could all be combined into one scale or needed to be
broken up into several subscales. Items which did not load on factors as expected were
discarded. Aﬁer factor analysis, alpha reliabilities were calculated to conﬁrm that items
in a scale were consistent enough to be combined. Then validity of each scale was tested
by examining its correlation with other scales that should have been similar. If
correlations were opposite of those expected, the scales were examined more closely for
problems. Often comments respondents had given over the phone or written in the
margins gave clues to why certain items were problematic. If justiﬁed, a scale was
discarded for this reason.

The second component of the analysis explored the agency level differences
among variables. Mean scores on all the variables were calculated for each agency.
Then one-way analysis of variance and posthoc paired comparisons were used to identify
agencies which were signiﬁcantly from each other. Posthoc comparisons were t-tests to

determine if variable means were signiﬁcantly different. The signiﬁcance level was

69

 

 

 

adjusted t<
comparisc

11
with simf

framewc

combini
lock 5‘
are 6111'
increas
the {w
IKline
four c
Eixpla
10 se
incr:
Clea
\‘aﬁ

ind

an;

(IQ

adjusted to take into account the number of comparisons made because when many
comparisons are made, some may be signiﬁcantly different by chance.

The third analysis examined the individual level relationships among variables
with simple correlations. This allowed testing of the bivariate relationships present in the
ﬁamework.

In the ﬁnal analysis, stepwise analysis of covariance was done in order to
combine the individual and agency levels of analysis. The analysis was conducted with
block stepwise linear regression. In block stepwise linear regression, groups of variables
are entered sequentially to see if adding an additional group of variables signiﬁcantly
increases the amount of variation explained in the dependent variable. For example, in
the ﬁrst regression four demographic variables might predict 8% of the variation in
fairness evaluations. In the second regression, the 3 situation variables are added to the
four demographic variables and the resulting equation with 7 independent variables may
explain 10% of the variation in fairness evaluations. An F-statistic can then be calculated
to see if the change in explained variance is signiﬁcantly different from zero. If the
increase in explained variation is signiﬁcant, then the category of situation variables is
clearly important and should be kept in the theory. If there is no increase in explained
variation then only demographic variables are needed and the theory simpliﬁes to
individual differences.

The stepwise regression described above is only at the individual level of
analysis. One way to include an agency level of analysis is with agency dummy
variables. The analysis then tests for signiﬁcant differences between the means of the

agencies on the dependent variable after adjusting for the individual level variables which

70

 

 

are equiva
analysis. 1
variables
be added

variable:

multipl
among
make :
faime:
the de
allow
Ofcc

dUm

lder

0n
1h

CC

are equivalent to covariates in an analysis of covariance (McClendon, 1994). A stepwise
analysis, for example, might ﬁrst enter just the agency dummy variables as independent
variables predicting fairness. Then in the second regression, citizen demographics could
be added as covariates to see if they explain additional variation. Finally, situation
variables might be added as covariates to give the full equation.

Using a stepwise analysis of covariance approach has several strengths. First,
multiple regression of any type divides explained variation in the dependent variable
among all the independent variables in the equation (Blalock, 1979). The analyst can
make statements like, ‘controlling for the age of the respondent, conﬂict decreases
fairness.” Second, the direction and relative inﬂuence of each independent variable on
the dependent variable can be compared using standardized regression coefﬁcients. This
allows statements like, “The inﬂuence of equal power on fairness is twice as large as that
of conﬂict” (Alwin & Hauser, 1981). Third, agency level factors can be included as
dummy variables along with the continuous individual level variables. This helps
identify agency level differences that can not be explained by individual level
phenomena. Conclusions can be reached like, “Delta township has a lower level of
perceived fairness than the MDNR even aﬁer controlling for citizen perceptions of power
equality.” Finally, interaction terms can be introduced which test whether the value of
one independent variable affects the relationship another independent variable has with
the dependent variable. For example, a signiﬁcant interaction might show that, “when
conﬂict is high, power equality increases fairness, but when conﬂict is low, power

equality has no effect on fairness.”

71

 

 

 

enter f1
second
(Georg
which i
and n3-

Cl'llfl' CI

Variabl

Pmpos
Varied

Questic
Phone.
meaSUl
c0mbir
Variabi

repreSe

There are also some weaknesses of the approach. For example, stepwise
regression needs a substantive theory to guide the decisions about which variables to
enter ﬁrst. If variables share a lot of variation (multi-collinearity), then the ones entered
second may not increase the explained variation signiﬁcantly and will be excluded
(George & Mallery, 1995). In other words, the order variables are entered may determine
which ones are signiﬁcant. Addressing this concern requires testing for multi-collinearity
and trying different orders of variable entry. Entering variables in blocks can be used to
enter collinear variables simultaneously and avoid some of these issues. Collinear
variables can also be combined into single indexes. This study uses both approaches.

In summary, this chapter has outlined the methods by which the framework
proposed in Chapter 4 is tested and elaborated. A variety of cases were chosen which
varied on the contextual factors. Citizens and agency employees were mailed
questionnaires that measured context, evaluation and consequences variables. Follow-up
phone calls were used to assess non-response bias. In the next chapter, the items
measured on the questionnaires are factor analyzed in order to see how they can best be
combined into computed variables. Then the reliability and validity of the computed
variables are assessed to determine how much each can be trusted as an accurate

representation of the underlying phenomenon.

72

CHAPTER VI: RESEARCH METHODS 2: COMPUTING THE VARIABLES AND
CHECKING RELIABILITY AND VALIDITY

Data for this study were collected through many items on several long
questionnaires. One reason for the long length was that each theoretical concept was
measured with several items. This is a common practice because it allows estimation of
the random measurement error associated with a variable - its reliability (Agresti &
Finlay, 1997). By using several items and then computing an average or a sum of scores
on those items, a more precise picture can be gained of the underlying (or latent)
theoretical construct. However, accuracy of measurement is not just an issue of random
error. Sometimes the items may all be highly consistent with each other (reliable), but
systematically measure the wrong thing. This kind of systematic error relates to the
question of validity which is the extent to which the items are actually measuring the
desired underlying construct rather than a different construct.

Prior to calculating reliability and validity, the items have to be examined to see
how they should be combined into computed variables. This is important because theory
predicts that some constructs may have sub-constructs. For example, the construct of fair
process was measured with 16 items. The items were chosen to represent the many fair
process principles proposed in the literature. It may be useful to compute both a general
fair process variable and several more variables measuring speciﬁc aspects of fair
process. Factor analysis is used to determine the natural sub-groupings of items for those
constructs measured with many items. Factor analysis is also useful for identifying
particular items which do not match the other items and should be excluded from the

computed variable. They may not match because they are actually measuring a different

73

construct or because they were poorly written so respondents can not give meaningful
answers.

In this chapter, once factor analysis is used to determine how items are to be
grouped into variables, reliability analysis measures whether respondents gave similar
answers to all items grouped into a variable. The more similar the answers, the less
random variation in the measure and the greater the reliability. Reliability also increases
if there are more items. This is analogous to sampling theory in that up to a point a more
accurate estimate of the population mean is obtained with a larger sample of individuals.
Reliability is calculated with coefﬁcient alpha (Alpha = kr/ 1+ (k-1)r where k is the
number of items and r is the average correlation between pairs of items) (George &
Mallery, 1995).

The validity of the computed variables is also assessed. Validity can be tested in
several ways. In this study, construct validity is measured (Babbie, 1998). A variable
which has construct validity is related to other variables in ways predicted by theory. For
example, theory predicts people who have a high level of involvement with an agency
will also have a higher than average number of relationships with employees in that
agency. If this relationship holds, both variables are likely valid. If it does not hold, then
one or both may not measure the intended construct, or the theory linking the two
variables may be wrong. Often researchers will deliberately include items in the survey
which already have highly established validity and can serve as comparison variables. In
this study this strategy was only used to a limited extent because the length of the survey

precluded inclusion of many extra items.

74

The exploration of the dimensional structure of constructs as well as reliability
and validity of the computed variables is organized according to the theoretical
framework in chapter 4 which divides variables into three groups. Variables in the ﬁrst
group together deﬁne aspects of the decision making context which may have inﬂuence
on the way citizens evaluate their decision making participation experiences. Variables
in the second group describe how citizen use fairness to evaluate their experience. The
third type of variable concerns a possible consequence for the agency of citizen
evaluations. In this chapter, the factor, reliability and validity analyses are presented
according to these groups, starting with the decision making context variables.

Context variables were measured in three ways. Some were assigned by the
researcher, others were derived from the survey sent to agency personnel, and others
were assessed on the survey sent to citizens. In the following analysis, context variables

ﬁom the three sources of data are considered separately.

Reliability and validity of researcher assigned context variables

Government level and participation technique were both assigned by the
researcher. When agencies were chosen they were chosen partly according to the level of
government they represented: federal, state or local. Because these are self-evident
categories there is no need for reliability or validity checks of government level.
Participation technique, however, does involve generalizing from very speciﬁc efforts
(e.g. Menominee River Management Plan) to general techniques (e. g. meeting). These
techniques were ordered according to the amount of discussion and personal interaction.

A check on reliability for this variable would involve ﬁnding out if someone else would

75

 

 

Fact

Were
com't
belie:
10 me

and d

Cliller
2c
7

b

3a
7

~

To get i
Marga!
emp I 0.1 ‘6
6590119
ef"lire!

assign the same speciﬁc participation efforts to the same techniques. Validity would
involve ﬁnding out if the techniques were indeed ordered from least to most interpersonal
interaction. It was difﬁcult to ﬁnd one person who was familiar with all the decision
making efforts covered by the survey, however informal conversations with several
agency employees suggested participation technique was reliable and valid because they

agreed with the way the participation efforts and techniques were classiﬁed.

Factor analysis and reliability of agency survey context variables

Some variables measured on the agency survey had many items so factor analyses
were conducted to ﬁnd out if there were important sub-groupings or if the items could be
combined into a single dimension. The ﬁrst set of four items measured agency employee
beliefs about citizen knowledge and reasons for participating (Table 8). The items sought
to measure the agency cultural beliefs that citizens have expertise (citizen knowledge)

and do not focus on societal needs (citizen self-interest).

Table 8. Items used to measure citizen knowledge and citizen self-interest.

 

g# Item wording Factor loadings
Citizens who participate in decision making are
2c . .. mainly concerned about their self-interests. .87 -.08
2d usually focused on the short-term. .84 -.16
2a usually able to understand and use technical information -.05 .86
2b usually in possession of the knowledge needed to make good -.20 .80
decisions

 

To get the scores used in the analysis, respondents were asked, “What are the
characteristics of citizens? Please indicate your perceptions of the beliefs of most other
employees in your unit. ”

Response scale: Deﬁnitely False, More False Than True, More True Than False,
Deﬁnitely True

76

Factor analysis of the four items which together comprised the citizen knowledge
and citizen self-interest variables showed that there were two dimensions. The citizen
knowledge items clustered on one factor and the citizen self-interest /short-term items on
the other factor. Alpha reliabilities for citizen knowledge (0.673) and citizen self-interest
(0.716) were barely acceptable according to a common rule of thumb which holds that
alpha values greater than 0.6 are questionable, greater than 0.7 are acceptable, greater
than 0.8 are good, and greater than 0.9 are excellent (George & Mallery, 1995).

The next set of eleven items concerned the perceptions of the beliefs of most
employees in the unit about the importance of bureaucratic principles and expertise
(Table 9). Factor analysis led to three dimensions. The six items aligned on the ﬁrst
factor generally concerned the importance of bureaucratic principles like authority,
consistency and efﬁciency. The second factor contained four items which were clearly
about the importance of expertise. The third factor only had one high loading item. This
item captured the concept of how well agency employees understood citizen needs, a
belief somewhat different ﬁom bureaucracy and expertise. Therefore it was not included
in the variables. The ﬁrst six items were averaged to make a bureaucracy variable with
an acceptable Alpha value of 0.77. However, the four expertise items received an alpha
of only 0.61. When the two items which had the lowest loadings on the expertise factor
(3c, 3e) were removed from the expertise scale, the reliability increased to 0.67.
Therefore the expertise variable was computed from only two items (3a, 3b).

In addition to measuring beliefs related to citizens and the importance of expertise
and bureaucracy, employees were asked about fairness. This was because it was

hypothesized that a particular agency’s cultural attitudes towards the importance of

77

Table 9. Factor structure for perceptions of the beliefs of most other employees on
bureaucracy and expertise variables.

 

Factor loadings
Q# Item wording 1 2 3
3i Well-established plans should not be changed in response to citizen .75 .08 .05
demands.
3h Public participation usually helps to build public support for the -.75 .06 .12
agency.
3g Consistent application of rules is more important than incorporating .69 .23 .05
public comments.
3j The beneﬁts of involving citizens outweigh the costs. -.66 -.08 -.14

3d Decisions made by a person in a position of authority should not be .58 .25 .23
challenged by citizens.

3k Involving citizens slows down decision making processes too much. .54 .41 -.17
3a Decisions should be made according to standard professional practices. .04 .81 .04
3b Experts should have the power to make decisions. .20 .75 .17
3e [Agency employees] understand the long term consequences of .34 .57 .05

decisions better than citizens.
3c When making decisions, correctness is more important than popularity. -.06 .48 -.41

3f [Agency employees] have a clear idea of public needs and desires. .05 .14 .87
Bold loadings indicate the items which loaded on the factor and combined into a variable

 

To get the scores used in the analysis, respondents were asked, “How much do you think
most other employees in your unit agree or disagree with the following statements?”

Response scale: Strongly Disagree, Disagree, Neither Agree Nor Disagree, Agree,
Strongly Agree

fairness would be related to how citizens judged the fairness of their experience with that
agency. For the same reason, employees were also asked how often they thought they
and other employees in their unit actually achieved fairness. Based on a theoretical
review of the fairness literature (discussed in Chapter 5) a set of 16 fair process items
and 5 fair outcome items were generated. These were then used to measure the
importance of fairness to employees (fairness importance) and how well their unit
achieved fairness (fairness performance). The same items were also used on the citizen

survey to assess how fair the decisions were from the citizen perspective.

78

The fairness items were meant to be combined into overall process and outcome
fairness but, because they were created to measure many speciﬁc principles, they might
also combine into sub-groupings of fairness items. Factor analysis was used to check for
sub-groupings. Factor analyses of the fairness importance and fairness performance
items related to the decision making process were done separately and led to slightly
different groupings (Table 10, Table 11). In order to be consistent in the analysis and
allow direct comparison, the same items were combined into the same subgroups for
fairness importance, fairness performance, and citizen judgements of fairness. This was
possible because the factor structures were reasonably similar and were consistent with
fairness principles mentioned in the literature.

Factor analyses using four factors led to the creation of fair process variables
about representation, neutrality, inﬂuence, courtesy, general fair process, and total fair
process. The items composing each and the resulting alpha reliability levels follow. Two
variables called representation importance (Alpha = 0.82) and representation performance
(A= 0.74) were created from items related to access to participation opportunities and
representation of everyone affected (d, e, f, g). Two variables called neutrality
importance (A = 0.73) and neutrality performance (A = 0.75) were created from items
related to the use of accurate information, reasoning, lack of bias, and honesty (m, n ,o,
p). Two variables called inﬂuence importance (A = 0.70) and inﬂuence performance (A=
0.74) combined items related to citizen inﬂuence over the decision making (h, i, j, k).
Two variables called courtesy importance (A = 0.61) and courtesy performance (A =
0.59) combined items relating to answering questions and polite treatment (1, q). Two

variables called fair process importance (A = 0.62) and fair process performance (A =

79

Table 10. Factor structure for fairness importance variables

Factor loadings

 

Q# Item wording l 2 3 4
5e The participation experience is convenient to attend. ._7_4 .22 .16 -.03
5h Citizens are able to participate directly in making decisions. .74 -.05 .14 .03
5f Everyone affected by the decisions has an opportunity to participate. £1 .27 .19 .08
5 g Local people are adequately involved. ,Q .39 .15 .08
5j Citizens have an inﬂuence on the choice of decision making process. .62 -.O7 -.09 .23
Si Citizens are able to have an inﬂuence on the decision outcomes. .60 .20 -.02 .35
5d Citizens are given sufﬁcient advance notiﬁcation of the opportunity to ._g .52 .14 .18
participate.
Sq Citizens are treated politely. -.03 .81 .20 .02
51 Citizen’s questions are answered. .31 .70 .19 .09
5p Agency employees are honest to citizens. .13 .58 .44 .25
5k Citizens' comments are seriously considered in the decision making. .43 .50 .03 .32
511 The decisions are well reasoned and logical. .04 .09 .84 .03
5m Information used to reach the decisions is accurate. .05 .28 .75 .08
50 There is a lack of bias toward particular interests, groups, and persons .21 . 12 .67 .11
5a The procedures used to make decisions are fair. .09 .01 .23 .85
5b Citizens are treated fairly. . .24 .32 .03 .69

 

All the bold numbers in a given column indicate the items which were combined to form

a variable. Underlined numbers indicate items which were combined into a separate

variable.
To get the scores used in the analysis, respondents were asked, “How important are each
of the following statements to you? ”
Response scale: Not at all Important, Somewhat Important, Important, Essential

80

 

 

 

>4
W
'2:
(D

r '0
4t

 

(J! U! £1
0 :3
:‘l :1

B
H
:3

a!
:lOODODeOd

£10
D.
O

J!
”‘1
C”

Table 11. Factor structure for fairness performance variables

Factor loadings

 

Q# Item wording 1 2 3 4
511 The decisions are well reasoned and logical. .75 .12 .17 .15
50 There is a lack of bias toward particular interests, groups, and persons. .70 .02 .25 -.O6
5m Information used to reach the decisions is accurate. .65 .01 .20 .27
Sa The procedures used to make decisions are fair. .63 .26 .09 .31
5b Citizens are treated fairly. .61 .07 .17 .44
5p Agency employees are honest to citizens. .60 . 13 .13 .44
Si Citizens are able to have an inﬂuence on the decision outcomes. .02 .80 .01 .13
Sj Citizens have an inﬂuence on the choice of decision making process. -.09 .73 .22 .03
5h Citizens are able to participate directly in making decisions. .34 .65 .16 .12
5k Citizens' comments are seriously considered in the decision making. .44 .59 .24 .23
Se The participation experience is convenient to attend. .15 .09 .78 .27
5d Citizens are given sufﬁcient advance notiﬁcation of the opportunity to .16 .21 .71 .30
participate.
5f Everyone affected by the decisions has an opportunity to participate. .34 .20 .70 .02
5 g Local people are adequately involved. .29 .38 .42 -.O7
Sq Citizens are treated politely. .26 .08 .09 .76
51 Citizen’s questions are answered. .17 .18 .28 .71

 

All the bold numbers in a given column indicate the items which were combined to form
a variable. Underlined numbers indicate items which were combined into a separate
variable.

To get the scores used in the analysis, respondents were asked, “In the section above you
indicated the importance to yourself of some statements. Now please indicate how
often you and other employees in your unit actually accomplish them. "

Response scale: Never, Seldom, Sometimes, Often, Always

81

 

0.73) “
proced:
perform

bdc.

were al
each se
outcorn
“The d:
citizens
citizens
importa

0.60) w

for Pub]
“MOM-t
analysis
FESQUrCE

I
alpha re}

Camions

0.72) were created using items asking about overall fairness of the decision making
procedures (a, b). Finally, total fair process importance (A = 0.87) and total fair process
performance (A = 0.88 ) were assessed with a combination of all the fair process items (a,
b, d, e, f, g, h, i,j, k, l, m, n, o, p, g).

The outcome items from the fairness importance and fairness performance scales
were also factor analyzed separately. The data easily combined into just one factor for
each set of importance and performance items. The items were: “Beneﬁts and costs of
outcomes are distributed fairly among citizens,” “The outcomes of decisions are fair,”
“The decisions reached are equally favorable to all citizens,” “The decisions beneﬁt the
citizens who are most deserving,” and “The decisions reached are consistent with
Citizens’ personal values.” Of the two variables computed from these items, fair outcome
importance (A = 0.71) had an acceptable reliability, but fair outcome performance (A =
0.60) was questionable.

Finally, agency employees were asked to indicate the level of resources available
for public participation in their unit. They rated three items, “Staff ’, “Training”, and
“Money” with response options of Almost None, A Little, Some, and A Lot. Factor
analysis of the responses yielded one factor. The resulting variable of participation
resources had an Alpha reliability of 0.76.

In conclusion, most of the constructed agency culture and resources variables had
alpha reliabilities above 0.70, indicating they could be used in further analyses, although

cautiously.

82

 

 

 

lbhdmi
Correlat
Lhexpe
variable
Chapter

perform

motivan
institute
so could
ﬁeld of 1
long ten
QPCOIC
mOre lOr
Slr0ng1}.
because 1

T

Validity of agency survey context variables

Validity was assessed by conducting bivariate correlations among the variables.
Correlations in the direction predicted by theory indicate the variables’ validity.
Unexpected correlations suggest the measures may be invalid. Correlations of agency
variables were checked at both individual and agency levels. This was done because in
Chapter 7 variables are used as aggregated means at the agency level. A variable which
performs differently when aggregated to the agency level may be invalid.

Theory predicts that there are a general set of beliefs about the ability and
motivations of citizens vis-a-vis agency employees. Historically, the civil service was
instituted to create a corps of employees who were not beholden to political parties and
so could work for the broader good of society. They were supposed to be experts in their
ﬁeld of work (Knott & Miller, 1987). In the area of land management, expertise included
long term planning. By deduction, one can hypothesize that employees centered in this
type of culture would feel that they are able to be more expert, less self-focused, and
more long-term focused than citizens outside the agency. Employees who feel this most
strongly may place less importance on principles of fairness like citizen inﬂuence
because they feel citizens are not well equipped to participate in decision making.

These arguments suggest that there should be negative correlations between the
belief that citizens are knowledgeable (citizen knowledge) and that their inﬂuence is
important (inﬂuence importance) on the one hand, and a set of beliefs on the other hand
that citizens are self-interested and short sighted (citizen self-interest), bureaucratic

principles are important (bureaucracy), and expertise is important (expertise).

83

These expectations were supported at the individual level because correlations
were in those directions although not all were signiﬁcant (Table 12). At the agency level
results were less clear because expertise and bureaucracy failed to have any signiﬁcant
correlations (Table 13). The questionable validity of the bureaucracy and expertise
variables is supported by qualitative data collected during the non-response bias phone
interviews. Respondents expressed confusion and contradictory answers when
attempting to answer many of the bureaucracy and expertise items. The items involved
making tradeoffs between citizen participation and other values and many respondents
felt both were important in different ways and couldn’t answer. Some terms in the
questions, e. g. “standard professional practices” and “expert” were open to multiple
interpretations. For these reasons, it was decided not to use the bureaucracy and expertise
variables in further analyses.

In contrast, the fairness importance and performance variables appeared to be
valid because they consistently correlated positively with each other. At the individual
level almost all the correlations were signiﬁcant. At the agency level, each variable
correlated signiﬁcantly with at least one other fairness variable. Thus the fairness
importance and performance variables are valid and reliable and can be used in further
analyses.

Finally, in addition to reliability and validity, it is useful to assess whether the
data for the variables are normally distributed. Normality is an important assumption of
statistics like t-tests and regression which are used in this study. In order to assess
normality of the data, statistics were calculated to see if the distribution for each variable

was skewed to one side (skewness) or had a larger or smaller than normal peak in the

84

mo.vne

 

 

 

 

.:.... .2. .2. .2. .R. .2. .2. :. .2. :. 8. .2. .2. .2. .2.-.2.- 8.- .2. 8058282282022
.2.. .E. .2. .2. .2.. .2. .2. .2. 2. .8. .2. .2. .2. .2.-.:.-.2. .2. 8.2828058522

.22. .2. .2. .2. .:. .2. .2.. .3. .2. .2. .2. .2. S.- .2: 8.- .2. oocasceeaauoeﬁarsﬁ

.2. .2. .2.. .2. .:. .2. .2. .2. :. .2. .2. S. .2.- 2. 8. 005.52.88.80

.2. .2. .2. :. .2. .2. .2. S.- .2. .2. 8-2.- 3.- :. 35.82.822.582

.2. .2.. .2. .2. .2. .2. .2. .2. .2. 22.-.:.- 2..- .2. Beater-.8825

.2.. .2. .2. .8. .2. .:. .2. .2. .2.- .2.- 8.- .2. 858.28 8228882

.2. .2. .2. .2. 8. .2. .2. 8. .2.- 2.- :. Becca-28.88522

.3. .2. .2. .2.. .2. .2. S.- :.- .2.- .2. 852888588322

.2.. .2.. .2. .2. .2. 3. .2.- 2..- .2. 8528828822233

.3. .2. .2.. .2. we. 8.- S. 2. 0858822358

.2. .:.. .2.. S. 3.- 3. 8. 838522832

.8. .3”. S. .2.- .2.- .2. 85285285225

.:.. 8.- 2.. 2.- 8. 852882 8:38.882

co. 8.. cor 8. 352.895 3085 ham

*3”. 2.0—. 3:... amicaxm

*2. Km: hoauosaosm

.2.- 2.2352... 526

282.65. 826

.52 .2.-:- 6 .22 2 .2 .2.. 52 2.5 6 E = E a... m m mo 05 use;

 

 

.mmm 2 mwm 28.2 momma 295m :23 33222, >25? 28:8 2022028 .26— 5220 .3 033-

85

.:.... 22...... :L :2 2. 2... ...... _ .:.. ......... .L .2 : .2 .1... L I V... .1.» 0.45:...)

6.... 02.3.. 0.95% a .:..3 94.3.2.3.» .Auzum... 352:... v..:...:....utcu 79>... kiiwwwx 4.-. 07::

 

mo.Vn*

 

 

 

 

8. we. 3. we. 3. .3. 3. 2. 3. 3.- 2. 3.. 8. 3. E.- .3.- 3.- 3. 8888883882...
.3. E. .3. 2w. .2. .3. 3. 2.. 2..- 3. 2. E. K. 3.- 3.- 3.- 3. 85::38358885
3. .3. 3. .3. .3. 2. 3. 2. 3 2.. 5 2. 3.- 3:3.- .3. 8§§ot888oaﬁzsﬁ

.3. S. E. 3. 3. 3. 3. .m. S. a. .3. 2..- 22.-.3.- .3. 85.52.882.80

G. .3. .3. 2.. R. 3. 9.. E. .3. .3. 3.- 2..- .3.- .3. 88:82.8 £382

3. 3. .3. 2. 3. 3.- 3 3. 2.. 2. 2. 3.- R. cognac-2.88822:

.3. 3. 3. 3. 3. on 3. .2. E.- Sc 3.. 3. 8:...8288358882

3. 8. 2. 9.. 3. .3. .2. $.- 3; 3.- .2. 85582888882...

3. 2. 2.- 3. 2. S. S. a. 3.- 3. 88885388828...

3. 3. .3. .3. .3. S.- 3; 3.- 3. 8:885_.88a§§£

8. 3. 2.. 2.. 3.- 2m. 3.- 2.. 8838538280

3. Q. 8. $.- 2..- z..- 3. 855853.332

3. 3. 3.- 3.- 3.- 8. 8:885:88:

.33. 3.- 3.. .3.- .3. 8828522558882

3.. 2..- La.- ..mm. eons-22.8.3882. 2n...

3. 2.. E.- 8283.

2. cm... xoﬁoamohsm

3a.- “mo-285.28 52.5

0380—305. 5220

.20.. .:.:- mo .22 2 .2 .:.-2 .8 E... 6 22 = E E m m 8 Mo 2855

 

 

.m .0 02m 2928 a 5.3 832...; 30:09. macaw mecca—oboe .26. zocow< .2 03m...

86

center of the distribution (kurtosis). For both the skewness and kurtosis statistics, values
under 1 are considered normal, values between 1 and 2 are acceptable and values over 2
indicate signiﬁcant departure from normality. All the agency variables had values for
both skewness and kurtosis under 1 except citizen self-interest, expertise, and fair
outcome performance variables which were between 1 and 2. Thus the agency variables

do not violate the important statistical assumption of normality.

Factor analysis and reliability of citizen survey context variables

The citizen survey measured several context variables, including conﬂict, equal
power, and prior relationships. It also assessed personal characteristics like agency
involvement, general involvement, age, gender, and education.

There were four items measuring the amount of conﬂict. Citizens were asked
how much they agreed with the statements on a scale of Strongly Disagree, Disagree,
Neither Agree Nor Disagree, Agree, and Strongly Agree. Factor analysis showed that the
items could equally well be combined into one dimension or into two dimensions. When
divided into two dimensions, two of the items were related to emotional intensity
(“Participants began the process with strong, deeply held positions.” “Participants
expressed strong emotions in response to disagreements”), and two were related to
incompatibility of positions (“Participants took positions that were very different from
other participants.” “Positions held by participants were highly incompatible with those
of others”). Since the need of this study was for fewer, more comprehensive variables,
the four items were combined into a single conﬂict variable with an alpha reliability of

0.77.

87

There were four items asking about the equality of power distribution among
citizens with the same agreement response categories used for conﬂict items. Factor
analysis gave two dimensions. The three items grouped on the ﬁrst dimension were,
“Power was distributed equally among citizens,” “Financial resources were equally
distributed among citizens,” and “The level of knowledge about how to get what they
want was equal among citizens.” The second dimension only had one item, “Access to
[agency employees] was equally available to all citizens.” Conceptually, it is reasonable
that the fourth item would receive a different pattern of ratings because it was dependent
on agency actions, not on the characteristics of the citizens participating. In the interest
of a conceptually clear variable, the item was excluded from the variable equal power.
The remaining three items had an acceptable alpha reliability of 0.79.

A set of six items assessed the number of prior relationships among citizens and
between citizens and decision makers. These asked the total number of citizens present,
the total number of agency employees present, and the number of citizens present who
opposed the respondent’s interests. For each of these three groups, respondents were also
asked with how many of them they had a respectful relationship before the participation
began. Because there was never an intention of creating just one variable, factor analysis
was not done. Instead variables were computed by taking the ratio of the number of each
type of person with whom the respondent had a relationship divided by the total number
present of that type. Thus the ﬁrst variable, percent decision maker relationships, was the
number of agency decision makers with whom they had a prior positive relationship
divided by the total number of agency decision makers present. The second variable,

percent citizen relationships, was the ratio of the number of citizens with whom they had

88

 

 

apﬁo
parcel
whom
preser

appror

charac
how m
during

10+ ho
hmlpm-
maximr
gﬂfﬂu
Calculat
doneeat
mﬁdai

a preVlO‘:

a prior positive relationship to the total number of citizens present. The third variable,
percent opposed relationships, was the ratio of the number of opposing citizens with
whom they had a prior positive relationship to the total number of opposing citizens
present. All these variables involved proportions of one item over another, so it was not
appropriate to calculate alpha reliabilities.

The ﬁnal set of context variables on the citizen survey related to citizen
characteristics. The level of involvement in the agency was calculated by asking people
how many times they had participated with that agency for each of several techniques
during the last ﬁve years. For example, they were asked to indicate, on a scale from 0 to
10+ how many times they had submitted written comment. The number of times they
had participated was added up across all the techniques and then divided by the
maximum score they could have received (i.e. circling 10+ for every technique). This
gave the percent involvement with the agency. The level of general participation was
calculated in a similar way, only with different items. They were asked if they had ever
done each of the activities in the past ﬁve years. The number they had checked yes was
divided by the total number of items to give a percent. The eleven items were taken from
a previous study (Smith & Propst, 2001) and included general politics, (e. g. “organized a
group of people around some political issue”), local natural resource decision making
(6. g. “signed a natural resource/ environment/ land use petition”), and regional natural
resource decision making, (e. g. “Attended a hearing or meeting of the Michigan
Department of Natural Resources or USDA Forest Service”). The full list of general

involvement items can be found in Appendix A, question 11 on the sample questionnaire.

89

 

not me
areas \\
activitil
sense 0
affirms}
analysi:
responc

single i‘

l'alidz'n

lDdEpen
r00m or
HOWeye
Suggest
who Opt:
ofrhe p e
Respon d
conﬂict ;

pOSlllVe (

TL

cOnt-1300

Items within each of the general involvement and agency involvement scales were
not meant to measure the same thing, but rather to measure activity in a broad number of
areas which could then be added up. It was expected that respondents would have done
activities in some items and not in others, so the items would not be “reliable” in the
sense of being consistent. In fact, this was the case as some items were answered
afﬁrmatively by only 8% of the people and others by 40%. For this reason, factor
analysis and alpha reliabilities were not calculated. Other context factors related to
respondent characteristics like gender, age, and education were only measured with

single items so reliabilities could not be calculated.

Validity of citizen survey context variables

Validity was hard to assess with these variables because they were meant to be
independent of each other and so should not show much correlation. There was also not
room on the survey for including other items just for the purpose of assessing validity.
However, certain patterns provide hints of validity. Two of the prior relationships items
suggest how much conﬂict was present. Speciﬁcally, dividing the number of citizens
who opposed most of my interests by the total number of citizens present gives a measure
of the percent of the people present who were perceived as being opponents.
Respondents who indicated a high percentage of opponents should have also rated
conﬂict as being high. The correlation between this ratio and conﬂict was signiﬁcantly
positive ( r = 0.30, p < .001), conﬁrming the validity of the conﬂict variable.

The validity of the prior relationships variables was suggested by their

correlations with agency involvement. Presumably someone who is more involved in a

90

panicu
96 dad:

I
were at
.001; r
signiﬁc
because
natural

ICSOUICt

relation
respond
Variable
ﬁlled ou
difﬁcult
did not r
term “re:
“respectl
they “Er
ClllZ'ms C
present a-
COnﬁlSiOr
or 012.0”

eXITEme C

particular agency will know more of the staff and the other citizens present. As expected,
% decision maker relationships, % citizen relationships, and % opposed relationships
were all positively correlated with agency involvement ( r =.349, p < .001; r = .213, p <
.001; r = .180, p = .01). Agency involvement and general involvement were also
signiﬁcantly positively correlated (r = .418, p < .001). This supports their validity
because one would expect a person who is generally highly involved in politics and
natural resource decision making to also be highly involved with a speciﬁc natural
resource agency.

These validity conclusions can only be considered tentative because the prior
relationships variables were themselves subject to questionable validity based on
respondent comments and pattems of response. The prior relationships items were the
variables most often left blank. Only 25% of the 852 people returning the questionnaire
ﬁlled out all the prior relationships items. Apparently the section was confusing or
difﬁcult so many left it blank. Many wrote in question marks, some indicating that they
did not remember how many people were there. Written comments also suggested the
term “respectful relationship” was unclear. Others did not see how they could have a
“respectful relationship” with someone who “opposed most of my interests.” Although
they were told to say how many people were present, maximum numbers like 300
citizens or 50 agency staff suggest they were thinking not in terms of who was actually
present at the meeting, but in terms of all the people involved in the decision issue. This
confusion was understandable for people whose participation consisted of letter writing
or one-on—one phone conversation. For these reasons, the results have to be viewed with

extreme caution.

91

survey.
except
distribL

around

almost .
Variable
Which r.
should 1
for em]
their ex;

are desc

F actor a

C
camure 1
Because
Could all
from the .
fair PTOce

Although validity was difficult to determine for context variables from the citizen
survey, it was possible to check the data distributions for normality. For all the variables
except one, skewness and kurtosis values were i 1, which indicates an essentially normal
distribution. Agency involvement had slightly larger skewness and kurtosis values of
around 1.3, but this still indicates acceptable normality.

In conclusion, the researcher assigned, agency, and citizen context variables were
almost all normally distributed, valid and reliable. The only exceptions were agency
variables of bureaucracy and expertise and citizen variables of prior relationships, all of
which had validity problems. When these variables are included in any analysis, they
should be interpreted with caution. However, the other variables are all good candidates
for explaining why citizens gave either positive or negative evaluations of the fairness of
their experiences. In the next section, variables related to citizen evaluations of fairness

are described and explored.

Factor analysis and reliability of citizen evaluation variables

Citizen evaluations of their experiences were measured by items designed to
capture both fairness of the decision making process and fairness of the outcomes.
Because there were so many fair process items, factor analysis was done to see if they
could all be grouped together. Factor analysis showed a pattern similar to that found
from the identical items on the agency survey, indicating that the constructs underlying
fair process judgements are robust to different situations and types of persons (Table 14).

Because of this similarity, items were grouped into the same sub-variables created for the

92

4g Lo
4f Ev

Nllmbe
Varia'

agency

counes

cOmbin
L'nme
the lien
and neu

Selim-1d

Table 14. Factor structure of fair process variables.

Factor loadings

 

Q# Item wording l 2 3 4
4n The decisions were well reasoned and logical. .70 .40 .22 .24
40 There was a bias toward a particular interest, group, or person. -.69 -.22 -.03 -.39
4m It appears that information used to reach the decisions was .68 .33 .30 .26
accurate.
4c Citizens were treated unfairly. ;.___Z -.02 -.18 -.02
4a The procedures used to make decisions were fair. £5 .37 .30 .21
4p [Agency] employees were dishonest. -.62 -.10 -.51 .02
4h Citizens were unable to have an inﬂuence on the decision - 53 -.48 .06 -.15
outcomes.
4j Citizens had an inﬂuence on the choice of decision making .19 .79 .26 .17
process.
4i Citizens were able to participate directly in making decisions. .18 .79 .19 .22
4k Citizens' comments were seriously considered. .44 .62 .37 .21
4q Citizens were treated politely. .28 .19 .78 .07
41 Citizens’ questions were answered. .27 .30 .60 .33
4d Citizens were given sufﬁcient advance notiﬁcation of the .06 .18 .59 .48
opportunrty to participate.
4e The participation experience was convenient to attend. .12 .11 .15 .79
4g Local people were adequately involved. .20 .21 .08 .76
4f Everyone affected by the decisions had an opportunity to .22 .25 .49 .59

participate.

Numbers in bold indicate items which loaded on the factor and were combined as a

variable. Underlining indicates items used separately to make a different variable.

agency variables. These were general fair process, representation, inﬂuence, neutrality,

courtesy, and total fair process.

The ﬁrst factor (Table 14) was divided into two variables. Items 4a and 4c were

combined to create fair process evaluation, a general measure of process fairness.

Unfortunately, this variable had a low Alpha reliability of 0.59, probably because one of

the items was stated in the negative. The other variable consisted of items about logic

and neutrality (4m, 4n, 4o, 4p) and was termed neutrality evaluation (A = 0.85). The

second factor contained items related to inﬂuence over the decision and process (4h, 4i,

93

4j, 41
whic
into c
repre
addit:

W85 C

“Ben:
were 1
decisi
Were t

and th

Validi-

Validit:
EXamp
fair “'1 1
also be
bl‘ﬁnar
V1111‘de
of the C4

Comm-m

4j, 4k) and was termed inﬂuence evaluation (A = 0.84) The third factor contained items
which measured polite treatment and answering questions (4q, 41) which were combined
into courtesy evaluation (A = 0.68). The fourth factor included items about access and
representation (4d, 4e, 4f, 4g) and was termed representation evaluation (A = 0.76). In
addition to these sub-groupings of fairness items, a total fair process evaluation variable
was computed by averaging all the items (A = 0.92).

The analysis of Fair Outcome items was much simpler. The ﬁve items were,
“Beneﬁts and costs were distributed fairly among citizens,” “The outcomes of decisions
were unfair,” “The decisions reached were equally favorable to all citizens,” “The
decisions beneﬁted the citizens who were most deserving,” and “The decisions reached
were consistent with my personal values.” A factor analysis found only one dimension

and this had an Alpha reliability of 0.86, indicating good reliability.

Validity of citizen fairness evaluation variables

Because no other variables had been included speciﬁcally for checking validity,
validity was assessed by comparing the evaluation variables among themselves. For
example, justice theory suggests that someone who evaluates a process as procedurally
fair will also tend to ﬁnd the outcomes fair. A process judged as generally fair should
also be judged as fulﬁlling speciﬁc principles of fairness like inﬂuence. Examination of
bivariate correlations should ﬁnd positive relationships among the fairness variables.
Validity of these variables is indeed supported by simple correlations (Table 15). Some
of the correlations are also fairly large, suggesting that evaluation variables should not be

combined in the same regression equation because of multicollinearity.

94

Table 15. Citizen level correlations among citizen evaluation variables.

 

 

Variable TFPE F PE RE IE NE CE
Total fair process evaluation

Fair process evaluation .77*

Representation evaluation .76* .41*

Inﬂuence evaluation .85* .56* .53*

Neutrality evaluation .88* .68* .51 * .68*

Courtesy evaluation .74* .51* .54* .58* .60*

Fair outcome evaluation .69* .55* .42* .61 * .73* .42*
* p < .01

In conclusion, the citizen evaluation variables appear to be reliable and valid.
They are also normally distributed, as shown by skewness and kurtosis statistics which
were between plus and minus 1 for all variables. In the next section the ﬁnal type of

variable is assessed.

Factor analysis, reliability, and validity of the evaluation consequence variable

Trust in the decision makers was measured with several types of items. There
were two items related to trust in the agency: “The [agency] can be trusted to make good
decisions,” and “I trust the [agency] to make good decisions without my input.” There
were also two items related to satisfaction with the job the agency does: ”The [agency]
does its job well,” and “I am satisﬁed with [agency] decision making.” There were also
items related to support of agency decisions: “I plan to actively oppose, appeal or sue a
decision reached in this decision,” and “I plan to support the decisions reached.” Finally,
there was one concerning the relationship between the citizen and the decision makers:

“As a result of this experience the relationship between [agency employees] and myself

95

 

hasi

they

WCI’C

andk

51mm:

and co

items 2
genera
which 1
Cl’aluati
reliabili
reliabilir
related v
Vallables
measures
Validity.
fratneu-o,

in [he 5111c

 

has improved." Factor analysis of these seven items yielded one dimension, suggesting

they could be combined. The seven items had a very good alpha reliability of 0.91.
Validity of the trust in decision maker variable was not determined because there

were no other related variables. However, the data were normally distributed. Skewness

and kurtosis values were within plus and minus 1 for all three consequences variables.

Summary

This chapter presented factor analyses, alpha reliabilities for calculated variables,
and correlations among variables. Factor analysis helped determine how to combine
items and led to the creation of variables which measured more speciﬁc aspects of the
general construct. This was particularly helpful in organizing the 16 fair process items
which were used to measure agency cultural attitudes towards fairness and citizen
evaluations of the fairness of their experiences. Calculations of inter-item Alpha
reliability coefﬁcients helped assure that variables were measured with sufﬁcient
reliability by the items. Correlations showed that variables which were theoretically
related were also empirically related in the data, conﬁrming the validity of most of the
variables. Table 16 lists all the variables, indicating what construct each variable
measures, the items used to calculate each variable, and variables with questionable
validity. In the next three chapters, these variables are used to test the contextual
ﬁ'amework of perceived fairness proposed in Chapter 4. Knowing that the variables used

in the analyses were reliable and valid increases conﬁdence in the accuracy of the results.

96

 

Table 16

Constrm

 

Context

 

 

Context
Beliefs a

 

lrnportar

 

Process :
import

Process 1
perforr

\

Table 16. A summary of variables and their validity.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Construct Variable Items used (see Low
surveys in @pendix) validity
Context variables: Research er assign ed
Government level
Participation technique Q3
Context variables: Agency survey
Beliefs about citizens Citizen knowledge Q2.a, b
Citizen self-interest Q2. c, (1
Importance of expertise Bureaucracy Q2.d, g, h, i, j, k X
Expertise Q2.a, b X
Process fairness Fair process importance Q5: a, b
importance Representation importance Q5: d, e, f, g
Neutrality importance Q5: m, n ,o, p
Inﬂuence importance Q5: h, i, j, k
Courtesy importance Q5: 1, q
Total fair process importance Q5: 3, b, d, e, f, h, i, j,
k, l, m, n, o, p, q
Process fairness Fair process performance Q6: a, b
performance Representation performance Q6: (1, e, f, g
Neutrality performance Q6: m, n ,o, p
Inﬂuence performance Q6: h, i, j, k
Courtesy performance Q6: 1, q
Total fair process performance Q6: a, b, d, e, f, h, i, j,
k, l, m, n, o, p, q
Outcome fairness Fair outcome importance Q5: t, u, v, w, x
Fair outcome performance Q6: t, u, v, w, x
Participation resources Participation resources Q8: a, b, c
Context variables: Citizen survey
Prior conﬂict Conﬂict Q10: a, b, c, (1
Prior equal power distr. Equal power Q8: b, c, e
Prior respectful % Decision maker relationship Q9: a, b X
relationships % Citizen relationship Q9: c, d X
% Opposed relationship Q9: e, f X
Respondent involvement Agency involvement Q2: a, b, c, d, e, f, g
in dCCiSiOIl making General involvement Q11: a to k
Demographics Age Q12: a
Gender Q12: b
Education Q12: (1

 

97

Table 16. Continued.

 

 

Construct Variable [tents used Low
validity

Citizen evaluation variables: Citizen survey
Process fairness Fair process evaluation Q4: a, c

Representation evaluation Q4: d, e, f, g

Neutrality evaluation Q4: m, n, o, p

Inﬂuence evaluation Q4: h, i, j, k

Courtesy evaluation Q4: l,q

Total fair process evaluation

Q4: a, c, d, e, f, h, i,j,
k. 1, m. n, o- p. q

 

 

Outcome fairness Fair outcome evaluation Q5: a, b, c, d, e

Consequence of citizen evaluation: Citizen survey

Trust in decision maker Trust Q1: a, c, d, e,
Q7: a, b, e

 

98

 

 

 

 

Cl

variabl
theoret
are ans
other a
trust in
Variabl
et‘alua
prediC'
frame.

might

CFOSE

CHAPTER VII: RESULTS AND DISCUSSION: COMPARING AGENCIES

How and why do agencies differ in terms of context, fairness, and consequence
variables? Are the agency level patterns among these variables consistent with the
theoretical framework proposed in Chapter 4? In this chapter, these research questions
are answered in order to understand how the particular agency making decisions affects
other aspects of the decision making context, citizen evaluations of fairness, and citizen
trust in decision makers. The agencies are compared in terms of mean values on the
variables measured on the survey. Then patterns of agency differences in context,
evaluation and consequence variables are compared to see if they correspond to the
predictions made in Chapter 4’s contextual framework. This is done to see if the
framework applies at the agency level of analysis. The ﬁrst aspect of the context which

might inﬂuence fairness was the agency making the decisions.

Cross-agency variation on agency variables

There were several aspects of the agency which might be important. The ﬁrst was
the level of government. As explained in Chapter 5, ﬁve agencies were chosen for this
study. They represented federal, state and local level natural resource management
agencies in Michigan. At the local level, three township planning commissions were
examined. Monitor, Delhi, and Delta townships were chosen because they were fast
growing, partly rural areas. The planning commissions were often faced with deciding
how to manage development on previously undeveloped land. At the state level, the
Forest Management Division of the Michigan Department of Natural Resources GVIDNR)

was chosen because it had to make decisions about management of the state’s forest

99

resources. At the federal level, the Huron-Manistee National Forest (HMNF), located in
northern lower Michigan, was chosen because it managed federally owned forest lands in
Michigan. By design, the agencies differed on level of government. In Chapter 4
contradictory predictions were made for how level of government might affect fairness.
Local levels should be evaluated as more fair because they would have an easier time
notifying and involving citizens. However, local levels also make decisions which affect
Citizens’ day-to-day lives and so would arouse stronger reactions, making it harder to
achieve fairness. In a later section, when context-faimess patterns are examined, support
for these rival hypotheses is examined.

Another aspect of the agency was the cultural attitudes of its employees. This can
be very noticeable to citizens. As part of a lengthy letter accompanying a returned
questionnaire about the Pigeon River Advisory Board, a citizen commented:

As a result of those meetings and my personal contact with the DNR, I have come
away with the impression that the mentality of the majority of the DNR personnel is
that an adversarial relationship exists between all “civilians” and those in the
department, and deep down inside they would rather we just left them alone instead
of complicating their lives with the decisions we make at these meetings.
The perceived attitude that citizens are an annoyance or even an adversary to good
management is an expression of broader cultural issues. The historical review in Chapter
2 suggested that agencies have struggled for many years with balancing neutrality and
accountability (Knott & Miller, 1987). Kweit and Kweit (1981) theorize that agency
cultures emphasizing neutrality will downplay citizen involvement. The quote above
suggests this may be occuning in the Michigan DNR. Looking at all the agencies in this

study in terms of the cultural variables measured, what is the balance between neutrality

and accountability?

100

The ﬁrst culture variables measured were the beliefs that citizens have and can
use technical knowledge (citizen knowledge) and that they are focused on short-term self-
interests rather than societal needs (citizen self-interest). They were measured on a scale
of Deﬁnitely False (-2), More False Than True (-1), More True Than False (1), and
Deﬁnitely True (2). Beliefs about citizen knowledge ranged from -0.1 for the Michigan
DNR to 0.9 for Monitor Township, and beliefs about citizen self-interest ranged from -
0.5 for Monitor Township to 1.2 for MDNR and Delhi Township (Table 17). However,
for most agencies, citizen knowledge was positive and citizen self-interest was also
positive. This suggests that decision makers see citizens as having some expertise, but
also being focused on the short term and on their own self-interests. If citizens are seen
as taking the short-term, personally advantageous position, they may not be trusted by the
agency to have inﬂuence over decisions. Decision makers may see themselves as the
neutral arbiters among the many conﬂicting citizen desires. Does this interpretation of
the primacy of neutrality over accountability ﬁnd support in measures of agency cultural
beliefs about fairness?

The agency questionnaires measured personal beliefs about the importance of
fairness principles and the extent to which the employee’s unit achieved fairness
principles. A set of 16 items, combined on the basis of factor analysis into 6 variables,
assessed employee personal beliefs about the importance of process fairness principles.
The items were rated on a scale of Not at all Important (0), Somewhat Important (1),
Important (2), and Essential (3). Courtesy importance and neutrality importance were
rated close to “Essential” by all the agencies, while inﬂuence importance was between

Somewhat Important and Important (Table 18). The other fair process variables hovered

101

 

Table '
aeener

Again
”1813i
MDNT
Delhi
Monin
Delta

 

Respor
and 1
Mon

Means
signi
mult

Table
farmes

A 29220
MDXr
Delhi

MOnitt
13%

'49an
MDN;
Delhi

Monitc
%

RESPOI
Nor
MePins
BOnf

Table 17. Means, standard deviations and signiﬁcant differences between
agencies on citizen knowledge and citizen self-interest variables.

 

Citizen knowledge Citizen self-interest
Agency Mean St.D. Mean St.D.
HMNF 0.0 1.0 1.1., 0.9
MDNR -0.1 1.0 1.23 0.8
Delhi 0.1 1.2 1.2 c 0.6
Monitor 0.9 0.8 05., d 1.8
Delta 0.6 1.0 0.01, 1.5

 

 

 

Respondents indicated if they believed citizens had the knowledge to participate
and that citizens were self-interested. Response scale: Deﬁnitely False (2),
More False Than True (-1), More True Than False (1), and Deﬁnitely True (2)

Means marked with subscripts a and b are signiﬁcantly different, and c and d are
signiﬁcantly different at p <.05 according to Bonferroni t-tests which correct for
multiple comparisons.

Table 18. Means, standard deviations and signiﬁcant differences between agencies on
fairness importance variables.

 

 

 

 

 

Total fair process Fair process Representation
importance importance importance
Agency Mean St.D. Mean St.D. Mean St.D.
HMNF 2.2 0.4 2.4 0.5 2.1 0.5
MDNR 2.2 0.4 2.4al 0.5 2.1 0.5
Delhi 2.2 0.4 2.6 0.5 2.3 0.5
Monitor 2.4 0.2 2.8 0.3 2.4 0.3
Delta 2.5 0.2 2.91, 0.2 2.5 0.4
Inﬂuence Neutrality Courtesy Fair outcome
importance importance importance importance

Agency Mean St.D. Mean St.D. Mean St.D. Mean St.D.

 

HMNF 1.7 0.5 2.5 0.4 2.6 0.5 1.2,. 0.6

MDNR 1.6 0.5 2.6 0.4 2.6 0.5 1.3 0.5
Delhi 1.7 0.7 2.6 0.4 2.3 0.6 1.1 0.7
Monitor 1.8 0.2 2.6 0.3 2.6 0.5 1.91, 0.7
Delta 2.0 0.6 2.8 0.3 2.8 0.4 1.3 0.4

 

 

 

 

 

Respondents were asked how important each fairness item was. Response scale:
Not at all Important (0), Somewhat Important (1), Important (2), and Essential (3).
Means marked with subscripts a and b are signiﬁcantly different at p <.05 according to
Bonferroni t-tests which correct for multiple comparisons.

102

around Important. The outcome variable, fair outcome importance, was rated as
Somewhat Important for all agencies except Monitor Township which rated it as
Important. Assigning a lower importance to inﬂuence, and the highest importance to
neutrality, is consistent with the interpretation that many agency cultures still value
neutrality over accountability.

The same patterns occurred with cultural variables asking if employees feel their
unit accomplishes fairness in its decision making (Table 19). Fairness performance items
were rated on a ﬁ'equency scale of Never (0), Seldom (1), Sometimes (2), Often (3), and
Always (4). When variables were compared with each other, neutrality was one of the
most frequently achieved principles (Often) and inﬂuence was the least achieved
(Sometimes). The similarity in rankings of fairness importance and performance
principles is no surprise as one would expect agencies to put the most effort towards
those things they value. In addition, cognitive dissonance theory (Festinger, 195 7) would
suggest that the greatest value is placed on those things most consistently and easily
achieved, setting up a reinforcing feedback loop. These results again support the primacy
of neutrality over accountability in these agencies. But do agencies differ in terms of
how they balance neutrality and accountability?

The extent to which agencies balance neutrality and accountability can be seen in
the ratio of the agency score on inﬂuence to the score on neutrality. Agencies with
greater ratios give more importance to accountability compared to others. For the ratios
of inﬂuence importance to neutrality importance, differences among agencies were very
small (HMNF = .7, MDNR = .6, Delhi = .7, Monitor = .7, Delta = .7). This same lack of

agency difference was present in the ratios of inﬂuence performance to neutrality

103

Table 19. Means, standard deviations and signiﬁcant differences between agencies on
fairness performance variables.

 

 

 

 

 

 

Total fair process Fair process Representation
performance performance performance
Agency Mean St. D. Mean St.D. Mean St.D.
HMNF 2.8a 0.5 2.9, 0.6 2.9 0.5
MDNR 2.8, 0.4 3.0,, 0.6 2.73 0.6
Delhi 3.0 0.3 3.4 0.4 3.3., 0.4
Monitor 3.4., 0.2 3.8., 0.4 3.5., 0.3
Delta 3.1 0.2 3.5 0.4 3.3., 0.4
Inﬂuence _ Neutrality Courtesy Fair outcome

performance performance performance performance
Agency Mean St.D. Mean St.D. Mean St.D. Mean St.D.
HMNF 2.3 0.7 2.88 0.5 3.3 0.5 2.0., c 0.4
MDNR 2.2 0.6 2.9, 0.5 3.4 0.5 2.1., 0.5
Delhi 2.2 0.9 3.2 0.2 3.3 0.4 2.4 a 0.2
Monitor 2.8 0.3 3.6., 0.4 3.8 0.4 3.1, 0.6
Delta 2.2 0.6 3.4., 0.4 3.7 0.4 2.3., 0.4

 

 

 

 

Respondents were asked how often their unit achieved the items. Response scale:
Never (0), Seldom (1), Sometimes (2), Often (3), and Always (4).

Means marked with subscripts a and b are signiﬁcantly different, and c and d are
signiﬁcantly different at p <.05 according to Bonferroni t-tests which correct for
multiple comparisons.

performance (HMNF = .8, MDNR = .8, Delhi = .7, Monitor = .8, Delta = .6). It appears
that the emphasis on neutrality occurs in many different natural resource agencies,
consistent with studies documenting cultures of bureaucracy and expertise among natural
resource managers in agencies like the Forest Service (Kaufman, 1960). The reforms of
the 1960s and 1970s may not have gone far enough, because neutrality is still a dominant
value in comparison to citizen inﬂuence.

Even though there appeared to be an overall cultural emphasis on neutrality, were

there any agency differences in absolute values of variables? The patterns of variation

104

among agencies on cultural beliefs about citizen knowledge and citizen self-interest were
similar to each other (Table 17). Monitor and Delta townships had the highest citizen
knowledge (0.9, 0.6) and the lowest citizen self-interest scores (-.05, 0.0). There were no
differences among Huron-Manistee, MDNR and Delhi Townships on citizen knowledge
(0.0, -0.1, 0.1) and citizen self-interest (1.1, 1.2, 1.2). These results suggest that Monitor
and Delta Township supervisors had higher regard for citizens as participants in decision
making. Is this conclusion supported by patterns in the fairness variables?

Among fairness importance variables, there were only two signiﬁcant (p<.5)
differences between agency means (Table 18). Delta township commissioners felt fair
process importance was signiﬁcantly more important than did the MDNR employees (2.4
vs 2.9). In addition, Monitor township commissioners felt fair outcome importance was
more important than did HMNF employees (1.9 vs 1.2). Although not signiﬁcant, the
same relative agency rankings held for total fair process importance, inﬂuence
importance, neutrality importance, and representation importance.

In contrast to the fairness importance variables, there were many signiﬁcant
differences between agencies on fairness performance variables (Table 19). However,
the pattern of agency differences in fairness performance matched the patterns in fairness
importance. On most fairness performance variables, HMNF, MDNR, and Delhi were not
different from each other, and Monitor and Delta were also similar to each other.
Monitor had higher fairness performance scores than HMNF and MDNR on all variables
and the differences were signiﬁcant for total fair process performance, fair process
performance, representation performance, neutrality performance, and fair outcome

performance. Delta had higher scores than MDNR and HMNF on all variables except

105

inﬂuence performance, and was signiﬁcantly higher on neutrality performance and fair
outcome performance.

These patterns in the fairness variables, which show Monitor and Delta with
higher fairness importance and performance rankings than the MDNR and HMNF, match
the patterns which show that Monitor and Delta commissioners have higher rankings of
citizens as knowledgeable and lower rankings of citizens as self-interested when
compared to the HMNF and MDNR. This suggests that there are consistent cultural
differences among agencies. What might explain these differences?

Agency culture could be affected by any of a number of factors. One explanation
would be based in level of government. The HMNF and MDNR were generally lower
than Monitor and Delta in regard for citizens and fairness. Perhaps the state and federal
government agencies had similar cultures because they are staffed by permanent civil
servants trained in natural resource management. As a result they rate fairness and
involving citizens as being less important than other decision making criteria such as the
perceived good of the resource. By comparison, planning commissions are composed of
appointed or elected local citizens. Their point of view may be more similar to citizens in
terms of emphasis on fairness. Since they are also locals, they may believe other citizens
possess the knowledge and motivation to participate. The only exception to this level of
government pattern would be Delhi Township which had cultural ratings very similar to
HMNF and MDNR. The difference between Delhi and the other planning commissions
may simply be a result of the speciﬁc makeup of the Delhi planning commission. A
larger sample of townships would help determine if Delhi is different from the usual

commission.

106

The third agency related context variable was perceived participation resources.
Employees were asked how many resources (money, staff, training) the agency possessed
for public participation. Resources were measured with a response scale of Almost None
(0), A little (1), Some (2), A Lot (3). In Chapter 4 it was predicted that greater resources
should lead to citizen evaluations that were more fair. Patterns in resource availability
showed that Delhi reported the most resources (Mean = 1.9, St.Dev. = .6), Monitor
(Mean = 1.8, St.Dev. = .5) and Delta (Mean = 1.7, St.Dev. = .6) were next, then HMNF
(Mean = 1.5, St.Dev. = .7), and ﬁnally MDNR (Mean = 1.2, St.Dev. = .7). Delhi was
signiﬁcantly greater than HMNF and MDNR ( p <.05). This pattern of perceived
resources, in which Delhi reports the most, is in stark contrast to the pattern of cultural
variables on which Delhi has the lowest scores. A possible explanation is that when
respondents were asked about resources, they responded in terms of how much they
thought they should have. Thus Delhi commissioners, who put low importance on
involving citizens, would feel they have plenty of resources, whereas employees of other
agencies, who want to conduct extensive participation, would feel their resources were

inadequate.

Cross agency variation on situation variables

Having shown that agencies do vary on agency factors, do they also vary in the
characteristics of the decision making situation? There were several aspects of the
decision making situation which potentially inﬂuenced citizen evaluations of fairness.

These were the participation technique used, the amount of conﬂict present, the extent of

107

power equality among citizens, and the prior relationships among citizens and between
citizens and decision makers. How did these variables vary across agencies?

The ﬁrst aspect of the situation was participation technique. Participation
technique may strongly inﬂuence a citizen’s experience. One long-time participant
described the Friends of the Forest -— a Huron-Manistee National Forest advisory board.

It gives the opportunity for various interest groups to debate issues and perhaps ﬁnd
common ground. Gives an opportunity to get to know forest managers and to have an
input in management decisions. Normally I feel very comfortable attending
“Friends” meetings. '
This Friends of the Forest participant liked the advisory board technique and felt it met
important needs of building relationships, debating issues, ﬁnding common ground, and
inﬂuencing forest management. This respondent was clearly satisﬁed with this
technique. However, there are many other ways to involve citizens. The possible ways
citizens could participate with each agency were uncovered during the compilation of the
mailing lists used in this study (Table 20). All agencies had records of some citizens who
had written in letters, as well as others who had called a decision maker or visited the
ofﬁce in order to have a one-on-one conversation. However, there were also agency
speciﬁc techniques.

The planning commissions held their regular meetings in public and citizens were
welcome to attend and speak if they wished to comment on the decisions being discussed.
These public meetings were essentially formal public hearings. The planning
commission in Monitor Township also held an all day Visioning session designed to help

update the township master plan.

108

Table 20. Participation techniques used by the agencies.

Agency and name of participation effort Participation technique

 

Huron-Manistee National Forest (HMNF)

Letters written
Phone call, ofﬁce visit one-on-one
Forest Plan Revision meeting
Friends of the Forest advisory board

 

Michigan Department of Natural Resources — Forest Management Division (MDNR)

Letters written
Phone call, ofﬁce visit one-on-one
Open house / Compartment reviews hearing
Presque Isle Management Plan meeting
Menominee River Management Plan meeting
Lake Superior Forest pilot project meeting
Pere Marquette Friends of the Forest advisory board
Pigeon River Country Advisory Council advisory board

 

Delhi Township planning commission (Delhi)

Letters written
Phone call one-on-one
Planning commission hearing hearing

 

Monitor Township planning commission (Monitor)

Letters written
Phone call one-on-one
Planning commission hearing hearing
Master plan Visioning workshop meeting

 

Delta Township planning commission (Delta)

Letters written
Phone call one-on-one
Planning commission hearing hearing

109

The Huron-Manistee National Forest in Michigan had several ways citizens could
be involved in addition to written letters and one-on-one conversations. The Friends of
the Forest was established as a response to challenges of the Huron-Manistee’s forest
plan. As part of the settlement, the HMNF promised to give biyearly updates on progress
towards achieving the plan. The group of appellants came to function as an advisory
board, although it was an open meeting and many other individuals and groups chose to
participate regularly. In addition to the Friends of the Forest, four years ago the Huron-
Manistee began its forest plan revision process by holding 30 public meetings. This was
followed by a series of in-depth workshops in which all participants were given
opportunities to share their concerns about the plan.

Forest planning by the Michigan Department of Natural Resource’s Forest
Management Division (MDNR) offered a variety of participation techniques. Citizens
sometimes sent in written comments or met MDNR personnel one-on-one in their ofﬁces
or in the ﬁeld. Open houses were often conducted as a prelude to a compartment review.
Compartment reviews were hearings in which management decisions were made for the
next 10 years for a small area. An initial evaluation conducted for the MDNR found
mixed support for this participation technique (McDonough & Thorbum, 1997),
suggesting a need to study it in more detail. The MDNR also conducted several local
planning efforts. These included the Presque Isle Management Plan, the Menominee
River Management Plan, and the Lake Superior State Forest pilot project. These projects
usually consisted of a series of meetings in which the same people met to negotiate
management decisions for a speciﬁc forest area. Thus, except for a smaller geographic

scale, they were similar in design to the Huron-Manistee forest plan revision workshops.

llO

The MDNR also had two long-running participation efforts. One was the Pere Marquette
Friends of the Forest, which was modeled after the Huron-Manistee Friends of the Forest.
The other was a very long term advisory board: the Pigeon River Country Advisory
Council. It was unique in that it was an appointed board that met regularly for 25 years
for the purpose of making policy recommendations to the MDNR about the Pigeon River
area.

Differences between agencies in the types of participation techniques used can be
partly explained by the nature of the resource decisions they make. Planning
commissions make decisions that affect small numbers of people in very speciﬁc areas.
Decisions are often made very quickly. It may have been for these reasons that they
chose to use formal public hearings. The MDNR also chose the formal hearing type
when it did compartment reviews because it was also approaching the management task
as a small location, small set of stakeholders issue. When issues were of broader
geographic or temporal scope, such as revising the township master plan or the forest
plan, one or a series of workshops was the approach often chosen. Thus the HMNF, the
MDNR, and Monitor township all conducted discussion-oriented meetings. Finally,
agencies doing larger scope plans needed to have the plans monitored and updated. The
advisory board, which allows the same group of people to discuss the plan repeatedly,
was the choice of the HMNF and MDNR. In conclusion, partly because of the nature of
the decisions they make, the HMNF and MDNR used a higher proportion of discussion-
based, relationship-building techniques than did the planning commissions. Of the
planning commissions, Monitor Township was the only one which experimented with a

workshop-type meeting. The rationale behind these choices of technique, while logical,

111

does not necessarily mean that the technique was the best or only possible choice. Some
participation techniques may be used simply out of tradition even though other more
effective possibilities exist.

In addition to participation technique, the decision making situation could vary in
terms of the amount of conﬂict between citizens. Conﬂict was measured with four items
and a response scale of Strongly Disagree (-2), Disagree (-1), Neither Agree Nor
Disagree (0), Agree (1), Strongly Agree (2). For all agencies, conﬂict existed at least
some of the time because the means were higher than zero (Table 21). This is reasonable
because citizens will be more motivated to participate when important decisions are at
stake and an opponent with a different interest exists. Thus some level of conﬂict can
often be expected in natural resource decision making that attracts citizen involvement.
However, the amount of conﬂict deﬁnitely varied among agencies. The Huron-Manistee
had the highest level of conﬂict (0.7) and this was signiﬁcantly higher than the amount
for MDNR (0.4), Delhi (0.4), and Monitor (0.2). The Huron-Manistee also had more
conﬂict than Delta (0.5) although the difference was not signiﬁcant.

One possible reason for the higher conﬂict rating of the HMNF was that it had
fewer routine types of interactions. For example, many citizens attending a planning
commission meeting are there to get a building approved, or a site plan changed. For
those types of meetings, the conﬂict might be relatively low. The MDNR also had
relatively routine open houses in which some citizens came and just collected information
or shared some opinions. There may not have been highly charged discussions. In
contrast, HMNF participation was often organized to speciﬁcally address a controversial

management decision. The regular advisory board meetings were attended by interest

112

Table 21. Means, standard deviations and signiﬁcant differences between agencies on
decision making situation variables.

Conﬂict* Equal power* % Decision % Citizen % Opposed

maker relationshipM relationship"
relationship“
Agency Mean St.D. Mean St.D. Mean St.D. Mean St.D. Mean St.D.
HMNF 0.7a 0.6 -0.5,, 0.9 69.3,l 39.4 56.0 37.9 40.4 45.4

MDNR 0.4., 0.7 -0.4 0.8 71.62, 36.3 63.11, 36.8 52.7 44.0

 

 

 

 

 

 

Delhi 0.4., 0.8 -0.4 0.8 41.8., 43.8 41.9., 41.7 27.9 44.7
Monitor 0.2., 0.8 -0.1., 1.0 45.5., 38.3 36.3., 38.9 44.4 46.4
Delta 0.5 0.7 -O.3 0.9 47.3., 44.1 44.7., 42.4 32.6 41.6

 

*Respondents were asked their agreement with statements indicating the existence of
conﬂict or power equality. Response scale: Strongly Disagree (-2), Disagree (-1),
Neither Agree Nor Disagree (0), Agree (1), Strongly Agree (2).

”Respondents were asked how many people [decision makers, citizens, opposed
citizens] were present and with how many of those they had prior relationships.
Dividing prior relationships by total present gave percent values.

Means marked with subscripts a and b are signiﬁcantly different at p <.05 according to
Bonferroni t-tests which correct for multiple comparisons.

group representatives who tend to be very expressive of strongly held views. While these
types of meetings and participants also existed for the planning commissions and the
MDNR, the HMNF may have had a higher ratio of them.

The extent of power equality among citizens was the third aspect of the decision
making situation. Citizens often perceived differences in power and attributed
differences in process fairness to these power differences.

I am concerned about preserving aesthetic values... speciﬁcally providing peace and
quiet through control of disruptive activities, i.e. snowmobiling through the forest.

Apparently the snowmobile lobby is just too strong. Why bother making
recommendations if they are ignored?

113

For this citizen as for many others, political power meant that citizen input was not
considered seriously, a violation of the fair process principle of inﬂuence. In this study
power distribution among citizens was measured with four items on the same agreement
response scale used for the conﬂict items. All agencies had negative means, suggesting
that power tended towards being unequal (Table 21). The only signiﬁcant difference
between agencies was that HMNF had less power equality (-0.5) compared to Monitor
Township (-0.1). When looking at the non-signiﬁcant trends, the pattern is similar to that
of conﬂict. HMNF has the highest power inequality (-0.5), Delta (-0.3), MDNR (-0.4 ),
and Delhi (-0.4) are in the middle, and Monitor is the lowest (01). The fact that
agencies high on conﬂict were also low on power equality suggests that the variables are
related. Theoretically, power equality can inﬂuence the ease with which conﬂict is
resolved and inequality can exacerbate existing conﬂict (Pruitt, 1998). This suggests
equal power may inﬂuence the level of conﬂict.

The ﬁnal set of situation variables is about relationships among participants in
decision making. The theoretical framework predicted that the more relationships of
respect exist among parties to a conﬂict, the greater the likelihood they will resolve their
differences. This would lead to perceptions of greater fairness. The ﬁrst relationships
variable, % decision maker relationship, was calculated by dividing the number of agency
employees with whom the respondent had a prior relationship of respect by the total
number of employees present. All agencies had means over 40% which showed
signiﬁcant prior relationships and suggested that many of the citizens had participated
more than once (Table 21). The HMNF and MDNR were not signiﬁcantly different from

each other, and the planning commissions were not different from each other. However,

114

the HMNF (69.3%) and MDNR (71.6%) had signiﬁcantly more relationships between
citizens and decision makers than did the planning commissions (41.8%, 45.5%, 47.3%).
A similar pattern held for the percent of citizens with whom the respondent had a prior
relationship. The HMNF and MDNR had more prior citizen-citizen relationships than
did the planning commissions, although the difference was only signiﬁcant for the
MDNR. Finally, the percent relationships with citizens who were opposed to the
respondents interests did not show any agency differences. Taken together, these results
suggest that the HMNF and MDNR do a better job of helping citizens and decision
makers to get to know each other and feel respect for each other. This is no surprise as
the HMNF and MDNR use participation techniques such as discussion meetings and
advisory boards to a much greater extent than do the planning commissions. These
participation techniques facilitate the building of relationships among citizens and

between citizens and decision makers.

Cross agency variation on citizen characteristics variables

Not only can the decision making situation differ, the citizens judging the
decision can also be different in important ways. In this study, four types of citizen
characteristics were measured: prior involvement in decision making, age, gender, and
formal education.

Citizen experiences with decision making were assessed in two ways. The ﬁrst
measured the involvement with the agency and the second measured involvement in
politics and natural resource decision making in general. Agency involvement was

calculated by adding up the total number of times the person was involved in the agency

115

and dividing it by the total potential amount they could have been involved. This gave a
percentage that could be compared among agencies (Table 22). The results showed that
MDNR had signiﬁcantly higher agency involvement levels (35%) than the planning
commissions and HMNF. Delhi (18.8%), Delta (19.2%), Monitor (15.2%), and HMNF
(17.5%) all had essentially the same agency involvement levels. These results suggest
that the MDNR tends to keep the same people involved repeatedly, while the other
agencies bring in new participants.

The level of general involvement was calculated by adding up how many of a set
of eleven general political and natural resource decision making activities the citizen had
done and dividing by the total possible. General involvement results showed that MDNR
(49.7%) was signiﬁcantly higher than the HMNF (44.2%), Delhi (22.5%), Monitor
(21.7%), and Delta (21.5%). The HMNF was also signiﬁcantly higher than the planning
commissions, but the commissions were not different from each other. This result
demonstrates that the MDNR and the HMNF attract citizens who show a higher level of
civic involvement and natural resource involvement than do the planning commissions.
The higher level of natural resource involvement is not surprising since citizens attending
planning commissions may be motivated by the desire to save their neighborhood, not
because of general environmental concerns. The barriers to local participation are also
not as great, making it easier for less experienced people to participate. For example,
planning commission hearings occur in the evening near citizens’ homes. HMNF and
MDNR meetings often occur during the day at the agency ofﬁce which may be several
hours drive from where citizens live. Thus it is easier for a citizen who is not highly

experienced in general political participation to attend a planning commission meeting.

116

Table 22. Means, standard deviations and signiﬁcant differences between agencies on
citizen characteristics variables.

Agency General Age Gender Education
involvement involvement
Agency Mean St.D. Mean St.D. Mean St.D. Mean St.D. Mean St.D.

 

HMNF 17.5., 21.3 44.2, 23.2 5.0 1.3 0.15, 0.4 3.5 1.2
MDNR 35.0, 23.9 49.7., 20.7 4.9 1.4 0.1., 0.3 3.5 1.2

 

 

 

 

 

Delhi 18.8., 16.2 22.5c 15.5 4.7 1.3 0.4., 0.5 3.3 1.2
Monitor 15.2., 14.8 21 .7c 15.3 5.0 1.4 0.3 0.4 3.4 1.3
Delta 19.2., 17.2 21.5c 14.6 5.0 1.3 0.3., 0.5 3.7 1.2

 

Agency involvement and general involvement are percentages of the total possible
participation in the agency and in general politics and natural resource decision
making. Age is measured in 10 year increments, so 5 equals an age from 50 to 59.
Gender is a variable where 0 is male and 1 is female. The education response scale is
Less than high school (1), High school graduate (2), 2 year degree (3), 4 year college
degree (4), Graduate or professional degree (5).

Means marked with subscripts a, b and c are signiﬁcantly different at p <.05 according to
Bonferroni t-tests which correct for multiple comparisons.

These results for involvement are also consistent with the results for gender. A
higher proportion of participants in the planning commissions were women (30 to 40%)
when compared to the HMNF and MDNR (10%). The gender differences might be
explained by difference in subject matter. Historically, women have been excluded from
forest management decision making, but have been very involved in decision making
related to home and neighborhood.

The ﬁnal citizen characteristics were age and education. Across all agencies, the
average age of participants was between 40 and 60 years old. The average level of
education was between a two year and a four year college degree. This suggests that
participants in natural resource decision making tended to be older and more educated.

However, there were no differences in age or education of citizens between agencies.

117

Cross agency variation on fairness evaluation and consequence variables

The ﬁnal set of variables which might express cross-agency variation were citizen
fairness evaluations and trust in decision makers. Several variables measuring the fairness
of the decision making process were constructed from 16 items. Total fair process
evaluation averaged the scores on all these items. Fair process evaluation averaged the
scores on two of the 16 items which asked about general fair procedures. Representation
evaluation averaged the scores on four items about the convenience of participation and
the involvement of diverse stakeholders. Inﬂuence evaluation used items measuring the
sense of control over decision making processes and outcomes and the feeling that citizen
input was seriously considered. Neutrality evaluation combined items about lack of bias
and use of accurate information. Finally, courtesy evaluation was calculated from two
items about polite treatment and answering of questions. There was one fair outcome
variable which used ﬁve items that asked about distribution of beneﬁts and costs and that
asked if the outcome was consistent with the citizen’s values. All of the fairness
variables had agreement response scales of Strongly Disagree (-2), Disagree (-1), Neither
Agree Nor Disagree (0), Agree (1), Strongly Agree (2).

The patterns in fair process evaluations across agencies showed that HMNF,
MDNR, Monitor and Delta were all evaluated about the same (Table 23). In particular,
the HMNF and MDNR were always very similar. However, Delhi often got signiﬁcantly
less fair evaluations. The biggest difference was for inﬂuence evaluation. Delhi received
—0.3 while HMNF got 0.2, MDNR got 0.0, and Monitor and Delta received 0.1.

Evaluations of outcome fairness showed a different pattern. HMNF received less fair

118

Table 23. Means, standard deviations and signiﬁcant differences between agencies on
citizen evaluation and trust variables.

 

 

 

 

 

 

 

Total fair process Fair process Representation Inﬂuence

evaluation evaluation evaluation evaluation
Agency Mean St.D. Mean St.D. Mean St.D. Mean St.D.
HMNF 0.3a 0.7 0.4 0.9 0.4 0.8 0.23 0.9
MDNR 0.3 0.7 0.5, 0.9 0.5 0.9 0.0 0.9
Delhi 0.1., 0.8 0.1., 1.0 0.3 0.8 -0.3., 0.9
Monitor 0.4 0.8 0.5 0.9 0.7 0.8 0.1 1.0
Delta 0.4 0.8 0.4 1.0 0.7 0.8 0.1 1.0

Neutrality Courtesy Fair outcome Trust in

evaluation evaluation evaluation decision makers
Agency Mean St.D. Mean St. D. Mean St. D. Mean St. D.
HMNF 0.2 0.9 0.8, 0.6 -0.2, 0.8 0.28 0.9
MDNR 0.3 0.9 0.8, 0.7 -0.1 0.8 0.1 0.8
Delhi 0.0 1.1 0.4., 1.0 -0.1 1.0 -0.1., 1.0
Monitor 0.3 0.9 0.7 0.9 0.2., 0.8 0.3., 0.9
Delta 0.2 1.0 0.7 0.8 -0.1 0.9 0.1a 0.9

 

 

 

 

 

Means marked with different subscripts are signiﬁcantly different at p <.05 according to
Bonferroni t-tests which correct for multiple comparisons.

outcome evaluations (-0.2) than did Delhi (-0.1), but this difference was not statistically
signiﬁcant. Monitor got a statistically higher fair outcome evaluation (0.2) than did
HMNF (—0.2), and also showed a non-signiﬁcant trend towards being higher than the
other agencies. The patterns in the fairness variables also carried over to the consequence
of trust. Delhi had signiﬁcantly lower trust scores (-0.1) than did HMNF (0.2), Monitor
(0.3), and Delta (0.1). Monitor had the highest trust score, although the difference was
not statistically signiﬁcant. What might explain these differences and similarities? The

theoretical framework proposes that differences in decision making context should

119

explain citizen evaluations. This theoretical prediction is tested in the next section, by

comparing patterns in context, evaluation, and consequence variables.

Testing the theoretical framework of perceived fairness

The theoretical framework proposes three types of context factors which
potentially affect citizen fairness judgements. The context factors include those related to
the agency making the decisions, those related to the decision making situation, and those
describing the characteristics of the citizen participants. To what extent do these factors
predict perceived fairness at the agency level?

The ﬁrst set of context factors are about the agency making the decisions. One
difference among agencies was the level of government. HMNF operated at the federal
level, MDNR at the state level, and the three planning commissions at the local level. In
Chapter 4 it was predicted that level of government would be important but the direction
of inﬂuence was hard to hypothesize. Considerations related to notiﬁcation and
representation suggested the local levels would be judged as more fair. Considerations of
legislative mandates for citizen participation suggested the local and the federal levels
would be perceived as most fair. Finally, the notion that local level issues would be more
involving to citizens suggested that state and federal levels would be seen as most fair.
Which of these theories did the data support?

Consistent with the many predictions above, the results of this study do not
provide a clear indication of the effect of level of government on fairness. When means
of evaluation variables calculated for each level of government were compared, there
were two signiﬁcant differences (Table 24). In both, the federal level had the highest

fairness and the local level the lowest fairness. The ﬁrst fairness variable showing

120

Table 24. Anova analysis of differences between government levels in fairness
evaluations.

 

 

Mean score F -test
Federal State Local DF p-value

Total fair process evaluation .34 .33 .26 (2, 766) .349
Fair process evaluation .42 .47 .30 (2, 733) .134
Representation evaluation .42 .46 .52 (2, 738) .367
Inﬂuence evaluation .20a -.03., -.09., (2, 707) .001
Neutrality evaluation .23 .30 .14 (2, 731) .219
Courtesy evaluation .84, .81, .55., (2, 705) .000
Fair outcome evaluation -.19 -.15 -.07 (2, 697) .212

 

Different subscripts in the same row show signiﬁcantly different means at the 0.05 level
according to Bonferroni posthoc tests.

signiﬁcant difference was inﬂuence evaluation. The lower inﬂuence evaluation at the
local level was consistent with the prediction that local levels are perceived as less fair
because the issues affect citizens more personally. With personally involving issues,
inﬂuence on the process and outcome would be more important, so citizens would react
more negatively to any perceived lack of fairness. However, the low score on inﬂuence
evaluation for the MDNR does not ﬁt that explanation because MDNR decisions are as
distant from citizens as are HMNF decisions. The second signiﬁcant difference was for
courtesy evaluation. Perhaps courtesy is lower at the local level because the more intense
issues inspire the decision makers to be tougher. Another possible reason is that HMNF
and MDNR decision makers are employees, and courtesy to citizens is easily noticeable
by their supervisors. However planning commissioners are unpaid volunteers, so they
can more easily get away with being rude. In conclusion, while level of government
may be a very important variable, this analysis does not provide straightforward

conclusions.

121

Agency culture is another context factor. An agency where employees hold
citizen abilities and motivations in low regard should also be evaluated more negatively
by citizens. In addition, if agency employees feel fairness has low importance, citizens
may also perceive less fairness. Finally, agencies which evaluate themselves as
achieving less fairness, may also be evaluated as being less fair by citizens. These
predicted relationships were tested with bivariate correlations between citizen evaluation
variables and agency culture and resources variables (Table 25). Because the correlations
were at the agency level, the sample size was 5, making it very difﬁcult to achieve
statistical signiﬁcance. The discussion focuses on correlations with p-values less than .3
in order to identify non-signiﬁcant trends which would need to be conﬁrmed with a larger
sample of agencies or management units.

The overall pattern largely conﬁrmed the predicted relationships. Of a total of
112 correlations between agency culture and fairness evaluations, 41 had p-values less
than .3, and of these 41, only 2 were in directions opposite of predicted. An interesting
pattern in the correlations was that representation evaluation and fair outcome evaluation
were the aspects of fairness most consistently related to agency culture. Representation
may be related most strongly because it can be readily judged by citizens and different
citizens are likely to reach the same evaluation. Judging representation involves looking
around the room and noting who is, and who is not, present. Because citizen judgements
are based on easily observed objective fact, they may be more accurate and thus be more
highly correlated with the amount of effort an agency actually expends on citizen
involvement. The same argument would hold for fair outcome evaluation. The decision

outcome may be publicly announced, facilitating citizen comparison of the relative

122

.3862: Mo 26250 ohm mecca—228 3:50:25 .:2626 wowomcoa 2: E 3:05 anﬁcmﬁ-co: Beam Eon E mecca—280

3. v wig-Q *

 

 

we». 9». m3 . clwrn 3m . um: Rm. . 9m.- “ S. .1 . mam . cor 3a.. NY- 82:82 sozmﬁoﬁam
NS. sma. c3... 3.- :N. om. mam. 2.- 3m. mm. 3:. 5. 3%. mo. coca—Eaton 2.88:0 can
3. 2. 2%. mo. 3. mm. 8%. S. VS. :3. an E. 22. mo. 855.22 32:50
35. mm. :w. omr mac. 2. e3. v0: m2. 5.. uma. co. emu. _N. 35:20.23 bzmbsoz
36. *Na. van. 2. mom. No. 3%. 2.. gm: mm. mm». 0v. 2». w? commute-tom 8:265
V: . 3.. gm- . wlmﬂ was . o ﬂ .- mmm. E - mmm . em. 3N. ONE m3. Nor eunuch—88¢ cozﬁsmoaom
as. S. Rm. ﬂ 2. 8. mm. a 2 .- ea. we. 3.. o _ .- as. «o. 8558.22 $88: .5
028.2823
awe. vNa. Mam: mm: 3m. mm. mg. 00. N2. 3.. 53.. I. 36. mm. $30.5 hm.“ Ego-H.
39. «ha. 2% m _. mu ~ . ac. Sc. mm. 2N . be. «3. . Nm. K M- . Nm. oogtoqem 220850 ham
3.». 8. Rm. em. in. S. :2. me. :2. an. 5. Nb. Re. 3. 8%:2E @2500
m3. 5.. cE. 3.- mm. 3.- Rm. 2 - Rm. em. 3:. 8. wk. 3. oogtoaé 5:22:02
NE... 2. woo. oo. SQ. mo. 03-. cm. c2. Nb. can. 3. mm». 34. 853.595 8:265
N3. me. New. mmr N3. we. 39. oo. VQQ. 5. Eu. co. En. om. oocmtomﬁm :ozﬂcOmoaoM
SN. 8. Re. om - ma. _ _. new. o _. :3. 8. 8%. 2 2. Se. mm. oocaeoaﬁ $88: Em
853.595
36. mm. hum. :.- mg. 3. 3K. 0N. Nag. vw. awn. _N. no». mv. $805 :8 35H
Ra. mm: 9mm. S. 33... 54.- 3%. emf N3. 36... was». er 2m: hm: 52225—8 coNEU
m3. 3. own. 2 - SA. mm. NE. mm. $5. on. m3. Om. 3.4. 9... 032305— 5230
A .— m .. a n m ._ m .— m .— m .—
gag—go
song—«>6 53.335 :oamﬁgo 5:23:30 coaxing coaxing 282m woﬁutg
2:830 SE $2500 3:230 Z 02—ch couﬁcomoaom $805 tam Ed :38- 82:82 a. 23:8 bosom/x

 

 

 

 

 

 

 

$225»: w H 70 82:82

cc: 2:23 mocowm 53, 2232:: 22:5 ENE: mo 20>“: oogocimﬁ big-q we: 2832228 season $32 5:03.. ”mm 035-

123

outcomes of others. However, neutrality, inﬂuence, and courtesy are more subjective.
Citizen judgements will be more inﬂuenced by personal and situation characteristics in
addition to what an agency actually does.

The third agency factor was participation resources. Agencies with more
resources for conducting citizen participation should be able to do a better job, leading to
greater perceived fairness. However, this variable predicts fairness poorly. In fact,
agencies with more perceived available resources have a trend of being evaluated as less
fair by citizens (Table 25). These results suggest that either the amount of resources for
participation is not related to fairness, or there is a difference between actual resource
availability and employee perception of resource availability. Because participation
resources was measured in a subjective way by asking employees how many resources
were available, their answers may have been affected by how much they thought there
should have been. Thus employees in agencies which value participation may have been
critical of the resources available because they wanted more. In actual fact, their
agencies may have had more resources than other agencies which did not value
participation. If this interpretation is correct, then more participation resources may
actually increase perceived fairness, but this study found an opposite result because it
assessed employee perceptions. In sum, the role of participation resources is uncertain.

The next set of context factors related to the decision making situation. One
aspect of the situation was the participation technique used. Techniques emphasizing
discussion and repeated interaction were predicted to lead to greater fairness.
Correlations between the average score on participation technique for each agency and

average agency fairness evaluations conﬁrmed the prediction (Table 26). In particular,

124

agencies with more discussion-based participation were seen as being more neutral (r =
.80, p = .105) and more courteous (r = .77, p = .128), as well as having a generally fair
process (r = .73, p = .163). There was a weaker trend towards citizens feeling they had
more inﬂuence (r = .45, p = .444). There was no relationship between participation
technique and fair outcome evaluation. This is logical because the technique used is an
aspect of the process, not the outcome.

Another situational factor was the amount of conﬂict. Conﬂict showed mixed
patterns among the fair process variables because there were an equal number of positive
and negative relationships, all of which were weak (Table 26). The only reasonably
strong correlation was with fair outcome evaluation (r = -.79, p = .113), suggesting that as
conﬂict increases, fair outcomes decrease. The fact that conﬂict is most strongly
associated with outcome is consistent with the idea that conﬂict often arises out of
differences in underlying interests. Usually a person’s interests are based in achieving a
particular outcome. Thus when conﬂict is more intense, differences in interests are
probably also more extreme, and it will be harder to ﬁnd a fair outcome.

Equal power also had small fairness correlations at the agency level, although
they were more consistent than those of conﬂict. With the exception of courtesy
evaluation, equal power was associated with greater fairness. It was most strongly
related to representation evaluation (r = .78, p = .122) and fair outcome evaluation (r =
.95, p = .012). The correlation with fair outcome may be statistically signiﬁcant because
the powerful can often get the outcome they want at the expense of those with less power.

The ﬁnal situation characteristics were the prior relationships among participants.

It was predicted that if citizens had more prior relationships with decision makers, with

125

 

.AVJUWQ-U~IW~W W) .1.” Z» V.rd\n\~w.--wtﬁ
Iv \V 4 . 4 v r 1.. HR a. H
:. F—Av—:.~:~—m za—B V.:ﬁ::.=——u>rv f..f..pv——L:wm :avN_——ch rﬂ—fvxiov— avovz-wrvw‘mp—Lmv ov=~uw>la —v:~w .I:Av.-~w~rvtnv“u -Aumh~wnvl ND) VN Ava» v-m< Vm .v\h- rNt
F—I demt *JF i . . - . . » ~ . r . .

60862: :0 0:50:50 2: macaw—2:00 005—2095 .:2626 02082: 0:: E £0.22: Euoaﬁwmm-co: 26am Eon 5 28:22.50

mo. v 029d ...

 

 

:3: 9... V::. :m. :3. mm. :3. E.- :::. :N. 3:. :N. 2:. 4:. E08285 3:09:
3:. .mlml- m2. 2.. :::. :m. 2:. mm. 3.. a..- :2. 2.. :Q. «N. 2080235 3200
:3. 2.- R:. :m. 2:. :m. ~:~. S. :3. 8. 3:. :m. :2. 2.. 858:8
:::. a. :8. t3.- ::. 3.- :8. 8.- :3. :o. :2. 2.- 2:. 5.- 5:50
:::. mm. ::. :5. :::. ma. :8. .8. :2. :h. 2:. .8. 8:. 3:. :3
:3. E. :2. 2.. :::. mm. 2:». :v. :2. S. ::. 2.. 2:. S. 32822 38:8 ex.
:2. .:Ihl. Nmm. 8. .:k. «N. :Q. .8. :3. 9V.- :::. mm. :3. S. 95:82:. 22:6 .x.
2:28:32
3:. an 3:. 3. as. :v. :3: 2. 3:. 3.. ~::. em. :::. :m. 5:9: .5365 .x.
2:. .2. E. a _ .- :2. 9. :3. 3. N2. 2. :::. mm. 3:. :m. .26: Ea:
:2. :7 V3,. 3. 5. cm; R». 9.. 2:. mm. :8. .8. 3:. 8. 85.50
:2. E. :2. F. :2. 8. 3.». :v. :::. 3. :2. mp. :::. «m. 83562 5:86:55:

a .— m .— m .— m .— m .— m .— m .—

nouns—g0

Gomwm=_m>o Gomams—w>o Fatwa—«>0 nomads—«>0 :Omum=_m>o Fatwa—9’0 mwOOOuQ
080030 ham $2.300 3:05:02 00:26.: comaﬁsom2m0m 2000-a 20m :3 =33-

 

 

 

 

 

 

 

.Amomocowm m u Zv 833.8.»

Congo 0:: 2033:: 5:5 SAVES—:0 3053 52:0 no m_0>0_ 00.82.“?me 0:_:>-n 0:: 28:22:00 282:0.“ _0>0_ >0=0w< 6N 03g.

126

other citizens, and with citizens who opposed their views, the process and outcome would
be more fair. Although the correlations were small, fair process was consistently
associated with a greater percentage of relationships. This pattern was the strongest for
relationships with opponents where correlations ranged from .60 to .85 and approached
signiﬁcance at the p < .05 level. When people have a relationship of respect with their
opponent they are more likely to cooperate, leading to a process that is more fair.

The ﬁnal set of context factors describe the individual characteristics of citizen
participants. The theory predicted that greater levels of involvement with the agency and
in natural resource decision making in general would lead to greater fairness perceptions
because the citizen would feel a greater sense of control. However, the correlations for
agency involvement and general involvement showed no clear trends, suggesting that
participation experience does not operate at the agency level. Age, education, and gender
did have clear and strong agency level patterns. Age was highly correlated in a positive
direction with all the fair process variables. Education was also positively correlated with
fair process. Gender was negatively correlated with all the fair process variables except
representation evaluation. This means that older people, more highly educated people,
and men perceived more fair process. These are also the categories of pe0ple who have
higher perceptions of control over government (Smith & Propst, 2001 ). A sense of
greater control may translate into a feeling that the process is fair.

In conclusion, at the agency level, there is support for the impact of context on
fairness. However, at the agency level, does fairness predict the consequence of trust in
decision makers? Correlations support the proposition that increased fairness is

associated with trust in decision makers. Speciﬁcally, total fair process evaluation (r =

127

.93, p = .021), fair process evaluation (r = .87, p = .056), representation evaluation (r =
.78, p = .124), influence evaluation (r = .89, p = .044), neutrality evaluation (r = .86, p =
.064), courtesy evaluation (r = .67, p = .213), and fair outcome evaluation (r = .56, p =

.330) all were positively correlated with trust in decision makers.

Summary

At the agency level, do agencies differ on important context variables? Do these
differences affect citizen evaluations, and do those evaluations impact trust in decision
makers? The results reviewed in this chapter show that the agencies did differ on context
variables. In addition, these differences in agency, situation, and citizen characteristics
did generally correlate with citizen evaluations of fairness as predicted by the theoretical
framework of perceived fairness. When an agency values fairness, reports that it
achieves fairness, and believes that citizens have the knowledge to participate and are not
self-interested, citizens also are more likely to report that the agency treats them fairly. A
decision situation characterized by a discussion-based participation technique, low
conﬂict, equal power, and prior relationships between citizens and their opponents was
perceived as leading to more fair decision processes and sometimes more fair outcomes.
Agencies with older, male, and highly educated citizen participants tended to be
evaluated more fairly. When agencies were evaluated as being more fair, their decision
makers were trusted. In the next chapter, the causes of individual citizen evaluations are
explored. The analysis focuses on situation and citizen characteristics, context variables
which were measured at the individual level. This is done to further understand why

some citizens perceive greater faimess than others.

128

CHAPTER VIII: RESULTS AND DISCUSSION: CITIZEN AND SITUATION
INFLUENCES ON FAIRNESS AND TRUST

The previous chapter examined the differences between agencies on context,
evaluation, and consequence variables in order to determine if the relationships among
variables at the agency level were consistent with the theoretical framework of Chapter 4.
The results indicated that context factors related to the agency making the decisions
explained variation in evaluation variables at the agency level. However, situation and
citizen characteristics did a relatively poor job of explaining agency differences in citizen
evaluations. This is not surprising as situation and citizen characteristics were measured
at the citizen level. At the citizen level do context factors related to the situation and the
citizens inﬂuence perceived fairness in the directions predicted by the theory? Do
fairness evaluations increase trust in the decision makers? In this chapter, these questions

are answered.

Citizen evaluations of fairness and trust in decision makers

Citizen evaluations of fairness are at the core of the theoretical ﬁamework
proposed in Chapter 4. The theoretical framework proposes that the decision making
context affects fairness evaluations and these in turn affect the level of citizen trust in
decision makers. The framework speciﬁcally proposes that fairness of the decision
making process and the decision outcome increase trust in decision makers. Much prior
research has shown that positive evaluations lead to positive outcomes for the agency

(e.g. Tyler & Degoey, 1995).

129

In Chapter 6, factor analysis and reliability analysis led to the creation of five
process fairness variables. Total fair process evaluation measured overall fairness by
averaging all 16 fair process items. F air process evaluation measured general fairness,
but did so with only two fairness items (“The procedures used to make decisions were
fair,” and “Citizens were treated unfairly”). The access of all affected citizens to
decision making was assessed with representation evaluation, the ability of citizens to
inﬂuence process and outcomes was assessed with inﬂuence evaluation, the neutrality of
the process was measured with neutrality evaluation, and the politeness of the agency was
assessed with courtesy evaluation. Fair outcome evaluation was the one measure of the
perceived fairness of the outcomes.

Bivariate correlations unambiguously supported the theoretical proposition that
fairness is positively related to trust in the agency. The correlations with trust were 0.78
for total fair process evaluation, 0.65 for fair process evaluation, 0.49 for representation
evaluation, 0.69 for inﬂuence evaluation, 0.78 for neutrality evaluation, 0.54 for courtesy
evaluation, and 0.69 for fair outcome evaluation. All of these were signiﬁcant at the p <
.01 level. These correlations show that inﬂuence, neutrality and fair outcomes have a

larger impact on trust than do representation and courtesy.

Participation technique and evaluation variables

Now that the relationship of fairness with trust has been established, how do
context variables impact fairness? There were four context variables related to the
situation in which decision making took place. These were the participation technique

used, the amount of conﬂict among citizens, the equality of power among citizens, and

130

 

 

 

theam

interac
hearin
that w

intera<

and Cl
had si
evaluz
but no
courte
Questi
10 ans'
politerl
inﬂUC.
becon

andh

|

 

ﬁlnda
that t}
inter'z

 

the amount of prior relationships among citizens and between citizens and decision
makers. The ﬁrst variable, participation technique, measured the amount of interpersonal
interaction. Ranked from lowest to highest, the techniques were written, one-on-one,
hearing, meeting, and advisory board. In Chapter 4, the theoretical prediction was made
that when citizens participated in decision making using a technique with more
interaction, they would rate the decision making process and outcome as more fair.

The correlations suggest a positive relationship between participation techniques
and citizen evaluations of fair process, as predicted (Table 27). Participation technique
had signiﬁcant positive correlations with fair process evaluation (r = .09) and courtesy
evaluation (.08) and all except one of the other citizen evaluation variables had positive,
but not signiﬁcant, correlations with participation technique. The connection between
courtesy and technique is not surprising because one of the courtesy items was “citizen’s
questions were answered.” A back-and-forth discussion would allow greater opportunity
to answer questions and the dynamics of ﬁiendly conversation would encourage
politeness. A conversation also allows feedback which can increase the sense of
inﬂuence because the citizens can see their comments being considered. As techniques
become more group-based, citizens are also able to see the diversity of citizens involved
and have a sense of the representativeness of the process. In addition, humans are
fundamentally social creatures, and so enjoy personal interactions. It is not surprising
that they would have more positive impressions of techniques which allow relaxed
interactions. The preference for group-based techniques has also been found in prior

research (Smith & McDonough, 2001).

131

 

due—Lit; 23:59 .5523 2:: .mu.£....:_..>

b

332.5 :C._:..::v. .v.::._:..E_..>.l.~|l. Cx.:.ﬁw:...3:.i .:.ﬂd...:.§:rd

.3855 35 2 26030 955322
a 2865 3382. 351095 .mmEmcomﬁ—B 5362:an 2: toga :03? :8“ 2: E 83:86 82? 2865 3028:: 20m

 

 

 

 

 

 

mo. V n— *
5.- «E. ewc. e:- No. «a n. 1:. :28”.va
*5 —.- an —.- «war «war 1:.- mo: « n “r poncow Basso-m
«2. 8. 8. 8. S. 8. S. ome-
«N —.- we. 5.- vor «N ~.- 5.- co.- EoEo>_o>E 22250
«we: £5. 5. - co. No. 55. 5.- EoEo>_o>E NA25wa
a. 8. co. 3.- S.- 8. 8. nieces». 8830 .x.
E.- 8- 8.- S.- vor 3.- S.- eﬁsosse 555 .x.
vor an n. no. co. co. «5 _. wo- QEmcotﬂg 35 .x.
L... «a. L:- én. an. .3. «m... sea .25
«NN.- 5.- « ~N.- not met «act a — ~.- ~05on
5.- $5. oo- No. 5. «as. 55. 0:35—08 coumamomtwm

283596

2623336 238396 $26396 zotuégo 26.23336 guess,» 3.83%

22836 but A8580 5.52362 monoébé zetanzmmogmox 88on 55% AEbSQ-N 353.23.

 

833...; net—3.56 nonED

ave—amt? axowcoo $530 28 Joint? c868 cassava. accusing nouns 5953 macaw—ebony SN @33-

132

 

 

Conﬂu-

Conflic

situatio

|
I-

OUICOIT-

neutral

dEClSlt‘
0UlC0r
0f the
relatix
Panic
be im
lﬂlEre

0fCor
Equa-

MOI-1‘

 

Conﬂict and evaluation variables

The second aspect of the decision making situation is the amount of conﬂict.
Conﬂict is predicted to reduce fairness evaluations because in a highly conﬂicted
situation people will hold stronger and more divergent positions, making it harder to ﬁnd
outcomes that all perceive as being fair. If outcomes do not turn out as desired, the
neutrality of the decision makers may be questioned.

The correlation results strongly support the hypothesis because all seven
evaluation variables are negatively correlated with conﬂict, and ﬁve of these are
signiﬁcant (Table 27). In particular, Conﬂict is signiﬁcantly correlated (p < .05) with
total fair process evaluation (-.11), fair process evaluation (r = -.08), neutrality evaluation
(r = -.21), and fair outcome evaluation (r = -.22). These results indicate that as conﬂict in
decision making increases, citizen satisfaction decreases. It does so both in terms of
outcomes, because citizens do not get what they want, and in terms of the general fairness
of the process and whether or not the process is seen as neutral and unbiased. The
relative sizes of the correlations suggest that the outcome function is more important,
particularly when one considers that one of the items measuring neutrality evaluation can
be interpreted in terms of outcomes, i.e. “There is a lack of bias toward particular
interests, groups, and persons.” This is not surprising because conﬂict often arises out

of competing interests, i.e. differing desired outcomes (Floyd, Germain & Horst, 1996).

Equal power and evaluation variables

The third aspect of the decision making situation is the extent of power equality

among citizens. Participation theories are oﬁen presented in terms of power and power

133

sharing so it is logical to propose that decision making would be seen as less fair if power
is less equal.

As predicted, all seven citizen evaluation variables were signiﬁcantly positively
associated with equal power (Table 27). Speciﬁcally, equal power was positively
correlated with total fair process evaluation (r = .43), fair process evaluation (r = .24),
representation evaluation (r = .3 8), inﬂuence evaluation (r = .37), neutrality evaluation (r
= .3 8), courtesy evaluation (r = .22) and fair outcome evaluation (r = .41). These
correlations were the strongest of any context variable, suggesting that equal power
among citizens is a key factor in achieving perceived fairness. This is no surprise
because power inequalities can lead to biased, unfair outcomes as well as processes
which are not neutral because they give greater inﬂuence or access to certain people at
the expense of others. Research on procedural justice found that unequal power
situations enhanced the importance of fair procedures (Barrett-Howard & Tyler, 1986),
implying that equal power situations were inherently more fair to begin with so additional
assurances of fairness were unnecessary. In conclusion, the literature as well as these
results makes it clear that the distribution of power is a central element of the decision

making context.

Prior relationships and evaluation variables

The ﬁnal situation variables measure the prior relationships of respect among the
participants. Because respect is an underlying principle of procedural justice (Lind &
Tyler, 1988), if people already know each other in a relationship of respect, they should

treat each other more fairly. This means that persons who claimed to have a prior

134

relationships with the decision makers and other citizens should view the process and
outcome as more fair.

Although one would expect prior relationships to be an important variable, the
results were mixed (Table 27). The percent of decision makers with whom the
respondent had a relationship was signiﬁcantly positively related to fair process
evaluation (r = .10) and courtesy evaluation (r = .13) as well as positively, but non-
signiﬁcantly, with four of the other evaluation variables. However, percent of citizens
relationships was signiﬁcantly negatively related to fair outcome evaluation (r = -.13), as
well as non-signiﬁcantly negatively related to all but one of the other evaluation
variables. The percent of opposed citizens with whom the respondent had a relationship
had no signiﬁcant correlations with evaluation variables. Perhaps the reason for the
positive signiﬁcance of relationships with decision makers lies in the fact that decision
makers are the ones with control of the process and the outcome so relationships with
them are the most important to fairness judgements. It is unclear why the percent of
citizen relationships was negatively related to outcome fairness. These results are
counter to ﬁndings that relationships lead to more cooperative strategies (Pruitt, 1998)
which should lead to more satisfactory outcomes. The results should be taken with
caution in view of validity concerns with the prior relationships variables as outlined in

Chapter 6.

Citizen characteristics and evaluation variables

The ﬁnal type of context variable was the characteristics of the citizens evaluating

the decisions. Because fairness evaluations are based on individual perceptions, it was

135

expected that individual differences among people would lead to differences in their
evaluations. It was predicted that males, people with more education, and people with
more participation experience would perceive greater fairness.

The results partially conﬁrm the predictions (Table 27). As predicted, men and
people with more education perceive greater fairness. Older people also gave more fair
evaluations, particularly in terms of fair outcomes (r = .13). Women were more negative
on total fair process evaluation (r = -.11), representation evaluation (r = -.10), inﬂuence
evaluation (r = -.08), neutrality evaluation (r = -.08), courtesy evaluation (r = -.13), and
fair outcome evaluation (r = -. 10). More educated persons were more positive on total
fair process evaluation (r = .10), fair process evaluation (r = .10), inﬂuence evaluation (r
= .14), neutrality evaluation (r = .08), and courtesy evaluation (r = .14). These results are
consistent with previous ﬁndings from a more general sample of Michigan residents in
which men and more highly educated people felt a greater sense of control over
government (Smith & Propst, 2001), implying greater satisfaction with government
decisions.

The results about the effect of prior participation experience were more mixed. A
higher level of involvement with a particular agency was positively correlated with
courtesy evaluations (r = .08), but negatively correlated with fair outcome evaluation (r =
-.08). General level of involvement was negatively associated with representation
evaluation (r = -. 12) and fair outcome evaluation (r = -.12). General involvement also
showed a non-signiﬁcant trend towards negative correlation with the other evaluation
variables. The negative evaluations of people with more experience are contrary to

expectation. They suggest that the more involved one becomes the more critical one gets,

136

or the people who tend to get more involved do so because they are more critical in
general. These results suggest that even though participatory experience increases the
sense of control over government (Finkel, 1985), this does not necessarily mean citizens

become more satisﬁed with the decisions or decision making processes.

Summary

The theoretical framework made a number of predictions, many of which were
supported at the citizen level. There was strong support for the association of fairness
evaluations with the consequence of trust in decision makers. In addition, a number of
context factors were found to affect fairness evaluations. There was some support for the
prediction that decisions made with participation techniques having more interpersonal
interaction will be evaluated as being fair. Increased conﬂict was clearly related to less
fair citizen evaluations, while equal power was associated with more fair citizen
evaluations. As predicted, prior relationships between citizens and decision makers led to
increased fairness. However, contrary to prediction, increases in citizen-citizen
relationships were associated with less fairness. Finally, women, younger people, and
people with less education evaluated their experience as more unfair. Contrary to
prediction, people with more prior participation experience perceived less fairness. In
conclusion these ﬁndings generally support the proposed theoretical framework. Context
factors do have an impact on citizen evaluations of fairness. Fairness, in turn, is
associated with greater trust in decision makers.

In the next chapter, a multivariate test of the framework is conducted. Using

stepwise analysis of covariance, the context variables are examined together in a set of

137

multivariate regressions. The multivariate analysis allows the testing of all context
variables together to see if they independently explain variation in fairness. The
regressions also include agencies as dummy variables in order to examine the agency and
the citizen levels of variation in fairness at the same time. Finally, the equations can also
directly test interactions among variables. As a result, a more realistic picture of the

inﬂuences and consequence of perceived fairness is constructed.

138

CHAPTER IX: RESULTS AND DISCUSSION: THE COMBINED INFLUENCE
OF AGENCY, SITUATION, AND CITIZEN FACTORS ON FAIRNESS AND
TRUST

The previous two chapters sought to determine if context factors affected fairness
and if fairness inﬂuenced trust in decision makers. This was done to test the theoretical
framework proposed in Chapter 4. The framework was developed for the purpose of
understanding why citizens perceive some decision making processes and outcomes to be
more fair than others. Once these factors are better understood, agencies will be better
able to adapt their decision making processes to the characteristics of the situation and
citizens. Achieving processes and outcomes that are more fair should help to build the
relationships of trust between citizens and decision makers that facilitate cooperative
resource management. In the previous chapters, relationships among variables were
examined at the agency and citizen levels.

Chapter 7 examined the framework at the agency level and showed that agency
level factors of government level and agency culture were inﬂuential in explaining cross-
agency variation in fairness. Situation and citizen characteristics also explained variation
in fairness at the agency level. Variation in fairness was also predictive of variation in
the consequence of trust in decision makers. In Chapter 8, situation and citizen
characteristics were correlated with fairness judgements at the citizen level. Situation
factors like power equality, amount of conﬂict, participation technique, and prior
relationships explained some variation in fairness. Citizen characteristics, like prior
participation experience, gender, age, and education also helped explain fairness.

Furthermore, fairness was highly correlated with the consequence of trust. In conclusion,

139

bivariate examinations of the data largely supported the theoretical framework at both the
individual and the agency level of analyses.

In this chapter, a multilevel, multivariate analysis technique is used. Multivariate
approaches can determine which variables are most inﬂuential because all the variables
are included in the same calculation. Multivariate analysis also allows a more direct test
of the structure of the theoretical framework because context, evaluation, and
consequence variables can be analyzed together. The proposition that context factors
inﬂuence trust by ﬁrst affecting fairness evaluations can be directly tested. Finally,
multivariate approaches enable the modeling of interactions among independent
variables.

A multilevel approach is incorporated because the factors themselves operate at
multiple levels. Chapter 7 showed that agencies do differ in fairness and these
differences can be partly explained by difference between the agencies. Chapter 8
showed that individuals also differ in fairness judgements, and that these can be
explained by citizen level factors like the situation and the respondent characteristics. A
complete picture of the relative importance of the different levels, and the interactions
between the levels, can only be obtained through a multilevel analysis.

The multilevel, multivariate approach is used to answer the following research
questions: 1) Does fairness mediate the effect of context factors on the consequence of
trust? 2) Which context factors have the greatest inﬂuence on fairness? 3) How do
context factors affect each other?, and 4) Can the differences between agencies in terms
of fairness be explained by individual and situation factors, or are there true agency level

factors that affect fairness?

140

Analysis overview

The mode of analysis used here was stepwise analysis of covariance. A multiple
regression equation was estimated with a certain set of independent variables. Then
another equation with some additional independent variables was calculated. For each
equation, the amount of variation explained in the dependent variable was estimated by
R2, the coefﬁcient of multiple determination. Subtracting the R2 values gave the
additional amount of variation explained by the variables which were added to the second
equation. An F-test was performed to show if the change in R2 was signiﬁcantly different
from zero. By adding sets of variables and observing changes in R2, the relative
inﬂuences of the different types of context variables were determined.

A multilevel dimension was added with dummy variables that represent the
agencies (McClendon, 1994). A dummy variable is coded with zeros and ones. Since
there were ﬁve agencies, four dummy variables were created. In this way each agency
was uniquely determined. For example, respondents who participated in Delta Township
received a one in the Delta variable, and zeros in the Monitor, Delhi, and MDNR
variables. Respondents who participated with the HMNF received a zero on all the
agency dummy variables. When a dummy variable was included in a regression, the
regression coefﬁcient it received was equal to the difference between the mean value of
the dependent variable for that agency and for the HMNF. If the coefﬁcient was
signiﬁcant, the difference between that agency and the HMNF was signiﬁcantly different
from zero. Including dummy variables was equivalent to conducting an analysis of

variance with agency as the factor, because it tested for differences among agency means.

141

Adding dummy variables was a simple way to examine agency level variation in
fairness, because it tested to see if agencies had signiﬁcantly different means for fairness.
Citizen level variables could also be used as regular independent regression variables. If
a coefﬁcient was positive and signiﬁcant, then higher levels of the independent variable
predicted greater fairness, aﬁer holding all other variables constant. Including citizen
level variables changed the interpretation of the dummy variables. The citizen level
variables acted as covariates which meant that the agency averages were adjusted for the
presence of any relationship between the covariate and the dependent variable. For
example, if conﬂict was included in the equation with the agency dummy variables, then
any differences between agencies existed after holding conﬂict constant. Thus
differences between agencies must have been caused by something else besides
differences in the amount of conﬂict.

The stepwise analysis of covariance did not use all the variables from the previous
chapters. Because the prior relationships variables had questionable validity and because
many people leﬁ them blank, they were not included. Including them would have
reduced the sample size by half. In addition, not all of the fair process variables could be
included because they were so closely related that they explained the same variation in
the dependent variable of trust. This multi-collinearity would pose problems for the
interpretation of the variable coefﬁcients because they would vary widely depending on
whether the other related variables were included (N etter, Wasserrnan, & Whitmore,
1993). The solution was to use two fairness variables in separate analyses. The total fair
process evaluation variable combined all the fair process items in a single index, thus

capturing the shared inﬂuence of the different aspects of fair process. The fair outcome

142

evaluation variable was also an index of outcome fairness items. Separate stepwise
analyses of covariance were conducted with the two types of fairness to see how their
relationships with the context variables and with trust differed.

The other variables not included were related to the agency. The agency level
was represented by dummy variables and if any other agency factors, such as level of
government or agency culture, were included in the same equation then they became
perfectly collinear with the dummy variables. The equation could not be solved. Thus
the analysis statistically tested for differences in agency means after adjusting for all the
citizen level variables. Then the pattern of differences in agency means could be
qualitatively compared with patterns of agency culture, participation resources, and level
of government to see if they might explain the differences.

In conclusion, the following variables were included in the regression equations:
agency dummy variables (Delta, Delhi, Monitor, MDNR), characteristics of the citizens
(age, education, gender, general involvement), characteristics of the situation
(participation technique, equal power, conﬂict), measures of fairness (total fair process
evaluation, fair outcome evaluation), and a measure of consequence (trust). In the next
section, the stepwise analyses of covariance which used these variables to answer the

research questions are presented.

Testing the mediation of fair process
A central aspect of the theoretical framework of perceived fairness is that the
inﬂuence of the context on trust in decision makers is indirectly channeled through

perceptions of fairness in decision making. Context factors operate on a particular event

143

which the citizen perceives and evaluates. Based on the evaluations, the citizen’s level of
trust in decision makers is modiﬁed. These considerations can be rephrased as the
question, does fair process mediate the effect of context factors on the consequence of
trust? Stepwise regression analysis can be used to answer this question.

The ﬁrst set of regression models demonstrate that context variables have very
little direct inﬂuence on trust after controlling for fair process. In the ﬁrst model, total
fair process evaluation is used as the only independent variable predicting trust. The
model is signiﬁcant and explains 64% of the variation in trust (Table 28). The second
model adds additional agency dummy variables, the four citizen characteristic variables,
and three situation variables. Although the change in R2 is signiﬁcant (p = .017), adding
these variables increases explained variation in trust by only 2%. The small direct
inﬂuence of context on trust can also be seen in the independent variable standardized
coefﬁcients, or Beta values (Table 29). Only equal power and conﬂict have signiﬁcant
coefﬁcients, and these coefﬁcients are small (0.10, -0.06). These results conﬁrm half of
the mediation effect: fairness does inﬂuence trust and context has very little direct effect
on trust. The second part of the mediation is to test whether context has an inﬂuence on
fairness.

The inﬂuence of context on fairness can be seen in the next set of stepwise
regressions in which fairness is the dependent variable. Four models were compared
which contained different set of variables (Tables 30 and 31). The third model is the
most parsimonious because it explains almost the same amount of variation (Adj R2 =
0.24) with fewer variables than the fourth model (Adj R2 = 0.25). The third model

contains seven independent context variables which have signiﬁcant effects on fairness

144

Sesame 25 55 @235 50389 3.2.: .8 2:9» :88 05 5 3:20.58 05 83w xocown :08 .85
808808 San 2E. ape—pat? >888 zocowm 08 M292 55 .289 .8282 .355 88956 5883.3 09 23 E 83:82

 

 

 

3.5.5 5.: .5 N 55.5 25.5 55.5 5.... N .5 R55 .5 55.5 .8“ .5 55m. .5 55% .5 555.5 MW.
55.5.. 55.5 5:. 5.0 Nod- 5.5 05.5 5.5- 5.5 55 No.0- mu... m N
88 we.
5a.: m ~
8.53598 58%
835.882 .8385 285539 5 8.862% N m
8.5589 855A858R 835% 8.8sz 888.6 nozuuanm wMV MZQ: EEQ .858: SEQ hubﬁcs 850—2

 

 

 

 

835.29. 585888:

838:9, 80555055 2: 8 28 as 88396
$808 85 :32 £5 80me .8385 E as: 033.5, 82838 85 <>OUZ< 3550?. 8 358588 Son 5388558 ”am 2%.—-

 

SE88 68888 88585th .8285 :83 £589,782: 128% coca—ow

 

.8588“. .owa £292 .380 .8832 .889 deans—«>0 $088 .85 :39 h 5.5 No.5 8.5 55.5 555.5 N
Scan—«.6 8085 LE 32 856 86 8.5 8.5 85.5 ~
5888 % ewes”? 38.: .8\
838.23 $8883 8:8 we. we a: .3. a: .198. 58:

“mo—pat? 80555055 25 .8 28
mm nouns—go $308 85 .89 £5 Eve—«E 86805 E 6.5 033:; 805885 85 <>OUZ< 0935on .8 $883 882 ”mm 2%.—-

145

88.588 288.» .8885 8 895888 25. v 8 888 888.58 5.885588 08: 888858 888.55 855 888w.“ .v 8588 8
588253.. $288 25 58 "922: 88385 85888 .8088 88.5 88 .8 88> :88 25 5 388.58 25 835 .888 88 8.5
888588 885 2.:- 85858, .8885 8858 88 M9292 58 .889 .8882 .889 885588 3805888 88 585 E 80882

 

V55 .5 555 .5 5N 5 .5 555 .5 m- 55 .5 N VN .5 N 55 .5 5N N .5 5 N m. .5 555 .5 N M 5 .5 555 .5 M55
55.5- 2.5- 55.5 3.5 585- 55.5- MN.5 55.5 3555- aNN.5- 585- £55- m 5
N N 5.5 N V5.5 555.5 N 55.5 M5N .5 555.5 5N N .5 N5N .5 555.5 5N 5.5 N 55.5 Ma.
5—.5- 55.5 and 585- 55.5- 585 55.5 55.5- 5N.5- ——.5- 55.5- m m
555 .5 E: .5 N 55 .5 N w 5 .5 555 .5 555 .5 N 5 N .5 5N N .5 55.5
5N.5- 55.5- 2.5 55.5 55.5 5N.5- 55.5- 55.5- m N

N35 555.5 555.5 53.5 5.8
55.5 5N5- 55.5 55.5 m 9

 

8.58358 58%
x 8.38% 8358.88 8.35% 888.595 H m
835% 8.5580 858N858m 835% 8.8sz 858C 85835.5 855‘ ~N>NQSN .2N8Q 8582 SNmQ 8582

 

 

 

 

 

8888; 5858855

85888 @888 :8 88 8888., 885885 8.5 <>OUZ< 88388 8 8885.88 885 58858585 9m 28,—.

 

85825». x 8385 8:98 .8580 6:68:08 853858d8305 8:88

 

8550335 .888 .528 £2.88... a»... .828: 8.5 8:52 .88 38 :8 3o Rd 88 8
8:88 628888 8588585 .838 8:98
8888238 88.8» .8585 8858358 .oma .5292 .289 .8582 .889 555.5 2.5 5N5 5N.5 555.5 m
828338 8885 .8585 828858 .858 £292 .289 .8582 .889 555.5 55.5 55.5 55.5 555.5 N
9292 .289 .8582 .889 N55.5 no.5 No.5 no.5 N555 9
.8888 ~N 8388 8508 8k
3353.. 38888 8.5.8 ma. .2 .8 NE. .8 .:.. .8. 88:

858.88 8088 .58 988 2588.» 885885 8.5 <>OUZ< 85588 .8 5888.8 8502 ”on 8:89.

146

(Table 31). Thus context factors clearly have an effect on fair process.

To summarize, to what extent does fair process mediate the impact of context on
trust? This can be calculated using a technique analogous to the calculation of direct and
indirect effects in path analysis (Alwin & Hauser, 1981; Pigozzi, pers. comm.). In path
analysis, the indirect effect of a variable is calculated by multiplying the Beta coefﬁcients
which lie along the path. In this situation, for example, the indirect effect of equal power
on trust is the multiplication of the Beta coefﬁcient for equal power when fairness is the
dependent variable (Table 31, Model 3) with the Beta coefﬁcient of fairness when trust is
the dependent variable (Table 29, Model 1). The product of the two Betas gives the
amount of inﬂuence which is mediated by fairness.

However, in this situation the combined inﬂuence of all the context variables on
fairness needs to be calculated. This combined inﬂuence is captured in the R statistic,
which is the multiple correlation. The R statistic is identical to the Beta coefﬁcient when
there is only one independent variable. By multiplying R statistics, the direct and indirect
effects of a group of variables can be compared.

The regression with context variables as the independent variables and fair
process as the dependent variable had an R of 0.506 (Table 30, Model 3). The regression
with fair process as the independent variable and trust as the dependent variables had an
R of 0.803. Thus the mediated inﬂuence of context on trust was R = .406 (.506*.803) In
contrast, the direct effect of the context variables on trust is equal to the change in R
when the context variables are added as independent variables to an equation already
containing fairness as an independent variables and trust as the dependent variable (Table

29). Without the context variables, the model with fair process as the independent

147

variable and trust as the dependent had an R of .803. When the context variables were
added, the model’s R was .813, an increase of .010. These results show that the indirect
(mediated) effect of context on trust (R = .406) was much greater than the direct effect (R
= .010). As the framework proposes, fair process does appear to mediate the inﬂuence of

context on trust.

The relative inﬂuences of context variables on fair process

Having found support for the mediating role of fair process, the next question is to
determine which aspects of the context cause the largest changes in citizen judgements of
fair process. A stepwise analysis of covariance was performed to answer this question.
As groups of variables were entered, the change in explained variation ( R2 ) was
examined. The agency dummy variables were entered in the ﬁrst model, followed by
citizen characteristics, and ﬁnally situation variables. Agency dummy variables were
entered ﬁrst to see if there was any agency level variance in fair process. Then citizen
background characteristics were included as a group. They were entered before the
situation variables because they were expected to explain some variation and were not of
primary importance. Finally, situation characteristics were entered.

When the agency dummy variables were entered by themselves they only
explained 2% of the variation in total fair process evaluation (Table 30, Model 1). All of
them had identical means except Delhi which was signiﬁcantly lower than HMNF (Table
31, Model 1). This was exactly the same pattern found in Chapter 7 when agency means
were compared for total fair process evaluation. It suggested that for some reason

citizens in Delhi Township were particularly upset with their planning commission.

148

However, the lack of differences among agency means and the small amount of explained
variation in process fairness, suggest that fairness perceptions are largely expressed at the
individual, not the agency level.

The next block of variables was citizen characteristics. Adding them to the
equation only increased explained variation in fair process by 5% (Table 30, Model 2).
Education was positively related to perceptions of fair process (Beta = .13, p <.05), while
higher levels of general political involvement was negatively related (B = -.20, p <.05).
Although not signiﬁcant, age was positively related to fairness (B = .08) and female
gender was negatively related (B = -.06). The directions of inﬂuence were the same as
found in Chapter 8 with zero-order correlations. However, the correlation between
general political involvement and total fair process evaluation was not signiﬁcant, while
the correlation for gender was. This inconsistency in results suggests the citizen
characteristics may interact enough to affect each other’s signiﬁcance level when
included in the same regression. The only variable which was signiﬁcant both in zero-
order correlations, and in multiple regression was education. Thus the effect of education
on fair process was quite reliable. In summary, citizen characteristics explain a small, but
signiﬁcant amount of variation in fair process.

The ﬁnal block of variables added was the situation characteristics. This set of
variables was clearly the most important, as it added 17% of additional explained
variation in fair process (Table 30, Model 3). Equal power (B =.39), participation
technique (B = .08), and conﬂict (B = -.10) were all signiﬁcantly related to fair process at
the p <.05 level. The signiﬁcance and directions of inﬂuence mirrored the zero-order

correlations in Chapter 8. These results suggest that the immediate environment of the

149

decision making is the aspect of the context which has the largest impact on citizen
evaluations of the decision making process.

Thus the ﬁnal model without interactions explained 26% of the variation in total
fair process evaluations. Examination of the regression coefﬁcients showed that the
relative inﬂuence on process fairness was greatest for equal power (B = .39), then
education (B = .18), then general involvement (B = -.l6), then conﬂict (B = -.10), and
ﬁnally participation technique (B = .08). Controlling for these covariates, there were now
signiﬁcant differences among the agencies. This suggests that differences among
agencies in terms of citizens and situations had been masking true agency level variation
in process faimess.

In conclusion, the stepwise analysis of covariance largely conﬁrmed the bivariate
relationships identiﬁed in Chapter 8 at the citizen level. Examination of changes in
explained variation among models suggested that agency, citizen, and situation
characteristics all explain variation in fair process. However, situation factors,
particularly the equality of power distribution, were the most inﬂuential. This result is
not surprising given that inequalities in power distribution could easily cause biased
outcomes, lack of neutrality, and lack of respect for some individuals. All of these results
are integral to perceptions of fairness. Having shown that straightforward context effects
exist, it is also possible that they may interact. The level of one context factor could
affect how another context factor inﬂuences fairness. The next section tests for statistical

interactions.

150

Interactions in the inﬂuences of context variables on fair process

How do context factors affect each other? Does the level of one context factor
affect how other context factors inﬂuence fair process? The effect of one factor on
another was tested by including statistical interaction terms in the regression model. The
ﬁrst step was to add all possible two way interactions among context variables to the full
basic regression equation. When this was done, only one interaction was signiﬁcant.
This interaction, between equal power and education, was then added by itself to the ﬁnal
regression equation to produce Model 4 (Table 30). Adding the interaction only
increased explained variation in fair process by 1%, but this increase was signiﬁcantly
different from zero. The interaction term’s coefﬁcient (B = -.3 8) also remained highly
signiﬁcant and large (Table 31 ). The probable reason for the small increase in explained
variation was that the term was calculated from two variables already entered in the
equation. Thus it did not add much overall explanatory power. However, the interaction
did highlight an important way the context variables operate and so was kept.

When an interaction term is added, it often affects the coefﬁcients of its
constituent variables. Thus in model 4, the coefﬁcient for power changed to .75, an
increase from .39, and education slightly decreased from .18 to .13. This is not a concern
because when a signiﬁcant interaction term is included, the coefﬁcients of its constituent
variables can no longer be interpreted on their own (Agresti & Finlay, 1997). Because of
the interaction, there is no longer a single, linear relationship between equal power and
process fairness so the coefﬁcient for equal power can not be interpreted. However, the
interaction term, the equal power term, and the education term can all be combined in a

simple linear formula which can be used to plot a graph of the interaction. Because the

151

equation does not include the other coefﬁcients or the intercept, the y-intercepts of the
lines are not accurate. However, the slope of the lines, and thus the relationships between
equal power, education, and fair process are accurately depicted. The equation, using

unstandardized beta coefﬁcients from Model 4, is:

Total fair process evaluation =

0.087*education + 0.643*equal power — 0.085*equal power*education

The graph of the relationship between equal power and fair process at three levels
of education shows that for people with lower education, power equality has a greater
impact on process fairness (Figure 3). People with lower education are also often the
people who have less power in society. Thus they may be more sensitive to power

differences.

 

 

 

 

1.5 ~
3.
E 0.5 ~
M
2 ° ‘
,I; -0.5 ~ , ,
a /’ ----<h1ghschool
g -1- /’ 2yearcollege
------- graduatedegree
-1.5 J
Equalpomr

Figure 3: The interaction of power equality with education on total fair process
evaluation.

152

Differences between agency means on fair process

Now that the regression model is fully speciﬁed, the multilevel aspect of the
model can be more fully explored. The ﬁnal research question was whether the
differences between agencies could be fully explained by individual and situation
variables, or if there was residual variation at the agency level. Any variation still
existing at the agency level might be explained by agency level variation in agency
culture, participation resources, and level of government.

In the previous discussion of the stepwise analysis of covariance, passing mention
was made of the fact that when the agency dummy variables were regressed on total fair
process evaluation without any covariates, only Delhi was signiﬁcantly lower than
HMNF (Table 31, Model 1). However, when the citizen characteristic of general
involvement was added to the regression (Model 2), Delta and Monitor now had lower
fair process means than HMNF and MDNR (although the differences were not signiﬁcant
in Model 2). This makes sense because citizens involved with the planning commissions
generally had lower levels of political and natural resource involvement. People with less
involvement experience tended to have more positive fairness evaluations (B = -.20).
Thus correcting the agency fairness means for general involvement would tend to make
HMNF and MDNR means more positive, and the planning commission means more
negative. By correcting for the important characteristic of citizen experience with
participation, agency level differences in fairness evaluations were uncovered.

So what were these agency differences? In the regression that included the citizen
and situation covariates, Delhi (B = -.24), Monitor (B = -.11), and Delta (B = -.09) were

signiﬁcantly lower than HMNF (Table 31, Model 3). Additional dummy variable

153

ll

analysis with Delhi as the reference agency showed that Delhi was also signiﬁcantly
lower than MDNR and Delta. In summary, the order from most to least fair process was
HMNF, MDNR, Delta, Monitor, and Delhi. What might explain these patterns in agency
differences? Do they match agency differences on agency variables?

Agency culture variables generally showed Monitor and Delta with higher
fairness importance and fairness performance scores than the HMNF and MDNR.
HMNF and MDNR were usually close to identical. The scores for Delhi fell between the
two groups, being signiﬁcantly different from neither. How do these patterns compare to
the patterns of citizen evaluations? The main similarity between patterns of citizen
evaluations and agency culture was that Delhi was usually the lowest of the three
planning commissions on both. In addition, HMNF and MDNR were essentially
identical to each other on both. The main dissimilarity was that HMNF and MDNR
received higher citizen fairness evaluations when compared to the planning commissions,
but had lower agency culture scores on fairness importance and performance when
compared to the planning commissions.

These patterns could be explained by level of government. The types of decisions
made at the local level may often be more problematic and involve citizens more
personally. As a result citizens might be more critical of the decision process and
outcome. The increased local level sensitivity to fairness might be a much more
inﬂuential factor than agency culture and so would reduce the positive impact of agency
culture at the local level. This would lead to lower citizen evaluations of fairness for
planning commissions, even though they had a culture more supportive of citizen

participation than did the MDNR and HMNF. For state and federal agencies, culture may

154

have a smaller inﬂuence on fairness evaluations, perhaps because there are other agency
level factors that inﬂuence the fairness with which decisions are made. For example,
regulations and decision making policies may require particular citizen participation
practices, and the use of citizen input when making decisions (Taylor, 1984).

When one examines patterns of culture and citizen evaluations between agencies
in the same level of government, they are very consistent. The HMNF and MDNR had
very similar cultures and very similar evaluations. Among planning commissions, Delhi,
the planning commission with the lowest citizen evaluations, also had the lowest agency
culture scores on many of the fairness importance and performance variables. Correcting
for the level of government, agency culture may have still have some impact on citizen
evaluations.

In conclusion, the stepwise analysis of covariance reviewed so far conﬁrmed that
fair process does mediate the inﬂuence of context on trust. Most of the context variables
measured had a signiﬁcant effect on fair process. At the individual level, the situation
and citizen characteristics inﬂuenced fairness. There was a signiﬁcant interaction
between the citizen characteristic of education and the situation characteristic of equal
power. At the agency level, a combination of level of government and agency culture
appeared to explain signiﬁcant differences in agency means on fair process. These
results provide strong support for the contextual theory of perceived fairness proposed in
this study. The test of the theoretical framework is not complete, however, because the
analysis only considered fair process. In the next sections, the analysis is repeated to see

if fair outcome evaluations follow the same contextual patterns.

155

 

Ibsnn

ofcor
evahr
predic
agenc
addec
addni
FusL
rnayl
tend!
more
great:
trust.
Used1
agenc
leveL
InSur
OVEra'
the co

SuPpo

Testing the mediation of fair outcome

In the previous sections, fair process was shown to mediate much of the inﬂuence
of context on trust in decision makers. Does the same hold true for fair outcome
evaluation? A regression model with fair outcome as the only independent variable
predicting trust explained 50% of the variation in trust (Table 32, Model 1). When the
agency dummy variables, the citizen characteristics, and the situation variables were
added, an additional 7% of the variation in trust was explained (Table 32, Model 2). The
additional variation was spread among a number of context factors (Table 33, Model 2).
First, people with higher education generally had greater trust in decision makers. This
may be because more educated people themselves possess more expertise and so would
tend to trust the highly expert decision makers. Second, participation techniques with
more discussion and decision situations with more equal power also were associated with
greater trust. Third, there were signiﬁcant differences among agencies in the level of
trust. Differences among agencies make sense because some of the questionnaire items
used to measure trust referred to general trust in the agency and the opinion that the
agency is doing a good job. For this reason, although trust was measured at the citizen
level, it is partly an agency level construct. Agency level variation should occur for trust.
In summary, although several of the context variables had signiﬁcant coefﬁcients, the
overall pattern is that most of the variation in trust was explained by fair outcome, and
the context factors had little direct impact on trust. For a mediation theory to be
supported, however, it also has to be shown that context inﬂuences fair outcome.

In fact, context factors did explain 21% of variation in fair outcome evaluation in

the ﬁnal model (Table 34, Model 4). The model contained seven variables with

156

48:03 05 0:0 EASE :002509 62.: .«o 02? :008 05 8 00:0..0t20 05 003w 5:0»: £000 8.“
8065000 809 05- .003083 >883 5:0?“ 0:0 M292 :80 :an £8802 .020D 8005880 b.8583 0.8 Ben 8 80882

 

 

 

830 80.0 80.0 50 02.0 08.0 32 3.0 08.0 08.0 03.0 80.0 me
So. 8.: 02. 8.0- 8.? 2.: 5o- 8...- mg- 8.:- 8...- 3.: m N
08.0 0:.
E... m _
26.36598 30%
03855000 036% 8050368.: 08828 n m
>5:an tenekamtum 335% 3.020% .0806 :88.»an 0wa “ZQ: EEQ x886: SEQ 58k 38:

 

 

 

 

0.038.203 8080885

0050.83 828080—08 05 mo 0:0 00 8:03:30
0888: :3 55 800—08 .5660: 8 .08.: 03083 809—080: 88 <>OUZ< 083980 .8 08065000 80a :0N_E0::Sm ”mm 050,—-

 

8580 630858 :ocmﬁowtam £026: 33:0 8080238 ~80:0ﬁ:0::0w

 

89:00.60 .0w0 .MZQE 8.0a £2822 .0209 80:02.30 0888: Ed ocod nod Fwd and oood N
80:02.30 0888: :0.“ coed end end and Good _
0w=§0 N 09:30 3008 .8.\
830.2% 28882 cox-is we. a: a: .3. a: are .08. 08:

0038:; 80:53—08 05 .8
0:0 00 830205 080080 Ea 53, 80038 87:00: 8 8.8 03083 83:03: :8 <>OUZ< 0086000 mo E0888 0:02 ”mm 033.

157

0.00:0m0 0:: 0:0 0:22: 50250: 830286 080080 :35 .50 0:_0> :008 0:: 8 0080:0530 0:: 003m 000:0m0 :000 :05
8085000 0:0: 0: .0 .00_:0_:0> 0088:: 000:0w0 0:0 M752 0:0 8:09 £00802 .0:0D 8005880 0:338:80 0:0 20: 8 0:08:82

 

 

 

0666 :66 6666 0.86 6666 «.36 M666 30.6 N666 60-06 036 N26 6006 M5
5.0.. mm... a :7 and mm... mod- :06- 006 and 006 5.0- no.0 cod m 0
666 6 0 m0 6 666 6 60:6 6 006 6 0 M0- 6 606 6 mm» 6 03. 6 m 0- ~ 6 an m 6 Ma.
r :6- cod mmé 09o- wad- vod mod 86 mod- bod 5d- m m
0.666 3.66 0.36 036 366 6.0.6.6 366 :06 Ma.
m — .9. 0:5- cod 0 ud o : .o 5.0- N 2. No.0 m N
N066 2.0.6 6666 0:6 mm.
:6 mod 0 :6 mod m :
0%: 00:000. 808 80%
N 0:08:00: 0:08:00: 830% 0308.. n m
2030800205 00:80.6 003080305 81% 8.00206 L080“: 00.00035 0MV SING: .:S0Q 80.80: SEQ N000:

 

 

 

 

 

0030.20: 800200305

0030205 080080 :05 05030> 80:80:00 :05 <>OUZ< 00350000 .50 080305000 0:0: 0003800000m ”mm 030,—-

 

:0000ﬁx 0:08:00: 003080380
.0w0 X 0:08:00: 00308038: 85:00 6:08:00: 830803.80 £0300 .0:00

 

0.080235 00.00 .0050 50820 .000 .0202 .200 .8052 .000 000.0 00.0 :00 0.0.0 000.0 0
85:00 6:08:00: 003080380 £030: 8:00
8080208: .8000» £0000m 80300300 6m0 .6752 8:00 £00802 .805 Good 26 oNd ad coo-o m
.5528: 0050 .0050 .8080? .000 .0202 .200 02.82 .000 000.0 00.0 00.0 00.0 000.0 0
0202 .200 09.82 .300 000.0 00.0 00.0 00.0 000.0 _
0M=0:0 x 0M0800 N008: :0\
030:? 20000005 8:8 m:- N: N: .000 N: .50 .00- 0002

0030205 080080 :35 0308; 80300003 :05 <>OUZ< 00350000 .:0 80888: :0002 #m 030-:-

158

signiﬁcant effects on fair outcome (Table 35, Model 4). Given that context factors
explained variation in fair outcome, to what extent did fair outcome mediate the inﬂuence
of context on trust? This can be estimated with R (the multiple correlation statistic) for
the models. The model with context factors as the independent variables predicting fair
outcome had an R of .462 (Table 34, Model 3). The model with fair outcome predicting
trust had an R of .710. Thus the indirect effect of context on trust was R = .328
(.462*.710). The direct effect of context on trust equaled the change in R when context

variables were added as independent variables to an equation with fair outcome as an

independent variable predicting trust (Table 32). This change was R = .050 (.760 - .710).
Thus most of the impact of context factors was mediated through fair outcome (R = .328)
instead of being direct (R = .050). These results support the mediation role of fair

outcome proposed in the theoretical framework.

The relative inﬂuences of context variables on fair outcome
Having found support for the mediating role of fair outcome, the next question is
to determine which aspects of the context have the greatest inﬂuence on fair outcome. A

stepwise analysis of covariance was performed in which agency dummy variables were

1"

added, followed by citizen characteristics, and ﬁnally situation variables.

Agency differences explained only 3% of the variation in fair outcome (Table 34,
Model 1). However, two of the agency coefﬁcients were signiﬁcant. Monitor (B = .17)
and MDNR (B = .11) had signiﬁcantly higher fair outcome means than HMNF. This

suggets that agency factors might be important in determining outcome fairness.

159

 

3960f‘
differe
Thus 5
in the
sectio:
involx
that o
greate
relati<

still b

The next regression model included citizen variables and explained an additional
3% of variation in fair outcome (Table 34, Model 2). MDNR was no longer signiﬁcantly
different from HMNF and the difference between Monitor and HMNF was reduced (.12).
Thus some of the agency differences in fair outcome were because the agencies differed
in the types of citizens they involved. This will be discussed in detail in a following
section. Of the citizen characteristics, age (B = .10), gender (B = -.09), and general
involvement (B = -.15) had signiﬁcant coefﬁcients (Table 35, Model 2). This suggests
that older people, men, and people with less political participation experience perceived
greater outcome fairness. These directions of inﬂuence mirror the ﬁndings of bivariate
relationships in Chapter 8, suggesting they may be relatively robust. However, it should
still be noted that adding citizen characteristics only increased explained variation by 3%.

Adding situation variables, however, boosted explained variation by 15% (Table
34, Model 3). Of the situation variables, equal power (B = .35) and conﬂict (B = -.17)
were signiﬁcant (Table 35, Model 3). Again, the directions of these relationships were
consistent with the bivariate correlations in Chapter 8. Adding the situation factors
eliminated any signiﬁcant differences among agency means and also made the
coefﬁcients for age and general involvement no longer signiﬁcant. These changes
suggested that agency differences in fair outcome had been largely based in agency
differences in citizen and situation characteristics. This ﬁnding will be explained further
below. The fact that the citizen variables of age and general involvement lost their
signiﬁcance suggests possible multicollinearity between citizen characteristics and
situation. This multicollinearity could take the form of statistical interactions, tested in

the next section.

160

According to the model, situations with more equal power and less conﬂict led to
greater fair outcome evaluations. Equal power was by far the most inﬂuential context
variable. Highly unequal power would increase the likelihood of outcomes that beneﬁted
one person over another and so violated the principle of outcome fairness. The
importance of conﬂict is understandable because in a high conﬂict situation pe0ple may
be more attached to their interests and perceive greater injustice if the outcomes were not
as they thought they should be. Finally, women perceived less fair outcomes than men.
Women, who are generally less powerful members of society, would be more likely to
have their needs overlooked and so experience an unfair outcome. This last possibility
suggests an interaction between gender and power. In the next section, the model is

ﬁnally ﬁilly developed by testing for context-context interactions.

Interactions in the inﬂuences of context variables on fair outcome

The initial search for signiﬁcant interactions used the complete model (Table 34,
Model 3) as a base and added all possible two-way interactions. There were no
signiﬁcant three-way interactions. This led to two signiﬁcant interactions: participation
technique with age and participation technique with gender. The model with the two
interactions explained an additional 2% of variation in fair outcome (Table 34, Model 4).

In order to understand the nature of the interactions, they were graphed using
equations based on the estimated unstandardized regression coefﬁcients. Because both
interactions involved the same variable of participation technique, both interaction terms
had to be included in the equations. For example, the plot of the interaction of

participation technique with age was calculated with gender equaling 0.5. This was done

161

"“1‘:*‘ F11

l4.
- _

to produce a plot that was gender neutral. The plot of the interaction of participation

technique with age had the following equation:

Fair outcome evaluation =

0.235age + 0.258technique — 0.060technique*age + 0.208technique*0.5

The graph of the interaction shows that for younger people, participation
techniques with more discussion led to greater fair outcome (Figure 4). For middle-aged
persons, discussion and interaction still increased fairness, but not as dramatically. For

the elderly, the participation technique was relatively unimportant.

 

 

 

 

1.8 -
g 1.6 .
5 1.4 e
g 1.2 —
0 1 _
E 0.8 -
g 0-6i ----age=2(20-29)
.:.; 0-4 * age = 4 (40-49)
é.” 0.2 ~ ....... age= 6 (60-69)
0 I l I l l
1 2 3 4 5
Participation technique

Figure 4: The interaction of age with participation technique on fair outcome evaluation

The second interaction was between gender and participation technique. The

graph was created with the following equation. In the equation, age is held constant at a

162

vdueo
responc

Fai

than f<
discus
increa
generz
they c

desire

Fl'gllre
eVa] ual

 

value of 4, which is the age group of 40 to 50 years old. Thus the plot applies to a
respondent of average age.

Fair outcome evaluation =

- 0.835gender + 0.258technique — 0.060technique*4 + 0.208*technique*gender

The graph shows that participation technique is much more important for women
than for men (Figure 5). For women, as the technique increases in the amount of
discussion and interaction, their perceptions of fair outcome increase. Men also show an
increase but it is very small. Women may prefer discussion-based methods because of a
general preference for building and maintaining relationships. In an environment where
they can interact more freely, they may be able to get outcomes closer to what they

desire.

0.4 —
0.3 —
0.2 -
0.1 - ;

 

-0.1 - 1 2 3 4 5
-0.2 -

-o.3 -

-0.4 ~ ,
-0.5 a —rmle
-0.6 3 -' ....... fame
-0.7 -

Fair outcome evaluation

 

Participation technique

Figure 5: The interaction of gender with participation technique on fair outcome
evaluation

In conclusion, the interaction terms show that participation technique has a large

positive inﬂuence on outcome fairness for people who are younger or female. Older

163

people and men don’t care much how they are involved. These interactions explain why
participation technique and age were not signiﬁcant by themselves. For some

respondents, participation technique was positively related to fair outcome and for others
it had no relation. On the average these would tend to cancel each other out, leading to a
small or non-signiﬁcant positive impact of participation technique on fair outcome. The

same attenuation of effect would occur for age and gender.

Diﬂerences between agency means on fair outcome

The ﬁnal item to explore is the multilevel aspect of the model. The major ﬁnding
was that for fair outcome there was no signiﬁcant agency level variation in fair outcome
after adjusting for citizen and situation characteristics. The ﬁrst model which just
contained the agency dummy variables, showed a signiﬁcant difference between Monitor
and HMNF (B = .17) and between MDNR and HMNF (B = .11) (Table 34, Model 1).
When citizen variables were added, the differences between the HMNF and the planning
commissions were reduced by 0.05. This can be explained by the level of general
involvement. People who were more involved perceived less fairness. Because the
HMNF had many highly involved people, when the agency means were corrected for
general involvement, the HMNF mean became more positive, closing the gap between
itself and the planning commissions. When situation characteristics were added, the gap
was closed even more. The HMNF had more conﬂict than the other agencies, and since
conﬂict reduced perceived fairness, correcting for conﬂict would increase the mean
fairness for HMNF. The HMNF also had less equal power. Since power equality was

positively related to fairness, correcting for equal power would also tend to increase the

164

 

11138.

othe

Sum

exar

frarr

deci
mos
mec‘
Situ.
Whe

pf0<

fain

P3111

mean fairness for HMNF. The net result was to close the gap between HMNF and the
other agencies enough to make the differences in fair outcome no longer signiﬁcant.
However, there was one non-signiﬁcant difference worth considering. Monitor
had much higher fair outcome evaluation than Delhi. This difference mirrored the
difference in agency cultures. Monitor had higher fair outcome importance and
performance than Delhi, although these differences were also not signiﬁcant. These
patterns suggest the possible existence of agency level variation in fair outcome which

can be explained by the agency factor of culture.

Summary

To what extent does the theoretical framework proposed in Chapter 4 hold when
examined using multivariate, multilevel statistics? This chapter explored the theoretical
ﬁamework to answer a number of research questions. The ﬁrst was, do the fairness
variables of fair process and fair outcome mediate the inﬂuence of context on trust in
decision makers? Stepwise regressions showed that the inﬂuence of context on trust was
mostly mediated by fair outcome and fair process. Fair process did have a slightly larger
mediation role (R = .406) than fair outcome (R = .328). This may have been because
situation factors of technique, equal power, and conﬂict describe the time and place
where the decision making process occurs, so they would have a greater effect on fair
process than on fair outcome. This would lead to more mediation by fair process.

The next questions asked which context factors had the largest inﬂuence on
fairness and did they interact. Stepwise regressions showed that situation factors,

particularly equal power, had the largest impacts on both fair process and fair outcome.

165

Situations with more equal power and less conﬂict had higher perceived process and
outcome fairness. Participation techniques were associated with increased fair process
and sometimes with increased fair outcome. This was because participation technique
interacted with both gender and age in its effects on fair outcome. Younger people
perceived greater fair outcome with discussion-based techniques than older people who
were relatively unaffected by technique. Men were also largely unaffected by the
technique in their judgement of fair outcome whereas women perceived much greater
fairness if discussion-based techniques were used. These interactions show that
participation technique was an inﬂuential factor for both fair process and fair outcome,
but its effects were speciﬁc to the type of citizen. Power equality was also citizen-
speciﬁc in its effect on fair process. Persons with less education showed a stronger
positive relationship between the extent of power equality and perceived fair process.
Taking into account these subtleties, situation factors had largely similar impacts on fair
process and fair outcome evaluations.

Citizen characteristics were less inﬂuential on process and outcome fairness, but
their inclusion as a group in the regressions still signiﬁcantly increased explained
variation in fairness. The directions of inﬂuence of each citizen characteristic were the
same for both outcome and process fairness. However, between the two types of
fairness, different citizen variables were signiﬁcant. Education signiﬁcantly increased
and general political and natural resource involvement lowered process fairness. In
comparison, being female signiﬁcantly decreased and being older signiﬁcantly increased
outcome fairness. These differences, however are simply a matter of degree and

somewhat unstable. The stepwise analysis showed that the signiﬁcance of the citizen

166

 

characteri:
equation.
both proc

Ti
correctin-
residual 2
process, I

had been

 

 

been a

n0n~si

faCIOf

characteristics could vary depending on what other variables were also included in the
equation. With a larger sample all the situation characteristics might be signiﬁcant for
both process and outcome fairness.

The last aspect of the context was the agency making the decisions. After
correcting for agency differences caused by the citizens or the situation, was there
residual agency level variation that could be explained by agency factors? For fair
process, correcting for citizen and situation actually uncovered agency differences that
had been hidden. The differences could be explained by a combination of level of
government and agency culture. The differences between the planning commissions and
the Huron-Manistee and MDNR probably existed because the planning commissions
dealt with issues which hit closer to home for citizens and so were more likely to lead to
negative evaluations if anything about the process went wrong. Within each level of
government, citizen evaluations followed agency culture. Agencies which placed higher
importance on fairness, which evaluated themselves as achieving greater fairness, and
which had positive beliefs about citizens tended to be more fairly evaluated by citizens.
Perhaps because the HMNF and MDNR had largely identical agency cultures, they also

showed no signiﬁcant differences in citizen evaluations of fair process.

In contrast, without including citizen or situation covariates there were signiﬁcant

differences among agency means on fair outcome. However, once the covariates had
been added, the agency differences were much smaller and were not signiﬁcant. Still,

non-signiﬁcant differences did exist and they matched cultural differences. Thus agency

factors are probably important for both process and outcome fairness.

167

In conclusion, the differences between fair process and fair outcome are only a
matter of degree. They both mediate the inﬂuence of context on trust, although fair
process mediates more. They both are most strongly inﬂuenced by situation factors, but
subtleties of how the situation interacts with citizens differ. Citizen characteristics also
have small but signiﬁcant impacts on both, although which characteristics are signiﬁcant

differ. Finally, both show agency level variation which can be explained with agency

 

level factors, although the agency differences are not signiﬁcant for fair outcome. Taken H
together, these results provide support for the contextual model of perceived fairness
proposed in this study.

E

168

CHAPTER X: SUMMARY AND CONCLUSION

Summary

How can trust in natural resource decision makers be increased? In the modern
world of resource management, collaborative management is increasingly promoted as a
way to accomplish ecosystem management (COS, 1999). Collaboration requires
relationships of trust. However, mistrust between citizens and decision makers oﬁen
exists (Fortman & Fairfax, 1991). A possible source of this mistrust is the historical
emphasis on neutrality in natural resource decision making.

Ironically, the rise of professionalism and neutrality in government agencies at the
beginning of the century was to increase trust in government. A system of political
patronage had introduced corruption and inefﬁciency in government agencies, and
establishment of a civil service based on hiring the best professionals for the job was seen
as a solution (Box, 1998). Decision makers were expected to remain detached from
relationships with citizens who were viewed as customers who lacked knowledge about
the resource and were focused on their personal interests. Thus public participation was
largely public relations. Decisions were left in the hands of professionals who tried to
achieve the greatest good for the greatest number (Kauﬁnan, 1960).

In the 19605 and 19705, citizens became dissatisﬁed with government and
demanded greater accountability from governmental decision makers. Public
participation was seen as one way of increasing accountability. New legislation, such as
the National Forest Management Act of 1976, prescribed the use of public participation

in a general way. Agency reactions were mixed as employees struggled with the

169

contradictions between the mandates for citizen participation and their cultural beliefs
which devalued citizen expertise (Frome, 1984). From the 19805 on, public participation
underwent a gradual transformation towards increasingly emphasizing direct citizen
participation in conﬂict resolution and collaboration. These strategies required new
attitudes in which professionals saw themselves as facilitators of processes that included
citizens in decision making (Shannon, 1987).

Over the course of the 20th century, agencies went from an emphasis on neutrality
towards a greater emphasis on accountability (Knott & Miller, 1987). This shift
represents a deeply seated tradeoff in the conception of justice. Justice involves treating
everyone equally (neutrality) and respecting their rights for self-determination
(accountability). In the search for trust from citizens, agency decision makers have tried
to balance neutrality and accountability in order to reach the sometimes elusive goal of
fairness. They have done so in the hope of increasing citizen trust in government
decisions.

Evidence for the importance of fairness in decision making does not just come
from a review of the political history of the US. Fairness has also been identiﬁed in the
social-psychological ﬁeld as a pervasive human need. The psychological study of
fairness has shown that both the process used to make decisions and the decisions that
result are evaluated according to principles of fairness. Process and outcome fairness are
important to people because they invoke instrumental needs like survival and symbolic
needs like social standing in society and basic dignity (Thibaut & Walker, 1975; Lind &
Tyler, 1988; Folger, 1993). Empirically, the wide-spread use of fairness to evaluate

corporate, political, and judicial decision making has been demonstrated (Lind & Tyler,

170

1988). Fairness is also important to citizens in judgements of natural resource decision
making (Tyler & Degoey, 1995; Lauber & Knuth, 1998; Smith & McDonough, 2001).
Theorists have proposed a wide range of principles for fair procedures including
neutrality, accuracy of information, and ethicality. Other process principles like
representation of all involved, voice, and direct participation in decisions are identical to
common prescriptions for conducting public participation. Outcome faimess has been
measured in terms of equity, equality, and need (Hegtvedt, 1992). Fairness has been
shown to have the consequences of increased citizen trust in decision makers and support
for their decisions (Tyler & Degoey, 1995).

The purpose of this study was to understand how perceived fairness inﬂuences
citizen trust in a decision making context. A major goal was to determine if aspects of
the context in which decisions were made inﬂuenced citizen judgements of fairness. A
better understanding of what inﬂuences perceptions of fairness would then lead to
suggestions for decision maker strategies. The result would be greater citizen trust and so
more successful natural resource management, particularly management utilizing
collaboration.

The central role of fairness was conceptualized in a theoretical framework. The
framework hypothesized that decision making context factors, such as the amount of
conﬂict among citizens, would affect perceptions of fair process and fair outcome.
Perceptions of fairness would then inﬂuence trust in decision makers. The framework
proposed three main types of context factors: aspects related to the agency making the
decisions, the situation in which citizens participated, and the characteristics of the

citizens judging the decisions. Directions of the bivariate relationships between context

171

variables and evaluation variables and between evaluation variables and the consequence
variable were speciﬁed. The context factors were partly identiﬁed from the history of
natural resource decision making in the United States.

There were three factors related to the decision making agency that might affect
fairness. The historical review showed that employee attitudes towards citizens changed
over the years. However, because different agencies might have changed at different
rates, agencies might differ in terms of how their culture valued citizen involvement and

fairness. Agencies might also differ in the amount of resources they devoted to

 

participation. Finally, this study included agencies at the local, federal, and state level to
see if there were level of government effects.

The situation in which participation and decision making occurred might also
affect fairness judgements. During the last century, participation became important
partly because of citizen conﬂicts over how resources should be managed, suggesting that
the amount of conﬂict, as well as factors related to the resolution of conﬂict, like power
distribution and prior relationships among participants (Pruitt, 1998), should affect the
way citizens evaluate their participatory experiences.

Finally, the citizens making the judgements of decision making fairness may

l‘ .

differ on important personal characteristics. The study examined typical demographic
characteristics like age, education, and gender. It also measured prior involvement levels
in political and natural resource decision making. Because fairness evaluations are
ultimately subjective perceptions, these personal characteristics might be very important.
The context, fairness, and trust variables were measured with written surveys sent

to decision makers and citizens. Five agencies at federal, state, and local levels were

172

studied. Reliability and validity of the variable operationalizations were assessed. Most
of the variables had adequate psychometric properties, with the exception of those
measuring prior relationships and some of the agency culture variables.

The framework was tested in three ways. First, differences among agencies on
the variables were measured and their patterns interpreted. This was done to see if the
theoretical framework functioned at the agency level of analysis. Then, the situation and
citizen context variables were correlated with citizen evaluations. This identiﬁed whether
citizen level bivariate relationships were consistent with the framework. Finally, agency
and citizen levels were combined in a stepwise analysis of covariance. This helped to
determine which context factors had the greatest inﬂuence on fairness. The analysis was
guided by a set of research questions. For this reason, the study results are most clearly

presented in terms of answering these questions.

How and why do agencies differ in terms of context, fairness, and trust variables?
Agency differences were examined for all the context variables, the fairness
evaluation variables, and the consequence variable of trust in decision makers. The ﬁrst
set of context factors concerned aspects of the agency making the decisions. Agency
culture was similar across all agencies in that they all gave more importance to neutrality
than to accountability/inﬂuence. However, there were agency differences. Compared to
the Huron-Manistee National Forest (HMNF) and the Michigan Department of Natural
Resources Forest Management Division (MDNR), Monitor and Delta planning
commissions generally felt fairness principles were more important. Monitor and Delta

also gave themselves a higher level of achieving fairness principles than did HMNF and

173

MDNR. These results suggested a level of government effect whereby planning
commissioners, who are generally volunteer local citizens, feel fairness is more
important. Agency employees might discount involving citizens and achieving fairness
because of a professional allegiance to making decisions that are consistent with their
disciplinary training.

For the category of situation context factors, there were also agency differences.
The HMNF and MDNR generally used more discussion techniques than did the planning
commissions. This may have been because they often made decisions with broader
geographic and temporal scope, requiring in-depth consideration of alternatives and the
involvement of many constituencies. The HMNF also had the highest conﬂict and the
lowest level of power equality. This may have been because, compared to the other
agencies, they had a higher proportion of controversial decision making. The planning
commissions and the DNR, while they had highly controversial decisions, also had a
large number of routine decisions where there may not have been any conﬂict. This
would lower their average levels of conﬂict. The ﬁnal situation variables were about
prior relationships among participants. HMNF and MDNR had more prior citizen-
decision maker and citizen-citizen relationships than did the planning commissions. Both
of these agencies also used more discussion techniques which tend to encourage
relationship building.

The ﬁnal situation context variables were citizen characteristics. The amount of
involvement of citizens in a particular agency was highest for the MDNR compared to
the other agencies. This suggests the MDNR tends to encourage the same people to

participate repeatedly. The involvement of citizens in general politics and natural

174

 

resource decision making was highest for both the MDNR and the HMNF. Since
participation in the HMNF and MDNR is generally less convenient than in local planning
commissions, and because it is more focused on environmental issues, it would tend to
attract those who already are very involved in natural resource topics. Finally, although
the agencies did not differ on age and education of involved citizens, the HMNF and
MDNR involved very few women compared to planning commissions. This may have
been because HMNF and MDNR deal with management of resources like wildlife,
timber and minerals, topics from which women have historically been excluded.

In addition to context variables, there were also differences in citizen evaluations
and the consequence of trust in decision makers. On fair process evaluations, the general
pattern was that Delhi received the lowest ratings by citizens. For fair outcome
evaluations Delhi was at the lower end, but essentially the same as MDNR and Delta.
However, HMNF received the lowest fair outcome evaluations and Monitor was the
highest. Finally, the consequence of trust combined the patterns from process and
outcome fairness. Delhi was the lowest, Monitor was the highest, and the others fell in
between. So what might be the explanations of these patterns in evaluations and trust?
The next question addresses this by examining patterns among context, fairness and trust

variables.

At the agency level, are the directions of inﬂuence among context, fairness, and
consequence variables consistent with the theoretical framework?

Patterns of contextual inﬂuence on fairness can be summarized in terms of the
three types of context: agency, situation, and citizen. Agency level factors seemed to

predict fairness evaluations. There appeared to be a level of government effect on citizen

175

evaluations. The planning commissions generally received lower faimess evaluations,
consistent with the hypothesis that local level decisions are personally relevant to citizens
so they form more extreme judgements. The state and federal levels often deal with
forest management topics that are geographically distant and do not affect most citizens’
day-to-day lives.

Agency culture was also somewhat consistent with citizen evaluations, although
there were not many signiﬁcant correlations. When looking at trends in the correlations,
there were 44 correlations out of 119 total which had p-values less than 0.3 and only ﬁve
of these were not in the predicted direction. These data tentatively support the conclusion
that agencies with cultures that believe citizens are not self-interested and have expertise,
receive more positive citizen evaluations of fairness. Agencies which believe fairness
principles are important, and which believe that they achieve fairness, are also evaluated
as being more fair. The fact that agency culture explained agency differences in fairness
supports the theoretical framework.

The results for situation variables at the agency level were similar. Situation
factors of participation technique and equal power were positively associated with
fairness evaluations, and conﬂict was negatively associated. Participation technique may
have been effective because the choice of technique is dependent on the types of
decisions made, a level of government issue. Agency differences in prior relationships
between citizens and decision makers and between citizens and citizens had small
correlations suggesting prior relationships led to greater process fairness, and this was

particularly the case for relationships with citizens who opposed one’s views.

176

The ﬁnal set of context factors was citizen characteristics. Agency differences in
citizen level of involvement in the agency and in general politics did not clearly correlate
with perceived fairness. However, although agency differences in age and education
were small, they did predict fairness evaluations. Older and more educated people
perceived decision making processes as being more fair. Women evaluated processes as
being less fair.

In conclusion, agency factors like level of government and agency culture
explained fairness differences among agencies. Situation and citizen characteristics also
helped explain agency differences. These results suggest that the theoretical framework
of perceived fairness does operate at the agency level. Does it also operate at the level of
individual citizen judgements? The next question examines patterns of relationship at the

citizen level of analysis.

At the citizen level, do context factors related to the situation and the citizens inﬂuence
perceived fairness in the predicted directions?

Bivariate relationships of situation and citizen variables with fairness evaluations
show clear support for the theoretical framework. Among the situation variables,
participation techniques that emphasize discussion were associated with greater fairness.
This is consistent with research showing that citizens prefer small focus group-styled
participation methods (Smith & McDonough, 2001). In addition, conﬂict decreases
perceived fairness and power equality increases fairness. This is consistent with
literature on conﬂict resolution which states that it is harder to reach a resolution if

conﬂict is intense and power is unequal (Pruitt, 1998). Another ﬁnding consistent with

177

conﬂict resolution research (Pruitt, 1998) was that more prior relationships with decision
makers was associated with positive fairness evaluations.

Citizen characteristics were also largely consistent with predictions. Women,
younger people and people with less education perceived lower levels of fairness. It may
be that because most planning processes are designed and conducted by people with more
education, they are not conducted in a way accessible to people with less formal
education. When pe0ple with less education participated they may have been less
comfortable and so rated their experience more negatively. Women may have a lower
sense of control over political processes (Smith & Propst, 2001) and, because one aspect
of fairness is control and inﬂuence, may perceive less fairness.

Contrary to prediction, people with more participatory experience actually had
more negative evaluations. Other research has shown that more participation experiences
create greater feelings of control over agency decision making (Smith & Propst, 2001;
Finkel 1985), which should lead to more positive evaluations of decisions. However,
because people are motivated to participate out of concern for the decision topic, those
who participate the most probably have the strongest concerns. Thus they may have the
most negative evaluations when they perceive any deviation from fairness in process or

01111001118.

At the citizen level, does perceived fairness increase trust in the decision makers?
When evaluations of fairness were correlated with trust in decision makers, the
relationship was strongly positive. This conﬁrms procedural justice research which ﬁnds

that trust in authorities increases when decision making processes are fair (Tyler &

178

Degoey, 1995). So how do context factors affect trust? The theoretical framework
proposes that the impact of context on fairness is mediated by fairness evaluations. The

next question tests this proposition

Does fairness mediate the effect of context factors on trust?

Do context factors directly impact trust or do they ﬁrst inﬂuence fairness
evaluations and the fairness evaluations then impact trust? This proposed mediation of
fairness is based on the premise that context factors are mostly relevant to speciﬁc
decision making experiences. Thus they should be closely related to evaluations of those
experiences. By contrast, trust is a more general belief that would be affected by the
speciﬁc fairness evaluations. The mediation role of fairness was tested through the
calculation of direct and indirect effects using values from stepwise analysis of variance.
An indirect effect is equivalent to mediation because it shows how the inﬂuence of an
independent variable on the dependent variable is transmitted through an intervening
variable. The indirect effect of context on trust as mediated by fair process was .406
which was much larger than the direct effect of context on trust (.010). In a similar way,
the indirect effect of context on trust as mediated by fair outcome was .328 which was
much larger than the direct effect of context on trust (.050). These results indicate that

the framework, which assigns a central mediating role to fairness, is supported.

Which context factors explain the most variation in perceived fairness?

The multivariate regressions found that many context variables had impacts on

both fair process and fair outcome. For fair process, situation factors of conﬂict, power

179

equality, and participation technique had more inﬂuence (R2 increase of .17), than did
citizen characteristics (R2 increase of .05), or agencies (R2 increase of .03). Fair outcome
showed an ahnost identical pattern (situation R2 increase of .15, citizen R2 increase of
.03, agency R2 increase of .03). Of the situation factors, power equality, with a
standardized coefﬁcient of .39 for fair process (.35 for fair outcome) was much more
inﬂuential than conﬂict (process B = —.10, outcome B = -. l 7) or participation technique
(process B = .08, outcome B = .0). Power equality may be important because it is so
closely related to fairness. Perceptions of unequal power can easily lead to perceptions of
bias in decision making process and outcomes that favor those with power.

Although citizen characteristics did not have large impacts on fairness, the
increases in explained variation for process and outcome fairness were still statistically
signiﬁcant. General political involvement (B = -.16) and education (B = .18) were the
only variables with signiﬁcant coefﬁcients for fair process. Gender (B = -.08) had the
only signiﬁcant coefﬁcient for fair outcome. Age showed a non-signiﬁcant trend towards
explaining fair outcome (B = .06, p = .06). Although there were differences in
signiﬁcance between process and outcome, the directions of the relationships between
citizen variables and both process and outcome fairness were the same. Thus it is likely

that all the citizen variables inﬂuence both outcome and process fairness in similar ways.

Does the level of one context factor aﬂect how other context factors inﬂuence fairness?
The analysis of covariance also tested the ways in which context variables might
statistically interact in their effects on fairness. There was only one signiﬁcant

interaction for fair process which suggested that for people with less formal education,

180

power equality has a larger inﬂuence on fair process. Because less educated people often
are the less powerful members of society, they may be more sensitive to power
differences. There were two signiﬁcant interactions for fair outcome: age with
participation technique, and gender with participation technique. Older people and men
were generally unaffected in their fair outcome evaluations by the amount of discussion
in a participation technique. However, women and younger people gave much higher
fairness evaluations if the technique emphasized discussion. Women may prefer a
discussion approach because it facilitates relationship building. Since commonly used

participation methods like formal hearings and written comments do not allow much

 

discussion or relationship building, this may help explain why women feel less satisﬁed L

with their experiences.

Can the diﬂerences between agencies in terms of fairness be explained by individual and
situation factors, or is there residual agency level variation in fairness that can be
explained by agency level factors?

This question examines the data from a multilevel perspective because it
considers the impact of citizen level and agency level factors simultaneously. In the
analysis of covariance, agency was the factor and situation and citizen variables were the
covariates. The analysis tested whether signiﬁcant differences among agencies in
fairness remained after controlling for the situation and citizens.

For fair process evaluation, controlling for situation and citizen uncovered
signiﬁcant agency differences. A combination of level of government and agency culture

seemed to explain the differences. The planning commissions may have received lower

evaluations than the Huron-Manistee and MDNR because planning commissions make

181

decisions about issues which affect Citizens’ daily lives and thus engender more critical
evaluations if anything goes wrong. However, state and federal agencies make decisions
about distant forests, so citizens are not motivated to be as critical. Within a given level
of government, agency culture correlated with citizen evaluations. HMNF and MDNR
had very similar agency cultures and very similar evaluations. Delhi had lower citizen
evaluations and its planning commissioners reported an agency culture which gave
relatively low fairness importance ratings. Compared to other planning commissions,
Delhi commissioners also believed that citizens had little knowledge and were very self-
interested.

For fair outcome evaluation, after controlling for citizen and situation, differences
among agencies were no longer signiﬁcant. However there was one fairly large, but non-
signiﬁcant, difference. Monitor received a more positive fair outcome evaluation than
did Delhi. This difference is best explained by differences among agency cultures in
which Monitor had the highest fair outcome importance and Delhi had the lowest. In
conclusion, for both outcome and process fairness, residual variation in fairness does

occur at the agency level and it can be explained by patterns in agency factors.

Study limitations

Although the results of this study are strong and quite clear, this study, like all
others, has its limitations. Perhaps the largest limitation is that it was conducted with the
population of people who are already actively involved in natural resource decision
making. It may not apply equally well to those who are not involved. However, the fact

that the importance of fairness was originally noticed in non-participants (Smith &

182

McDonough, 2001) suggests similar patterns may apply to others. Another limitation is
the cross-sectional nature of the data. The model proposed and analyzed makes
suggestions about causality which can really only be conﬁrmed in a longitudinal study.

One of the strengths of the research design was also a limitation. In an attempt to
compare across a variety of cases and develop general theory, the questionnaires were
written as generically as possible. Some respondents complained that questions were too
vague or did not ﬁt their experience. For example, several people involved in the Huron-
Manistee Forest Plan Revision commented that the process was not ﬁnished (it had
already been ongoing many years) so they could not judge the decision outcomes. This
was in contrast to local planning where the decision was sometimes made in one meeting.
It appears that increasing generalizability decreased validity, an often acknowledged

dilemma (McGrath, 1982).

Future research

This study generated a plethora of research needs. One clear need was to deve10p
better measures of many of the constructs. Participation technique was an objective but
very rough measure of the amount of discussion and interaction. New measures which
assess these more accurately need to be devised. For example, the amount of discussion
and interaction could be assessed more directly by asking citizens and facilitators rather
than relying on the researcher assigned values used in this study. One set of items used in
this study actually did measure prior relationships amongcitizens, an aspect affected by
participation technique. However, these items clearly did not work because they were

left blank so often. New ones should be tried and pretested extensively. In hindsight, a

183

much more detailed, longer set of questions would have been necessary to assess the
relationships people have and the nature of those relationships.

The items measuring agency culture of bureaucracy and expertise also suffered
from validity problems so they need to be revised. Participation resources was measured
on a very subjective general scale which may have confused actual resources available
with the respondent’s desired level of resources. These two aspects need to be clearly
separated. A more objective measure of resources, like comparison of agency budgets,
might be more useful. Finally, the items used to calculate trust that asked about support
for the decision did not distinguish between support based on active acceptance of the
decision and support based on passive resignation to a decision that cannot be changed.

Perhaps the most interesting variable suffered from design limitations. An entire
survey was created to measure agency culture, but this was then summarized to average
values for just ﬁve agencies. Analysis of the correlations between agency culture and
fairness uncovered some patterns. However, a larger number of agencies and units
within those agencies would have led to clearer statistical results. In order to minimize
the number of people responding to multiple surveys, the units should as much as
possible not share the same citizens and employees. This will reduce respondent burden.
At least 20 units should be sampled, although that would be an expensive, complex
undertaking.

In addition to the measurement and design needs, the study also raised substantive
questions for further exploration. For example, why does education correlate positively
with fairness? Why were regional governments evaluated more positively than local

ones? What is the role of conﬂict in fairness evaluations? These research questions are

184

similar in that they try to understand and explain patterns found in the quantitative data.
They are often best answered through in-depth qualitative studies.

The study presented here established the role of fairness in increasing citizen trust
in decision makers and identiﬁed a large number of variables which affect fairness. The
next steps are two-fold. First, the quantitative measures should be reﬁned and re-
administered to new and more diverse samples. Secondly, the observed patterns among
variables need to be examined qualitatively to understand the relationship mechanisms

more clearly.

Management implications

So what are the management implications of the answers to these research
questions? How does support for the theoretical framework translate into everyday
decision making? How do managers get higher fairness evaluations from citizens? This
study has several implications for natural resource managers and agencies.

The ﬁrst implication is that citizen evaluations of fairness in decision making
have substantial impacts on trust in the agency and support for decisions. Bivariate
correlations showed that total fair process evaluation and trust had a correlation of 0.78
and fair outcome evaluation and trust were correlated by 0.69. This result conﬁrms an
earlier qualitative study (Smith & McDonough, 2001) which found that citizens naturally
used fairness when reaching judgements about an agency. The current study identiﬁed
many aspects of process fairness, such as representation, neutrality, inﬂuence, and
courtesy, which were highly correlated to trust. Decision makers would be wise to focus

on fairness principles like ensuring adequate representation, conveying their own

185

neutrality, giving citizens inﬂuence over the decision, and treating citizens respectfully.
Many decision makers are probably already trying to achieve these principles. These
results suggest that it is important to communicate to citizens about the efforts being
made to achieve these principles. It is also important to conduct frequent evaluations of
decision making fairness, to see where improvements are needed.

Given that fairness is so important, how do context factors affect fairness and
what are the implications for decision makers? The ﬁrst set of factors was related to the
agency. There was evidence that agencies in which employees believe citizens are
knowledgeable and not self-interested receive higher fairness evaluations. Leadership
within agencies should consider ways to change agency culture. Shannon (1992) found
that leaders in Forest Service forests were able to increase the responsiveness of their
staff to citizens by modeling participatory behaviors in internal decision making. Other
efforts like training existing staff, instituting appropriate incentives, and hiring new
employees all can contribute to cultural change.

Achievement of many of the fairness principles could be greater with training.
For example, staff could learn creative and effective ways to use the media and local
informal networks in spreading the word about decision making, thus increasing
representation. The high percentage of women participating at the planning commission
level may be because of grassroots organizing. One woman, after receiving the survey,
called with a question and indicated that she was working with a neighborhood
organization that had been going door-to-door generating opposition to a proposed
development. If employees could learn how to tap into this kind of local network in a

positive way they might get much higher turnout at public meetings.

186

The effects of situation factors on fairness also have important implications. The
signiﬁcant interaction between gender and participation technique showed women’s
outcome fairness evaluation increases noticeably in more discussion-oriented sessions,
but men’s increases only slightly. A similar interaction showed that younger people also
gave more fair outcome evaluations when discussion was used, although older people did
not show any differences. Examination of the demographics of survey respondents
showed that women and younger people were not well represented in agency decision
making. There was only one woman for each ﬁve male participants, and at the state and
federal level, this reduced to a 1 to 10 ratio. The average age of respondents was 50 years
old. The strong implication is that one way to increase the representativeness of decision
making, a key fairness principle, would be to use more discussion techniques. This
would provide a more conducive environment for younger people and women and might
encourage them to participate.

In addition to participation technique, the amount of power equality had a very
large impact on fairness evaluations for all respondents. The interaction between power
equality and education showed that people with less education were more sensitive to
power differences. This may be because, as the saying goes, knowledge is power. In
natural resource decision making, citizens with expertise may be more successful in
persuasively communicating their preferences to decision makers. Citizens who lack
education, and therefore the power that comes from expertise, are more aware of how
their lack of relative power may inﬂuence the fairness of the process. The connection
between power and expertise was actually part of the measure used to assess power. In

this study, power equality among citizens was measured in terms of equality of ﬁnancial

187

resources, equality of knowledge about how to get what one wants, and general power
equality.

Given the importance of power to fairness judgements, how can decision makers
inﬂuence the distribution of power? While they can not hand out money to make
ﬁnancial resources more equal, they could provide other types of assistance that equalize
power. Decision makers could hold educational meetings for citizen activists to show
them how decisions are made, and so how they can best inﬂuence them. Some decision
makers have deliberately found money to hire scientists that can collect and interpret data
for use by all citizens. This has been particularly effective in disputes over hazardous
pollution between citizens and industry, but has also been used in city planning (Renn,
Webler, & Wiedemann, 1995). It has been demonstrated that when citizens are given
accurate, easy to understand information, they are able to make reasonable science-based
decisions (Doble & Richardson, 1992). In addition, when access to knowledge was
similar among all participants, differences in opinion could be recognized as representing
different values, not just different levels of expertise. Differences in values are harder to
discount and ignore, leading to the possibility of serious consideration of citizen interests
and inﬂuence on decisions. Thus fairness of the process should be increased. In addition
to trying to reduce power inequalities, the connection between power inequality and
fairness could be minimized. Keeping all decision making very transparent and in the
open could reduce concerns that power inequalities affect decision making processes.

Conﬂict was the third aspect of the decision making situation that could be
managed to increase fairness. The results showed that conﬂict had a moderately negative

impact on fairness evaluations. This, however, does not mean that conﬂict is bad. The

188

presence of conﬂict often draws attention to genuine differences of interests and opinions
among people. Conﬂict may be necessary to make sure that the full range of options is
considered before a decision is made. However, once conﬂict has served its role of
ensuring all perspectives are considered, successﬁrl resolution of that conﬂict will
probably reduce the negative impacts on fairness. Resolution would mean that the parties
in opposition voluntarily agree to a settlement that meets the interests, but not necessarily
the positions, of all. If underlying interests are met, then outcome fairness judgements
should be high. A resolution process which meets everyone’s interests would also
necessarily allow people to share their concerns, and have them seriously considered in a
neutral way. Thus a conﬂict resolution process should meet the fair process criteria of
neutrality, inﬂuence, courtesy, and representation.

There are a broad array of conﬂict resolution techniques ranging from very
informal discussions to very formal mock trials (Priscoli, 1990). Decision makers faced
with conﬂict may wish to receive training in some of these techniques or secure the
services of an outside facilitator. When using someone from outside, however, very
careful attention must be paid to how the decision will be implemented. A review of
environmental mediation cases found that mediation led to a large number of settlements,
but, after the outside facilitator left, implementation was attempted in a traditional,
unsuccessful manner (Sipe, 1998).

In conclusion, the theoretical ﬁ'amework conﬁrmed by this study points the way
towards a broad array of management strategies that may improve the relationships
between citizens and decision makers. The strong correlation between fairness and trust

in decision makers suggests managers should use fairness principles as a way to guide

189

and evaluate their decision making. The relationships between context factors and
fairness evaluations suggest a variety of ways to increase fairness. Agency cultures can
be developed which value fairness and which hold citizens in high regard. Participation
techniques which emphasize discussion can be used to encourage women and younger
people to participate, increasing the representativeness and therefore fairness of the
process. Power equalities, particularly those arising in education and expertise
differences, can be directly addressed through the provision of easy-to-understand
technical information for all citizens. Finally, conﬂict can be reduced and fairness
increased through careful use of conﬂict resolution methods. A focus on achieving
fairness through practices like these will help natural resource managers and agencies
conduct participation that leads to greater citizen satisfaction with decisions and perhaps

even greater trust in government.

190

APPENDICES

191

APPENDIX A

CITIZEN QUESTIONNAIRE

192

F

Public Participation in Huron-Manistee
National Forest Decision Making

 

This survey is being sent by the Forestry Department at Michigan State University to all the
people who are on the mailing list for the Huron-Manistee National Forest Plan Revision. We
would like to know about your experiences with USDA Forest Service decision making and
public participation. Even if your knowledge about the Forest Plan Revision is limited. we are
very interested in your opinions.

Your completion of the survey is entirely voluntary and your speciﬁc responses are conﬁdential
and available only to the researchers at Michigan State University working on this project.
Forest Service personnel will not see individual responses. You indicate your willingness to
participate by filling out and returning this survey.

Thank you very much for your time!

Question 1: How do you feel about USDA Forest Service decision making in the Huron-
Manistee National Forest during the past 2 years? Please check a box to show how you feel
about each of the following statements.

Strongly Disagree Neither Agree Agree Strongly Don‘t
Disagree Nor Disagree Agree know
The Forest Service can be trusted to
a’ make good decisions. 0 D D D D D
b My experiences with the Forest Service
have been mainly negative. CI D D D C] D
C The Forest Service does its job well. D C] D D D D
d I trust the Forest Service to make good
decisions without my input. Cl CI D D D D
c i am satisﬁed with Forest Service
decision making. 0 Cl C] D D Q

 

 

Definition of terms used in the remaining questions:
“participant" I both citizens and Forest Service employees.
“citizen ” - people who participated but were not Forest Service employees.

 

 

 

193

Question 2: Over the past 5 years, how many times were you involved in the following
Huron-Manistee Forest Plan Revision (FPR) opportunities?

Circle the # of times involved in the past 5
years (or write a number if more than 10)
Received mailings related to the Forest Plan
Revision (FPR). 0 1 2 3 4 5 6 7 8 9 10 __times

0 Submitted written comment related to the FPR. O l 2 3 4 5 6 7 8 9 10 times

C Talked one-on-one (phone or in-person) with a
Forest Service employee about the FPR. 0 l 2 3 4 5 6 7 8 9 10 times

«F Attended a FPR meeting that primarily

consistedofWhesmadeby 0 1 2 3 4 5 6 7 8 9 10 times
participants (e. g. informational meeting,

public hearing, listening session).

+2 Attended FPR meeting that primarily consisted o l 2 3 4 s 6 7 8 9 10 times

of citizsnaftesluissuasine 188% among
themselves (e. g. Working Group Session).

C HMMgMQJOvI o‘F ﬁtFOFhavsyauW? 0

Question 3: What was your one mm Forest Plan Revision-related participation
experience? Please check only ONE item from the following list.

 

0- l 0 Submitted a written comment related to the Forest Plan Revision (F PR).\ If you checked one

2 D Talked one-on—one (phone or in-person) with a Forest Service employee of these, please ﬁll in
about the FPR. the box below

3 0 Attended a FPR meeting that primarily consisted of mm
who: made by participants (e. g. informational meeting, public
hearing, listening session).

‘l 0 Attended FPR meeting that primarily consisted of W

dim issues among themselves (e. g. Working Group Session). /

 

 

 

5 D None of the above b

--‘

 

When did this experience occur? Please indicate as exact a date as possible.
Year ? Month ? Day ?

What was the location (city, building)?

13

 

 

 

 

 

 

 

 

 

 

 

C What was the topic, subject or purpose of the participation?
Please base your answers to questions 4-10 on written Please base your answers to
material and mailings you received about the Forest Plan questions 4-10 on this specific
Revision. Answer the questions in terms of the overall FPR experience.
process.

 

 

 

 

 

194

L

F~ >§9 "nm

3NPs-Qv"

\

Question 4: Wm Please describe and evaluate the M the Forest

Service used to make decisions related to your most recent participation experience
(identiﬁed in Q3) in the Forest Plan Revision. Please check a box for each statement.

Strongly Disagree Neither Agree Agree Strongly

Disagree Nor Disagree Agree
The procedures used to make decisions
were fair. 0 0 Cl Cl C]
I am satisﬁed with the process used to
reach decisions. Cl C] Cl Cl 0
Citizens were treated unfairly. O C] Cl 0 0
Citizens were given sufﬁcient advance
notiﬁcation of the opportunity to D 0 Cl 0 D
participate.
The participation experience was
convenient to attend. Cl Cl Cl Cl E]
Everyone affected by the decisions had
an opportunity to participate. 0 Cl D 0 Cl
Local people were adequately
involved D D D D D
Citizens were unable to have an
inﬂuence on the W D D D D D
Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
Citizens were able to participate
directly in making decisions. Cl U Cl 0 Cl
Citizens had an inﬂuence on the choice
of W 0 D D 0 0
Citizens' comments were seriously
considered. 0 Cl Cl Cl 0
Citizens' questions were answered. 0 D D CI 0
It appears that information used to
reach the decisions was accurate. Cl 0 D D
The decisions were well reasoned and
logical. Cl Cl Cl Cl Cl
There was a bias toward a particular
interest. group, or person. Cl C] D 0 Cl
USFS employees were dishonest. Cl 0 D C] 0
Citizens were heated politely. Cl Cl C] Cl CI

195

 

Don’t
know

Don’t
know

DD

000 U D

U

Question 4 continued: ’ ion Ma 1 Proce

Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
y- Citizens listened to and understood
each other. Cl Cl 0 Cl C]
5 Citizens successfully worked together
to reach agreement Cl C] Cl Cl Cl
t I do not understand the process that
was used to reach decisions. Cl Cl Cl Cl Cl
Question 5: mm Please describe and evaluate the outcomes of decisions
related to your most recent participation experience in the Forest Plan Revision.
Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
4' Beneﬁts and costs were distributed
fairly among citizens. Cl Cl 0 0 Cl
b The outcomes of decisions were unfair. Cl 0 D D D
6 The decisions reached were equally
favorable to all citizens. Cl 0 Cl 0
4 The decisions beneﬁted the citizens
who were most deserving. 0 Cl 0 D D
e The decisions reached were consistent
with my personal values. C] D C] Cl 0
F The decisions reached did not meet my
personal interests. D Cl Cl 0 D
I do not know what decisions were
/ ultimately reached. Cl Cl C] CI Cl
Question 6: Other topics related to your most recent participation experience.
Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
0' I had very little background knowledge
about the decision topics. CI 0 0 Cl 0
b The decisions involved were of minor
importance to me. D D D CI Cl
O I was too busy to be involved with the
Forest Plan Revision. CI Cl CI Cl Cl

196

 

 

Don‘t
know

Don't
know

0 D D D U

 

nmnb-p

+7

nv-P

Question 7: How has your most recent participation experience affected you and others?

 

 

 

Strongly Disagree Neither Agree Agree Strongly Don’t
Disagree Nor Disagree Agree know
I plan to actively oppose. appeal or sue
a decision reached in this experience. Cl 0 D D 0 Cl
I plan to support the decisions reached. Cl D D Q Q D
This experience made me upset 0 Cl C] 0 Cl Cl
This experience made me happy. Cl 0 CI Cl 0 D
As a result of this experience. the
relationship between Forest Service Cl Cl Cl Cl Cl Cl
employees and myself has improved
As a result of this experience. there is
now more respect between other Cl 0 D D Cl Cl
citizens and myself.
Questions 8. 9. 10 ask about the situation pm: to your most recent participation.
Question 8: Please describe your perceptions of the relative amounts of influence
participants had MEL: your most recent participation experience began.
Strongly Disagree Neither Agree Agree Strongly Don’t
Disagree Nor Disagree Agree know
Power was distributed equally between
the agency and citizens. D D 0 Cl 0 Cl
Power was distributed equally among
citizens. D D D E] El El
Financial resources were equally
distributed among citizens. Cl Cl Cl C] 0 Cl
Access to Forest Service employees
was equally available to all citizens. Cl Cl Cl Cl Cl Cl
The level of knowledge about how to
get what they want was equal among 0 Cl 0 Cl 0 Cl
citizens.
Question 9: For your most recent participation experience, please give the number of
people who were present and the number with whom you had a respectful relationship
m the participation began.
Total it # with whom i had a respectful
present relationship before the participation began
Forest Service staff. a, b
an citizens. a of

Citizens who gm most of my interests. L

197

a:

mnﬁbp

e>~§o «hm

hr

Question 10: Please describe the level of conflict and disagreement between participants

that W of your most recent participation experience.

Participants began the process with
strong, deeply held opinions.

Participants took positions that were
very different from other participants’

Participants expressed strong emotions
in response to disagreements.

Positions held by participants were
highly incompatible with those of
others.

Strongly Disagree Neither Agree Agree
Disagree Nor Disagree
Cl D D Cl
C] Cl Cl Cl
E! El Cl Cl
Cl D D Cl

Strongly
Agree

C]

The remaining questions are not about the Forest Plan Revision, but give us

important background information we need to interpret your answers.

 

Don 't
know

Question 11: Enlarging your experiences with the Forest Plan Revision, during the past 5
years in which of the following ways have you participated in general politics as well as

natural resource decision making?

Organized a group of people around some political issue.

Worked on a local political campaign.
Ran for public ofﬁce.

Regularly attended meetings of a natural resource,

environmental or outdoor recreation related organization.

Signed a natural resource lenvironment/land use petition.

Attended a local zoning/ land use hearing or meeting.

Attended a hearing or meeting of the Michigan Department

of Natural Resources or USDA Forest Service.

Communicated (phone call, letter, or visit) with the Michigan
Department of Natural Resources or USDA Forest Service to

give a suggestion or complaint.

Wrote a letter to an editor of a newspaper about a natural

resource or land use issue.

Organized a group of people around a natural resource or

environmental issue.

Served on a natural resource/land-use commission, advisory

board, or planning team.

198

N00
NOD
NOD

N00
N00
N00

N00

NOD

N00

N00

N00

YesD
YesD
YesCI

YesD
YesCl
YesD

YesCJ

YesCl

Yes D

YesD

YesD

 

Question 12: Please tell us about yourself:

What is your age? (Please check a category)

CI 10-19 0 20-29 C] 30-39 CI 4049 CI 50-59
C] 60-69 C] 70-79 C] 80-89 D 90-99

Are you:
C] Male 0 Female

What is your racial/ethnic background? (Please check one category)

Cl Asian American 0 Native American Cl Other
[:1 African American Cl Latino/Hispanic /
Cl Caucasian Cl Mixed ethnicity ) 2C5

What is the highest level of formal education you have completed? (please check one)

Cl Less than high school 0 Four-year college degree
Cl High school graduate (or equivalency) Cl Graduate or professional degree
Cl Associates or other 2 year degree

Question 13: Friends of the Forest follow-up survey.

We are also comparing the Forest Plan Revision process with the Forest Service Friends of the
Forest meetings for the Huron-Manistee Forest.

Have you attended a USDA Forest Service Friends of the Forest meeting?
No Yes

If yes. would you be willing to ﬁll out a follow-up 3 '/z page survey about the Friends of the
Forest?
No D Yes CI

Question 14: Do you have any further comments about the Forest Plan Revision?

 

 

 

Thank you very much for your time and thoughts!
This survey does not require an envelope and is postage prepaid. Please just fold it in half, staple
or tape it, and mail it.

   

199

APPENDIX B

AGENCY QUESTIONNAIRE

200

Public Participation in the Huron-Manistee
National Forest: A Survey of Employees

 

This survey is being sent by the Forestry Department at Michigan State University to employees
of the Huron—Manistee National Forest. This is an independent study initiated by the researchers
at MS U in order to better understand public participation and provide suggestions to resource
managers.

Your specific responses are conﬁdential and available only to the researchers at Michigan State
University working on this project. USF S personnel will not see individual responses. The
questionnaires are numbered so that we can avoid sending unnecessary reminders to those who
have already responded. Once data collection is completed. the numbered lists will be
destroyed.

Your completion of the survey is entirely voluntary and you indicate your willingness to
participate by ﬁlling out and returning this survey. It should take about 20 minutes to complete.
If you have questions please call Patrick Smith (51 7) 353-5103 or Maureen McDonough (51 7)
432-2293.

Thank you very much for your time and thoughts!

 

Definitions of terms used in this survey:

“Decision” 8 a natural resource management or regulatory decision.

“Citizen” = anyone who is not a state or federal agency employee and
who could potentially participate in USFS decision
making.

 

 

 

Preliminary Part: While working do you ever have contact (even informally) with citizens?

-o

Q D Yes —> Please ﬁll out the survey

Cl No —> You do not need to ﬁll out this survey, please mail it back to us blank.

Myamiﬂiru ? Dddlifﬁpeﬁ'htkw.

0) myw currently an- «I. conduit-sine .

201

We

\D'prD—n

Part 1: Over the past 12 months, how many times have you done the following public
participation activities?

Please circle the number of times done in
the past 12 months
Responded to a citizen‘s question or comment about a
decision by talking one-on-one (phone or in-person). 0 1-3 4-6 7-9 10-12 13-15 16-18 20+

Initiated a phone call or in-person visit to a citizen in

order to gather input aboutadecision. 1-3 4-6 7-9 10-12 13-15 16-18 20+

Readacitizen’s written comment to the USFS. 1-3 4-6 7-9 10-12 13-15 16-18 20+

0
0

AttendedaUSFSOpenhouse. 0 1 2 3 4 5 6 7 8 9 10 11+
0

AttendedaUSFS public meeting. 1 2 3 4 5 6 7 8 9 10 11+

Attended a meeting of a citizen’s organization as a
USFS representative. 0 1 2 3 4 5 6 7 8 9 10 11+

Incorporated a citizen suggestion that was new to you
intoaplan,document,orreport. 0 1 2 3 4 5 6 7 8 9 10 11+

Part 2: What are the characteristics of citizens? 0n the LEFT SIDE. please circle the number
that best represents your personal beliefs about each statement. On the RIGHT SIDE. please
circle a number to indicate your perceptions of the beliefs of most other USF S employees in your

unit.
Deﬁnitely More False More True Definitely

False Than True Than False True
-2 -1 l 2
My own Beliefs of most
beliefs other employees
in my unit

Citizens who participate in USFS decision making are...
-2 -1 1 2 usually able to understand and use technical information. -2 -1 1 2 4.
-2 -l 1' 2 usually in possession of the knowledge needed to make good -2 -l 1 2 b

decisions.

-2 -l 1 2 mainly concerned about their self-interests. -2 -1 1 2 C
-2 -1 1 2 usually focused on the short-term. -2 -1 l 2 ‘1

Citizens fail to participate because...
-2 -1 1 2 .. they don’tcare much about the important decisions we make. -2 -1 1 2 e
-2 -l l 2 .. they feel they are too busy to participate. -2 -l 1 2 F
-2 -1 1 2 .. they are satisﬁed with the decisions. -2 -1 l 2 j
-2 -1 1 2 .. theydo not trust the USFS. -2 -1 l 2 It

202

O

Q.

Part 3: On the left side please indicate how much you personally agree or disagree with the
following statements. On the right, rate how much you think most other L'SF S employees in
your unit agree or disagree with the statements. The statements measure detailed differences. so
please rate them all even if they seem Similar.

Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
SD D N A SA
My own Beliefs of most
beliefs other employees
in my unit
SD D N A SA Decisions should be made according to standard 0— SD D N A SA
professional practices.

SD D N A SA Experts should have the power to make decisions. A SD D N A SA
SD D N A SA When making decisions, correctness is more important C SD D N A SA
than popularity.

SD D N A SA Decisions made byaperson inaposition of authority J SD D N A SA
should not be challenged by citizens.

SD D N A SA USFS employees understand the long term consequences of 3 SD D N A SA

decisions better than citizens.

SD D N A SA USFS employees haveaclear idea of public needs and desires. ‘F SD D N A SA

SD D N A SA Consistent application of rules is more importantthan SD D N A SA
incorporating public comments.
SD D N A SA Public participation usually helps to build public [1 SD D N A SA
support for the agency.
SD D N A SA Well-established plans should notbechanged inresponse L SD D N A SA
to citizen demands. ‘
SD D N A SA The beneﬁts of involving citizens outweigh the costs. J SD D N A SA

SD D N A SA Involving citizens slows down decision making processes too much. k SD D N A SA

 

Part 4: How many times have you ever done the following? Please circle the number for each.

Attended a Forest Plan Revision working group

 

session. 0 1-3 4-6 7-9 10-12 13-15 16-18 20+
Attended a public Forest Plan Revision listening

session or informational meeting. 0 1-3 4-6 7-9 10-12 13-15 16-18 20+
Participated in internal discussions about citizen input

to theForestPlanRevisionprocess. 0 1-3 4-6 7-9 10-12 13-15 1618 20+
AttendedaUSFS FriendsoftheForest meeting. 0 1-3 4—6 7-9 10-12 13-15 16.18 20+

 

 

203

Part 5: How important are each of the following statements to you? Please check a box for
each statement.

Not at all Somewhat Important Essential
important Important

 

a. The procedures used to make decisions are fair. 0 D D D
P Citizens are treated fairly. Cl Cl 0 Cl
C Citizens are satisﬁed with the process used to reach Cl C] U C]
decisions.
d Citizens are given sufﬁcient advance notiﬁcation of the Cl C] 0 Cl
opportum'ty to participate.
e The participation experience is convenient to attend. D D D O
4? Everyone affected by the decisions has an opportunity to Cl 0 Cl Cl
participate.
; Local people are adequately involved. 0 Cl D Q
A) Citizens are able to participate directly in making decisions. Cl Cl 0 D
1 Citizens are able to have an inﬂuence on the 51:21:19!) 0 D D Cl
931m
3 Citizens have an inﬂuence on the choice of W Cl 0 O C]
amass.
k, Citizens' comments are seriously considered in the decision Cl C1 C1 C1
making.
1, Citizens' qgestions are answered. 0 Cl 0 Cl

 

Not stall Somewhat Important Essential
Important Important

 

M. Information used to reach the decisions is accurate. 0 D D U
The decisions are well reasoned and logical. Cl C] D C]
9 There is a lack of bias toward particular interests, groups, Cl Cl C] Cl
and persons.
f Agency employees are honest to citizens. Cl Cl CI Cl
89 Citizens are treated politely. U D D D
r Citizens listen to and understand each other. CI Cl 0 Q
3 Citizens successfully work together to reach agreement. 0 CI Cl Cl
f Beneﬁts and costs of outcomes are distributed fairly among 0 Cl Cl 0
Citizens.
Li. The outcomes of decisions are fair. D D D 0
if The decisions reached are equally favorable to all citizens. Cl 0 Cl C]
1.) The decisions beneﬁt the citizens who are most deserving. Cl 0 D D
K The decisions reached are consistent with citizens’ personal 0 Cl Cl D
values.
y The decisions reached meet citizens’ personal interests. 0 Cl 0 D

204

Part 6: In the section above you indicated the importance to yourself of some statements.
Now please indicate how often you and other L'SF S employees in your unit actually
accomplish them. Please Circle 0 word on the right for each statement.
Frequency of accomplishment in my unit
The procedures used to make decrsions are fair. Never Seldom Sometimes Often Always

Citizens aretrcarcd fairly. Never Seldom Sometimes Often Always

Citizens are satisﬁed with the process used to reach
decisions Never Seldom Sometimes Often Always

Citizens are given sufﬁcient advance notiﬁcation of the

opportunity to participate. Never Seldom Sometimes Often Always

The participation experience is convenient to attend. Never Seldom Sometimes Often Always

Mix Q- “WP

Everyone affected by the decisions has an opportunity to
pal-nap”; Never Seldom Sometimes Often Always
Local people are adequately involved, Never Seldom Sometimes Often Always

Citizens are able to have an inﬂuence on the 922151211 V . 0f
9112911195.- . ever Seldom Somenmes ten Always

Citizens are able to participate directly in making decisions. Never Seldom Sometimes Often Always
Citizens have an inﬂuence on the choice of W

 

03§N n~Lvi~n>So

 

am Never Seldom Sometimes Often Always

C itizens' comments are seriously considered in the decision

making. Never Seldom Sometimes Often Always

Citizens' questions are answered. Never Seldom Sometimes Often Always

Information used to reach the decisions is accurate. Never Seldom Sometimes Ollcn Always

The decisions are well reasoned and logical. Never Seldom Sometimes Often Always

There is a lack of bias toward particular interests, groups.

and persons. Never Seldom Sometimes Often Always
I Agency employees are honest to publics. Never Seldom Sometimes Often Always
2 Citizens are treated politely. Never Seldom Sometimes Often Always
r’ Citizens listen to and understand each other. Never Seldom Sometimes Often Always
5' Citizens successﬁrlly work logger to reach agreement. Never Seldom Sometimes 0M Always
rt Beneﬁts and costs of outcomes are distributed fairly among

citizens. Never Seldom Sometimes Often Always
[4, The outcomes of decisions are fair. Never Seldom Sometime! 0M Al"?!
V The decisions reached are equally favorable to all citizens. Never Seldom Sometime! 03m Always
it) The decisions beneﬁt the citizens who are most deserving. Never Seldom Sometimes 0M Always
K The decisions reached are consistent with Citizens’ personal

values. Never Seldom Sometimes Often Always
} The decisions reached meet citizens’ personal interests. Never Seldom Sometimes Often AlWIYS

205

 

= Part 7: How big an impact does each of the following factors have on the seriousness with which

citizen input is considered in your unit? Please circle a word/or each factor.

f

I
l

 

Extent of impact

—-.---- ..

 

 

 

 

 

 

206

Q Rules and policies. Zero Small Medium Large Dominating
b Leadership of superiors. Zero Small Medium Large Dominating
c Personal convictions of the decision maker. Zero Small Medium Large Dominating
A Resources (staff, money), Zero Small Medium Large Dominating
(, Pressure/expectations of equal-level colleagues. Zero Small Medium Large Dominating
Part 8: How much of the following resources are available to your unit for public
participation? Please circle an amount for each resource.
Amount available
a. Staff Almost None A Little Some A lot
b Training Almost None A Little Some A lot
0 Money Almost None A Little Some A lot
Part 9: In your unit how frequently is there... i
a, discussion among employees about how to ’
improve citizen involvement in decision making? NW“ Seldom Sometimes 03¢" Always
discussion among employees about citizen
comments in an attempt to identify the Never Seldom Sometimes Often Always
underlying concerns?
C a climate that makes subordinates feel it is 01: to
openly disagree with superiors? Never Seldom Sometimes Often Always
A debate between subordinates and superiors about
technical issues? Never Seldom Sometimes Often Always
Part 10: Please tell us about yourself:
What is your age? (Please check a category)
6L 0 10-19 D 20-29 D 30-39 Cl 40.49 CI 50-59 CI 60-69 Cl 70+
Are you:
b D Male Cl Female

Tb

o~p~§n *3

Part 10 continued:

What is your racial/ethnic background? (Please check one category)

Cl Asian American :1 Native American Cl Other (please specify)
Cl African American Cl LatinoxT-lispanic >C_i.
D Caucasran D Mixed ethnicity (please specify)

What is the highest level of formal education you have completed? (please check one)

0 Less than high school C] Four-year college degree
CI High school graduate (or equivalency) Cl Graduate or professional degree
0 Associates or other 2 year degree

If you received professional training or a degree.
what was the subject(s) or ﬁeld(s)?

 

Do you belong to the Society of American Foresters? D Yes C] No
Do you belong to any other professional societies? (please list)
All told. how long have you worked for the USFS? years.

How long have you worked on the Huron-Manistee National Forest? years.

Part 11: Do you have any further comments about public participation in the USFS or
about this survey?

 

 

 

 

 

Thank you very much for your time and thoughts!

This survey does not require an envelope. and postage is prepaid. Please just fold it in half,
staple or tape it. and mail it.

 

207

APPENDIX C

SHORT FORM (PHONE SURVEY) QUESTIONNAIRES

208

 

USFS Citizen Nonresponse phone interview: Name Q1 # _ Date
Phone #
Reason you did not return survey:

Would you be willing to send in completed Questionnaire? Yes No

 

Should I send new Questionnaire? Yes No

Mention conﬁdentiality.
Question I: How do you feel about USDA Forest Service decision making in the
Huron-Manistee National Forest during the past 2 years?

 

 

Strongly Disagree Neither Agree Agree Strongly Don't
Disagree Nor Disagree Agree know

The Forest Service can be trusted to

make good decisions. 0 CI Cl D Cl D
My experiences with the Forest Service

have been mainly negative. 0 Cl D Cl C] D
The Forest Service does its job well. D Cl D D D Cl
I trust the Forest Service to make good

decisions without my input. 0 Cl D CI 0 D
I am satisﬁed with Forest Service

decision making. D D U D D D
Question 2: Over the past 5 years, how many times were you involved in the
following Huron-Maulstee Forest Plan Revision (FPR) opportunities?
Received mailings related to the Forest Plan Revision (FPR). ___times
Submitted written comment related to the FPR. ___times
Talked one-on-one (phone or in-person) with a Forest Service employee about the

F PR. _times
Attended a FPR meeting that primarily consisted of W made

by participants (e.g. informational meeting, public hearing, listening session). _times
Attended FPR meeting that primarily consisted of WW issues _times

among themselves (e.g. Working Group Session).

209

 

 

If person has participated ask about experience:

 

 

Strongly Disagree Neither Agree Agree Strongly Don’t
Disagree Nor Disagree Agree know

The procedures used to make decisions

were fair. Cl C] D Cl Cl 0
The outcomes of decisions were unfair. Cl C] Cl 0 Cl 0
The decisions reached were consistent

with my personal values. Cl Cl Cl Cl 0 Cl
I plan to actively oppose, appeal or sue

a decision reached in this experience. 0 D D 0 Cl 0
Power was distributed equally among

citizens. D Cl D Cl Cl Cl
Participants began the process with

strong, deeply held opinions. Cl Cl C] Cl Cl Cl
Question 11: mm: your experiences with the Forest Plan Revision, during the
past 5 years in which of the following ways have you participated in general polities
as well as natural resource decision making?
Organized a group of people around some political issue. No Cl Yes Cl
Worked on a local political campaign. No D Yes Cl
Ran for public ofﬁce. No D Yes Cl
Regularly attended meetings of a natural resource, environmental or outdoor
recreation related organization. No D Yes Cl
Signed a natural resource lenvironment/land use petition. No Cl Yes 0
Attended a local zoning! land use hearing or meeting. No D Yes 0
Attended a hearing or meeting of the Michigan Department of Natural
Resources or USDA Forest Service. No D Yes Cl
Communicated (phone call, letter, or visit) with the Michigan Department of
Natural Resources or USDA Forest Service to give a suggestion or complaint. No D Yes D
Wrote a letter to an editor of a newspaper about a natural resource or land use
issue. No D Yes D
Organized a group of people around a natural resource or environmental issue.

No D Yes Cl

Served on a natural resource/land-use commission, advisory board, or planning
team. No D Yes D

What is your age? (Please check a category)

CI 10.19 C) 20-29 C] 30-39 D 40.49 C] 50-59
CI 60-69 D 70.79 CI 80—89 D 90.99
Are you: 0 Male Cl Female

210

What is your racial/ethnic background?

 

 

Cl Asian American Cl Native American Cl Other

Cl African American Cl Latino/Hispanic

D Caucasian Cl Mixed ethnicity

What is the highest level of formal education you have completed?

0 Less than high school Cl Four-year college degree

Cl High school graduate (or equivalency) Cl Graduate or professional degree

Cl Associates or other 2 year degree

Other comments:

211

 

USFS Agency Nonresponse phone interview: Name Q1 # Date
Reason you did not return survey:

old

Would you be willing to send in completed Questionnaire? Yes No
Should I send new Questionnaire? Yes No

Part I: Over the past 12 months, how many times have you done the following
public participation activities?
Please circle the number of times done in
the past 12 months
Responded to a citizen’s question or comment about a
decision by talking one-on-one (phone or in-person). 0 1-3 4-6 7-9 10-12 13-15 16-18 20+
Initiated a phone call or in‘person visit to a citizen in

 

 

 

orderto gatherinputaboutadecision. 0 1-3 4-6 7-9 10-12 13-15 16-18 20+
Readacitizen‘s written comment to the USFS. 0 1-3 4-6 7-9 10-12 13-15 16-18 20+
AttendedaUSFSopenhouse. O l 2 3 4 5 6 7 8 9 10 11+
AttendedaUSFSpublicmeeting. O 1 2 3 4 5 6 7 8 9 10 11+
Attended a meeting of a citizen’s organization as a

USFSrepresentative. O 1 2 3 4 S 6 7 8 9 10 11+
Incorporated a citizen suggestion that was new to you

intoaplan,document,orreport. O l 2 3 4 5 6 7 8 9 10 11+
Part 4: How many times have you ever done the following? Please circle the number for each.
Attended a Forest Plan Revision working group

session. 0 1-3 4.6 7-9 10-12 13-15 16-18 20+
Attended a public Forest Plan Revision listening

session or informational meeting. 0 1-3 4-6 7-9 10-12 13-15 16-l8 20+
Participated in internal discussions about citizen input

to the Forest Plan Revision process. 0 1-3 4-6 7-9 10-12 13-15 16.18 20+
AttendedaUSFS FriendsoftheForestmeetinE 0 1-3 4-6 7-9 1012 13-15 16-18 20+

 

 

212

Part 3: How much do you personally agree or disagree with the following statements.

Strongly Disagree Neither Agree Agree Strongly
Disagree Nor Disagree Agree
SD D N A SA
SD D N A SA Decisions should be made according to standard
professional practices.
SD D N A SA Experts should have the power to make decisions.
SD D N A SA When making decisions. correctness is more important
than popularity.
SD D N A SA Decisions made by a person in a position of authority
should not be challenged by citizens.
SD D N A SA USFS employees understand the long term consequences of

decisions better than citizens.
SD D N A SA USFS employees have a clear idea of public needs and desires.

SD D N A SA Consistent application of rules is more important than
incorporating public comments.
SD D N A SA Public participation usually helps to build public
support for the agency.
SD D N A SA Well-established plans should not be changed in response
to citizen demands.
SD D N A SA The beneﬁts of involving citizens outweigh the costs.

SD D N A SA Involving citizens slows down decision making processes too much.

What is your age? (Please check a category)
0 10-19 D 20.29 CI 30-39 D 40.49 D 50.59 CI 60-69 D 70+
Are you: 0 Male 0 Female

What is your racial/ethnic background? (Please check one category)

 

Cl Asian American Cl Native American Cl Other (please specify)

CI African American 0 Latino/Hispanic

Cl Caucasian Cl Mixed ethnicity (please specify)

What is the highest level of formal education you have completed? (please check one)
D Less than high school 0 Four-year college degree

Cl High school graduate (or equivalency) 0 Graduate or professional degree

D Associates or other 2 year degree
If you received professional training or a degree, what was the subject(s) or ﬁeld(s)?

 

Do you belong to the Society of American Foresters? D Yes Cl No
Do you belong to any other professional societies? (please list)
All told, how long have you worked for the USF S? years.

How long have you worked on the Huron-Manistee National Forest? years.

213

REFERENCES

Agresti, A. & B. Finlay. 1997. Statistical methods for the social sciences, third edition.
Upper Saddle River, NJ: Prentice Hall.

Alwin, D.F. & RM. Hauser. 1981. The decomposition of effects in path analysis. In
Linear models in social research, ed. P.V. Marsden, pp. 123-140. Thousand Oaks,
CA: SAGE.

Amstein, SR. 1977. A ladder of citizen participation. In Citizen participation
certiﬁcation for community development, ed. P. Marshall, pp. 40-49. National
Association of Housing and Redevelopment Ofﬁcials, Washington DC.

Babbie, E. 1998. The practice of social research, eighth edition. Belmont, CA:
Wadsworth Publishing Co.

Barrett-Howard, E. & T.R. Tyler. 1986. Procedural justice as a criterion in allocation
decisions. Journal of Personality and Social Psychology 50(2): 296-304.

Benest, F. 1996. Serving customers or engaging citizens, what is the future of local
government? Public Management 78(2): A6-A10.

Blahna, D.J. & S. Yonts-Shephard, 1989. Preservation or use? Confronting public issues
in forest planning and decision making. In: Hutcheson, J ., F. Noe & R.Snow (eds).
Outdoor recreation policy: pleasure and preservation., p 161-176. Westwood, CT:
Greenwood Press.

Blalock, HM. 1979. Social statistics, revised second edition. New York: McGraw-Hill.

Box, RC. 1998. Citizen Governance: Leading American Communities into the 21”
Century. Thousand Oaks, CA: SAGE

Brockner, J. & P. Siegel. 1996. Understanding the interaction between procedural and
distributive justice: The role of trust. In Trust in organizations: Frontiers of theory
and research. eds. RM. Kramer & T.R. Tyler, pp. 390-413. Thousand Oaks, CA:
SAGE.

Camevale, P.J. & D.G. Pruitt. 1992. Negotiation and mediation. Annual Review of
Psychology 43: 531-582.

Coleman, J .S. 1988. Social capital in the creation of human capital. American Journal of
Sociology 94 Supplement 895-8120.

214

COS, Committee of Scientists. 1999. Sustaining the people '5 lands: Recommendations

for stewardship of the national forests and grasslands into the next century.
Washington, DC: USDA.

Culhane, P.J. 1981. Public lands politics: Interest group inﬂuence on the Forest Service
and the Bureau of Land Management. Baltimore: Johns Hopkins University Press.

Doble, J. & A. Richardson. 1992. You don’t have to be a rocket scientist... Technology
Review Jan: 51-54.

Ewick, P, & S. Silbey. 1998. The Common Place of Law. Chicago: University of
Chicago Press.

Festinger, L. 1957. A theory of cognitive dissonance. Evanston, 111.: Row Peterson.

Finkel, SE. 1985. Reciprocal effects of participation and political efﬁcacy: A pane
analysis. American Journal of Political Science 29(4): 891-913.

Floyd, D.W., R.H. Germain & K.T Horst. 1996. A model for assessing negotiations and
mediation in forest resource conﬂicts. Journal of Forestry 94(5): 29-33.

F olger, R. 1993. Reactions to mistreatment at work. In Social Psychology in
Organizations. ed. J .K. Murnighan, pp. 161-183. Englewood Cliffs, NJ: Prentice
Hall.

Fortmann, L. and SK. Fairfax. 1991. Forest resource policy. In Rural policies for the
1990 ’s. eds. C.B. Flora & J .A. Christenson, pp. 270-280. Boulder, CO: Westview
Press.

Frome, M. 1984. The Forest Service, 2"" Ed. Boulder, co: Westview Press.

George, D. and P. Mallery. 1995. SPSS/PC+ Step by Step: A Simple Reference Guide.
Belmont, CA: Wadsworth Publishing Co.

Hegtvedt, K.A. 1992. When is a distribution rule just? Rationality and Society 4(3):
308-33 1 .

Hegtvedt, K.A. & B. Markovsky. 1995. Justice and injustice. In Sociological perspectives
on social psychologa eds. K.S. Cook, G.A. Fine, & J .S. House, pp. 396-420. Boston:
Allyn and Bacon.

Hofstede, G. 1998. Attitudes, values and organizational culture: Disentangling the
concepts. Organization Studies 19(3): 477-492.

Homans, GO 1961. Social behavior: Its elementary forms. New York: Harcourt, Brace
& World, Inc. pp. 404.

215

Jones, E.S. & P. Mohai, 1995. Is the Forest Service keeping up with the times?: Interest
group and forestry school perceptions of post-NFMA change in the United States
Forest Service. Policy Studies Journal 23(2): 351-371.

Kaufman, H. 1960. The Forest Ranger: A study in administrative behavior. Baltimore:
Johns Hopkins Press.

Kemmis, D. 1990. Community and the politics of place. Norman: University of
Oklahoma Press.

Knott, J .H. & G.J. Miller. 1987. Reforming bureaucracy: The politics of institutional
choice. Englewood Cliffs, NJ: Prentice-Hall.

Kweit, M.G. & R.W. Kweit. 1981. Implementing citizen participation in a bureaucratic
society. New York: Praeger.

Laird, F.N. 1993. Participatory analysis, democracy, and technological decision making.
Science, Technology, & Human Values 18(3): 341-361.

Lauber, T.B. & B.A. Knuth. 1997. F aimess in moose management decision-making: the
Citizens’ perspective. Wildlife Society Bulletin 25(4): 776-787.

Lauber, T.B. & B.A. Knuth. 1998. Measuring fairness in citizen participation: A case
study of moose management. Society and Natural Resources 12(1): 19-37.

Lawrence, R.L., S.T. Daniels, & G.H. Stankey. 1997. Procedural justice and public
involvement in natural resource decision making. Society & Natural Resources 10:
577-589.

Leventhal, GS. 1976. F aimess in social relationships. In Contemporary topics in social
psychology. eds. J .W. Thibaut, J .T. Spence, & R.C. Carson, pp. 211-239. Morristown,
NJ: General Learning Press.

Leventhal, G.S., J. Karuza Jr. & W.R. Fry. 1980. Beyond fairness: a theory of allocation
preferences. In Justice and social interaction. ed. G. Mikula, pp. 167-218. New
York: Springer-Verlag.

Lind, E.A., C.T. Kulik, M. Ambrose & M.V. de Vera Park. 1993. Individual and
corporate dispute resolution: Using procedural fairness as a decision heuristic.
Administrative Science Quarterly 38: 224-251.

Lind, E.A. & T.R. Tyler. 1988. The social psychology of procedural justice. New York:
Plenum Press.

216

Lind, E.A., T.R. Tyler, & Y.J. Huo. 1997. Procedural context and culture: Variation in
the antecedents of procedural justice judgments. Journal of Personality and Social
Psychology 73(4): 767-780.

McClendon, M.J. 1994. Multiple regression and causal analysis. Itasca, IL: F.E.
Peacock Publishers.

McDonough, M.H. 1991. Integrating the diversity of public values into ecosystem
management. Pp 133-137 In: Lemaster, D.C. & G.R. Parker (eds). Ecosystem
management in a dynamic society. West Lafayette, IN: Purdue University.

McDonough, M.H., KE. Vachta, S.L. Funkhouser, & A.L. Gieche. 1994. Creating
community-forestry partnerships: a participatory approach. USDA Forest Service
Urban Forestry Center for Midwestern States.

McDonough, M.H. & M. Thorbum (1997) An evaluation of, and recommendations for,
public participation and citizen advisory mechanisms in the Michigan Department of
Natural Resources Forest Management Division. MSU Foresu'y Department
Working Paper.

McGrath, J .E. 1982. Dilemmatics: the study of research choices and dilemmas. In
McGrath, J .E., J. Martin, & R.A. Kulka (eds). Judgement call in research. Beverly
Hills: SAGE.

Molotch, H.L. 1976. The city as a grth machine: Toward a political economy of place.
American Journal of Sociology 82: 309-78.

Netter, J ., W. Wasserrnan & G.A. Whitmore. 1993. Applied statistics. Boston: Allyn and
Bacon.

Parker, J .D. & M.H. McDonough, 1999. Environmentalism of African Americans: An
analysis of the subculture and barriers theories. Environment and Behavior 31(2):
155-17 7.

Pateman, C. 1970. Participation and democratic theory. Cambridge: Cambridge
University Press.

Pollak, PB. 1984. Planning decisions: Does citizen participation make a difference? A
case study. Planning & Administration 11(2): 63-77.

Priscolli, J .D. 1990. Public Involvement, conﬂict management, and dispute resolution in
water resources and environmental decision making. Ft. Belvoir, VA: Institute for
Water Resources, US. Army Corp of Engineers. IWR Working Paper 90-ADR-WP-
2.

217

 

Pruitt, D.G. 1998. Social Conﬂict. In Eds. Gilbert, D.T., S.T. Fiske & G. Lindzey. The
Handbook of Social Psychology, 4'” Ed, Vol 2, pp 470-503.

Renn, O., T. Webler, & P. Wiedemann. 1995. Fairness and competence in citizen
participation: Evaluating models for environmental discourse. Boston: Kluwer.

Robinson, 60. 1975. The Forest Service: A study in public land management.
Baltimore: Johns Hopkins Press.

Russell, J .W., C. Blye, J. Caplan, H.A. Deutsch, O.D. Grossarth, M. Lunn, E. Schultz, R.
Scott & T. Stewart. 1990. Public participation: Volume 5 in critique of land
management planning. Washington, DC: USDA Forest Service.

Rylander H, R.G., D.B. Propst, & T.R. McMurtry. 1995. Nonresponse and recall biases in
a survey of traveler spending. Journal of Travel Research (spring): 39-45.

Salwasser, H. 1994. Ecosystem management: can it sustain diversity and productivity?
Journal of Forestry 92(8): 6-1 1.

Schumaker, J .R., J. O’Laughlin, & J .C. F reemuth. 1997. Why don’t federal employees
use alternative dispute resolution more often? Journal of Forestry 95(1): 20-22.

Shannon, M. 1987. Forest planning: Learning with people. In Social science in natural
resource management systems. eds. M.L. Miller, R.P. Gale & P.J. Brown, pp. 233-
252. Boulder, CO: Westview Press.

Shannon, MA. 1992. Foresters as strategic thinkers, facilitators, and citizens. Journal
of Forestry Oct:24-27.

Sipe, N.G. 1998. An empirical analysis of environmental mediation. Journal of the
American Planning Association 64(3): 275-285.

Sirmon, J ., W.E. Shands & C. Ligget. 1993. Communities of interests and open
decisionmaking. Journal of Forestry 91(7): 17-21.

Slover, BL. 1996. A music of opinions: Collaborative planning for the Charles C. Deam
Wilderness. Journal of Forestry 94(5): 18-23.

Smith, P.D., M.H. McDonough, & M. Mang. 1999. Ecosystem management and public
participation: lessons from the ﬁeld. Journal of Forestry (Oct): 32-38.

Smith, P.D. & M.H. McDonough. 2001. Beyond public participation: fairness in natural
resource decision making. Society and Natural Resources.

218

Smith, P.D. & D. Propst. 2001. Are topic speciﬁc measures of socio-political control
justiﬁed? Exploring the realm of citizen participation in natural resource decision
making. Journal of Community Psychology.

Steele, EH. 1987. Participation and rules: The functions of zoning. American Bar
Foundation Research Journal 1986(4): 709-755.

Syme, G.J. & E. Eaton. 1989. Public involvement as a negotiation process. Journal of
Social Issues 45(1): 87-107.

Taylor, S. 1984. Making bureaucracies think: The environmental impact statement
strategy of administrative reform. Stanford: Stanford University Press.

Thibaut, J. & L. Walker. 1975. Procedural Justice: A psychological analysis. New York:
John Wiley & Sons.

Tuler, S. & T. Webler. 1999. Voices from the forest: What participants expect of a public
participation process. Society and Natural Resources 12(5): 437-453.

Tyler, T. 1990. Why people obey the law. New Haven: Yale Univ. Press. pp. 273.

Tyler, T. 1994. Psychological models of the justice motive: Antecedents of distributive
and procedural justice. Journal of Personality and Social Psychology 67(5): 850-
863.

Tyler, T. & P. Degoey. 1995. Collective restraint in social dilemmas: Procedural justice
and social identiﬁcation effects on support for authorities. Journal of Personality and
Social Psychology 69(3): 482-497.

United States Forest Service. 1977. Forest Service inform and involve handbook. USDA
Forest Service.

Van den Bos, K., E.A. Lind, R. Vermunt & A.M. Wilke. 1997. How do I judge my

outcome when I do not know the outcome of others? The psychology of the fair
process effect. Journal of Personality and Social Psychology 72(5): 1034-1046.

Walster, E., E. Berscheid & G.W. Walster. 1973. New directions in equity research.
Journal of Personality and Social Psychology 25(2): 151-176.

Webler, T. & S. Tuler. forthcoming. Fairness and competence in citizen participation:
Theoretical reﬂections ﬁ'om a case study. Administration & Society.

Wondolleck, J .M. 1988. Public lands conﬂict and resolution: Managing national forest
disputes. New York: Plenum Press.

219