w

5&1 m.
am?

 

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

Recreational Boaters' Awareness and Responses to Low
Lake Michigan Water Levels.

presented by
Tzu-Ching Chang

has been accepted towards fulﬁllment
of the requirements for

Ph . D . PRTR

degree in

 

 

552W

 

Major professor

Date ﬂO// 4723/, 0n3
// /

MSU is an Afﬁrmuu'vr Action/Equal Opportunity Institution 0- 12771

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

 

*- OQM 32am

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c1ClRC/DateDuep65—p. 1 5

 

RECREATIONAL BOATERS’ AWARENESS AND RESPONSES TO

LOW LAKE MICHIGAN WATER LEVELS
By

Tzu-Ching Chang

A DISSERTATION

Submitted to
Michigan State University
In partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Park, Recreation and Tourism Resources

2003

ABSTRACT
RECREATIONAL BOATERS’ AWARENESS AND RESPONSES TO

LOW LAKE MICHIGAN WATER LEVELS
By

Tzu-Ching Chang

Boating is an extremely important recreation activity and tourism industry in the
Great Lakes. In 2001, over 2-million recreation watercraft were registered in Michigan,
Wisconsin, Indiana and Illinois. Great Lakes water levels regularly ﬂuctuate and periods
of extreme low and high water reduce boating quality and accessibility and negatively
impact the boating industry by increasing cost and reducing revenues. While several
recent studies have focused on the impacts of low Great Lakes water levels on boating,
they did not improve understanding of whether and how boaters incorporate low water as
part of their decision making on whether to boat, how much to boat, and where to boat.
A better understanding of how low water inﬂuences boaters and boating will aid boating
agencies and businesses reach decisions on long-term (e. g., locating and re-locating
facilities) and short-tenn (e. g., dredging, boater education) strategies to mitigate the
negative impacts of water level ﬂuctuations, which may actually become even more
extreme because of climatic changes.

The purposes of this study are to develop and evaluate five LOGIT models
designed to identify various factors, which inﬂuence recreation boaters’ awareness and
responses to low Lake Michigan water levels. The ﬁve LOGIT models focus on factors

that inﬂuence (1) boater awareness of Lake Michigan water levels, (2) their perceptions

of the extent that water levels had dropped, (3) if concerns relating to low water actually
inﬂuence boaters and more speciﬁcally, (4) whether boaters reduce their amount of
boating in response to low water, and (5) if boaters change the locations where they boat
because of low water.

The results of the LOGIT models indicate that boater’s awareness of water levels
is inﬂuenced by their socioeconomic characteristics, and also as would be expected on
whether and how much they operate their boats on Lake Michigan. Boater’s perceptions
about the degree to which water levels have dropped are related to the type of boats they
own and where they store these boats during the boating season. There is a relationship
between the probability that boaters are inﬂuenced by the low water and where they store
their boats and whether they acquired information about water levels. Boat
characteristics and the locations they operate their boats are related to the likelihood they
will change their boating locations in response to low water levels. The probability that
boaters will change boating locations varies depending on the type of boats they own and
the locations where they operate these boats. The probability that a boater will respond to
low water by reducing their amount of boating is related to the number of different
locations and the number of days they boat.

The results indicate that the LOGIT models can be used to calculate the
probabilities of recreational boaters’ awareness, assessments, and different coping
responses to low water levels given boat types and boater characteristics. This study
demonstrates some of the potential beneﬁts of modeling boater awareness and responses
to environmental factors, and the need to expand modeling to include other types of

environmental factors including water quality, access, aesthetics and crowding.

Copyright by
Tzu-Ching Chang
2003

ACKNOWLEDGMENTS

I wish to express my appreciation to Dr. Edward M. Mahoney, my academic
advisor and dissertation chairman for providing extensive guidance and encouragement
during all stages of my doctor study. Thanks also go to Dr. Daniel J. Stynes for his many
helpful ideas, suggestions, and criticisms during this study and through my doctoral
program.

I am also wish to thank Dr. Sandra Batie and Dr. Joseph J. Fridgen for their
valuable assistance, not only in my dissertation, but also in other aspects of my doctoral
program as well.

Special thanks are also extended to the faculty and staff members in the
Department of Park, Recreation, and Tourism Resources for their assistance and helps
during my doctoral program at Michigan State University. I would also like to thank
personally my friends for their supports and encouragement. Finally I wish to gratefully

acknowledge my parents and family for their continuous support.

TABLE OF CONTENTS

Page

LIST OF TABLES ................................................................................................................. Ix

LIST OF FIGURES .............................................................................................................. XIII
CHAPTER 1

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

1.1 INTRODUCTION ........................................................................................................ 1

1.2 PROBLEM STATEMENT ............................................................................................. 7

1.3 STUDY OBJECTIVES ............................................................................................... 11

1.4 RESEARCH QUESTIONS .......................................................................................... 12

1.5 STUDY HYPOTHESES .............................................................................................. 13

1.6 STUDY ORGANIZATION .......................................................................................... 14
CHAPTER 2

LITERATURE REVIEW .......................................................... 16

2.1 ENVIRONMENTAL FACTORS AND THEIR INFLUENCE ON RECREATIOAN

BEHAVIOR/DEMAND .............................................................................................. 16

2.2 STUDIES OF FLUCTUATING GREAT LAKES WATER LEVELS ................................... 24

2.3 PREVIOUS RECREATIONAL BOATING STUDIES ....................................................... 35

2.4 THE BINARY LOGIT MODEL ................................................................................. 41
CHAPTER 3

METHODS 00.0.0000. 0000000000000000000000000000 O 00000000000000 000000000 0000000 0.00048

3.1 STUDY POPULATION .............................................................................................. 49

3.2 SAMPLING PROCEDURE .......................................................................................... 52

3.3 DATA COLLECTION ................................................................................................ 64

3.4 SURVEY INSTRUMENTS .......................................................................................... 65

3.5 RESPONSE RATE AND DATA PREPARATION ............................................................ 67

3.6 DEFINITION AND FORMATION OF VARIABLES ........................................................ 68

3.7 STATISTICAL ANALYSIS AND SPECIFICATION OF THE BINARY LOGIT MODELS 74

vi

CHAPTER 4

RESULTSOOOOIOOOOOOOOOOOOOOOIOOOOOOOO 0000000000000 00000.0... ....... .0000. ........ 00.087
4.1 DESCRIPTION OF DATA OBTAINED FROM THE 2001 LAKE MICHIGAN POTENTIAL
DAMAGES STUDY MAIL AND TELEPHONE SURVEYS .............................................. 87
4.2 RECREATIONAL BOATER AWARENESS AND PERCEPTIONS OF LAKE MICHIGAN
WATER LEVEL CHANGES (STUDY OBJECTIVE #1) ................................................. 95
4.2.1 AWARENESS OF LAKE MICHIGAN WATER LEVEL CHANGES ........................ 95
4.2.2 ASSESSMENT OF LAKE MICHIGAN WATER LEVEL CHANGES ...................... 107
4.3 WHO WAS AFFECTED BY Low WATER LEVELS (STUDY OBJECTIVE #2) .............. 1 18
4.4 RECREATIONAL BOATER BEHAVIORAL CHANGES RESULTING FROM
LOW WATER LEVELS (STUDY OBJECTIVE #3) ..................................................... 130
4.4.1 LOGIT MODEL TO PREDICT/CLASSIFY BOATERS WHO CHANGE THEIR
BOATING LOCATIONS IN RESPONSE TO Low WATER LEVELS ...................... 131
4.4.2 LOGIT MODEL To PREDICT/CLASSIFY BOATERS WHO REDUCE THEIR
BOATING IN RESPONSE TO Low WATER LEVELS ......................................... 138
CHAPTER 5
CONCLUSIONS .0... ........ 0... 00000000000000 O ......... 0.. ....... .0... ...... 0.00149
5.1 SUMMARY OF MAJOR FINDINGS .......................................................................... 150
5.2 INTERPRETING AND APPLYING THE RESULTS OF THE LOGIT MODELS ............... 161
5.3 LIMITATIONS ........................................................................................................ 165
5.4 OVERALL IMPLICATIONS AND RECOMMENDATIONS ............................................. 167
APPENDICES
A. APPLICATION FOR CERTIFICATE OF WATERCRAFT TITLE AND/OR REGISTRATION
OF MICHIGAN ....................................................................................................... 172
B. APPLICATION FOR CERTIFICATE OF WATERCRAFT TITLE AND/OR REGISTRATION
OF WISCONSIN ..................................................................................................... 174
C. APPLICATION FOR CERTIFICATE OF WATERCRAFT TITLE AND/OR REGISTRATION
OF INDIANA .......................................................................................................... 176
D. APPLICATION FOR CERTIFICATE OF WATERCRAFT TITLE AND/OR REGISTRATION
OF ILLINOIS .......................................................................................................... 178
B. THE 2001 LAKE MICHIGAN POTENTIAL DAMAGES STUDY MAIL SURVEY ........... 180

F. THE 2001 LAKE MICHIGAN POTENTIAL DAMAGES STUDY TELEPHONE SURVEY 187

vii

G. FIVE BINARY LOGIT MODELS RESULTS ............................................................. 201

BIBLIOGRPHY ................................................................... 21 1

viii

Table 2.1:
Table 2.2:
Table 2.3:
Table 2.4:
Table 2.5:

Table 3.1:

Table 3.2:

Table 3.3:

Table 3.4:

Table 3.5:

Table 3.6:

Table 3.7:

Table 3.8:

Table 3.9:

LIST OF TABLES

The mean lake levels in September for the ﬁve Great Lakes .......................... 26
The expected water levels in the Great Lakes in 2003 .................................... 27
Lake Michigan Potential Damages Study projects in 2000 and 2001 ............. 29
Number of bOats vying for each ramp in different water level ........................ 33
Previous Michigan Boating related studies and literature reviews .................. 36
Information contained in the watercraft registration lists in the four Lake

Michigan states ................................................................................................ 51
Number of the watercraft with the valid 2001 registrations including their

expiration year ................................................................................................. 53
Type of use of the valid 2001 registered watercraft ........................................ 54

Number of the 2001 registered watercraft and recreational/pleasure
watercraft ......................................................................................................... 55

Number of the valid 2001 registered recreational/pleasure watercraft to be
sampled in each of the four states .................................................................... 55

Number of the valid 2001 registered recreational/pleasure watercraft
sampled by county zone and state for the 2001 Lake Michigan Potential
Damages Study mail survey ............................................................................ 61

Number of the valid 2001 registered recreational/pleasure watercraft of
different sizes in different zones in each of the four states for the 2001
Lake Michigan Potential Damages Study mail survey .................................... 62

Number of the valid 2001 registered recreational/pleasure watercraft
interviewed by the 2001 Lake Michigan Potential Damages Study
telephone survey by county zone and state ..................................................... 63

Number of the valid 2001 registered recreational/pleasure watercraft of
different sizes in different zones in each of the four states for the 2001
Lake Michigan Potential Damages Study telephone survey ........................... 64

ix

Table 3.10: Response rate to the 2001 Lake Michigan Potential Damages Study mail

survey ............................................................................................................. 68
Table 3.11: Types of 2001 boating seasons storage facilities ........................................... 70
Table 3.12: Types of 2001 boating seasons storage access to different waters ................ 70

Table 3.13: Boat storage segments from the 2001 Lake Michigan Potential Damages

Study mail survey .......................................................................................... 71

Table 3.14: Boat storage segments from the 2001 Lake Michigan Potential Damages

Study telephone survey .................................................................................. 72

Table 3.15: Recreational boaters’ responses about Lake Michigan water level changes. 73

Table 3.16: Recreational boaters’ awareness about Lake Michigan water level changes 73

Table 3.17: Recreational boaters’ perceptions about the degree of the Lake Michigan

water level changes ........................................................................................ 74

Table 3.18: The dependent variables for ﬁve binary LOGIT models ............................... 77
Table 3.19: The potential independent variables for ﬁve binary LOGIT models ............ 78
Table 4.1: Registered boaters who completed the 2001 Lake Michigan Potential

Damages Study mail survey ............................................................................ 89
Table 4.2: Registered boaters who answered the 2001 Lake Michigan Potential

Damages Study telephone survey .................................................................... 90
Table 4.3: Characteristics of registered watercraft owned by mail survey respondents... 91
Table 4.4: Characteristics of registered watercraft owned by telephone survey

respondents ...................................................................................................... 92
Table 4.5: The comparison between the 2001 Lake Michigan Potential Damages Study

mail and telephone surveys .............................................................................. 93
Table 4.6: Relationships between boater and boat characteristics and boater awareness

of Lake Michigan water level changes ............................................................ 97
Table 4.7: The result of the ﬁnal binary LOGIT model for boaters’ awareness of Lake

Michigan water level changes ....................................................................... 103
Table 4.8: The probability of awareness of Lake Michigan water level changes by

selected boater characteristics ....................................................................... 104
Table 4.9: Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who were and

were not aware of Lake Michigan water level ﬂuctuations ........................... 107

Table 4.10:

Table 4.11:

Table 4.12:

Table 4.13:

Table 4.14:

Table 4.15:

Table 4.16:

Table 4.17:

Table 4.18:

Table 4.19:

Table 4.20:

Table 4.21:

Table 4.22:

Table 4.23:

Table 4.24:

Table 4.25:

Boaters’ perceptions about the extent of Lake Michigan water level

ﬂuctuations .................................................................................................. 1 07
Relationships between boater and boat characteristics and boater

perceptions of the magnitude of Lake Michigan water level changes ......... 109
The result of the ﬁnal binary LOGIT model for boaters’ assessments

of the degree of Lake Michigan water level changes .................................. l 15
The probability for the boaters who assessed Lake Michigan water level

had dropped a lot by selected boater characteristics .................................... l 16

Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who assessed
the Lake Michigan water levels had dropped a lot and remained the same 118

Relationships between boater and boat characteristics and whether a
boater’s behavior was inﬂuenced by concerns relating to low water levels 120

The result of the ﬁnal binary LOGIT model for classifying boaters
who were inﬂuenced and not inﬂuenced by concerns relating to
low water levels ........................................................................................... 126

The probability for the boaters who were inﬂuenced by concerns relating
to low water levels by selected boater characteristics ................................. 127

Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who were
inﬂuenced and not inﬂuenced by concerns relating to low water levels ..... 130

Relationships between boater and boat characteristics and whether
boaters changed their boating locations in response to low water levels 132

The result of the ﬁnal binary LOGIT model for predicting/classifying
boaters who changed boating locations in response to low water levels ..... 136

Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who changed
and did not change their boating locations in response to low water levels 138

Relationships between boater and boat characteristics and whether
boaters did less boating in response to low water levels ............................. 140

The result of the ﬁnal binary LOGIT model for predicting/classifying
boaters who did less boating locations in response to low water levels ...... 145

The probability for the boaters who did less boating in response to low
water levels by selected boater characteristics ............................................ 146

Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who did and
did not do less boating in response to low water levels ............................... 148

xi

Table 5.1: Summarized results for the ﬁve Lake Michigan water level binary
LOGIT models ............................................................................................... 151

Table 5.2: A summary of statistically Signiﬁcant independent variable parameters
comprising the Lake Michigan water level binary LOGIT models ............... 153

Table G. l: The results of eight binary LOGIT models for boaters’ awareness of
Lake Michigan water level changes. ............................................................. 202

Table G2: The results of ﬁve binary LOGIT models for boaters’ assessments of
Lake Michigan water level changes .............................................................. 204

Table G3: The results of nine binary LOGIT models for the boaters who were
inﬂuenced and not inﬂuenced by concerns relating to low water levels ...... 206

Table G.4: The results of six binary LOGIT models for the boaters who changed their
boating locations in response to low water levels ......................................... 208

Table G5: The results of nine binary LOGIT models for the boaters who did less
boating in response to low water levels ........................................................ 209

xii

LIST OF FIGURES

Figure 1.1: The variablility of the water precipitation in Lake Superior and Lake

Michigan-Huron from 1996 t02001 ................................................................. 4
Figure 2.1: The water levels on the Great Lakes from 1918 to 1998 ............................... 25
Figure 2.2: The logistic function curve ............................................................................. 44
Figure 3.1: Lake Michigan counties in the four states ...................................................... 50
Figure 3.2: Sampling proportions for the three sampling zones in Michigan .................. 57
Figure 3.3: Sampling proportions for the three sampling zones in Wisconsin ................. 58
Figure 3.4: Sampling proportions for the three sampling zones in Indiana ...................... 59
Figure 3.5: Sampling proportions for the three sampling zones in Illinois ...................... 60
Figure 3.6: The framework for the study .......................................................................... 75
Figure 3.7: The sequence of steps for developing and evaluating LOGIT models .......... 76

xiii

CHAPTER 1

INTRODUCTION

1.1 Introduction

The Great Lakes including Lake Superior, Lake Michigan-Huron, Lake St.Clair,
Lake Erie and Lake Ontario form the largest fresh surface water system on earth,
covering more than 94,000 square miles and draining more than twice as much land (The
Great Lakes Information Network, 2002). Lake Michigan is the third largest Great Lake,
surrounded by four states: Michigan, Wisconsin, Indiana and Illinois. This system
greatly affects the way of life, as well as all aspects of the natural environment, from
weather and climate to wildlife and habitat. Many outdoor recreational activities and

outstanding tourism opportunities are available on and near Lake Michigan.

From 1996 to 2000, the number of registered watercraft in Michigan increased by
5.7 % (National Marine Manufacturers Association, 2001). In 2001, there were 1.1
million watercraft on the Michigan Secretary of State’s registration list of which 825,260
were currently registered for operation. It is estimated that registered boaters logged
more than 22 million days of boating in 2001 (Planning & Zoning Center, Inc. et al.,
2002). Recreational boaters support a major industry that includes boat manufacturers,
retailers and builders, marina operators, maintenance services and marine business
suppliers. Recreational boating also has a direct connection to sport ﬁshing and accounts

for one billion dollars of regional economic impact. More than 600 marinas (public,

commercial, and private clubs) and 450 boat dealerships are in Michigan. A conservative
estimate Of spending by recreational boaters in 2002 is estimated to be $2.24 billion
including $860 million on boating trips, $502 million in annual watercraft spending', and
$880 million in boat sales2 (Mahoney et al., 2003). Talhem et al. (1998) estimated that
local expenditures by boaters who rented seasonal slips in Michigan marinas were about
$7,000 per boat, or $200 million in Michigan. Lee (1999) estimated that total spending
by owners of registered boats was approximately $635 million on trips in 1998 within

Michigan.

The Great Lakes have a close relationship with the economy and lifestyle of those
in Michigan, and elsewhere throughout the US. which has more than 3,000 miles of
coastline and more than 30,000 jobs depend on the Great Lakes-St. Lawrence water
transportation system (Kurth, 2002). Lake Michigan’s tourism and recreational boating

industry is dependent both on water quality and quantity.

Fluctuations in Great Lakes Water Levels and Impacts on Boating

Over the past ﬁve years Lake Michigan has experienced lake levels that have
approached the all-time recorded low received in 1964. The Great Lakes Environmental
Research Laboratory (2001) reported “Lake Michigan and Lake Huron are experiencing
the lowest water levels in 35 years. These water levels have adversely affected

recreational boaters, marinas, commercial navigation and hydropower” (Sellinger, 2002).

 

’ The annual watercraft spending includes the spending on equipment/accessories, marina slip rental fess,
repair and maintenance, put-in and haul-out fees, insurance, and storage (off-season).

2 The boat sales are the spending by Michigan registered boat owners on new and pre-owned boats in 2002
and include new watercraft and engine sales, pre-owned watercraft sales, and new trailer sales.

From 1997 to 2001, the water levels in Lake Michigan and Lake Huron dropped by 3.4
feet. Although the water levels of Lake Michigan and Lake Huron were lower than those
experienced in several decades, the water levels of 1999 through 2001 were still not the
lowest experienced during this century and were several feet above a calculated potential

extreme low3 (Planning & Zoning Center, Inc. et al.4, 2001).

Great Lakes water levels can change seasonally each year and even vary over
longer periods. Short-tenn changes are generally of greater magnitude than the monthly
averages. The range of seasonal water level ﬂuctuations from low levels in winter to
high levels in summer on the Great Lakes averages 12 to 18 inches. Long-term
ﬂuctuations occur over periods of continuous years. Over the last century, the range from
extreme high to extreme low water levels has been nearly 4 feet for Lake Superior and

between 6 and 7 feet for the other Great Lakes (Keillor, 2002).

Two primary factors, which could cause major changes in Great Lakes water
levels, are a long period of heavy or low precipitation and evaporation rates. For
example, low Lake Michigan water level during 1999 to 2001 was caused by the natural
cycles of precipitation within the watershed and from a combination of factors such as
lower precipitation, lower runoff, higher evaporation and higher air temperatures during

1997, 1998, 1999, 2000, and 2001 (The Great Lakes Information Network, 2002).

 

3 The extreme low water deﬁned by Lake Michigan Potential Damages Study (LMPDS) is projected to be
574.3 feet.

" This study was conducted by Planning & Zoning Center, Inc., department of Park, Recreation and
Tourism Resources of Michigan State University and EPIC-MRA and is a partial study of Lake Michigan
Potential Damages Study funded by US. Army Corps of Engineers, Detroit District.

The quantity and the timing of rain or snow can affect water levels. Water levels
rise in the spring when rain combines with melting snow. Therefore, in an average year,
each of the Great Lakes receives its peak water supply between March and May. For
example, the heavy snow of 1995 and 1996 caused the water levels to reach close to
maximum records on most lakes in 1997. From 1998, drought conditions emerged in the
Great Lakes and basin and water levels changed from near record highs on each of the

upper lakes to their lowest levels in 35 years in 2000 (Sellinger, 2002).

Figure 1.1 illustrates the variability in precipitation that has occurred over the past
seven years on Lakes Superior and Michigan-Huron. Precipitation on Lakes Superior

and Michigan-Huron provided long-term supplies throughout the system.

 

Precipitation
Percent above or below average
1995-2001

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I.—

 

1995 1996 1997 1998 1999 2000 2001
[:1 Lake Superior I Lake Michigan-Huron

 

 

 

 

 

 

Figure 1.1. The variability of the water precipitation in Lake Superior and Lake
Michigan-Huron from 1995 to 2001.

Source: US. Army Corps of Engineers, Detroit District, 2002

The other primary factor causing water level changes is evaporation. During the
peak evaporation months (September to December), the Great Lakes can lose one to two
inches of water per week. Evaporation occurs when cold dry air hangs over the warm
lake bodies. The temperatures in summer that warm the lake water affect the rate at
which a lake evaporates. When the cool autumn temperatures are over the warmer lake
surfaces, evaporation begins in earnest. With the warm air temperatures over the past
four years, water temperatures on all the Great Lakes also have been above average
precipitation to decrease water levels steadily (US. Army Corps of Engineers, Detroit

District, 2002).

The ﬂuctuating water levels not only affect to the environment but also the local
economy or related business industry such as boating (Planning & Zoning Center, Inc. et
al., 2002). For example, the ﬂuctuating water levels can affect the extent of ﬂooding,
shoreline erosion and shoreline property damage, wetland acreage, depth of navigation
channels, hydroelectric power output, water quality, and aquatic ecosystem. Moreover,
both low and high water levels can have signiﬁcant ﬁnancial impacts on Lake Michigan
businesses and communities such as commercial navigation, recreation boating, marinas,

beaches, ﬁshing, cottage, and homeowners.

The Lake Michigan Potential Damages Study (LMPDS) categorized the impacts
of low water levels to three types (Planning & Zoning Center, Inc. et al., 2002). First, the
potential social impacts could be land use management and shoreline management
inﬂuence. US. Environmental Protection Agency ofﬁcials reported that falling levels
will lead to shifts in currents and depths that may cause some intakes and create problems

with water taste, odor and even public health. Second, water level changes can bring the

environmental impacts to archaeological and special natural features, habitat biodiversity,
ﬁsheries or threatened and endangered species. Third, there can be the ﬁnancial impacts
on the related recreation boating industry and boating tourism business. For example,
extreme ﬂuctuations in lake levels can limit recreational boating (e. g., launch, navigation)
because low (or high) water levels may make some areas inaccessible for boat launches

and some navigation channels inaccessible to boaters.

Marinas located in the Great Lakes coastal counties spend millions to dredge boat
slips, channels, and harbors to sustain the use, but may still lose customers. For example,
during 1999 and 2000 private boat owners were affected through reduced use of their
boats because of loss of access to some marina slips, and fewer boat launching
opportunities at public ramps (Planning & Zoning Center, Inc. et al., 2001). Not only
private marinas, but also federal harbors located at the mouth of a river or along a
coastline need dredging to keep the water level deep enough for watercraft. Without
dredging, most rivers and harbors would be inaccessible for recreational boating and

commercial navigation.

The impacts of low water levels have been of concerns to legislators, recreational
boating organizations, and many Lake Michigan communities. Because Lake Michigan
has serious erosion problems and was the most damaged lake during the previous high
water periods in the 1970s and 19808, US. Army Corps of Engineers started to conduct
the Lake Michigan Potential Damages Study (LMPDS) to assess potential damages due
to extreme low and high water levels (a range over 9 feet) since 1996. The objective of
the LMPDS is to create a modeling procedure to estimate ﬁnancial effects of lake level

changes and related social, environmental, and cultural consequences. The main focus of

the LMPDS is the ﬁnancial effects and inﬂuences caused by low Lake Michigan water
levels, which are assessed for residential, commercial, industrial, and institutional uses
and include recreational boating, municipal water supply and wastewater treatment, and
navigation. The LMPDS only provided information regarding the revenue losses on the
marina industry and boat dealers, assessed for Lake Michigan coastal counties in 2000

and 2001 because of low Lake Michigan water levels.

Based on a series of studies from the LMPDS completed in 2000 and 2001,
Mahoney et al. concluded that “there is a need for a continuing, coordinated effort to
identify and proﬁle Great Lakes marinas and evaluate their vulnerability to ﬂuctuating
water levels; better understand Great Lake boater behavior and likely response to changes
in boating opportunities; and estimate the ﬁnancial impacts of different levels of boating
activity on local communities” (Planning & Zoning Center, Inc. et al., 2002, p56).
Because water levels differ between years and areas, and the boating market and behavior
of boaters can change signiﬁcantly within a year, only data gathered during the same

boating season can be used to estimate the potential impacts of low water correctly.

1.2 Problem Statement

Studies conducted by Mahoney et al. (2001 and 2002) for the Lake Michigan
Potential Damages Study (LMPDS) mainly focused on the revenue and cost impacts of
low water on marinas and boat dealerships. Those included unusable slips, slips that
could not accommodate the size boats they were designed to accommodate, dredging, and

services such as fueling made inaccessible because their ﬁxed positions were inaccessible

(because of increasingly inadequate water depth in relation to boat draft). That study
documented that Lake Michigan water levels (both low and high water) had a negative

ﬁnancial impact on recreational boating.

That study estimated that there would be around $12 million to $28 million losses
in the 297 marinass, located in the 34 coastal counties, depending on the decrease in Lake
Michigan water levels. Some commercial marinas had or were likely to go out of
business because of dropping water levels. This situation will further reduce boating
access and the availability of support services (Planning & Zoning Center, Inc. et al.,
2002, p4). The same study also reported increased repair and maintenance costs related
to low water caused damages. Besides the ﬁnancial impacts on marinas and dealerships,
Lake Michigan low water level also affected the use of the boating infrastructure such as

boat launch ramps, which was important to sustain the recreational boating industry.

In the 2001 LMPDS report, 24% of mail survey respondents and 18% of
telephone survey respondents reported boating more days in 2000 than in 2001. For the
boaters who responded boating fewer days, 23% of them cited low water as a primary
reason. IfLake Michigan water levels were to drop one, two, or three feet below the
2001 levels, 40%, 60% or 70% of respondents believed that low water would force them
to ﬁnd alternative launch sites. That report not only concluded that decreases of these
magnitudes would result in substantial reduction in the number of people who go boating

or buy boats but also suggested that the boating industry should consider carefully the

 

5 The estimated number was based on the 297 marinas responding to the 2001 survey. They are also
located in Lake Michigan coastal counties in four states. The ﬁnancial impacts included the ﬁnancial loss
from the unusable wet slips, wet slips that could not handle the size boats they were designed for,
dredging cost, damage to docks, piers, break walls, replacement and repairs cost, inaccessibility cost, the
cost of new or rebuilt facilities, and other damage/impacts/losses because of low water levels.

perceptions of recreation boaters about low or high water levels (Planning & Zoning

Center, Inc. et al., 2002, p60).

While those studies have been useful to the boating industry in helping them win
passage of helpful legislation (e. g., low interest dredging loans) and by providing
information about the ﬁnancial impacts in boating and tourism industries, they failed to
provide a comprehensive understanding of the impacts of low water levels on recreational
boaters. Some boaters who were surveyed were aware of the changing of water levels,
but others were very unaware. The behaviors of some boaters were affected or
inﬂuenced in different ways by low water. Low water induced behavioral changes
included boating fewer days, going to different locations or stopping boating. Those are
still unresolved aspects, which are also pointed to in the LMPDS conclusion (2002):
“There needs to be a continuing coordinated effort to, thus ... better understand Great
Lakes boater behavior and likely response to changes in boating Opportunities” (Planning

& Zoning Center, Inc. et al., 2002, p56).

This study focuses on the questions that have not been adequately addressed or
answered in previous studies including (1) what is the awareness and perceptions among
recreational boaters of water level changes in Lake Michigan, (2) what boaters are more
likely to be affected by low water concerns, and (3) what types and degree of boating
behavioral changes are induced by low water levels. Estimation of the potential impacts
and inﬂuences of ﬂuctuations in Lake Michigan water levels cannot be complete without
understanding recreational boaters’ awareness and perceptions relating to water levels,
the size and types of boats/boaters most likely to be impacted by low water levels, and

the types of behavioral changes induced by low water such as decreased boating activity,

different boating locations, and changing where boats are stored during the boating

8638011.

Purposes of This Study

1.

This study has the following three purposes:

Previous lake level studies have only focused on the cost and revenue impacts on
boating related business industry such as marinas and boat dealerships. The analyses
have focused primarily on supply side impacts and do not provide signiﬁcant
information on the relationships between low water levels and the “demand” for Lake
Michigan boating. While impact assessment serves an important purpose, the
recreational boating industry and boating agencies also require information on the
potential short-term and long-term inﬂuences or low water levels (and other
environment related factors) on boating “demand.” Projecting the type and range of
the potential impacts of future changes in water levels on recreational boating require

a more complete understanding of boater awareness and responses.

To produce a better understanding of how recreational boaters perceive and assess
water levels, and whether and how they adjust their volume of boating days and
boating locations as input to the policy, investment and management decisions by

public boating authorities and agencies, and the recreational boating industry.

This study will provide a more comprehensive and systematic understanding of how
boaters change their behaviors due to the changes in the quality and quantity of

boating opportunities.

10

1.3 Study Objectives

Low water levels appear to have negatively impacted boating experiences and
impacted boater behaviors. However the extent to which low water levels and boater
awareness of water levels inﬂuences boater behaviors including how much and where
they boat is not well understood. Therefore, this study is intended to test and understand
the relationships between low water levels and various boater behaviors including the

development of different binary LOGIT models.
Three objectives guide the conduct of this study.

Objective 1: To assess recreational boaters awareness and appraisal of the

magnitude of changes in Lake Michigan water levels.

Objective 2: To identify which types of boaters are most likely to be inﬂuenced
by water level changes and what factors are associated with the
extent of the inﬂuence of water level changes on various boating

behaviors.

Objective 3: To identify the relationships between the possible behavioral
changes influenced by water level changes and boat and boater
characteristics, boat storage segments and locations and boating

days and locations.

11

1.4 Research Questions

A set of research questions is formulated for each of the three study objectives.

These questions are used to determine data requirements and methods of analysis.

The research questions related to objective 1 are: “To what extent are boaters
aware of the water level changes in Lake Michigan?” and “How do they assess the degree
of water level changes?” These questions are answered by two binary LOGIT models
with two different dependent variables: boaters’ awareness and perceptions toward the
Lake Michigan water level changes and some independent variables including boater
characteristics, boat characteristics, boat storage segments and locations, volume of

boating days and boating locations.

The primary question related to objective 2 is “what type(s) of boaters are most
likely to be inﬂuenced by water level changes?” To answer this question a binary
LOGIT will be formulated with “low water inﬂuenced their boating behavior” as the
dependent variable and boater characteristics, boat characteristics, boat storage segments

and locations, volume of boating days and boating locations as independent variables.

The third objective will be accomplished by addressing two related questions: (1)
what types of boater behavioral changes (e. g., stopped boating, decreased boating
volume, changed boating locations, moved the boat to a different storage location) did
change low water levels cause? and (2) does a relationship exit between different types of
behavioral changes and boater and boat characteristics? Two binary LOGIT models will
be developed to predict the probabilities of different types of behavioral changes for

different types of boaters and boats. The possible “behavioral changes in response to low

12

water levels” are the dependent variables including changing the boating locations and
reducing the volume of boating days for two binary LOGIT models. The independent
variables are also boater characteristics, boat characteristics, boat storage segments and

locations, volume of boating days and boating locations.

1.5 Study Hypotheses

Three different hypotheses are formulated and tested.
Hypothesis 1

Boater awareness and assessment of changes in Lake Michigan water levels are

signiﬁcantly inﬂuenced by:

1. Boater socio-economic characteristics: age, education, income, gender,

permanent state of residence, and location of permanent residence.

2. Boat characteristics: length and type of their boats and number of years the boat

has been owned.

3. Boat storage characteristics: boat storage type, where (state and county) the boat

was stored during the boating season.

4. Boating characteristics: whether or not the boat has been Operated during the
past three years, where it has been operated and the volume of their boating

activity.

13

Hypothesis 2

Boater socio-economic characteristics, boat characteristics (e.g. boat type, length
and years owned), boat storage segments and locations, and boating locations and boating
volume differences exist between recreational boaters who were inﬂuenced and not

Signiﬁcantly inﬂuenced by the low water levels during the 2001 boating season.
Hypothesis 3

Behavioral changes induced by the low water levels are signiﬁcantly inﬂuenced
by boater socio-economic characteristics, boat characteristics (e. g. boat type, length and
years owned), boat storage segments and locations, and locations where they boat and

boating volume.

1.6 Study Organization

This dissertation is presented in ﬁve chapters. Chapter one deﬁnes the study
problem and research questions and also presents study objectives and study hypotheses.
Chapter two provides a review of some relevant literature including a series of statewide
boating studies conducted over thirty years and recent studies focused on the impacts of
low Great Lake water levels. In addition chapter two also reviews literature relating to

development and application of the binary LOGIT model.

The methods used to collect and prepare the data used to specify the binary
LOGIT models are described in chapter three including two different survey methods
(e. g., study population, sampling procedures, data collection, survey instruments, and

response rate). Furthermore chapter three discusses the formation of the variables that

14

comprise the model, the deﬁnition for the dependent and independent variables, and the

modeling speciﬁcations related to three different study objectives.

Chapter four presents the results of the analyses conducted for each of three study
objectives including the hypothesis testing results, the process employed to develop the
binary LOGIT models, the speciﬁcation of the different models, the model parameter
estimations and the model evaluations and predictions of boater awareness and
perceptions regarding the Lake Michigan water level changes and the inﬂuence of low
water on different boater behaviors. Chapter ﬁve summarizes the study results and
applications, discusses the study’s limitations, and offers some conclusions and

suggestions for future research.

15

CHAPTER 2

LITERATURE REVIEW

This literature review is divided into four sections. Since water level is an
environmental factor, the ﬁrst section reviews literature relating to how environmental
factors inﬂuence or impact recreationist’s behaviors and demand. The second section
focuses on previous studies concerning the inﬂuences and impacts of low Great Lake
water levels on recreational boating and the boating industry. The third section reviews
previous Michigan boating studies on such topics as the economic impact (e.g., boater
Spending) of recreational boating, boating demand and boating patterns. The ﬁnal section
reviews literature relating to the binary LOGIT model with an emphasis on applications

related to this study.

2.1 Environmental Factors and Their Inﬂuence on Recreation

Behavior/Demand

All people have their own ways of valuing their leisure time and many spend a
great deal of it in various recreation environments. A growing number of persons are
opting to live in ‘planned recreational communities,’ which provide them with tennis
courts, swimming pools, tennis courses, and even boating and skiing in certain cases
(Fisher et al., 1984). The environment in which we live inﬂuences our perceptions,

attitudes or values about the environment, and the changes in the recreation environment

16

also inﬂuence our recreational choices and behaviors (Garden & Stern, 1996). Hanson
and Hatch (1998) found that a one-foot reduction in water level in the Apalachicola-Flint-
Chattahoochee River basins would result in a signiﬁcant decrease in annual recreational
visitation frequency (Hanson, 1998; Hanson & Hatch, 1998). A linear model created by
MacGregor (1988) identiﬁed the change in boating activity related to sediment
accumulation in a lake and estimated that the total value loss across all 46 of Ohio’s state

park lakes was over $470,000 (MacGregor, 1988).

When studying the impact of changes in the recreation environment,
psychologists frequently focus on identifying and describing the relationships between
behavior and environment (Fisher et al., 1984). Conversely, economists tend to focus on
how changes in the environment impact the supply of recreational opportunities and as a
result on the behaviors of recreationists (Fisher et al., 1984). The following literature that

is reviewed relates to how changes in the environment inﬂuence recreational behaviors.

Changes in Great Lake water levels and understanding boater sensitivities and
adaptive responses are deﬁnitely related to environmental psychology. The main focus
of environmental psychology is to understand better the relationships between people and
their physical environment (Weigel, 1983; Howell & Laska, 1992). Heimstra and
McFarling (1978) deﬁned environmental psychology as the discipline that is concerned
with the relationships between human behavior and the physical environment. Bell et al.
(2001) deﬁned environmental psychology as the study of the relationships between

behavior and experiences and the built and natural environments.

17

The ﬁeld of environmental psychology is extremely broad including identifying
people’s environmental perceptions, cognitions, and attitudes, the theories of
environment-behavior relationship and evaluation of people’s response to environment
changes (Fisher et al., 1984). Current themes with the environmental psychology
research include the topics about awareness and attention, perceptions and cognitive
maps, preferred environments, environmental stress and coping, participation, and
conservation behavior (Evans et al., 1982; Kaplan & Kaplan, 1982; Garling & Golledge,

1993)

A major ﬁeld of environmental psychology involves the study of environmental
awareness and attention, and assessments and perceptions. Since environmental
awareness study is to understand human behavioral, it needs to start with understanding
how people notice the environment (Sommer, 1972). The environmental perceptions
study is to understand how people picture the natural environment and the environmental

changes.

Among the factors, which may affect environmental perceptions, include personal
characteristics (perceptual ability and experience), cultural and environmental effects and
other effects on environmental experience. A number of studies have investigated the
relationship between socio-demographic factors and differences in opinion and attitudes
toward the environment (Samdahl & Robertson, 1989; Arcuy & Christianson, 1990;
Lyons & Breakwell, 1994). Van Liere and Dunlap (1980) proposed some hypotheses
about the determinants of environmental concern including age, social-class, residence

and gender hypothesis.

l8

Sometimes the “environment” is considered as a factor in shaping behavior in
terms of a one-way cause. For example, the cause-“low water levels” has an effect on the
boating environment, which might inﬂuence boaters’ behaviors. Environmental
psychologists suggest that it is important to study the interrelationship of
cause/effect/inﬂuence more carefully. On the other hand, most environmental
psychologists argue that behaviors are not determined only by environmental factors, and
people can change the way they behave or the environment in which they live, work and
play (Bell et al., 2001; Tolley, 2001). For example, many boaters have the capability to
change the locations where they boat if factors in the boating environment change (e. g.,
low water levels change the quality of the boating environment). Recognizing this fact
boating business including marinas can invest in mediating low water effect through the

dredging and the design of facilities and services.

Five basic assumptions underlie a signiﬁcant number of environmental psychology
studies and articles: (1) different groups are affected differentially by various
environmental inﬂuences, (2) environmental factors have differential effects and varying
extent of inﬂuence, (3) many environmental inﬂuences are unconscious, (4)
representation involves mental ‘images’ and (5) Space and place have different symbolic
meanings for people (Tolley, 2001). For example, low water may be differentially
perceived by different boaters depending on how often they boat, where they live (e. g.,
near the Great Lakes), or where they boat (e. g., Lake Michigan, inland lakes or other
Great Lakes). Even though ﬂuctuating Great Lake water levels have gathered a great
deal of media attention, some boaters have been immune to this information and

remained unaware of the extent or location of low lake levels. In other cases, boaters and

19

non-boaters alike form perceptions of low water (e.g., severity, impacts) over time, and

often subliminally.

It is frequently difﬁcult to determine precisely the behavioral changes that are
caused by changes in the environment due to a number of intervening factors. One
difﬁculty is where and what do we want to attempt to measure. Also the problem of
representativeness is present. Real behavior happens in real time, and many behaviors

are often very transient (Garden & Stern, 1996).

Three types of research methods used to conduct the environmental psychology
research are: (1) experimental research (Veitch & Arkkelin, 1995; Bell et al., 2001), (2)
correlation research (Bell et al., 2001), and (3) descriptive research (Fisher et al., 1984;

F ransson & Garling, 1999; Bell et al., 2001). The method used in this study is correlation
and descriptive research. Correlation research is to identify the relationship between
situational variations and some other variables. Descriptive research is to report their
characteristics or reactions in a particular situation or any environmental changes.
Environmental psychologists usually use descriptive research to report reactions that
occur in a particular situation. Environmental psychologists often rely on descriptive
research in an attempt to identify behaviors in response to speciﬁc environmental
changes. According to Fisher et al. (1984), descriptive research is common because it is
not constrained by a need to infer causality or association, and it can often be generalized

to other settings.

20

Modeling Recreation Substitutes

Descriptive research has certain advantages, but only describes and does not
explain why recreationists (e. g., boaters) react in certain ways to changing environmental
factors. The ﬁeld of recreation modeling is developed in an effort to understand how
recreationists respond to changes in recreational opportunities and environmental

attributes.

A basic assumption underlying most recreation modeling is that individuals (e. g. a
recreational boater) will seek to maximize the utility (value) derived from a set of
possible recreational choices (Siderelis, 1995). Loomis (1995) contends that recreational
utility may be affected by environmental factors that comprise the attractiveness of a
recreation destination and attractiveness is a function of factors such as air quality, water
quality and clarity, degree of naturalness, abundance of ﬁsh and wildlife, forest cover,

etc.

If factors that comprise a recreational destination /location are diminished or
taken away, recreationists may (if substitute locations/sites are available) adjust their
behaviors (e.g., selection, number of visits) to maximize their utility. ISO-Ahola (1986)
proposed a theory of substitutability — the tendency to substitute one recreation activity or
location for another. The two primary substitution determinants are an individual: (1)
perception or analysis of a reason (e. g., low water levels) for a substitution, and (2)
perception or analysis of the qualities (e. g., attributes, beneﬁts) of the leisure activity or
location to be replaced compared to the available alternatives (Iso-Ahola, 1986). Thus,

the ﬁrst step in modeling recreational choice is to determine awareness or perception of

21

possible reasons to consider substitutes. The next step is to determine tendency or

willingness to make a substitution decision.

According to Loomis (1995) the quality of the natural resources and environments
can inﬂuence recreation choices and behaviors including: (I) participate or not
participate, (2) where to recreate - site selection, (3) trip frequency, and (4) length of stay.
Extending this to lake level ﬂuctuations suggests that if boaters are aware of the
magnitude and implications lake level ﬂuctuations, this factor can inﬂuence them to stop

boating, change their boating locations, and reduce the amount of boating.

Decisions whether to participate in a recreation activity, or to Visit a particular
recreation site, is frequently modeled as a “zero (not participate or visit)—one @articipate
or visit) variable.” A commonly used zero-one (dependent variable) model is the binary
LOGIT or PROBIT model (Hof & Kaiser, 1983; Peterson et al., 1983; Stynes et al., 1984;
Loomis, 1995; Lupi et al., 2000). The dependent variable in a LOGIT or PROBIT model
is “one” for a decision to participate or Visit, and “zero” for decision not to participate or
Visit. The independent variables often include personal characteristics such as physical
ability, attitudes, preferences, and incomes (Loomis, 1995). When it comes to modeling
recreational site/location choices, conditional or multinomial LOGIT models are usually
used given that recreational site/location choices commonly involve more than two

possible alternatives (Morey, 1991; Loomis, 1995; Sidererlis, 1995; Lupi et al., 2000).

Loomis (1995) developed a model for predicting trip frequency decisions based
on a Travel Cost Model. The purpose is to observe how trip frequency changes in
relationship to Site characteristics such as visitor facilities, water quality, or wildlife

abundance (Loomis, 1995). Contingent Valuation Method is frequently used to assess

22

the impacts of site attributes on recreational use behaviors. Hanson and Hatch (1998)
developed a regression equation-relating to how water level changes on Lake Martin
impacted users. The result showed that a one-foot reduction in water level would cause a
decrease in annual visitation frequency per person of 4.486 days (Hanson, 1998; Hanson
& Hatch, 1998). The problem was that a low R2 value indicated that individual response

varied greatly.

Summary

According to environmental psychology and the substitute theory of recreation
behavior, the recreational user’s choice/behavior is inﬂuenced by the recreation resources
and environment quality. Thus the Great Lakes have a close relationship with tourism
and recreational boating industry. Furthermore the Great Lakes water quality and
quantity will inﬂuence the recreation boating industry and boaters’ behavior. The
following section of this chapter clearly identiﬁed low water level problems caused
ﬁnancial impacts and revenue losses to the boating business industry because of the

losses of their consumers (recreational boaters).

23

2.2 Studies of Fluctuating Great Lakes Water Levels

Fluctuations in Great Lake water levels, both high and low level, come in cycles
but not in regular or predictable cycles (Keillor, 2002). In the twentieth century, high
water levels were ﬁrst experienced in the 19505. Two other high-water periods occurred
during 1973-74 and 1985-86 on the Great Lakes. Besides Great Lakes, some other areas
such as Lake Minnetonka, are experiencing the low water levels Simply because of
climate changes. The Twin Cities area, as of April 27th (2003), has only received 4.12
inches of precipitation, compared with 8.36 inches during the same period last year and
the low precipitation cause the water level in Lake Minnetonka drops below 928.6 feet
above sea level. Also southwestern and eastern Ontario have experienced lowest surface
water levels and driest soils recorded for several decades because of the extended periods
of low rainfall and high temperatures in recent years (Ontario Ministry of Natural

Resources, 2002).

Figure 2.1 provides a historical record of water levels for Lake Superior, Lake
Michigan-Huron, Lake St. Clair, and Lake Erie from 1918 to 1998 (U .8. Army Corps of
Engineers, 2002). It shows that the high water levels occurred coincidentally in all Great
Lakes during the same periods. Low water levels occurred at different times in the
various Great Lakes. For example, on Lake Superior, the low water level occurred
around 1926 to 1927. Low water levels occurred on Lake Michigan during 1926-1927,
1933-1938, 1963-1965, and 1999-2001. Lake St. Clair experienced low water in 1924-
25 and 1934-35. Lake Erie experienced low water in 1934 to 1935. Overall, the
historical record of water level indicates that low water is a more common occurrence for

Lake Michigan.

24

 

 

Lake Superior

 

177.4
177.8
178.8
178.2
175.8

 

75.4
19181923 19319331938 19431%195319581%3 196819731978 1983198819931998

Lake St. Clair

    

175.8
175.4
175.0
174.5 " ’3
174.2

173.8
1918 1923192819331938 1943194819531958 1%31968197319781983 198819931998

Lake Erie

 

.1918 1923192819331938 1943194819531958 19631%8197319781983 198819931998

 

Figure 2.1. The water levels on the Great Lakes from 1918 to 1998.
Source: US. Army Corps of Engineers, Detroit District, 2002.

25

 

Table 2.1 reports water levels on the Great Lakes for 2000, 2001 and 2002. The

table also shows the historical maximum high water level and minimum low water levels

from 1918 to 2001 for the Great Lakes. Water levels during 2001 and 2002 were not as

low as the historical low water level but were lower than the average levels from 1918 to

2001. Great Lakes water levels were signiﬁcantly higher in 2002 than in 2001. Lake

Michigan-Huron and Lake St. Clair are eight or to ten inches above the levels in 2001,

but remained ten and three inches, respectively, below their long-term averages (US.

Army Corps of Engineers, Detroit District, 2002). Even though Great Lakes water levels

increased in 2002, they were still much lower than the average levels between 1918 and

2001 (Table 2.1).

   

Michigan-

Table 2.1. The mean lake levels in September for the ﬁve Great Lakes.

 

Superior Huron St. Clair Erie Ontario

2002 lake level. 601.84 578.89 573.62 570.73 244.39
2001 lake level 601.38 577.53 573.26 570.28 244.52
2000 lake level 601.67 577.85 573.91 571.19 245.43
History maximum" .

Level 603.38 582.35 577.30 573.95 246.78

(Year) (1958) (1986) (1986) (1986) (1945)
History minimumM

Level 600.72 576.44 571.75 568.57 242.49

(Year) (1926) (1964) (1934) (1934) (1934)
Average lake level ” 602.13 579.04 575.09 571.10 244.78

"‘ The lake levels are measured in feet.
" The lake level of history maximum and minimum and average is for period 1918 to 2001.

Source: US. Army Corps of Engineers, Detroit District, 2002

26

The expected Great Lakes water levels in 2003 are shown in Table 2.2. In that
table, the expected water levels (on June 2003) on the Great Lakes, as well aS the period-
of-record average levels for the Great Lakes, are given in inches above (+) or below (-)
Low Water Datum (LWD), which are given on International Great Lakes Datum, 1985
(IGLD 1985’). According to the expected water level, the Lake Ontario level on June
2003 will keep the same as the level during the 1990 to 1990. The 2003 water levels in
Lake Erie, Lake St. Clair and Lake Superior will be higher than the water level of IGLD
1985 but still lower than the level during 1900 to 1990. The Lake Michigan water level

on June 2003 is estimated to be almost the same as the Low Water Datum of IGLD 1985.

Table 2.2. The expected water levels in the Great Lakes in 2003.

     

Lakes

 

Lak‘? Lake Erie Lake .5" Michigan/ L31“?
Ontarro Clair Superror
Huron
Period of Record
Average Levels +34 +31 +27 +22 +7
(1900-1990) June 5
Expected Levels
June 5, 2003 +34 +25 +19 -1 +1
Low Water Datum
IGLD 1985 243.3 569.2 572.3 577.5 601.1

Source: US. Army Corps of Engineers, Detroit District, 2003

 

’ IGLD is International Great Lakes Datum. The IGLD 1985 deﬁned by Great Lake Information Network
is ‘Because of movement of the earth’s crust, the ‘datum’ or elevation reference system used to deﬁne
water levels previous within the Great Lakes-St. Lawrence River system must be adjusted every 25 to 35
years’. The current datum was known as the International Great Lakes Datum, 1955 (IGLD 1955). These
brieﬂy explain the development and impacts of the revision to this datum, known as the International
Great Lakes Datum, 1985 (IGLD 1985). The date, 1985, is the central year of the period 1982-1988
during which water information was collected for preparing the datum revision.

27

The US. Army Corps of Engineers is studying the potential effects over the next
50 years of a wider range of Lake Michigan levels than those experienced in the past 30
years (Planning & Zoning Center, Inc. et al., 2002). The range of ﬂuctuation they are
anticipating is 9 feet, which would extend from a low level of 574.3’ (still water without
storm-induced draw down, IGLD, 1985) to a high level of 583.4’. During the last 30
years the range of ﬂuctuation has been 6.27 feet. The lowest monthly average lake level
recorded during the last 30 years (576.05 IGLD 1985 International Great Lakes Datum)
occurred in March 1964. The highest monthly average lake level recorded (582.32 IGLD
1985) occurred in October 1986. The all-time-high recorded water level of Lake

Michigan (582.64 IGLD 1985) occurred in June 1886.

The Review of Lake Michigan Potential Damages Study

The Lake Michigan Potential Damages Study (LMPDS) is an approximately 10-
year initiative aimed at assessing the potential damages (e.g., erosion, water level
ﬂuctuations) along all of the US. Great Lakes shorelines. In that study, Lake Michigan
was chosen to be the ﬁrst study area, since it had severe erosion problems and was the
most damaged lake during the previous high water periods in the 19705 and 19805

(Planning & Zoning, Inc. et al., 2002).

The LMPDS incorporates studies of the inﬂuence of water level changes on the
boating industry, lake use, community and different economic sectors. Table 2.3
summarized the main focus of those different studies conducted as part of the LMPDS

during 2000 and 2001.

28

Table 2.3. Lake Michigan Potential Damages Study projects in 2000 and 2001.

 

 

 

 

 

 

Year Conducted by Major Finding
2000 Christian J. Stewart Study region: Grand River at Grand Haven, MI;
Consulting Lake Macatawa at Holland, MI; Kalamazoo Lake
and River at Saugatuck/Douglas, MI; and the
Sheboygan River in Sheboygan, WI.
Togic: Inventory and mapping of shoreline
protection and boating structures in drowned river
mouth areas of Lake Michigan by using Arc View
G18.
2000 NTH/WTA Joint Study region: Ottawa and Allegan counties in
Venture Michigan; Ozaukee, Sheboygan and Manitowoc
counties in Wisconsin
T ogic: To develop an assessment of the potential
damage to typical shoreline structures due to the
changes of lake levels.
2000 W.F. BAIRD & Study region: Lake Michigan harbors especially on
Associates Ltd. Saugatuck county, MI and Manitowoc, WI.
T ogie: To assess the potential impacts of different
(high and low) water level scenarios (as deﬁned by
GLERL, 2000) on the maintenance/repair costs for
the harbor facilities
2001 Planning & Zoning Survgy: Marinas, boat dealers and charter boats
Center, Dep. of Park, , , _ .
Recreation & Tourism M: Economic Impact of Lake MIchIgan water
Resources of MSU & levels on recreational boating and charter ﬁshing in
EPIC-MRA ﬁve counties in Michigan and Wisconsin
2001 University Of Wisconsin Investigation region: Three Wisconsin counties
Sea Grant Institute/Lake . , .
. T0216. Update land use Inventory and cover change
Informatron Computer .
. . . analysrs
Graphics Facrlrty-
University of

 

Wisconsin-Madison

29

 

Table 2.3. Lake Michigan Potential Damages Study projects in 2000 and 2001 - cont’d.

   

 

 

Year Conducted by Major Finding

 

2002 Planning & Zoning Study Region: Review the current coastal land use
Center, Inc. and Wade- conditions in Michigan, Wisconsin, Illinois and
Trim Indiana.

Togic: To evaluate whether other land use
management measures currently exist in the ﬁve
prototype counties that should be included in order
to reduce the damages and estimate the loss to
structures and property because of water level
changes

 

2002 Planning & Zoning Study Region: Michigan, Wisconsin, Indiana and
Center, Dep. of Park, Illinois.
Recreation & Tourism
Resources of MSU &
EPIC-MRA

Surv_e_y: Marinas, dealership, and recreational
Boaters

T ogic: To estimate the economic impacts of Lake
Michigan low water on recreation boating industry
including the ﬁnancial impacts and revenues on
marinas and dealerships and also the inﬂuence on
recreational boaters.

 

One of the projects involves an assessment of the economic impacts of Lake
Michigan water level on the recreation boating and charter ﬁshing industries. The
principal investigators are the Planning & Zoning Center, Inc., department of Park,
Recreation and Tourism Resources of Michigan State University and EPIC-MRA. As
part of the project marinas, dealerships, and charter boats were surveyed in ﬁve counties
in Michigan and Wisconsin in 2000. The impacts of low water levels on marinas,
dealerships, charter boats, and recreational boaters in four Lake Michigan states were

estimated and identiﬁed in the 2001 study.

30

Impacts on magmas

Both studies indicates that marinas suffer measurable losses such as unusable
slips, dredging cost and/or reduced revenues because of low water levels. In the 2000
studies conducted in ﬁve Lake Michigan counties in Michigan and Wisconsin, Mahoney
et al. (2001) indicated that low water levels had a negative impact on recreational boating
especially a ﬁnancial effect on Michigan marinas such as a ﬁnancial loss to marinas in
the ﬁve study areas of between $2 to $4 million, which included the total lost slip
revenue, increased dredging costs, and damage to dock and other structures (Planning &
Zoning, Inc. et al., 2001). It was estimated that an additional 12” decrease (below 2000
water levels) in water levels would have a catastrophic impact on many of the marinas
serving Lake Michigan boaters and this in turn would reduce boat sales and negatively
impact local business that sell products and services to boaters including the local tourism

industry (Planning & Zoning, Inc. et al., 2002, p.59).

The LMPDS (2002) estimates direct and indirect losses at marinas2 located in 34
Lake Michigan coastal counties were around $12 million in 2001. If Lake Michigan
water level were to drop an additional 12” below 2001 water level, the ﬁnancial impact to
those marinas is estimated to be around $21 million. If Lake Michigan water level were
to drop an additional 18” below 2001 water level, the estimated ﬁnancial impact would
be approximately $28 million. Low water level related damage to boats is substantial,

estimated at about $26 million in 2001.

 

2 The estimate of the direct and indirect losses is based on the 297 marinas identiﬁed by that study.

31

Impacts on public boat launch ramps

As part of the 2002 LMPDS, the Planning & Zoning Center (PZC) visited and
assessed all known public boat launch sites located on Lake Michigan in four states. The
assessment included 97 boat launch ramps. The assessment determined that ﬂuctuations
in water levels signiﬁcantly impacted on the accessibility of these facilities. In 2001, the
97 boat launch ramps with access to Lake Michigan remained open when water levels
were at 577.6’. Based on measurements it was determined that if water levels were to
reach a potential extreme low (574.4’ IGLD 1998) only 33 ramps would be accessible to
Lake Michigan. Conversely, if levels increased to the extreme high water level (583.4’
IGLD 1998), only 17 of the 97 boat launch sites that access Lake Michigan would be
open. At extreme high water levels, ﬂooding could restrict access to some ramps and
marina slips and there might be debris in the water that would represent a risk to boaters.
On the other hand, with extreme low water levels, many boat launch ramps and marina
slips would not have sufﬁcient water depth to launch and navigate boats to gain access to

deeper waters from protected harbors.

According to the report, communities on the west side of Lake Michigan are
better to deal with low water level than the communities on the east side because of the
better design of the boat ramps in Wisconsin. For example, there were plans for a new,
modern ramp to be built in 2001 in Muskegon State Park. Besides ramps with the
ﬂoating docks used in Wisconsin could be more usable with lower water level than the

ramps with the permanent docks.

 

3 This number is calculated by dividing the total number of 370,000 possible trailer launched boats into the
number of open boat launch ramps. The 370,000 trailer boats are the registered watercraft in the Lake
Michigan coastal counties in four states and are smaller than 39’ in length.

32

Thus extreme high and low Lake Michigan water levels would make some boat
launches unusable and bring the pressure for boaters on boat launches. A decreased
number of the available launch Sites would require recreational boaters to travel greater
distances (higher cost) to ﬁnd another open launch site or private marinas to launch their
boats. As a result, available boat launches and the waters they access will then become
more crowded and the quality of recreational experiences may be diminished and
pressure on resources may increase (Table 2.4). Overall, high and low water levels are

expected to increase the cost of boating.

Table 2.4. Number of boats vying for each ramp in different water level".

   

Estimated number for Lake Estimated number of boats

 

Michigan Launch Ramp vying for each ramp
Water level Open
2001 Water Level 97.2 3,824

(577.6’ IGLD 1985)

Potential Extreme Low
Water Level 33 1 1,240

(574.3 ’ IGLD 1998)

Potential Extreme High
Water Level 17 21,818

(583.4’ IGLD 1998)

‘1 The estimated number is based on the boat launch sites located in 34 Lake Michigan coastal counties
and an estimate of 370,909 trailer launches boats owners reported using on Lake Michigan in the 34
shoreline counties and calculations based on inspections and measurements taken at boat launch ramps.

*2 This is the actual number according to the investigation result of Planning & Zoning Center.

Sources: Planning & Zoning Center, Inc. (2002)

In addition to making boat launches inaccessible, low and high lake levels will
cause structural damage. It is estimated that there will be a loss of structure values for the
seawalls/bulkheads in a 50-year period ranges from 10% to 75% for an extreme low lake

level and from 75% to 90% for an extreme high lake level (Planning & Zoning, Inc. et

33

al., 2001). And the range Of estimated construction costs for water level changes on Lake
Michigan harbor structures will be from $4,500 to $1,900 per linear foot according to the

design of the structure.
Impacts on recreational boaters

In the 2000 LMPDS, EPIC-MRA found that 29% of recreational boaters, who
responded to their telephone survey, cited low water levels as the most important problem
confronting bOaters. More than 40% of respondents had used their boats less over the
two or three years because of low water levels. Boaters were also asked how further
drops in water levels would impact their boating. If Lake Michigan water levels were to
drop about one, two or three feet, 33%, 86% or 90% of the respondents indicated that an
additional drop of 12” would cause them to ﬁnd alternative launch sites, which would

have a major negative ﬁnancial impact related to a decrease in boating.

The 2001 Lake Michigan Potential Damages Study telephone survey found that
9% of the boaters cited low water as the reason they were boating less in 2001 than
during the previous few years ago. The study concluded that low water levels could
substantially reduce boating activity because of low water and perceptions of low water
related problems. They recommended that the boating industry should carefully consider
the perception of recreation boaters about low or high water levels (Planning & Zoning,
Inc. et al., 2002, p60). While two LMPDS indicated low water levels have a negative
impact on boating, Mahoney (2002) suggested that we still need more information about
what types of boater are inﬂuenced by low water and what are their potential boating

behavioral changes because of low water. Mahoney (2002) went on to suggest that a

34

ﬁrture effort to understand better Great Lakes boater behavior and the likely response

changes in boating opportunities.

2.3 Previous Recreational Boating Studies

Michigan State University is a nationally recognized leader in the conduct of
recreational boating studies. In addition to the low water studies, ten major statewide
recreational boating studies have been conducted in 1965, 1968, 1971, 1974, 1977, 1980,
1986, 1994, 1998 and 2001. These studies are described in Table 2.5. In general these
surveys collected and provided information on (1) the characteristics of the behaviors of
boat owners, (2) boater behaviors and travel patterns, and (3) economic impacts of

boating.

Spatial Patterns of Recreational Boating in Michigan

A number of these studies provide information relating to the spatial distribution
of recreational boaters and their boating activity. In an earlier study by Chubb and Chubb
(1975) and later studies by Talhelm et al. (1986) and Stynes et al. (1995), all provided a
description of the Spatial patterns of recreational boating. Those studies identiﬁed a
predominate south-north movement of boaters with many of the boaters in southeast
Michigan traveling to the northwest and northeast regions of the State. Also, more boats
owned by boaters who live in the southern part of Michigan are stored in northern

counties of Michigan.

35

Table 2.5. Previous Michigan Boating related studies and literature reviews.

 

Year of

 

 

 

 

 

 

 

study Conducted by Major focus & findings
1965 Michigan Waterways 1. To provide a description of boating spatial
Division patterns
2. The boats/boaters in the southeast counties of
Michigan had highest level of boating use
1968 Department of Park and 1. To make recommendations concerning the
Recreation Resources adequacy of the techniques and methods and to
of Michigan State formulate a general planning process
UUiVCYSit)’, by ChUbb 2. More variables should be investigated in
J .E. Oakwood (Thesis) connection with determination and projection
of demand for boating opportunities and the
signiﬁcance Of out-of state as well as in-state
boaters should be investigated
1971 Recreation Resource 1. To provide a description of boating use
Consultants characteristics and patterns at the time of the
studies
1974 Recreation Resources 1. To provide a description of boating spatial
Consultants, By Chubb, patterns
M- & ChUbb, H 2. Boats registered in southern Michigan counties
comprised the largest share of recreational
boating use
Warner (Thesis) 1. The analysis of recreational boating
expenditures
2. An increase in income, family size, and age of
the craft owner had a positive impact on
recreational boating
1977 Michigan Waterways 1. To provide a description of boating use
Division characteristics and patterns, the boating days
and the total number of private and public slips
for each county in Michigan
1980 Michigan Sea Grant, 1. A comprehensive summary of previous
By Stynes & Safronoff boating research and information collected

   

prior to 1980

2. Fishing is the most popular boating activity

. Southeastern Michigan continues generated the

majority of boat days in Mnchiga

 

36

Table 2.5. Previous Michigan Boating related studies and literature reviews — cont’d.

 

Year

 

 

 

 

 

 

study Conducted by Major focus & ﬁndings
1986 Travel, Tourism and . To provide a description of boating spatial
Recreation Center, patterns
Michigan State . More populated southern Michigan counties
University, by Talhelm, and counties with more boating opportunities
Jordan, & HOICCCk located near population centers experienced
the highest amount of recreational boating use
1994 Department of Park, . To measure patterns of boating activity as a
Recreation and basis for evaluating current and anticipated
Tourism Resources, future needs
Michigan State . The northern LP and the straits areas are net
University, by Stynes, importers of boats
Wu and Mahoney.
Wu (Dissertation) . A system model for estimating recreational
boating use in Michigan counties
. The system models produced estimates of the
number of boats in different types of storage,
the number of boats kept in Michigan counties,
and the number of boat days in destination
counties by boat storage segments
1998 Department of Park, . To provide a description of boating use
Recreation and characteristics and trip expenditures by
Tourism Resources, recreational boaters in Michigan Marinas
Michigan State . There were signiﬁcant differences in both the
University levels and patterns of use and spending by
Lee (Dissertation) storage pattern
2001 Planning & Zoning . To provide a description of boating use

 

center, Department of
Park, Recreation and
Tourism Resources,
Michigan State
University & EPIC-
MRA

characteristics and the inﬂuence of low water
on marinas, boat dealers, recreational boaters
and boat launch facilities

37

 

A considerable variation in the rate of recreational boating participation existed
among the registered watercraft owners in Michigan according to the result of the 1974
boating study (Fiske, 1974). A non-linear relationship was found existing between the
family income of boaters and boating participation. Among boaters’ socio-economic
characteristics, income, family size, occupation, and age were signiﬁcantly correlated
with boating participation in one or more county regions in Michigan. In addition to
variables relating to socio-economic characteristics, the storage location during the
boating season was also found to be signiﬁcantly correlated with boating participation in

certain Michigan counties.

Fiske (1974) also found that boat length as an important factor in public policy
concerning the construction of public boat marinas and other facilities because larger
boats usually require more care and handling equipment than smaller boats. Moreover
Fiske (1974) also concluded that boat type was another signiﬁcant variable relating to the
boating participation. The effect of transportation upon watercraft use was entered as a
dummy variable and exhibited a signiﬁcant inﬂuence on boating participation in
Michigan. According to previous ﬁnding, the boating participation and behavior is
inﬂuenced by boater’s characteristics, watercraft characteristics, boat storage

characteristics, and boat transportation (Fiske, 1974).

38

The variable about surface water acreage of county was identiﬁed having a
positive correlation with boating participation in “bottom thirty’ ’4 origin counties. The
value for this variable consisted of the total surface water area contained in each county
in selected water categories: (1) natural lakes and ponds, (2) natural lakes with a dam, (3)
artiﬁcial lakes, (4) artiﬁcial ponds, (5) hydro-electric reservoirs, (6) small lakes, and (7)

ﬂood control reservoirs.

The 1994 Michigan boating study provided the most current information on
statewide boating use at the county level (Stynes et al., 1995). The researchers used
survey results including where boats were stored and used, boat registration data,
information on seasonal homes and inventory of marina slips to estimate the number and
type of boats stored in different counties during the boating season. This information is
important because the registration data only includes the residence of the owner, but not
where the boat is stored or used during the boating season. The information on boat
storage locations was used to produce the estimates of boating days for individual
counties by applying segrnent-speciﬁc parameters from the survey to the estimating

numbers of watercraft of each type stored in a given county (Stynes et al., 1995, p. E3).

Wu (1995) used the data from the 1994 Michigan boating survey to develop a
system of models to estimate the recreational boating use in Michigan counties. The
models included a classiﬁcation model, boat allocation model, a trip generation model,

and trip distribution model.

 

4 The “bottom thirty” counties in Michigan include Kalkaska, Lake, Osceola, Oscoda, Missaukee, Arenac,
Luce, Alcona, Ontonagon, Otsego, Montrnorency, Baraga, Sanilac, Menominee, Gogebic, Ogemaw,
Tuscola, Alger, Isabella, Lapeer, Clare, Crawford, Schoolcraft, Huron, Gladwin, Presque Isle, Oceana,
Mecosta, Benzie, Iron counties.

39

Four segments were used as the basis for the system of models because boating
use and spatial patterns of use different between boats in different type of storage (Wu,
1995). The four storage segments used in that study were marinas, second homes,
waterfront homes and non-waterfront homes, which will be used in this study also. For
the classiﬁcation model, the independent variables used to classify the boat storage type
included the characteristics of boats and boat owners: (1) length of boat, (2) type of boat,
(3) residence location, (4) ownership of second home, (5) income and (6) age. The type
of boat was grouped into inboards, outboards, pontoon boats, and canoes. The location of
owners’ residence was grouped into seven regions: south-coast region, central coast

region, north-coast region, Upper Peninsula and out-of state.

The 1998 Michigan Boating Survey focused on spending patterns of recreational
boaters. It was estimated that owners of registered boats spent an estimated $635 million
on trips within Michigan in 1998. A typical boater spent $23 a day on day trips and $60 a
day on overnight trips, averaging about $35 per day. Boaters keeping their boats at
marinas spent $76 on boating trips, while at the other extreme boaters storing their boats
at waterfront primary homes spent $20 a day (Lee, 1998, p.108). Wu (1995) found that
boats kept at marinas during the boating season generated more in terms of local
spending and economic impacts than boaters who kept their boats at permanent
residences and trailered to boating locations due in part because bigger boats are
generally kept at marinas and spent more money on marina slip rental fee, storage fee,

maintenance or service fees.

40

2.4 The Binary LOGIT Model

This section provides a description of the binary LOGIT model and how to apply

the binary LOGIT model to three different study objectives.

Many areas of social science research involve dependent variables with two
possible values/outcomes (Wrigley, 1985; Agresti & Barbara, 1986; Cramer, 1991;
Powers & Xie, 2000). A dependent variable with two alternatives is called a binary
choice, usually represented as “Yes” or “No” or coded as “l” and “0”(Stynes & Peterson,
1984; Ben-Akiva & Lennan, 1985). Such choices are common in recreation research,

e. g., Whether to participate in a recreation activity, whether to go ﬁshing, and whether to
Visit a state park. In the case of this study recreational boaters can react to low water

problems by continuing to boat or not boat.

In situations with dependent variables having two possible values or outcomes,
some problems arise in using a standard regression model (linear model) (Cramer, 1991;
Powers & Xie, 2000). The predictions in a linear regression model are unbounded and
are not necessarily restricted to (0,1). Thus the error structure will be heteroscedastic and
non-normal, which will cause inefﬁcient estimates (Smith & Munley, 1978; Stynes &

Peterson, 1984; Cramer, 1991; Gujarati, 1995).

The most widely used binary response models are those using a logit
transformation approach (the logistic function) to resolve those problems caused by using
a linear regression model (Hosmer & Lemeshow, 2000). With binary dependent
variables, the goal is to estimate or predict the probability of success or failure (yes or

no), conditional on a set of independent variables (Powers & Xie, 2000). For those

41

models using logit transformation approach, the larger the value of the index is; the
greater the probability that the event (Yes) in question will occur (Cramer, 1991; Powers
& Xie, 2000). In this study, events include different awareness (or perceptions) toward
water level changes and various reactions (related to three study objectives) to low water
level issues. For example, one event would be that boaters reduced the amount of their
boating days in response to the low water levels, or they changed locations of boating. A
monotonic relationship between the value of the index and the probability of the event
occurring can be assumed. Under these assumptions, the “true” probability function has

the characteristic shape of a cumulative distribution function (CDF).

The two most commonly used CDFS are the normal (PROBIT model) and logistic
(LOGIT model) (Stynes et al., 1984; Cramer, 1991). Since the logistics cumulative
density function can closely approximate that of a normal random variable, there is
usually little difference in the empirical results produced by the two models. Unless there
is theoretical justiﬁcation for preferring the normal to the logistic cumulative distribution
function, the LOGIT model is preferred to the PROBIT model when repeated

observations are available (Judge et al., 1980).

Assumptions about the logistic regression

Logistic regression is an Often used method in part because it enables the
researchers to overcome many of the restrictive assumptions and requirements of OLS
regression. First, instead of assuming a linear relationship between the dependents and
independents variables, logistics regression is capable of handling nonlinear effects.

Second, the dependent variable need not be normally distributed. Third, the dependent

42

variables do not to be homoscedastic for each level of the independent variables and
normally distributed error terms are not assumed. Fourth, logistic regression does not
require interval independent variables. Fifth, logistic regression does not require that the
independents be unbounded. However there are still some other constraints for logistic
regression. For example, the coding for dependent variable should be meaningful such as

l and 0. Also, the error terms are assumed to be independent (independent sampling).

Another important consideration that must be recognized is that of
multicollinearity. To the extent that one independent is a linear function of another
independent, multicollinearity will occur in logistic regression. The greater the
correlation between independent variables, the more the standard errors of the logit
(effect) coefﬁcients will be inﬂated. High standard errors ﬂag possible multicollinearity.
In LOGIT modeling, convergence is usually achieved in 4 or 5 iterations. A general
warning is that if the number of iterations exceeds 10 or 15 this may be a signal of
multicollinearity in the data set. A result may be relatively high estimated standard errors

on the estimation output (Wooldridge, 2000).

The Logistic Function and Logistic Transformation

LOGIT model is based on the logistic function (Equation 2.1) and the logistic

curve is shown in Figure 2.2.

 

y = + 4.1.4...) : T7117; (2.1)

43

Where y = A dependent variable
x = An independent variable

a,b = The estimated coefﬁcients

 

 

 

 

 

 

 

 

Figure 2.2. The logistic function curve.

The graph shows that the value of “y” is bounded asymptotic by approaching 1
and 0 when “x” approaches positive and negative inﬁnity, respectively. In the function
form, “y” is considered as dependent variable and “x” is independent variable. The slope
of the logistic function (equation 2.2) is directly proportional to the product to the product
of the distances to the two asymptotes with b the constant of proportionality and can be
used to measure how far about the two distributions are. The slope of the logistic curve is

governed by “b”, and at the point of inﬂexion it equals “b/4”.

44

dX/dy : by(1 — y ) (2.2)

Equation 2.1 represents the simple logistic function with one independent variable
and two parameters. Frequently often more than one independent variable associated
with the dependent variable is present. The multivariate logistic function is written as

Equation 2.3.

1 a+blxl+b2x2+uu+bnxn

= 6
)Or y 1 + ea-l-bl.xl+b2x2+...+bnx’l (203)

 

 

y : 1+ e_(a+blxl+b2x2+...+bnxn

Where y = A dependent variable
XI , x2 , 36’3"" xn = Independent variables

a, b1, b;, b 3,. . .bn, = The estimated coefﬁcients

LOGIT Models

The logistic function is restricted to the (0,1) interval, which allows it to be used
as a probability function and the logistic regression model with using the logit
transformation can be converted to a convenient linear form. If we assume the
probability (P) of success (yes, Y=1) is a logistic function of the set of independent

variables (Equation 2.4), the probability of failure (no, Y=0) will be “l-P (Equation 2.5).”

45

1 ea+b1xl+b2x2+...+bnxn

’(a+b1xI+b2XZ+'”+bnxn) _ 1+ea+b1xl+b2x2+m+bnxn (2’4)

 

 

P:
1+6

1
1 + ea+b1x1+b2x2+m+bnxn (2.5)

 

1—P=

Instead of ﬁtting a model for P, we could use the Odds of a “success (Yes)”
outcome. The odds are deﬁned as the probability of “success (Yes=1)” divided by the
probability of “failure (No=0)”. The Odds can be a value between 0 to 00. The odds are

shown in Equation 2.6.

Odds = ::EII::0S; : -——1 f) : ea+b1x1+b2X2+-~+bnxn (2.6)
17

Taking natural logarithms Of both side of Equation 2.5 yields a linear equation
(Equation 2.7). These LOGITS are the log odds of success (yes) vs. failure (no). In
addition, they are unbounded and linearly related to the independent variables. Thus by
using a logistic ﬁmction, the probabilities of “Y” can be restricted to (0,1) and also

related to the independent variables: X1,. . .,Xn.

log(%_ P): 10g(ea+b.x1+b2x2+~+b.xn)= a + :1), x.- (2.7)

46

Estimation from Individual Data: Maximum Likelihood Method

Two basic methods can be used to estimate the LOGIT model. The maximum
likelihood method can be used with either individual observation or group frequencies.
The weighted least square method uses group frequencies (Stynes et al., 1984). The
maximum likelihood method is preferable when, in situations like this study, probability
models are estimated from survey data, which provide large samples of independent
observations with a wide range of variation of the regressor variables (Berkson, 1953).
One advantage of maximum likelihood method is that the parameter estimates can be
almost any analytical speciﬁcation of the probability function, and the result of
estimations are consistent and asymptotically efﬁcient, together with ready estimates of
their asymptotic covariance matrix (Cramer, 1990). Because the likelihood of the
parameters given, the data are deﬁned to be equal to the probability of the data given the
parameters, the maximum likelihood estimation can ﬁnd the parameter value(s) that

make(s) the observed data most likely (Berkson, 1953).

Summary

This section reviewed the theory and application of the LOGIT model, which will
be used a statistical tool to develop different models for three study objectives. Then the
modeling speciﬁcations and model evaluation tools for ﬁve binary LOGIT models of

three objectives will be discussed in the last section of chapter three.

47

CHAPTER 3

METHODS

Two types of data are used to develop ﬁve LOGIT models for three study
Objectives. For study objective 1, the date from the Lake Michigan Potential Damages
Study telephone survey is used to develop two LOGIT models: (1) “Awareness of Lake
Michigan Water Level Model” and (2) “Perceptions of the Drop in Water Levels Model”.
For study objective 2 and 3, the data from the Lake Michigan Potential Damages Study
mail survey is used for develop another three models: (1) “Concerns about Low Water
Level Inﬂuences Boaters Model”, (2) “Low Water Levels Change Boating Locations”
model, and (3) “Low Water Levels Reduce Boating Amount” model. Both survey are

originally designed for the Lake Michigan Potential Damages Study.

This chapter provides a description of (l) the survey methods employed to collect
the data that is used to both test and formulate the binary LOGIT models, (2) the
formation of the variables that comprise the model, and (3) the speciﬁcation process for
ﬁve LOGIT models. The description of the survey methods includes the study
population, sampling procedures, data collection, survey instruments and response rate.
The model speciﬁcation section will have descriptions about the deﬁnition of dependent
variables, the selection of independent variables and model function forms and evaluation

tools for model performance.

48

3.1 Study Population

The study population consisted of recreational boaters, who owned
recreational/pleasure watercraft longer than 17 feet that were registered to operate in
2001 in Michigan, Wisconsin, Indiana and Illinois. However, the sampling was weighted
to obtain a greater percentage of recreational boaters, who reside or registered their
watercraft in 34 Lake Michigan coastal counties (Figure 3.1). Some states like Michigan
register boats where their owners permanently reside, others where their boats are stored
during the boating season. The 34 Lake Michigan coastal counties are Michigan
counties: Menominee, Delta, Schoolcraft, Mackinac, Cheboygan, Emmet, Charlevoix,
Antrim, Grand Traverse, Ieelanau, Benzie, Manistee, Mason, Oceana, Muskegon,
Ottawa, Allegan, Van Buren, and Berrien; Wisconsin counties: Marinette, Oconto,
Brown, Door, Manitowoc, Kewaunee, Sheboygan, Ozaukee, Milwaukee, Racine, and
Kenosha; Indiana counties: La Porte, Porter and Lake, and Illinois counties: Lake and

Cook.

The 2001 registration lists for the four states were obtained with the cooperation
of the US. Army Corps of Engineers. Information obtained in the four registration lists
differed (Table 3.1). All four states provided information about boat length, type of
watercraft], mode of powerz, boat use3 and the location of the boat owners’ residences.
Wisconsin and Indiana also provide information on where the watercraft is kept during

the boating season. Recreational watercraft registrations are renewed every three years.

 

’ “Type of watercraft” could be open boat, cabin boat, sail boat, houseboat, pontoon or canoe.
2 “Mode of power” could be inboard, outboard, in/outboard motorboat, or non-motorized boat.
3 “Boat use” could be for recreational use, commercial use, livery use, ﬁshing use or government use.

49

Copies of the watercraft certiﬁcate application/title form for four states are included in

Appendix A, B, C, and D.

 

Schoolcraft Chippewa

.50 '5»

Lake
Michigan

Q

o
Antrim

La
Pane

 

Indiana

100 O 100 200 Kilometers

 

 

 

Figure 3.1. Lake Michigan counties in the four states.

50

Table 3.1. Information contained in the watercraft registration lists in the four Lake

 

My;
Owner’s name
Address

City

State

Zip

County code
Make

Length (feet)
Length (inches)
Hull ID #

Hull code
Propulsion code
Use code

Type code

Fuel code

Filler

Michigan states.

Wisconsin

Registration ID
Registration status
Registration type
Expiry year
Length (feet)
Length (inches)
County kept

Hull ID

Hull ID veriﬁcation
Hull make

Make name
Model name
Model year

Boat type

Engine type

Fuel type
Propulsion type
Use type

Purchase date
Information disclosure
Last name

First name

Initial

Address

City

State

County

Zip

   

Indiana
Registration #
HULL ID

HIN veriﬁcation
Owners name
Owner type
Commercial status
Primary owners name
Primary owners
Address

Primary owners
AddressZ

Primary owner city
Primary owner state
Primary zip

Primary extended zip
Additional owner
Owner county
Storage county
Storage address
Storage city

Storage state
Storage Zip

Storage extended zip
Storage township
Class

Purchase date

Make

Make2

Model

Model year

Length

Propulsion

Fuel type

Hull type

Use

Type

Name

Address

City

State

Zip code
County code
Boat type
Propulsion code
Boat length
Manufacturer
Hull ID number
Registration #

51

3.2 Sampling Procedure

The primary Objective of the sampling method for both 2001 Lake Michigan
Potential Damages Study mail and telephone surveys that gathered data for this study was
to identify a high proportion of boats that could potentially be Operated on Lake
Michigan. No information on where boats are used is contained in the registration
databases and a range of different types and sizes of boats that can be operated on the
Great Lakes. Another problem is that not all watercraft contained on the registration lists
are currently registered. For example, Michigan’s list contains boats whose registrations
have been expired for up to three years. These boats cannot legally be operated and
therefore were not included in the sampling procedure. Three sampling criteria were
established to sample a higher proportion of currently registered recreational watercraft
that were likely to be operated on Lake Michigan. First, they must be registered currently
and able to operate legally during the 2001 boating season. Secondly, they must be
recreational or pleasure watercraft, not commercial or government owned watercraft.
Thirdly, the size of a watercraft should be longer than 17 feet, capable of being operated

on Lake Michigan, stored or registered in counties near Lake Michigan.

Step 1 of the Sampling Procedure for the 2001 Lake Michigan Potential Damages

Study Mail Survey

Registration procedures and databases differ across states. States like Michigan
maintain boats with expired registrations in their databases for more than two years in

some cases. These boats cannot be legally operated, but the procedure saves the state’s

52

money if the boat owner decides to renew the registration. So, the ﬁrst step in sampling
was to identify watercraft with valid 2001 registrations (Table 3.2). Michigan has a
three-year registration so watercraft whose registrations would expire in 2002, 2003 and
2004 were included in the sample. Since the expiration month in Michigan is April,
watercraft whose registrations expired in 2001 (April) were not included in the sample

since they were not eligible to be operated at the time (June, 2001) the sample was drawn.

The Wisconsin registration database included watercraft with valid registration in
2002, 2003 and 2004. Indiana registration database only included the watercraft that
could legally be operated in 2001. Since the expiration month of the watercraft
registration in Illinois is August, all watercraft with 2001 registration list were included in
the sample because they could be operated for most of the boating season even if their

registrations were not renewed.

Table 3.2. Number of the watercraft with the valid 2001 registrations including their
expiration year.

     

 

Michigan Wisconsin Ma _Il_lipo_is
N % N % N % N %
2001 - - - - - - 39,336 9%
2002 286,628 35% 44,712 7% - - 130,921 30%
2003 266,305 32% 277,734 46% - - 133,820 31%
2004 21232—7 33% 281264 47_°/q ; ; m Eff/2

 

Total 825,260 100% 603,710 100% 245,371 100% 433,756 100%
m

53

Step 2 of the Sampling Procedure for the 2001 Lake Michigan Potential Damages

Study Mail Survey

The second sampling requirement was that the watercraft had to be recreational
and pleasure watercraft but not commercial, livery, dealership or government watercraft.
Ninety-ﬁve percent of the watercraft on the registration databases for the four states was
registered as recreational/pleasure watercraft. Of 825,620 watercraft with valid 2001
registration in Michigan database, 97% were used for recreational or pleasure purpose.

Table 3.3 shows the “use type” for the 2001 registered watercraft.

Table 3.3. Type of use of the valid 2001 registered watercraft.

 

 

 

Michigan Wisconsin In_diﬂ mi_nO_is_

N % N % N % N %
Pleasure 801,681 97.1% 587,713 97.5% 243,925 99.0% 424,851 97.9%
Commercial 4,392 0.5% - - - - - -
Boat dealer 3,080 0.4% 900 0.1% - - 862 0.2%
Livery 14,790 1.8% - - - - - -
Passenger 696 0. 1% - - - - - -
Rental - - 14,280 2.4% - - 2,488 0.7%
Other A 0.1% 3 ; 1&5 my. ;5_5§ 1240
Total 825,260 100% 602,893. 100% 245,371 100% 433,756 100%

   

"' In Wisconsin registration list, there were 817 watercraft without ’use type’ information.

The sampling procedure resulted in the identiﬁcation of the 2,058,170 2001
registered recreational/pleasure watercraft (Table 3.4). Michigan watercraft comprised
39% of these watercraft followed by Wisconsin (29%), Illinois (21%), and Indiana

(12%).

54

   

 

Table 3.4. Number of the 2001 registered watercraft and recreational/pleasure watercraft.

 

 

Michigan Wisconsin Indiana Illinois Total
Number “watercraft "S‘ed 1,170,517 603,710 533,924 416,422 2,724,573
In regrstratron database
Number Of’he val‘d 200’ 825,260 603,710 245,371 433,756 2,108,097
regrstered watercraft
Number of the valid 2001
registered recreational 801,681 587,713 243,925 424,851 2,058,170

(pleasure) watercraft

 

Table 3.5 shows how the sample (size) was allocated across states. A total of
2,058,170 recreational/pleasure watercraft in the four states was eligible for sampling. Of
these, 3,200 were selected to comprise the sample for the 2001 Recreational Boating Mail
Survey. Another sample of 1,870 recreational watercraft owners was interviewed for the

telephone survey.

Table 3.5. Number of the valid 2001 registered recreational/pleasure watercraft to be

   

 

2001 registered Pimen‘ag" 0f 2901 Sample size
t' l t ﬁ registered recreational
Michigan 801 ,68 1 39% 1,249
Wisconsin 587,713 29% 896
Indiana 243,925 12% 396
Illinois 424,851 21% 6_5_9
Total 2,058,170 100% 3,200

55

Step 3 of the Sampling Procedure for the 2001 Lake Michigan Potential Damages

Study Mail Survey

Since the purpose of the study was to determine the impacts of low Lake
Michigan water levels on boating, the sampling procedure was designed to select Lake
Michigan boaters. The third step was the identiﬁcation of the valid 2001 registered
recreational/pleasure watercraft, which were most likely to be operated in Lake
Michigan. The assumption was that boats registered or stored closer to Lake Michigan
are more likely to be operated on Lake Michigan than boats registered further away. The
complication is that many boat owners trailer their boats to Lake Michigan from inland
counties and many owners, who live in inland counties, store their boats on Lake
Michigan during the boating seasons. Watercraft less than 40’ in length are frequently

trailered and launched on Lake Michigan.

To maximize the number of registered recreational/pleasure watercraft that were
operated on Lake Michigan, the sampling proportion was stratiﬁed to allow for the
selection of the higher proportion of watercraft registered or stored in Lake Michigan
counties or Lake Michigan adjacent counties and nearby counties. Again, in Michigan
and Illinois, the registration data only included the address of the owner. Wisconsin and

Indiana registration data included the “during the boating season storage county” address.

Three different sampling zones were established for each state. Zone 1 included
all counties that bordered Lake Michigan. Zone 2 consisted of all counties immediately
adjacent to Lake Michigan counties. The remaining counties in each of the states

comprised Zone 3 (Figures 3.2, 3.3, 3.4, and 3.5).

56

 

 

     

Legend
1:1 Lake Michigan
coastal counties

Countles adjacent to
Lake Michigan coastal
counﬂes
The remaining
counties

 

 

 

 

 

Figure 3.2. Sampling proportions for the three sampling zones in Michigan.

57

Legend
1:1 Lake Michigan
coastal counties

Counties adjacent to
Lake Michigan coastal
counties
The remaining
counﬂes

   
  
  
  
     
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

{Milwauxee
. * .- MILWAUKEE

 

 

Figure 3.3. Sampling proportions for the three sampling zones in Wisconsin.

58

     
   
 
 
 

HENRY WAYNE
VERMILLION

9’

   

RUSH

OHIO

swrrzar-
LAND
Legend

FLOYD El Lake Michigan
coastal counties

Counties adjacent to

VANDER- Lake Michigan coastal
3011511 counties
The remaining
counﬂes

Figure 3.4. Sampling proportions for the three sampling zones in Indiana.

59

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

   

 

 

 

Legend

Lake
E Michigan
coastal
counnes
I Counties {dunno FERRY w Am
adjacentto Lake ”mﬂm-
Michigan coastal ‘ WWW SAUN

counties ‘
[:j The remaining ”m" 1111mm
counhes
MWDUI

Figure 3.5. Sampling proportions for the three sampling zones in Illinois.

 

 

 

 

 

 

 

 

 

 

60

In Michigan, half of the valid 2001 registered recreational/pleasure watercraft
were registered to persons living in 34 Lake Michigan counties (Zone 1); 25% were
registered in Zone 2; and 25% were owned by persons living in Zone 3. The number of
the valid 2001 registered recreational/pleasure watercraft sampled in each zone for each

state is reported in Table 3.6.

Table 3.6. Number of the valid 2001 registered recreational/pleasure watercraft sampled
by county zone and state for the 2001 Lake Michigan Potential Damages Study

       

 

mail survey.

Zone 11 Zone 22 Zone 33 Total Percentage
Michigan 665 292 292 1,249 39%
Wisconsin 468 214 214 896 28%
Indiana 218 89 89 396 12%
Illinois 349 155 155 659 M
Total l,7_09 159 25_O 9,299 100%
Percentage 53% 23% 23% 100% -

 

1 Zone 1: Lake Michigan coastal counties.
2 Zone 2: All counties adjacent to Lake Michigan coastal counties.
3 Zone 3: The remaining counties.

Step 4 of the Sampling Procedure for the 2001 Lake Michigan Potential Damages

Study Mail Survey

The fourth step of the sampling procedure was to stratify the valid 2001 registered
recreational/pleasure watercraft in Zone 1, 2, and 3 by boat size. Watercraft under 17’ in
length were not sampled because they are much less likely to be operated on the Great
Lakes. The three (boat) size strata are: 17’ to 25’, 26’ to 39’ and over 40’. Table 3.7

shows the results of the stratiﬁcation sampling including the number of different size

61

boats in different zones in each of the four states that comprised the ﬁnal sample.
Watercraft 17’ to 25’ in length are often trailered to launch sites compared to watercraft
bigger than 40’, which are normally kept at Great Lakes marinas or waterfront homes.
Watercraft bigger than 40’ were sampled at a higher proportion to identify a greater

number of watercraft that were operated on Lake Michigan.

Table 3.7. Number of the valid 2001 registered recreational/pleasure watercraft of
different sizes in different zones in each of the four states for the 2001 Lake

 

 

 

 

 

 

State Boat Size Zone 1 Zone 2 Zone 3 Total

Michigan 17’ to 25’ 511 266 256

26’ to 39’ 67 23 32

40’ + 131 3 5

Total 665 292 292 1,249
Wisconsin 17’ to 25’ 378 204 204

26’ to 39’ 45 9 8

40’ + £15 I 2

Total 468 214 214 896
Indiana 17’ to 25’ 150 83 79

26’ to 39’ 19 2 3

40’ + 4_9_ 5 Z

Total 218 89 89 396
Illinois 17’ to 25’ 247 139 145

26’ to 39’ 45 13 8

40’ + 4_7_ 3 Z

Total 349 155 155 659

62

Sampling Procedure for the 2001 Lake Michigan Potential Damages Study

Telephone Survey

A similar sampling procedure was used for the 2001 Lake Michigan Potential
Damages Study telephone survey conducted by EPIC MRA4. The only difference was
that larger sized (over 40’) watercraft were not over-sampled. The total number of the
valid 2001 registered recreational/pleasure watercraft owners was 1,870 boat owners who
were interviewed through the telephone by EPIC from December 2001 to January 2002.
For the telephone survey, the results of interviewed registered watercraft owners located

in each zone in each state are reported in Tables 3.8 and Table 3.9.

Table 3.8. Number of the valid 2001 registered recreational/pleasure watercraft
interviewed by the 2001 Lake Michigan Potential Damages Study telephone
survey by county zone and state.

 
 

 
 

 
 

     

 

Zone 1 ' Zone 22 Zone 33 Total Percentage
Michigan 298 134 174 606 32%
Wisconsin 230 105 1 1 1 446 24%
Indiana 68 69 142 279 15%
Illinois 155 1 17 267 539 291/9
Total 159 @ 6_99 M 100%
Percentage 40% 23% 37% 100% -

W
1 Zone 1: Lake Michigan coastal counties.

2 Zone 2: All counties adjacent to Lake Michigan coastal counties.
3 Zone 3: The remaining counties.

 

" Telephone survey was designed and supervised by EPIC MRA, a Lansing-based survey research firm.

63

Table 3.9. Number of the valid 2001 registered recreational/pleasure watercraft of
different sizes in different zones in each of the four states for the 2001 Lake
Michigan Potential Damages Study telephone survey.

Number of recreational watercraft surveyed

   

 

 

 

 

 

State Boat Size Zone 1 Zone 2 Zone 3 Total
Michigan 17’ to 25’ 266 119 154
26’ to 39’ 3O 13 18
40’ + Z Z 2
Total 298 134 174 606
Wisconsin 17’ to 25’ 203 92 97
26’ to 39’ 25 13 12
40’ + g _2_
Total 230 105 l 1 1 446
Indiana 17’ to 25’ 60 66 141
26’ to 39’ 6 1 1
40’ + Z 2 Q
Total 68 69 142 279
Illinois 17’ to 25’ 141 109 252
26’ to 39’ 12 7 15

 

3.3 Data Collection

On October 23, 2001 a six-page questionnaire was mailed to the owners of
stratiﬁed 3,200 2001 registered recreational/pleasure watercraft that were sampled. A
follow-up questionnaire was sent to 3,200 boaters on November 12, 2001 because no
identiﬁcation mark on the returned questionnaires to identify respondents and non-

respondents was evident.

64

Since some of the boat owners owned more than one watercraft, the registration
number and size of the sampled watercraft were included on the mailing label to better
insure that the owners answered for the sampled watercraft. A postage paid business
reply envelope was also included as part of the mailing. The 2001 Lake Michigan
Potential Damages Study telephone survey interviewed the owners of 1,870 2001

registered recreational/pleasure watercraft between December 2001 to January 2002.

3.4 Survey Instruments

Many of the questions contained on the 2001 Lake Michigan Potential Damages

Study mail and telephone surveys were similar.

The 2001 Lake Michigan Potential Damages Study Mail Survey Instrument

The 2001 Lake Michigan Potential Damages Study mail survey was six pages in
length (Appendix E). The ﬁrst section/page collected information relating to the
ownership of the sampled watercraft and boating done during the 2001 boating season.
The questions included: (1) whether they still owned the watercraft that was sampled,
type and size of the watercraft (conﬁrmation information) and the year they acquired the
watercraft and (2) “when and where they operated the boat during the 2001 boating

season and if they did not put it in the water, why not? ”

The second section/page focused on the number of days the boaters boated during
the 2001 boating season, the location of their boating (states and counties) and the

distribution of those days between the Great Lakes including Lake Michigan and inland

65

lakes and streams. The third section/page of the questionnaire collected information
about the location where the owners stored their boats during the boating season (i.e.,
permanent residence, cottage/second home, marina, or yacht club), type of storage (i.e.,
on-land, in a dry stack facility, or in the water), what type of water bodies that can be
accessed from the storage location (i.e., waterfront site with access to Lake Michigan,
waterfront Site with access to other Great Lakes, access to an inland lake waterfront site,
a river or stream waterfront site), the distance from the storage location to their
permanent residence, whether they changed their storage 2000 location and whether low

water was a reason for the change of storage location.

The next part/section of the questionnaire emphasized the impacts of low water on
boating volume, behaviors and locations. These questions included: (1) trailering and
launching the boat (number of launches and locations of launches) and whether they
encountered any low water related launching problems, (2) the inﬂuence of low water
related problems on the volume of boating days and boating locations including if they
stopped boating or do more or less boating in certain counties, and (3) whether and how
the boaters requested/received any information concerning water levels. The ﬁnal
section/page collected information on the socio-economic characteristics of the registered
boat owners such as the location of their permanent residence, their ages, number of
adults and children in their families, gross annual income, and number of watercraft they

0WD.

66

The 2001 Lake Michigan Potential Damages Study Telephone Survey Instrument

The questionnaire for the 2001 Lake Michigan Potential Damages Study
telephone survey consisted of 70 questions (Appendix F), which paralleled the questions
comprising the 2001 Lake Michigan Potential Damages Study mail survey. In addition,
the telephone survey asked more questions relating to boat launch sites including the
perceived quality of public launch sites. About half of questions inquired concerning
boaters’ knowledge and perceptions about water level changes in Great Lakes. Boaters
were also asked about their willingness to pay to alternative water level problems, which
government agencies should be responsible for addressing the water level problems,
agency performance dealing with water level issues and their possible behaviors if water

levels ﬂuctuated by certain amounts.

3.5 Response Rate and Data Preparation

The response rate for the mail survey was approximately 50%. The ﬁrst mailing
resulted in 951 returned surveys, which is a response rate of 30%. After the second
mailing a total of 1,481 completed surveys had been returned for a response rate of
almost 50% (Table 3.10). In Michigan response rate was 57% (the highest of four states)
followed by Wisconsin (47%). The lowest response rate (39%) was from Illinois

registered boat owners.

There were a host of problems associated with the timing of the survey that
lowered the response rate. First, the initial surveys were mailed on October 23, only 45

days after the September 11 (2001) terrorism attacks. This timing not only slowed the

67

mail but persons were obviously pre-occupied with international issues. As a result, the
data collection process took approximately two months compared to the projected one

month.

Table 3.10. Response rate to the 2001 Lake Michigan Potential Damages Study mail

survey.
@

Completed Original Undelivered Delivered Response

 

survey sample mail sample size1 rate2
Michigan 683 1249 52 1197 57%
Wisconsin 405 896 35 861 47%
Indiana 157 396 10 386 41%
Illinois 2% 6_59 ﬂ 6_01_ Mg
Total 1,481 3,200 155 3,045 49%

 

2 Response rate = completer survey / delivered sample size

The returned mail questionnaires were coded as they were received using a
Microsoft EXCEL program. Response ranges were used to establish valid responses and
to identify possible coding errors and outliers. EPIC administered the 2001 Lake
Michigan Potential Damages Study telephone survey. They provided an electronic ﬁle of

their database to this study.

3.6 Definition and Formation of Variables

This section discusses the deﬁnitions of boating days and the formation of boater
awareness and assessment about the Lake Michigan water level changes and boat storage

segments.

68

Deﬁnition of Boating Days

For the purposes of this study, a boating day is any day, which a boat is taken out
under power or sail. If the boat is Visited (e.g., entertaining, lodging, maintenance) but it
is not taken out under power or sail, this does not count as a boating day. On the 2001
Lake Michigan Potential Damages Study mail survey registered boat owners were asked
about (1) the number of days the boat had been operated up until the time of the survey
and (2) the number of days the boat was likely to be operated in 2001 after they had
completed the survey (Question 4, mail survey in Appendix E). Because of the late date
of the survey, most owners had completed their 2001 boating seasons and had already
stored their boats for the Winter. For the small percentage of boaters who planned to
continue boating after they had completed the survey, their 2001 boating days were

calculated as the sum of the “before and after the survey” boating days.

Formation of Boat Storage Segments

Boaters, who responded to the study provided information on the location Where
their boats were stored during the 2001 boating season including type of storage
(permanent residence, cottage/second home, public marina, commercial marina,
condominium slip, and yacht/boat club) and water access from the storage location (Lake
Michigan, other Great Lakes, inland lakes, and rivers or streams) (Question 6, mail

survey in Appendix E). Storage information is shown in Table 3.11 and Table 3.12

69

 

Percent

 

Permanent residence

Cottage/second home

Public marina

Commercial marina

Owned space in marina/dockominium
Yacht/boat club

Other

Total

52%
17%
12%
7%
3%
3%
9%

1 00%

 

Table 3.12. Types of 2001 boating seasons storage access to different waters.

 

 

 

Percent

 

 

A waterfront site with access to Lake Michigan

A waterfront site with access to other Great Lakes
A waterfront site with access to inland lakes

A waterfront site with access to a river or stream
A non-waterfront site

Total

 

Based on this information, boats were classiﬁed into one of four storage

segments: (1) marinas, (2) second homes, (3) permanent waterfront home, and (4)

 

permanent non-waterfront home. The same segments were utilized by Wu (1995) to

develop recreational boating models. Boats kept at “other” type of storage facilities, and

boats with missing storage information were excluded from the analysis. The four

segments are (Table 3.13):

70

1. Marina segr_nent: boats stored at rented space in public marinas and commercial
marinas, boats kept at owned slips in commercial marinas, and boats kept in

yacht clubs during the boating season.

2. Second home segment: boats kept at cottages or second homes during the

boating season.

3. Permanent waterfront home segment: boats kept at permanent waterfront

residences during the boating season.

4. Permanent non-waterfront home segment: boats kept at non-waterfront

 

permanent residences during the boating season.

Table 3.13. Boat storage segments from the 2001 Lake Michigan Potential Damages
Study mail survey.

 
 

 
 

    

 

Number Percent

 

Marina 328 26%
Second home 225 18%
Permanent waterfront home 525 42%
Permanent non-waterfront home & 94%

 

The telephone survey did not gather information about water access from the

storage location (Lake Michigan, other Great Lakes, inland lakes, and rivers or streams).
As a result three, not four, storage segments were formed. The three storage segments

are shown in Table 3.14.

71

Table 3.14. Boat storage segments from the 2001 Lake Michigan Potential Damages
Study telephone survey.

Number

       

Percent

 

Marina 1 84 32%
Cottage/second home 52 8%
Permanent home _3_7_9 60%

Total 615 100%

 

Recreational Boater Awareness and Assessment of Changes in Lake Michigan

Water Levels

Objective 1 of this study is to identify how recreational boaters perceive the
extent of changes in Lake Michigan water levels. The telephone survey included a
question that asked boaters ﬁrst whether they were aware of Lake Michigan water levels,
and if so, they were asked about their perceptions regarding the degree of Lake Michigan
water level changes during the previous three years (Question 11 and 57, telephone
survey in Appendix F). They were asked to choose from ﬁve possible responses about
the Lake Michigan water level changes: (1) water levels had dropped a lot, (2) water
levels had dropped just a little, (3) water levels had remained the same, (4) water levels
had increased a little or (5) did not know (Table 3.15). Thirteen boaters, who believed

that the Lake Michigan had increased a little, were excluded from the analysis.

72

 

Number Percent

 

Lake Michigan water levels had dropped a lot 870 52%
Lake Michigan water levels had dropped just a little 182 11%
Lake Michigan water levels had remained about the same 125 8%
Lake Michigan Water levels had increased a little 13 1%
Did not know about the Lake Michigan water level changes @ M
Total 1,659 100%

 

The four responses were grouped into two categories according to their awareness
about the Lake Michigan water level changes: (1) The recreational boaters were aware of
the water level changes (e.g., water level had dropped a lot, a little, or remained the same)
and (2) The recreational boaters did not know about the Lake Michigan water level

changes (Table 3.16).

Table 3.16. Recreational boaters’ awareness about Lake Michigan water level changes.

     

Number Percent

 

Recreatronal boaters were aware of the Lake MIchIgan 1,177 72%
water level changes
Recreational boaters were not aware of the Lake 0

. . 4_69 28 /o
MIchIgan water level changes
Total 1,646 100%

Furthermore, for the boaters who were aware of the water level changes, they
were regrouped to two other categories according to their different perceptions about the
degree of the Lake Michigan water level changes: (1) Lake Michigan water levels had

dropped “a lot”, and (2) Lake Michigan water levels had remained the same or dropped

73

just “a little” (Table 3.17). Awareness of Lake Michigan water level changes and
perceptions of the extent of water levels changes were two dependent variables in two

binary LOGIT models of objective 1.

Table 3.17. Recreational boaters’ perceptions about the degree of the Lake Michigan
water level changes.

Percent

       

Number

 

The Lake Michigan water levels had dropped a lot 870 74%
The Lake Michigan water level had remained the same

. . all 26%
or just dropped a lrttle
Total 1,177 100%

 

3.7 Statistical Analyses and Speciﬁcation of the Binary LOGIT Models

Figure 3.6 shows the relationship between each of the objectives and the ﬁve
binary LOGIT models and also describes the statistical analyses performed and the binary
LOGIT models formulated to achieve each of the three study objectives. It also identiﬁes
the primary research question relating to each of the objectives and the source of the data
used to formulate the binary LOGIT models and on which various statistical analyses are
preformed. Frequencies and t-tests/chi-square tests are performed as a precursor to the

binary LOGIT models.

The sequence of steps employed to formulate and test the binary LOGIT models
are graphically presented in Figure 3.6. The binary LOGIT model results in Chapter four

will be presented in the same sequence.

74

 

0.0.3:. ...000 .02.... 2s» .U
00.00..0> 20000022. 000
0.00..0> 2220000 000200
0.0000020. 05 052.200 0 ..
u 300. 20:00.20 . .02.... .N
203000 500300.". .F

>30 00. .00 0.83080... 00... .o.m 059.0

00.00. .203]

 

>0>.:w ..05.
>035 0000200
0.2200
50.5.... l

 

\

 

1 F20... .80 0.: Li

 

 

 

I?

J

.82.. :03 .02.... 2.0.0
00.00t0> 2220022. 000
0.0002. 20000000 0002000
0.0000000. 0... 00.2.0000 0...

u 303 20:60.20 300.1% .N

0.02000 800000.... .w

 

 

 

.50. 2 0000000. ....
0.2000 .000..00.00.
.0 009.000
.0.0.>0000 .0228
0... 0.0 a

 

0000000. 000 0>00 9.0000

000 000.0000. 000 0200.000 000.20
.000 00.00.020.000 .2000 0:0 .000
.20 009.000 .0>0. .20.: >0 000002....
009.000 0.030000 0.0.0000 05
e822. 8.58.0.2 a... 0.500. 0..

 

 

 

\

 

 

 

K . .0350, a

...I

 

\2235 :03 .02.... 2s... .0]
00.00..0> 20000000... .20
0.00.:0> 20000000 0003.00
0.0000000. 00. 2.2.200 0»
800. 0.0000000 . «03... .N
0.02000 500300.... .w

 

>0>....w =05.
>006 0000200
.0220...
50.8.5.
00.0.. Sam 00...

>025 08000.0...
>005 0000200
.0.220n.
c0020....

 

 

 

 

 

0:232. ., .. 9.0800

 

0009.000 .0>0.
.203 >0 000002....
00 2 >2... 0.02 0.0

“dragging

 

 

J

 

 

\

 

F 9.0.. .80 at .1

 

.cvrv.p. V.. .vh....-1a»..x>1..vw..~¢uau: .... 4

 

0050.2 00.00.05 All. 00.:0m 0.00 ‘I.

 

 

 

 

 

 

a 8.... 55......

J

..............

 

00000000 0005
2000 002200.00
..05 0.0 .003 .20
009.000 .0>0. .203
.0 0.030 0.0 as

\ J

 

 

 

 

, 0.00.0100»

 

 

\

0.030000 9.0000 0320..

...0 009.000 .0>0. .203 .0 00:00.2.
05 .0 22.8 00. 0...... 020.00000

0.0 0.200. .003 000 009.000 .0>0.
.20.... >0 0000022. 00 2 >2... .002
0.0 0.208 .o 08>. 5...; 0.2.0... op

 

 

I . He‘s

. a , «Waxes... . . ..

 

 

 

0.0>0. .203 000.00.:

00.0.. 0. 009.000 .0 002.0002
0... .0 00.2000 000 002.2030
0.2000 2.0.0200. 000000 0...

 

 

 

f 003030 1.0.0000. ....

. . _

 

00000030
00.0000”.

 

 

 

 

 

 

 

4

 

002000.30 00.0000”.

 

 

75

Identify dependent variables

Determined based on the hypotheses

Identify potential independent variables
Determined based on the hypotheses
Statistical tests on alternative independent variables

1. ANOVA - t-test for continuous variables
2. Cross Tab - chi-square for categorical variables

Estimate LOGIT models
Stepwise — backward method
Maximum likelihood parameter estimates

 

Evaluate and explain results of LOGIT models

     
     
       
 

1. Number of cases
Block/Model chi-square
Nagelkerke’s R2

Model prediction

.u~ eww

Parameter estimation

Figure 3.7. The sequence of steps for developing and evaluating LOGIT models.

76

Identify Dependent Variables

Each binary LOGIT model has a different dependent (dummy) variable. The

following ﬁve (dependent) variables were recoded as dummy variables shown in Table

3.18.

Table 3.18. The dependent variables for ﬁve binary LOGIT models.

     

 

No=0

Question Number

 

Objective 1

Dependent
variable 1

Dependent
variable 2

Objective 2

Dependent
variable 3

Objective 3

Dependent
variable 4

Dependent
variable 5

Boaters aware of Lake
Michigan water level
changes

Boaters perceived or
assessed that Lake
Michigan water levels
had dropped a lot

Boaters are inﬂuenced
by concerns relating to
low Lake Michigan
water levels

Boaters changed
boating locations in
response to low Lake
Michigan water levels

Boaters reduced their
boating (days) in
response to low Lake
Michigan water levels

Boaters not aware of
the Lake Michigan
water level changes

Boaters perceived or
assessed that Lake
Michigan water levels
had remained the
same/dropped a little

Boaters are not
inﬂuenced by
concerns relating to
low Lake Michigan
water levels

Boaters did not
change boating
locations in response
to low Lake Michigan
water levels

Boaters did not
reduce their boating
(days) in response to
low Lake Michigan
water levels

Telephone survey: Q
11 and Q57: “Based on
your experience or what
you have heard from
others, would you say
that Lake Michigan water
levels had dropped a lot
or remained about the
same over the past few
years in your usual
boating waters?”

Mail survey: Q 10:
“Did low water concerns
inﬂuence your boating
during the 2001 boating
season?”

Mail survey: Q 10b:
“Did low water inﬂuence
the locations where you
boated?”

Mail survey: Q Ma:
“Did you do less boating
in 2001 because of low
water?”

77

Identify Independent Variables

According to previous boating studies, the boating participation and boaters’
boating behavior are signiﬁcantly related to boater socio-economic characteristic,
watercraft characteristics, boat storage characteristics and boat days and locations. Those
variables are potential independent variables to develop ﬁve LOGIT models and will be
determined to enter in different LOGIT models according to the results of the hypotheses
testing. Besides the independent variables with category characteristics are recoded as

dummy variables to develop the LOGIT models. The description of those independent

variables was shown in Table 3.19.

 

 

Variable name Question Q“; 812011
Type of data # in mail
(Category) survey telephone
survey
Boaters Socio-economic characteristics
Age Continuous data Q16 Q67
Education
Under college Category data NA.2 Q68
College
Graduate (Default)"
Income
Under $45K (T) / Under $60K (M) ’3 Category data Q18 Q69
$45K to $60K (T)/ $60K to $99K (M)
Over $6OK(T)/ Over $100K (M) (Default)
Gender
Male (Coded as l) Dummy variable NA Q70
Female (Coded as O)

‘1 For the category data, the last category of variable is considered as the reference (default) category.

*2 “NA” means the information is not collected in this survey.

*3 The category for income variable is different in telephone and mail survey. The “T” is refereed to the
telephone survey. The “M” is referred to the mail survey.

78

Table 3.19. The potential independent variables for ﬁve binary LOGIT models — cont’d.

 

Variable name Question Q“; i: on
Type of data # in mail
(Category) survey telephone
survey
Permanent state of residence
Michigan Dummy variableM Q14 Recorded
Wisconsin Dummy variable illiterviewer
Indiana Dummy variable
Illinois Dummy variable
Location of permanent residence
Lake Michigan coastal counties Dummy variable Q14 Recorded
Next to Lake Michigan coastal counties Dummy variable Eiyterviewer
Other Counties Dummy variable
Boat characteristics
Boat type
Outboard motor boat Dummy variable Q2a Q5
Inboard-outboards Dummy variable
Inboard motor boat Dummy variable
Powered sailboat Dummy variable
Pontoon Dummy variable
Boat length Continuous data Q2b Q3
Number of years the boat has been owned Continuous data Q2c Q4
Boat storage characteristics
Boat storage type
Marina Dummy variable Q6 Q20
Cottage and Second home Dummy variable
Permanent waterfront home.5 Dummy variable
Permanent non-waterfront home.S Dummy variable
Boat storage state location
Michigan Dummy variable Q6 Q18
Wisconsin Dummy variable
Indiana Dummy variable
Illinois Dummy variable

79

 

 

Table 3.19. The potential independent variables for ﬁve binary LOGIT models - cont’d.

 

 

 

 

 

80

 

Variable name Type of data gilnemii Quﬁgon
(Category) survey telephone
survey
Boat storage county location
Lake Michigan coastal counties Dummy variable Q6 Q18
Next to Lake Michigan coastal counties Dummy variable
Other Counties Dummy variable
Boating days and locations
Number of boating days
Total number of boating days Continuous data Q4 NA
Lake Michigan boating days Continuous data Q4d Q13
Other Great Lakes boating days Continuous data Q4f NA
Boating experiences during the past 3 years
Operated the boat during the past 3 years Dummy variable Q3 Q6
Operated on inland lakes Dummy variable Q4d, Q4f Q7
Operated on other Great Lakes Dummy variable Q4f Q8
Operated on Lake Michigan Dummy variable Q4d Q9
Number of different boating counties Continuous data Q4a NA
Other information
{:elcéussEtYegsgy; information concerning water Dummy variable Q12 NA
The boat was trailered to launch (Yes=1) Dummy variable Q9 NA
The boater own a waterfront home (Yes=l) Dummy variable Q15 NA

*4 The category data is recorded as dummy variable. For example, if the respondent lives in Michigan, the
answer is recorded as “Yes=l”. If not, the answer is recorded as “O” for No.

*5 The storage types of permanent waterfront home and permanent non-waterfront home are combined in
one variable for telephone survey.

 

 

Estimate LOGIT Models

Potential independent variables for the binary LOGIT models included: (1)
registered boat owner socio-economic characteristics (e.g., income, gender, permanent
state and location of residence) (X 1), (2) boat’s characteristics (e.g., boat length, type,
number of years, the boat has been owned) (X2), (3) boat storage segments and locations
(X3), (4) volume of boating days and boating locations (X4) and (5) other information
(e.g., requested/received any information concerning water levels, trailing boats) (X5). A
set of hypotheses was tested to determine the signiﬁcance of the relationship between the
different dependent variables and hypothesized independent variables. T- tests were used
to test relationships between the dependent variables and the continuous independent
variables and chi-square tests were employed to test relationships between the dependent

variables and the categorical and dummy independent variables.

Suppose Y is a dummy variable, coded as 1 for the outcome, which the
respondents checked “Yes”(Y=l) and “Y=O” for the other possible outcome. The

probability of a “Yes” answer is equal to p. The probability of a “No” outcome is then
equal to 1- p. Some explanatory variables (such as X: ,X2,..., X5) are used to explain
the probability of the dummy variable (Y). If we assume the probability of an even (such
as ﬁve dependent variables shown in Table 3.18) occurring (Y= “Yes”=1) is a logistic
function of the independent variable as Equation 3.1. Then the probability of an event

not occurring (Y=”No”=0) is as Equation 3.2.

81

 

(ﬁre/Yemxzeemx.)
IMY=D=p= e
1+ 6(ﬂ0+ﬂ1X1+ﬂ2X2+"'+ﬂ5X5) (3.1)

1

1+ 6(ﬂo+ﬂle+ﬂ2X2+"'+ﬂ5X5) (3.2)

 

Where p = The probability of an event occurring (Y=”Yes”=l)
l- p = The probability of an event not occurring (Y=”No”=0)
X1 = The independent variables - socioeconomic characteristics
X2 = The independent variables - boat characteristics
X3 = The independent variables - boat storage segments and locations
X4 = The independent variables - volume of boating days and boating locations
X5 = The independent variables - other information

50, = Constant

Br, [32, B3, B4, [35 = Model parameters

The odds are deﬁned as the probability of an “event occurring” divided by the

probability of an “event not occurring” (Equation 3.3).

odds = M = —— = e(ﬂ0+ﬂ1X1+lB2X2+m+ﬂ5X5)

p
PdNb) l-p a»

82

Taking natural logarithms of both sides yields a linear equation for the “logits” L.

L = log[1—p—] = 108179):ﬂ0+ﬂ,X1+,32X2+-~+ﬂ5X5 (3,4)

’ P
So these logits are the log odds. The logits L are unbound and can have a linear

relationship with those independent variables. Then software - SPSS 11.0 will be used to

estimate those logits and odds as the Equation 3.5

A 5
logit[p)=ﬂ0+;ﬂixi (3.5)

Where i = An individual recreational boater

’60, ’6', ’62,- ’85 = The estimated coefﬁcients from SPSS

Evaluation and Explanation of LOGIT Models

The t-tests and chi-square tests identiﬁed a set of independent variables that have
statistically signiﬁcant relationships with ﬁve different dependent variables. These
served as the independent variables to formulate those ﬁve binary LOGIT models. A
stepwise- backward method was used to develop the models for two principal reasons.
First, a stepwise method is commonly used in situations where little or no previous
research exists to base the formation and speciﬁcation of LOGIT models (e.g.,
identiﬁcation of statistically signiﬁcant independent variables). A stepwise method is

also frequently employed in situations where the primary purpose is to determine a model

83

that ﬁts the data, rather than determining causality (Agresti & Barbara, 1986; Huberty,

1989)

Then it was necessary to decide between either a backward or forward stepwise
method. The backward method is preferable to the forward method because of
suppressor effects which occur when a predictor has signiﬁcant effect, but only when

another variable is held constant (Field, 2000).

The backward method begins with a model that includes all the
independent/predictor variables, which are determined to be statistically related to ﬁve
different dependent variables. Next, independent/predictor variables that can be removed
from the model without substantially reducing how well the model ﬁts the observed data
are identiﬁed. These variables are deleted from the model sequentially with the ﬁrst

variable removed being the one that least impacts how the model ﬁts the data.

The efﬁciency and reliability of each of the binary LOGIT models are evaluated
by three statistical results: (1) classiﬁcation table, (2) measures of the proportion of
variation explained, and (3) the coefﬁcient estimates of the independent variables (Field,
2000; Long & Freese, 2000). Those four numbers are considered as the “Goodness-of-ﬁt
Measure”, which is used to compare alternative models forming in terms of the

composite contributions of the independent variables.

The classiﬁcation table is a simple tool for assessing how well the model ﬁts the
data by counting up how many observations the model can predict correctly. The higher
the overall percentage of correct predictions the better the model. Besides the value of
the percentage of correct predictions provided by the LOGIT model, two other criterions:

maximum chance criterion and proportional chance criterion are also suggested to use to

84

evaluate the performance of model’s predictive accuracy (Hair, Anderson and Tatham,
1987). The ﬁrst one is the “maximum chance criterion”, which indicated that the percent
of correct classiﬁcation from the LOGIT model should be higher than the percent of
observed group member in the largest group. Another criterion is the “proportional
chance criterion”, which needs to take into account of the ability of the LOGIT model to
classify correctly subjects/objects into smaller group as well as the largest group. Again,
the percentage of correct predictions provided by the LOGIT model should be higher

than the proportional chance criterion.

For the measures of the proportion of variation explained, SPSS 11.0 produces
two numbers: Cox and Snell’s R2 and Nagelkerke’s R2. Nagelkerke’s R2 is an
improvement over Cox and Snell’s R2 that can attain a value of one when the model
predicts the data perfectly (Cox & Snell, 1989; Nagelkerke, 1991). For binary choice
model with an alternative speciﬁc constant, those two R2 should be between 1 and O

6619,

(Ben-Akiva and Lerman, 1985). Besides R2 may not achieve value of even when the

models predicts all the outcomes perfectly (Field, 2000). The higher the “R2” the more

the model can explain for the variations in the data.

For the independent variables, the estimated coefﬁcients in the LOGIT models
should have a value different from “0” (positive or negative value) for the function to be
evaluated to be reliable and efﬁcient. A “B” parameter is derived for each of the binary
LOGIT models. An “Odds Ratio” (Exp(B)) is calculated by computing the natural
logarithm of the “B” parameter. The “Odds Ratio” is an indicator of the changes in odds
of an event occurring (e. g., reducing boating days due to low water levels) resulting from

a unit change in the independent variable (e.g., boat length, type of boat, boat storage

85

segment). An “Odds Ratio> 1” indicates that the value of the independent variable
increases the probability of an event occurring also increases. For example, as boat size
increases the probability that low water concerns will inﬂuence their boating will increase
too. On the other hand, an “Odds Ratio< 1” indicates that the value of the independent

variable decreases the probability of an event occurring also decreases.

86

CHAPTER 4

RESULTS

This chapter provides the results of ﬁve LOGIT models related to the three study
objectives. This chapter begins with a description of the types of data produced from two
2001 Lake Michigan Potential Damages Study surveys described in chapter 3. This
includes descriptive analyses conducted on these data. The second section describes the
results for “Awareness of Lake Michigan Water Level Model” and “Perceptions of the
Drop in Water Levels Model” for objective 1. The third section describes the results for
“Concerns about Low Water Level Inﬂuences Boaters Model” for objective 2. The forth
section describes the results for “Low Water Levels Change Boating Locations” and
“Low Water Levels Reduce Boating Amount” models for objectives 3. The results for
ﬁve LOGIT models are presented in four stages: (1) hypothesis testing, (2) model
speciﬁcation, (3) LOGIT model parameter estimations, and (4) LOGIT model evaluation

and correct prediction.

4.1 Description of Data Obtained from the 2001 Lake Michigan
Potential Damages Study Mail and Telephone Surveys

The 2001 Lake Michigan Potential Damages Study mail and telephone surveys
conducted for this study produced the most current information on boating activities and

patterns in Michigan, Wisconsin, Indiana and Illinois. The two surveys also produced

87

information on the characteristics of boaters (Tables 4.1 and 4.2) and registered

watercraft (Tables 4.3 and 4.4).

The average age of respondents from the mail survey was around 53 years old.
Sixty four percent of registered boat owners, who responded to the mail survey, have at
least one child in their households, with an average of two children per household. Two
thirds (66%) of registered boat owners who responded to the mail survey and almost 40%
of registered boat owners who answered the telephone survey had annual household
incomes of over $60K. A quarter of the mail survey respondents (Table 4.1) owned a
permanent waterfront home and almost 20% had a waterfront seasonal home in the state
where they register their boats. Sixty percent of telephone survey respondents (Table
4.2) had a college degree. Ninety-one percent of telephone survey respondents are males.
The average age of respondents from the telephone survey was around 52 years old. A

third (33%) are retired.

Tables 4.3 and 4.4 provide a summary of the characteristics of the registered
watercraft owned by respondents to the 2001 Lake Michigan Potential Damages Study
mail and telephone surveys. Just 3% of the mail survey respondents own a watercraft
less than 16 feet in length and is regrouped in the category of 17’ to 20’. This low
percentage of small watercraft is a result of the sampling method that deliberately over-
sampled larger watercraﬁ. Seven percent of the boats from the mail survey were over 40
feet and one percent of the telephone respondents own a boat over 40 feet. The majority
of mail survey respondents owned inboard-outboards. A higher proportion of the
telephone survey respondents owned outboard motor boats and this was due in part to the

sampling method. Respondents to both surveys owned their boats an average of 7 years.

88

Table 4.1. Registered boaters who completed the 2001 Lake Michigan Potential Damages
Study mail survey.

“
Variables Percentage/Mean

Age of registered boat owners 53 years

 

Number of adults in the household

1 adults 13%
2 adults 73%
3 or more than 3 adults %
100%
Household with children
No children 36%
1 child 25%
2 children 25%
3 children or more than 3 children _14"/_o
100%
Household income
Under $60K 34%
$6OK-$99K 32%
Over $100K 351%
100%
State where boat owners’ permanent residence is
Michigan 44%
Wisconsin 24%
Indiana 11%
Illinois 19%
Other states %
100%
Waterfront home
Own a waterfront permanent home 25 %
Own a waterfront cottage or seasonal home 19%
Number of owned watercraft.
l watercraft 44%
2 watercraft 30%
3 watercraft 16%
4 or more than 4 watercraft 1%
100%

m
" Number of owned watercraﬂ included the watercraft sampled in the mail survey

89

Table 4.2. Registered boaters who answered the 2001 Lake Michigan Potential Damages

     

 

 

Study telephone survey.
Variables. Percentage/Mean
Age of boat owner 52 years
Educational level
Under college 40%
College 22%
Higher than college M
100%
Income
Under $45K 21%
$45K to $60K 40%
Over $60K £019
100%
Employment
Employed 62%
Unemployed 3%
Retired 33%
Homemaker 2%
100%
Gender
Male 91%
Female 2°/_o
100%

m
*The response - refused is treated as 3 missing value in the analysis.

90

Table 4.3. Characteristics of registered watercraft owned by mail survey respondents.
E

 

Variables Percentage
Boat length.
17' to 20' 59%
21' to 25' 21%
26' to 39' 13%
More than 40' M
100%
Boat type
Inboard motor boat 19%
Inboard-outboards 30%
Outboard motor boat 27%
Powered sailboat 8%
Pontoon boat M
100%
Year the watercraft was purchased"
Earlier than 1985 11%
1986 to 1990 12%
1991 to 1995 18%
1996 to 1998 25%
1999 13%
2000 13%
2001 &
100%
Average number of years the boat has been owned 7.41 years
New or pre-owned
New boat 45%
Pre-owned boat mg
100%
State where the watercraft was registered
Michigan 46%
Wisconsin 27%
Indiana 11%
Illinois mg
100%

m
* The percentage of boat length is not the same as the percentage in the original sample because 10% of the
mail survey respondents did not own the sampled watercraft and owned another registered watercraft,
which was used to answer the survey.
** This variable referred to the year the watercraft was purchased. It could be either new or pre-owned
watercraft.

91

Table 4.4. Characteristics of registered watercraft owned by telephone survey

 

respondents.
Variables Percentage
Boat length.
17' to 20' 63%
21' to 25' 26%
26' to 39' 10%
More than 40' _1_°_/(_)
100%
Boat type
Inboard motor boat 13%
Inboard-outboards 34%
Outboard motor boat 37%
Powered sailboat 4%
Pontoon boat mg
100%
Year the watercraft was purchased
Earlier than 1985 19%
1986 to 1990 14%
1991 to 1995 28%
1996 to 1998 21%
1999 10%
2000 4%
2001 M
100%
Average number of years the boat has been owned 7.29 years
State where the watercraft was registered
Michigan 32%
Wisconsin 24%
Indiana 15%
Illinois 2%
100%

 

 

 

 

 

"‘ The telephone surv—y did not sample the bpsmaller tan 16 feet. Of all respomsd,only 1% did not
own the sampled watercraft but owned another registered watercraft. Of 1%, only 1 respondent owned
another registered watercraft smaller than 16 feet and was excluded from the analysis.

92

Although two 2001 Lake Michigan Potential Damages Study surveys were
conducted during the same period and had the same sampling frame (owners of currently
registered recreational watercraft in the four states), the samples were still different.
Statistical tests were ﬁrst performed to determine any statistically signiﬁcant differences
in the characteristics of telephone and mail survey respondents including their ages,
incomes, state of registry, boat length, boat type and number of years the boat has been

owned. The results of the statistical tests are presented in Table 4.5.

Chi-square analyses showed that there is no signiﬁcant difference on the average
age of boat owners between mail and telephone survey respondents. Statistically
signiﬁcant differences existed in the income, state of registry, boat length, boat type and

number of years the boat has been owned between mail and telephone surveys.

Table 4.5. The comparison between the 2001 Lake Michigan Potential Damages Study
mail and telephone surveys.

     

 

Mail Telephone x2 (dt) P-value
By age
<29 33% 67% 7.962(6) 0.241
30-39 40% 60%
40-49 43% 57%
50-59 43% 57%
60-65 45% 55%
>66 331/9 171/9
42% 58%

93

Table 4.5. The comparison between the 2001 Lake Michigan Potential Damages Study
mail and telephone surveys - cont’d.

     

 

 

 

 

Mail Telephone x2 (df) P-value
By income
Under $20K 24% 76% 91.906(4) 0.000
$20 to $40K 47% 53%
$40 to $60K 45% 55%
$60 to $100K 43% 57%
Over $100K 6% M
42% 58%
By state of registry
Michigan 53% 47% 115.835(3) 0.000
Wisconsin 47% 53%
Indiana 36% 64%
Illinois 3_0°_/9 M
42% 5 8%
By boat length
17’ to 20’ 39% 61% 136.774(4) 0.000
21 ’ to 25’ 37% 63%
26’ to 39’ 49% 51%
Over 40’ ﬂ 232/2
42% 58%
By boat type
Outboard motor boat 28% 72% 196.988(4) 0.000
Inboard-outboards 39% 61%
Inboard motor boat 61% 39%
Powered sailboat 57% 43%
Pontoon boat 4_5_% 5_5_°_/9_
42% 58%
By year the watercraft was
purchased
< 1980 32% 68% 189.449(7) 0.000
1981 to 1985 25% 75%
1986 to 1990 39% 61%
1991 to 1995 32% 68%
1996 to 1998 47% 53%
1999 49% 51%
2000 68% 32%
2001 6_2_°/2 38%
42% 58%

W—
—

94

4.2 Recreational Boater Awareness and Perceptions of Lake Michigan

Water Level Changes (Study Objective #1)

This section reports on the analyses employed in the development of two binary
LOGIT models developed to explain recreational boaters’ awareness and appraisal of the
magnitude of changes in Lake Michigan water levels. The models utilized data generated
by the 2001 Lake Michigan Potential Damages Study telephone survey. Boaters (1,646)
were ﬁrst organized into two groups based on their awareness of Lake Michigan water
level changes. Then the 1,177 boaters, who indicated an awareness of Lake Michigan
water level changes, were further divided into two groups according to their assessments
about how much Lake Michigan water levels had changed. Potential independent
variables included boater and boat characteristics, boat storage characteristics, boating

locations and amount/volume of boating days.

4.2.1 Awareness of Lake Michigan Water Level Changes

Of the 1,646 boaters who responded to the 2001 Lake Michigan Potential
Damages Study telephone survey, almost three quarters (72%) of boaters were aware of

Lake Michigan water level changes.

Hypothesis Testing
The hypothesis is that recreational boaters’ awareness of Lake Michigan water
level changes is signiﬁcantly inﬂuenced by their socioeconomic characteristics, boat

characteristics, boat storage types and locations, their boating activity level (days) or

95

boating locations during the past three years. Table 4.6 reports the results of hypothesis
testing of the relationship between possible independent variables (for the binary LOGIT

model) and boaters’ awareness of Lake Michigan water level changes.

Statistically signiﬁcant relationships were found between recreational boaters’
awareness of Lake Michigan water level changes and their socioeconomic characteristics
including education, income, gender, where their permanent residence is located and their

boat’s characteristics including type and size.

Owners who have inboards and inboard-outboards are much more aware of
changes in Lake Michigan water levels. Large powered sailboats are even more aware of
water level changes given that the draft of these boats makes them more susceptible to
water level impacts. No statistically signiﬁcant relationship between lake level
awareness and the type of ‘during the boating season storage’ and location where they
stored their boats is present. Their Great Lake boating experiences during three previous
years are statistically related to their awareness of Lake Michigan water level changes.
As expected, active boaters are more likely to be aware of ﬂuctuations in Lake Michigan

water levels.

96

Table 4.6. Relationships between boater and boat characteristics and boater awareness of

Lake Michigan water level changes.
w
Aware of LM Not aware of xz/

water level LM water level F-value P-value

 

ﬂuctuations ﬂuctuations (df)
Boater characteristics
Age (Mean) 53 51 2.269(1) 0.132
Education
Under college 67% 33% 16.62(2) 0.000
College 69% 31%
Graduate 171/9 Mg
72% 28%
Income
Under $45K 63% 37% 13.81(2) 0.001
$45K to $60K 70% 30%
Over $60K M 251%
72% 28%
Gender
Female 65% 35% 330(1) 0.040
Male _72_°/g 280A
72% 28%
Permanent state of residence
Michigan 89% 1 1% 168.4(3) 0.000
Wisconsin 77% 23%
Indiana 60% 40%
Illinois 5_4% M
72% 28%
Location of permanent residence
LM coastal counties 89% 11% 255.5(2) 0.000
Next to LM coastal counties 81% 19%
Other counties 50% 50%
72% 28%
Boat characteristics
Boat type
Outboard motor boat 68% 32% 2 1 . 1(4) 0.000
Inboard-outboard 75% 25%
Inboard motor boat 73% 27%
Powered sailboat 90% 10%
Pontoon 6_6% 342/9
72% 28%

 

97

Table 4.6. Relationships between boater and boat characteristics and boater awareness of

   

AwaM " aeo x2/
water level LM water level F-value P-value

 

ﬂuctuations ﬂuctuations (df)
Boat length 22 20 643(1) 0.010
Number ofyears, the boat has 7.4 70 148(1) 0255
been owned
Boat storage characteristics
Boat storage type
Permanent home 61 % 79% 2.269(2) 0.433
Cottage and second home 9% 0%
Marina 3% 231/9
72% 28%
Boating behavior
Boating days (Mean) 24 13 1.513(1) 0.219
Operated the boat during the past
three years
Yes 72% 28% 1592(1) 0.000
No 31°19 M
72% 28%
Operated the boat on inland lakes
during the past three years
Yes 71% 29% 561(1) 0.010
No 2% M
72% 28%

Operated the boat on other Great Lakes
during the past 3 years

Yes 92% 8% 220.0(1) 0.000
No 59% 41%
72% 28%

Operated the boat on Lake Michigan
during the past 3 years

Yes 98% 2% 332.3(1) 0.000
No 56% 44%
72% 28%

W

98

Model Speciﬁcation

Variables having statistically signiﬁcant relationships with the awareness of Lake
Michigan water level ﬂuctuations for use in the binary LOGIT model are (1) Income, (2)
Education, (3) Gender, (4) Permanent state of residence, (5) Location of permanent
residence, (6) Boat length, (7) Boat type, and (8) Where boats were operated (e. g. on
Lake Michigan & on other Great Lakes). The default category for income variable is
over $60K. For education variable, the default category is having a graduate degree.
Each category (Michigan, Wisconsin, Indiana, & Illinois) of the “location of permanent

residence” variable is recorded to four dummy variables.

Of the 1,646 survey respondents, only 1,142 cases were available to develop the
binary LOGIT model. Approximately 500 cases were excluded because they had a
missing value for one or more of the independent variables. The stepwise- backward
method begins with all 11 independent variables. Statistical tests are then performed to
assess whether removing any of these predictors can be done without appreciably
affecting how well the model ﬁts the observed data. This backward-stepwise method
produced eight separate binary LOGIT models. The following variables were
sequentially removed: ”boater education,” “boat length,” “operating experience on inland
lakes,” “boat type: inboard-outboards, outboards and pontoon,” and “operating
experience on other Great Lakes.” Eight signiﬁcant independent variables remaining in
ﬁnal LOGIT model are permanent state of residence-Michigan, Wisconsin and Indiana,
location of their permanent residence- Lake Michigan coastal counties, counties adjacent
to Lake Michigan coastal counties, Lake Michigan boating experience, annual income

(45K to 60K) and gender.

99

LOGIT Model Parameter Estimations
The estimate of LOGIT model is (Table 4.7):

logit(p) = -1.409 +1.885M1 + 0.380WI + 0.624IN +1.627LMC +1.419NLMC
+ 2.7780LM — 0.456INCOME + 0.651MALE

P = The probability that a boater was aware of Lake Michigan water level
changes

MI, WI, IN = Permanent state of residence-Michigan, Wisconsin and Indiana

LMC = Location of permanent residence for Lake Michigan coastal counties

NLMC = Location of permanent residence for counties adjacent to Lake

Michigan coastal counties
OLM = Operated the boats on Lake Michigan
INCOME = Income - $45Kto 60K

MALE = Male boater

The equation can be used to predict the odds and the probability of a recreational

boater who was aware of Lake Michigan water level changes for given values of those
independent variables. A positive “[3” indicates that as the value of the independent
variable increases the probability of being aware of water level changes also increases. A
negative “[3” indicates that as the value of the independent variable increases the
probability will decrease. The larger the value of “B“, the greater the relative importance
of this variable in predicting awareness. The Wald tests whether the coefﬁcient is

signiﬁcantly different from zero.

100

Using the ﬁrst signiﬁcant variable to illustrate — permanent state of residence “
Michigan” as the example, the estimate of “B” is 1.885 with a standard error of 0.268 and
is signiﬁcant according to the Wald’s test. The interpretation of the coefﬁcient ,6] (e. g.,
1.885) is the change in the log odds of awareness for a resident of Michigan relative to a
non-Michigan resident. The “Odds Ratio” (Exp(,61)) is calculated by computing the
natural logarithm of the “B” parameter and is an indicator of the awareness of Lake
Michigan water level changes resulting from a unit change in the independent variable (in
this case, living in Michigan, rather than Illinois).

For a recreation boater who did not live in Michigan, the calculation would be:

log(0dds) = -1 .409

Therefore, the odds of the awareness of Lake Michigan water level changes is:

odds(awareness) = exp(-1.409) = 0.244

And the probability of awareness is:

p(awareness) = $14— = 0. 96

1+ 0.244

For a recreation boater who lived in Michigan, the odds and probability of his

awareness of Lake Michigan water level changes is:

log(odds) = —1.409 +1885 *1 = 0.476,
0dds(awareness) = exp(0.476) = 1.6096,

p(awareness) = ﬂ—gé— : 0.6168

1 + 1.6096

For a recreation boater who lived in Michigan, the probability of being aware of
Lake Michigan water level changes is around 62%, compared to only 20% for non-

Michigan residents.

101

Examining the coefﬁcient of county location variable, the probability of being
aware of Lake Michigan water level changes for a boater who lives in Lake Michigan
coastal counties is around 55%. Moreover, boaters who live in Lake Michigan coastal
counties are almost 5 times higher to be aware of Lake Michigan water level changes

than boaters who do not live in the Lake Michigan coastal counties.

The most important factor to inﬂuence the boaters’ awareness of Lake Michigan
water level changes is whether they operated their boats on Lake Michigan. If the boater
who lived in Michigan and also operated the boat on Lake Michigan, the probability of
being aware of the water level changes will be around 96% with the default values (0) of
all other variables. In this case, the probability of being aware of Lake Michigan water
level changes increases from 62% to 96%. Male boaters are more likely to be aware of

the Lake Michigan water level changes than female boaters.

Table 4.8 illustrates the calculation of probabilities of being aware of Lake
Michigan water level changes using the LOGIT model. The examples from one to four
are the probability of being aware of Lake Michigan water level changes for boaters who
lives in Michigan but have different characteristics. The examples from ﬁve to eight are
the probability for boaters who live in Wisconsin but have different characteristics.
Probabilities of being aware of Lake Michigan water level changes increase from 62%
for Michigan residents with the default values of all other variables (did not live in Lake
Michigan county, did not operate on Lake Michigan, income less than $45K or more than
60K, female) to 89% if living in a coastal county or 96% if operated a boat on Lake
Michigan, and ﬁnally 98% if all of the earlier characteristics and male. For the

Wisconsin residents, probabilities of being aware of Lake Michigan water level changes

102

increase from 26% with the default values of all other variables (the same as the setting
for Michigan residents) to 65% if living in a coastal county or 85% if operated a boat on

Lake Michigan, and ﬁnally 92% if all of the earlier characteristics and male.

Table 4.7. The result of the ﬁnal binary LOGIT model for boaters’ awareness of Lake

 

 

Permanent state of residence

Michigan 1.885 0.268 49.552 0.000 6.586

Wisconsin 0.3 80 0.225 2.836 0.092 1.462

Indiana 0.624 0.215 8.396 0.004 1.867
Location of permanent residence

LM coastal counties 1.627 0.224 52.736 0.000 5.091

Next to LM coastal counties 1.419 0.213 44.569 0.000 4.132
096”}th the boat on Lake 2.778 0.304 83.650 0.000 16.090
Michlgan
Income

Over $60K - - 4.167 0.124 -

$45K to $60K -0.456 0.229 3.982 0.046 0.634

Under $45K -0.092 0.187 0.244 0.621 0.912
Male 0.651 0.281 5.376 0.020 1.918
Constant -l.409 0.329 18.398 0.000 0.244
Summary Statistics

Model chi-square [di]=459.5[9]'
Nagelkerke R2=0.48

% Correct predictions=82%

@
* P-value < 0.05

103

 

104

 

$3 $3 $3 <er r: a
one? 4%? :2 .w 23 $53
mood #4346. m2 .N ciao £60 E
oov. T _ mow. T _ oov. _ - _ aov. ﬂ - _ 3&6 aov. T 8888
306 _ o o o o o o Ba; :06 232
. . £8-65;
o o o o o o o o :6 o 84 o- 6885
EA
mid ﬂ wimm _ o o o o 393 wtm so :89 2:
3:325
8958
o o o o o o o o mﬂé o2: .888
$3 2 :82
. . . "22:58
o o o o Re _ _ o o as m Re _ 588 s:
o o o o o o o o cow. w vmod «.565
o o o o o o o 0 new. ﬂ w m6 £8893
wwwg _ mmwA _ mmwé _ 3w; _ owmé www; Semioaz
Axeoxm x Axeoxm x Axeoxm x 99.5 x :25 a
o gwﬁwmz u on :mwﬁoaz 813 no Eon 05 w
E am 33 5 c “E o 888% 98 33:22 E 5 332 5w 53:22
can .8 £22 E 8:550 8:550 2338 53:22 8:58 338 cm £22 5 mg: on? Samoa <
338 Sewing: 8.3 815 S 8.:— oaB 882. < 8:3 5 m9»: on? 388 <
5 $3: on? 5:89 29: < . .
Hang nleag a gag

.moﬂmtouoﬁmno .683 3828 23 mowcmno .023 ..on3 :meosz 8.3 .«0 $8898 mo 43:938.“ 05. .wé 2an

 

105

 

 

$3 :3 some :2 w a:
20.: 9.2 $3 $3 9-3}
83 $5 mono 3:. wave 5
ooe. _- Se. _ .. _ o8. _ - _ 8e. _- _ $3 ooe. _ - 8688
God _ o o o o :3 God 222
. . Axonéﬁvv
o o o o o :e o Se o- 9:85
23
we: MEN o o o 35.2 MEN 8 :86 2:
BEBQO
monsoon.
o o o o o mm _ .e 23 388
23 on :82
o o o o 83 _ o o owed RE _ewowoeezem
o o o o o o o o 83 ego 885
”no _ woo _ woo _ wmd _ me: woo 3882;
o o o o o o o o 33 23 53:92
62.33 x Gavan x Gavan x caves x avoxm a
_EwEomZ a dawn—~22 03 so Eon 05
03 co Eon 95 BEE o voEuomo BE 59.8%? E Emcoomna E 59885»
BE 53835 5 36:38 355.8 2338 cam—n22 won—Boo _Smmoo Ewmnoaz E mo>n on? noEon <
2338 EwEon—z 03 03 S 82. on? BEon < 03 5 mg: on? BEon < . .
5 mg: on? uoEon 23.. < . .
3m 3 3 3

   

 

.nzwoo I monmtoowﬂmno Beanbag no; mowomno 35.3 5E3 .Ewanz 03 mo oEBE 0.8 on? EoEon 05 com bEnEnoa onh. .3» 033.

LOGIT Model Evaluation and Correct Prediction

According to model chi-square statistics reported in Table 4.7, the ﬁnal LOGIT
model is strongly signiﬁcant (x2=459.5). The value of Nagelkerke’s R2, the measures of

the proportion of variation explained is around 48% for the ﬁnal LOGIT model.

Overall, the percentage of correct predictions of boaters’ awareness of Lake
Michigan water level changes (Table 4.9) substantiates the predictive power of the Model
8, which yields a hit ratio of 82% with correct classiﬁcation in the diagonal cells and
incorrect classiﬁcations in the off-diagonal cells. In other words, this model correctly
predicts almost 82% of respondents’ awareness. This binary LOGIT model correctly
classiﬁes 87% of boaters who were aware of the water level changes, and 70% of the

boaters who were not aware of the water level changes.

Compared to the maximum chance criterion-72%, the percent of correct
classiﬁcation of the LGOIT model resulted in an increase of 10%. The percent of correct

classiﬁcation (82%) is also higher than the proportional chance criterion (60%).

106

Table 4.9. Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who were and were
not aware of Lake Michigan water level ﬂuctuations.

   

 

 

 

Model classification/prediction
Aware of LM Not aware of LM

 

Percentage
Ob d water level water level C
serve ﬂuctuations ﬂuctuations orrect
Aware of LM water 713 108 821
level ﬂuctuations 0 V 8 0
(Row Pct.) (8 7/o) (13 o) 7 A)
Not aware of .LM water 97 224 321
level ﬂuctuations 300/ a 700
(Row Pct.) ( 0) (70 A) /0
Model Predicted (total) 810 332 1,142
Summary Statistics
Percentage of cases correctly classiﬁed 82%
Maximum chance criterion 72%
Proportional chance criterion 60%

4.2.2 Assessment of Lake Michigan Water Level Changes

Of 1,177 recreational boaters, who were aware of Lake Michigan water level
ﬂuctuations, 74% believed that Lake Michigan water level had dropped “a lot”, while

26% believed that they had either remained constant or dropped “a little” (Table 4.10).

Table 4.10. Boaters’ assessments about the extent of Lake Michigan water level

       

 

ﬂuctuations.
Water level had dropped “a lot” 870 74%
Water level had remained the same or had dropped a little ﬂ 210/2
Total. 1,177 100%

 

107

WI:

The hypothesis that framed the development of the second binary LOGIT model
is that boaters’ assessment of the extent of Lake Michigan water level changes is
signiﬁcantly inﬂuenced by their socioeconomic characteristics, type and size of the boats
they own, the number of years they have owned their boats, where they stored their boats,

boating volume and where they operated their boats, and types of their boating activities.

Hypothesis tests revealed no signiﬁcant relationship between assessment of the
degree of Lake Michigan water level changes and any of the socioeconomic
characteristics including age, education, income and gender because 1,177 boaters are all
aware of Lake Michigan water level changes. Two of the boat characteristics — “number

of years the boat has been owned” and “length” inﬂuence their perceptions.

The state and county where boaters kept their boats are statistically related to their
“extent of change” perceptions. Type of storage are statistically related to how much
they perceived Lake Michigan water level had changed, but not related to their awareness
about the Lake Michigan water level changes at all. Boaters who stored their boats in
marinas or who kept their boats in Lake Michigan coastal counties are both more likely to
assess Lake Michigan water level had dropped a lot. The types of boating activities they
participate in (e. g., ﬁshing and cruising) do not signiﬁcantly determine their perceptions.
However, their Great Lakes boating experiences during the previous three years are, as
was expected, signiﬁcant factors in determining their perceptions of the magnitude of

changes in Lake Michigan water levels.

108

Table 4.11. Relationships between boater and boat characteristics and boater perceptions
of the magnitude of Lake Michigan water level changes.

 

 

 

 

Perceptions of Lake Michiga_n XZ/
water level changes
_ F-value P-value
Dropped Remained (df)
“a lot” the same
Boater characteristics
Age 53 51 1.959(1) 0.162
Education
Under college 67% 33% 0.171(2) 0.918
College 69% 3 1%
Graduate m 21’/_0
72% 28%
Income
Under $45K 76% 24% 1.859(2) 0.395
$45K to $60K 70% 30%
Over $60K M 213/2
72% 28%
Gender
Female 65% 35% 0.005(1) 0.526
Male 221/9 2%
72% 28%
Boat characteristics
Boat type
Outboard motor boat 75% 25% 4.389(4) 0.356
Inboard-outboards 73% 27%
Inboard motor boat 70% 30%
Powered sailboat 84% 16%
Pontoon boat 12_°/_o _28_%_
72% 28%
Boat size 22 19 725(1) 0.010
Number of years, the boat has 8 7 3. 13(1) 0.077
been owned
Boat storage characteristics
Boat storage
Permanent home 79% 21% 15 .92(2) 0.001
Cottage and second home 56% 44%
Marina §_l_°/_o 1%
72% 28%

109

Table 4.11. Relationships between boater and boat characteristics and boater perceptions
of the magnitude of Lake Michigan water level changes — cont’d.

 

 

 

Perceptions of Lake Michiga_n xz/
water level changes
, F -value P-value
Dropped Remained ((11)
“a lot” the same

Boat storage location (state)

Michigan 82% 18% l2.109(4) 0.017

Wisconsin 80% 20%

Indiana 80% 20%

Illinois 65% 35%

Others m M

72% 28%

Boat storage location (county)

Lake Michigan coastal counties 82% 18% 8.437(1) 0.002

Other counties 22_°/g M

72% 28%

Boating activity

Fishing 79% 21% 1.359(1) 0.143

Cruising 76% 24% 0.746(1) 0.218

Skiing 70% 30% 4.374(1) 0.027
Boating pattern
Boating days (Mean) 23 26 1.161(1) 0.204

Operated the boat during the past three years

Yes 74% 26% 1.303(1) 0.179
No 64% M
72% 28%
Operated the boat on the inland lakes during the past three years
Yes 70% 30% 1.527(1) 0.126
No 172/2 .21_°/9
72% 28%
Operated the boat on other Great Lakes during the past three years
Yes 76% 24% 7.925 (1) 0.003
No 7% 303/2
72% 28%
Operated the boat on Lake Michigan during the past three years
Yes 76% 24% 1 .89(1) 0.096
No 7_2_°/2 2.8%

 

110

Model Speciﬁthiqg

Variables that have signiﬁcant differences between the boaters with different
“extent of change” perceptions are used to develop the LOGIT model. Those are: (1)
boat length, (2) boat type, (3) number of years the boat has been owned, (4) where the
boat was operated (e. g. on Lake Michigan, on other Great Lakes), (5) boat storage type,
(6) boat storage state location, (7) boat storage county location, and (8) water skiing. The
category variables (boat type, boat storage type, boat storage state and county location)

are recorded as dummy variables before entering the LOGIT model.

Of the 1,177 boaters who were aware of changes in Lake Michigan water level,
only 569 were available for model formation because some cases had a missing value for
one or more of the independent variables. Again, a stepwise-backward procedure was

used to identify alternative forms of the model and to arrive at the “best” ﬁnal model.

Four variables were excluded from stepwise-backward procedure — (1) boats
stored in marinas during the boating season, (2) water skiing is a boating activity, (3)
outboards as a type of boat, and (4) boats stored in Illinois during the boating season.
Eleven variables remained in ﬁnal model but one variable (pontoon) was not signiﬁcant

according to Wald’s test.

Ten signiﬁcant variables incorporated as part of the ﬁnal binary LOGIT model
have signiﬁcant effects on the probability that boaters assess Lake Michigan water levels
had dropped “a lot” (Table 4.12). Those are boat length (at the 0.05 level), number of
years the boat has been owned (at the 0.05 level), boat type - inboard-outboards (at the
0.10 level), and inboards (at the 0.05 level), operating experience on other Great Lakes

(at the 0.10 level), state where the boat was stored during the boating season - Wisconsin

111

(at the 0.05 level), Indiana (at the 0.10 level) and Michigan (at the 0.05 level), county
where the boat was stored during the boating season — Lake Michigan coastal counties (at

the 0.10 level), and type of storage - cottage and second home (at the 0.01 level).

LOGIT Model and Parameter Estimations
The estimated LOGIT model is (Table 4.12):

log it(p) = 0.035 + 0.052LENGTH + 0.048YEAR — 0.4321NOUT — 1.0451NBOARD
- 0.68006L + 0.790WI + 0.724IN + 0.89 1M] + 0.377SLM — 1.267SECOND

P = The probability that a boater assessed Lake Michigan water level had

dropped a lot
LENGTH = Boat size
YEAR = Number of years the boat has been owned
INOUT, INBOARD = Boat type — Inboard-outboards and inboards
OGL = Boats were Operated on other Great Lakes

WI, IN, M1 = Boats were stored in these states during the boating season —

Wisconsin, Indiana, and Michigan

SLM = Boats were stored in Lake Michigan coastal counties during the boating

season

SECOND = Boats were stored at cottage or second homes

According to the results, the values of the independent variables including boat

length, number of years boat has been owned, boat storage state location (e. g. Wisconsin,

112

Indiana & Michigan), boat storage county location increase, the probability of the

boaters’ assessment that Lake Michigan water level had dropped “a lot” increases too.

The results from the LOGIT model reveal some important expected and
unexpected relationships between different independent variables and perceptions of the
degree of Lake Michigan water level changes (Table 4.12). Some of these relationships
are easy to explain, others are much more difﬁcult. For example, owners of larger size
boats are more likely to have perceived that Lake Michigan water levels had dropped “a
lot” in large part because the draft and maneuverability of their boats they are more likely
to be impacted by low water. With every 1 foot increase in boat size, the probability that
the boater assess that Lake Michigan water level had dropped “a lot” will increase around
1%.

The results also indicated that boaters, who owned their boats for a longer time,
are more likely to assess that the Lake Michigan water levels had dropped “a lot.” One
year longer, which the boaters owned their boats, will increase 1% in the probability of
their assessment that Lake Michigan water level had dropped “a lot”. The low water
level signiﬁcantly happened since 1997. If the boaters had the boating experience earlier
than 1997, they would be able to compare the change of Lake Michigan water levels

before and after 1997.

The LOGIT model also shows differences in perceptions depending on which
Lake Michigan states (e. g., Michigan, Wisconsin, Indiana and Illinois) where the boat
was stored at during the boating season. The owners who stored their boats in Michigan
during the boating season are almost 2.5 times more likely to assess that Lake Michigan

water level had dropped “a lot” than boaters who did not keep their boats in Michigan.

113

This is consistent with previous low water studies, which found that Lake Michigan water
levels have had a greater impact on boating in Michigan than other Lake Michigan states.
Besides boat owners who kept their boats in Lake Michigan coastal counties are more

likely to assess that water levels had dropped “a lot.”

Unexpectedly, boaters who stored their boats at cottages or second homes are
more likely to perceive that Lake Michigan water levels had not dropped a lot or dropped
a little. This is in part due to the fact that many of the second homes and cottages are
located on inland lakes and these boats are used more often on inland lakes. It is not
clear why the boaters who had either “inboard-outboards” or “inboards” are less likely to
assess the Lake Michigan water level had dropped a lot. While this makes sense for
inboard-outboards since they are generally smaller boats, it is unexpected for inboards,

which are on average larger boats.

By using the results of model parameter estimates, the probability of boaters’
assessment that Lake Michigan had dropped “a lot” can be calculated for different types
of boaters. For example, the probability of assessing that Lake Michigan had dropped “a
lot” is 83% for a boater who owned a 20’ boat for ten years with the default values of all
other variables (Table 4.13). For a boater who owned a 30’ boat for ten years, the
probability will increase from 83% to 89%. Furthermore, the probability will decrease if
the boater owned an inboard-outboard motorboat. The probability of assessing that Lake
Michigan had dropped “a lot” is 84% for a boater who owned a 30’ inboard-outboard for
ten years. Then if a boater owns a 30' inboard-outboard boat for ten years and also kept

the boat in Michigan, the probability will increase from 84% to 93%.

114

Table 4.12. The result of the ﬁnal binary LOGIT model for boaters’ assessments of the

 

 

Boat length 0.052 0.022 5.402 1 0.020 1.053
Number Of years the boat has 0.048 0.020 5.739 1 0.017 1.049
been owned
Boat type
Inboard-outboards -0.432 0.244 3. 128 1 0.077 0.649
Inboards -l.045 0.337 9.602 1 0.002 0.352
Operated the boat on other Great _0. 680 0.355 3.663 1 0.056 0.507
Lakes
Boat storage state location
Wisconsin 0.790 0.290 7.412 1 0.006 2.204
Indiana 0.724 0.400 3.280 1 0.070 2.062
Michigan 0.891 0.281 10.024 1 0.002 2.437
Boat storage county location -—
Lake Michigan coastal counties 0.377 0.220 2.916 1 0.088 1.457
Boat Storage type ‘ -1 .267 0.341 13.781 1 0.000 0.282
Cottage/second home
Constant 0.035 0.624 0.003 1 0.955 1.036

Summary Statistics

Model chi-square [de =61 . 157[9]"

Nagelkerke R2: 0.156

% Correct predictions=79%

 

P—alue < 0.05
*" P-value < 0.10

115

 

 

$8 :3 $3 $8 Q a
48.2 E .m 33 23. ESE
3.0m mvnn m3.~ mmmn mono 2
one; 986 E8980
mmod _ mmod _ mmod _ mmod _ Nwmd >34. 882 9808 E 38. mEB Eom
. Ewﬁoaz E 8958
o o o o o o o O was Em C .988 33.8 as. 8..» 88
o o o o o o o o £3 33 58292
_owd n o o o o o o mnoN and E563
o o o o o o o 0 Sad and £3823
o o o o o o o o .958 omEHonm Eom
. . moxEA
o o o o o o o o mom c me o- E20 Enno co Eon on“ BEEQO
o o o o o o o o $2. $64. 2585
mmvd- _ mmvo- _ o o o o 94nd mmvd- EEEonEoéEonE
. . . . . . Ego =85
w... o 2 me o 2 was o 2 me o 2 3o _ mg o 8: 88 2: 88» no 8:52
own on em; om cm; om we; om mmon Nmod 5mg. Eom
Axeaxm x caves. omega x omega x @qu a

 

116

8292 5 EB 2:
88. 2E mon S no.“
Eon oEonEoéEonE
.3 E 35.6 EEon <

a

€on 2 now Eon
nEonEoﬁEonE
.3 E cacao EEon <

m 29:53—

mon 2 now Eon
.3 E 3:26 BEon <

mon 3 now Eon
.om E Ego ooEon <

N «_mEExm ~ 2mEExH

 

.monmtonoEEno RES @8028 an “on E 3936 En _o>o_ 6E3 EwEBE SEQ 3383 on? EBEon on» .5.“ bEnEnoa BE. .2 .v BEE

LOGIT Model Evaluation and Correct Prediction

According to model chi-square statistic, the ﬁnal LOGIT model is signiﬁcant at
the 0.01 level (x2=61.157) in terms of overall model ﬁt (Table 4.12). The value of model
chi-square is still strongly signiﬁcant. The performance of the Nagelkerka R2 is about

16%.

The percentage of model correct predictions (79%) of the degree of Lake
Michigan water level changes provides a measure of its predictive power (Table 4.14).
The table shows correct classiﬁcations in the diagonal cells and incorrect classiﬁcations
in the off—diagonal cells. Overall, this model correctly predicts almost 79% of
respondents’ assessments of the extent of Lake Michigan water level changes.
Furthermore this model correctly classiﬁes 97% of boaters who assessed the Lake
Michigan water levels had dropped “ a lot” but only 12% of boaters, who assessed the
Lake Michigan water levels had remained the same or just dropped “a little.” The
possible reason of low prediction percentage (12%) is the fewer cases of boaters who
assess the Lake Michigan water level had remained the same. Besides higher variations

are usually found existing in the respondents who checked the “No” answer.

This model only increase 5% in correct prediction by comparing the difference
between the values of overall model correct prediction (79%) and maximum chance
criterion (74%). But this model can still adequately predict boaters’ assessment of Lake
Michigan water level changed because the overall model correct prediction percentage

(79%) is still higher than the value of proportional chance criterion (53%).

117

Table 4.14. Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who assessed the
Lake Michigan water levels had dropped a lot and remained the same.

   
     

 

 

Mtgel classiﬁcation/prediction

 

Observed Remained the same Dropped “a lot” Percentage Correct
Remained the same 15 111 126
(Row Pct.) (12%) (88%) (12%)
Dropped “a lot” 12 432 444
(Row Pct.) (3%) (9 7%) (9 7%)
Model Predicted
(total) 27 543 570

Summary Statistics

Percentage of cases correctly classiﬁed 79%
Maximum chance criterion 74%
Proportional chance criterion 53%

 

4.3 Who was Affected by Low Water Levels (Study Objective #2)

This section reports the analyses focused on identifying signiﬁcant relationships
between water level related boating behavior changes and various independent variables
including: boat and boater characteristics, boat storage location and type, amount/volume
of boating days and boating locations. It also reports on the binary LOGIT model
formulated to predict boaters who are affected by low water levels. The data for the
analyses is taken from the 2001 Lake Michigan Potential Damages Study mail survey.
Approximately a third (33%) of the 1,386 recreational boaters, who responded to the mail

survey, were inﬂuenced by low water levels, 67% were not.

118

Hypothesis Testing
It was hypothesized that various boater and boat characteristics, boat storage
location and type, and boating volume and locations would inﬂuence whether they would

be inﬂuenced by low water level concerns. The hypotheses testing result is listed in

Table 4.15.

There is no statistically signiﬁcant relationship between low water responsive
behaviors and boater ages or incomes and also total number of boating days. Type and
size of boats are statistically related to whether a boater’s behavior is inﬂuenced by low
water levels. Generally the owners of larger boats, as would be expected, are more
inﬂuenced behaviorally by low water than small boat owners. Whether a boater’s
behavior is inﬂuenced by low water levels is signiﬁcantly related to where the boats are
stored during the boating season. A higher percentage of the owners whose boats were
stored at marinas were inﬂuenced by low water levels in large part because of dredging

problems in waterways accessing many Lake Michigan marinas.

Owners of boats stored in Michigan and especially in Lake Michigan counties
were also more likely to alter their behaviors due to low water levels. Again this is
consistent with studies showing that low water problems were more dramatic in Michigan
than other Great Lakes states. The behavior of boaters who boated more days on Lake
Michigan and the other Great Lakes were more likely to have been impacted by low
water levels. Also boaters who trailered their boats to boating locations were more likely
to be inﬂuenced by low water levels because many of the launch sites were negatively

impacted by low water levels.

119

Table 4.15. Relationships between boater and boat characteristics and whether a boater’s
behavior was inﬂuenced by concerns relating to low water levels.

   

 

 

Inﬂuenced xz/
by low water concerns F-value P-value
Yes No (df)
Boater characteristics
Age (Mean) 53 53 0.051(1) 0.822
Income
Under $60K 33% 67% 0.534(2) 0.766
$60 K-$99K 31% 69%
Over $100K % @019
32% 68%
Boat characteristics
Boat type
Inboard motor boat 35% 65% 23.434(4) 0.000
Inboard-outboards 37% 63%
Outboard motor boat 29% 71%
Powered sailboat 44% 56%
Pontoon boat 2_2_°/_o M
32% 68%
Boat length (Mean) 24.3 21.6 34.215(1) 0.000
Boat storage characteristics
Boat storage
Marina 44% 56% 29.549(3) 0.000
Waterfront permanent home 31% 69%
Cottage & second home 24% 76%
Non-waterfront home Mg M
32% 68%
Boat storage state location
Michigan 3 8% 62% 13 .404(3) 0.004
Wisconsin 27% 73%
Indiana 28% 72%
Illinois ﬂ 2329
32% 68%
Boat storage county location
Lake Michigan coastal counties 36% 64% 9.291(1) 0.00]
Other counties 21% w
32% 68%

120

Table 4.15. Relationships between boater and boat characteristics and whether a boater’s
behavior was inﬂuenced by concerns relating to low water levels — cont’d.

     

 

 

Inﬂuenced XZ/
by low water concerns F -value P-value
Yes No ((11)
Number of boating days (Mean)
Total boating days 32 32 0.000 (1) 0.998
Lake Michigan boating days 14 9 15 .97 1 ( 1) 0.000
Days on other Great Lakes 5 2 9.296(1) 0.002
”um?“ 0f different boating 2.2 1.7 25.328(1) 0.000
count1es/locations
The boat was trailered to launch
Yes 42% 58% 28.371(l) 0.001
No M ﬂ
32% 68%
Requested/received any information
concerning water levels
Yes 58% 42% 79.921(1) 0.000
No m 73%

 

Model Speciﬁcation

The results of the hypothesis testing identiﬁed a statistically signiﬁcant
relationship between whether they were inﬂuenced by low water levels and some
variables including (1) boat type, (2) boat length, (3) boat storage type, (4) boat storage
state location, (5) boat storage county location, (6) number of Lake Michigan boating
days, (7) number of boating days on other Great Lakes, (8) number of different boating
counties/locations, (9) whether the boat was trailered to launch, and (10) whether the
boater requested any information concerning low water levels. Those variables are then

used to develop the LOGIT model. Again the category variables (boat type, boat storage

121

type, boat storage state location, and boat storage county location) are recorded to

different dummy variables before entering the LOGIT model.

Information on 743 was available to develop the binary LOGIT model. The
stepwise-backward method formed alternative models, each with a different collection of
independent variables, which were determined by the result of hypothesis testing. The
ﬁrst model contained all the entire set of statistically signiﬁcant hypothesized
independent variables. The stepwise process eliminated “boat storage type: waterfront

,9 ‘6

permanent home, number of boating days on other Great Lakes and on Lake
Michigan,” “boat type - outboards,” “number of years the boat has been owned,” “boat

storage type — marina,” and “boat storage location: Illinois and Wisconsin.”

The seven remaining signiﬁcant independent variables incorporated as part of
ﬁnal LOGIT model are (1) boat length (at the 0.05 level), (2) boat type: inboard-
outboards (at the 0.05 level), (3) boat storage type: non-waterfront home (at the 0.10
level), (4) state where theirs boat are stored: Michigan (at the 0.105 level), (5) number of
different boating counties/locations (at the 0.05 level), (6) whether information
concerning water levels was requested/received (at the 0.05 level), and (7) whether they

trailered their boats to launch (at the 0.05 level).

122

LOGIT Model Parameter Estimations
The estimated LOGIT model is (Table 4.16):

log it(p) = —2.504 + 0.028LENGTH — 0.456INOUT + 0.529NWH + 0.42 11141
+ 0.135NDBC + 0.979REQUEST + 0.546TRAIL

P = The probability that a boater who was inﬂuenced by low water levels
LENGTH = Boat size

INOUT = Boat type — Inboard-outboards

NWH = Boats were kept at non-waterfront homes

M1 = Boat storage state location - Michigan

NDBC = Number of different boating counties/locations

REQUEST = Requested/received any information concerning water levels

TRAIL = Boat was trailered to launch

As would obviously be expected the owners of bigger boats are more likely to be
inﬂuenced by low water levels. Owners of inboard-outboards are just somewhat less
likely to be inﬂuenced by low water levels. This result is consistent with the result from
previous model. The boaters who owned inboard-outboards are not only less likely to be
inﬂuenced by low water level but also less likely to assess Lake Michigan had dropped “a
lot” because most inboard-outboards are smaller than other types of boats (e. g., sailboat,

and inboards).

123

Owners of boats stored at non-waterfront homes during the boating season are
more likely to be inﬂuenced by low water levels. Boaters who kept their boats at non-
waterfront home are 1.698 times more likely to be inﬂuenced by the low water level
concerns than the boaters who did not keep their boats at non-waterfront homes. This
result is in part because boats kept at non-waterfront homes must be trailered and

launched and as previously stated many launch sites were low water impacted.

This result is also conﬁrmed by the model ﬁnding that boaters who trailered their
boats to operate on Lake Michigan are 1.73 times more likely to be inﬂuenced by the low
water level concerns than the boaters who did not trailer their boats to operate on Lake
Michigan. According to the investigation of the 2001 LMPDS, eighteen ramps of the 55
ramps located in Allegan and Ottawa counties in Michigan and Manitowoc, Ozaukee and
Sheboygan counties in Wisconsin were closed due to the drop or little water for a sizable
boat (Planning & Zoning Center, Inc. et al., 2001). Moreover in 2002, thirty-ﬁve of the
134 public launch ramps located in 34 Lake Michigan coastal counties in Michigan,
Wisconsin, Indiana and Illinois were closed due to the same reason (Planning & Zoning

Center, Inc. et al., 2002).

The results of the LOGIT model also show that owners of boats kept in Michigan
during the boating season were signiﬁcantly more likely to be inﬂuenced by the low
water level concerns than the owners who kept their boats in Wisconsin and Illinois. The
more different locations a boat is Operated during the boating season, the more likely the

owner/operator will be inﬂuenced by low water levels.

124

Interestingly, boaters who requested/received any information concerning water
levels are more likely to be inﬂuenced by the low water concerns than boaters who did
not. However, the question is whether they request/received any information because
they are sensitive to low water levels, or they alter their behaviors because of some
combination of their heightened sensitivities and the information they receive. So boaters
who requested any information about low water levels are 2.662 times more likely to be

inﬂuenced by the low water level concerns than the boaters who did not.

Different examples showed different probability that boaters can be inﬂuenced by
the low water level concerns with different characteristics (Table 4.17). For a boater who
owned a 20-feet boat, the probability that he/she will be inﬂuenced by low water levels is
about 13% with the default values of other variables (not inboard-outboards, did not keep
the boat at non-waterfront, did not keep the boat in Michigan, did not request any
information concerning low water levels and did not trailer the boat to launch) (Example
1). The probability will decrease from 13% to 8% if he/she owned a 20-feet inboard-
outboard motor boat (Example 3). If this boater owned a 20-feet inboard-outboard motor
boat and also kept this boat at non-waterfront home in Michigan, the probability of being
inﬂuenced by the low water level concerns will be around 19% (Example 4). If this
boater trailered his/her boat to launch, operated this boat in three different locations and
also requested the information about low water levels, the probability of being inﬂuenced

by low water level concerns will increase to 62% (Example 7).

125

   

Table 4.16. The result of the ﬁnal binary LOGIT model for classifying boaters who were

 

s.E

inﬂuenced and not inﬂuenced by concerns relating to low water levels.

Wal

 

[3 df Sig. Exp(B)

Boat length 0.028 0.012 5.363 1 0.021 1.028
Boat type

Inboard-outboards -0.456 0.194 5.501 1 0.019 0.634

Powered sailboat -0.419 0.312 1.797 1 0.180 0.658

Pontoon boat 0.274 0.284 0.930 1 0.335 1.315
Boat storage type

Non-waterfront home 0.529 0.279 3.598 1 0.058 1.698
Boat storage state location

Michigan 0.421 0.169 6.183 1 0.013 1.524

Indiana 0.865 0.559 2.389 1 0.122 2.374
Boat storage county location

Lake Michigan costal counties 0.214 0.175 1.495 1 0.221 1.238
Number Of d‘ffemm boatmg 0.135 0.067 4.017 1 0.045 1.145
count1es/locat1ons
RequeStfd/rece‘ved any mfom‘at‘o“ 0.979 0.212 21.278 1 0.000 2.662
concemmg water levels
The boa“ “(as "3116er to 13mm“ to 0.546 0.206 7.023 1 0.008 1.727
Lake M1ch1gan
Constant -2.504 0.625 16.040 1 0.000 0.082

Summary Statistics
Model chi-square [d1] =85.92[11]m

Nagelkerke R2: 0.15
% Correct predictions=69.0%

* P—value<0.05

126

 

 

 

127

$2 $2 .xa $2 Q a
2.3 E _ .o 83 m: .o 93:»
83. ES- 3. $8. _- 88 a
$3. _ 33- _ 32- _ 82- _ «mod 33- 6888
o o o o o o o o 8: $3 882 8 E8828 83 88 9:.
m_o>o_
o o o o o o o o 83 mad 883 E55888 88882
EE noioootnoomoscom
. . mango—$2238
o o o o o o o o m: _ m: o E538 688% E0 8282
E3 _ o o o o o o 82 SE 98.392 a as. 83 8m
. . . . 08%
an o _ ENE o _ o o o E SE _ ENE o 28.2938: 3 £8. 8; 38
ESE- _ on»? _ ESE- _ o o «SE 8%. 888886885
Ewe om one om one om one 8 ENE ESE 582 8m
£35 £33 omega £38m $3 a
:Ewmnoaz 5 oEon oEon EoﬂbEBéoo
EoﬁEEBéo: E £8. E 38. EB nan? Eon SSE oEonEo
EEB noEB .Eon SSE .Eon SSE EEonEo -oEonE .3 E Eon .om E
@255 on? EEon <
88883385 .8 -2895 .8 a cacao 2? 888 <
E E256 on? EoEon < @236 on? EoEon <
v 2955— n SQEEEH N o-ﬂEExm— g 29:36—
a

.monmtouoEEno
EoEon BEE—om .3 £32 BE? 32 on onE 3538 an Boobs—«E 803 on? EoEon o5 now bEnEnoE onh .N. _ .v BER.

EC 3

 

128

 

 

$8 $2. EEN
ES. EEEE NEEE $33
EEE EEEE- EVE..- 8.8 E.
..EEN- . ..EEN- . EEEN- . NEEE «EN. .8380
SE . E E E E ENE SE 855. s 8.2.8 83 as. 2....
E_o>o_
ENEE . ENEE . E E NEEN ENEE 8:3 E58880 8.355...
3E oozoooabmoocom
. . . . . mooEoo<Eo5§oo
EEE E E E. E E EEE E E E... . EE. E EEEEEE ass? .a .8837.
.NEE . .NE . .NEE . ENE. .NEE 838.2 E. as. :3 8m
. . . . . 2:2.
ENE E . ENE E . ENE E . NEE . ENE E 888.83-an a as. 8 3 8m
EEEE- . EEEE- . EEE- . EEEE ESE- 8.888-283
EEE EN EEE EN EEE EN ENE. ENEE 5E8. aom

caaxm x .anxm x cavern. x aaxm a
E_o>o_ E_o>o_ ooE3 3o_ EonE

EconEoo_ anotno
3:: E @2823
E3 9E :Ewanz
5. oEon EoocnoE3
-ooo E Enou— E3 523
.Eon oEonEoéEonE E

EoE3 3o. EonE coEEEomE
o5 noomoscoa Eoc3o Eon
o5 9E Ecocho. Eooobzo ooh: E
oEEooo o. @2385 EE3 Eon on...
.5359. E oEon EoEtoE3éo:
E 38. .Eon onnEoéEon—n
.om E ooo3o on3 ..oEon <

b oEEExm

5:958... o5 ooEEoodo.
ocE mooEoB Eooobﬁ
ooh: E Eon o5 ooEoooo
EoEon £5 ncE .cEwEﬁ—z
E oEon EoEtoE3éo:

E Boo— .Eon oEonEoéEonE

.8 E oogo on3 ooEon < .om noo3o on3 EoEon <

e oEEExm m 39:83—

.PEoo I EonEtoEEEno
EoEon noEBoE En E_o>o_ ooE3 3o. on wEEE $588 an nooooomﬁ ooo3 on3 EoEEon ono o8 bEnEnoE of. .N. _ .v o_nE._.

LOGIT Model Evaluation and Correct Prediction

According to model chi-square statistic, the ﬁnal LOGIT model is signiﬁcant at
the 0.01 level in terms of overall model ﬁt (Table 4.16). The Nagelkerka R2 for ﬁnal

LOGIT model is around 15%, an acceptable number in social science research.

The classiﬁcation presented in Table 4.18 shows how well the model predicts
whether boaters were inﬂuenced (e. g., their behaviors) by low water level concerns. The
LOGIT model correctly classiﬁes 446 boaters (90%), who were not inﬂuenced by low
water level concerns and misclassiﬁes 47 of these boaters (10%). On the other hand,
model 9 correctly classiﬁes 67 (27%) of the boaters inﬂuenced by low water level
concerns but misclassiﬁes 183 boaters. The overall accuracy of classiﬁcation is the

weighted average of the two values - 69%.

This model only increase 2% in correct prediction by comparing the difference
between the values of overall model correct prediction (69%) and maximum chance
criterion (67%). But this model can still adequately predict whether boaters were
inﬂuenced by low water level concerns because the overall model correct prediction

percentage (69%) is still higher than the value of proportional chance criterion (56%).

129

Table 4.18. Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who were

inﬂuenced and not inﬂuenced by concerns relating to low water levels.
m

Model classiﬁcation/prediction

 

Inﬂuenced by low Not inﬂuenced by Percentage

Observed water levels low water levels Correct
Inﬂuenced by low water 67 183 250
levels 270/ o 27y
(Row Pct.) ( a) (73 A) ( a)
Not inﬂuenced by low 47 4 4 6 493
water levels [0‘7 900 900/
(Row Pct.) ( 0) ( A) ( a)
Model Predicated (total) 114 629 743
Summary Statistics
Percentage of cases correctly classiﬁed 69%
Maximum chance criterion 67%
Proportional chance criterion 56%

 

4.4 Recreational Boater Behavioral Changes Resulting from Low Water

Levels (Study Objective #3)

This section reports on the analyses employed in the development of two binary
LOGIT models developed to explain recreational boater behavioral changes resulting
from low water levels. The models utilized data generated by the 2001 Lake Michigan
Potential Damages Study mail survey. Only the data from 451 boaters who actually
experienced low water related problems on Lake Michigan is used to formulate two
binary LOGIT models, one to predict/classify boaters who changed locations as a result
of low water problems and the other to predict/classify boaters who boated less when

they encountered these problems.

130

4.4.1 LOGIT Model to Predict/Classify Boaters Who Change Their Boating

Locations in Response to Low Water Levels

More than three quarters (79%) of the 451 boaters, who were inﬂuenced by low

Lake Michigan levels, changed their boating locations.

MW

Statistical tests on a set of various hypothesized independent variables determined
that age was the only socio-economic characteristic signiﬁcantly related to whether
boaters changed where they boated in response to low water levels (Table 4.19). Older
boaters are less likely to change their boating locations. In part this was because most of
them stored their boats at second homes and cottages making a move very difﬁcult.
Younger boaters are more likely to store their boats at their permanent homes and trailer

them to various boating locations.

The length and type of boats and number of years the boat has been owned are
also related to the likelihood that an owner will change the boating location in response to
low lake levels. Powered sailboats are most likely to change locations due to low water
levels. Because rarely are they ever operated on the Great Lakes, owners of pontoon
boats are less likely to change locations in response to low water levels. Contrary to
other LOGIT models, state and county (e. g., Lake Michigan county) where the boats
were stored during the boating season does not statistically signiﬁcantly change the
likelihood that their owners will change their boating places. Boaters with their boats
kept in Michigan during the boating season are still more likely to change boating

locations in response to low Lake Michigan water levels.

131

Amount of boating days is related to whether a boater changes locations in
response to low water levels. More frequent boaters, and boaters who boat more often on
the Great Lakes and in more different locations/counties are statistically more prone to
change where they boat in response to low water levels. Moreover boaters who requested
or received any information concerning water levels or do not own a waterfront home are

more likely to change their boating locations.

Table 4.19. Relationships between boater and boat characteristics and whether boaters
changed their boating locations in response to low water levels.

 

 

 

 

 

Changed boatinglocations 78/ P-
Yes No F-value(df) value
Boater characteristics
Age (Mean) 51 58 15.478(1) 0.000
Income
Under $60K 33% 67% 1.158(2) 0.560
$60K to $99K 31% 69%
Over S 100K M w
32% 68%
Boat characteristics
Boat type
Inboard motor boat 35% 65% 13.6 1 3(4) 0.009
Inboard-outboards 36% 64%
Outboard motor boat 29% 71%
Powered sailboat 44% 56%
Pontoon boat 22% M
32% 68%
Boat length (Mean) 25 21 6.108(1) 0.014
Number of years the boat has 9 ll 6.960(1) 0009

been owned (Mean)
m

132

Table 4.19. Relationships between boater and boat characteristics and whether boaters
chan ed their boatin ; locations in res .onse to low water levels — cont’d.

    

 

 

 

Changed boating locations 967 P-
Yes No F-value(df) value
Boat storage type
Marina 44% 56% 6.824 (3) 0.078
Waterfront permanent home 31% 69%
Cottage and second home 24% 76%
Non-waterfront home M M
32% 68%
Where boat is stored during season
Michigan 38% 62% 1.513 (3) 0.679
Wisconsin 27% 73%
Indiana 28% 72%
Illinois 2% 233/9
32% 68%
Boat storage county location
Lake Michigan costal counties 36% 64% 0.028(1) 0.498
Other counties 28_°/9 12%
32% 68%
Boating days & locations (Mean)
Total number of boating days 33 25 4.378(1) 0.037
Lake Michigan boating days 16 6 11.420 0.001
gfgfﬁisbwmg days on Other 4.4 2.4 1.514(1) 0.219
Number of different boatin
counties/locations g 2'3 1'5 19'855( 1) 0000

Requested or received any information
concerning water levels

Yes 58% 42% 4.593 0.020
No M 7_2°/_o
32% 68%
Own a waterfront home
Yes 27% 73% 6.724(1) 0.008
No me 93%.
32% 68%

133

mm

Variables having signiﬁcant relationships with boaters who changed and did not
change where they boated for use in the LOGIT model are (1) age, (2) boat length, (3)
number of years the boat has been owned, (4) number of different boating
counties/locations, (5) number of boating days, (6) boat type, (7) whether the boat owner
requested any information concerning low water levels, and (8) whether the boat owner

owns a waterfront home. The default category for boat type is pontoon boat.

The stepwise- backward begins with those independent variables described above.
Seven variables have signiﬁcant effects on the probability that a boater changed his/her
boating locations in response to low water levels. Those variables are number of years
the boat has been owned (at the 0.01 level), number of different boating
counties/locations (at the 0.01 level), requested/received any information concerning
water levels (at 0.10 level) and boat types: inboards (at the 0.01 level), inboard-outboards

(at the 0.01 level), outboards (at the 0.01 level) and sailboat (at the 0.01 level).

LOGIT Model Parameter Estimations
The estimated LOGIT model is (Table 4.20):

log it(p) = -143.795 + 0.072YEAR + 0.471NDBC + 0.782REQUEST + 1.4 141NBOARD
+1.217INOUT +1.8160UTBOARD +1.996SAIL

P = The probability that a recreational boater will change the boating locations in

response to low water levels
YEAR = Number of years the boat has been owned

NDBC = Number of different boating counties/locations

134

REQUEST = Requested/received any information concerning water levels

INBOARD, INOUT, OUTBOARD, & SAH. = Boat types

The type of boats owned by boaters is signiﬁcantly related to the likelihood that
they will move “where they boat” because of their concerns about low water levels. The
owners of sailboats are much more likely (7.4 times) to move locations in response to
ﬂuctuating water levels than the owners having pontoon boats. As would be expected,
owners of pontoon boats are unlikely to move where they boat in large part because these

boats are rarely found on the Great lakes and they are not very transportable.

Boaters who boat at more locations/counties are more likely to change boating
locations to avoid problems associated with low water. The more different counties
where they operate their boats, the more likely they are to move where they boat in
response to low water levels. The probability that they will change their boating
locations increases 1.6 times for one more different county/location where they boat. A
majority of these boaters, who operated their boats in more than four different
counties/locations, own outboards (average size 22 feet) that can be trailered to different
locations, and many of them stored their boats at permanent homes (around 65%) and
trailered (74%) them to boating locations rather than keeping them at marinas or second
homes.

Boaters who requested information about low water are also more likely to
change their boating locations. The boaters who requested any information about low
water levels are 2.186 times more likely to change their boating locations in response to

low water levels than the boaters who did not request the information about low water

135

levels. Again, it is not clear whether these boaters changed boating locations because

they obtained this information, or they requested the information because they are just

more sensitive and responsive (including changing boating locations).

Table 4.20. The result of the ﬁnal binary LOGIT model for predicting/classifying boaters

   

who changed boating locations in response to low water levels.

 

13 S.E Wald df Sig. Ex(p)
N‘m‘be’ °f years the boat has 0.072 0.023 9.930 1 0.002 1.074
been owned
Numlier 0f d‘ffemm boatmg 0.471 0.164 8.221 1 0.004 1.601
count1es/locat10ns
Requested/received any
information concerning water 0.782 0.414 3.566 1 0.059 2.186
levels
Boat type
Inboard motor boat 1.414 0.527 7.198 1 0.007 4.1 11
Inboard-outboards 1.217 0.457 7.081 1 0.008 3 .378
Outboard motor boat 1.816 0.540 1 1.305 1 0.001 6.147
Sailboat 1.996 0.728 7.511 1 0.006 7.361
Pontoon - - 14.900 4 0.005 -
Constant -143.795 45.462 10.005 1 0.002 0.000

Summary Statistics
Model chi-square [df] =44.048[1 l]’
Nagelkerke R2: 0.212

% Correct predictions=85%

W

* P-value < 0.05

136

LOGIT Model Evaluation and Correct Prediction

According to model chi-square statistics reported in Table 4.20, the ﬁnal LOGIT
model is strongly signiﬁcant (x2=44.048). The value of Nagelkerke’s R2, the measures of

the proportion of variation explained is around 21% for the ﬁnal LOGIT model.

The percentage of correct predictions of whether a boater would change where
they boat in response to low water levels is reported in Table 4.21. The LOGIT model
has an overall prediction accuracy of 85%. As for previous models, correct
classiﬁcations are shown in the diagonal cells and incorrect classiﬁcations in the off-
diagonal cells. This binary LOGIT model correctly classiﬁes 98% of boaters, who
changed boating locations due to their experience with low water levels but it only
correctly distinguishes 13% of boaters who did not change boating locations. It
incorrectly classiﬁes 2% of boaters who changed their boating locations as not changing
locations and 87% of boaters who did not change their boating locations as changing their

boating locations.

Compared to the maximum chance criterion—79%, the percent of correct
classiﬁcation of the LGOIT model resulted in an increase of 6%. The percent of correct

classiﬁcation (85%) is also higher than the proportional chance criterion (70%).

137

Table 4.21. Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who changed and

did not change their boating locations in response to low water levels.
m

Model classiﬁcation/prediction

 

Changed the Dld not change Percentage
boatin location the boating C t
Observed g location orrec
Changed the boating 277 5 232
location
(98%) (2%) (98%)
(Row Pct.)
Did not change the boating 46 7 53
location (8 7‘7) (13 °/) (13 V)
(Row Pct.)
Model Predicted (total) 273 12 285
Summary Statistics
Percentage of cases correctly classiﬁed 85%
Maximum chance criterion 79%
Proportional chance criterion 70%

 

4.4.2 LOGIT Model to Predict/Classify Boaters who Reduce their Boating in

Response to Low Water Levels

Of 451 boaters, who were inﬂuenced by low water levels, over half (54%) did less

boating in response to low water levels.

Hypothesis Testing
The same independent variables used to develop the LOGIT model that
classiﬁes/predict boaters who change boating locations in response to low water levels

are also used to formulate this model.

138

The result of the hypotheses testing identiﬁed a statistically signiﬁcant
relationship between whether a recreational boater will reduce the boating days in
response to low water and their ages, number of years, a boat has been owned, total
number of boating days, number of Lake Michigan boating days, number of different
boating counties/locations, whether a boater requested/received any information

concerning water levels, and whether a boater owns a waterfront home (Table 4.22).

Again age is the only socio-economic variable related to whether boaters reduced
their boating (days) in response to low water. Boat type is not statistically signiﬁcantly
associated with reduced boating levels in reaction to low water levels. The number of
years the boat has been owned and boat length are statistically related to a low water
induced decrease in boating activity. The boaters who reduced their boating days have
owned their boats longer than the boaters who did not reduced their boating days. This

could in part be attributable to the fact that they are “older boats.”

During the boating season boat storage location (state or county) and type of
storage (e.g., marina, waterfront permanent home, second home and non-waterfront
home) are both not statistically signiﬁcantly related to a propensity to reduce boating in
response to low water levels and related problems. However, a greater proportion (about

44%) of the owners of boats kept at marinas reduced their boating.

Less active boaters, that is boaters who do less days of boating, and boaters who
boat less days on Lake Michigan and other Great Lakes are more likely to reduce their
boating as one response to low water levels. In all probability due to the fact that more
frequent boaters are more experienced and knowledgeable about lake conditions

including waste level and its impact on boating access and quality. Or, it may be due to

139

the fact that frequent boaters are more enthusiastic and just will not let water levels get in

the way of their boating.

Table 4.22. Relationships between boater and boat characteristics and whether boaters
did less boating in response to low water levels.

 

 

 

 

Did less boating xz/ P-
Yes No F-value(dt) value
Boater characteristics
Age (Mean) 54 50 6.494(1) 0.001
Income
Under $60K 33% 67% 0.330(2) 0.330
$60K to $99K 31% 69%
Over $100K E4; 6%
32% 68%
Boat characteristics
Boat type
Inboard motor boat 35% 65% 5.517 (4) 0.238
Inboard-outboards 36% 64%
Outboard motor boat 29% 71%
Powered sailboat 44% 56%
Pontoon boat 22°19 M
32% 68%
Boat length (Mean) 22 24 3.550(1) 0.055
Number of ears the boat has been
owned (Mean) 10 8 3.400(1) 0.066
Boat storage characteristics
Boat storage
Marina 44% 56% 2.399(3) 0.494
Waterfront home 31% 69%
Cottage and second home 24% 76%
Non-waterfront home 2_9_‘Zq M
32% 68%

140

Table 4.22. Relationships between boater and boat characteristics and whether boaters
did less boating in response to low water levels - cont’d.

 

 

 

 

Did less boating xz/ P-
Yes No F-value(df) value
Boat storage characteristics
Boat storage state location
Michigan 38% 62% 1.926(3) 0.588
Wisconsin 27% 73%
Indiana 28% 72%
Illinois 2% M
32% 68%
Boat storage county location
Lake Michigan coastal counties 36% 64% 0.910(1) 0.200
Other counties M 121/9
32% 68%
Boating days & locations (Mean)
Total number of boating days 26 37 13.689(l) 0.000
Number of Lake Michigan boating days 10 16 10.638(l) 0.001
Egrer of boating days on other Great 3.5 4. 8 0.194(1) 0.660
21:33:? different boating counties 19 2.4 7.717(1) 0. 006
Requested/received any information
concerning water levels
Yes 58% 42% 2.929(1) 0.054
No M 72%
32% 68%
Own a waterfront home
Yes 27% 73% 7.020(1) 0.005
N0 3% M
32% 68%

141

MRI—6W

Variables having statistically signiﬁcant relationships with the awareness of water
level ﬂuctuations for use in the binary LOGIT model are ( l) boater’s age, (2) boat length,
(3) number of years the boat has been owned, (4) total number of boating days, (5)
number of different boating counties/ locations, (6) boating days (e. g. number of boating
days on Lake Michigan and on other Great Lakes), (7) whether the owner requested or
received any information concerning water levels, and (8) boat owner owns a waterfront

home.

The stepwise-backward method begins with those independent variables
described above. Then variables eliminated through the stepwise-backward models are
“age,” “type of boat,” “number of years the boat has been owned,” “number of boating

days on other Great Lakes,” and “whether they requested/received any information

concerning water levels.”

Five different variables have signiﬁcant effects on the probability that a boater
will reduce the volume of boating days in response to low water levels. The ﬁve
variables are boat length (at 0.01 level), total number of boating days (at 0.05 level),
number of different boating counties/locations (at 0.10 level), number of Lake Michigan
boating days (at 0.10 level), and whether the boat owner also owns a waterfront home (at

0.05 level).

142

LOGIT Model Parameter Estimations

The estimated LOGIT model is (Table 4.23):

log it(p) = —0.350 + 0.044LENGTH — 0.012NBD - 0.159NDBC — 0.017NBDLM + 0.5890WH

P = The probability that a recreational boater will reduce the volume of boating

days in response to low water level

LENGTH = Boat length
NBD = Number of boating days

NDBC = Number of different boating counties/locations
NBDLM = Number of Lake Michigan boating days

OWH = The boater owns a waterfront home

The results show that more active boaters are less likely to respond to do less
boating in response to low water levels. Every additional boating day reduces the
probability that a boater will decrease their boating in response to low water levels. So
one more day a recreational boater operate his/her boat, the boat owner is 0.988 times

less likely to reduce the amount of his/her boating days in response to low water levels.

Boaters who boat in more different locations are also less likely to reduce boating
days when confronted with low water levels. Possibly, these boaters are aware of and
have more experiences at different locations and therefore have more alternatives
available to them. One more different county/location a recreational boater operates
his/her boat, the probability that the boat owner will reduce the amount of his/her boating

days will decrease around 4%.

143

The model also indicates that boaters who own waterfront homes are more likely
to reduce the number of days they boat because of low water. Boaters who own
waterfront homes are almost 2 times higher to reduce the amount of boating days in
response to low water levels than boaters who do not own waterfront homes. This is
most likely because they are less able to move their boating to other locations. According
to the result of previous LOGIT model, this variable is not related to the prediction of
whether boaters will change their boating locations in response to low water levels. For a
boater who owns a waterfront home, the possibility of reaction on reducing the boating

days is signiﬁcantly higher than the reaction on changing boating locations.

The equation produced by the LOGIT model can be used to predict the odds and
the probability of a recreational boater who did less boating in response to low water
levels for given values of those independent variables. Table 4.25 illustrates the
calculation of probabilities that recreational boaters who did less boating in response to
low water levels. If the boater owned a 20-feet boat, the probability for this boater to
reduce the boating days in response to low water level is 63% with the default values (0)
of four other variables (number of boating days, number of different boating locations,
number of Lake Michigan boating days and own a waterfront home). The probability
decreases to 45% if the boater operated the boat for 60 days during the boating season. If
this boat was Operated in three different locations, the probability will change from 45%
to 30%. The probability will be 48 % if the boat owner has all of the earlier

characteristics and also has a waterfront home.

144

Table 4.23. The result of the ﬁnal binary LOGIT model for predicting/classifying boaters

who did less boating locations in response to low water levels.

 

 

0 SE Wald df Sig. Exp(B)
Boat length 0.044 0.016 7.860 1 0.005 1.045
Number of boating days -0.012 0.005 5.775 1 0.016 0.988
Numb.” °f d‘tjferem boatmg -0159 0.096 2.753 1 0.097 0.853
counties/locations
Number Of Lake M‘Ch‘ga“ -0017 0.009 3.685 1 0.055 0.983
boatrng days
Own a waterfront home 0.589 0.262 5.069 1 0.024 1.802
Constant -0.350 0.375 0.873 1 0.350 0.704
Summary Statistics

Model chi-square [df] =31.244[5]‘
Nagelkerke R2: 0.126

% Correct predictions=64%

* P-value < 0.05

145

 

 

 

$3. $3. $3 :8 Q E
ENEE E_E.E ENE 83 EEK
wnod- $90- on .o- mmd mono E
mmd- _ mmd- _ mmd- _ mmd- _ mend mmd- EEEcoU
owmd n o o o o o 0 New; mwmd oEon EoﬁooEB E 850
. . - EEE wEEon
E E E E E E E E EEE E :E E 58522 83 E6 882
. - . - . . - moonEooimoccsoo
E. E E E. E E E E E E EEE E EEE E 888 888% .8 88:2
and- on mud- on and- on o o wwod N56- 3% wEEon mo nonESZ
wwd om wwd om wwd om wwd om mvon god n&co_ Eom
898 x caves. x 89.5 caves 38.8 E
oEon EoEtoEB .888
E nogo ocE :oEEom wEEon o5 moron oomEom wEEon
wEEon o5 moron Eon on noz— mconEoE o5 moron 9E“. on
88:82 Eonobno ooh: Econoccno ooh: now Eon o5 ooEnomo Eon .om
E EzEn on no.“ Eon E Eon o5 ooEEooo 9E Eon .3. E 226 on? .oEon <

o5 ooEEooo .Eon .om
E £56 on? EoEon <

V O—maﬂxnﬂ

 

.monECoEEEno EoEon nonoﬁom En E_o>o_ EEB Bo. on omoommon E wEEon 83 En on? EoEon onn now bEnEnoa onh .vmé 29$.

28 88 .EN

a wank/O 0:3 hoamon <

m o_mEExH

E 326 on? EoEon <

N 29:85—

m oEEExH

6

4

l

LOGIT Model Evaluation and Correct Prediction

According to model chi-square statistics reported in Table 4.23, the ﬁnal LOGIT
model is strongly signiﬁcant (x2=3 l .244). The value of Nagelkerke’s R2, the measures of
the proportion of variation explained is around 13% for the ﬁnal LOGIT model. The low
R-square indicates there is a need for adding more other independent variables to

improve this model’s performance.

The percentage of right classiﬁcation/prediction of whether recreational boaters
will reduce their boating days in response to low water levels substantiates the predictive
power of the ﬁnal LOGIT model, which yields a hit ratio of 64% with correct
classiﬁcation in the diagonal cells and incorrect classiﬁcations in the off-diagonal cells
(Table 4.25). It correctly classiﬁes 67% of boaters, who did less boating because of low
water levels, and 61% of boaters who did not respond to low water levels by reducing
their boating activity level. However, the model incorrectly classiﬁes 33% of boaters
who did less boating as boaters who did not reduce their boating activity levels.
Conversely, 39% of boaters who did not reduce their boating because of the low water

were incorrectly classiﬁed as having reduced their boating.

Compared to the maximum chance criterion-54%, the percent of correct
classiﬁcation of the LGOIT model resulted in an increase of 10%. The percent of correct

classiﬁcation (64%) is also higher than the proportional chance criterion (50%).

147

Table 4.25. Accuracy of ﬁnal LOGIT model classiﬁcations of boaters who did and did

 

 

 

Did less boatin Did not do less Percentage
Observed g boating Correct

Did less boating 106 52 158
(Row Pct.) (6 7%) (33%) (6 7%)
Did not do less boating 61 96 157
(Row Pct.) (39%) (61%) (61 %)
Model Predicted (total) 167 148 315
Summary Statistics

Percentage of cases correctly classiﬁed 64%

Maximum chance criterion 54%

Proportional chance criterion 50%

 

148

CHAPTER 5

CONCLUSIONS

This ﬁnal chapter is presented in four sections: (1) a summary of the principal
study ﬁndings from the ﬁve binary LOGIT models, (2) the implications and applications
of the results from the data analyses and ﬁve binary LOGIT models, (3) the limitations of

the study, and (4) the recommendations for future research.

Boating is an extremely important recreation activity and tourism industry in the
Great Lakes. In 2001, over 2 million recreation watercraft were registered in Michigan,
Wisconsin, Indiana and Illinois. Of these boats, over 370,000 were registered to owners
in 34 Lake Michigan coastal counties. Recreational boaters support a major boating
industry and bring the positive economic impact including $760 million from boating
trips and $635 million from boat purchases in 2002 for Michigan. There is a mounting
concern relating to a host of challenges confronting recreational boaters and the boating
industry including efforts to limit access to boating waters, regulations limiting the
development and improvement of boating facilities and services, insufﬁcient funding of
state boating programs, lack of a comprehensive boating plan, and ﬂuctuating Great
Lakes water levels. Great Lakes water levels ﬂuctuate over time because of the
differences in precipitation, temperature and other climatologic factors. However, over
the past ﬁve years, Lake Michigan-Huron are experiencing the lowest water levels in 35
years. This fact, in combination with other factors, is having a major impact on marine

businesses, especially many small marinas that service boaters on the Great Lakes.

149

A number of recent studies focused on the actual revenue and cost impacts of
ﬂuctuating Great Lakes water levels. These economic and ﬁnancial impacts were caused
in part by changes in boater behavior (e. g., reduced boating) in response to low Great
Lake water levels about the affects of boater behaviors. This study instead focused on
creating a better understanding whether and how boaters perceive Lake Michigan water
levels and how these perceptions affect their boating. Five binary LOGIT models were
estimated in order to assess and determine various factors, which inﬂuence recreational
boaters’ awareness and assessment about water level changes. The LOGIT models also
assess the relationship of various boat and boater characteristics in determining how
boaters perceive, are inﬂuenced by, and if and how they adjust their boating (e. g.,

amount, locations) in response to their awareness of low Lake Michigan water levels.

5.1 Summary of Major Findings

The main purpose of this study was to develop a series of LOGIT models to better
understand the relationship between boater and boating characteristics and awareness,
perceptions and responses to low Lake Michigan water levels. The ﬁve binary LOGIT
models that were developed and evaluated are summarized in Table 5.1. The summary
includes the relevant study objective, data sources, number of boaters included in the
model building analyses, dependent variables, the number of stepwise LOGIT models
that were formulated, signiﬁcant independent variables comprising the ﬁnal models, and
the prediction percentage for each LOGIT model. The parameters and P-values for the
signiﬁcant independent variables incorporated in the ﬁve ﬁnal LOGIT models are

presented in Table 5.2.

150

 

oEon EomtoEB

mozm noSE. o. 33:8...

oE mEon 852.36.
:ocEEEoE ..oEB Bo.

 

 

 

 

 

 

 

 

E ago 5:26 Eom .w .8338 no @8838
53958:. ..oEB 3o. ..oEso o5 H8.35.7 .m
.8288 ..o .8838 comEoo. Among? ..oEB
EoEso o5 658.3 ... 2:2. .058 owEEoom Eom .w .wo. EEEE wEEomd
$8.5 E80 Eonuo co EoﬁoEB E oo§0.w .5382 5:82 $8.3 E80
BE .nEwEo.2 8.8. :ozEEEomE 8E? 3o. 88» own—9E Eom .N. .958 own—Sm Eoms nonoo so an 535.2
:o 33 wEEon .89 8288 .8 .ooEozcoE oMEEoE Eom .0 :oquo. 8E... no .w.ov ooEnoao
o5 .wov WEE wEEom .n Ego o5 858.3 .h $8.3 E80 Eonuo 88E owEooE Eomn 8o? EEon 88.? .w
328828.858 25 Eomd .8 BE .5359. 8E... o9: owESE Eom.m o3 Eom ...
wEEon ExEn moEon ..o EonEEZd :o 996 wEEon .89 38.8. E80 nnwco. Eom .n
acouobmv mo 33:52 .m mcomumoo<mo$§oo 05 .wdv 9A8 waﬁmom .m 550 no dummy—om: Donovan.—
meo w:.Eon wiEon mongoo wEEon 8E4 co .wov .8an Boa—Egon mo :0.“qu .m wagon
..o Eons—o: .88. .v EoEbE ..o EonEEZé EconobE ..o EonEEZ .v 8o? EEon 88.3 .v 3.5.58 momonoomzn
.856 coon «En Eon .ooEso coon mEn Eon noEso goon mEn Eon ooEso coon En Eon mo oEE quEEnom .v »n Beacon.
o5 88» ..o nonEEZ .m o5 PEPE ..o nonEEZd o5 go» ..o EonEEZ .m o5 88» ..o EonE=Z.m 8.280 .m wo.nE..E>
58». ESE .N 58.... 8m .N 25 86m .N 25 8m .N 8.888 .N 88888.
ow< .. ow<.. ﬁmco. Eom .. 5&5. Eom. oEooc. .. nag—mama
m.o>o. EoEB :Ew.no.$. 296. £26. no. E oommouo
8E... :6. on omcoamon EoEB 585.2 8.3 Bo. EEB EwEﬁE 8.8. En m.o>o. 5E3 :Ew.no.2 895.0
a. 3.33 wEEon on omcoamon a. 98:83 26. on 9528 858.8 8E... E5 .ooEEoEmE .o>o. BE? :Ew.no.2 mo.nE..E>
:8: 3268 ﬁoEom onon .5953 888m an Bozo—Ea. oE EoEom no .523qu ﬁoEom 8E: mo oEBE oE EoEom Eoocoaon.
EoEEon .mv EoEon .nv ﬁoEon own... 88Eon up... 8oEon 3%.. ooNHoWMw
.333 ..ES. Aotsm ..ES. .333 ..ES. xogm ooonmo.oh >255 oconQo.o... ooSom EEG
28:82 4% ENE mnEon .88 28:82 .EE 3:8:on mEno .o>o.
owEuoE Eon .mozmtoEoEEno EoEon .EE Eon .ocE EoEB tan Boson—E. on m.o>o.
Eochno .o>o. BE? An nooooocc. mochno .8o.>Enon on b8... .88 oE EoEon ..oEB :EwEoGn 8.3 a. moonno ..o onBEwEE o5 ..o
o.n.mmon. o5 :ooEon 3.55328 o5 amazon. o .. ..o womb no.3? bacon. o... .EwEaaE 9E mmocoEBE .EoEon .EooEonoo. 888 o ... 25830
E 3.8.25 N 828.3 _ 3.8.3

 

 

 

 

 

.E.oooE .5001. bEﬁn .86. BE? Ewing). 8.3 o>m o5 now $.38 Bugs—85m ...m o.nE.—.

151

 

 

 

 

 

 

 

 

 

A8388
.588 838.2 8.8.
:. ..wd. comEom mﬁEon
88 8.82 on 882.8 o5 98.8.. 88%
oE mEon 852:5... 8E mEon 8on3 .0580...
8888.8. 8E3 3o. 38.8.:
8388 8 .8888 do .EEBE 8.888.?
8856 o6 858.38 ..wov comEoE wa.Eon
.886... E. 826.2 65 E88 .0868 88.8.2
..wd. 888 MEEon 8E EEon 8on3 oEnmd 8.8. 8 ..wov ©8828
oﬁ 8.88 .888 A282. 8E mEon 88.36
8888.8. SE? 28. 8E EEon 8on3 oEEmd .888 an owEtoo ..wo. A8388
8388 8 .8888 .88.. owEEoE Eon .8 2.3.6 838.2 8E... on :8:
oEon .8888» 888 on. 852.3% 288895-82 ..wd. Goo—E... E80 a. 8.288 838.2
E 83o .oEso Eomd .8988 a. Eon..Em owEnoE Eon .8 8.3.4.. 85o 8 ..wd. .8888 8E... ..wd. 88.88
ExEo wEEon 888.2 £8888 £8888 E8.Eoo. 8E .88 88.36 82888. .8 888......
8.3 .8 88.526 -oEons. ..wov 85 Eom.m Bongoo wEEon .858 coon mEn Eon E865 .oooE
82.828.28.88 m8.Eoo.\Eo.8=8 888.5. .8 88:82..” o5 88» .8 88826 a. 8.888.? 68.8.2 .95 o5
wEEon w=.Eon €8qu CEon 898E. ..wov 88.88 8.8888
888.8 .8 Eons—52.». 888.... .8 89:52.... a. Eon..EE £8888 a. moEonq. $8888 .8 oEE 888$...” Eo.nE..E>
ExEv wEEon ..o ..onE:Z.m .858 soon 8.. Eon -.oEonc. ..wov o3 Eom.~ -.EEona. ..wov 85 Eom.~ nooao0.m 88.88.88.
5M8. Eom. o5 .8th ..o 8882.. 588. Eom. nwwao. Eom. 2885.. .Eowﬁwﬁ
2888
.5004
£888 a E.o.ooE n 2288 a $288 w 2888 w 88638
-8588
..o 88:82
E 8.8.30 N 258.30 _ 8.8.30

 

 

8.88 I £888 :00. 8:5 .o>o. 8E3 888.2 8.3 95 o5 8.. 8.88 BEEEEdm ...m 28...

152

 

 

 

 

 

 

 

 

 

 

 

. . . . .8...8
«8 o 8.. o to o ...... o 82. 8.. 8.. o... .88 E 88.52 .m
Eon 88.8.. Av
N8... 8.... 8.. ....m .m
:5... 3.... 8o... 2...... 882.... .N
.8... :N. as... 8.....- to... 22.. 8888.88... ..
8.3 Eom. .N
3.... as... .8... .8... 8.... NS... .88. 88. ..
monummhuuoahap—u .30”
8o... 2.... 8.88 .588 2. o. 88...... .N
8.... R... 8.258 .588 .2. ..
08.0.8.8. 808.88. .o 8.8.8. .v
2...... .8... , .8...... 2
N8... 2...... 5268.? .m
8.... 2.... 838.... 2
02.0039. .«0 0.8m “Soggom .M
o8... 8.8.00 .m
8.... 8.3- 88 o. 8.3 .N
882.. ..
momentouuahaﬂu heuacm
.8 a .m.m a ....m a Em a .8 a
.88.. .082
«5.03 9.03803 m..o.m0m @0253? .082 w_o>oxH .0>QH .033
wEEom. 3.63. wEEom. 09.8.0 .o>o. 8E3 30. BE 3 ... 8.0 .Ewﬁoa). 88...
30>». SE 3 26. £38. SE 3 3Q. .88. 2.8880 8.. .8 888.080.. .8 wmoroEB...
n 8.8.8 N 8.8.8 . 8.8.8

 

 

.983 .o>o. .oEB .Ew.no..>. 8.... on. 8.6.888 88088.... o.nE..E> 8088.8... 885.8... >..Eo..m.EE ..o 8883 < .~.m 03....

8.0.8.... ...001.

153

 

 

 

 

 

 

 

 

 

 

 

Rod :3- 3% wages 53:22 33 a
22. ~86. £8 wagon _soh 2
what wcuaom .N
. . . . . . 303303358
:8 o a: o- «8 o 5 o «3 o m: o wagon Samba Mo “3:52 ._
was wanna:
.3 hon—5:. can £82.82 «:53:
”mod RS 8258 3&8 59:22 8.3
cam—mom wagon 05
macaw @8on 8m 38s Eon? 5550 .m
Eod 3.3 «565 Am
83 85o 59885, a
23 53 «8.0 5M3 Swag: :
c938 wagon 05
mac—6 @2on 8a Smog 2053 88m .N
Sod EN..- 2.8; 2.83. a. owazoo AN
33. 32 use: EEthBéoZ C
omﬁowm «non mo 09$. 4
83283.35 «MP—3m :3:
ma a @m a Em a .wa a .3 a
“55:2 80:83 Eoumomwwwmguﬁ Eve—2 £054 B>W~vmwmm>>
wﬁgaom 335M wﬁaaom owﬁﬁo 354 883 33 BE? 5 QED gwﬁoaz 8:3
£264 .583 33 £33 33>? 33 Soon 5:350 05 mo Eongoeom .wo mmo=033<
m 3:8...5 N 3.8.25 ~ 228.25

 

 

SEQ :32 533 ammEBE 8:3 05 wcﬁtqﬁoo $8098.82 oﬁmtg «coucaovﬁ Emu—bums bﬁotmuﬁm no E883 < .N.m 2an

.Paoo I £085 .53

154

 

 

 

 

 

 

 

 

 

 

 

$8 $2 $8 $8 $8 :825 8:20 Reggae;
$$ $3 $8 $§ $2 8:85 852E 8:852
$$ $3 $3 $3 $~w gauge 380:8 was ‘8 28.8
$2 $8 $2 $2 $3 Na 8:33wa
E _ m c :3 z :8 E 8 En? c3 Beaming 382
and 83- ~85 ME ME Sod 33- 23 20.0 83 8v. _- .5325

. . 2:2
vmo o 3% o E95895 a .330 8850 “mama

. . 3% £252

woo o 3m o 8 nob—Eb 8m 38s .oﬁonz/N

. . . . nouns—~85 .883 32 3382
mmo o mwh 0 COO o aha o .5 6030570“ Her—>90 0:“ uoﬁogé
Stan—~85 .3 89¢ .350

$3 33. 8x3 .85 55o 5 AN

83 £2 53:22 8.3 co 2
E5830 0.3 $89 22:5 .m
99:. ”£225
no nun—E:— EE 28:33 ”—5.25

.me a .ma a .ma a .3 a .wmm n
382
“55:2 $5383 Euaaom 325:9: $602 233 E62 354
mﬁuaom 335M wcgom omSEU Esq 33>? 33 .883 E QEQ .833 53:22
£33 833 33 £33 86>? Boa Sons 3.5250 05 mo mcocaoouum BRA mo mm0533<
m «Ease N «Essa _ 228:5

 

 

>535 .o>o. 883 53:32 834 on“ wEmEQEOo $8028an 033.?» “concomocﬁ Eggnwi biocmtﬁm mo 5853 < dd 2an

.PEOU I 20on ,EOOA

155

The LOGIT Models and Related Findings

The ﬁrst LOGIT model reveals that boat owners, who have higher incomes (more
than $60K), register their boats in Michigan and especially in Lake Michigan coastal
counties, and operate their boats on Lake Michigan, are much more likely to be aware of
Lake Michigan water level changes. This could be anticipated because the water level
situation on Lake Michigan received a signiﬁcant amount of media coverage, which is
more than the other three states. Also, and as will be discussed later, previous studies
including the Lake Michigan Potential Damages Study indicate that low water levels had
a more dramatic impact on boating facilities on Michigan’s Lake Michigan boating

facilities and access to boating including marinas and boating services.

As might be anticipated, the more frequently boaters boat on Lake Michigan, the
more likely they are to be aware of ﬂuctuations in Lake Michigan water levels. Owners
of boats that are operated on Lake Michigan are 16 times more likely to be aware of Lake
Michigan water levels than registered boaters who do not boat on Lake Michigan. Also,
as would be expected the closer boat owners live relative to Lake Michigan, the greater
the likelihood they are aware of changes in the lake’s water levels. Interestingly, the
characteristics of the boats they own and where they store their boats during the boating
season do not appreciably shape their awareness of Lake Michigan water levels. This is
unexpected given that owners of boats kept at marinas and waterfront homes and cottages
are more susceptible to low water because they are less able in the short term to shift
locations (e.g., trailer their boats to different locations) to avoid low water problems.
Also, larger boats and sailboats because of their drafts are more likely to be impacted by

low water conditions.

156

Conversely, the second LOGIT model that focused on boaters’
assessment/perceptions of the extent of Lake Michigan water level changes shows that
assessment/perceptions of the extent of Lake Michigan water level ﬂuctuations is
signiﬁcantly related to the characteristics of the boats that owners operate and also the
type and locations where boats are stored during the boating season. Boater
characteristics are not signiﬁcant in this second sequential model because only boaters
who were aware of low water levels (which is related to socio-economic characteristics)
were included in the development of the model so their effects are already incorporated.
Owners of larger boats who are more liable to be impacted by low water levels are more
likely to perceive that Lake Michigan water levels had dropped “a lot.” This is in part
due to the fact that even relatively small drops in water levels can have a signiﬁcant
affect on access by larger boats. Also, large boats are generally stored at marinas and
many marinas, because of their locations and reduced dredging, were signiﬁcantly
impacted by lower water levels. Boaters who store their boats in lake Michigan coastal
counties, especially in Michigan, are much more prone to perceive that water levels had
dropped “a lot.” This substantiates previous studies discussed in the literature review that
shows the impact of low water levels (e.g., on marinas and boat launches) was much

greater in Michigan than the other Lake Michigan states.

On average boaters who perceive that water levels had dropped “a lot” estimate
they dropped 3 feet during the past three years compared to a 1.6 feet drop estimated by
boaters who perceive they had remained constant or dropped just “a little.” Interestingly
the estimate/perception that the Lake Michigan levels had dropped of 3 feet is very close

to the actual change in Lake Michigan water levels, which was 3.4 feet between 1997 and

157

2001. This indicates that the owners of larger boats (who were more likely to perceive
that water levels had dropped a lot) are more accurately informed about Lake Michigan
water levels. Also, larger boats are more likely to be outﬁtted with more sophisticated
equipment including depth indicators and boat-to-boat communications. Small boat
owners, since they are impacted less directly by low water levels, appear to be less

informed about the degree of water level changes.

The third LOGIT model is intended to determine factors that inﬂuence whether
boaters’ behavior is actually inﬂuenced by low Lake Michigan water levels. Similar to
the previous model, whether a boater is inﬂuenced by low water levels is inﬂuenced by
the characteristics of their boats and where they store their boats, and is also related to
whether they trailered their boat to the launch site and if they sought information relating
to low water levels. As would be expected, owners of larger boats, boats that are kept in
Lake Michigan coastal counties, and boats that are trailered are more likely to be
inﬂuenced by low water levels. Low water has a major impact on accessibility of boat
launch sites. In 2002, thirty-ﬁve of the 134 public launch ramps located in 34 Lake
Michigan coastal counties were closed due to the same reason (Flaming & Zoning
Center, Inc. et. a1, 2002). It is also more probable that boaters who requested information
about water levels are inﬂuenced by low water levels. Whether boaters request
information is a result of low water conditions, or they are aware of low water and
change behaviors because they requested and had available the information is not evident
from the results. It is assumed that boaters who are more sensitive to low water, or who
had experienced low water problems (e.g., damage, inaccessible facilities) are more

prone to seek out information on the topic.

158

The fourth LOGIT model shows that boaters are more likely to change their
boating locations because of low water if they own inboard-outboards, inboards, or
sailboats and if they boat at more different counties. Sailboaters are most likely to
change where they boat to avoid low water situations because they are very susceptible to
being damaged or denied access because of insufﬁcient water levels. Boaters who boat at
more locations appear more ﬂexible, capable and willing to change where they boat. So,
while low lake levels signiﬁcantly impacted access to many Lake Michigan boat launch
sites used by trailered boats, the fact that these boats can be moved more easily to
different locations provides them access to more substitute areas. This, in part, conﬁrms
the importance that boating programs in the Great Lakes states has been directed at

providing boating access to different locations along the Great Lakes.

Boaters who boat more often and at more locations are less likely to reduce their
boating in response to low water. This may be because of their enthusiasm for boating
(“low water is not going to get in their way”), or because they are more knowledgeable
about locations where low water will not impede their boating. Boaters who own larger
boats are more prone to reduce their boating days because of low water levels. Also,
boaters who own waterfront homes are also more likely to reduce their boating because
of low water levels in large part because they are less able to move to other boating
places. During 2001 and 2002 there was a backlog in dredging and many channels that
provide access to larger boats were either inaccessible or had very limited access. Many
privately owned docks and piers where large boats are stored were also inaccessible

because of an inability to do maintenance dredging.

159

Evaluations of the LOGIT Models

The models were assessed for multicollinearity by looking at the number of the
iterations and estimated standard errors. The number of iterations for ﬁve models (10 for
Model 1, 5 for Model 2, 13 for Model 3, 6 for Model 4, and 9 for Model 5.) indicate there
is no a signal of multicollinearity in the data set. Besides, the value of the estimated
standard errors for all estimation outputs are less than 1, which also indicates no

multicollinearity in the data set.

Each of the ﬁve ﬁnal LOGIT models developed through the stepwise-backward
process is signiﬁcant at the 0.01 level (chi-square statistic). Again this suggests that the
amount of the variation in boaters’ awareness and perceptions about the Lake Michigan
water level changes, whether they are inﬂuenced by low water level concerns, and if they
respond to low water levels by changing boating locations and/or reducing their amount
of boating that are explained by the binary LOGIT models is signiﬁcantly different from

zero.

Nagelkerke’s R2 (a measure of success of predicting the dependent variable from
independent variables) is another way of assessing LOGIT models. The “Awareness of
Lake Michigan Water Level Model” explains about 48% of the variation in the data, the
“Perceptions of the Drop in Water Levels Model” explains 16% of the variation and the
“Concerns about Low Water Level Inﬂuences Boaters Model” explains 15%. The last
two models “Low Water Levels Change Boating Locations” and “Low Water Levels
Reduce Boating Amount” Models explain 20% and 13%. In the social sciences, it is
quite rare to have the highest R2 (e.g., over 90%), which might indicate the problem of

high multicollinearity and unreliable of the estimations of those independent variables.

160

McFadden (1979) has suggested that Pseudo R2 values1 between 0.2 and 0.4 should be
taken to represent a very good ﬁt of the model. The values of 13%, 15% and 16% are
still accepted number in the social science research. The small R-square (like 13%)

might indicate we need to include more other independent variables in the model.

The binary LOGIT accurately predicts 85% of boaters who change their boating
locations due low water levels. The “Awareness of Lake Michigan Water Level Model”
correctly predicts 82% of the boaters who are aware of Lake Michigan water level.

These two models (about their awareness model and the changes on the volume of their
boating days) have almost the same values (both over 60%) in prediction percentages for
the respondents who answered “Yes” and “No”. However three other models (the
assessment, whether boaters are inﬂuenced by low water, and whether they changed their
boating location) have extremely high prediction percentage in one group, whose

response is equal to “YES” but extremely low prediction percentage in the other group.

5.2 Interpreting and Applying the Results of the LOGIT Models

The LOGIT model can be used to identify associations between dependent
variables and independent variables. The LOGIT model can also be used to explore how
various explanatory variables affect the probability of an event (e.g., boaters reduce their
boating or change locations) occurring. They can be used additionally to forecast or
make predictions concerning the likelihood that boaters will be aware of Lake Michigan
water level, be inﬂuenced by low water, change their boating locations or reduce their

amount of boating given certain boat owner, boat storage and boat characteristics.

 

’ Pseudo R2 is another type of R-square, produced by the LOGIT models and very similar to the
Nagelkerke’s R2.

161

Since the LOGIT model is nonlinear, the magnitude of the change in the outcome
probability associated with a change in one of the independent variables depends on the
levels of all of the independent variables incorporated as part of the model. This makes
interpretation of the estimates of the parameters for the independent variables more
difﬁcult. It is even more difﬁcult to interpret the estimates of parameters for categorical

data.

The study shows that certain boater segments (e.g., large boats, sailboats, boats
stored at waterfront homes) respond to low water levels in various ways and that some
boaters (e. g., boaters who live in coastal counties) are more aware of water levels. The
ﬁve LOGIT models can provide agencies and marine businesses a better understanding of
how ﬂuctuating Great Lake water levels can affect on boater’s behavior including how

much boating they do and where they boat.

The model also provides additional veriﬁcation of the ﬁndings from previous
studies. The Lake Michigan Potential Damages Study showed that Lake Michigan
boating facilities in Michigan were more severely impacted by low water. In part this is
because many of these facilities were built many years ago during a period of relatively
high water levels. Many were situated in locations that are very susceptible with lake
level ﬂuctuations using technologies (e.g., ﬁxed docks, launch ramps) that do not adjust
well to water level ﬂuctuations. Conversely in Wisconsin, Lake Michigan boating
facilities were situated in locations less vulnerable to low water using technologies like
ﬂoating docks that adjust better to ﬂuctuations. As a result, boaters in Michigan are more
aware (because they are more impacted) of water level ﬂuctuations and more likely to

reduce their boating and change their boating locations because of low water. This

162

suggests that boating agencies and private marine businesses need to be more cognizant
of lake level ﬂuctuations when situating and designing boating facilities. It may be
economically justiﬁable to consider relocating and redesigning boating facilities to make
them less susceptible to ﬂuctuating water levels especially in light of the implications of
possible long-term weather changes (e. g., climatic warming). Further development and
testing of LOGIT models intended to understand factors that inﬂuence boater behaviors,

could aid in deciding where to best invest redesign and relocating investments.

The LOGIT model results also suggest the potential long-term importance of
factors that may reduce the accessibility and quality of boating opportunities. The
LOGIT models suggest that more avid and involved boaters are less likely to let low
water reduce their boating. However, the models show that less involved and inactive
boaters are more prone to reduce or stop boating when confronted by additional costs and
barriers. Given the large number of inactive boaters (as many as 400,000 annually of 1.1
million Michigan registered boaters) and concerns about increasing dropout rates, the
impacts of low water and deterioration in other environmental factors (e.g., water quality,
crowding) could have signiﬁcant long-term impacts on future boating participation and

the well-being of the boating industry.

The LOGIT models show that boaters are far from homogenous in terms of their
sensitivity and responsiveness to environmental factors that inﬂuence accessibility and
quality of their recreational experiences. Some boaters will reduce their boating activity
and change the locations where they boat if environmental factors, in this case water
level, change or deteriorate. The water level is just one of many environmental factors

(e.g., water quality, aesthetics), or product attributes, that inﬂuence the “demand” for

163

recreational boating opportunities. Better understanding how recreation boaters — the
demand side of boating industry are aware and perceive about this water environment
changes, how they are inﬂuenced by this water environment and how they react to this
water environment are all very important to the policy, investment and management
decisions for public boating authorities and agencies, and the recreational boating
industry.

The LOGIT models can also assist businesses, agencies and coastal communities
anticipate and take action to mediate the negative impact of low water. It also provides
evidence that boaters may be more sensitive to other boating environmental factors than
originally believed. Given that the Great Lakes water levels are cyclic and impacted by
climatic conditions, boating agencies and businesses must more seriously consider
investments in faculties and maintenance to mediate against low and high water
conditions. The impact of low and high water conditions should be taken into
consideration when setting boating facilities and services. Also, the Department of
Natural Resources should use the results of the Lake Michigan Damages Study to

evaluate redesign of their Great Lake boat launch sites.

The LOGIT models also reveal that boaters indicate that most boaters have
varying perceptions of the degree that water levels had dropped. Only 15% of all boaters
requested the information about Lake Michigan water level changes. However, more
than half (58%) of the boaters, who were inﬂuenced (location, amount of boating),
requested the information about water levels. The results indicated that the majority of
boaters have inadequate information and incomplete knowledge concerning water level

changes in the Great Lakes. Many are making decisions including whether and where to

164

boat based on inaccurate perceptions. And, these decisions have a signiﬁcant economic
impact on marine businesses and coastal communities. Making information concerning
water levels and their impact on boating facilities and services more conveniently
available can minimize the negative inﬂuence of the low water level on recreational
boaters. This will require an effort from public agencies, boating organization and
business to develop and disseminate information about the water levels and the resultant

impact on boating facilities and services.

5.3 Limitations

Several limitations related to the method and model utilized as part of this study
are evident. First the data used to develop the LOGIT models was collected as part of a
larger study with multiple purposes. While a primary purpose was to develop a more
comprehensive understanding of how low water impacts on boater behaviors, the two
surveys were also required to collect data to produce the Lake Michigan Potential
Damages Study. If the sole purpose of the study had been to develop and test the LOGIT

models, additional different information could have been collected.

In relation to the potential limitations of the data, two distinct collection methods
(mail, telephone) were utilized where each has its own purpose, sample, and method. So,
there might be some differences between the two data sets. The “Awareness of Lake
Michigan Water Level” and “Perceptions of the Drop in Water Levels” Models were
developed using data collected in the 2001 Lake Michigan Damages Study telephone
survey. The other three binary LOGIT models were developed by using the data from the

2001 Lake Michigan Damages Study mail survey. Although both surveys relied on

165

samples of registered boats the sampling proportions were different. There might be a
bias by using two different datasets. Furthermore with the same dataset, more statistical

models such as nested LOGIT model can be considered to use in this study.

The 2001 Lake Michigan Potential Damages Study was designed to assess boater
awareness and perceptions of Lake Michigan water level and the impact of the low water
levels on recreational boating. The emphasis was on collecting information from Lake
Michigan boaters. However, no list of Lake Michigan boaters was available to be used as
a sampling frame. The only list available was the Registered Boat Owner list from four
states. A stratiﬁed sampling method was employed to identify as many Lake Michigan
boaters as possible. Half of the sample was boaters who registered their boats in 34 Lake
Michigan coastal counties. A higher proportion of owners of larger boats were drawn for
the 2001 Lake Michigan Potential Damages Study mail survey. The concern is that the
owners of smaller boats who live in inland counties but boat on Lake Michigan were less
likely to be sampled and surveyed. The extent of the bias that could have been

introduced because of this sampling method is unknown.

A primary emphasis of study is to determine if, and how water levels induce
changes in boating behavior — how much boater’s boat and where they boat. The
inﬂuence of low water on other boater behaviors was not analyzed. For example, some
boaters did not put their boats in the water in 2002 because of low water, and some
moved where they stored their boats (e.g., marinas) because of low water. Even though a
large number of registered boaters were sampled and a high response rate was realized,
not enough responses from persons exhibiting these low water induced behaviors to

develop LOGIT models.

166

5.4 Overall Implications and Recommendations

This study only focused on modeling recreational boaters’ awareness and
response to a certain boating related environmental factor — water levels. Obviously,
while water level is important there are many other environmental factors that can
inﬂuence boating behaviors such as water quality/ pollution, lake erosion, boating
crowding, and aesthetics. This study suggests that there is both a need and potential to
conduct additional studies that focus speciﬁcally on boater responses to various
environmental factors. Future studies employing LOGIT models could model the
boaters’ attitudes or responses to the water quality changes. Boaters could be studied to
determine their awareness and sensitivity to various measures of quality including clarity
and sight depth (Adamowicz, 1994; Lupi, 1998; Siderelis & Moore, 1998; Soutukorva,
2000). Since a great deal of boating is ﬁshing related, it would also be useful to conduct
studies to better understand how contaminants and reduced catch rates of ﬁsh inﬂuence

boater behaviors.

There is also the potential to monitor awareness and behavioral responses to
quality and quantity of boating facilities and services and also boating accessibility.
According to LMPDS results, many marinas’ slips and public boat launch sites are not
accessible due to the ﬂuctuating water levels. In some areas of the state the quality of
boating facilities has declined and it would be interesting to determine whether boaters
are sensitive to these changes and how the changes affect the amount and location of their
boating. It would also be interesting to assess boater awareness and responsiveness to

crowding and conﬂicts with other water users.

167

Some previous studies have found that often people do not change their behaviors
in ways that are consistent with their perceptions related to environmental changes
(Tarrant & Green, 1999). So while they may perceive reduced environmental quality,
they may not reduce their boating or the locations in a way that is obviously consistent
with their perceptions. There may be mediating circumstances such as available
substitutes and the awareness of these substitutes that need to be identiﬁed and better
understood. Unfortunately, because this study utilized two different data sets to develop
the different LOGIT models, it is not possible to completely understand and describe the
relationship between boater awareness and behavioral response to low water levels.
Future research on low water levels and other environmental factors and their relationship
to boating behavior should collect data in such a way to permit the researchers to assess a

correlation between perceptions and behaviors.

The study also suggests the boating industry and boating agencies that develop
and operate boating facilities and services (e.g., fuel facilities, launch ramps) need to
focus more attention on water ﬂuctuations and environmental factors in relationship to
how they design and where they locate boating facilities. Actual and anticipated changes
in environmental factors may warrant consideration of changing the design and location
of existing boating facilities. This study and the Lake Michigan Potential Damages Study
indicate that some boaters are impacted by low water and some actually alter their
behaviors in response to low water conditions and probably other environmental factors.
Recognizing this, businesses and communities that are heavily involved and dependent

on recreational boating need to be more cognizant of the implications of changes in

168

environmental factors on boater behaviors and incorporate this information in their

decisions regarding investments in boating facilities and services.

This study also suggests that state and federal boating agencies with the support
of the boating industry must be more involved in efforts to protect and enhance the
environment that supports Great Lakes boating. Current industry efforts to secure funds
for environmental protection, dredging and redevelopment of boating facilities (both
public and private) in response to changes in the environment are supported by this study.
Industry efforts such as the Clean Marinas Program are also consistent with the ﬁndings

of this study.

The ﬁndings also show that only certain segments of boaters were aware of low
Lake Michigan water levels, and many did not have an accurate understanding of how
signiﬁcantly water levels had dropped up until 2001. Clearly, there is a beneﬁt of more
carefully targeting information aimed at boaters in order to develop a broader and more
accurate understanding of water level ﬂuctuations. This same conclusion applies to

changes in other environmental factors.

Another recommendation to future research is to increase the sample size and
collect more exact data on which to develop the LOGIT models. Modeling boater
response to various environmental factors, facilities and services, access, and possibly
crowding will require more precise information on their perceptions and behaviors in
response to changes in these factors. The sample size (1,3 86 boaters who responded to
the mail survey) was still not large enough to develop LOGIT models for speciﬁc types
of behavioral changes (e.g., changes of boat storage locations). A larger number of cases

would have permitted the development of nested LOGIT model. A three-level nested-

169

LOGIT (aware or not aware of low water levels, inﬂuenced or not inﬂuenced by low
water levels, and the different types of response to low water levels) could have been

attempted.

Data should be collected immediately after the boating season or possibly through
a diary of boaters’ boating activity. The 2001 Lake Michigan Potential Damages
telephone survey interviewed recreational boaters from December 2001 to January 2002.
Questions should exact information about how they changed boating (amount, locations)

when they perceived changes in various environmental factors.

Most previous boating studies have focused on boater characteristics, use levels
and boater spending. This study has demonstrated the potential beneﬁts of modeling
boater awareness and behavioral responses to environmental factors. In the future, more
attention and effort need to be given to the modeling of recreational boaters’
awareness/attitudes/responses toward the quality of boating environment including water

level, water quality, boating facilities and services or boating accessibility.

170

APPENDICES

171

APPENDIX A

Application for Certiﬁcate of Watercraft Title and/or Registration

of Michigan

172

 

ELI

 

tit Application for Michigan Watercraft Title

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TRANSACTION TYPE MG NUMBER EXPIRES ON: COUNTY OF RESIDENCE CODE REG FEE
YEAR MM; LENGTH HULL IDENTIFICATION NUMBER W15 FEE
Ft. ln
HULL MATERIAL CODE TYPE DODE powER c005 use . CODE TAX
FUEL . CODE MODEL 0R SERIES NUMBER OWNER'S DRIVER LICENSE NUMBER DATE OF BIRTH REG. TRANSFER
OWNER‘S NAME(S) AND ADDRESS TOTAL
FUU. RIGHTS To SURVIVOR
FIRST SECURED PARTY FIUNG' DATE "C In“ N'N Ink-n PARTY FIUNG DATE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CLAIM FOR TAX EXEMPTION USE TAX RETURN PURCHASE DATE:
REASON: ‘
‘ 1. Purchase price or retail value,
‘ I h .
, w” ° W" ” 9mm SELLERS NAME AND ADDRESS:
2. 6% Tax
I certiiy the tax exemption shown 3, end“ for tax paid to a
above is valid. initial bOXI reciprocal state (proof attached)
l certiiy i own this watercraft and all information 4.7“ Being pald
on this application is correct to the best of my
knowledge.
New Owners/Applicants Signature ' ’ HULL M ATERI Al. TY E OW SE FUEL
l. Wood 1. Open 1. inboard 1. Pleasure 1. Gas
X 2, steel 2. Cabin 2. Outboard 2. Commercial 2, Diesel
.. 3. Fiberglass 3. Sail 3. Sail 3. Sail 3. Electric
4. Aluminum 4, Flow 4. Salt/w Power 4. Other
5- Other 5. Canoe 5. Other Power 5- Commercial Freight
s. Pontoon 6. No Power 6. Commercial Fishing
X 7. Personal WC 7. Jet Propulsion

 

 

Contact a Secretary of State branch ofﬁce if you do not receive your title within 60 days.
Final determination of the correct tax liability will be made by the Michigan Department of Treasury. You may be required to document your tax
return or prove you are entitled to the exemption claimed. If you cannot support your claim, minimum penalties include the added tax, a negligence
penalty, plus interest from the date of filing this application. Additional penalties can be imposed including criminal prosecution or assessing up to
175% of the taxrdue. '
EXEMPTION - TRANSFERS BETWEEN RELATIVES: An exemption from use tax is allowed when the new owner is the spouse, father, mother, brother.
sister, child, stepparent, Stepchild, stepbrother, stepsister, half brother, half sister, grandparent, grandchild, legal ward, or legally-appointed guardian of
the previous owner. Documentation proving the relationship may be requested by the Michigan Department of Treasury.

VALIDATION:

  

Candice S. Miller, Secretary of State

Authoritv aranted under Public Act 160 and Public Act 303 of 1967 as amended.

 

 

APPENDIX B

Application for Certiﬁcate of Watercraft Title and/or Registration

of Wisconsin

174

 

 

SAT

Wisconsin Boat Registration and Titling Application Fo,m9400_193 (Rm)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Please Check Application Types — See instructions Leave Blank - DNR Use Only .
BOAT REGISTRATION (Including ﬂeet boats) ate Veriﬁed Tax Due Veriﬁed Total Amount Rec'd
D Original Registration l:l Transfer and Renewal
Transfer Renewal Reg. # Type

Fleet Number (If any) Dealer ID E Sigma] 10:“ WS BR BF BD DV
enewa en' Verification C M Flt. Ct. # Att

REPLACEMENT MATERIAL Reason:

El Certiﬁcate El Title D Decals N O S T W .. N Y

DNR Hunting / Fishing Cust ID (9 digits) (optional) Trans- Type ‘mi‘als

 

 

OTRDCDDDT RT

 

V
I Owner's Name - Last, First, Middle Initial(Print) Daytime Telephone

Additional Owners (DO not list previous owner's names)

 

 

 

 

 

 

) 1.
Mailing Address Check here If address change D Date of Birth 2.
| I 3.
City State ZIP Code County State of Principal Use US. Citizen
WISCONSIN Eves E] NO
Wis Registration # (if Any) Boat Hull Identification # (Verify numbers on boat) Boat Make or Manufacturer

 

Out of State Registration # (It Any) Series or Model Boat Length Model Year Purchase Date County Where Boat Kept

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

it In
Engine Propulsion Type of Boat Hull Material Fuel Type of Use
1.EIOutboard 1i: Propeller 1D Open 6D Pontoon 1. DWood 5D Rubber 1. D Gasoline (E) Pleasure 5‘ D Other
2.[:]Inboard ZEWater J6” 2‘ D Cabin 7' Cl Sailboat 2. [:lAluminum 6. C] Other 2. El Diesel 2.[:l Rental 'gzggemia'
3.l:]inboard- 3.l:Air Thrust 3D House 8.[:] Inflatable 3' [:]Steel 3 El EI tr' 3": Commercial
Sterndrive (V0) 4.: Manual 4E] Canoe/Kayak 4 |:]F'b | ' ec 'C Passenger
4.E]Airboat 5i: Sail Only 5i: Personal Watercraft ’ l erg ass 42L CommP'f-in' FiFl‘ing
BOAT LIEN INFORMATION - MANDATORY for Titled Boats 16’ and Over
Lien on Boat? Lien Holder Name Address City, State and Zip Code
El No D Yes:
Note: ' Boats OHIY lMake of Engine (‘“_b°a'dl'“b°afd 0‘" Engine Serial #
_ , . . Drive Only) _

Documentation Boat Name Hailing Port Engine 1. Engine 1.
# Engine 2. Engine 2.

 

 

U ———‘—-—>4—O-—>4

Sales Tax information and Fee Computation - See Back Page For Instructions

Full Purchase Price (Include motor, trailer, accessories) 3

 

Trade—In (Description of boat or property traded) Year, Make & Hull l.D. #

 

Less Trade-In Allowance

 

 

Taxable Receipts (Line 1 minus line 2)

 

If Tax Exempt, Enter Exemption Code and Reason(See instructions on
5% State Tax (Line 3 x .05) baCk) Cod Reason

 

1/2% County Tax (Line 3 x .005) (If applicable)

 

 

“2% Football Stadium Tax (Line 3 X .005) (If applicable)

 

This Boat Was Obtained From: (Please complete information below)

1/10% Baseball Stadium Tax (Line 3 x .001) (Ilapplicable) Dealer or Seller Name

 

 

 

 

 

 

 

 

 

 

 

V
E
l

a

S bt t ITax Due 3 El Private Sale
u o a
El Dealer

8. Lien Filing Fee $5.00

_ . Address Seller's Permit / Tax #
9. Registration Fee (See Fee Schedule)
10. Replacement Decal Fee $2.50
11. Replacement Certiﬁcate Fee $2.50 0'er Slalev ZIP COde Telephone Number
12. Replacement Title Fee $5.00 ( )
13. $1.00 Voluntary Contribution For Lake Research/ Signature of Dealer Date Signed

Clean Water‘

14. Total Tax and Registration Due $

 

 

 

 

I certify with my signature that to the best of my knowledge the information and statements on this application are true and correct. Any person who

knowingly makes a false statement on an application for title may be punished under 3. 30.547, Wis. Stats, by fines up to $5,000 or imprisonment Up to
five years or both.

E) Withhold personal identiﬁers collected on this form from disclosure on any list of 10 or more
individuals that the DNR is requested to provide to another person [S 23.45, Wis. Stats]
Signature of Owner (Additional owner's signature is not required) Dale SIQI'IEd

Social Security NUmber (required on behalf of Department
of Revenue for tax purposes)

 

You may make a voluntary contribution of $1.00. This fee is to be in addition to the registration - . ~ - - -

fee and will be used exclusively for research conducted by the Dept. Of Natural Resources to Mail to. Attn. Boat Registration and Tltlmg
determine methods Of improving the quality of the lakes in Wisconsin.
AP
UN

Department of Natural Resources

PO B 7238
PLICANT MUST KEEP A COMPLETED COPY of this page while boat is being OPERATED Madisooxn WI 53707-7236
TIL certiﬁcate is received. '

 

 

 

APPENDIX C

Application for Certiﬁcate of Watercraft Title and/or Registration

of Indiana

176

 

 

1LT

AFFIDAVIT OF POLICE OFFICER
Physical Inspection of an Indiana Resident's WATERCRAFT

State Form 39530 (R3 /9-96)

 

INSTRUCTIONS: Use reverse side for vehicle.
Indiana Code does not permit watercraft inspections by watercraft dealers.

NOTE: There are no provisions in the Indiana Code for assessing an inspection fee.

NOTE TO LAW OFFICER

1. A Title or Certificate of Origin does not have tO be present to complete this affidavit.

2. You are required to physically inspect the watercraft to verify the existence and condition of the Hull Identification Number.

 

 

 

Name of owner Address Of owner

 

 

. WATERCRAFT INFORMATION . .

   

I, the undersigned law enforcement officer, hereby swear or affirm that I have personally examined the watercraft described
as follows:

 

Year Make Model Length

 

 

Indiana Registration Number 7 f k U A W Hull Identification Number (HIN)

 

 

 

       
 

 
 
  

 
 

‘ ' NOTE TO LAW OFFICER
THE BMV WILL NOT ACCEPT THIS AFFIDAVIT IF THE INFORMATION REQUESTED IN THIS BOX IS NOT

Printed name of inspecting officer Title Of inspecting Ofﬁcer

  

comm-E AND Lactate. ’ ‘ "

 

    

 

Name ofdepartment City State

 

ID number Telephone number

( )

I swear or affirm that the information I have entered on this form is correct. I understand that making a false statement on this

form may constitute the crime of perjury.

 

Signature Of inspecting Ofﬁcer Date signed (month, day, year)

 

 

 

 

APPENDIX D

Application for Certiﬁcate of Watercraft Title and/or Registration

of Illinois

178

 

ILLINOIS

 

 

 

 

 

DEPARTMENT OF
ﬁ THIS AREA FOR OFFICE USE ONLY ﬂ

 

 

   
   
    
    

 

 

  
 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

    
  
 
 
   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

N AT U R A l.
RESOURCES
H NEW C] TRANSFER RENEWAL D RENEWAL RENEWAL BY PHONE Ci UIIITEDSTATESCOASTGUARD oocuueurgn
\' See other side for fee See other side for fee See other side Ior lee $ SEE 0TH SI Se ee other Sidef or rSU CGe as
O and enter amount -§ $—__— and enter amoun $ and enter amount —. FOR DETAIR LS and enter an mount H
I D DUPLICATE ENTER DOCUMENTATION
El DEALER OR I] DUPLICATE TITLE $5 CORRECETED U NUMBER
MANUFACTURER $10 (COMPLETE A, B. E& F) HERE —§
(NO TITLE) $5
ILLINOIS ' l l ' ' OTHER STATE ' l l ' ' ' ' CURRENT REGISTRATION EXPIRES
REGISTRATION NUMBER II :|I_I REGISTRATION
(enter NONE if none I I I I BER—r I I I I I I 5 - 30 -
I InIVUI AI IUIIILI|I NAM I I I I I I I I , MOPEL (NAME 1 i i r | | ‘I I I l I I I I I I J
l I I | l I I l I l | l l L l l l l l l | l l l L
HULL IDI:N I IHCAIIUN NUMBER * SEE OTHER SIDE FOR FORMAT" I I I I I I I I
l l I l l I l | l l I l l 1 l l | l l J
LENGTH~FEET ONLY MODEL YEAR MIO. DAY I YEIAR I
PURCHASE
DATE
1 l l l l
INSEI)R'|I'I_II HULL MATERIAL va i annm i IYPI: USE PROPULSION FUEL
(PRIMARY) 1 OPEN 5 PONTOON i PLEASURE i OUTBOARD i s HORSEPOWER WATERCRAFT COLORS-SEE OTHER
I WOOD 4 FIBERGLASS 2 CABIN 6 CANOEI EBOAT DEALER 2 INBOARD 2 DIESEL SIDE FOR COLOR CODES
2 STEEL 5 OTHER 3 HOUSEBOAT KAYAK 4 MANUFACTURER 4 34M 3 OTHER .
6 iNFLATABLE A SAILBOAT 7 OTHER 5 OTHER 5 OTHER l l l '
I I I | | J l I
YES III THE OWNERS SHOWN CHECK ONE BOX AND ENTER NUMBER—ﬂ
EVVLO WIS TO BE DRIVERS
RECORDED AS OWNERS {:1 DRIVERS L|CENSE I I I I i T I I I I I I I T I I I I I LICENSE :1
IN ”JIOIINLTIETNrAINCV I I:] SOCIAL SECURITY I i I I I I I i l l I I I I I I i i I STATE L
IVLJ ;
SURVIVORSHIP
————*——~«l I: TAX ID DATE OF BIRTH
OWNER(S) LAST NAME, FIRST NAME, MIDDLE INITIAL MO. DAY YEAR
l l | l l l l l I l l I l I I l l l I I I I I I I I I I I I
(1) I
7‘ I I I I i I I I I I I I I I I I i I I I I I I I I I I I I I i
B (2) I
| l l l | I I l l I l l l l L l l l I I I l I l l I l l l l l i
l I I I l l l I l I I I I l I I I l l I l l l l I I
MI I
l l | I l l l J_ | | | I I L i l l | I l l l | | I |
l l I l I I l | l l l l l I l l l I I I I l l l I |
HI I
l J l l l l l J l l l I I L l I I I l l I l l I l l
STREET ADDRESS OR RR. & Box NUMBER COUNTY OF RESIDENCE
I j l l l | l l l l l | l I l I l l I I I I I I l I l | l I I 2::
| l l J | l l l l l l J | l l l l I l i l l L | l l I J J_
CITY STATE ZIP CODE
I—III I I I l I I I I I I I T I I I I I I I I: I I T I I T l T
l l l l l l | I l | I I I l
I IF HOII DFRI NAME (enter NIONE III none) I I I I I I I I I I I I I I I I I I I MI DAY I YEIAR I
SECURITY
AGMT. DATE
1 I I l l l l I l l I | l l l l l l l I l I | l I
STRIEET IADDRESSI
I l I l l I I I I I l I l I I l l I I l I I l I l
C I
I l I l l l I | l I J I l | l l l l I | l I l l l l l l J
CITY STATE ZIP CODE
I’I l I I I I I I l I I I I I I I I I I I I I I:I:I I I I I I I I
I l l | l l l l l l l | l l I l l l
ICITY , w. ..WLD FROM STREET ADDRESS
COUNTY STATE ZIP CODE
D
SIGNAT DATE SIGNED
THE SELLER HEREBY TRANSFERS INTEREST IN THE ABOVE DESCRIBED WATERCRAFT TO THE NEW OWNER R(S)
SIGNATURE OF PREVIOUS OWNER NOT REQUIRED WHEN TITLE Is PROPERLY ENDORSED AND SURRENDERED WITH THIS APPLICATION
(1) (2)
SIGNATURE S
E ( )(3I (4) DATE SIGNED _—
INVE HEREBY AFFIRM THAT THE INFORMATION PROVIDED IS TRUE AND CORRECT. YOUR SIGNATURE AUTHORIZED THE DEPT. OF NATURAL RESOURCES
TO LOWER THE AMOUNT OF YOUR CHECK IF FEE SUBMITTED IS GREATER THAN THE REO JIRED FEE.
PLEASE GIVE US A PHONE NUMBER where you may be reached on weekdays.
STATE REASON(S) FOR TITLE SEARCH OR
F NEED FOR DUPLICATE 0R CORRECTED TITLE
Phone —
(area code) (number)

 

 

 

SEND THIS COP-Y ONLY TO DEPARTMENT OF NATURAL RESOURCES

 

 

 

APPENDIX E

The 2001 Lake Michigan Potential Damages Study MaiI Survey

180

02 U 00> U mmcom E 00003 05 E 0000 £5 :6 0. $00003 :0» 00< .0m

 

 

 

 

 

 

 

©0025 .0050 D 0300000 000 003 000m B w§0300000_~m0w000 w0000m D 0003000 000000000 0050 _U
w00000 5 00000000 0004 D 0000 £5 :00 00 000005 B 05000.— 0 E 000% 0 00m 00 03003 D 00.03 32 5000 00000000 D
00500? 0000 D 0000000 0000000m D 300 .00 0000 D 0000000 02003 D
2&0 05 EW 002000 zoom s 0303 20 a 08 as :0 8: .5» 20 33 An
All 02 a
_0m 0050000 0. 03m. 5002 Soom 5 0000 $5 300 00 00300 00003 0000000 000.0 :0» 20 5000— 0003 5 .0m A| 00> D
:08 a 80322 5 as; as 5 .20 as 05 33 .m
0003005 U 302 D 000030000 00 300 0 00000000 :0» ED .3 000> $000 $5 000.000 00> 20 0003 .0N
500 20:2 :80 £0
308003 0050 D 3.0 00.. .wdv 000000003 3000000 U 30m 00 00000 D 0000000 U
038 as, .30 D 8538:: dam D 20850 D 208302 B 2085 D

08.. .8 25 023 0N
0000— $05000 05 00 00.00002 .000 00000230 05 3000 000000005 >00? 0: £00 00003 .N

 

00000. 0000005000 05 0.0000 00003 00000000 M08502 05 003000 000 on .02 D
.3 00:02.0 8 030. 0:0 .00.. 0050 05 00.— ~3.0.0.0 £5 0.0.0500 00> D 00> U
0003022 5 00000390 3000000 0000 005000 030 :0» 0D .0. _ 0 2 D

2002 w0m=000 05 00 000.3002 0000 05 030 :00 :0» 0D .—

smmaoa a: $336.09

.00» 0.00.; 6000000 300300000.—

..0 0050000000 05 5:5 3 00:00:85 $5 0050» 000 03 >03 5.00 00,—. $0200 0030000: £5 00000 000 00030000 00 000.000 000 00.00 ”mg .Aow_~.mmm.§ 3
$00.35 00802 w0_>_0>0~ 00000m0m 00 00000800 03000303 05 .00 0000 .92 .00003— 0Em< .09 00 00800000 00 000 x0055 £5 00 00.200 Snwt 000»

w0€00w00 00000000 .000.:mE®0>00000E =0E-0 00 2 _x cm: m.mmm.:m 00 00000 ‘3 300002 0m .5 0000000 $0200 00: .00 00000000: 05 w0=000w00 00000000

>00 0>0n :0» .2 00000000 05 00 00.8 003000 00 000 000000 >08 :0.» 000 .0302? E >955 £5 5 0000300000 S0300 00029000 000» 00 000030 0000000030
0000 00 0800 0:0» 00 000000 0 0>00 000 :05 0 >9 000050000 E00000 EB 00> 00003 32 00 0000—00 00003000 00000000000 0>00 >000 :0» 00003 000 0000 000» 00000—

.02 so» 223 25082 050052 a so» 00% 9 30: BEBE 23; 03 .>0>0Dm 000 00000200 00> 0020 0000320 000 00 08000 0.200

.mm<mqm :59} 02:22 mmh ZO m2<Z ~50». m>Om< QmEFZme H<Om demhmamd E: OF >AA<UE~Ummm mh<qmm 44:5 mZOmeDO

mm; 00 mEOm 200:: 000 000600 060000;? 003022 E 000030 0000 00.000030 009m 00 0000000000000 w0€000 .200 000 03 0000000 0000095 b0> mm

8000000 50> 0000000 00000000000 00 00003000 05 800000000 .05 29000 0 £050 00 $000 006 000000.50 029000 0 >2 .2000 000» 5 0000000 000000030 .00 a: 0 .0000

00000—00 30000000 0000 0>0n 00> 0000200 000 w0_000_0 3:000 000 00000— 000 000000 $00000 E 00: 00.0 00_000w0 w0s000 _0000000000 00000 000 00000mw0m .00 0900

b00< .m.D 05 00 0000000 00 EB A080 05 0.0 0:800 00:. .3000 0000000 000 000000. 00280—0000 00 30300 000205 005 >030 00mg 0 .00 :00 mm 00E. 00030000
000 53000 w00000 00 m_0>0_ 00003 32 00 300900 05 00 0:000 500% 0 0:3 w00000 _0000000000 .00 A030 0 w00000000 00 9000303 008m 0030022

>m>¢3m 02:.(00 A<ZO.h<m¢Omm $0.10.: N? EON

 

 

 

181

.880 no 8&5: _n 5.8— : Sagan :
3830 3 £58: on :59: R 5:526 2
2353 S 89:35 S 8.5.2: 3 5.5: a «.2250 2
2&3 S assume 3 83 3 8.58: _n 36 3
59.233 a 235 x. 5.93: S 013:: on 538 2
Eam§> S 93an G 8323 8 .85 a :85 a
28.5 3 $552 3 3:23.. 3 .5; 220 3 5E8 :
uaazﬁm E Swans: G 5&1 3 2.28 2 255m 2
c2333 R 99582 8 33 a. sage 8 .am a
85am 2 5.8.8: a 33833. a. 83250 3 ban a
.332 d an 92.5.2 mm as. ... team a g h
.16 ..m : 8“...st S 533. 3. Sam 2 832 e
33.95 2 25.: em San—53. on gauges - 552 n
55% 2 855952 a 89.2: an 5.5 _N 3&2 v
92 335 K 5882 a «:38. R 2835 on 59.2 n
.355 E .83: a .5: on 8.56 a. b»? N
3390 ac 3.8.52 um 88. an 56 M: ao2< _
Hui. aﬁi é.
, «L < -. [1. b . I
. .M. rgﬁﬁﬁa 3.3
... “.141“: Navy: . _ r15
lam—wzmxpiamT‘ 2.1.25 I
{xxx .1; , ‘ _ A
rﬁzrwlz “2.5.5224. J ...
33. 5.. r74: _. ‘
\ h I” , i?Vfl¢’.§
, 6. , . u . _
3 41- ,. _ ¢ ,
3!? WW v.
.3 3,. ﬂ” 3,33
in; Ed... ..
4.? . . W 5.31 x ..
.Wﬂw » ., ,6 \_ A
1 i. /, as. M W 1 .. 1%.“
.5 ‘ . ‘_ 55.1Limw/(l)
..2393 $4 . _., #185“ a... //:
8.2 n, l . V g , x _ ‘lfzxw
.UJ :1 51.; “.34).. .35 /
W ‘ 155: Lgr, r
. . . .\tL_.ﬂJ _
.3...
. I.
(3:; .1

“be 3853
$.20 magma $558325 £332
$332 10% a “a vow: .wdv mgr—own: «on :5 .833 05 3 $3 2 cos? Sow
5 wow: 83 £5 ma; Amt/Om? QmMmE/OO H96 was .2836? >53: 30m .m

.3c&
:35 v.85 S .368: 32% mksoxn. 35 ENE! 38 .435
9.8 Six :2qu 38 5.80.5. 5A3: gm. 2: SEQ gm.
33 6.2% £23.?” £951 33 no 3cm 5 wagon m0 925
Soom 5 E29» mﬁauoqaoo
8 8di $20 880 so @2830 Eon ma ma? mxav >58 Bow ..3.

 

Sam 5 53:32 “is :o wagon we when

 

$333
58228 5 5 E22 813 no “98830 Eon £5 83 93v ban. 30E .3

 

 

 

182

wagon 3:3qu
350 D E 3285 63 D 83858 590 U
.53? >23 D Hague?» gooﬁ— D 3093“ ogonoom U
mango“ cocoon “mom D 35..“ mo “moo D £832 5:33 U
258 a Ea 88 5 $6 $2 38 as was? :3 3 BB .3

 

 

 

Fl mxwv 33 D

$36 no 52:5: 25% U
9% 802 D
258 E Ev so» 55 58 E 95% 32 Ho 22: Son m3“ 280% so» BO .3

 

Ann 253 8:580 280M
Eu: 39 208 E Eon £5 SE30 :9“ Eu 83:58 “an? E .3.

>955
£5 35358 5cm Son 05 Eugene 38E 5? so» what mo 53:52
G o 3 3% :9. :a 62823 a: a; x E 88 a mason ‘8 939
Soon E ha 8 £8 5 530m SUE: «.8823 “won £5 mm? 93¢ >58 Bo: .v

.53.: R
25>» a.
5.05:3... 3.
550:; o.
:83: a.
82.32.... ..
0.28:8 R
Siam 0.
.332 ..m ...
EC ..m E
32.0% m.

38.0 me
«.0080 no
:omucoEO cc
igumo no
58000 we

$3.1... .43.... a...»

93:5 3
093302 no
50332 s
Dun—09:02 8
...—18:02 am
02:22 mm

85.52 _m
5.8.: on
02:82 3
33 a.
8:055 S.
8323 3.
3:303 m...
.893 3.
8.3 2.

if “mu grﬂwgywﬂa.

...r

.chzaawuli... ..
.. ...-..2 ..11
... I. a?! SS»
5......33 :
. _. w§§.3§. ..
ﬂaisﬁati ...: ... .239.
> y _. iﬂv’i.j. ”in:
:6! H , . h

in .14... ‘ ‘
E. was. 33......“

2.3:: o.
.820 3

SE... “EEO am

050...: R
5326 3
0030200 mm
.25.: E
SE a
.5220: S

 

5.1.:-

SUD _N
2003:.0 ow
:9:=0 m.
0:20 2
13000.20 2
50.326 2
59015.0 m—
30 Z
.5310 2
.355 ~—
ﬁEBm :
oﬁcum o.
E. a

.50 w
ﬂag: .
0823. 0
E55. n
5.0:: v
§m0=< n
0&7. ~
«Suﬁ —

 

3: . -8: < D
0% 0:00.03?

Eugm ho ho>.C < D

 

 

 

 

 

 

 

 

 

0:. 0:05.033
0.3 0525 :< D I 008m
2033 @625 ...050 U
w::00::00 a. 000.3 .5003 ..050 D 5:0 “womb—30> D a: 00 v
0005 .050 9 $000: .002 .093 83550—00335:— bﬂo m
55 05 0:050:03 < D a :0 .0 9 50:08:... D E 000:. 50:30 D 0
...—0.03 MES—E8 028—00: .w:tooE «ESE 30:05:00 D
a. 53:22 0:305 .303 05 E D «:52: 0:55 D
025 2 $08: 3:00.“ 0.05.. b: a E D 2:0: 0:000m\0w§00 D tom \ .30
55. 0:. 0:05.50; .4. D 0::— :0 D 00:023. 0:05:50: D
8003. 0923. ..0 09:. g...— .0 00.3. 55:00.—
macwaom 5.002 coon 05 5.5: $02 $5 000. :0 E: 0055 .nw
00:02.0. 0:05.50: b: 00.62 D .50... .033 20.0.5: 8 $000.4. D
.050 D 0:8: 0:05:50 8 0.20 02 oh D 00.. :30 E09¢EEE 5 00.00.05 D
.0003 33 U 0:5: 3:800. .0 093.8 20m D 08:. 220 .:0§B:_.8E .00 0.03 D

93:: E5 :0 200:8 «20:82 Ema—go :0.» .33 3:030. 05 was “5;? .aw

8.. _u

_a 558:0 0. 03m. doom E 80: 05 :30 #505 H .8 50:82 050m .02 D
0:800. wings ooom MES: :82 «:5 50.9. :0» 0:05» :5 :05002 0:20me a 25 m: .w

2:0: .0 02:52

$000.: w_ a 0.0:? .o 4:00— .. .002 50»

0.0:? 8 00:02.0. E05250: 50» :55 _0>nb 8 :0.» 0x8 .5850: a 0000 mg. 30m .2.
...: 0m: DEBS: :oz 0.0:?
.0 .302 0:3 $02 50% 0.055 8 00:023. 50:05.0: 50> Eod a 0. 3:8 >52: 30: .h

.03: .00 02:52

 

0:0 . -:o: < D
0% E0583?

adobm yo .9»: < D

 

 

 

 

 

0:. 0:05.80?
003 0:25 :< D 0.8m
$033 @6055 .050 D
3:00:50 a. .023 .9003 ..050 D 220 000%:0m> D a: 00 V
“00.0 .050 8 $08: «no: .092 EEEEﬁooEactmE b”... w
5.5. 0:. 305.20? < D a :0 8 3 505083. D E 000% 00:30 E 0
E0095 95005.00 02300: $5.52: 05.3.: 320500 _H_
um 53:22 0:305 .033 05 E D «:55 0.3:: D
00.3 2 00.08: 550nm 203m .0: a 5 D 0E0: 0:000.ng00 _U to: \ .30
5S, 0:. 0:05.033 < _U can. :0 D 00:02.0. E0550: D
0.0003. 0928» ..0 093. 550.... ..0 0:5. 55:00.—

 

 

 

 

 

   

0588 3.8.. 88 0:

 

55383.: 8.. .5 E. 22;. .0

 

183

 

avd
2d

9.2.
:2—

N23
0v;
mi?
3.5,
mm;
on;
I”;
2:3
2:5

0000

ow =2
3. :2
cs :2
we :2
3:2
352
mm 52
3:2
3:2
9:2
VN .2
_N :2
n. 52
I 52
S 52
n :2

05.00

:95».— 5>
cﬂsoogum
035.0
0:300
50905052
005E052
530.2
82:02
00505002
3:203
.0555
8.09
596125
=0.Eom
uﬁcum
Era—2
:0m0__<
0E0:

 

 

 

 

 

Am», 0008 0050550 E8005 A502
009 505—0305 00.03 26. 0000055050 :0» 005 00302 30000? $500580
00 53:22 00.3 8 000000 0365 .23 00:0 05 0003 0050550 “an? 5 .ma

©_0o5m .650 D 00:05:. 8 020 “0002035 D 503.5 50585 D
.0555 3onch D £30058 2: I >030 00:55. D 50:52 2 030:3 D
_05520 00000950 D 000.. 03.52 8 :03 D 0000.0 83 8a .5053 D

95550 005 :0 0.00505 500055050 00> 20 0550305 50 05b 09;? .55
00> _u .2 .5085 2 9.20. oz D
5.000003 3:005:00 00 535052

0053 8 000000 0E>o5 005 0020 00 080305 00003 32 b8 00250050 :05 2Q .05
an 080v 009500 000005
A502 009 5000003 95005500 535052

005015 2 000000 00305 88 00:0 .0 .009 So» 50:50— :05 20 005580 0053 E .05

._.I é a
_3 50:00.5 3 555. 02 U
203 E 00000? $500580
5 amwﬁoaz 00.3 8 000000 0E>o5 35 00:0 >50 00 «son 50» 505:: no» 05 .05
0008050 55000305 E 088050 5:002 5 35.00 D 0080050 0080275— U
30% as =.W 08:00 $03: so» 20 8%. 55.2 0o 8.5 as,» .55
| 5500000 95003

 

 

 

 

 

 

53 05 wage 00:0 5053— 2 809 WE“ 00:05 so» 20 88: E000ME. >008 30: .05

 

m0>D

_3 50:00.5 8 55m. 02 E
558000 ”Eamon SON 05 wESv z $0305.8— 50 000555 06 55 0:0 0 9 Hon 055 00:05 :0» ED .5

 

 

 

 

026 003 D
05% 50 0090.5: 058m D
05% 052 D
. :0 050 0.00 00 05E : 0m: :0» _ .
088 5 0 _ . PD 3 AI 00>D
a 508.5 e .50. oz D
5:00.000

wagon Son 05 9550 80m. $5 0008050 no» 05% mo 005:5: 05 005055: 00003 32 2Q .ow

184

 

 

 

0.800 .0200 D

00w05 00 3 .H. .0000 00m D 580 0.0.2 B 00..0.0000< m0..00m .U 00.000w0 M29 D

00...000>.0D D 0.00”. D 00.0.5.0... D 00.00w02 D 005050302 D
000..0E00..0. 00. ..0 0000000 00. 0003 .003 .0N.
00300000000000 000 0.0 00008080. ..0 005%. .003 .0N.

 

 

Al 8.5

 

n. 000800 3 00.0. 02 D
0000000 w0..000 Sam 00. $0.000 0.0>0. 00.03 w000000000 0000000000. 000 0>.0000 00 .000000 000 0.9 .m.

 

 

 

A0... 0000. 0000000 000000.
A502 009 0080.005 00.0.00 00.03 30. 0000. 000000500 000 0.0 0000000 .003 0. .0. .

 

0.800 .0200 D
0.00500005 0. w0...0m 0.0000 ... _H. 050005 .00... 0. w0..00m 0.0000... D 0000000. 00000 .0 .000 0. 0.0003 D
00.000w000 w0..000 0003. D .000 00.000..0m. m 00.0500\0w0..000 .00m U 0.0.0000000. 003 w0.0.000\w0.000$. B

00200000030000.0055 000 0.0 0000.005 ..0 0050. .003 .0. .

 

 

AI 300

.0. 00.08.00 3 00.0. 02 n.
0000000 90.000 .000. 00. w0.000 0.00.0005 00.0.00 00.03 30. >00 300.005.800.000000 0.0000005 00» 0.0 .00..0 00000. .0 000. 000.0 .. .

 

02.“.

 

.0... 0000. 0000000 0000005 A502 00% 0w0..000 00000 00 000 0.0 00.00000 .003 0. .00 > B
000.03 30. ..0 0000000 0000000. 0.0.000 .0 w0..000 00000 00 00» 0.9 .00.

020

 

 

.00... 0000. 0000000 00003. .50.). 009 0w0..000 0000 000000 00 50.0 000 0.0 0000000 .003 0. .00 > B
000.03 30. ..0 0000000 0000000. 0.0000 .0 w0..000 000000 00 50.0 000 0.0 .00.

:— 0030000 0. 505. 07. D

00> D
000.000 000 00003 0000000. 00. 00000.00. 00.03 30. 0.9 .00.

02 D

00> B
000.03 30. ..0 0000000 Sam 0. w0..000 000. 00 000 05 .00.

 

AI 8.6

 

.0. 8002.0 8 00.0. 02 D
0000000 w0..000 Sam 00. w0.000 900000 0000 00000.00. 00000000 00.03 30. 0.9 .o.

185

 

60.00000 0.0.000 0.0.9.2000". 0. 02.0.0000... 0.00:_0> 00.2005 ...3 03 5.0: 0:0... 5.0.. 05:00 25
000325 050.038 0.05-00.07.05 0... .... 0.0.0005 00 0000 00 .3200 00.0.5800 000.. 000.00 0000.5
$00.0 ...—0.005.... 0.... 0. 05.05.0055 00. 00> x000...

 

 

 

 

 

 

 

80.0w .0 00>0 D 000.005 - 08.8; _H. 000.000 -ooodom U
gadmméoodvm D 500.2% -08de .H. 08de 00003 .H.
0008 0. 0.0000. 0.0000000 0000w .00000 000» 003 .00 3 .w.
AOM0 ..0 0000.. x. 00000. 000.30 0..00< 00.000002. 000» 0. 00.000 0.000000 9000.0... 6.5005 00000 30... .0.
0000» .00. >0>000 0.0. 00.0.5080 000 .00. .000 0000.0.w00 00. ..0 00030 0... ..0 0m< .0.
02 _H.
0.0000 20% 000.000. .. 0. 0.0000 0% 0.0.0 .003 0. 00> D
00w0..00 00 0.00.. 00000550000000 000500.03 .0m.
02 .H.
0.0000 0.0.m 000.000. .. 0. 0.0000 0.0 0.0.0 .003 0. 00 > .H.
00.00.. 000500.03 .000000005 .0m. 00> _H.
.0. 02.3.5 3 00.0. 02 _u

00m0..00 00 0.000 00000033000000 00 000.00.03 .000000005 0 :30 3.000000 00» on. .m. 6

 

 

 

 

 

 

8
0.0000 0.8m 000000.000 .00000005 0000 0. 0.0000 000 0.0.0 .003 0. .0.
000D 000A. 000D
0.8m 0.. ..00 00 00305 00000 00.00050 0000. 0.8m 0. ..00 00 00305 00000 00.00050 000m. 0 . CON 0. ..00 00 00305 00000 00.00050 000m.
.00. m... .000 .0 00% .00. N0 .00050 00.m .00.. .... .00050 00%
000.0 D 000.0 D 000.0 D
20.0 .0.. .w.0. 000000.03 .0000005 U ....0 .0. .w0. ...00000.03 .0000005 D 20.0 .0.. .m0. 000000.03 .0000005 D
30.. 00 0000 U .H. 30M. 00 00000 _H. 30.. 00 00000 D
000.005 D 000.005 _H. 000.005 D
00305 ....3 0.0m D 00305 003 ...0m _H. .0305 003 0.0m D
000030500 ...0m D 000030500 ...0m D 000030500 ...0m D
00000.00 D 00000.00 D 00000.00 D
0000053000000. .H. 0000052000000. D 0000050000000. D
00000:. E 000000. E 000000. B
n u .00.. N ... .00.. . 0. .005

 

 

 

 

..000. w0...000 00. 00 00.....000. .000 00. w:.00.0xm. I 000000 000.0 0000. .0 0.0005530 00.>005 0000.5 .00..
0.00m. ..0 0. 000M323. 0. 0000.0.M00 3.000000 00 .00. 030 00% 00 0.000 000.0 0000. 30.. ..000. M050... 0... 00 00.....000. .000 00. M00008 .07. .m.

APPENDIX F

The 2001 Lake Michigan Potential Damages Study Telephone Survey

187

REVISED BY PZC AND EM 12/06/2001
Dec. 2001 Survey on Lake Michigan Boaters

 

 

 

 

SEQ# ZIPCODE: DATE: / /
PHONE STATE
COUNTY JURIS WARD PRECINCT

 

Hello, this is (NAME) from EPIC-MRA, a Lansing-based survey research firm.
The Army Corps of Engineers has commissioned a random survey of boat owners to
inquire about boating issues in [State being surveyed]. I am not selling anything, and
you will not be contacted again because of your participation in this survey. The
survey takes just a few minutes to complete, and I would like to include your

opinions.

 

May I please speak to (boat owner from list)? [IF NOT AVAILABLE, ASK] Is there
a time when I can call back tonight when (he/she) might be available?

[IF YES, CALL BACK TIME] [IF NOT AVAILABLE -- TERMINATE]
_01. In cooperation with The Army Corps of Engineers the State of [State Being
Surveyed] has provided us with boat registration records for our use during this survey.
These records indicate that you own a [Description of Foot Size of Boat From Call Sheet
-— e. g. twenty-seven foot boat] is that still correct?

(1) Yes ------- GO TO Q6
(2) No
(3) Refused — TERMINATE INTERVIEW

_02. Do you own another boat registered in your name in the State of [Name of State]?

(1) Yes
(2) No ---------- TERMINATE INTERVIEW
(3) Refused TERMINATE INTERVIEW

_03. What is the size, in feet of that boat?

# of feet IF UNDER 17 FEET, TERMINATE INTERVIEW

 

_04. For how many years have you owned your boat?

Years

188

_05. Can you please tell me what kind of boat that is, is it a. . .?
[READ 01 TO O9--CODE RESPONSE]

(01) Outboard motorboat

(02) Inboard/Outdrive motorboat
(03) Inboard motorboat

(04) A powered sailboat

(05) Pontoon boat

 

 

(06) An unpowered sailboat --------------- TERMINATE
(O7) Canoe or row boat TERMINATE
(08) Personal watercraft (jet skis) -------- TERMINATE
(09) Or something else TERMINATE
(10) Undecided/Don't know --------------- TERMINATE

_ 06. Have you operated the boat on the water in the past three years?

(1) Yes
(2) No GO TO Q. 10

 

_07. Have you operated that boat on the water on an inland lake in the past three years?

(1) Yes
(2) No
(3) Undecided/Don't know

_08. Have you operated that boat on the water on any of the other Great Lakes or a
waterway connecting to the Great Lakes in the past three years?

(1) Yes
(2) No
(3) Undecided/Don't know

_09. Have you operated that boat on the water on Lake Michigan or a waterway
connecting to Lake Michigan in the past three years?

 

(1) Yes GOTOQ13
(2) No CONTINUE
(3) Undecided/Don’t know ------ CONTINUE

 

189

_10. What is the main reason why you did not take your boat out in the past three
years? [DO NOT READ -- CODE RESPONSE]

( 1) Other competing interests (PROBE)
Write Comment as Stated or Code Appropriately

 

(IF REASON WAS LOW WATER LEVELS OR BOAT LAUNCH NOT
AVAEABLE, CODE “2” OR “3” & CONTINUE)

(2) Didn't use it because of lower water levels --CONTH\IUE INTERVIEW

(3) Boat launch or marina slip was no longer available - CONTINUE
INTERVIEW

(4) Lost interest (PROBE) ”Did you lose interest because of a speciﬁc reason?

Write Comment as Stated or Code Appropriately

 

(IF REASON WAS LOW WATER LEVELS OR BOAT LAUNCH NOT
AVAILABLE, CODE “2” OR “3” & CONTINUE)

(5) Became too expensive ------------ GO TO Q. 66
(6) Didn't use it for other reasons---- GO TO Q. 66
(7) Undecided/Don't know ----------- GO TO Q. 66

_11. Based on your experience or what you have heard from others, would you say that
Lake Michigan water levels have dropped or remained about the same over the
past few years in your usual boating waters? [IF DROPPED, ASK] Do you think
they have dropped a lot or just a little?

(1) Water levels dropped a lot

(2) Water levels dropped just a little

(3) Water levels have remained about the same
(4) Undecided/Don’t know

_12. Speciﬁcally, over the past two or three years, how much, either in inches or in feet,
would you say Lake Michigan levels have dropped?

feet inches

GO TO Q. 66

I90

_ 13. How many days did you operate the boat on Lake Michigan or connecting waters

in 2001?

Number of Days

_ 14. Is this more days in 2001 than in 2000 or less days?

_15.

 

(1) More

(2) About the same number of days in 2001 ----- GO TO Q. 18

(3) Less days GO TO Q 16

What would you say is the main reason why you were out on the water more this

past season than one or two years ago?

[DO NOT READ — CODE OR WRITE IN OTHER RESPONSE & GO TO Q. 18]

_16.

17.

(1) More income

(2) More free time

(3) Retired

(4) Has a new boat

(5) Other (Please specify):
(6) Undecided/Don't know/Refused

 

What is the main reason why you took your boat out fewer days in 2001 on Lake
Michigan or on a connecting body of water to Lake Michigan than in 2000 or
1999? [DO NOT READ -- CODE RESPONSE]

(1) Boat isn't large enough for Great Lakes waters

 

 

 

 

(2) High fuel cost GO TO Q. 18
(3) Lower lake levels GO TO Q. 18
(4) Have no place to launch or keep boat ------------ GO TO Q. 18
(5) Public access problems/lack mooring space ----- GO TO Q. 18
(6) Other (Please specify): GO TO Q. 18
(7) Undecided/Don't know GO TO Q. 66

When you say that your boat isn’t large enough for the Great Lakes waters, do you
mean that it cannot safely operate on the Great Lakes on a windy or stormy day and
that it is ﬁne on a calm day, or do you mean it is not big enough regardless of the
weather?

(1) Cannot safely operate on the Great Lakes on a windy or stormy day
(2) Not big enough regardless of the weather -------- GO TO Q. 66
(3) Undecided/Don’t GO TO Q. 66

 

191

_18. In what county and state do you usually keep your boat during the boating season?

(01) (Please specify):

 

(02) Undecided/Don’t know

_19. Can you tell me in what port city or community you usually boat out of during the

boating season? [WRITE COMMENT AS STATED]

 

_20. At what type of place or facility did you keep your boat in the 2001 boating

21.

22.

23.

season? [READ & ROTATE l — 6]

(1) Permanent residence

(2) Cottage/second home

(3) Rented slip in a publicly operated marina
(4) Rented slip in a privately operated marina
(5) Owned space in a marina or dockaminium
(6) Yacht/Boat Club

(7) Other (please specify:)
(8) Refused

 

How did you keep the boat during the 2001 boating season? — Did you keep it . . .
[READ & ROTATE l - 3]

(1) On land outdoors

(2) In a dry stack facility
(3) In the water -- that is, in a wet slip or at a dock or moored in the water
(4) Other (please specify:)

 

During the 2001 boating season, was the location you kept your boat different than
in the 2000 boating season?

(1) Yes
(2) No ------ GO TO Q. 24

What was the main reason why you moved your boat?
[DO NOT READ -- CODE RESPONSE OR WRIT E-IN UNDER OTHER]

(1) Lower water levels

(2) Other (Please specify):
[PROBE to see if “other” reason is related to low water, if so, Code “1”]

(3) Undecided/Don't know/Refused

 

192

_24. At what type of place or facility did you store your boat last winter?

25.

_26.

_27.

28.

[READ/ROTATE l — 7]

(1) Permanent residence

(2) Cottage/second home

(3) Dry stack at a marina

(4) Outdoor rented space at a marina

(5) Outdoor rented space at a storage yard
(6) Indoor rented space
(7) Yacht/Boat Club

(8) Other (please specify
(9) Refused

 

From what type of place or facility did you typically operate your boat for trips
onto Lake Michigan or waters connecting to Lake Michigan?

(1) From a dock or mooring site in front of a waterfront home? GO TO Q 45
(2) From a marina slip ----- CONTINUE

(3) From a boat launch ramp GO TO Q. 28

(4) Other (please specify) -- GO TO Q 45

(5) Refused GO TO Q 45

 

 

Can you tell me the name of the marina?

(1) Marina [Name]

 

Based on what you know or have heard or read, how available are slips in the area
where you usually use your boat - very available, somewhat available, only
slightly available, or not available at all?

(1) Very available

(2) Somewhat available
(3) Only slightly available
(4) Not available at all

(5) Undecided/Don’t know

GO TO Q. 45

Can you tell me the name of the County where that boat launch is located?
[CODE RESPONSE OR WRITE IN UNDER OTHER]

(1) (Please specify):
(2) Undecided/Don't know

 

193

_29. Do you usually use the same boat launch or do you use others?
[CODE RESPONSE OR WRITE IN UNDER OTHER]

(1) Yes, always uses the same boat launch
(2) Uses other boat launches
(3) Undecided/Don't know

30. Thinking about the times that you used the facilities of public boat launches while
taking your boat out during peak use times such as weekends and holidays...

Overall, how would you rate the facilities — such as launch ramp, parking, rest
rooms, picnic areas —- available at the public boat launches you may have used in
the past couple of years — would you give the facilities a positive rating of excellent
or pretty good, or a negative rating of only fair or poor?

(1) Excellent

(2) Pretty good

(3) Only fair

(4) Poor

(5) Undecided/Don’t know
I’m going to read several characteristics of boat launch sites and I would like you to tell
me how important each one is to you in your evaluation of a boat launch sites - Please

tell me if the item Very Important, Fairly Important, Only a Little Important or Not
Important at All in making your overall evaluation of a boat launch site?

Very Fairly Little Not. DK/
[READ AND ROTATE Q’S 31 TO 37] Impt. Impt. Impt. Irnpt Undec

_31. The availability of parking ................. (1) (2) (3) (4) (5)
_32. The condition of the ramp surfaces... (1) (2) (3) (4) (5)
_33. The repair and upkeep of the ramps. . .. (l) (2) (3) (4) (5)
_34. The quality and upkeep of restrooms. .. (1) (2) (3) (4) (5)
_35. The condition and upkeep of the docks. (1) (2) (3) (4) (5)

_36. The signs provided to help you ﬁnd the
facility ...................................... (l) (2) (3) (4) (5)

_37. Water depth at the end of the ramp. . (l) (2) (3) (4) (5)

194

Which of the following problems have you experienced while using a launch site?

[ROTATE Q’S 38 TO 43] Xe_s ﬂ DK
_38. A launch site closed because it was at capacity ........ (l) (2) (3)
_39. Long wait launching ..................................... (l) (2) (3)
_40. Delays in retrieving your boat ............................ (1) (2) (3)
_41. Dangerous channels out to open water ................... (l) (2) (3)
_42. Boat congestion ........................................ (1) (2) (3)
_43. Reduced area to maneuver your boat .................... (l) (2) (3)

_44. What would you say is the longest amount of time you are willing to wait to use a
boat launch before you can’t wait any longer?

[WRITE IN MINUTES OR HOURS]
Minutes Hours

45. How far a distance in miles is it from your home to where you most often operate
or launch your boat for boating on Lake Michigan?

- Miles

46. How long does it normally take you to travel from your home to the place where
you most often operate or launch your boat during the boating season for boating
on Lake Michigan - a half hour or less, a half hour to an hour, more than an hour
but less than 2 hours, more than 2 but less than 4 hours, or more than 4 hours?

(1) A half hour or less

(2) A half hour to an hour

(3) More than an hour but less than 2 hours
(4) More than 3 but less than 4 hours

(5) More than 4 hours

(6) Undecided/Don’t know

47. What boating activities do you like to do in Lake Michigan or connecting waters?

(1) Fishing

(2) Cruising
(3) Water skiing, tubing or wakeboardin g
(4) Other (please specify)

 

195

_48.

_49.

_50.

51.

52.

_53.

How well have ofﬁcials at the federal, state, county, city and township levels dealt
with low Lake Michigan water levels? Did they deal well with low water or deal
poorly with low water?

(1) Dealt well with low water
(2) Dealt poorly with low water
(3) Undecided/Don’t know

Tell me in terms of a percentage amount, How much of your boat registration fee
you believe goes to help maintain launch facilities and channels?

(1) All of them or, 100%

(2) More than half of them, or 50% to 100%
(3) Less than half or, 1% to 50%

(4) None of them, or 0%

(5) Not sure

Would you be willing to pay a fee in which the revenues go toward dealing with
water level problems and boating facilities?

(1) Yes
(2) No --- GO TO Q. 52
(3) Uncertain

How much would you be willing to pay in fees or taxes for improved facilities to
deal with water level problems? [WRITE 1N AMOUNT]

per year per use
When you use your boat on Lake Michigan or connecting waters, how many
people, on average, usually travel on the boat with you, including family and

friends?

people

What organization or department, as you understand it, is in charge of boating in
your state? [WRITE COMMENT AS STATED]

196

_54. Which of the following sources of revenue are used to pay for the development and

maintenance of public launch facilities . . . is it
[ROTATE 1— 4, Code all responses]

(1) Boat registration fees

(2) A portion of the fuel taxes

(3) Fees paid at the launch sites

(4) From the general fund appropriated to the DNR
(5) Other (Specify):
(6) Undecided/Don’t know

 

The Federal, State, and Local governments all use some taxpayer dollars to pay for public
facilities and general welfare projects involving Lake Michigan and its connecting
waterways. Speciﬁcally, can you tell me which level of government — Federal, State, or
Local, is primarily responsible for: [ROTATE QUESTIONS 55-56]

_55.

_56.

57.

Developing and maintaining public boating access facilities — is that ﬁmction the
primary responsibility of the [READ & ROTATE 1-3 ]

(1) Federal government

(2) State government

(3) Local county, city or township government
(4) A combination of them (V olunteered)

(5) Undecided/Don’t know

Dredging channels and harbors — is that function the primary responsibility of the
[READ & ROTATE 1-3 ]

(1) Federal government

(2) State government

(3) Local county, city or township government
(4) A combination of them (Volunteered)

(5) Undecided/Don’t know

Based on your experience or what you have heard from others, would you say that
Lake Michigan water levels have dropped or remained about the same over the past
few years in your usual boating waters? [IF DROPPED, ASK] Do you think they
have dropped a lot or just a little?

( 1) Water levels dropped a lot

(2) Water levels dropped just a little

(3) Water levels have remained about the same ----- GO TO Q 60
(4) Undecided/Don’t know

197

_ss.

_59.

60.

61.

62.

Speciﬁcally, over the past two or three years, how much, either in inches or in feet,
would you say Lake Michigan levels have dropped?

feet inches

What is the main problem you have noticed as a result of lower water levels over
the past two or three years? [WRITE COMMENT AS STATED]

 

In 2001, the water level of Lake Michigan as about 1 foot above its record low
level recorded in 1964. If Lake Michigan water levels were to drop a foot below
this past year 2001 levels, would that force you to seek alternative sites to Operate
your boat, or, would you say that you would not have a problem?

(1) Would have to seek alternative sites ----- GO TO Q. 63
(2) Would not have a problem
(3) Undecided/Don't know

If Lake Michigan water levels were to drop two feet below this past year 2001
levels, would that lower water level force you to seek alternative sites to operate
your boat, or, would you say that you would not have a problem?

(1) Would have to seek alternative sites ----- GO TO Q. 63
(2) Would not have a problem
(3) Undecided/Don't know

If Lake Michigan water levels were to drop three feet below this past year 2001
levels, -- in other words, two feet below the lowest level recorded in 1964 -- would
that lower water level force you to seek alternative sites to operate your boat, or,
would you say that you would not have a problem?

(1) Would have to seek alternative sites

(2) Would not have a problem
(3) Undecided/Don't know

198

_63. How about the opposite problem? What kind of an impact would there be on
boating on Lake Michigan if there was an increase in water levels in the next 10 or
so years of about a foot over previously recorded high levels, which were recorded
in 1986, which would be nearly 6 feet higher than this past summer -- would it
have a major impact, a minor impact, or no real impact at all?

(1) A major impact
(2) A minor impact
(3) No real impact at all ---------- GO TO Q. 65
(4) Undecided/Don't know ------ GO TO Q. 65

64. What would you say would be the main impact of that kind of an increase in water
levels? [WRITE COMMENT AS STATED]

 

_65. What would you say is the main problem confronting boaters in (Name of State)?
[WRITE COMMENT AS STATED]

 

Finally, I would like to ask you a few questions about yourself, for statistical purposes
only.

_66. In terms of your job status, are you employed, unemployed but looking for work,
retired, a student, or a homemaker?

(1) Employed

(2) Unemployed

(3) Retired

(4) Student

(5) Homemaker

(6) Other

(7) Undecided/don't know
(8) Refused

199

_67. Could you please tell me in what year you were born?
[RECORD YEAR HERE AND THEN CODE BELOW]
(1) 18 to 24 years -- (1977 to 1983)
(2) 25 to 29 years -- (1972 to 1976)

(3) 30 to 35 -------- (1966 to 1971)
(4) 36 to 40 -------- (1961 to 1965)
(5) 41 to 49 -------- (1952 to 1960)
(6) 50 to 55 -------- (1946 to 1951)
(7) 56 to 64 -------- (1937 to 1945)
(8) 65 and over ----- (1936 or before)

(9) Don't know/Refused

_68. What is the last grade or level of schooling you completed?
(DON'T READ-CODE RESPONSE)

(l) lst to 11th Grade

(2) High School Graduate

(3) Non-college post high school (technical training)
(4) Some college

(5) College graduate

(6) Post graduate school

(7) Undecided/Don't know

(8) Refused

 

_69. Would you please tell me into which of the following categories your total yearly
household income falls --- including everyone in the household? Please stop me
when I get to the category that applies to you? (READ LIST --- 1 to 7)

(1) Under $15,000

(2) $15,000 to $30,000

(3) $30,000 to $45,000

(4) $45,000 to $60,000

(5) $60,000 to $75,000

(6) $75,000 to $100,000

(7) Over $100,000

(8) Retired (VOLUNTEERED - ASK: But is there an income category I read that
would apply to your household?)

(9) Refused

_70. Sex of respondent (BY OBSERVATION ONLY)

(1) Male
(2) Female

THANK RESPONDENT FOR HIS OR HER TIME AND TERMINATE

200

APPENDIX G

Five Binary LOGIT Model Results

201

 

.. .. -- I l t -- Sod owe—Bo SE5

-- -- -- -- -- -- -- ~86 ewéeo

-- .. -- -- -- -- I -- 633680
cormostm

god. 390. mwod. :: .0. wood. Rod. 2: .o- mﬁ .o- vam 895

$35. ...omvd- ...med. .553- *vad- ...momd- ...N—md- ...wmmd- Move 8 Mmg

.. l 1 l l ... ... ... Meow .5>O
0885

-- -- Nomd mend 38.0 Sod ...oood ... ﬁ 36 Son cooﬁom

-- -- -- -- cnmd mcmd owmd wmmd mEmoﬁsoébwonE

-- .. -- m _ Nd awvd .3an ...mwwd «936 Ron .8on 28330
as “mom
-- -- -- -- I -- mood- mood- Ewan: “mom

3:: ...mmvg soil *vaA 191 ...wﬂl $91 New; 83:28 24 8 :82

sumo; .334 L394 .53; echo.“ 1.3.— ...owog sumo; 3:566 53:32 8:3
8528.. “5.88qu mo £583

.386 *m E ed $3.0 ... ﬂ Nod .L mod ... ﬂ mod ... _ $6 .336 2865

.536 cmmd .0.de _de wmmd mmmd mmmd wand c5885»

...mwwg *mmw._ Lwn; LE; _5ch swan; *2»: *mmw; SawEomE
8:368 m0 033 “accustom

w _0562 h .252 w .232 m .252 v _02.—Z m .252 N .0632 _ 1:52

.mv gnu _D>0— Mount, Cm _30—2 033 MO mmOCOHNBm .mhowwOD MOM 2060—: :00..—

  

es 3 _e we £52 2:. ...e 2.5

202

2.0 V magi ......
mod V migi ..

 

 

 

 

 

5% I .55 .52. .55 .55 .58 Raw .59. 52.2.55 5850 R.
GE 0.823
L235. 225...: L2: .2 ..258. EN; EM. .2 .25.... .23.: .5 529.5580
5.... we... ...... 5.5 we... we... ...... we... .2 8.55.532
m3 mm... 2.5 m2 2.5 5.5 5.5 5.5 a: :65 .... .80
2.3.. 2.3.. 8.5.. 5.8.. 5.5.. 525.. 2.5% 8.5.. 52.28... meg N-
.533. .82....3. ... 2.5.. .3233. .5298. .5233. .5232. .5255. E2 5535-..... 5.52
.533. .82m. 5.. ... 2.5.. .3255: .555. .5239. .5233. .555. :3 65:55.50 .85
..22- ..22- ..2....- .22.... ..23- ..23- ..23- 2.255.. :3 653.50 Ram
.33 .35.» ...m. ... as... .3; 52.5 .55? .39. .5355
LE... .83 .23 .SR .82 5me .82. .82. 5552.2 8.... co
-- $2. 83 En... 3m... . woo $3 Sm... m8.3 520 555 so
-- -- -- -- -- 83. 82.. two- $5.... 55.... :0
2882.0 225 3.89 82.3
. . we... .35... ..S... .25... L8... .95... .95... .355 5.550
a 5.5:. b 5.52 5 5.52 m .252 v 5.52 m 5.52 N 5.52 . 5.52

....Eoo I mowgno .o>o. .855 Suwﬁoaz 8.3 we 52.85.55 .5882 82.225... .Dmvmbacﬁ Em... mo 838. BE. ...0 2an

203

 

 

 

 

. .255 . 52.25 . ..2 ~35 . .~2 ~55 . .2. ~25 a... 8.82.8 .252
.555 .8235 225.5 ..~2~.m5 .52 ~25 a... 88528 V.8...
.223. .258? .2585. .2355. ..~2~mm5 a... 883.8 58m
.85 5m~5 -55 ~25 525 52250
.. -- -- 28- 28- 2...... 22. I 5.2.8 225m
.-m5 8.5 .5..5 .585 .225 8.258 .288 82222 8.5.. ... 58. 83 55m
29.5 .55 ~55 .55 ~85 22.8.2
....~5- ~..~5- -~.- -~5- 28....
...-5 .~m5 ~55 285 555 2.2.5.
.82. 3.5 8.5 5.5 225 25823
22.80. .0550 owﬁoa .mom
.~5~..- .52- .-~..- .5-..- ...5~. .- 8.5.. .588 .5. .2250
-- -- -- -- 255- 8.52
02b owes. .50m.
.555- .555- .255- .255- .255- .85.. 580 .825 ..o 5a.. 8.. 8.28.0
~52 ...~.m ~22 5.... ~25 58 58.8..
.25. .- .25. .- :55- 8.5- ~25- .8. 55:. .888.
.~m..5- ...~..5- .~5~5- .~..~5- .5~5- 858.8558...
-- -- 5 . 5 5m . 5 ..~ . 5 .8. 55... E2550
on... .53.
.255 .~..55 .~..55 .255 .~..o5 .5:>8 88. m5.. 58. 8.. .28.. .o 52.22
.~55 ..55 .55 .~55 ..85 22.2 55m
m .252 v 3.5:. m .25: N $63). 2 .253).

 

.mowcmno .o>o. .82.» 252222 8.3 .0 222.288.. .2252. .8 20.50... .5004 2.5:... P»... ..o 2.28. 2.... Nd 0.5.5...

204

o. .o V mégi .1.
36 v mﬁgi ...

 

 

 

 

2...? $3: 2.. .9 $2... 2.3:. 82.2.5... .550 .x.

$55 52.... .... . 2 .... .3. 5.5.5. :3 253-30 2.2935580

52... S. .o 5.5 R... S. ... N: 8.52252

8.... 8.... 8.... 8.... 8.... N2 ..8m 2 .80

$.23 .223 5.5% $.55 52.5% 82:8... 2: N-
m .252 v .252 m .252 N .252 . .252

E
63:00 I mowcano .o>o. 2033 5&322 873 go 3:02.383 .8889 28 2255 E00: 3:5 9»... .«o 3.89. on... ”Nd 29¢.

205

828:8 .888 8.892.).

206

 

EN... EN... 95... gm... 3N... Cm... «mm: 936 3m...
8:... 8 EB. 83 .89 09,—.
.8... .8... .8... .8... .8.... ..3... .8.... .35... 8...... 2:2. 558253-87.
0:89
- .. - - - - - - ..8...
808280: .:o:.t8:>>
- - - on . .o- :9. .o- mm _ .o- Sm..- wow... 9:. .o- «8.82
0.5 “$808 Som
.. - - .. .. - - 8.... So... 8...... .85 59.0 :O
- - - - - - 8.....- 8.....- 8.....- 828.2 8.3 =0
93.. 89.89 .8 .098: 2
8058232880 w:..mo9
.26 .26 .m. .o .26 .Nm..o .92... .mm. .o .mm..o .wm..o . . .
880...... .8 598:2
Rd mnmd 92d mmmd mmmd :26 9:6 99:. V9.6 :ooucom
3.0- omvd- wmvd- wand- oomd- o3..- wit..- 9%.... 82¢. 809 :mm
- - - - - :2 .o- 3 . .o- :2 .0. mm 9 .o- 9:09 .888 9.809930
.36. .36. .mde. .33.. .936. .wmmd- .vmmd- ...de- .wmmd- 88098068095
25 8.5
90:30 :39 mm:
- - - - wood- wood- wood- wood- wood-
:89 09. 88% mo 898:2
.8... .8... .8... .8... .m8... .5... ...8... ...8... .5... 582 5cm
5 .252 .. .252 h .252 e .252 m .252 v .252 _ .252

n .252 N ...—.32

 
 

‘ 22,12 :88» 3:.
o. gum—o: 8:00:00 .29 98:02.8. .0: .8: 98:02.8. 225 093 8889 09. ...... 22.08 .2009 .9889 «8: mo 39:8: 09,—. .m.O 298.

o_.o V 0383.2 ......
no.0 V 03.97% ...

 

 

 

 

 

 

 

 

207

a a .88 ......8 ......8.,.. 2N8 8...... 23.8. ...N... 82.2.5... 5250 ....
2.3.8.2 .... .0... 3.8.2 .....N... 3.3... 3.3.2 3.3.2 3.8.2 3.3.... -..... ...-...meme
m... m... m... m... m... m... 2... m... m... «2 8.52.082
...0 .... .... .... .... .... .... .... .... 2:08.280
2.83 2.83 8.8.. 0.83 00.3 3.3 .23 0.23 2.3 900....8... mo: N-
2. ..83 22.8.3 22.8.3 22.3.3 22.8.3 22.8.3 22.3.3 22.8.3 22.3.3 ...... 05:03-20 .0902
2. ..83 22.3.3 22.8.3 22.3.3 22.3.3 22.8.3 22.3.3 2......an 22.3.3 E9. 05:03-20 0.02m
22.....- 2:3..- 22......- 2..3..- 2:8... 2:2..- 2:3..- 2:8... 22.3.3 :9. 05:03.20 new
.83. .2 ON. 83- .38.- 2...: 3...... 8.... EN... 3 . .... .8800

90:8...
.23... .23... .83... .33... .03... ...3... .83... ..3... .53... 0. 90.2.2. 33 .89 09.2
20.6. 20.03
.88... ....a... .38... .33... .N8... ...8... .58... .N8... .8... 8.88:8 8.580....
2:0 90300029050802
3.... 33... 88.... 53.3 2.... 33.3 ..8... 38... 3.... 82.5.
- - 2 ..m 32 ..2... 82.3 N2... ......3 3.3 2.5....
- 2.... 3. ..m 55.3 3.3 .2... 3.... 8.3 E...” 28823
22.... .32. 33.3 2....” an... 33.3 30.3 .8... 83.3 80222
£03.30—
28:00 02203 .mom
N .252 0 .252 m .252 v .252 m .252 N .252 . .252

w .0902

 

 

9.800 I 3.0.20. .083 30.
0. 89.20.. 3:200:00 29 900:0:98 .0: 9:: 900:0:98 0.03 093 80.009 09. ..0. 3.0908 .2009 2.889 0:8 .20 02:00: 09..- .m.O 0.90..-

 

 

 

o. .o V 050.72 ...... no.0 V 03.0.72 ...

 

 

 

 

 

.205. .033 30. 0. 03:030. :. 3:09.80. @9009 .909. 90w:..90 093 20.009 09. .02 2090:. .5009 2.8:.9 .93 ..0 2930.. 092- .....0 0.9....-

333 .8.... .8.... .8.... .33 3...... 8090.00... .8900 3
3.03.3 3.30.2 3.33.. ......82 3.03.2 3.00.... c... 2.32... ...-3.203000
23 03.0 33.0 ...-N... 3m... 3m... ... 0.902.032
32.0 32.. 32.0 ..2... 3.... 3.0 ... =03 ... .80
33.2.. 03.0... 33...... 3033 30.2.. 20.2.. 0002.00... 30. -
. 3.2.0.... . 3.300.. . 3.0.0.2. . 3:03.... . .. ..032. . .2332. c... 0.33-20 .0002
. 2.2.0.... . 3.300.. . 3.0.0.2. . .....032. . .. ..032. . .2332. a... 0.33-0.0 .00...
...33..- ...300..- 2.02.- E03.- ...2....- 22.33.... a... 0.83-9.0 00.3
.329..- .9392- .0322..- .9922:- .. 9 9.00- .1020...- 9:30:00
32.. 200- 82.- 0.00.. 2.09.0.2» . 030
20>0. .083 m:.:-.00:00
.32. .3090 .33... .03... 030... 30... 0090920.... 2.... 90200050300002
20.0 20.0 0.0.0 ......0 20.0 .20.. 2.9.09 5.290.... 8.... .0 .0990...
.00...- 3209 w:9009 ..0 .098: Z
...... 2...... ...... .3... ...80 .22. 209.00.300.08 2.....09 2.0.0.2.. .0 590.02
.03.. .08.. .0. n. 8...... .3... .3. .009 ....m
.03.. ...... .83.. .32... .33.. ...3. .009 .90... 9309.00
.2~.. .30.. .2... ..3... .33... .03... 39809209309...
.2... ..0... .3... .30.. .8... 38.0 .89 .80... 9.89...
.NS... .990... .30... .mmo... .08... .98... 90:30 0:? .009 09. 30>
m2... m2... 9&5. .mom
3......- 9.o.o- 2.....- So...- 0w<

0 .0002 m .0002 .- .0002 m .0002 N .0002 . .0002

208

 

 

.ommo- owed. .ooZ- mmmim- omw.~m- cmmom- mwmsm- wowsm- oom.wm- 59200

.33 .too .33 .32. .33 .33 .82 So .ovho 08o: Egret“; m :30

£26— Boa?

- - - - ommo cmmo momo N_m.o :mo wEEoocoo sotmgoﬁ:

.98 ouZuooSoBmozcom

- - - - - - 0:3 82 82 382mm

- EB 2:3 8% 2 no .womo .mmco .Soo 89o 83 888 Emoaso

- - - - - - - - mmoo $3330-?885

- - - - - 23 83 $2 80o Eon 888 2895

223832258 maroon

.oﬂ .o- .3 _ .o- .o: .o- .o: .o- .Q o .o- .3 _ .o- .wf .o- m: .o- 2 _ .o- Subway .8 H3832

8&3 380 550

- - - - - - - oooo- oooo- co 3% wagon mo 55:52

mbwo maroon

.23. .23. .29? .Eoo- .Eoo- .m_o.o- ._~o.o- .mmoo- .mmoo- camwnozz 8133 89:52

.23.. .23. .Eoo- .23. .23.. .29? .Soo- .Qoo- .N_o.o- 3% maroon oo bag—Z

3:30 :89 was

- - - mmoo omoo wmoo wmoo wmoo mmoo Son 2: Bo» mo 82:52

.33 .ﬁoo .33 .83 .83. .33 .qmoo .33 .33 5&62 28m

- - o_o.o Eoo Eoo Boo Boo Boo Boo ow<
o .252 m 3.52 h .352 o .252 m .252 v .252 m .252 a .252 _ .252

.233 $33 32 8 8:032 E wagon 32 2c on? $885 05 8m 2358 .5004 3:5 05: mo 3:58 of. 6.0 053.

 

209

 

2.0 v mégi ......
36 v 3:91 ...

 

 

 

.\ . . . . . . . . . . . o . . . . . 8038a
1 E \cm 8 \1 we so 8 x: we \..o 8 \..m 8 \..m 8 \cm 8 8050 X
. . . . . . . . . :2 Maggie
ES o E? e E3. 2 EMS m Ewen M 5.8 w Ea: m Em: m 5.3 n Eooéoéoow
83 33 33 3.3 $3 $3 $3 on; am; we vioﬁowmz
goo 53 «So mo; 2 _.o 2 o M: o oﬂo oﬁo NM zoom a. xoo
218v mnoaov 398v mmooov wwmdﬂ £33m ooosam Knoom 83% 30:28.: m3 m-
. . . . . . . . . oz
.5va _m .ESo 2 .Eg X .HavS om .53m 2 .838 mm .2 :20 a ham: 9. .5: : 8 08:33“. Boo:
Eva: €892 Saw: 5:82 5 33». 8:26? : 329% 5:23 5: _ 2:. c2
. . o . n o o o o Dhﬂmﬂﬁvmlmno MOO—m
ESE- €33- ESE- ESE- ESE- 238.7 338. 383. .2: o 2:. CE
. 8533-50 3%

o 7:52 m .0632 h 3.52 o .252 m .232 v .332 n .352 N Eco—z — .252

.PEoo I £26— 583 32 9 omcomme E wagon 32 U6 on? 2889 2: 8m £0on .5004 .983 2:: mo 8:58 22. .md 2an

210

BIBLIOGRAPHY

Adamowicz, W., Louviere, J ., & Williams, M. (1994). Combining revealed and stated
preference method for valuing environmental amenities. Journal of
Environmental Economics and Management, 26, 271-292.

Agresti, A. & Barbara, F. (1986). Statistical Methods for the Social Sciences. San
Francisco: Dellen Publishing Co.

Arcury, T. A. & Christianson, E. H. (1990). Environmental Worldview in Response to
Environmental Problems. Environmental Behavior, 22, 387-407.

Arcury, T.A. (1990). Environmental Attitudes and Environmental Knowledge. Human
Organization, 49(4), 300-304.

Baird, W. F. (2000). Potential Impacts and Costs of High and Low Lake Level Scenarios
on Lake Michigan Harbor Structures. Detroit: US. Army Corps of Engineers,
Detroit District.

Bell, P. A., Greene, T. C., Fisher, J. D., & Baum, A. (2001). Environmental Psychology
(5th ed.): Earl McPeek.

Ben-Akiva, M. & Lerman, S. R. (1985). Discrete Choice Analysis-Theory and
Application to Travel Demand. Cambridge: The MIT Press.

Berkson, J. (1953). A Statistically Precise and Relatively Simple Method of Estimating
the Bioassay with Quanta Response, Based on the Logistic Function. Journal of
American Statistical Association, 48, 565-599.

Chubb, M. (1967). A System Analysis and Spatial Demand Approach to Statewide
Recreation Planning: A Case Study of Boating in Michigan. Unpublished doctoral
dissertation. Michigan State University, East Lansing.

Chubb, M. (1968). 1968 Michigan Recreational Boating Study. Recreation Research and
Planning Unit. Department of Park and Recreation Resources, Michigan State
University, East Lansing.

211

 

Chubb, M. & Chubb, H. (1975). 1974 Michigan Recreational Boating study. East
Lansing: Waterways Commission and the Law Enforcement Division, Michigan
Department of Natural Resources.

Cramer, J. S. (1991). The LOGIT Model: an Introduction for Economists. Great Britain:
Edward Arnold.

Cox, DR. and Snell, E]. (1989). Analysis of Binary Data (2nd edition), London:
Chapman & Hall.

De Young, R. (1999). Environmental Psychology. In D. E. Alexander & R. W. Fairbridge
(Eds), Encyclopedia of Environmental Science. Hingham: Kluwer Academic
Publishers.

Evans, G., Smith, C., & Pezdek, K. (1982). Cognitive maps and urban form. American
Planning Association Journal, 48, 232-244.

Field, A. (2000). Discovering Statistics Using SPSS for Windows. Thousand Oaks: SAGE
Publication Inc.

Fisher, J. D., Bell, P. A., & Baum, A. (1984). Environmental Psychology (2nd ed. ). New
York: CBS College Publishing.

F iske, P. R. (1974). A Regional Analysis of Recreational Boating Participation in The
State of Michigan. Unpublished Ph.D. Dissertation, Michigan: Michigan State
University. East Lansing.

Fransson, N., & Garling, T. (1999). Environmental Concern: Conceptual, Deﬁnitions,
Measurement Methods, and Research Findings. Journal of Environmental
Psychology, 19, 369-382.

Garling, T., & Golledge, R. (Eds). (1993). Behavior and Environment: Psychological
and Geographical Approaches. Amsterdam: North Holland.

Garden, G. T., & Stem, P. C. (1996). Environmental Problems and Human Behavior.
Needham Heights: MA: Allyn & Bacon.

Gujarati, D. N. (1995). Basic Econometrics. New York: McGraw-Hill, Inc.

Hair, J.F. Jr., Anderson, RE, and Tatham, R.L. (1987). Multivariate Data Analysis. New
York: Macmillan.

212

 

Hanson, T. R. (1998). Economic Impact of Environmental Characteristics on
Recreational Demand. Auburn University, Aubrum, Alabama.

Hanson, T. R., & Hatch, L. U. (1998). Impact of Reservoir Water Level Changes on
Lakefront Property and Recreational Values. Department of Agricultural
Economics, Mississippi State University and Auburn University.

Heimstra, N., & McFarling, L. (1978). Environmental Psychology (2nd ed. ). Monterey,
CA: Brooks/Cole.

Herriges, J. A., & Kling, C. L. (Eds). (1999). Valuing Recreation and the Environment.
Northampton: Edward Elgar Publishing, Inc.

Hof, J. G., & Kaiser, H. F. (1983). Long-Term Outdoor Recreation Participation
Projections for Public Land Management Agencies. Journal of Leisure Research,
15(1), 1-14.

Hosmer, D. W., & Lemeshow, S. (2000). Applied Logistic Regression. New York: John
Wiley & Sons.

Howell, S. E., & Laska, S. R. (1992). The Changing Face of the Environmental Coalition:
A research note. Environmental and behavior, 24, 134-144.

Huberty, C. J. (1989). Problems with Stepwise Methods: Better Alternatives. Advances in
social science methodology, 1, pp. 43-70.

Illinois Department of Natural Resources (2003). Application for Watercraft title and
registration. Illinois Department of Natural Resources, Retrieved January 12, On
line access: http://www.dnr.state.i1.us/admin/systems/Boats/pdf/boatappOZ.pdf

Indiana Bureau of Motor Vehicles (2003). Application for Vehicle/Watercraft Title.
Indiana Bureau of Motor Vehicles, Retrieved January 12, 2003, on line access:
http://www.in.gov/bmv/fonns/46449.pdf

Iso-Ahola, S. E. (1986). A Theory of Substitutability of Leisure Behavior. Leisure
Sciences, 8(4), 367-389.

Jackson, EL. (1986). Outdoor recreation participation and attitudes to the environment.
Leisure Studies, 5, 1-23.

Judge, G. G., Hill, C. R., William, E., Grifﬁths, Lutkepohl, H., & Lee (1980). Theory and
Practice of Econometrics. New York: John Wiley & Sons.

213

 

Kaplan, S. & Kaplan, R. (1982). Cognition and Environment. NY: Praeger.

Keillor, P. (2002). Twenty Frequently Asked Questions about Great Lakes Water Levels
and Coastal Erosion. UW Sea Grant Advisory Services. Retrieved October 5,
2002, on line access: http://www.seagrantwisc.edu/communications/LakeLevels/

Kurth, J. (2002). Shrinking Great lakes Threaten Michigan Way of Life: Tourism,
Shipping and Fishing Industries Feel the Impact The Detroit News.

Lee, H. C. (1999). Trip Expenditures of Recreational Boaters in Michigan. Unpublished
PhD dissertation, Michigan State University, East Lansing.

Loomis, J. B. (1995). Four Models for Determining Environmental Quality Effects on
Recreational Demand and Regional Economics. Ecological Economics, 12, 55-65.

Long, J. S., & Freese, J. (2001). Regression Models for Categorical Dependent Variables
Using Stata. College Station: A Stata Press Publication.

Lupi, F. (1998). Valuing the effects of water quality on recreational resources:
Discussion. American Journal of Agricultural Economics, 80(5), 1038-1041.

Lupi, F., Hoehn, J. P., Chen, H. Z., & Tomasi, T. D. (2000). The Michigan Recreational
Angling Demand Model. Paper presented at the Microbehavior and Macroresults
IIFET 2000, Corvallis, Oregon.

Lyons, E., & Breakwell, G. M. (1994). Factors Predicting Environmental Concern and
Indifference in 13 to 16 years old. Environmental Behavior, 26, 223-238.

McFadden, D. (1979). Quantitative Methods for Analyzing Travel Behaviour of
Individuals: Some Recent Developments. In D. A. Hensher & P. R. Stopher
(Eds), Behavioural Travel Modelling (pp. 279-318). Croom Helm, London.

MacGregor, R. D. (1988). The Value of Lost Boater Use and the Cost of Dredging: Two
Aspects of Sedimentation in Ohio 's State Park Lakes. Unpublished PhD
dissertation, Ohio State University, Columbus.

Mahoney, E., Chang, T.C., & Palmer, R. (2002). Economic Impact of Lake Michigan
Levels on Recreation Boating in Michigan. Livonia: Michigan Boating Industry
Association.

Mahoney, E., Chang, T. C., & Palmer, R. (2003). Economic Impact of Recreation
Boating in Michigan: A Report for Michigan Boating Industry Association.
Livonia: Michigan Boating Industry Association.

214

 

Mahoney, E., Stynes, D., Chang, T. C., & McCelleis, T. (2003). The Economic
Importance of Michigan 's Recreational Boating Industry: Co-Chairs of Michigan
Legislative Boating Caucus.

Michigan Secretary of State (2003). Watercraft/Outboard Motor Titling & Registration.
from Michigan Secretary of Sate, Retrieved January 12, 2003, on line access:
http://www.michigan.gov/documents/wr1 lc_wc_title_app_25 93 1_7.pdf

Morey, E. R., Shaw, W. D., & Rowe, R. D. (1991). A Discrete-Choice Model of
Recreational Participation, Site Choice and Activity Valuation when Complete

Trip Data Are Not Available. Journal of Environmental Economics and
Management, 20, 181-201.

Nagelkerke, N.J.D. (1991). A Note on a General Deﬁnition of the Coefﬁcient of
Determination. Biometrika, 78(3), 691-692.

National Marine Manufacturer Association (2001). 2000 US. Recreational Boat
Registration Statistics. Chicago, Illinois: Market Statistics Department.

NTH/W TA Joint Venture (2000). Assess Cost of Higher and Lower Design Crest
Elevations on Future Shore Protection. Detroit: US. Army Corps of Engineers,
Detroit District.

Oakwood, J. E. (1968). Planning for Recreational Boating in Michigan: An Analysis of
the Waterways Division '3 Planning Program. Unpublished Master Thesis,
Michigan State University, East Lansing.

Peterson, G. L., Dwyer, J. F., & Darragh, A. J. (1983). A Behavioral Urban Recreation
Site Choice Model. Leisure Sciences, 6(1), 61-81.

Flaming & Zonning Center, Inc., & Wade-Trim. (2002). Assessment of Alternative Land
Use Management Practices. Detroit: US. Army Corps of Engineers, Detroit
District.

Flaming & Zoning Center, Inc. et al. (2001). Lake Michigan Potential Damages Study-
Economic Impact of Lake Michigan Levels on Recreational Boating and Charter
Fishing in Five Counties. Detroit: US. Army Corps of Engineers, Detroit District.

Planning & Zoning Center, Inc. et. al. (2002). Lake Michigan Potential Damages Study-
Economic Impact of Lake Michigan Levels on Recreational Boating. Detroit: US.
Army Corps of Engineers, Detroit District.

215

Powers, D. A., & Xie, Y. (2000). Statistical Methods for Categorical Data Analysis.
London: Academic Press.

Samdahl, D.M., & Robertson, R. (1989). Social Determinants of Environmental
Concerns: speciﬁcation and test of the model. Environmental Behavior, 21, 57-81.

Siderelis, C., Brothers, G., & Rea, P. (1995). A Boating Choice Model for the Valuation
of Lake Access. Journal of leisure research, 27(3), 264-282.

Siderelis, C., & Moore, R. (1998). Recreation demand and the inﬂuence of site
preference variables. Journal of leisure research, 30(3), 301-318.

Sellinger, C. (2002). Great Lakes Water Levels. Muskegon: The Great Lakes
Environmental Research Laboratory.

Smith, V.K. & Munley, V.G. (1978). The Relative Performance of Various Estimations
of Recreation Participation Functions. Journal of Leisure Research, 10, 165-176.

Sommer, R. (1972). Design Awareness. San Francisco: Rinehard Press.

Soutukorva, A. (2000). The Value of Improved Water Quality- A Random Utility Model
Recreation in the Stockholm Archipelago. The Beijer International Institute of
Ecological Economics. Retrieved, 2002, from the World Wide Web:
http://www.bei j er.kva.se/publications/pdf-archive/artdisc 1 3 5 .pdf

Stewart, C. J. (2000). Lake Michigan Potential Damages Study: Inventory and Mapping
of Shoreline Protection and Boating Structure in Drowned River Mouth Areas of
Lake Michigan. Detroit: US. Army Corps of Engineers, Detroit District.

Stynes, D.J., & Safronoff, D. (1980). 1980 Michigan Recreational Boating Survey. East
Lansing: Michigan Sea Grant.

Stynes, D. J., & Peterson, G. L. (1984). A Review of LOGIT Models with Implications
for Modeling Recreation Choices. Journal of leisure research, 16, 295-310.

Stynes, D.J., Wu, T.C., & Mahoney, E. (1995). 1994 Michigan Boating Survey. East
Lansing: Michigan State University.

Talhelm, D., Jordan, 8., & Holecek, D. (1986). 1986 Michigan Recreational Boating
Survey. East Lansing: Michigan State University.

216

Talhelm, D., Mahoney, E., Bishop, G., Lee, H.C., & Wu, T.C. (1998). Great Lakes
Marina Census and Marina Needs. East Lansing: Michigan Agricultural
Experiment Station, Michigan State University.

Tarrant, M. A., & Green, G. T. (1999). Outdoor Recreation and the Predictive Validity of
Environmental Attitudes. Leisure Sciences, 21, l7-30.

TEACH Great Lakes (2002). Water Levels on the Great Lakes. Great Lakes Information
Network. Retrieved, 2002, on line access: http://wwwgreat-
lakes.net/teach/envt/levels/lev_1 .htrnl

The Great Lakes Information Network (2002). The Great Lakes. Retrieved November 12,
2002, from The Great Lakes Information Network on line access:
http://www.great-lakes.net/lakes/#overview

Tolley, M. (2001). Environmental Psychology. University College Northampton.
Retrieved 22 July, 2002, on line access:
http://nli .northampton.ac.uk/ass/psych/psy1 OO l/Environmental/Index.htm

US. Army Corps of Engineers, Detroit District (2002). Great Lakes Update-Are Great
Lakes Water Levels Recovering? US. Army Corps of Engineers, Detroit District.
Retrieved September 20, 2002, on line access:
http://huron.lre.usace.army.mil/levels/updatel48.pdf

US. Army Corps of Engineers, Detroit District (2002). Monthly Bulletin of Lake Levels
for the Great Lakes. US. Army Corps of Engineers, Detroit District. Retrieved
September 20, 2002, on line access:
http://huron.lre.usace.army.mil/levels/cbulletin.pdf

Van Liere, K. D., & Dunlap, R. E. (1980). The Social Bases of Environmental Concern: a
Review of Hypotheses, Explanations and Empirical Evidence. Public Opinion
Quarterl, 44, 181-197.

Veitch, R. & Arkkelin, D. (1995). Environmental Psychology: An Interdisciplinary
Perspective. Englewood Cliffs: Prentice-Hall, Inc.

Wade, C., & Hard, D. (2001). Land Use/Cover Chang Analysis for the Lake Michigan
Potential Damages Study, Wisconsin Coastal Zone. University of Wisconsin Sea
Grant Institute and Land Information Computer Graphics Facility of University of
Wisconsin-Madison.

217

 

Warner, TD. (1974). An Analysis of Recreational Boating Expenditures (A study of Lake
Michigan Boaters). Unpublished master’s thesis. Michigan State University, East
Lansing.

Weigel, R. H. (1983). Environmental Attitudes and the Prediction of Behavior. In N. R.
F eimer & E. S. Geller (Eds), Environmental psychology: Directions and
Perspectives (pp. 257-287). New York: Preager.

Wisconsin Department of Natural Resources (2003). Boat Registration and Title
Instructions. Wisconsin Department of Natural Resources, Retrieved January 12,

 

2003, on line access: .-
http://www.dnr.state.wi.us/org/caer/cs/servicecenter/forms/9400193.pdf

Wooldridge, J. M. (2000). Introductory Econometrics: A Modern Approach, South-
Westem College Publishing.

Wrigley, N. (1985). Categorical Data Analysis for Geographers and Environmental I

Scientists. New York: Longman Inc.

Wu, T. C. (1995). A System of Models for Estimating Recreational Boating Use in
Michigan Counties. Unpublished Ph.D. Dissertation, Michigan State University,
East Lansing.

218