I P.
13w. ii:

 

{Lama
. a h
. . :. ti. 5 .V
5.. t. .I..l¢.
41rd.

5.3...
3.140,,95

.2: it s» .5! .
{3...}. .A... . .I
y . . .. ..

s an. s
"aﬁrétﬁ.
17.1.23.—
1:2...

.4
in
13b“

r .I .. n
13).}...32

.Itihn
:9

\ I." .3. . :p.
25.3. a. .i .3 .
guuﬂ‘pd.“nnh«ﬂ $5...:.._.h.Jn.rl.

.15....{3
>3. ...A:x.‘a .

5.2::

. .....u_....... 3.2.1.

5.: a. .2 {31...
Fiji.
22...? x.

(2
. {1:31.
In): . . .H85’3L3310a ..

it. If S
nlv ‘

1|\\.\
a». v2.2..-
lit-.1
.. fill) I!

i...
.onxl‘..|2l:!§'$ If}. 36!
. vlLA:ti.( its.
x..|)lh..:..l $\

 

 

. -. ... Luzﬂar, .HA. ‘v.
, ,, , gram"?
, ., ,, ‘ , km? E; , .memmx. ‘ . ‘ . . ‘ ‘
n t 1 n ‘ A ‘ A . . . , :5 . .
g.....“._.sau.,,”..z.§=.3._2E3 214?. V.

. .. ; :1: v...:

 

 

 

Illltﬂllltliﬂﬂilllciizwiiﬂiiilﬂi|l

LIBRARY

Michigan State
University

.3500

 

 

 

 

This is to certify that the

dissertation entitled
Factors Inﬂuencing Snowmobilers' Aberrant Behaviors:
A Comparison of 1998-97 Convicted Snowmobile Law Violators

and Other Snowmobilers
presented by

Joel Anthony Lynch

has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in Paris. Beareation and
Tourism Resources

WM

Major professor

77/3/00

MS U is an Afﬁrmative Action/Equal Opportunity Institution 0- 12771

 

PLACE IN REI'URN 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

 

p3 Wﬂéﬂ

@2

 

 

JAN 0 ‘7 261%

 

 

 

 

 

 

 

 

 

 

 

11/00 M.“

 

FACTORS INFLUENCING SNOWMOBILERS’ ABERRANT BEHAVIORS:
A COMPARISON OF 1996-97 CONVICTED SNOWMOBILE LAW VIOLATORS
AND OTHER SNOWMOBILERS

By

Joel Anthony Lynch

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

2000

ABSTRACT

FACTORS INFLUENCING SNOWMOBILERS’ ABERRANT BEHAVIORS:
A COMPARISON OF 1996-97 CONVICTED SNOWMOBILE LAW VIOLATORS
AND OTHER SNOWMOBILERS

By

Joel Anthony Lynch

In the 1990’s, increases in snowmobiling activity and fatalities over the previous
decade heightened snowmobiler safety concerns. Attempts to curb accidents through law
enforcement had been only moderately successful, due to limited personnel and budgets
coupled with snowmobiler apathy toward regulations. A lack of knowledge about
underlying factors contributing to risky or illegal snowmobiling behaviors was a further
complication to snowmobile managers and law enforcement personnel. Most previous
snowmobile safety research was restricted to accident circumstances or comparisons
between accident victims and non-victims.

The goal of this study was to identify salient variables that discriminate between
convicted snowmobile safety law violators and other snowmobilers. The factors
hypothesized to differentiate these two groups were conceptualized using the PRECEDE
Model, which postulates that behavior is inﬂuenced by predisposing, enabling, and
reinforcing factors. Variables for these factors were derived from studies examining
snowmobile and vehicle accidents as well as from folk crime research, which compared

trafﬁc and ﬁsh and game law violators and non-violators.

Mail questionnaire data ﬁom a census of the 983 individuals cited by Michigan
Conservation Ofﬁcers in winter 1996-97 and convicted of a snowmobile safety violation
was contrasted with data ﬁ'om a sample of 3,325 other snowmobilers not cited during the
same winter. A series of logistic regression models was used to identify differentiating
variables the two groups.

Based upon this analysis, the 1996-97 convicted snowmobile law violators were
younger, snowmobiled more and spent more on snowmobiling than other snowmobilers.
They were also more critical of law enforcement and trail management, more likely to
perceive the dangers of snowmobiling as someone else’s failure, and members of their
households were less likely to have taken a snowmobile safety class. In an equal sized
sub sample, the ﬁnal model correctly classiﬁed about two-thirds of 1996-97 convicted
snowmobile law violators and other snowmobilers. Model diagnostics suggest that those
incorrectly classiﬁed as other snowmobilers may have been unwitting violators who had
limited knowledge of their violation and no intention to violate. Conversely, many of
those incorrectly classiﬁed as violators apparently were willing to violate, but were not
apprehended in winter 1996-97 by Conservation Ofﬁcers. Recommended intervention
strategies target intentional, situational, and unwitting violators.

Future research recommendations include comparing violators over a period
longer than one winter with other snowmobilers and examining driving and other
recreation related violations on a subject’s record. Also, clarifying the proportion and
makeup of violator segments based on motivating factors (situational, intentional or
unwitting) would assist managers to better target scarce educational and enforcement

resources.

DEDICATION

T o my Mother, Father, and Wife;
F or all their endearing love and support.

iv

ACKNOWLEDGEMENTS

I would ﬁrst like to thank Herb Burns, Laura Willard, William Leutscher, Susan
Kapello, Dave Purol, and Curt Bacon of the Michigan Department of Natural Resources
(MDNR) Law Enforcement Division and Dan Moore, Margo Fuller, and Hector Chiunti
of MDNR Forest Management Division for their ﬁnancial support and assistance
throughout this project. I would also like to thank the Michigan Agriculture Experiment
Station for their ﬁnancial support.

Secondly, I would like express my profound appreciation to Dr. Charles Nelson
for his belief in me and my abilities throughout my tenure as a graduate student in the
Department of Park, Recreation and Tourism Resources. I am especially thankﬁil for his
candor, his ability to keep things grounded in reality particularly when things became
difﬁcult, and for the innumerable contributions he made throughout my educational
endeavors. I am also grateﬁil for his guidance through this research and all the other
countless research projects I was able to get my hands into during my years at Michigan
State University.

Thirdly, I cannot adequately express my deep gratitude to my wife Kristine for her
all her support in maintaining my sanity while I pursued my degrees, especially my
doctoral. I am also extremely appreciative of her taking time away from her own
enormous workload to provide feedback and editorial assistance on this dissertation.
Additionally, much of the support and encouragement to continue onward in my graduate
studies at Michigan State University came as a direct result of my mother and father. I
cannot truly express my appreciation to them for the impacts they made in my life from
square one. Undoubtedly, without it I would not have achieved this degree. Additionally,
I am appreciative of my brother Jason, older sister Juice, and my twin sister Jill, to whom
I owe equal thanks for their encouragement and support. I would like to include all my
friends for all the good times and the Michigan State University Hockey and basketball
teams for the much needed intermissions ﬁ'om the toils of school, work, and of course
this dissertation!

Finally, a big thanks goes to my committee members, Don Holecek, Gail Vander
Stoep, and Larry Leefers as well as other faculty at Michigan State University whom, in
their own ways, contributed to my success and my education.

TABLE OF CONTENTS

CHAPTER 1: INTRODUCTION ........................................................................................ 1
Evolution of Snowmobiles and Snowmobiling .............................................................. l
Snowmobilcr Safety and Law Enforcement .................................................................... 2
Folk Crimes ..................................................................................................................... 5
State of Knowledge ......................................................................................................... 6
Conceptual Framework ................................................................................................... 9
Problem Statement ........................................................................................................ 11
Study Objectives ........................................................................................................... 12
Key Study Limitations and Delimitations ..................................................................... 12

CHAPTER 2: REVIEW OF LITERATURE ..................................................................... 14
Overview of Criminal and Deviant Behavior Theories and Models ............................. 15
Predisposing Inﬂuences ................................................................................................ 17

Past Behaviors ........................................................................................................ l7
Perceptions of Risk ................................................................................................ 18
Attitudes and the Theory of Planned Behavior ...................................................... l9
Personality .............................................................................................................. 21
Emotions ................................................................................................................ 23
Demographic Factors ............................................................................................. 23
Enabling Inﬂuences ....................................................................................................... 24
Exposure and Vehicle Characteristics .................................................................... 24
Skill Level .............................................................................................................. 25
Reinforcing Inﬂuences .................................................................................................. 26

CHAPTER 3: RELEVANT FACTORS AND HYPOTHESES ........................................ 29

Predisposing Factors ..................................................................................................... 29
Age ......................................................................................................................... 29
Perceptions of the Dangers of Risky Behaviors and Situations ............................. 30
Attitudes Towards Management and Law Enforcement ........................................ 30

Enabling Factors ............................................................................................................ 31
Exposure and Active Participation ......................................................................... 31
Performance Capabilities of Snowmobiles ............................................................ 32

Reinforcing Factors ....................................................................................................... 32
Family Based Activity ........................................................................................... 32
Investment in Snowmobiling ................................................................................. 33
Lack of Concern for Safety Education ................................................................... 33

Summary of Hypothesized Distinguishing Variables .................................................. 33

CHAPTER 4: METHODS AND PROCEDURES ............................................................ 36

Subjects and Sampling Protocol ................................................................................... 37
1996-97 Convicted Snowmobile Law Violators .................................................... 37
Other Snowmobilers .............................................................................................. 38

Data Collection Method ................................................................................................ 40
Technique ............................................................................................................... 4O

Questionnaire Design and Administration .................................................................... 41
Questionnaire Design ............................................................................................ 41
Questionnaire Administration ................................................................................ 43

Questionnaire Response Rate ........................................................................................ 43
1996-97 Convicted Snowmobile Law Violator’s Response Rate .......................... 43
Other Snowmobiler’s Response Rate .................................................................... 44
Survey of Non-Respondents .................................................................................. 44
Administration of Non-Respondent Survey ........................................................... 45

Data Analysis .................................... 46
Descriptive Analysis .............................................................................................. 46
Logistic Regression Analysis ................................................................................. 47

Modeling Strategy ......................................................................................................... 49
Characterizing 1996-97 Convicted Snowmobile Law Violators and Other
Snowmobilers ........................................................................................................ 50

Logistic Regression Diagnostics ................................................................................... 53
Nonlinearity in the Logit ........................................................................................ 53
Collinearity ............................................................................................................ 54
Evaluations of Residuals ........................................................................................ 54

Model Discrimination ................................................................................................... 55

CHAPTER 5: DESCRIPTIVE RESULTS ........................................................................ 57

Response and Non-Response Comparison .................................................................... 57
1996-97 Convicted Snowmobiler Law Violator Non-Response ........................... 57
Citation Data Summary and Response and Non-Response Comparison ............... 62
Other Snowmobiler Non-Response ....................................................................... 69

1996-97 Convicted Snowmobile Law Violator and Other Snowmobiler Comparison 72
Household and Respondent Characteristics ........................................................... 72
State and Region of Principal Residence ............................................................... 74
Household Snowmobiles and Use ......................................................................... 75
Rating of Dangerous Snowmobiling Behaviors and Situations ............................. 79
Rating of Fatality Reduction Initiatives ................................................................. 79
Household Snowmobiling Expenditures and Safety Education ............................ 82
Perceptions of Snowmobile Accident Factors ....................................................... 83

Assessment of Independence Between Independent Variables .................................... 84

Descriptive Summary .................................................................................................... 88

vii

CHAPTER 6: LOGISTIC REGRESSION ANALYSIS RESULTS ................................. 93

Phase 1: Univariate Analysis ........................................................................................ 93
Phase 2: Initial Logistic Regression Analysis ............................................................... 95
Backwards Variable Selection Model .................................................................... 95
Single Block Entry Model ................................................................................... 100
Initial Logistic Regression Model Diagnostics .................................................... 106
Initial Logistic Regression Analysis Summary .................................................... 115
Phase 3: Logistic Regression Analysis: Model Discrimination .................................. 118
Model Discrimination .......................................................................................... l 19
Model Discrimination Summary .......................................................................... 121
CHAPTER 7: SUMMARY, IMPLICATIONS, AND CONCLUSION .......................... 123
Study Background ....................................................................................................... 123
Summary of Univariate Analysis Findings ................................................................. 124
Predisposing Factors ............................................................................................ 124
Enabling Factors .................................................................................................. 125
Reinforcing Factors .............................................................................................. 126
Summary of Logistic Regression Analysis Findings .................................................. 127
Study Implications and Recommendations ................................................................. 129
Dissemination of Information ............................................................................. 132
Education ............................................................................................................. 133
Law Enforcement ................................................................................................. 134
Legislation and Regulations ................................................................................. 137
Management ......................................................................................................... 139
Research Recommendations ....................................................................................... 141
Delimiting Violators and Non-Violators and Characterizing Different
Violators ............................................................................................................... 142
Additional Information ........................................................................................ 145
Conclusion .................................................................................................................. 146
APPENDICES ................................................................................................................. 148
Appendix A: Michigan Snowmobile Questionnaire ................................................... 149
Appendix B: Other Snowmobiler Cover Letter .......................................................... 154
Appendix C: 1996-97 Convicted Snowmobiler Law Violator Cover Letter .............. 156
Appendix D: Other Snowmobiler Follow-up Cover Letter ........................................ 158
Appendix E: 1996-97 Convicted Snowmobiler Law Violator Follow-up Cover
Letter ........................................................................................................................... 160
Appendix F: Michigan Snowmobiling Non-Respondent Telephone Survey ............. 162

viii

REFERENCES ................................................................................................................ 165

GOVERNMENT REFERENCES ................................................................................... 175

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

LIST OF TABLES

Scale and data type of potential predisposing variables differentiating
1996-97 convicted snowmobile law violators and other snowmobilers. .......... 34

Scale and data type of potential enabling and reinforcing variables
differentiating 1996-97 convicted snowmobile law violators and other
snowmobilers. ................................................................................................... 25

Household size and snowmobile activity, age, and residency of 1996-97
convicted snowmobile law violator respondents and non-respondents who
snowmobiled in Michigan during that winter. ........................................ 58

Selected characteristics of 1996-97 convicted snowmobile law violator
respondents and non-respondents household snowmobiles used in Michigan
during that winter. ............................................................................................ 59

Regional household snowmobiling activity of 1996-97 convicted
snowmobile law violator respondents and non-respondents in Michigan
during that winter. ............................................................................................. 60

1996-97 convicted snowmobiler law violator respondents and
non-respondents rating of support for selected fatality reduction initiatives. ...61

Age, sex, and state and region of principal residence for all 1996-97 convicted
snowmobile law violators and respondents cited by Michigan Conservation
Ofﬁcers during that winter. ............................................................................... 63

Month, day of week, time of day, and region citations were issued to all
1996-97 convicted snowmobile law violators and respondents cited by
Michigan Conservation Ofﬁcers during that winter. ........................................ 65

Number and type of moving violations issued to 1996-97 convicted
snowmobile law violators and respondents cited by Michigan Conservation
Ofﬁcers during that winter. ............................................................................... 67

Table 10.

Table 11.

Table 12.

Table 13.

Table 14.

Table 15.

Table 16.

Table 17.

Table 18.

Table 19.

Number and type warnings above and beyond the citation issued to all 1996-97
convicted snowmobile law violators and respondents cited by Michigan
Conservation Ofﬁcers during that winter .......................................................... 68

Household size and snowmobile activity, age, and residency of other
snowmobiler respondents and non-respondents who snowmobiled in
Michigan during winter 1996-97. ..................................................................... 70

Selected characteristics of other snowmobiler respondents and
non-respondents household snowmobiles used in Michigan during
winter 1996-97. ............................................................................................ 71

Regional household snowmobiling activity of other snowmobiler
respondents and non-respondents in Michigan during 1996-97 winter. ........... 71

Other snowmobiler respondents and non-respondents rating of support for
selected fatality reduction initiatives. ............................................................... 72

Household size, snowmobiling activity, and respondent age for 1996-97
convicted snowmobile law violators and other snowmobilers who
snowmobiled in Michigan during winter 1996-97 ............................................ 73

State and region of principal residence for 1996-97 convicted snowmobile
law violators and other snowmobilers who snowmobiled in Michigan
during winter 1996-97 ....................................................................................... 74

Snowmobile ownership and use in Michigan by 1996-97 convicted
snowmobile law violators and other snowmobilers during winter 1996-97. ....75

Selected characteristics of 1996-97 convicted snowmobile law violators and
other snowmobilers who snowmobiled in Michigan during winter 1996-97. ..76

Average estimated household miles and days of snowmobiling for 1996-97
convicted snowmobile law violators and other snowmobilers in Michigan
during winter 1996-97 ....................................................................................... 77

xi

Table 20.

Table 21.

Table 22.

Table 23.

Table 24.

Table 25.

Table 26.

Table 27.

Table 28.

Regional household snowmobiling activity for 1996-97 convicted snowmobile
law violators and other snowmobilers in Michigan during winter 1996-97. ....78

Rating of danger ﬁom selected snowmobiling behaviors and situations by
1996-97 convicted snowmobile law violators and other snowmobilers in
Michigan during winter 1996-97. ..................................................................... 80

Rating of support for selected fatality reduction initiatives by 1996-97
convicted snowmobile law violators and other snowmobilers in Michigan
dining winter 1996-97 ....................................................................................... 81

Household expenditures for snowmobiling and participation in snowmobile
safety training for 1996-97 convicted snowmobile law violators and other
snowmobilers in Michigan during winter 1996-97 ........................................... 82

Assessment of selected factors contributing to a personal injury or fatal
snowmobiling accident with which 1996-97 convicted snowmobile law

violators or other snowmobilers were familiar in Michigan during winter
1996-97. ............................................................................................................ 83

Bivariate correlation between independent variables measuring support for
selected fatality reduction initiatives for 1996-97 convicted snowmobile law
violators and other snowmobilers combined. ................................................... 85

Bivariate correlation between independent variables measuring danger from
selected snowmobiling behaviors and situations initiatives for 1996-97
convicted snowmobile law violators and other snowmobilers combined ......... 86

Bivariate correlation between independent variables measuring danger from
selected snowmobiling behaviors and situations and support for selected
fatality reduction initiatives for 1996-97 convicted snowmobile law violators
and other snowmobilers combined. .................................................................. 87

Bivariate correlation between independent variables measuring household
snowmobiling activity, safety class, respondent age, snowmobile ownership
and selected characteristics, and spending for 1996-97 convicted snowmobile
law violators and other snowmobilers combined .............................................. 89

xii

Table 29.

Table 30.

Table 31.

Table 32.

Table 33.

Table 34.

Table 35.

Table 36.

Table 37.

Predisposing variables and their statistical signiﬁcance between 1996-97
convicted snowmobile law violators and other snowmobilers. ........................ 90

Enabling and reinforcing variables and their statistical signiﬁcance between
1996-97 convicted snowmobile law violators and other snowmobilers. .......... 91

Likelihood ratio and likelihood ratio chi-square tests between the dependent
and predisposing independent variables. .......................................................... 94

Likelihood ratio tests between the dependent and enabling and reinforcing
independent variables ........................................................................................ 95

Logistic regression statistics for the predisposing, enabling, and reinforcing
variables selected in the backwards stepwise logistic regression analysis
between 1996-97 convicted snowmobile law violators and other
snowmobilers. ................................................................................................... 97

Logistic regression statistics for predisposing, enabling, and reinforcing
variables using single block entry in logistic regression analysis between
1996-97 convicted snowmobile law violators and other snowmobilers. ........ 101

Logistic regression model statistics with, without, and combined
non-signiﬁcant variables identiﬁed in the single block entry in logistic
regression analysis between 1996-97 convicted snowmobile law violators

and other snowmobilers. ................................................................................. 102

Logistic regression statistics for predisposing, enabling, and reinforcing
variables with the average engine size of household snowmobiles and danger
associated with the speed of snowmobiles removed using single block entry in
logistic regression analysis between 1996-97 convicted snowmobile law
violators and other snowmobilers. .................................................................. 105

Nonlinearity in the logit results using the Box and Tidwell transformation of
the relevant predisposing, enabling, and reinforcing variables distinguishing
1996-97 convicted snowmobile law violators and other snowmobilers. ........ 106

xiii

Table 38.

Table 39.

Table 40.

Table 41.

Table 42.

Table 43.

Table 44.

Table 45.

Nonlinearity in the logit results using the Box and Tidwell transformation of
relevant and reconﬁgured predisposing, enabling, and reinforcing variables
distinguishing 1996-97 convicted snowmobile law violators and other
snowmobilers. ................................................................................................. 107

Logistic regression statistics for ﬁnal model with the reconﬁgured
predisposing, enabling, and reinforcing variables using single block entry in
logistic regression analysis between 1996-97 convicted snowmobile law
violators and other snowmobilers. .................................................................. 108

Colinearity results for the reconﬁgured predisposing, enabling, and reinforcing
variables distinguishing 1996-97 convicted snowmobile law violators and
other snowmobilers. ........................................................................................ 109

Mean age and proportion of the household completing a snowmobile safety
class between cases with Studentized residuals values greater than 2.00,
1996-97 convicted snowmobile law violators, and other snowmobilers. ....... 110

Percentage of support for more intensive enforcement of existing snowmobile
regulations and rating of the danger associated with the public trail conditions
between cases with Studentized residuals values greater than 2.00, 1996-97
convicted snowmobile law violators, and other snowmobilers. ..................... l 11

Percentage of grouped average number of estimated snowmobiling days per
household snowmobile and household expenditures for snowmobiling
equipment, maintenance, insurance, and storage between cases with
Studentized residuals values greater than 2.00, 1996-97 convicted

snowmobile law violators, and other snowmobilers. ..................................... 112

Means for the general predisposing, enabling, and reinforcing variables
between cases with leverage values greater than 0.012, 1996-97 convicted
snowmobile law violators, and other snowmobilers. ..................................... 113

Percentage of support for more intensive enforcement of existing snowmobile
regulations and rating of the danger associated with the public trail conditions
between cases with leverage values greater than 0.012, 1996-97 convicted
snowmobile law violators, and other snowmobilers. ...................................... 114

xiv

Table 46.

Table 47.

Table 48.

Percentage of grouped average number of estimated snowmobiling days per
household snowmobile and household expenditures for snowmobiling
equipment, maintenance, insurance, and storage between cases with leverage
values greater than 0.012, 1996-97 convicted snowmobile law violators, and
other snowmobilers. ........................................................................................ l 15

Classiﬁcation table for the ﬁnal independent variables logistic regression
model using single block entry between 1996-97 convicted snowmobile law
violators and other snowmobilers. .................................................................. 119

Distribution of predicted probabilities for the ﬁnal independent variables
logistic regression model using single block entry between 1996-97 convicted
snowmobile law violators and other snowmobilers. ....................................... 121

XV

LIST OF FIGURES

Figure 1. Three Michigan Snowmobile Regions ............................................................ 60
Figure 2. Five Michigan Snowmobile Regions .............................................................. 64
Figure 3. Histogram of Observed Group and Predicted Probabilities .......................... 120

xvi

CHAPTER ONE
INTRODUCTION

Evolution of Snowmobiles and Snowmobiling
Described as a light-weight sit-upon machine capable of traveling over snow and

ice, the 'motor toboggan' developed by Carl Eliason in 1925 is considered the predecessor
of the modern snowmobile (Horney, 1970). Yet, it was not until the reﬁnement of the
two-stroke engine in the late 1950’s that the snowmobile became economically feasible to
mass produce (Sheridan, 1979). The development of the snowmobile, along with
increasing participation in outdoor recreation during the 1960’s, dramatically transformed
winter recreation patterns. Now, in the late 1990’s, snowmobiling represents a major
wintertime recreational pursuit and a thriving industry (Zesiger, 1997).

Nowhere else has this been more evident than in Michigan. In a 1997 study of
Michigan snowmobiling, it was estimated that during the 1996-97 winter season,
snowmobilers rode over 170 million miles during 2.3 million days of snowmobiling
(Nelson et al., 1998). The equipment purchased and travel spending to support this use
generated an estimated $321 million for Michigan’s economy (Stynes et al., 1998).

Snowmobiling has undergone many changes since its emergence in the early
1960’s. The ﬁrst commercially produced snowmobiles were difﬁcult to handle due to
inadequate suspension systems and a high center of gravity. These early snowmobiles
likewise suffered from other deﬁciencies, including poor traction on ice, insufﬁcient
lights, and poor brakes (Waller & Lambom, 197 5). Trails for snowmobiling were also
initially inadequate, forcing snowmobilers onto open ﬁelds, unplowed roads, and frozen
waterways. Consequently, most snowmobiling occurred on private lands relatively close
to home, with participants averaging four hours of riding per snowmobiling day (Bulter,
1974)

Since then, however, snowmobile innovations and enhanced opportunities have
dramatically transformed snowmobiling. For instance, snowmobiles were revolutionized
in the early 1990’s by the development of suspension systems (Klim, 1996). This not
only improved handling, but also the stability and comfort of snowmobiles. Current
snowmobiles are also much more powerful, efﬁcient, and considerably more durable than
their predecessors. Such improvements have meant quicker acceleration, greater speeds,
and longer driving ranges.

Opportunities to ride also have been greatly enhanced. An estimated 200,000
miles of marked and groomed trails in the United States are open to snowmobiling
enthusiasts. This system is comprised of a network of interconnected trails and links to
communities providing lodging, food, fuel, and other amenities. Improvements related to
trail systems and the comfort of snowmobiles have facilitated snowmobilers taking more
extended trips involving overnight stays than ever before. These changes also have
helped recruit new participants. In the 1990’s, a larger number of women and children
became snowmobiling enthusiasts (Hermance, 1995). These newcomers have meant a '
growing snowmobile market. The International Snowmobiling Manufacturers
Association has reported a resurgence in snowmobile sales since 1990 (Klim, 1997).
However, even since its beginning, snowmobiling has not come without management

challenges, especially as it relates to the safety of snowmobilers.

Snowmobiler Safeg and Law Enforcement
As the popularity of snowmobiling grew in the 1960’s, so too did concerns about

the safety of the sport. By the late 1960’s, many states reported alarming incidents of
non-fatal and fatal snowmobile accidents (Dewar, 197 3). Early examination of accident
reports suggested that the causes of these accidents were primarily unsafe behaviors such
as alcohol use, night-time operation, and excessive speeds as well as inadequacies related

to snowmobile design and riding locations (Withington & Hall, 1970; Wenzel et al.,

1973). Studies in the 1980’s and 1990’s examining the relationship between accidents,
rider behaviors, and the shortcomings of snowmobiles had similar conclusions (Wenzel
& Peters, 1986; James et al., 1991; Rowe et al., 1992; Avis, 1994; Rowe et al., 1994).

To reduce the incidence of accidents, states initially instituted snowmobile
operation and safety regulations as part of their comprehensive snowmobile legislation.
By 1971, nearly all states with snowmobiling activity had enacted legislation outlining
such regulations (U SDI, 1971). The intent of these regulations was to deﬁne a standard of
operation to safeguard snowmobilers, bystanders, and the environment. These regulations
can be segmented into three categories: registration, operation, and safety. Registration
regulations cover machine licensing and use permit policies. Operation regulations guide
the proper use of snowmobiles. These regulations under Michigan laws include setting of
a minimum age for operating snowmobiles, prohibiting use on highways, operating while
intoxicated, and disturbing the peace and wildlife. Safety regulations set minimum safety
standards for snowmobiles, such as the presence of lights, as well as helmet requirements.
Typically, a violation of these regulations results in a ﬁne, while more serious offenses
such as operation under the inﬂuence of alcohol could mean an arrest.

Enforcement of snowmobile regulations and laws, especially on public lands, is
primarily the responsibility of a state’s natural resource agency or some equivalent.
County sheriff and municipal police departments, and, in some cases state police
agencies, also have the authority to enforce snowmobile laws and often provide a
presence within communities and along roadways. Accordingly, the enforcement of
snowmobile laws on Michigan public lands and waterways had been delegated to the
State’s Depm'tment of Natural Resources (DNR) Law Enforcement Division (LED).
Michigan also has established a snowmobile enforcement grant program that in FY 1997-
98 provided $205,000 in matching monies to participating county sheriff departments to
assist their enforcement efforts. In addition, Michigan Public Act 297 of 1998 extended

grant eligibility to city, village, and township police departments for the winter of 1999-

2000. It also increased Michigan snowmobile registration fees to provide almost $1
million annually to counties and other local units of government.

While snowmobile regulations and their enforcement have done much to improve
the safety of snowmobiling, since the mid 1990’s concerns about the safety of
snowmobiling have once again intensiﬁed. This increase is the result of the higher
numbers of fatal and non-fatal accidents and the rising rate of participation in
snowmobiling (IL-DC, 1996; CDC 1997; MN-DNR, 1997; WI-DNR, 1997). The
magnitude of the predicament is particularly signiﬁcant, since the number of
snowmobiles registered in the 1990’s is considerably lower than the all time highs of the
late 1970’s and early 1980’s, while accident numbers are higher.

Michigan, as have other states, has experienced an increase in the number of fatal
snowmobile accidents. During the winter of 1995-96 more than 40 snowmobiling related
fatalities were recorded, doubling the totals from the winters 1993-94 and 1994-95
(Nelson et al., 1998). While the safety of snowmobilers has always been a priority of
DNR-LED, the increase in fatal snowmobile accidents magniﬁed this concern.

As a result, the DNR-LED began an intensive patrol initiative during winter 1996-
97 to assess increased ofﬁcer presence and target enforcement to reducing snowmobile
accidents. A heavily used three-county area (Kalkaska, Wexford, and Grand Traverse
Counties) in the northwestern lower peninsula was chosen for this initiative. During
weekends, two units of ﬁve to seven ofﬁcers patrolled trails and roadways targeting
unsafe behaviors. While no empirical evidence was gathered on the effectiveness of this
initiative, the LED concluded that such efforts were effective in reducing accidents in this
area (Willard, 1998). Nevertheless, statewide fatal and non-fatal accidents remained

steady that year.

While attempts to mitigate the unsafe behaviors associated with snowmobile
accidents by instituting and enforcing operation and safety regulations may be effective,
it has its limitations. For instance, many regulations have not been amended since the
early 1970’s and may not reﬂect snowmobiles and snowmobilers in the let century. In
addition, the effectiveness of law enforcement is hampered by the expense of more
personnel, over-time pay, and the immensity of the task. A vast system of intertwined
trails and frozen waterways is difﬁcult to patrol with few ofﬁcers. Lastly, enforcement of
any law is contingent upon the individuals being regulated and their disposition toward

the law in general, as well as the behaviors the law is intended to regulate.

Folk Crimes

The term ‘folk crimes’ was ﬁrst used by Ross (1961) to describe violations, such
as trafﬁc oﬁ'enses, that, while legally punishable, are not perceived by society as deviant
behavior. A broader concept, ‘mundane crime,’ is used by Gibbions (1983) to depict a
wider array of activities that include petty vice and public disorder offenses, work place
crimes, environmental abuse offenses, and folk crimes.

The characterization of certain violations as less than deviant emerges from
fundamental differences between the nature of these acts and what many consider more
serious forms of deviant behavior. For instance, the sheer volume of trafﬁc violations
committed each year is overwhelmingly larger than other offenses such as theﬁ (Ross,
1961). As a result, folk crime violations are largely viewed by society as part of some
norm, especially since the typical penalty is a minor ﬁne. F urthennore, a folk crime
violator is not stigmatized by society since the violation is considered a relatively

innocuous technical aberration that lacks immoral intent.

A wide range of motives and situational inﬂuences, such as knowledge about the
law, are also thought to be at the root of such acts, further compromising compliance and
effective enforcement (Gibbions, 1983). This notion is central to Gramann and Vander
Stoep’s (1987) segmentation of violators’ rationale for depreciative behaviors in outdoor
recreation settings and Goldstein’s (1996) illustration of the motivations for vandalism.
The persistence of these folk crime activities also is tied to the interaction between law
enforcement personnel and violators. Ofﬁcers typically have considerable discretion in
determining whether to cite an individual for violating a law. For instance, police ofﬁcers
use ‘personal tolerances’ to deﬁne what constitutes a violation of the law (Rothe, 1994).
Thus one ofﬁcer may let a Speeder go while another enforces the law. Creating such

inconsistencies reduces the internal integrity of the law (Ross, 1983).

State of Knowledge
Curtailing snowmobile accidents is a formidable task given the magnitude of the

enforcement situation and the perspective some individuals take toward risky
snowmobiling behaviors and the regulations governing them. However, this task is
particularly hampered by a lack of knowledge about the relationship between risky
snowmobiling behaviors and fundamental inﬂuences or circumstances driving such
behaviors.

As of the late 1990’s, research concerning the safety of snowmobilers has been
focused primarily on describing the nature of snowmobile injuries and the circumstances
surrounding the accidents, especially as they relate to the role of alcohol (Rowe et al.,
1993). The limiting factor in these studies is that they utilized retrospective data derived
from accidents reports. While these studies have contributed to our understanding of
snowmobile accidents, few have considered how factors such as a snowmobiler’s
attitudes, perceptions, experience, and other characteristics are related to aberrant

behaviors and the likelihood of one being involved in an accident (Rowe et al., 1993).

One of the ﬁrst studies to do so was conducted by Carlson and Klein (1971) on
snowmobile owners in Michigan. They compared the trafﬁc records of snowmobilers
who had reported a snowmobile accident with those who did not report one. They found
that those with accidents had more trafﬁc violations and automobile crashes than the
other group.

A 1972 comparison of driving records of a sample of snowmobile owners and
non-owners from a single county in Vermont was made by Waller (1975). However, no
difference in the number of trafﬁc violations or accidents was found among the groups.
Yet, differences were detected in a second study the following year (Waller & Lambom,
1975). In this study, they conducted extensive interviews with injured snowmobilers
during the winters of 1971-72 and 1972-73. These snowmobilers were compared with
non-injured snowmobilers interviewed during the same season. In terms of miles driven,
injured snowmobilers had either ridden very few miles or had ridden 1,000 or more miles.
Lack of familiarity with the machine or environment, higher horsepower snowmobiles
driven at greater speeds, previous snowmobile accidents, greater likelihood of alcohol
consumption, and being more prone to ﬁ'ustration were all factors that differentiated
injured and non-inj ured snowmobiles. Moreover, injured snowmobilers also were
younger and lacked snowmobiling experience, skills, and judgment about safety.

More recently, a 1993 mail survey was conducted with a sample of snowmobilers
from Sudbury, Ontario, Canada to develop a proﬁle of snowmobilers at risk for
involvement in a snowmobile accident (Rowe et al., 1993). The results revealed that
snowmobiling greater distances and a greater number of times per year, as well as
frequently exceeding posted speed limits on lakes and groomed trails, put snowmobilers
at greater risk to be involved in an accident. The use of high performance (greater than a

500 cc engine size) machines was not associated with accident involvement.

Such studies have begun to identify factors that inﬂuence aberrant behaviors that
potentially contribute to a snowmobiler’s risk of an accident. Yet much of the knowledge
gained from these studies is based upon snowmobiling patterns and snowmobile
characteristics of the 1970’s. Since the mid 1990’s, snowmobiles and snowmobiling have
changed considerably. Furthermore, these studies have focused on snowmobilers who
were involved in a snowmobiling accident and have overlooked those who violate
snowmobile regulations.

Trafﬁc safety researchers have long been interested in how underlying factors,
especially the role an individual’s psychological disposition, have inﬂuenced risky
driving behavior and the likelihood of involvement in an accident (Elander et al., 1993).
Early research, particularly Tilrnan and Hobbs’ (1949) study of cab drivers, identiﬁed a
general association between social deviance and accident involvement. Similarly, through
their work with adolescents, J essor and J essor (1977) assert that various deviant
behaviors, such as dangerous driving and drug use, resonate from a general deviant
disposition.

Recent studies of those who disobey trafﬁc laws have revealed not only a prior
pattern of law violations, but also a greater likelihood of involvement in trafﬁc accidents
than law abiding drivers (West et al., 1993; Lawton et al., 1997a; Meadows et al., 1998).
Evidence of these and other distinguishing factors associated with aberrant behaviors and
accident involvement have been identiﬁed by researchers through comparisons between
folk crime law violators and non-violators. Key comparisons have also been made in

research involving trafﬁc and ﬁsh and game laws.

Conceptual Framework

Conventional comparative research of violators and non-violators of trafﬁc and
ﬁsh and game laws has often been limited to a small number factors related to an
individual's disposition. For example, studies have been conducted about how attitudes or
personality differ among the respective groups. Yet, researchers have often speculated
that a complex array of interactive factors inﬂuences an individual’s choice to violate a
given law or put them at risk for being involved in an automobile accident. In an attempt
to account for the inﬂuences of a multitude of factors, Gabany and associates (1997) have
turned to the PRECEDE Model to develop a more robust inventory of factors that
inﬂuence a driver’s speeding behavior. This model was developed originally by Green
and colleagues (1980) to evaluate health education programs, but it has been shown to
have other applications, especially in studying factors affecting unhealthy behaviors
(Mullen et al., 1987). Thus, this model is applied to conceptualize and analyze differences
between 1996-97 convicted snowmobile law violators and non-violators. The structure of
the model is seen as a heuristic framework within which inﬂuences of behaviors are
conceptualized as resulting from three components (Mullen et al., 1987, p. 974);

l. Predisposing factors: Preferences and prior motives that people bring to an

experience, including knowledge, attitudes, beliefs, values, and perceptions
that either support or inhibit behavior.

2. Enabling factors: Objective characteristics of an individual, community, and
environment that facilitate action on behavior.

3. Reinforcing factors: Rewards or punishments, including social support that
follow a behavior or are anticipated as a consequence of it.

Potential variables conceptually differentiating snowmobile law violators and
non-violators are derived from research examining snowmobiling accidents as well as
research comparing trafﬁc, ﬁsh, and game law violators and non-violators. The variables
identiﬁed in the trafﬁc safety and ﬁsh/wildlife literature are conceptually generalizable to
studying snowmobile law violators and other snowmobilers due to the similar nature of

these aberrant behaviors.

Differences are segmented using the three components of the PRECEDE Model.
In general, predisposing factors suggest that violators differ from non-violators in past
behaviors, perceptions of risk, attitudes, personality, and emotion factors. Regarding past
behaviors, violators have been shown to have histories of prior trafﬁc violations as well
as accidents (Garretson & Peck, 1982; Cooper, 1997). Moreover, violators have been
shown to have different perceptions of risk. For instance, they believe that they are less
likely to be involved in accidents because they consider themselves to be more competent
than the average driver (DeJoy, 1993; Guppy, 1993). In terms of attitudes, violators oﬁen
have negative attitudes toward law enforcement and laws regulating behaviors (Andersen,
1987; Kanellaidis et al., 1995). Psychometric measurements of violators’ personalities
have found them to be more sensation seeking, impulsive, and hostile (McMillen et al.,
1992). Lastly, more positive feelings towards violating laws resulted in increased
tendency to engage in violations @awton et al., 1997b).

Exposure, deﬁned as the duration of time one engages in an activity, is one
enabling inﬂuence that differentiates trafﬁc law violators from non-violators. Violators
tend to drive more than non-violators (Kanellaidis et al., 1995). While vehicle
characteristics have received limited attention, a weak relationship between engine size
and the propensity for trafﬁc accidents has been demonstrated (Rothengatter, 1988).
Lastly, greater skill has been identiﬁed as a salient factor of hunters apprehended for
violating game laws versus non-violating hunters (Gray & Karninski, 1994).

Reinforcing inﬂuences characterize those factors that provide one with an
incentive or reward for deviant behavior. Peer and family support have been shown to
inﬂuence hunters and ﬁshermen to violate ﬁsh and game laws (Jackson, 1980; Curcione,
1992). Moreover, while the prime motive for poaching is usually to provide food,
research also has shown poaching offered pleasure, achievement, and excitement (F orsyth

et al., 1998).

10

Problem Statement

In the 1990’s, higher numbers of fatal and non-fatal snowmobile accidents,
coupled with increased participation and more trails, have renewed concerns about
snowmobiler safety. As a general rule, snowmobile regulations were instituted to
establish a standard of safe operation and conduct. Attempts to curb snowmobiling
accidents solely through the institution and enforcement of such regulations have been
difﬁcult due to limited law enforcement personnel and budget, as well as antipathy
toward regulation by many snowmobilers.

The situation is further complicated by insufﬁcient knowledge about the
inﬂuences and circumstances contributing to risky or illegal snowmobiling behaviors.
Typically, research conceming the safety of snowmobilers has relied on accident data,
focusing mostly on descriptions of accident circumstances. The few studies that were
designed to distinguish underlying factors that inﬂuence aberrant snowmobile behaviors
are out-of-date and restricted to individuals involved in snowmobiling accidents.

Nevertheless, managers and law enforcement personnel are required to ensure the
safety of snowmobilers and bystanders. Comprehending key factors affecting
snowmobilers’ aberrant behaviors would be useful in improving snowmobiling safety.
For example, law enforcement may be able to more easily identify high risk individuals
or particular characteristics that contribute to unsafe behaviors. Managers may be able to
learn more about the unsafe snowmobilers’ riding habits and institute changes in design,
informational messages, safety education, and policies to improve the situation.
Consequently, this study attempts to identify salient variables that predispose, enable, and
reinforce aberrant snowmobiling behaviors. It then seeks to create a proﬁle of a

snowmobiler who is inclined to violate snowmobile laws.

11

Study Obiectives
Three speciﬁc objectives are deﬁned for this study.
1. Compare characteristics of those individuals cited for snowmobiling safety
violations by Michigan DNR-LED Conservation Ofﬁcers during the winter of
1996-97 and convicted for such offenses, (here-after called 1996-97 convicted

snowmobile law violators) with a sample of Michigan snowmobilers not cited
and convicted that winter (here-after called other snowmobilers).

2. Using those variables identiﬁed as having overt inﬂuences on snowmobilers’
behavior, examine the PRECEDE model’s ability to discriminate between
1996-97 convicted snowmobile law violators and other snowmobilers.

3. Based on the above two objectives, propose management actions to reduce
risky snowmobiling behavior, and by inference, accidents.

Key Study Limitations and Delimitations

Limitations are shortcomings of the study methodology, while delimitations are
researcher imposed boundaries placed upon on the study. In this study, the key limitation
was the sample population of convicted snowmobile law violators. Those convicted
violators included in the study are assumed to be representative of all those convicted of a
Michigan snowmobile law violation in 1996-97. However, this can not be proven. This
group only includes those who were cited by DNR Conservation Ofﬁcers during winter
1996-97 and convicted. County Sheriffs and Michigan State Police also issued an
unknown number of citations to snowmobilers during the winter 1996-97. Statewide data
on their activities were unavailable at the time of the study, although following winter
1999-2000 such records are likely to be available from DNR-FMD. At the time of the
study, such citations were reportedly kept on ﬁle at the individual county level or with the
Michigan State Police. Hence, violators who were cited by either County Sheriffs or
Michigan State Police, but not Conservation Ofﬁcers, were excluded from this study
because it was impractical to acquire these citations.

The key delimitation of this study related to the deﬁnition used to segment 1996-
97 convicted snowmobile law violators and other snowmobilers. In this case, convicted

snowmobile law violators included only those that were cited and convicted during the

12

1996-97 winter season. Inevitably, there were other snowmobilers during winter 1996-97
who violated the law but were not cited and convicted. Further, those cited and convicted
of violations in the pervious winters were not included as violators. Finally, at the request
of the DNR-LED, other snowmobilers or 1996-97 convicted violators were not asked

about past convictions on snowmobile law violations.

13

CHAPTER TWO
REVIEW OF LITERATURE

The purpose of this study is to identify, through a comparison of 1996-97
convicted snowmobile law violators and other snowmobilers, salient variables that
predispose, enable, and reinforce aberrant snowmobiling behaviors and to create a proﬁle
of a snowmobiler who is inclined to violate snowmobile laws. Research pertaining to
traffic law violators and ﬁsh and wildlife law violators has identiﬁed a number of
characteristics differentiating violators from other participants in those activities.
Accordingly, these bodies of literature are examined to aid in formulating a conceptual
set of variables likely to distinguish snowmobile law violators from other snowmobilers.

Researchers studying trafﬁc and ﬁsh and wildlife violators have developed their
own deﬁnitions and research procedures. As a result, there are some fundamental
differences between these two bodies of research that are important to note. For instance,
offenders tend to be characterized differently. Trafﬁc safety researchers often deﬁne
violators based upon a speciﬁed quantity of trafﬁc law convictions or the seriousness of
their violation(s). A deﬁnition involving quantity is used because some trafﬁc researchers
generally feel that some driving violations are unintentional errors so that delineating
individuals with more than one citation tends to segment the willful violator (Reason et
al., 1990). In addition, trafﬁc researchers classify violators as ‘high risk’ drivers. This
spectrum also includes those involved in accidents and those who engage in dangerous
behaviors such as speeding. Fish and wildlife researchers, on the other hand, tend to
characterize violators based on a single conviction. However, these violators are not
necessarily at high risk to repeat violations or a hunting accident.

Lastly, offenders are oﬁen distinguished from non-offenders based upon self-
reported behaviors or self-numerated convictions. Thus, members of comparative groups

literally classify themselves and may not be mutually exclusive. Research comparing ﬁsh

14

and wildlife law violators from non-violators does so with convicted violators (as noted
by a court of law) as a separate group, not relying on self-classiﬁcation.

The review of literature presented begins with a brief overview of the theoretical
foundations and models used to explain deviant and criminal behavior. The rest of the
chapter is a synopsis of key distinctions between folk crime violators and non-violators as
well as accident victims and non-victims. This section is organized using the three
conceptual components of the PRECED Model. It begins with the predisposing
inﬂuences, which includes past behaviors, perceptions of risk, attitudes and the theory of
planned behavior, personality, emotions, and demographic factors. Next, enabling
inﬂuences cover topics pertaining to exposure, vehicle characteristics, and skill levels,

while reinforcing inﬂuences include factors related to associations and motives.

Overview of Criminal and Deviant Behavior Theories and Models

There are a wide range of theories, and subsequently conceptual models,
concerning the causation of criminal and deviant behaviors (Cullen & Agnew, 1999;
Williams & McShane, 1999). Typically, the emphasis of these theories and models
corresponds to major scientiﬁc disciplines such as biology, which looks to explain
behaviors using genetic, chemical, neurological, or physiological variables; psychology,
based upon personality or emotional aspects; and sociology, which explains behaviors
using cultural and socio-demographic traits (Akers, 1994). Over the years, new theories
and models have emerged integrating biology, psychology, and sociology elements
(Miether & Meier, 1994; Williams & McShane, 1999).

In addition, these theories and models are often further segmented based upon on
their analytical dimension (Akers, 1994). For instance, some of the models focus on the
individual’s human nature. They attempt to explain how someone becomes a criminal.
Typically, these models focus on fundamental attributes, such as disposition, to explain

differences between individuals who commit versus those who refrain from committing

15

criminal acts. Contrary to these models, the second dimension focuses on the social
phenomenon, relating to social and cultural factors, of offender groups. As a result, the
variables and structural relationship between model components vary considerably,
ranging ﬁom simple sequential linear models to interactive and dialectical models
(Einstadter & Henay, 1995).

These criminal and deviant theories and models have limited applicability in
studying folk crimes. For example, most theories focus exclusively on serious crimes
involving willful intent (e. g. murder, rape) versus technical inﬁactions (e.g. trafﬁc and
public disorder offenses) (Ross, 1961 ). Moreover, they tend to concentrate on the explicit
aspects of a criminal’s behavior, investigating only a few key variables rather than a
totality of circumstances. However, integrative behavioral models focusing on targeted
behaviors (legal or illegal) have proven more useful in studying common risky and illegal
behaviors. For instance, the Theory of Reasoned Action (Ajzen, 1985), which postulates
behavioral intentions are a function of attitudes, subjective norms, and perceived control,
has been used by trafﬁc safety researchers as a way to examine those involved in a wide
number of risky and illegal trafﬁc behaviors (Parker et al., 1996). More recently the
PRECEDE Model, which postulates our behaviors are inﬂuenced by predisposing,
enabling, and reinforcing factors, has been applied to develop an inventory of inﬂuences
and circumstances related to driver’s speeding behavior (Gabany et al., 1997). The merit
of the PRECEDE model, over the Theory of Reasoned Action Model, is that accounts for
a wider range of prebehavioral inﬂuences and circumstances. As a result, it is adopted
here to conceptualize and analyze differences between 1996-97 convicted snowmobile

law violators and other snowmobilers.

l6

Predisposing Inﬂuences

Aberrant driving behaviors, and in many cases trafﬁc accidents, often have been
attributed to an individual’s disposition. Consequently, the psychological functioning of

the driver has been a major subject of study of trafﬁc safety researchers.

Past Behaviors

Some of the earliest trafﬁc research of those involved in accidents detected a
moderate, but reliable, correlation to prior trafﬁc violations. For example, Jonah and
colleagues (1981) conducted interviews with a sample of motorcycle operators from
Ontario, Canada a year after they were licensed to ride. In a comparison of those involved
in an accident with those who were not, they found that the accident group was more
likely to a have trafﬁc violations than the non-accident group. Moreover, they found the
accident group was signiﬁcantly younger, rode more powerful motorcycles, rode more
miles, and reported that they had ridden after using alcohol. Of these factors, the age of
the operator was found to be highly correlated with accident involvement, leading the
authors to conclude that younger riders may lack experience and be more likely to engage
in risk-taking behaviors.

Similar ﬁndings about the relationship between trafﬁc violations and accidents are
noted with automobile drivers. For example, Garretson and Peck (1982) used
discrinrinant analysis to differentiate a sample of fatally injured California drivers from a
sample of drivers from the general population based on the total number and seriousness
of prior trafﬁc violations. They concluded that generally young male drivers with prior
convictions were more likely to be involved in a fatal accident than others.

Other researchers have examined the link between prior trafﬁc violations and
speciﬁc risky behaviors. Evans and Wasielewski (1982) compared the driving records of
individuals identiﬁed from license plates in photographs measuring the time interval

between them and the preceding vehicle in the same lane. As they predicted, drivers who

17

1eﬁ less than one second between themselves and the preceding vehicle had signiﬁcantly
more points associated with trafﬁc law violations. Using speeding as a form of risky
behavior, Cooper (1997) matched the driving records of individuals involved in trafﬁc
accidents between 1991 through 1994 and found that drivers with more total speeding
convictions were associated with a higher number of crashes. This was especially true if
the convictions were for excessive speeds, deﬁned as 40 km/h or more over the speed
limit. The relationship between the abundance and seriousness of prior trafﬁc violations
and accident involvement has been the crux of much of the trafﬁc safety research. This
led to the question, what are the fundamental psychological inﬂuences on these risky
drivers? One such area of trafﬁc safety research that logically grew from this relates to
the perception of risk.

Perceptions of Risk

Trafﬁc safety researchers turned to investigating trafﬁc offenders’ perceptions of
risk and the association of that with skills, apprehension, and accident likelihood. In one
such study of male drivers in the United Kingdom, Guppy (1993) asked violators and
non-violators, classiﬁed by self-reported drinking and driving and speeding behaviors, to
rate the probability of being apprehended by police and being involved in an accident.
Through an analysis of variance, Guppy concluded that both those who drink and drive
and those who speed considered themselves less likely to be involved in an accident or
being apprehended than those who did neither.

The likelihood of an accident was also the subject of a study by DeJoy (1992).
Through a questionnaire, he asked male and female college students to judge their driving
safety, skills, and accident likelihood against others in their own age group and the
average driver. Additionally, they rated ﬁﬁeen deviant driving behaviors on frequency,
seriousness, accident likelihood, and apprehension likelihood. Finally, they self-reported

their involvement in accidents and their convictions for speeding, driving under the

18

inﬂuence, and other moving violations. Using correlation analysis, DeJoy found that
males considered themselves to be more competent and safer than females. While both
males and females held similar perceptions about the frequency and likelihood of
accidents resulting from deviant behaviors, males tended to perceive the behaviors as less
serious. A violator index, created by summing all self-reported trafﬁc convictions, was
used as the dependent variable in a follow-up regression analysis of perceived driving
safety, skill, and accident likelihood. It was concluded that those with worse records rated
themselves as less skillful and safe.

Jonah (1986) synthesized much of the early research related to young drivers and
risky behaviors such as speeding or driving under the inﬂuence. He concluded that, even
when controlling for exposure and age, younger drivers, particularly males (18-25 years
old), engage in risky deviant behaviors more than others. This was the same conclusion

reached by Elander and others (1993) in another review of trafﬁc safety literature.

Attitudes and the Theog of Planned Behavior

Attitudes also have been investigated by researchers to identify distinctions
between violators and non-violators. In one such study of Michigan deer hunters,
Andersen (1987) used a mail questionnaire to compare attitudes towards wildlife laws
and management between all those cited in 1982 for violating deer laws with a random
sample of deer hunters derived from the same hunting year license sales. Additionally, he
used ten socio-economic variables to explore their relationship to the formation of these
attitudes. Through multivariate analysis he found partial support for the hypothesis that
violators generally have unfavorable attitudes towards wildlife laws and management. In
the analysis of socioeconomic variables, he found that lower occupation status, lower
income, younger, long-time rural residents, and less educated individuals had more
negative attitudes towards law enforcement and were more likely to violate. Similar

attitudes towards wildlife laws, enforcement, and management as well as socioeconomic

l9

variables have been corroborated in other studies comparing apprehended violators with
lawful hunters (Stoll, 1975; Melynk, 1977).

Attitudes also have been the subject of investigation in research concerning trafﬁc
violators and non-violators. Kanellaidis and others (1995) surveyed driver attitudes
toward speed limits in Greece to determine a proﬁle of speed limit violators. Using a mail
questionnaire, data were collected about the driving habits, attitudes and views about
speed limits, self-reported compliance with speed limit laws, demographic information,
and attributes of the vehicle driven. Discriminant analysis was used to proﬁle violators,
while factor analysis was used to analyze the reasons for speeding. The discriminant
analysis revealed that those who reported seldom or never complying to speed limits were
younger males, had higher education level attainment, and annually drove a higher
number of miles. Three factors were revealed by the factor analysis regarding violator’s
attitudes toward speed limits. The ﬁrst factor described violators who show off or speed
as a way to release emotions. The second factor identiﬁed external reasons such as the
absence of police or wanting to keep up with trafﬁc, while the last factor showed a
general disagreement with speed limits.

Similar conclusions were reached in a study of Illinois motorcycle riders who had
taken a safety course versus those who had not in during 1980-1983 (Mortimer, 1988).
Using regression analysis with individuals cited for moving violations, Mortimer
concluded that those with violations were younger, less likely to wear helmets, and more
likely to be involved in accidents. The author concluded that “attitudes about safety gear,
inexperience, and other aspects of youth and inadequate skill and judgment are related to
motorcyclists’ violations” (Mortimer, 1988, p. 194).

Donovan and others (1983) summarized several studies relating driver attitudes,
violations, and accident occurrences. Their synthesis identiﬁed six attitudes that enhanced
driving risk. These included: 1) attribution of the causes of accidents to factors beyond

personal control, 2) driving as a means to reduce tension, 3) driving as a means of

20

increasing the perception of personal efﬁcacy, status and power, 4) driving as a means of
increasing personal conﬁdence, 5) positive evaluation of speed, risk-taking, and sensation
while driving, and 6) an aggressive attitude while driving through which acute and
chronic anger and resentment are expressed (Donovan et al., 1983, p. 410).

F ishbein and Ajzen’s (1975) model, the Theory of Reasoned Action, and its
updated version, the Theory of Planned Behavior (Ajzen, 1985) have been used by trafﬁc
safety researchers as a way to examine the impacts of attitudes, subjective norms, and
perceived control on behavioral intentions related to law violations. One such study
assessed the intentions of drivers from southern England to engage in four types of
deviant traffic behaviors (Parker et al., 1992). These included drinking and driving,
speeding, close following, and risky overtaking. Based on a series of regression analyses
using a number of variables measuring driving experience and demographics, they
concluded that violators were more likely to be males, have high opinions of their skill,

and be less concerned about other people’s opinions.

Personalig
Many trafﬁc safety researchers have explored the function of decision making

styles as independent measures, beyond beliefs and attitudes, of tendencies to violate
trafﬁc laws (Elander et al., 1993). Accordingly, psychometric measurements of
personality related to sensation seeking, impulsively, hostility, and others have been used
in numerous studies of trafﬁc law violators and non-violators. One of the early studies to
use measures of violator personalities was Shoham and associates (1984). In their study,
they compared risk-taking, irnpulsiveness, anxiety, sensation seeking, driving habits, and
other variables of known violators and non-violators. As postulated, non-violators
(described as anxious drivers) and violators (depicted as reckless drivers) were found to
differ from each other on all personality measures except risk-taking. They also noted that

the reckless drivers tended to display deviant behaviors and break the law more often.

21

McMillen and others (1992) examined personality traits of drinking and non-
drinking college students using scales measuring sensing-seeking, social responsibility,
and hostility. In addition to the personality traits, students were asked about driving
habits, accidents, trafﬁc offenses, drinking behaviors, and perceptions of risk-taking by
driving aﬁer drinking. Through an analysis of variance, they concluded that those
individuals apprehended for driving under the inﬂuence of alcohol (DUI) were
signiﬁcantly different on all personality measures ﬁom the other group.

Wilson (1992) found similar personality traits in a comparable study of convicted
DUI drivers in Ontario, Canada. However, this study differed from others because, in
addition to the DUI and control group, it also examined drivers having three or more
accidents and drivers with a high number of traffic law violations. The researchers
obtained the driving records of all subjects and ascertained driving attitudes and
motivations, deviant lifestyle habits, and social-environmental inﬂuences through
interviews. As expected, the DUI group differed by having a more deviant lifestyle (using
drugs, alcohol, etc.) from the accident and violation group, which was found to differ
from the control group. However, with the exception of the control group, no differences
were found with regard to driving-related attitudes.

Other studies of trafﬁc violators have examined the role of social deviance. For
example, a series of questionnaires was administered to 108 drivers in England over a
three-year period starting in 1988 to collect data on personality, demographic details,
driving habits, accidents, and a scale measuring general social deviance (West et al.,
1993). This scale asked drivers to report the likelihood that they would be tempted to
engage in a variety of deviant behaviors where no one was harmed. Over the three-year
period, young males were more likely be involved in an accident and report engaging in
many forms of deviant behaviors. Lawton and associates (1997a) replicated the West et
a1. (1993) study and expanded it to incorporate self-reported violations. Their study used
a sample of 830 drivers from Reading and Manchester in the United Kingdom. Their

22

regression analysis conﬁrmed the association between social deviance, accidents, and
driving violations. They also concluded that driver age was negatively associated with

accident likelihood and the tendency to commit violations.

Emotions

The emotional state of individuals also has been shown to be a salient factor of
trafﬁc accidents, but it has not been widely examined with trafﬁc offenders until the
1990’s. Emotions were the focus of a study by Lawton and colleagues (1997b) in which a
small sample of questionnaires was distributed on cars parked at shopping areas in
Manchester, United Kingdom. The questionnaire ﬁrst asked drivers to evaluate their
engagement in twelve common trafﬁc violations; they were asked to rate four (2 positive
and 2 negative) affective items about each of the trafﬁc violations. The last parts of the
questionnaire measured personality, demographic characteristics, and driving habits.
Using principle component analysis on the self-reported behaviors, three factors were
identiﬁed. These include fast driving, assertive driving, and hostility directed at other
road users. Analysis of variance with the demographic and driving habits revealed that
young drivers had more tendencies to violate trafﬁc laws, and because of assertive
driving, males were more prone to violate trafﬁc laws than females. Finally, in a
regression analysis with the violation scores as dependent variables and affective scores
as independent variables, they found that more positive feelings towards violating

resulted in an increased tendency to violate.

Demographic Factors

Numerous studies of trafﬁc violators and non-violators have sought to identify
stable independent demographic variables that contribute to law violations or accident
risk. Age and gender (i.e. young males) constantly have been correlated with the

propensity to violate trafﬁc laws (Jonah, 1986; Elander et al., 1993). However, Reason

23

and others (1990) used demographic variables to make distinctions between deliberate
violations of the law and errors of judgment that led to violations. In order to obtain a
wide range of ages, 520 drivers in their study were given mail back questionnaires at their
cars parked along streets or in parking lots. This questionnaire was designed to elicit
demographic information, driving habits, and self-assessments of driving skill, law
compliance, and likelihood of making driver errors. In addition, they also were asked the
frequency with which they committed 50 different violations categorized as either ‘slips’
caused by misinterpretation, ‘mistakes,’ ‘unintended violations,’ or ‘deliberate
violations.’ The factor scores from three distinct conditions identiﬁed as deliberate
violations, dangerous errors, and silly errors, were used as dependent variables and
regressed against the demographics, driving habits, and self assessments factors. Being
young, male, driving a high number of annual miles, and believing one was a better driver
than others were found to be signiﬁcant predictors to the commission of deliberate

violations.

Enabling Inﬂuences
Enabling inﬂuences are generally thought of as factors that help provide the

means that lead to deviant behaviors (Gabany et al., 1997). These factors include

exposure (the miles driven each year), characteristics of vehicles used, and driving skill.

Exposure and Vehicle Characteristics

The role of exposure, deﬁned as the annual miles driven and time spent driving,
has been shown by some trafﬁc safety researchers to be positively correlated with the
propensity for trafﬁc violation and accident involvement. Taken at face value, it is logical
to contend that the more one drives, the greater the likelihood that they will receive a
trafﬁc citation. As a result, such variables distinguishing trafﬁc law violators from non-

violators have been challenged as independent factors. Such criticism has merit, because

24

early studies comparing violators and non-violators ignored exposure as a potential
source of bias (Elander et al., 1993). In spite of this, even though most studies today
control for exposure, a positive correlation between annual miles driven still has been
demonstrated as a distinction between trafﬁc law violators and non-violators (West et al.,
1993; Kanellaidis et al., 1995; Lawton et al., 1997a). A second factor that has received
limited attention by trafﬁc safety researchers has been the role of vehicle characteristics
as enabling inﬂuences leading to trafﬁc law violations. In most studies in which such
information has been incorporated, very little signiﬁcance has been demonstrated
(Kanellaidis et al., 1995). However, in some cases, vehicle characteristics, such as engine
size, have been found to be mildly associated with the propensity for trafﬁc accidents

(Rothengatter, 1988).

Skill Level

Skill also has been identiﬁed as differentiating those involved in trafﬁc accidents
ﬁ'om those who are not, although the association related to operating automobiles is
primarily limited to hazard perceptions and visual skills (Elander et al., 1993).

Skill level is different when it comes to differentiating those hunters apprehended
for violating game laws from legal hunters. Such was the case in a study reported by
Shafer and others in 1972. In their study, a mail questionnaire was sent to 1,200 New
York deer hunters, half of whom had been convicted of a deer law violation in the past
ﬁve years, while the other half had no violations recorded. The results indicated that
violators were more successful at harvesting deer and presumably more skillful than non-
violators. Other distinctions noted were that violators had more contact with conservation
law enforcement ofﬁcers during their years of hunting and felt that it was appropriate to
illegally harvest deer if it was for food. Two similarities between violators and non-
violators also were found in their study. These included similar attitudes toward hunting

laws and regulations as well as investments in hunting equipment.

25

Similar distinctions regarding skill were noted by Jackson and his colleagues
(1979) in a comprehensive study of waterfowl violators in Wisconsin. Through concealed
ﬁeld observations, they identiﬁed and quantiﬁed game law violations and conducted ﬁeld
interviews with waterfowl hunters. An additional post-season at home interview was
conducted with those who agreed, to collect data on demographics, behavior scales, and
open-ended questions. Several distinctions were made in their comparison of violators
and non-violators. Demographic information revealed that violators were younger and
had hunted fewer years than non-violators. In addition to better seasonal and daily
successes, violators had more opportunities to shoot at game and tended to shoot more
times when given the chance. Violators also were considered more dedicated to their
sport, practicing more at skeet ranges, using duck calls, and reading technical information
about waterfowl hunting. Moreover, violators were considered locals of the areas where
they violated and knew the areas well. These same general conclusions and distinctions
also were made in an extensive study carried out by Gray and Kaminski (1994) in 1990 in
which they compared waterfowl violators and non-violators from twelve different states
comprising the Mississippi F lyway. They also noted that violators hunted more

frequently than non-violators.

Reinforcing Inﬂuences

Reinforcing inﬂuences are considered to be those factors that provide individuals
with an incentive or reward for conducting themselves in a deviant manner (Gabany et
al., 1997). Examples of such factors are group associations (including family), lack of law
enforcement presence, and others. Jackson (1980) reported in his study of waterfowl
violators about developmental inﬂuences on hunter behaviors and characteristics. Results
indicate that violations tended to occur more ﬁequently with the presence of one or more

hunting companions. This suggests that hunters’ associations may in some way

26

encourage violations. He also found that hunters graduating from hunter safety courses
were no more likely to comply with wildlife laws than those not taking such courses.

Illegally taking ﬁsh and wildlife also has been examined by sociologists applying
more dynamic approaches to explain these types of violators. Such investigations have
turned to the broader area of folk crimes to characterize ﬁsh and game violators.
Typically such research relies upon anecdotal interviews of small groups of admitted ﬁsh
and game law violators and law enforcement ofﬁcials. One of the earliest such studies
was conducted by Wilson (1983) in an investigation of commercial shell ﬁsherman
known to violate conservation laws in New England. Through observations and
interviews with poachers and state ﬁsh and game wardens, she isolated characteristics
leading to a pattern of mutual accommodation by violators and enforcement personnel
allowing the persistence of this form of deviant behavior. In this case, accommodation for
violators “involves not ﬂaunting their violations, exercising self-regulation in time and
place of offense, not embarrassing enforcers, and remaining within the bounds of the
intent of the law” (Wilson, 1983, p. 128). As for the enforcers, accommodation “is
premised on wide discretion in enforcement and involves avoidance of the crime scene
and revealing their presence” (Wilson, 1983, p. 128).

Wilson’s work help set the stage for similar studies that have detailed other
characteristics of ﬁsh and game law violators. For example, a cohort of sixteen known
marine recreational ﬁsherman who poached in southern California were interviewed by
Curcione (1992) over a period from 1987 to 1990. This group ranged from 22 to 56 years
olds and all indicated that they were introduced to ﬁshing through a family member at an
average age of 10 years old. These same mentors also exposed them to deviant values and
situations where ﬁshing laws were regularly violated. Because these violators were
characterized as skilled ﬁsherman, they received special amenities (e.g., ﬁee trips, etc.)
ﬁom boat captains as motivation to ﬁsh on their boats. This was because boat captains

wanted to increase catch rates to show that they were successful boats, thereby increasing

27

business. Lastly, no one had criminal records, other than a speeding ticket, nor did they
perceive any likelihood of being caught. A similar approach with similar ﬁndings was
used in a study of 36 poachers in rural Louisiana by Forsyth and Marckese (1993). While
the prime motive for poaching was for food, they found some indication that it also was a
form of pleasure and excitement, offering a way to test their skills in outwitting the
wardens. Forsyth and associates (1998) have replicated their previous study of poachers
in rural Louisiana and augmented it with interviews of game warders. The motives of
poachers was still primarily to provide food for their family, but other reasons included
money, excitement, and tradition. Game wardens were more lenient on poachers who
hunted for food, and less so on those who poached for money or excitement, especially if
it involved a threatened or endangered species. These same motives of poachers as well

as others are outlined by Muth and Bowe (1998) in a synthesis of literature on poaching.

28

CHAPTER THREE
RELEVANT FACTORS AND HY POTHESES

The review of research examining the circumstances of snowmobiling accidents
as well as research comparing trafﬁc and ﬁsh and game law violators with non-violators
has identiﬁed a number of salient factors that predispose, enable, and reinforce trafﬁc
accident involvement and aberrant behaviors. Accordingly, these factors are used here to
conceptualize relevant distinctions between 1996-97 convicted snowmobile law violators
and other snowmobilers. Based upon the review of literature, an explanation of the
potential distinguishing factors and rationale for their inclusion are furnished. A
hypothesis statement describing the relationship of each factor between 1996-97

convicted snowmobile law violators and other snowmobilers follows.

Predimsing Factors

Age

Consistently, evidence ﬁom previous snowmobile and trafﬁc safety and ﬁsh and
wildlife research has indicated that 16 to 25 year olds are more likely to commit law
violations and be involved in trafﬁc and snowmobile accidents (e.g., Dewar, 1973;
Jackson et al., 1979; Anderson, 1987; James et al., 1991; Elander et al., 1993). However,
while youthﬁrl age generally has been found to be highly correlated with such events,
many researchers believe that it is not age itself that is the cause, but some combination
of this age group’s tendency to engage in risky behaviors and/or their lack of experience
and skills (Wallach & Kogan, 1961; Jonah, 1986). Nevertheless, it is hypothesized that:

(1) 1996-97 convicted snowmobile law violators are younger than other
snowmobilers.

29

Perceptions of the Dangers of Risky Beh_aviors and Situa_ti_on_s

Past studies of trafﬁc law violators and injured snowmobilers have shown that
they have an inadequate perception about risky behaviors and situations. For example,
Waller and Larnbom (1975) found that injured snowmobilers were twice as likely to
report that they had been ‘drinking heavily’ prior to the accident and that they had a
greater tendency to operate their snowmobile at higher speeds than uninjured
snowmobilers. These were the same factors that Rowe and associates (1993) found in
their study of injured snowmobilers.

Likewise, trafﬁc researchers have generally concluded that young drivers who
have been convicted of a trafﬁc violation and/or been involved in accidents (Mortimer,
1988; McMillen et al., 1992) perceive deviant driving behaviors (speeding, drinking and
driving, not wearing helmets, etc.) as less serious, or they lack experience in judging the
seriousness of such behaviors, than other drivers (Jonah et al., 1981). Moreover, there is
also an assertion by some researchers that individuals will purposefully underestimate the
risk of certain behaviors, such as speeding, in order to exonerate risky behaviors (Douglas
& Wildavsky, 1982). As a result of these ﬁndings, it is hypothesized that:

(2) 1 996-97 convicted snowmobile law violators will consider behaviors and

situations related to alcohol use, speeding, skill of operators, operating on

roadways, other vehicles on seasonal roads, other non-motorized trail users,

and trail conditions and design as less dangerous than other snowmobilers.
Attitudes Towards Management and ng Enforcement

Opposing attitudes toward management, law enforcement, and statutes regulating
behaviors also have been found to correlate with those convicted of violating wildlife
laws (Stoll, 1975; Melynk, 1977; Andersen, 1987) and trafﬁc laws (Mayer & Treat, 1977;
Donovan et al., 1983; Kanellaidis et al., 1995). In addition, several researchers have
found that those citizens who have had contact with police and/or received a citation

generally have a more unfavorable attitude toward police ofﬁcers than those who have

not had contact (Cox & White, 1988). As a result, it is hypothesized that:

30

(3) 1 996-97 convicted snowmobile law violators will likely be less supportive of
fatality reduction initiatives that pertain to establishing speed limits, licensing
snowmobile operators, requiring personal liability insurance, and more intensive
enforcement of existing snowmobile laws than other snowmobilers.

Enabling Factors

Expgsure and Active Participation

Presumably, the more one engages in an activity, such as snowmobiling, the
greater the likelihood events like an accident or citation will occur. Researchers refer to
this as exposure and generally deﬁne it in terms of annual miles driven or time spent
engaging in an activity. The role of exposure has been examined extensively by trafﬁc
safety researchers and most believe that greater exposure leads to greater risk of being
involved in a trafﬁc accident or being cited for violating a trafﬁc law (West et al., 1993;
Kanellaidis et al., 1995; Lawton et al., 1997a). However, a few researchers have reported
a negatively accelerating curve in the association of total annual miles driven with
increases in accident and citation risk (see Elander et al., 1993). This is due to the
assertion that as one travels more, they gain more experience and likely adopt safe
driving habits (Elander et al., 1993). In addition, they also spend more time on limited
access highways which are safer.

Studies of snowmobilers injured as the result of an accident indicate that they had
ridden more miles and spent more time snowmobiling than uninjured snowmobilers
(Waller & Larnborn, 1975; Rowe et al., 1993). Likewise, Gray and Karninski’s (1994)
comparative study between hunters who violated wildlife laws and non-violators found
that violators hunted more frequently than non-violators. As a result of these conclusions,
it is hypothesized that:

(4) I996-97 convicted snowmobile law violators will have more snowmobiling
days than other snowmobilers.

(5) 1996-97 convicted snowmobile law violators will have ridden more miles than
other snowmobilers.

31

Performance Capabilities of Snowmobiles

Previous research regarding snowmobile and trafﬁc accidents has found limited
support for the assertion that the greater a vehicle’s power, the more likely it is to be
involved in an accident. However, there are conﬂicting conclusions about this
relationship. For instance, Waller and Larnbom (1975) found that higher horsepower
(HP) snowmobiles (30 HP or more) differentiated injured and non-injured snowmobilers,
while Rowe and associates (1993) concluded that the use of high performance
snowmobiles (greater than 500 cc) appeared not to adversely inﬂuence accident
proneness. Incorporating automobile engine size as an enabling inﬂuence leading to an
increase in trafﬁc law violations or accidents also has limited support (Rothengatter,
1988). In addition, research associating motorcycle engine size with accident or citation
risk has found conﬂicting conclusions. For instance, Jonah and associates (1981) found
that engine size was statistically signiﬁcant with those motorcyclists involved in an
accident, but not with those who received a citation. The opposite conclusion was reached
in another study of motorcyclists by Jonah and colleagues (1982). In both studies, engine
size did not add much in the way of explanation to models distinguishing violation
involvement. Nevertheless, it is hypothesized that:

(6) 1996—97 convicted snowmobile law violators will have a higher than average
engine size (cc) than other snowmobilers.

Reinforcing Factors

Family Based Activig

The developmental inﬂuences of hunter behaviors also have been the subject of
studies of hunters and ﬁshermen who have violated game and ﬁsh laws. As a result of
this work, researchers have found that violators were both introduced to the activity and
exposed to deviant values and situations by family members (Curcione, 1992; Forsyth &

Marckese, 1993). As a result, it is hypothesized that:

32

(7) The proportion of household members who snowmobiled in 1996-97 will be
will be greater for convicted snowmobile law violators than other snowmobilers.

Investment in Snowmobiling
In Jackson and colleagues (1979, p. 316) comprehensive study of waterfowl

violators, they concluded that violators were “deeply involved and dedicated to hunting.”
Violators were found to practice more at skeet ranges, use duck calls, and read technical
information about waterfowl hunting than non-violators. However, Shafer and his
colleagues (1972) found that violators and non-violators were similar in their investments
in hunting equipment. Nevertheless, it is hypothesized that:

(8) 1996-97 convicted snowmobile law violators will have more snowmobiles than
other snowmobilers.

(9) I996-97 convicted snowmobile law violators will have newer snowmobiles
than other snowmobilers.

(10) 1996-97 convicted snowmobile law violators will have spent more on
snowmobiling equipment than other snowmobilers.

Lack of Concern for Safeg Education

Jackson (1980) further reported in his study of waterfowl violators that those
hunters who graduated from a hunter safety course were not any more likely to comply
with wildlife laws than those who did not take such a course. Nevertheless, it is
hypothesized that:

(I 1) 1996-97 convicted snowmobile law violators will have been less likely to

have had someone in the household complete a snowmobiling safety class than
other snowmobilers.

Smnmagy of Hypothesized Distingpishing Variables
A summary of the scale and data type of the hypothesized predisposing, enabling,

and reinforcing variables potentially differentiating 1996-97 convicted snowmobile law

violators and other snowmobilers is displayed in Tables 1 and 2.

33

Table 1. Scale and data type of potential predisposing variables differentiating 1996-97
convicted snowmobile law violators and other snowmobilers.

 

 

Data
Predisposing Variables Scale Type
Respondents age 14-81 Ratio
Dangerous behavior or situation rating '
Operation of snowmobiles by intoxicated persons 1-5 Ordinal
Drivers lacking skill in operating their machines 1—5 Ordinal
Cars or trucks on seasonal roads 1-5 Ordinal
Other uses of snowmobile trails (e.g. X-C skiing) 1-5 Ordinal
Speed of snowmobiles 1-5 Ordinal
Operation of snowmobiles by persons who have been 1_5 Or din al
drinking, but are not intoxicated
Public trail conditions 1-5 Ordinal
Public trail design 1-5 Ordinal
Snowmobiling on county or state roads 1-5 Ordinal
Fatalig reduction initiatives rating b
Mandatory training for all ﬁrst year snowmobilers regardless 1_5 Or dinal
ofage
Mandatory personal insurance for all snowmobilers 1-5 Ordinal
More intensive enforcement of existing snowmobile -
regulations 1-5 Ordinal
Licensing required for all snowmobilers 1-5 Ordinal
Enforcement of a snowmobile speed limit on public lands, 1_5 Or din al

trails, and waters

 

a. Rating scale: 5= extremely dangerous, 4= highly dangerous, 3= moderately dangerous,
2= slightly dangerous, 1= not dangerous.

b. Rating scale: 5= strongly supports, 4= moderately supports, 3= neutral, 2= moderately
oppose, l= strongly oppose.

34

Table 2. Scale and data type of potential enabling and reinforcing variables differentiating
1996-97 convicted snowmobile law violators and other snowmobilers.

 

 

Factor
Variables Scale Data Type
Enabling
Average number of estimated snowmobiling days per
household snowmobile ’ 1'1 15 Interval
Average number of estimated miles ridden per household 10-5,333 Interval
snowmobile ’
Average engine size (cc) of household snowmobiles ’ 90—1,100 Ratio
Reinforcing
Proportion of the household 12 years and older who
snowmobiles during the winter of 1996-97 0-17'1-00 Interval
Total nmnber of snowmobiles owned by the household 1-6 Interval
Average age of the of household snowmobiles 1965-1997 Ratio

Total household expenditures for snowmobiling equipment, .
maintenance, insurance, and storage 0’126’550 Rat“)

Proportion of the household 12 years and older with a

snowmobile safety class 0-13'1 '00 Interval

 

a. Averaged based on those snowmobiles used in Nﬁchigan during the 1996-9'7 wrnter.

35

CHAPTER FOUR
METHODS AND PROCEDURES

The data used to accomplish the goal of this study to identify salient variables
discriminating convicted snowmobile safety law violators from other snowmobilers was
complied in two separate projects. First, data on other snowmobilers used in this study
were compiled in a winter 1996-97 statewide assessment of those who purchased a
Michigan Snowmobile Trail Permit (Nelson et al., 1998). A Snowmobile Trail Permit is
required to operate any snowmobile in Michigan on public lands (including road right-of-
ways) and ﬂown waters of the state except for the sole purpose of ice ﬁshing. This
study’s purpose was to examine resident and non-resident snowmobiling activity,
household spending, opinions about management, support of policy initiatives,
perceptions of dangerous situations, and factors contributing to snowmobile accidents.

The 1996-97 convicted snowmobile violator data were compiled using the same
instrument as the study of trail permit purchasers from a census of violators cited by
Michigan DNR-LED during winter 1996-97 and convicted of violating safety regulations
(Lynch & Nelson, 1998b). The purpose of this violator study was to better understand
those who engage in risky snowmobiling behaviors by comparing them to other trail
permit purchasers. Of the trail permit purchasers ﬁom the initial sample, those few who
had been convicted of a 1996-97 snowmobile safety violation were eliminated. Both
these studies were administered by the Department of Park, Recreation, and Tourism
Resources at Michigan State University with funding from the DNR Forest Management
Division (F MD) and Law Enforcement Division (LED), with additional support from the
Michigan Agricultural Experiment Station.

36

The research reported here uses this information to distinguish underlying factors
that potentially inﬂuence aberrant snowmobiler behavior by 1996-97 convicted
snowmobile law violators. This will be accomplished through a comparison of 1996-97
convicted snowmobile law violators and other snowmobilers on a number of elements

based on the aforementioned hypotheses.

Subjects and Sampling Protocol

1996-97 Convicted Snowmobile Law Violators

Convicted snowmobile law violators are deﬁned as those individuals who were
cited by Michigan DNR Conservation Ofﬁcers during the winter of 1996-97 and
convicted of violating operation and safety regulations as outlined by the Michigan
Snowmobile Law P.A. 74 of 1968 as amended. These citations are kept on ﬁle at the
DNR-LED in Lansing, Michigan. Copies of all 997 citations on ﬁle as of May 1, 1997
were procured by the researchers from Michigan State University and used in this study.
The name, address, age, type of offense, number of offenses and warnings, and county
where the citation was issued were entered into Microsoft’s Excel spreadsheet program.
Seven individuals received more than one citation during the winter of 1996—97 and thus
duplicate names were removed ﬁ'om the database. Additionally, the names of seven other
1996-97 convicted snowmobile law violators were removed because they had been
selected in the sample of other snowmobilers. As a result, there were 983 individuals in
the 1996-97 convicted snowmobile law violator group.

It is important to note that while the convicted snowmobile law violators included
in this study are assumed to be representative of 1996-97 convicted Michigan
snowmobile law violators, this can not be proven. Rather, they are those who were caught
by DNR Conservation Ofﬁcers and received a citation. County Sheriffs and Michigan

State Police also issued an unknown number of citations to snowmobilers violating

37

snowmobile laws during the winter 1996-97. Statewide data on their activities were
unavailable at the time of the study, although following winter 1999-2000 such records
are likely to be available ﬁ'om DNR-LED. At the time of the study, such citations were
reportedly kept on ﬁle in each of the 83 counties or with the Michigan State Police.
Hence, individuals who were cited by either County Sheriffs or Michigan State Police,
but not Conservation Ofﬁcers, were excluded from this study because it was impractical
to acquire these citations. Moreover, no attempt was made to ask respondents about
whether or not they had received a citation for any other offense this past winter or any
other time. While such information may have proven useful, it was deemed too sensitive

to request by the Michigan DNR-LED, the contracting agency.

Other Snowmobilers

Other snowmobilers are deﬁned as those individuals who purchased a Michigan
Snowmobile Trail Permit between October 1, 1995 and September 30, 1996 and
snowmobiled during winter 1996-97 and were not cited by the DNR-LED for a
snowmobile safety violation during that winter. A trail permit is required for a
snowmobile to operate on public lands, waters, and roadway shoulders where permitted.
In ﬁscal year 1995-96, approximately 212,000 permits were sold by a host of agents,
including the Michigan Snowmobile Association; various local snowmobiling clubs in
Michigan; license agents such as sporting goods stores, snowmobile retailers, or chain
stores (such as Meijers, K-Mart, etc.); the Michigan Secretary of State; and the DNR. The
record of the sale, including the name, address, and telephone number of the permit
holder is maintained by the FMD of the DNR. While it appears that some permit records
held by the issuing agent were not returned to the F MD during the ﬁscal year 1995-96,
the vast majority were. These records served as the sampling frame for this study since
complete permit records for the winter 1996-97 were not available at the time of the

study.

38

A sample of 3,325 snowmobilers was systematically selected by taking every 30th
legible name with complete address from a random starting point. Sampling procedures
were employed to ensure that commercial businesses (e.g., snowmobile rental
establishments) and duplicate names were not selected in the sample. Additionally,
procedures ensured that those who had purchased multiple permits were no more likely
than those purchasing a single permit to be selected. This provided data with a margin of
sampling error of :l: 1% at the 95% conﬁdence interval. The names, addresses, and phone
numbers were then entered into Microsoft’s Excel spreadsheet program.

The rationale for choosing trail permit purchasers versus snowmobile registrants
as the sampling ﬂame is threefold. First, Michigan registrants do not include the vast
majority of non-residents who snowmobile in Michigan. It was estimated that about a
third of the snowmobilers in Michigan are non-residents. Secondly, registrations may or
may not include snowmobiles that are operated, or are intended to be operated, on public
lands. Conversely, a trail permit is required to operate on public lands and speciﬁc trails
for which public land managers are responsible. Lastly, if registrants were used, then a
sample of registrations from many surrounding states would need to be generated to
capture the range of non-residents who snowmobile in Michigan. This would be
complicated since the proportions of non-residents from each state snowmobiling in
Michigan was unknown.

While these other snowmobilers are characterized as non-violators, as they were
not cited by DNR-LED Ofﬁcers in the winter 1996-97, they can not be regarded as never
having violated snowmobile laws or having been cited during another year. Like the
1996-97 convicted snowmobile law violator group, no attempt was made to ask
respondents about whether or not they had received a citation for any other offense this
past winter or any other time. While such information may have proven useful, it was

deemed too sensitive to request by the Michigan DNR-LED, the contracting agency.

39

Data Collection Method

Technique

In choosing the most appropriate method to elicit the necessary data, several
constraints were taken into consideration. First, since this was a comprehensive
assessment, nine broad objectives were outlined for this project. To ensure the objectives
were met, it was necessary to employ lengthy and somewhat complicated items to rate the
array of programs and potential options as well as enumerate details about snowmobiles,
activities, and spending habits. Moreover, some of the questions were personal in nature
and required a level of detail involving considerable thought by the respondents.
Secondly, the magnitude of Michigan’s snowmobiling population is substantial and
encompasses residents as well as non-residents from a number of states and provinces. Of
the 212,000 trail permits sold during 1995-96, an estimated 79,000 were sold to non-
residents (Nelson et al., 1998). Lastly, ﬁnancial and personnel resources to conduct this
study were limited.

As a result of these constraints, it was decided that a self-administered mail
questionnaire was the most appropriate means to collect the desired data. Mail
questionnaires have been used successfully in a number of studies comparing violators
and non-violators (Shafer, 1972; Kessel, 1974; Andersen, 1987; West et al., 1993; Gray
& Kaminski, 1994). A mail questionnaire also has many advantages (Mangione, 1995).
For example, it allows a researcher to contact a larger sample over a wide geographic area
relatively inexpensively. Larger samples make it possible to make more precise
statements about the group. Moreover, mail questionnaires give respondents time to
answer the questions in private and at their own convenience. They also allow an
individual an opportunity to collect information about questions that may not be readily

available.

40

Against these advantages, there is a serious limitation of self-administered mail
questionnaires that needs to be noted. As Mangione (1995, p. 60) points out, “non-
response error is the single biggest impediment to any survey study, but it is particularly a
risk for mail surveys.” As a result, a phone survey of a sub-sample of non-respondents

with a shortened questionnaire was used with both groups to determine the effects of non-

response bias.
Questionnaire Desigp and Administration
Questionnaire Desigr

Questionnaire design can affect the quality of the data as well as the quantity of
responses (Babbie, 1995). As a result, special attention was given to devising a simple
and concise questionnaire that effectively achieved all study objectives. The
questionnaire initially drafted by the researchers was revised during two rounds of
reviews by DNR staff from LED and FMD. It was also reviewed and approved by the
Michigan State University Committee on Research Involving Human Subjects.

The result was a four-page questionnaire comprised of 31 questions measuring
124 items (Appendix A). About equal numbers of open-ended and closed-ended
questions were employed. The vast majority of open-ended questions asked respondents
to enumerate speciﬁcs about their household, snowmobiles, snowmobile activity, and
household spending on snowmobiling. To simplify questions for ease of answering, a
table format was used as much as possible. Four-other open-ended questions provided an
opportunity for respondents to list other details or ideas not covered in previous
questions. This was done to ensure an adequate range of ideas was captured. Closed-
ended questions varied from checking “yes” or “no” to rating some aspect on a Likert
scale (e.g., l= strongly oppose through 5= strongly support). Here again, tables were used

to reduce the length of the questionnaire as well as to facilitate responses.

41

The questionnaire was divided into six sections: information about households
and their snowmobiles, snowmobile activity, spending related to snowmobiling, opinions
about snowmobile management, support of selected policy initiatives, and information
about snowmobile safety and risk. Additionally, explicit directions with examples and
deﬁnitions were provided to fully detail the nature of the information requested. Finally,
the questionnaire was printed in green ink against a white background on 11x 17 inch
paper that was folded in half. No pretest of the questionnaire was conducted due to time
constraints.

The quality and content of the cover letters accompanying the questionnaire can
also inﬂuence response (Mangione, 1995). Separate cover letters were composed for
other snowmobilers (Appendix B) and 1996-97 convicted snowmobile law violators
(Appendix C). While this study was conducted in cooperation with the DNR, the cover
letters were written on a Michigan State University Department of Park, Recreation, and
Tourism Resources letterhead to ensure impartial responses, especially to questions
asking respondents to rate DNR management. The purpose for conducting the study, as
well as the value of the respondent’s reply, was given in the cover letter. Additionally, the
letter included the process of how their names were selected, the voluntary nature of their
participation, a guarantee of conﬁdentially, uses to be made of the information, time
necessary to complete the questionnaire, and the researcher’s contact information. The
cover letters in the follow-up survey to non-respondents were slightly modiﬁed to
highlight the importance of their response and value of their information. The other
snowmobiler’s cover letter can be found in Appendix D and the 1996-97 convicted

snowmobile law violator’s cover letter is in Appendix E.

42

Qpestionnaire Administration

The other snowmobiler’s questionnaire was sent on June 3, 1997, and the
questionnaire to 1996-97 convicted snowmobile law violators was sent on June 12, 1997.
This initial mailing was sent by ﬁrst class mail with a postage paid business reply
envelope to ensure ease of response (Mangione, 1995). All questionnaires were
individually numbered so that respondents could be eliminated from a follow-up of non-
respondents. This follow-up mailing was sent by certiﬁed mail to other snowmobilers on
July 10, 1997, and the follow-up was sent to convicted snowmobile law violators on July
17, 1997. Relying on one subsequent mailing was done because of previous success in
obtaining adequate response rates through certiﬁed mail and also because substantial
effort and cost is required in sending out reminders (Nelson & Lynch, 1994; Nelson &
Lynch, 1995). Out of those not responding to the ﬁrst and second mailing, a random sub-
sample of 495 other snowmobilers and 139 convicted snowmobile law violators was

selected for the non-respondent telephone follow-up survey.

Questionnaire Response Rate

1996-97 Convicted Snowmobile Law Violator’s Remnse Rate
Of the 983 snowmobilers cited by DNR Ofﬁcers and convicted of a safety

violation in 1997, 61 (6.2%) had an invalid address (e.g., no forwarding order, forwarding
order expired, or no such person at this address) as reported by the United States Postal
Service. Of the 922 (93.8%) with valid addresses, 367 (39.8%) returned questionnaires.
From these, three questionnaires were excluded because the respondents ﬁlled out only a
portion of the questionnaire. Furthermore, to certify that the individual who received the
citation ﬁlled out the questionnaire, the age on the citation was compared to the age on
the questionnaire. As a result of this analysis, four respondents were excluded due to
signiﬁcant inconsistencies (greater than 2 years). Consequently, there were 360 (39.0%)

convicted snowmobile law violators used in this study.

43

Other Snowmobiler’s Response Rate
Of the 3,325 trail permit purchasers sampled, 173 (5.2%) had invalid address

(e.g., no forwarding order, forwarding order expired, or no such person at this address) as
reported by the United States Postal Service. Of the 3,152 valid addresses, 1,535 (48.7%)
returned questionnaires. Of those returning questionnaires, 27 (1.8%) reported they had
sold their snowmobiles and were no longer snowmobiling. The remaining 1,508 (98.2%)
were still involved in snowmobiling. However, in order to make meaningful comparisons
between other snowmobilers and1996-97 convicted snowmobile law violators, only those
individuals who snowmobiled in Michigan during the winter of 1996-97 were included in

this study. Of the 1,508 respondents, 1,348 (89.4%) ﬁt this deﬁnition.

Survey of Non-Remnants
With any self-administered mail questionnaire, the data gathered can be subject to

bias as a result of differences between those who do and those who do not respond
(Mangione, 1995). There are widely differing opinions about what constitutes an
adequate response rate for a self-administered mail questionnaire (Tull & Hawkins,
1993). For instance Babbie (1995) contends that 50% is adequate for analysis, while
Brown and associates (1989) assert that 70% is a feasible target and thus a reasonable
response rate. In spite of these conﬂicting differences, response rates are often a function
of factors such as the complexity of the questionnaire, respondents’ interest in the subject
matter, or type of population being sampled (Crompton & Tian—Cole, 1999). These
factors may help explain the response rates attained for 1996-97 convicted snowmobile
law violators and the other snowmobiler group.

The 1996-97 convicted snowmobile law violators response rate of nearly 40% is
more subject to non-response bias. Nevertheless, this response rate is comparable to that
of other studies of violators in which a self-administered mail questionnaires were used to

collect data. For example, Shafer and associates (1972) and Andersen (1987) had

response rates of 28% and 39.3% respectively. Regardless, as with the other snowmobiler
survey, a follow-up survey of non-respondents is considered desirable to assess bias
associated with non-response.

The other snowmobiler response rate of nearly 49% is considered adequate given
the late spring timing of the mailing and the intricacy of the questionnaire. This response
rate falls in the middle of other snowmobile studies in which a self-administered mail
questionnaire was used (MN -DNR, 1970; Lanier, 1974; WA-SPRC, 1988; Robertson,
1996). Nevertheless, non-response biases may be evident and thus a survey of non-

respondents is warranted.

Administratipn of an-Resppndent Survey

A telephone survey, using a shortened version of the mail questionnaire
(Appendix F) with a sub-sample of non-respondents from both groups, was used to gather
data to determine non-response bias. Since phone numbers are not included on citations,
1996-97 convicted snowmobile law violator telephone number were obtained ﬁom
PhoneDisc, a CD-ROM database of telephone directories for midwest states, located at
the Michigan State University Library. Of the 555 (60.2%) 1996-97 convicted
snowmobile law violators who did not return a questionnaire, 85 (15.3%) were selected
for the follow-up study. However, of the 85 names, only 55 (64.7%) had listed phone
numbers that corresponded with the address. Calls to 1996-97 convicted snowmobile law
violators were made during October and November 1997.

Phone numbers for the other snowmobilers were included as part of the trail
permit information. Of the 1,617 (51.3%) non-respondents for the other snowmobilers
group, 166 (10.3%) were selected for the follow-up study. Calls to this group were made
during September and October 1997.

45

Data Analysis
The Statistical Package for the Social Sciences (SPSS) version 8.0 was used to

analyze the data. First, basic descriptive statistics are used to characterize the 1996-97
convicted snowmobile law violators and other snowmobiler groups. Logistic regression
analysis was then used to identify key variables differentiating 1996-97 convicted
snowmobile law violators from other snowmobilers and predicting respondent group

membership.

Descriptive Analysis

The data used in this research were compiled for a multi-objective study of
Michigan snowmobiling during winter 1996-97. One of the principal aims of this study
was to better understand those snowmobilers who engage in illegal and risky
snowmobiling behaviors. To recap, two distinct groups of snowmobilers were
operationally deﬁned for comparison. The ﬁrst group, labeled 1996-97 convicted
snowmobile law violators, consisted of all snowmobilers who were cited by Michigan
DNR Conservation Ofﬁcers during the winter season of 1996-97 and convicted of
violating operation or safety regulations. Conversely, snowmobilers who purchased a
Michigan Snowmobile Trail Permit between October 1, 1995 and September 30, 1996
and were not cited by Michigan DNR Conservation Ofﬁcers for a snowmobile safety
violation during the winter 1996-97 were deﬁned as other snowmobilers. While both
1996-97 convicted snowmobile law violators and other snowmobilers are part of a larger
population of winter 1996-97 Michigan snowmobilers, they are considered as mutually
exclusive groups since sampling procedures insured that no snowmobiler was included in
both groups. Given this distinction and the objective of the research to identify
differences between the 1996-97 convicted snowmobile law violators and other
snowmobilers, the groups are deemed to be independent samples so that tests of

signiﬁcance are used to accomplish the research objectives.

46

The statistical tests used to analyze differences between 1996-97 convicted
snowmobile law violators and other snowmobilers included the two sample t-test of
means for the interval or ratio data and the Mann—Whitney U test for ordinal data and the
chi-square for nominal data were also used. A signiﬁcance level of 0.05 was used to
established differences between the groups for these statistical tests. This signiﬁcance
level was chosen because it is not considered exceedingly stringent or lax, thereby
allowing general differences between the two groups to be identiﬁed.

In addition to these tests, bivariate correlation matrices were used to assess
independence of the explanatory independent variables. The Speannan’s rho (r,) was used
as a measure of independence for ordinal variables since they bear a close resemblance to
continuous interval or ratio data, in this research (Healey, 1993). For interval or ratio data
the Pearson’s r statistic was used to measure the independence of the explanatory
independent variables. Those variables with r of $0.40 were deemed to be independent.

Where r > 0.40 only one of the variables was included in the analysis.

Logistic Reggession Analysis

A variety of statistical analyses and methods have been used to identify key
variables discriminating various types of folk crimes (Gabor, 1986). Trafﬁc safety
research methodology has typically relied upon linear-based ordinary least squares (OLS)
models of regression and discriminant analysis. While these statistical techniques have
been used successfully, there are a number of constraints associated with them that limit
their useﬁrlness. For example, linear regression models require the dependent variable to
be continuous, unbounded, and measured on an interval or ratio scale (Menard, 1995).
Moreover, linear regression is extremely vulnerable to violations of OLS model
assumptions (Demaris, 1992; Menard, 1995). Discriminant analysis, on the other hand,
allows the use of dichotomous dependent variables and has fewer assumptions than linear

regression analysis. However, the researcher is faced with using only interval, ratio, or

47

dichotomous independent variables in the analysis. Furthermore, discriminant analysis is
intolerant of violations of multivariate normality and similar covariance matrices (Press
& Wilson, 1978).

Since the 1980’s however, new statistical techniques have been devised to analyze
categorical data, especially dichotomous dependent variables (Demaris, 1992;
Kleinbaum, 1994). The value of such techniques is that they require fewer assumptions
than do regression and discriminant analysis, making them more statistically robust
(Press & Wilson, 1978; Demaris, 1992; Menard, 1995).

One such statistical technique is the probit analysis. This technique is more
appropriate for designed experiments, since its ﬁrnction is to measure the relationship
between the strength of a given stimulus and the proportion of cases manifesting a certain
response to the stimulus (Norusis, 1997). Closely related to probit analysis, are a couple
of statistical techniques based upon the logit model and logistic ﬁrnction. Two such
techniques include loglinear regression analysis and logistic regression analysis. In
situations where the independent variables are categorical, the loglinear regression
analysis is best (Gabor, 1986). On the other hand, where there is a mixture of numerical
and categorical independent variables, logistic regression analysis (LRA) is particularly
appropriate (Demaris, 1992).

The logistic regression analysis, like discriminant analysis, is designed to ﬁnd the
best ﬁtting and most parsimonious model to describe the relationship between a
dichotomous dependent variable and a set of independent variables. It is also used in
classifying individuals into group associations (Press & Wilson, 1978). As a result of
these attributes, LRA is beginning to be used in distinguishing wildlife (Gray &
Kaminski, 1994) and trafﬁc law (Cooper, 1997; Meadows et al., 1998) violators from

non-violators. Thus it is adopted here as the preferred method of statistical analysis.

48

Mathematically, the logistic regression model is formulated from the logistic
function (Kleinbaum, 1994). The product derived from this function yields estimates
restricted between 0 and 1, thereby allowing a dichotomous dependent variable to be
formulated as a probability function. To use this probability in regression, it must ﬁrst be
re-conceptualized as an odds ratio to contend with the possibility that the observed values
of the probability will be less then 0 or greater then 1 (Menard, 1995). As a result of this
conversion, the dependent variable in this context is expressed as the odds or ratio of the
probability that a given snowmobiler will be a violator (Y = 1) versus the probability that
they will be a other snowmobiler (Y = 0), given a set of independent variables. This can
be rewritten as:

Odds(Y = 1) = P(Y = 1) / [1 - P(Y = 1)]
where (P) is the probability (Menard, 1995). Furthermore, transforming P(Y =1) / [1 -
P(Y =1)] with the natural logarithm, called Logit(Y), produces a value that in theory
ranges ﬁ'om +/— inﬁnity, thus allowing the dependent variable to be interpreted as a linear
model (Menard, 1995). This in turn produces the logistic regression equation of
Logit(Y) = or + [3,X, + 02X, + kak.

In logistic regression, the maximum likelihood method is used to have the greatest
probability of getting the observed results. The overall signiﬁcance statistics in logistic
regression assess how well the model ﬁts, the importance of the independent variables,
and the proportion of the cases classiﬁed correctly. Typically, a 0.50 split is used to

classify cases.

Modeling Strategy
There are two basic goals of this research. The ﬁrst is to identify those

predisposing, enabling, and reinforcing variables that best differentiate 1996-97 convicted
snowmobile law violators and other snowmobilers and secondly to examine the model’s

ability to discriminate between 1996-97 convicted snowmobile law violators and other

49

snowmobilers. In order to accomplish these goals, a properly devised modeling strategy
is imperative. Hosmer and Lemeshow (1989) outline such a strategy for selecting
variables in LRA. This basic modeling strategy is adopted here to accomplish the

research goals.

Characterizing 1996-97 Convicted Snowmobile Law Violators and Other Snowmobilers
Developing a parsimonious predictive logistic regression model begins with a
properly deﬁned procedure for selecting independent variables (Hosmer & Lemeshow,
1989; Kleinbaum, 1994). The goal of this process is to eliminate irrelevant variables and
minimize their numbers to only those that measurably add to the best ﬁtted model. This is
done to avoid biases and numerically unstable estimates (Gabor, 1986; Hosmer &
Lemeshow, 1989; Menard, 1995). Consequently, Hosmer and Lemeshow (1989) suggest
a thorough univariate analysis be conducted using contingency tables between the
dependent variable and each of the independent variables. The likelihood ratio chi-square
(LR-x2) test with k - 1 degrees of freedom was used to test the association between the
dependent variable and independent ordinal and interval/ratio variables with less than or
equal to ﬁve integers. This test statistic was chosen because the likelihood ratio chi-
square is “exactly equal to the value of the likelihood ratio test for the signiﬁcance of
coefﬁcients for k - 1 variables in a univariate logistic regression model” (Hosmer &
Lemeshow, 1989, p. 83). To test for signiﬁcance with interval or ratio variables with
more than ﬁve integers, a univariate logistic regression model was used to obtain the

likelihood ratio test (LR). Those explanatory independent variables with a signiﬁcance
level of $0.25 were included in the multivariate logistic regression model. Mickey and

Greenland (1989) have shown that a signiﬁcance level of $0.05 often fails to identify

relevant independent variables.

50

The logistic regression analysis was then applied with all cases and relevant
independent variables identiﬁed in the univariate analysis to ascertain which variables
best differentiated 1996-97 convicted snowmobile law violators from other
snowmobilers.

Considering that there were a number of potential variables identiﬁed in chapter
three and the theory behind comparative research between 1996-97 convicted
snowmobile law violators and non-violators is limited, this analysis used a backwards
likelihood-ratio stepwise variable selection approach with a 0.15 criterion entry level of
signiﬁcance (Hosmer & Lemeshow, 1989). The backwards elimination method helps
lessen the risk of excluding independent variables that may have a statistically signiﬁcant
effect in predicting the dependent variable only when another variable is included in the
model (Menard, 1995, p. 55). In such circumstances, both variables are already generally
in the model. The likelihood-ratio test (LR) involves “estimating the model with each
variable eliminated in turn and looking at the change in the likelihood-ratio when each
variable is eliminated” (Norusis, 1997, p. 53). The choice of setting the statistically
signiﬁcance criterion for entry at 0.15 is to “prevent the failure to ﬁnd a relationship
when one exists” (Menard, 1995, p. 55).

Due to similar results between the likelihood-ratio test (LR) and Wald statistic
(Wzk), the latter was used to evaluate the signiﬁcance of the individual independent
variables to the prediction of the dependent variable.

Considering that the aim of this research is to develop a parsimonious predictive
logistic regression model, any independent variables not contributing to the model were
eliminated and a new LR model calculated using the single block entry approach. The
single block approach enters all independent variables concurrently in the order in which
they emerged in the backwards stepwise logistic regression model. This second step is
conducted to verify the signiﬁcance of each of the independent variables included in the

original model. This is done by comparing the regression coefﬁcients and the Wald

51

statistics of the individual independent variables (Hosmer & Lemeshow, 1989). In cases
where the change in coefﬁcients, Wald statistics, and signiﬁcance level of the
independent variable were noticeably different, a comparison of the logistic regression
model with and without the independent variable(s) in question was conducted in order to
determine if their inclusion is statistically signiﬁcant.

The adequacy of the overall model will be assessed with the model chi-square
(Model X2). This provides a test for the null hypothesis when all the coefﬁcients in the
regression equation take on a value of zero (Menard, 1995). A 0.05 level of signiﬁcance
is used to test the model’s adequacy. In addition, the proportional reduction in the model
chi-square (R2,) will be used to evaluate “how much the inclusion of the independent
variables in the model reduce the badness-of-ﬁt of the model chi-square statistic”
(Menard, 1995, p. 22). Where R2L is 0, the independent variables are useless in predicting
the dependent variable and where it is 1, the independent variables perfectly predicts the
dependent variable.

Provided there is theoretical relevancy, the interaction between two independent
variables can also be incorporated in logistic regression analysis (Demaris, 1992; Menard,
1995). This cross-product term, which is calculated within SPSS, is computed by
multiplying the two independent variables together (Norusis, 1997). The term is then
added to the logistic regression model and new parameters computed. The change in the
model chi-square and proportional reduction in the model chi-square ﬁ'om the original LR
model to the new LR model, as well as the signiﬁcance of the variable, will be used to

assess the interaction effects of the cross-product term (Menard, 1995).

52

Logistic Reggession Diagpostics

Violations of the assumptions of logistic regression analysis can potentially
produce biased coefﬁcients, ineffective estimations of the coefﬁcients, or invalid
statistical inferences (Menard, 1995). As a general rule, tests for nonlinearity in the logit,
collinearity between independent variables, and the evaluation of residuals should be

conducted to identify any of these issues (Hosmer & Lemeshow, 1989; Menard, 1995).

Nonlinearig in the Lo git

One of the assumptions of logistic regression is that there is a linear relationship
between the natural logarithm of the dependent variable [logit(Y)] and the interval or
ratio data independent variables (X) (Hosmer & Lemeshow, 1989). That is, “if the change
in logit(Y) for a one-unit change in X is constant and does not depend on the value of X,
we say the logistic regression model has a linear form, or that the relationship is linear in
the logit, and the change in logit(Y) for a one-unit change in X is equal to the logistic
regression coefﬁcient” (Menard, 1995, p. 60). While there are a number of ways to
evaluate nonlinearity in the logit, one relatively easily procedure is the Box-Tidwell
transformation (Hosmer & Lemeshow, 1989). This transformation adds a “term of the
form X[ln(X)] to the model” (Hosmer & Lemeshow, 1989, p. 90). In this case, the term
X[ln(X)] will be added to the logistic regression model using the single block entry
approach. Where the coefﬁcient of X[ln(X)] is statistically signiﬁcant (p= $0.05) there is
an indication of nonlinearity in the logit (Menard, 1995). Although the Box-Tidwell
transformation is simple to conduct, it is not overly sensitive in identifying minor
deviations from linearity (Hosmer & Lemeshow, 1989). Nevertheless, it should be
included as part of the logistic regression diagnostics. In cases where there is evidence of
nonlinearity of a particular independent variable, the values of the variable in question

will be grouped into the ﬁve categories used in the descriptive analysis comparing 1996-

53

97 convicted snowmobile law violators and other snowmobilers. The model would then

be recomputed, testing the reconﬁgured variable for nonlinearity in the logit.

Collineariy

The issue of collinearity results from the correlation between independent
variables, which in turn can potentially yield biased logistic regression coefﬁcients and a
higher RZL. While there is no procedure for assessing collinearity among the independent
variables in the logistic regression function of SPSS, it can easily be checked using the
tolerance statistic in the linear regression function, since we. are merely interested in the
relationship among the variables and not the functional form of the model (Menard,

1995). Tolerance values less than 0.20 are an indication of collinearity (Menard, 1995).

Evaluations of Residuals

The residual is generally deﬁned as “the difference betWeen the observed
probability of the event and the predicted probability of the event based on the model”
(Norusis, 1997, p. 57). According to Menard (1995, p. 71), “the principle purpose for
which residuals analysis is used in logistic regression is to identify cases for which the
model works poorly, or cases that exert more than their share of inﬂuence on the
estimated parameter of the model.” As a general rule, the analysis of residuals should
include a review of the Studentized residual, the leverage statistic, and dbeta (Menard,
1995)

The Studentized residual is used to identify cases that may ﬁt the model poorly.
Generally, residuals less than -3 and greater than +3 should necessitate an examination of
the case. Less of a concern, but still enough to require a review of the case, are residual
values less than -2 and greater than +2. The leverage statistic values are used to indicate
cases that may be unduly inﬂuencing the parameters of the logistic regression model

(Hosmer & Lemeshow, 1989; Menard, 1995). These leverage statistic values will fall

54

between zero, meaning no inﬂuence, and one, meaning a complete inﬂuence of the
parameters. Those cases where the leverage statistic values are larger than 0.012 may be
potentially inﬂuential and thus need to be reviewed. Any case with values of 0.018 or
more should be examined closely (Menard, 1995). Lastly, the dbeta values denote
changes in the regression coefﬁcients when cases are deleted from the modeling process
(Menard, 1995; Norusis, 1997). Dbeta values greater than one indicate that the case
should be examined more closely.

Without fail, an analysis of residuals will identify problem cases. This is
especially relevant in the analysis between the 1996-97 convicted snowmobile law
violators and other snowmobilers, since the deﬁnition used to determine group
membership is deﬁnitive, but may not accurately characterize behavior. For example, a
1996-97 convicted snowmobile law violator, while characterized as an offender, may
have unwittingly violated the law and would not have done so if they were aware of the
rule. Likewise, another snowmobiler may be a willful violator who simply did not get
caught and did not receive a citation during the 1996-97 winter. Consequently, the
residuals may not necessarily indicate a problem using the legalistic deﬁnition of group
membership, but may be problematic when considering behavior or motivation for

behavior. Thus caution should be used when deciding to delete the case.

Model Discrimination
To further assess the performance of the model, the predicted and observed group
membership of the cases was examined for accuracy in distinguishing classiﬁcation of
1996-97 convicted snowmobile law violators and other snowmobilers (Norusis, 1997). In
cases where there is a large disparity between the sample size of each of the two groups,
as is here, the classiﬁcation procedure is potentially biased since most individuals will be
classiﬁed in the larger group (Morrison, 1969). Consequently, the smaller sample of the

two groups is used to determine the effective sample size for the model to classify 1996-

55

97 convicted snowmobile law violators and other snowmobilers. This provided an equal
chance for a given snowmobiler to be classiﬁed. A logistic regression analysis was then
used with the independent variables from the ﬁnal model. Model discrimination is
evaluated using a classiﬁcation table. This table indicates the “observed and predicted
group membership when cases with a predicted probability of 0.5 or greater are
classiﬁed” as violators (Norusis, 1997, p. 44). The c statistic is used to measure the
signiﬁcance of this model’s capability in distinguishing between the two groups (Norusis,
1997). According to Norusis (1997, p. 61), “it can be interpreted as the proportion of
pairs of cases with different observed outcomes in which the model results in a higher
probability for the cases with the event [being a violator] than cases without the event.”
The value of the c statistic ranges from 0.5 to 1. Values closer to 1 mean that the model
assigns higher probabilities to cases with the event [being a violator] than without
(Norusis, 1997). In addition, a histogram of the estimated probabilities of being a violator

is used to assess the probabilities of being incorrectly classiﬁed.

56

CHAPTER FIVE
DESCRIPTIVE RESULTS

This chapter includes the response and non-response comparisons as well as the
descriptive analysis for other snowmobilers and 1996-97 convicted snowmobile law
violators. In addition, bivariate correlation matrices assessing the independence of the

explanatory independent variables is also included in this chapter.

Remnse and Non-Response Comparison

Non-response bias for both the 1996-97 convicted snowmobile law violators and
other snowmobilers was assessed by means of a telephone survey using a shortened
version of the mail questionnaire (Appendix F). Calls were made to the other
snowmobiler group during September and October 1997 and during October and
November 1997 for the 1996-97 convicted snowmobile law violator group. While these
data were compared with those from the mail questionnaires to identify discrepancies, no
statistical tests of signiﬁcance were performed since responses in both groups were too

small for meaningful comparisons.

1996-97 Convicted Snowmobile Law Violator Non-Response

Of the 55 1996-97 convicted snowmobile law violator non-respondents selected
with phone numbers that corresponded with the address given on the Snowmobile Trail
Permit, 8 (14.5%) refused to participate, while 14 (25.5%) responded to the shortened
survey. The other 33 (60.0%) non-respondents where not able to be reached on ﬁve
different occasions at different times of the day because the phone number had been

disconnected or was wrong.

57

Of these 14 non-respondents contacted, the average number of household
members 12 years and older was higher (3.5 vs. 2.5 people) than those who responded to
the mail questionnaire (Table 3). However, the average proportion of the household
members who snowmobiled in Michigan during the winter of 1996-97 was 60%
compared to 83% for respondents. The percentage of those who were residents of
Michigan was similar, at a little over 80%, for both groups. The mean age of non-
respondents was 30.1 years old, which is slightly lower than respondents who averaged
34.2 years old. A possible explanation for this difference was that there were
proportionally more 26 to 35 year olds and no 55 and older individuals in the non-
respondent group compared to the respondent group.

Table 3. Household size and snowmobile activity, age, and residency of 1996-97

convicted snowmobile law violator respondents and non-respondents who snowmobiled
in Michigan during that winter.

 

 

Respondents Non-Respondents
Household and Respondent Characteristics (N= 360) (N= 14)
>‘< 2
Number of household ’ 2.5 3.5
Pr0portion of household who snowmobiled ’ 0.83 0.60
% %
Michigan resident 81.9 85.7
Age in years
525 18.8 21.4
26-35 41.9 57.1
36-45 25.6 21.4
46-55 10.7 0.0
255 3.1 0.0

2 = 34.2 (10.0) b 2 = 30.1 (6.5) b

 

a. Includes only rnmviduals 12 years an oldﬁ.
b. Standard deviations in parentheses.

58

The mean number of snowmobiles owned (1.64 vs. 2.14 snowmobiles) and used
(1.43 vs. 2.11 snowmobiles) during winter 1996-97 by the household was less for non-
respondents than respondents (Table 4). Non-respondents had considerably more
powerful engines, averaging 620 cc, than respondents, which averaged 538 cc.
Table 4. Selected characteristics of 1996-97 convicted snowmobile law violator

respondents and non-respondents household snowmobiles used in Michigan during that
winter.

 

Respondents Non-Respondents

 

Household Snowmobile Characteristics (N= 360) (N= 14)
>7 >‘<
Number of snowmobiles owned 2.14 1.64
Number snowmobiles used 2.11 1.43
Average engine size (cc) ' 537.6 620.1
Average model year ' 1992.4 1993.6
Average estimated annual miles ' 1,061.5 1,465.4
Average estimated annual machine days ‘ 18.2 28.5

 

a. Averaged based on ﬂiose snowmcﬂiiles used in Mﬂﬁgan duringﬂie 1996-9'7 wmter.

The average age of the machines was also different as non-respondents tended to have
newer machines (1994. vs. 1992) than respondents. Both the average estimated annual
miles (1,465 vs. 1,062 miles) and number of days (28.5 vs. 18.2 days) each snowmobile
was operated was much greater for non-respondents than respondents. Correspondingly,
the total number of non-respondent household snowmobiling days was considerably more

than respondents, primarily in the Upper Peninsula (Table 5) (Figure 1).

59

Table 5. Regional household snowmobiling activity of 1996-97 convicted snowmobile
law violator respondents and non-respondents in Michigan during that winter.

Respondents Non-Respondents

 

Region ' (N= 360) (N= 14)
>’< >7
Number of Upper Peninsula days 9.5 18.9
Number of Northern Lower Peninsula days 21.3 21.4
Number Southern Lower Peninsula days 0.7 0.9
Total number days 34.6 41.2

 

a. See Figure 1 for display of regions.

 

Peninsula

 

 

 

Southern Lower ,
Peninsula 1 | I

 

 

 

 

 

 

 

 

 

 

Figure 1. Three Michigan Snowmobile Regions

60

When asked about their support for selected fatality reduction initiatives, non-
respondents and respondents differed slightly (Table 6). Non-respondents were less
supportive of mandatory training for all ﬁrst year snowmobilers regardless of age than
respondents. Overall however, they were similar to violator respondents in that they were
generally less supportive of fatality reduction initiatives than snowmobilers not cited for a
safety violation in winter 1996-97.

Table 6. 1996-97 convicted snowmobiler law violator respondents and non-respondents
rating of support for selected fatality reduction initiatives.

 

 

Respondents Non-Respondents
Fatality Reduction Initiatives ' (N= 360) (N= 14)
>7 % Support " )7 % Support b

Mandatory training for all ﬁrst year

snowmobilers regardless of age 3-86 64-4 3-36 50-0
Mandatory personal insurance for all
More intensive enforcement of existing

snowmobile regulations 3’01 343 3-36 50-0
Licensing required for all snowmobilers 2.83 35.6 3.07 35.7
Enforcement of a snowmobile speed limit

on public lands, trails, and waters 2-45 24-6 2-43 21-4

 

 

amaﬁng scale: 5 = strongly supports, 4= moderately suppommoderately
oppose, 1= strongly oppose.
b. Percentage of individuals strongly or moderately supporting the initiative.

The comparison of 1996-97 convicted snowmobile law violator respondents from
the mail questionnaire and the sub-sample of non-respondents may indicate some
moderate differences. For example, non-respondents were slightly younger, owned newer
and much more powerful snowmobiles, and rode them more often than respondents.
However, the exact nature of these differences may be strongly inﬂuenced by the
relatively small number of non-respondents who completed the shortened survey.

Following is a more revealing examination of possible differences between respondents

and non-respondents based on the actual citation records.

61

Citation Data Summg and Response and Non-Remnse Comparison
The information from all 983 snowmobile operational and safety citations on ﬁle

at the DNR-LED was available for this analysis. This was compiled in order to
characterize the entire population of 1996-97 convicted snowmobile law violators and
compare those who did and did not respond to the mail questionnaire. This provided
accurate information on the age, sex, and state and region of principal residence of the
total 1996-97 convicted snowmobile law violator population and is compared to
respondents in Table 7.

The average age of all 1996-97 convicted snowmobile law violators was 33.0
years old and the mean age of respondents was similar at 34.4 years old. Considering age
ranges, those from 36-55 were slightly more likely to respond and those under 25 were
slightly less likely to respond. This result is similar to the comparison between
respondents from the mail questionnaire and the sub-sample of 1996-97 convicted

snowmobile law violator non-respondents.

62

Table 7. Age, sex, and state and region of principal residence for all 1996-97 convicted
snowmobile law violators and respondents cited by Michigan Conservation Ofﬁcers
during that winter.

All 1996-97 Violators Violator Respondents

 

Category (N= 983) (N= 360)

Age in years N % N %
$25 288 24.9 69 19.2
26-35 375 40.9 146 40.7
36-45 208 22.7 95 26.5
46-55 78 8.5 37 10.3
255 28 3.1 12 3.3

>‘<= 33.0 (10.2) ' >‘<= 34.3 (10.0) ‘

Sex
Male 945 96.3 344 95.6
Female 36 3.7 16 4.4

State of residence
Michigan 81 1 82.5 295 81.9
Out of State 172 17.5 65 18.1

Region of county residence b' ‘

Western UP 33 3.4 9 2.5
Eastern UP 46 4.7 12 3.4
Northwestern LP 91 9.3 32 8.9
Northeastern LP 62 6.4 17 4.7
Southern LP 579 58.9 223 62.3
Out of state 172 17.5 65 18.1

 

a. Standard deviations in parentheses.
b. See Figure 2 for display of regions.
c. For Michigan residents.

63

There were also few differences between the total population of violators and

respondents concerning gender and state and region of origin (Figure 2).

 

 

 

 

 

 

 

 

 

 

 

 

Southern LP f i ”—1-

Alll

Figure 2. Five Michigan Snowmobile Regions

 

 

 

 

 

 

 

 

 

Only minor differences were found between the population of all 1996-97
convicted snowmobile law violators and respondents concerning the month, day of week,
time of day, and region in which the citation was issued (Table 8). Slightly more than
95% of citations issued to 1996-97 convicted snowmobile law violators occurred between
December and February, with January accounting for more than 40% of the citations.
Nearly all citations issued during 1996-97 were given on the weekend between 12:00 PM

and 1:00 AM. The largest number of citations was issued in the northwestern Lower

Table 8. Month, day of week, time of day, and region citations were issued to all 1996-97
convicted snowmobile law violators and respondents cited by Michigan Conservation
Ofﬁcers during that winter.

 

All 1996-97 Violators Violator Respondents

 

Category (N= 983) (N= 360)
Month N % N %
October 1 0.1 0 0.0
November 13 1.3 4 1.1
December 253 25.7 98 27.2
January 401 40.8 152 42.2
February 287 29.2 100 27.8
March 27 2.7 5 1.4
April 1 0.1 1 0.3
Day of week
Sunday 207 21.1 67 18.7
Monday 20 2.0 10 2.8
Tuesday 27 2.7 10 2.8
Wednesday 10 1.0 6 1.7
Thursday 32 3.3 12 3.3
Friday 145 14.8 58 16.2
Saturday 541 55.1 196 54.6
Time of day
1:00AMto 5:59AM 10 1.0 3 0.8
6:00 AM to 11:59 AM 63 6.4 24 6.7
12:00 PM to 5:59 PM 488 49.9 183 51.1
6:00 PM to 12:59 AM 416 42.6 148 41.3
Region citation issued '
Western UP 121 12.3 51 14.2
Eastern UP 199 20.3 73 20.3
Northwestern LP 312 31.8 110 30.6
Northeastern LP 251 25.6 90 25.0
Southern LP 99 10.1 36 10.0

 

a. See Figure 2 for display of regions.

65

Peninsula (31.8%), followed by the northeastern Lower Peninsula (25.6%), and the
eastern Upper Peninsula (20.3%) (Figure 2). The larger proportion of citations issued in
the northwestern Lower Peninsula may be a function of the previously described DNR-
LED intensive patrol initiative during winter 1996-97.

Of all the citations written in 1996—97 by Conservation Ofﬁcers, 904 (92%) were
single offenses, while 70 (7.1%) were for two offenses and 9 (0.9%) were for three (Table
9). The proportions of the number of offenses written for respondents was similar to those
of all 1996-97 violators. Slightly more than half of the citations written during 1996-97
were for operating on a public roadway, while 14.4% were for other roadway related
offenses, such as operating against the ﬂow of trafﬁc, failing to operate in a single ﬁle
order, or failing to be at the extreme right of the roadway going with the ﬂow of trafﬁc.
Of the 79 (8.0%) snowmobiles cited for more than one offense, most citations related to
riding on public roadways, operating against flow of trafﬁc, not operating in a single ﬁle,
or not operating to the extreme right of the roadway, coupled with registration problems
such as no proof of registration or improper placement or display of snowmobile
registration. Respondent citations were pr0portionally similar to all 1996-97 convicted
snowmobile law violators in the type of violation.

Along with receiving a citation, 8.9% of the 1996-97 convicted snowmobile law
violator population were warned about other offenses (Table 10). The frequency and type
of warnings received by respondents were in similar proportion to all 1996-97 convicted
snowmobile law violators. Based upon this extensive review of citation data comparing
respondent violators to all 1996-97 convicted snowmobile law violators, no signiﬁcant

non-response bias is noted.

66

Table 9. Number and type of moving violations issued to 1996-97 convicted snowmobile
law violators and respondents cited by Michigan Conservation Ofﬁcers during that
winter.

 

 

All 1996-97 Violator
Violators Respondents
Number and Type of Violation Issued (N= 983) (N= 360)
Number of violations N % N %
One 904 92.0 330 91.7
Two 70 7.1 26 7.2
Three 9 0.9 4 1.1
Type of violation ‘
Operation on public roadway 516 52.5 188 52.2

Operation against ﬂow of trafﬁc, not in a

single ﬁle, at extreme right/left 142 14-4 61 17-0

Operator or passenger without helmet 70 7.1 24 6.7
Operation on railroad right of way 63 6.4 25 6.9
Operation too fast for conditions 54 5.5 19 5.3

Failure to stop for stop sign, road,
crossing or railroad crossing

Minor operating snowmobile w/o supervision,

50 5.1 21 5.8

safety certiﬁcate, or crossing public roadway 16 1'6 4 1’1
Operating under inﬂuence of intoxicating liquor 16 1.6 4 1.]
Operation within 100 ft of person/ice shanty 12 1.2 1 0.3
Operation in forest reproduction or 11 1 1 4 1.1

restricted area, prohibited area trespass '

Operation without head light or tail light 7 0.7 2 0.6
Fleeing and eluding, not stopping for

ofﬁcer’s signal 8 0'8 2 0'6
Operation on airport property 6 0.6 0 0.0
Operation with improper mufﬂer 4 0.4 2 0.6
Pursuing or harassing wildlife 3 0.3 1 0.3
Operation during closed hours or ﬁrearm season 2 0.2 1 0.3
Operation without event permit 2 0.2 1 0.3
Failure to report an accident 1 0.1 0 0.0

 

a. Where there were multiple Violations, the ﬁrst Violation listed was used.

67

F"II

Table 10. Number and type warnings above and beyond the citation issued to all 1996-97

convicted snowmobile law violators and respondents cited by Michigan Conservation
Ofﬁcers during that winter.

 

 

All 1996-97 Violator
Violators Respondents
Number and Type of Warnings Issued (N= 983) (N= 360)
Number of warnings N % N %
None 896 91.1 323 89.7
One 60 6.1 27 7.5
Two 27 2.7 10 2.8
Type of warning '
No proof of registration, improper '
placement or display of registration 32 36'8 12 32'4
Expired or no trail permit or failure to
display 17 19.5 5 13.5
Operation against ﬂow of trafﬁc, not in a
single ﬁle, at extreme right/left 13 14'9 5 13.5
Operation on public roadway 8 9.2 6 16.2
Operation too fast for conditions 7 8.0 4 10.8
Operator or passenger without helmet 3 3.4 1 2.7
Operation within 100 ft of person/ice shanty 3 3.4 1 2.7
Failure to stop for stop sign, road crossing, 2 2 3 1 2 7

or railroad crossing

Operation in forest reproduction or 1 1 1 1 2 7
restricted area, prohibited area trespass ' '

a. Where there were multiple warnings, the ﬁrst listed was used.

68

Other Snowmobiler Non-Response

Of the 166 other snowmobiler non-respondents selected, contact was made with
70 (42.1%). The 96 (57.9%) non-respondents not contacted were either unavailable on
ﬁve different occasions, the phone number had been disconnected, or it was a wrong
phone number. This inability to contact non-respondents suggests that many individuals
may not reside at the street address provided with the Snowmobile Trail Permit.
Consequently, some mail questionnaires may not have reached the targeted individual,
even though the U. S. Postal Service did not return them as undeliverable.

Of the 70 contacted, 18 (25.7%) refused to participate, 5 (9.6%) had sold their
snowmobiles and stopped snowmobiling, and 47 (67.4%) responded to the shortened
survey. The percentage of non-respondents that had sold their snowmobiles was
considerably greater (9.6% vs. 1.8%) than those who responded to the mail questionnaire.
This suggests that some of the non-response may be related to those who have stopped
snowmobiling.

Of the 47 non-respondents still involved in snowmobiling, the average number of
household members 12 years and older was higher (3.26 vs. 2.49 pe0p1e) than
respondents to the mail questionnaire (Table 11). However, these non-respondent
households do not appear to be as active in snowmobiling, since the proportion who
snowmobiled in Michigan during the winter of 1996-97 was slightly lower (71% vs.
84%) than respondents. The percentage of non-respondents who were residents of
Michigan was 78.7%, which is greater than for respondents, as 63.3% were residents of
the state. Despite the similarity in average age between these two groups (40.0 vs. 39.1
years old), there were more 36 to 45 year olds and fewer 46 year olds and older non-

respondents than respondents.

69

Table 11. Household size and snowmobile activity, age, and residency of other
snowmobiler respondents and non-respondents who snowmobiled in Michigan during
winter 1996-97.

Respondents Non-Respondents

 

Household and Respondent Characteristics (N= 1,348) (N= 47)
>7 >7
Number of household ' 2.49 3.26
Proportion of household who snowmobiled ' 0.84 0.71
% %
Michigan resident 63.3 78.7
Age in years
£25 8.1 8.5
26-35 32.3 31.9
36-45 30.5 40.4
46-55 18.4 10.6
255 10.8 8.5

>2: 40.0 (11.7) " >7= 39.1(11-4)b

 

a Ecludes oﬁly indmduals 12 years an older.
b. Standard deviations in parentheses.

The mean number of snowmobiles per household owned and used during
Michigan’s 1996-97 winter was slightly less for non-respondents (1.96 vs. 2.11
snowmobiles) than respondents (Table 12). The non-respondent’s household
snowmobiles were on average slightly less powerful (506 vs. 525 cc) and older (1991 vs.
1992) than respondents. While the average estimated annual miles each snowmobile was
operated was slightly greater for respondents than non-respondents (814 vs. 788 miles),
predictably the average estimated number of days each snowmobile was used was almost
10 more days for non-respondents due to the higher proportion of Michigan residents.
Correspondingly, the mean number of household snowmobiling days for non-respondents
was considerably more at 40.7 days than respondents, who rode their snowmobiles for an

average of 23.5 days (Table 13).

7O

Table 12. Selected characteristics of other snowmobiler respondents and non-respondents
household snowmobiles used in Michigan during winter 1996-97.

Respondents Non-Respondents

 

Household Snowmobile Characteristics (N= 1,348) (N= 47)
>7 >7

Number of snowmobiles owned 2.11 1.96
Number snowmobiles used 1.97 1.83
Average engine size (cc) ' 524.8 505.8
Average model year ' 1992.2 1991.1
Average estimated annual miles ' 814.3 788.5
Average estimated annual days ' 12.6 22.3

a. Averaga 538a on ﬁfOSC SHOWIIIOBIICS 118a 1n “10138311 during ETC 1993-g, Wlntel'.

Table 13. Regional household snowmobiling activity of other snowmobiler respondents
and non-respondents in Michigan during 1996-97 winter.

Respondents Non-Respondents

 

Region ' (N= 1,348) (N= 47)
>7 5?
Number of Upper Peninsula days 9.1 7.5
Number of Northern Lower Peninsula days 12.1 30.3
Number of Southern Lower Peninsula days 2.4 3.1
Total number of days 23.5 40.7

 

 

a See Figure I for display of regons.

When asked about their support for selected fatality reduction initiatives, non-
respondents and respondents were similar (Table 14). Both groups were most supportive
of mandatory training for all ﬁrst year snowmobilers regardless of age, and less
supportive of an enforced snowmobile speed limit on public lands, trails, and water. Of
all the similar mean ratings, the percentage strongly or moderately supporting the

initiatives were less for the non-respondents in three of the ﬁve initiatives.

71

Table 14. Other snowmobiler respondents and non-respondents rating of support for
selected fatality reduction initiatives.

Respondents Non-Respondents
Fatality Reduction Initiatives ' (N= 1,348) (N= 47)

 

>2 % Support b 2 % Support b
Mandatory training for all ﬁrst year

 

 

snowmobilers regardless of age 4-03 72-7- 3-71 66-6
More intensive enforcement of existing

snowmobile regulations 3.69 59.1 3.38 53.3
Mandatory personal insurance for all

snowmobilers 3.66 59.4 3.18 46.7
Licensing required for all snowmobilers 3.06 40.7 3.00 44. 5
Enforcement of a snowmobile speed limit

on public lands, trails, and waters 299 41-4 2-98 42-2
a. Rating scale: 5= strongly supports, 4= moderately supports, 3= neum, 2= moderately

oppose, 1= strongly oppose.
b. Percentage of individuals strongly or moderately supporting the initiative.

While differences exist between the other snowmobiler respondents of the mail
questionnaire and the sub-sample of non-respondents, they are minor. Furthermore, based
upon the number of disconnected and wrong phone numbers, it appears that many of the
non-respondents may have moved even though the questionnaires were not returned as
undeliverable or the forwarding order expired by the US. Postal Service. Consequently,
the response rate may be higher than it appears. In total, non-response bias does not

appear to adversely affect the other snowmobiler data in this project.
1996—97 Convicted Snowmobile Law Violator and Other Snowmobiler Comparison

Household and Respondent Characteristics
The total number of individuals 12 years old and older in households of 1996-97

convicted snowmobile law violators and other snowmobilers was comparable, averaging
2.5 people (Table 15). Likewise, the proportion of the household that snowmobiled in
Michigan during the 1996-97 winter was similar between these groups. While the age

range between the two groups was similar, proportionally, 14 to 35 year olds accounted

72

for 61% of the 1996-97 convicted snowmobile law violators, while they only made up

40% of the other snowmobiler group. Consequently, the mean age of 1996-97 convicted

snowmobile law violators was statistically signiﬁcantly younger at 34.2 years old than

other snowmobilers at 40.0 years old (F -8.61 l, p= 0.000).

Table 15. Household size, snowmobiling activity, and respondent age for 1996-97
convicted snowmobile law violators and other snowmobilers who snowmobiled in

Michigan during winter 1996-97.

 

 

1996-97 Violators Others
Household and Respondent Characteristics (N= 983) (N= 1,348)
Number of household ‘ N % N %
One 52 14.4 162 12.1
Two 172 47.8 709 52.8
Three 71 19.7 226 16.8
Four 44 12.2 178 13.3
Five or more 21 5.9 68 5.1
>7= 2.5 (1.2)b >7= 2.5(l.1)b
Proportion of household who snowmobiled ‘
525% 15 4.2 23 1.7
26% to 50% 70 19.4 318 23.7
51% to 75% 34 9.4 99 7.4
276% 241 66.9 903 67.2

Age in years *
$25
26-35
36-45
46-55
255

>-<= 0.83, (0.26) b

67 18.8
149 41.9
91 25.6
38 10.7
11 3.1

>2: 34.2 (10.0) b

52= 0.84 (0.24) b

107 8.1

429 32.3
405 30.5
244 18.4
143 10.8

>2: 40.0 (11.7) b

 

3. Includes orﬂy individuals 12 years an older.
b. Standard deviations in parentheses.

"' Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance.

73

State and Region of Principal Residence
The state of principal residence was statistically signiﬁcantly different between

the groups (Z= —6.700, p= 0.000), as 80% of 1996-97 convicted snowmobile law violators
were Michigan residents compared to 63% for the other snowmobilers (Table 16). A
statistically signiﬁcant difference between the two groups was also found for the region
of county residence in Michigan (Z= -4.789, p= 0.000). Sixty-two percent of the 1996-97
convicted snowmobile law violators were from the southern Lower Peninsula, compared
to 42% of the other snowmobilers.

Table 16. State and region of principal residence for 1996-97 convicted snowmobile law
violators and other snowmobilers who snowmobiled in Michigan during winter 1996-97.

 

 

1996-97 Violators Others
Principal Residence (N= 360) (N= 1,348)
State of residence * N % N %
Michigan 295 81.9 853 63.3
Out of state 65 18.1 495 36.7
Region of county residence " *
Western UP 9 2.5 59 4.6
Eastern UP 12 3.4 28 2.2
Northwestern LP 32 8.9 120 9.3
Northeastern LP 17 4.7 54 4.2
Southern LP 223 62.3 538 41.6
Out of state 65 18.2 495 38.3

 

a. See Figure2 for display of regions.
* Signiﬁcant difference using the Mann-Whitney U test at 0.05 level of signiﬁcance.

74

Household Snowmobiles and Use

Snowmobile ownership, characteristics, and use are shown in Tables 15 through
18. On average, both 1996-97 convicted snowmobile law violator and other snowmobiler
households owned slightly more than 2 snowmobiles (Table 17). Proportionally though,
14.5% of 1996-97 convicted snowmobile law violator households owned four or more
snowmobiles, compared to 10.9% of other snowmobiler households. While similar in
ownership patterns, the mean number of household snowmobiles used during Michigan’s
1996-97 winter was signiﬁcantly greater (t= 2.239, p= 0.025) for violators, with 2.11
household snowmobiles used versus 1.98 for other snowmobilers. Average engine size of
household snowmobiles used during winter 1996-97 was also signiﬁcantly higher for
convicted snowmobile law violators (538 cc vs. 525 cc) than the other snowmobilers (t=
2.107, p= 0.035) (Table 18). There is no statistically signiﬁcant difference in the age of
snowmobiles owned by 1996-97 convicted snowmobile law violators versus other
snowmobilers, with the mean snowmobile being ﬁve years old.

Table 17. Snowmobile ownership and use in Michigan by 1996-97 convicted snowmobile
law violators and other snowmobilers during winter 1996-97.

 

 

1996-97 Violators Others
Household Snowmobiles (N= 360) (N= 1,348)
Number of owned N % N %
One 104 29.6 380 28.4
Two 145 41.3 599 44.8
Three 51 14.5 213 15.9
Four 34 9.7 104 7.8
Five 17 4.8 41 3.1
>7= 2.2 (1.1)'l >7= 2.1 (1.0)‘1
Mean number used b" >2: 2.11 (1.06) ‘ >-<= 1.98 (0.95) *

 

a. Standard deviations in parentheses. .
b. Averaged based on those snowmobiles used in Michigan during the 1996-97 wrnter
* Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance.

75

Table 18. Selected characteristics of 1996-97 convicted snowmobile law violators and
other snowmobilers who snowmobiled in Michigan during winter 1996-97.

 

 

1996-97 Violators Others
Household Snowmobiles (N= 360) (N= 1,348)
Average engine size (cc) " * N % N %
$450 70 20.5 309 24.4
451-500 71 20.8 259 20.5
501-550 54 15.8 210 16.6
551-600 84 24.6 302 23.9
2601 62 18.2 184 14.6

>7= 537.6 (104.5) b >7= 524.8 (98.7) b
Average model year '

 

$1980 17 5.0 52 3.9
1981-85 22 6.4 95 7.2
1986-90 51 14.9 202 15.2
1991-93 40 11.7 223 16.8
1993-97 212 62.0 755 56.9

>7= 1992.4 (5.4) b 52: 1992.2 (5.0) b
a. Averaged based on those snowmobiles used in Nﬁchigan during the W

b. Standard deviations in parentheses.
* Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance

76

One of the key statistical differences between 1996-97 convicted snowmobile law
violators and other snowmobilers was the average annual mileage (t= 5.863, p= 0.000)
and number of days (t= 6.268, p= 0.000) household snowmobiles were used during
Michigan’s 1996-97 winter (Table 19). In terms of mileage, the snowmobiles owned by
1996-97 convicted snowmobile law violators were ridden approximately 20% more miles
and used for approximately 30% more days than the snowmobiles owned by other
snowmobilers.

Table 19. Average estimated household miles and days of snowmobiling for 1996-97

convicted snowmobile law violators and other snowmobilers in Michigan during winter
1996-97.

 

 

1996-97 Violators Others
Household Snowmobile Use (N= 360) (N= 1,348)
Average estimated miles ‘” * N % N %
$500 100 28.7 579 44.0
501-1,000 101 28.9 388 29.5
1,001-2,000 113 32.4 285 21.7
2,001-3,000 30 8.6 48 3.7
23,000 5 1.4 15 1.1
>7= 1,061.5 (732.9) b >7= 814.3 (691.4) b '
Average estimated days " *
$5 62 17.8 459 34.9
6-10 82 23.5 348 26.5
11-20 109 31.2 305 23.2
21-30 40 11.5 104 7.9
231 56 16.0 99 7.5
>7= 17.9 (17.6) " >7= 12.5 (13.5) b

 

a. Averaged based on those snowmobiles used in Michigan during winter 1996-9'7.
b. Standard deviations in parentheses.
* Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance.

77

Regionally, 1996-97 convicted law violator snowmobile activity was greatest in
the northwestern Lower Peninsula, averaging 14.4 days per household or 41.4% of the
days by households (Table 20). This was nearly twice the mean number of days per
household by other snowmobilers in that same region. The northwestern Lower Peninsula
includes the Kalkaska, Traverse City, and Cadillac areas. This was the region where the
most intensive DNR-LED enforcement activity occurred during the winter of 1996-97
and may have had an inﬂuence on this skewed proportion of 1996—97 convicted
snowmobiler law violator snowmobiling days to other snowmobilers. Statistically, 1996-
97 convicted snowmobile law violator household snowmobile days were signiﬁcantly
greater statewide and across each region except the western Upper Peninsula (Figure 2).

Table 20. Regional household snowmobiling activity for 1996-97 convicted snowmobile
law violators and other snowmobilers in Michigan during winter 1996-97.

 

 

1996-97 Violators Others
Regional Snowmobiling Days (N= 360) (N= 1,348)
Region” )7 (s) ‘ % >7 (3) ' %
Western UP 3.5 10.7 10.2 4.9 12.4 20.9
Eastern UP "' 6.0 12.0 17.4 4.1 10.0 17.6
Northwestern LP * 14.4 23.1 41.6 7 .8 16.9 33.1
Northeastern LP "' 6.9 17.7 20.0 4.3 11.6 18.3
Southern LP * 3.7 9.7 10.8 0.29 3.7 10.1
Statewide * 34.6 34.4 100.0 23.5 25.4 100.0

 

a. Ttandard deviations.
b. See Figure 2 for display of regions.
"' Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance.

78

Rating of Dangerous Snowmobiling Behaviors and Situations

To better understand snowmobilers’ perceptions about the dangers of
snowmobiling, respondents were asked to rate the danger of a variety of behaviors and
situations (Table 21). Both 1996-97 convicted snowmobile law violators and other
snowmobilers considered the operation of snowmobiles by intoxicated persons and
snowmobilers lacking skill in operating their machines as the two most dangerous
situations. Conversely, both thought that public trail conditions and snowmobiling on
county or state roads were least dangerous. The primary difference between the two
groups was that 1996—97 convicted snowmobile law violators perceived that the speed of
snowmobiles and the operation of snowmobiles by persons who have been drinking, but
are not intoxicated, as less dangerous than other snowmobilers. Overall, 1996-97
convicted snowmobile law violators differed signiﬁcantly from other snowmobilers
concerning the perceived risks of the operation of snowmobiles by intoxicated
snowmobilers (Z= -2.349, p= 0.019); the operation of snowmobiles by person who have
been drinking, but are not intoxicated (Z= -2.695, p= 0.007); the speed of snowmobiles
(Z= -3.358, p= 0.001); public trail conditions (Z= -3.507, p= 0.000); and snowmobiling
on county or state roads (Z= -2.756, p= 0.006).

Rating of Fatality Reduction Initiatives
Respondents were asked about their support for ﬁve selected fatality reduction

initiatives (Table 22). There was signiﬁcant statistical difference on initiatives between
groups because 1996-97 convicted snowmobile law violators were less supportive than
other snowmobilers of any initiative. Overall, both groups were most supportive of
mandatory training for all ﬁrst year snowmobilers regardless of age and least supportive

of enforcing a snowmobile speed limit on public lands, trails, and waters.

79

ZN
m.mm
wAN
_.N_
wd
5.:
wd
w.—
wd

“OZ

mom
5mm
adm
_.m~
WE
QNN
3N
ﬁx
5N
EEG

Qmm
9mm
ﬁg.
NSN
93
6.2
5:”
—.o~

m6

0. 58006.2. _. AME 01%

wé
We
56
md—
93
#6—
mg.—
mHN

ad—

ad

QN

m6.
Qw—
ﬁwm
VAN
Q:
WNV

ﬁns

docs—maﬁa mo .00. mod 00 .000 D 05232802 05 wEm: 0050.230 Emu—mama _.

mNm
wan
QB
ﬁg
”2
92
Q:
5;
ﬁ—
am

mdm
odm
odm
Qom
mdm
0.:
mdm
cg.

md

ﬁw—
NNN
«in
v.3
mNm
od—
_.mN
Know

NA:

3%

ad

w.w

ad—
o6.—
v.2
WE
v.0—
mSN
wd—
ea.

oé

mé
ad—
@6—
NH—
ﬁmm
N.m~
0.9.

ode

08.0.5. xlm

 

msgmga 00 mass— o\o

.. .0? 890E058 so wﬁznogoem
.008 :8 233
.. 0:206:00 :«b 0:95
... 00.030225 «o: 0.8 :5
.96—5.6 :009 0%: 055 28003
.3 02505308 mo eon—00090
... 02508305 Mo 0025
E... 0-x .0582. 08 0.3
0:05 2305308 .3 000: 0050
0000.. 35800 no 0.0.0.5 8 €00
005508 005
wczﬁao E :30 mew—00— 0002.5

.. 0:803 00830805
.3 00.308308 no 5300090

 

34m; nzv
.850

88 170

0.83—05 8-33

.3603 00053 9:26 camiomz 5 002508305 0050
0:0 820—03 32 2308305 0000350 3-82 .3 055035 23 083309 32505308 008200 80h Sweat .«o waved AN 030p.

eozmamm co uom>0n0m

80

0080050080 00 .0>0. 3.0 00 0000 D 00853-8002 05 90.80 000000-0000 80058w5 ...

 

... 80003
08 .820 000a.
2320 8 00a:

 

06m 06. 0.0. wd. 0.0m v.2“ m.w. m0. 0.0. Em.

.0000 0.308305
0 ..0 8080000080“

. . . . ._ 80.308305 ..0
. mm o m. 0 mm m m. N.m~ wsm v.2 N.N~ 0.0. 0.0. 00.. 0008.00 80.8000...

.. 000000.800

0.5083000

Wm ms Nana m.m~ 0.0-m 0.x. 1N. v.00 0.x. m0. ..0 808080.80
028008. 000$.

_. 80.508308

3.. 0... 0.0. 0.0. Wmv 0.8 0.0 wém 0.0. .6». ..0 00.0 00:00:08
.8082. 8000.803.

_.. 0w0 .00 80.00090

80.508305

m0 5m Ow. m.- 0.0.0 0.x 0.0 m0. Nw. N00 0000 08m ..0 00..
88.08 0.080.802

Mg and mam-.4 wad—0'0 a.“ a a 3 maﬁa mama
0,80. 030%. 0000402 0% 3 000.0% 0420.00.12 maulm
80.00”. .x.

.83 "z. 89“ "z. 82003
8050 8000.0; 00-000. 00.00200”. N3:000".

 

80-000. 008.? M0086 00030.82 8 80.508305
0050 .80 8000.03 30. 0.308308 0000.800 00-000. 3 00300.08. 00.000000 00:80.0 00000.00 00.. 0000000 ..0 M053. .mm 0.00..-

81

Household Snowmobiling Expenditures and Safeg Education
As indicated in Table 23, 1996-97 convicted snowmobile law violators spent

signiﬁcantly more (t= 4.043, p= 0.000), an average of $5,456 for winter 1996-97, on
snowmobiling equipment, maintenance, insurance, and storage versus other
snowmobilers who spent an average of $3,752. While the proportion of the household 12
years and older who had completed a snowmobiling safety class was slightly less for
1996-97 convicted snowmobile law violators than for other snowmobilers (14% vs.
16.9%), the difference was not statistically signiﬁcant (t= -l .570, p= 0.117).

Table 23. Household expenditures for snowmobiling and participation in snowmobile

safety training for 1996-97 convicted snowmobile law violators and other snowmobilers
in Michigan during winter 1996-97.

 

 

1996-97 Violators Others
Household Spending and Safety Class (N= 360) (N= 1,348)
Annual expenditures on snowmobiling " * N % N %
$5500 80 23.4 487 36.2
$501 to $2,000 56 16.4 266 19.7
$2,001 to $5,000 55 16.1 163 12.1
$5,001 to $10,000 105 30.7 300 22.3
2$10,001 46 13.5 131 9.7

>‘<= $5,456 (10,472) ° >7= $3,752 (5,742) °
Proportion of household previously
completing a safety class "

None 275 76.4 956 71.2
325% 7 1.9 42 3.1
26% to 50% 44 12.2 205 15.3
51% to 75% 13 3.6 44 3.3
276% 21 5.8 96 7.1

>7= 14.2% (28.5) ° >7= 16.9% (30.1) °
a. Includes equipment, maintenance, insurance, and storage expenses.
b. Includes only individuals 12 years an older.
c. Standard deviations in parentheses.
"' Signiﬁcant difference using the independent-samples t-test at 0.05 level of signiﬁcance.

82

Perceptions of Snowmobile Accident Factors

Of the 1996-97 convicted snowmobile law violators, 193 (53.6%) knew of a
winter 1996-97 personal injury or fatal snowmobiling accident as did 707 (52.4%) of the
other snowmobilers. When asked the circumstances surrounding the accidents, 1996-97
convicted snowmobile law violators and other snowmobilers had widely differing views
(Table 24). For example, 1996-97 convicted snowmobile law violators cited lack of
operator skill, trail conditions, and visibility as the top three factors that contributed to the
accident. This contrasted with other snowmobilers who cited speed, alcohol, and
operator’s lack of skill as the main factors of the accident. With the exception of
visibility, the proportion citing each of these factors was signiﬁcantly different between
the two groups.
Table 24. Assessment of selected factors contributing to a personal injury or fatal

snowmobiling accident with which 1996-97 convicted snowmobile law violators or other
snowmobilers were familiar in Michigan during winter 1996-97.

 

 

1996-97 Violators Others

Factor (N= 193) (N= 707)

N % ' N % ‘
Lack of driving skill * 104 53.9 293 41.4
Trail conditions "' 82 42.5 _ 160 22.6
Visibility 78 40.4 268 37.9
Speed * 69 35.8 462 65.3
Alcohol * 67 34.7 338 47.8
Ice conditions 44 22.8 148 20.9
Snowmobile on roadway 28 14.5 130 18.4
Fatigue 19 9.8 46 6.5
Non-snowmobilers on trail 11 5.7 43 6.1

 

a. Multiple factors were often perceived to be involved, hence percent coTumns add to
more than 100.0%.
"' Signiﬁcant difference using the chi-square test at 0.05 level of signiﬁcance.

83

Assessment of Independence Between Independent Variables

Bivariate correlation matrices, using the Speannan’s rho (r,) and Pearson’s r tests
of association, were computed to assess the independence of the explanatory independent
variables.

Four pairs of variables were correlated (r,>0.40) for the ratings of support for
selected fatality reduction initiatives and the ratings of the danger of snowmobiling
behaviors and situations (Gray & Kaminski, 1994) (Tables 25 - 27). Of these variables,
rated support for the creation and enforcement of a snowmobile speed limit on public
lands, trails, and waters correlated with the variable measuring support for more intensive
enforcement of existing snowmobile regulations (r,= 0.487, p= 0.000) (Table 25) and the
rated danger associated with the speed of snowmobiles (r,= 0.645, p= 0.000) (Table 27).
Consequently, rated support for enforcement of a snowmobile speed limit was eliminated
in favor of the other two variables. This was because the creation and enforcement of a
speed limit can be considered analogous to more intensive law enforcement, while the
danger associated with the speed of snowmobiles is an important factor in determining
the snowmobiler’s perceptions of the safety of snowmobiling. The third set of correlated
variables (r,>0.40) included the danger associated with the operation of snowmobiles by
intoxicated persons and the operation of snowmobiles by persons who have been
drinking, but are not intoxicated (r,= 0.514, p= 0.000) (Table 26). Since it is widely
known that operating a snowmobile when intoxicated is dangerous, the variable
measuring the danger of operating a snowmobile by persons who have been drinking is
included in the logistic regression analysis. The last pairs of correlated variables (r,>0.40)
pertained to public trail conditions and trail design (r,= 0.565, p= 0.000). The rating of
trail conditions was selected for inclusion in the analysis since it is inﬂuenced by design

and other factors.

84

.8305 803.2. .23 A 0. 0500208 2..
88.0.0800 ..0 .00. 000 0.008000% .0

 

000.03 .0 .0000. .0000. 000:0 =0 .08..

 

ooo.. 8.. 00.3 mmmo o0. .o 00o.o ..8... 000mm 00000 0.5083000 0 ..0 8080000..:m.
. . . . 000000800 0.3083080
ooo . v0~ o 0mm o 0.~ o 808000005. 30.0000 ..0 808080.80 03000.8 000.).
80.. 3.3 800 8:8... 002.8838 =0 0o. 02.002 8.88...
. . 8000083000
ooo . 0mm o 00:00:08. ..0 00-. 00:00:08 .000800 000.005.).
. 0w0 ..0 000.000w00 82.008380
ooo . 0000003. 0000 .80 ..0 00.. 80.800. 000.0000:
..8... 0025 8080000000.. 00:00... 00:00:00. 0000003.. 0.00.00> 000.008. 0000003. 0.00.00.

 

. 0080800 82.008380 000.0 0000 80.0.03 30. 0.5083000 00.0.88

00-000. 00.. 003.008. 00000000 0.00.0. 00.00.00 00.. .000000 8.00.0008 00.00.00> 800000008 8003.00 0000.00.80 0.0.0.35 .mm 0.00 .0

85

.00.00.00> 00003000 803 A 0. 00.00.0800 0....
80000000000 .00 0000 000 0.8050000m .0

 

. . . 00000 00000
ooo . 0.N o mm. o wmmo wwmo 0w..o 0v~.o NN..o .m..o 003000.”. 00 38000 000 w:...00830=m
ooo. .1. m0m.o .36 mod oNNo ommo mm . .o m.o.o 000.000. 000.000 ..000 000000.
ooo. wood- 08o wmmo ammo .0. .o .voo- 0000.0.0000 000000.008 ..000 000?.

0000000000. 00:

000 000 £80.80 .0000

ooo.. homo 08o mnoo o. ..o I. 00. 0o 90.0.0000. 0>00 003 00000000 00
0000083000 .00 00000000
ooo. wvoo- omoo .moo «so 000% 00.000830000 .00 000%

. . . . .0500 0-x 0.00 0.0000
80: 000 o 002 0000 00800050 2.853800088 0200

o 0 o mge
ooo . 00.. o who o 00.0.00> .0000000 :0 00.0000 00 0000
. . 0080.008 0.000 $500000
ooo . o: o :06 8 :30 $00.00. 000309

. 00000000 000000808 00
ooo . 00000000000. 00.308305 ..0 80.000000

86

 

003003. 08.000. 0000.00.00 080.800. 000% 00003 00.0.00> :05 0000000000. 0.00.00> 5.003%
00000 00 00.>000m. 0:000w0000.

 

. 000.0800 80.308308 00000 0000 00000.00, 30. 2.0083000 00000800 .0000. 00.. 00300.08.
0000000000 0000 000000000 0800083000 00.00.00 800.0 00300 w80=0008 00.00.00.» 800000.000. 000300 00.00.0008 000000005 .00 0.00 ...

02.05., 0828 .80 A 0 802208 i.

003005 00 0800002. 80 00 as 00 8300 80 ..0
80.00.8000 ..0 0000 000 0000800005 .0

 

 

 

000.0 08.0 08.0- 80.0 2.. 00.0.0 08.0- 08.0 08.0- :3 00a: 0800
0.0.0 0.8.0 08.0- :8 003 08.0- 08.0 08.0 20.0 05.820000
000.0 08.0 08.0 80.0 000.0 08.0 03.0 000.0 0:0 8:8:
0.8.0 0.8.0 08.0- 000.0 02.0 08.0 80.0 :00 00 0.0 85505
08.0 08.0 08.0 000.0 03.0 08.0 08.0 000.0 03.0 8032.0
0032.00 00080 250080 0:00:00 0.800 008: 0200 802$ :00 808002.00 .. 200.59
. 0080800

002308305 000.00 000 00000.03 30. 0000830000 0000.800 00-000. 00.. 0030008. 000000000 0.0000,. 00000.00 00.. 0000.00 000
000.0030 0000 000300.00 90050830000 00000.00 800.. 000000 90000008 00.00.00> 800000.008 00003000 00.00.00.000 000.00>.m. ...N 0.00 .H

87

The bivariate correlation matrices for independent variables measuring household
snowmobiling activity, safety certiﬁcate possession, snowmobile ownership, selected
snowmobile characteristics, respondent age, and non-trip spending is shown in Table 28.
Of the variables, the average age and engine size of household snowmobiles used in
Michigan during the 1996-97 winter are correlated (t= 0.492, p= 0.000). The latter
variable was selected since the average age was found not to be statistically signiﬁcance
between the 1996-97 convicted snowmobile law violator and other snowmobiler groups.
The other set of correlated variables involve the average annual miles and days household
snowmobiles were used during Michigan’s 1996-97 winter (r= 0.534, p= 0.000). The
average annual snowmobile days was selected for inclusion in the modeling process
because estimating annual mileage is difﬁcult and days seem to be a more accurate

measure of snowmobiling activity.

Descriptive Summm
Of the ﬁfteen predisposing independent variables theorized to differentiate 1996-

97 convicted snowmobile law violators and other snowmobilers, 11 (73.3%) were found
to be statistically different between the two groups at the 0.05 level of signiﬁcance (Table
29). These included age and ﬁve of the nine dangerous behaviors or situations relating to
drinking, the speed of snowmobiles, trail conditions, and riding on roadways. The
variables measuring snowrnobilers’ perceptions of the dangers of operating snowmobiles
when intoxicated and the public trail design were found to be correlated with other
variables and thus were eliminated ﬁ'om the modeling process. The list of statistically
different variables also contained all of the fatality reduction initiatives. However, the
variable measuring support for the creation and enforcement of a snowmobile speed limit
on public lands, trails, and waters was eliminated ﬁom the modeling process since it was

correlated with two other variables.

88

.8300? 08380 .000 A 0 0000.280 2..

000000000 0w00000 000 6000-808 6000000808 80808.00 0000.00. .0

0000.3 0.0-000. 00.0 mar—:0 00300.3. 0. 0000 0000083000 00000 :0 0000.. 00w000>< .0.
80.00.8000 .00 0000 0 0.000000 .0

 

. . . 9 3.0008380
0000 000 0 0000 :00 000.0 000.0 000.0- 000.0 000.0 0000800 8083080800080.
000.. 8.0.00.0 30.0 000.0 000.0. 000.0- 000.0 000.0 802 .8000 0800508 80082
000.0 000.0 000.0 000.0- 000.0- 000.0 000.0 0000 00000 020808 000830
000.0 8.0000 000.0- 000.0- :00 03.0 80.0 .8332: 00082
000.0 000.0- 000.0- 000.0- 000.0- 00 .000 2008 000830
000.. 000.0 000.0 000.0 8030 8030 80502

01803300030

 

 

000.. 000.0- 000.0 8000 .080 000 00808080
08.. «8.0 000.0 000.0 000.000 0 :03 000000000
ooo. 00.308305 00.508305 0:3 80.800000
0.0 880 0m 200800:
0000800 8:2 0000 00< 00 08:30 o0< 80.0 800053000 20000> .980
.000”.

 

. 000.0800 00000083000
000.00 000 00000.03 30. 0.5083000 0000.800 3.000. 00.. 00.00000 0:0 00000000000005 00000.00 0:0 00.000030 0.5083000 .0w0
0000000000 000.0 000.00 500.000 303083000 0.00.0002. 80.000008 00.00.50, 0000000000. 0003000 00.00.0800 00000005 .wm 0.0.00.

89

Table 29. Predisposing variables and their statistical signiﬁcance between 1996-97

convicted snowmobile law violators and other snowmobilers.

 

 

Signiﬁcantly
Predisposing Variables different ‘
Respondent’s age S
Dangerous behavior or situation rating "
Operation of snowmobiles by intoxicated persons ‘ S
Drivers lacking skill in operating their machines NS
Cars or trucks on seasonal roads NS
Other uses of snowmobile trails (e.g. X-C skiing) NS
Speed of snowmobiles
Operation of snowmobiles by persons who have been drinking,
but are not intoxicated
Public trail conditions S
Public trail design ‘ NS
Snowmobiling on county or state roads S
F atalig reduction initiatives rating °
Mandatory training for all ﬁrst year snowmobilers regardless of age S
Mandatory personal insurance for all snowmobilers S
More intensive enforcement of existing snowmobile regulations S
Licensing required for all snowmobilers S
Enforcement of a snowmobile speed limit on public lands, trails, S

and waters ‘

 

a. S = signiﬁcant statistical difference at 0.05 level, NS = not signiﬁcant.

b. Rating scale: 5= extremely dangerous, 4= highly dangerous, 3= moderately dangerous,
2= slightly dangerous, 1= not dangerous.

c. Rating scale: 5= strongly supports, 4= moderately supports, 3= neutral, 2= moderately
oppose, 1= strongly oppose.

d. Correlated variable eliminated ﬁ'om modeling process.

90

As for those potential enabling variables theorized to differentiate 1996-97
convicted snowmobile law violators and other snowmobilers, all were found to be
statistically signiﬁcantly different (Table 30). However, since the average estimated
annual miles and days household snowmobiles were used during Michigan’s 1996-97
winter were correlated, only the estimated days variable was included in the logistic
regression analysis.

Table 30 also displays the potential reinforcing variables. Only one variable,
annual household expenditures on snowmobiling, was found to be statistically different.
The average age of household snowmobiles was eliminated since it was correlated with
the average engine size of household snowmobiles.

Table 30. Enabling and reinforcing variables and their statistical signiﬁcance between
1996-97 convicted snowmobile law violators and other snowmobilers.

 

 

 

Factor Signiﬁcantly
Variables different ‘
Enabling
Average number of estimated snowmobiling days per household S
snowmobile b
Avg. number of estimated miles ridden per household snowmobile b' ° S
Average engine size (cc) of household snowmobiles b S
M00002
Proportion of the household 12 years and older who snowmobiled NS
during the Winter of 1996-97
Total number of snowmobiles owned by the household NS
Average age of the of household snowmobiles b’ c NS
Total household expenditures for snowmobiling equipment, S
maintenance, insurance, and storage
Proportion of the household 12 years and older with a snowmobile NS
safety class

 

a. S = signiﬁcant statistical difference at 0.05100, NS = not signiﬁcant.
b. Averaged based on those snowmobiles used in Michigan during the 1996-97 winter.
c. Correlated variable eliminated from modeling process.

91

While only about two-thirds of the hypothesized variables were identiﬁed as
being statistically different between 1996-97 convicted snowmobile law violators and
other snowmobilers, all (except those correlated with others) will be included in the
logistic regression analysis. This is because different tests of signiﬁcance with a larger

acceptable signiﬁcance level will be applied.

92

CHAPTER SIX
LOGISTIC REGRESSION ANALYSIS RESULTS

Three phases were involved in the logistic regression modeling process in order to
identify those variables best differentiating 1996-97 convicted snowmobile law violators
from other snowmobilers and predicting each group’s membership. First, a univariate
analysis was conducted between the dependent variable and all independent explanatory
variables. Secondly, those independent variables potentially contributing to the
explanation of the dependent variable were then used in a series of logistics regression
(LR) models to elicit those variables that best discriminated the 1996-97 convicted
snowmobile law violators from other snowmobilers. Third, the model’s accuracy in

predicting each group’s membership was examined.

Phase 1: Univariate Analysis
Of the twelve predisposing independent variables included in the univariate

analysis (Table 29), eleven were statistically signiﬁcant with a p-value less than or equal
to 0.25 (Mickey & Greenland, 1989) (Table 31). The respondent’s age was highly
signiﬁcant (LR= 78.069, d% 1, p= 0.000). Of the seven dangerous behaviors or
situations, all but operator’s lacking driving skill was statistically signiﬁcant. Two had
signiﬁcance levels of 0.000. These included the speed of snowmobiles (LR-x2= 39.964,
df= 4) and trail conditions (LR-x2= 20.928, df= 4). All of the fatality reduction initiatives
were statistically signiﬁcant, with mandatory personal insurance (LR-x2= 79.941, df= 4,
p= 0.000) and more intensive law enforcement (LR-x2= 25.088, df= 4, p= 0.000) as most

signiﬁcant.

93

Table 31. Likelihood ratio and likelihood ratio chi-square tests between the dependent

and predisposing independent variables. '

 

 

 

Predisposing Variables Value (if p
Respondent’s age " 78.069 1 0.000 *
Dangerous behavior or situation rating b‘ °
Drivers lacking skill in operating their machines 0.180 4 0.996
Cars or trucks on seasonal roads 6.812 4 0.146 *
Other uses of snowmobile trails (e.g. X-C skiing) 7.149 4 0.128 *
Speed of snowmobiles 39.964 4 0.000 *
Operation of snowmobiles by persons who have
been drinking, but are not intoxicated ““800 4 0'019 *
Public trail conditions 20.928 4 0.000 *
Snowmobiling on county or state roads 11.394 4 0.022 "'
Fatalig reduction initiatives °' °
Mandatory training for all ﬁrst year snowmobilers ..-
regardless of age 8'804 4 0'066
Mandatory personal insurance for all snowmobilers 25.088 4 0.000 *
More intensive enforcement of existing a:
snowmobile regulations 7994] 4 0000
Licensing required for all snowmobilers 6.997 4 0.136 *

 

 

a. Dependent variable coded 0= other snowmobilers and 1=T99637 convicted violators.
b. Rating scale: 5= extremely dangerous, 4= highly dangerous, 3= moderately dangerous,

2= slightly dangerous, l= not dangerous.

c. Rating scale: 5= strongly supports, 4= moderately supports, 3= neutral, 2= moderately

oppose, 1= strongly oppose.
d. Likelihood ratio test.
e. Likelihood ratio chi-square.

* Signiﬁcance at p3 0.25.

Both the enabling independent variables, average household snowmobiling days

(LR= 33.647, df= 1, p= 0.000) and engine sizes (LR= 4.413, df =1, p= 0.036), were

statistically signiﬁcant (Table 32). Only two of the four reinforcing independent variables

were found to be statistically signiﬁcant with the dependent variable (Table 32). These

were total household spending on snowmobiling (LR= 14.479, df= 1, p= 0.001) and the

proportion of household 12 years and older having attended a snowmobiling safety class

(LR= 2.445, df= 1, p= 0.118). The proportion of household 12 years and older who

snowmobiled during the 1996-97 winter and the total number of snowmobiles owned by

the household were not statistically signiﬁcant and thus were eliminated from the

modeling process.

94

Table 32. Likelihood ratio tests between the dependent and enabling and reinforcing
independent variables. '

Factor
Variables Value df p

Enabling

Average number of estimated snowmobiling days
per household snowmobile b 33-647 1 0.000 *

Average engine size (cc) of household snowmobiles b 4.413 1 0.036 *

 

Reinforcing
Proportion of the household 12 years and older who 0 330

snowmobiled during the winter of 1996-97 1 0566
Total number of snowmobiles owned by the
household 1.108 1 0.293
Total household expenditures for snowmobiling *
equipment, maintenance, insurance, and storage ”A79 1 0-001
Proportion of the household 12 years and older 2 445 l 0 118 a:

with a snowmobile safety class

a. Dependent variable coded0= other snowmobﬂers and 1= 1996-97 convicted Violators.
b. Averaged based on those snowmobiles used in Michigan during the 1996-97 winter.
* Signiﬁcance at pS 0.25.

 

Phase 2: Initial Logistic Regression Analysis

Backwards Variable Selection Model

The initial LR modeling process, encompassing all cases and signiﬁcant
independent variables identiﬁed in the univariate analysis, employed the backwards
stepwise variable selection approach using a 0.15 criterion entry level of signiﬁcance.
The choice of setting the statistical signiﬁcance criterion for entry at 0.15 is to “prevent
the failure to ﬁnd a relationship when one exists” (Menard, 1995, p. 55).

Of the ﬁfteen independent variables, seven were not statistically signiﬁcant (p=
0.15) predictors of the dependent variable (distinguishing 1996-97 convicted snowmobile
law violators from other snowmobilers). These non-signiﬁcant variables included four
variables describing the dangers related with drinking and snowmobiling, vehicle use of

seasonal roads, riding on public roadways, and other users of snowmobile trails. The last

95

three non-signiﬁcant variables pertained to respondent’s support for initiatives to reduce
snowmobiling relates fatalities. These included licensing operators, requiring personal
insurance, and mandatory training for ﬁrst year snowmobilers. Since these independent
variables were not statistically signiﬁcant predictors of dependent variable, they were
eliminated from the modeling process.

The logistic regression statistics for those predisposing, enabling, and reinforcing
independent variables that were signiﬁcant predictors of the variable distinguishing 1996-
97 convicted snowmobile law violators from other snowmobilers are displayed in Table
33. Consistent with the descriptive analysis, age was best at differentiating 1996-97
convicted snowmobile law violators from other snowmobilers (W 2k= 28.5376, p=
0.0000). This was expected since 1996-97 convicted snowmobile law violators were on
average approximately six years younger than other snowmobilers. Youthfulness of
violators is also consistent with past snowmobile, trafﬁc safety, and ﬁsh and wildlife
research on law violators and accident casualties (Dewar, 1973; Jackson et al., 1979;
Anderson, 1987; James et al., 1991; Elander et al., 1993).

The second of the signiﬁcant predisposing independent variables that divided the
two groups was support for more intensive enforcement of existing snowmobiles
regulations (W 2,= 23.9049, p= 0.0000). As one might expect, 1996-97 convicted
snowmobile law violators were less likely to support more enforcement. Researchers
have found that, in general, individuals who have received a citation have a more
unfavorable attitude toward the police and their presence than those who have not
received a citation (Cox & White, 1988). This has been corroborated in other studies
comparing apprehended wildlife law violators with lawful hunters (Stoll, 197 5; Melynk,
197 7).

96

.000 00 00 808000000 _.

0883 3-000. 000 @0000 00300.2 00. 000: 02.008305 00000 :0 00000 00w000>< .0

0000.0...000 00.00.00.000 .0800 n 0 000000.880

n 0 800000.. ..0 000090 n ..0 6.00.0000 0.0..» u .03 .0000.0m...000 00.000800 ..0 00000 00000000 u.m.m 00.00.00.000 000.0000m00 H m .0

 

000.0 000.00 00008305 0

 

 

003.c- .. 38.0 000.0 man-Nd 0000.0. 00. 3 000.0 000 00000 N. 0.0000000 00. ..0 00.000080
. . . . . 0w00000 000 60:00:00. 600000.808 .8080.000
0- .00 o 0. .000 o 0000 0 .0000 o 808 o m0...008305 00.. 0000000000000 0.0000000 .000 ..-
0.8800040
$00.0 .. 0000.0 0000.0 0000.0 200.0 .. 80005308 2008000 00 .80 80.... 2008 0082.0.
. . . . . 02.008305 0.0000000 00
008 o .. 0.8 o 003 o. .08 0 mm... o 0 000 w:...008305 00008.000 ..0 000800 0w000><
00.140301.
mono-o- .. .08... 090.0 0000.0 mmm ..o- 00..008305 ..0 00000 000 00.3 020.0500 009000.
800.0 _._ moood 030:0... 0000.0 5.2. 000.0.0000 ..80 00000 000 00.3 080.0500 00w000n.
. - . . . . 000000.800 2.008305
.3. o 0. 88 o 030 mm 003 .0 30m 0. 90.00.00 ..0 058080.00 000.0008. 0008 00.. 000005
30......- _. coco... 00mm.w~ 008.0 83.0.- 0w0 0.0500500.
Mada-0%
0 0 .03 .m.m m 0.00.00>
00000 0.

 

. 0000008305 00000 05 00000.0.> 30. 00008305 0000.800 3.000. 0003000 0.0.0.0000 00.0000w00 0.0080.

00.30000 000030.000 000 8. 00000.00 00.00.00> 90.000.00.00 000 08.0000 000.005.0000 000 00.. 00.00.0000 00.000800 0.00800 .mm 0.00 ..-

97

Of the dangerous behaviors and situations, the two most signiﬁcant independent
variables from the univariate analysis, public trail conditions (W2k= 12.9278, p= 0.0003)
and the speed of snowmobiles (W2k= 4.3436, p= 0.0371), were signiﬁcant predictors of
the dependent variable. In this model, the 1996-97 convicted snowmobile law violators
considered the speed of snowmobiles as less dangerous and public trail conditions as
more dangerous than other snowmobilers. Logically, with greater the speed of the
snowmobile, it is more likely that imperfect trail conditions will inﬂuence safety and
control. In addition, comparative studies of individuals who have not and those who have
violated wildlife laws concluded that violators are generally more unfavorable towards
any management activities (Melynk, 1977; Andersen, 1987). This suggests that 1996-97
convicted snowmobile law violators are more likely to be dissatisﬁed with trail
management (e.g., public trail conditions) than other snowmobilers.

As postulated, the average number of days each household snowmobile was used
for snowmobiling during the 1996-97 winter (W2k= 10.2557, p= 0.0014) and the average
engine size of household snowmobiles (W2k= 3.7093, p= 0.0541) were statistically
signiﬁcant enabling predictors in differentiating 1996-97 convicted snowmobile law
violators from other snowmobilers. In this model, 1996-97 convicted snowmobile law
violators had more snowmobiling days than other snowmobilers. This is comparable to
other studies where injured snowmobilers spent more time snowmobiling than uninjured
snowmobilers (Waller & Lambom, 1975; Rowe et al., 1993) and wildlife law violators
hunted more than non-violators (Gray & Kaminski, 1994). While there are conﬂicting
conclusions about the signiﬁcance of engine size as it relates to law violations, this model
found that 1996-97 convicted snowmobile law violators had slightly larger engines than

other snowmobilers.

98

Of the original ﬁve reinforcing variables hypothesized to differentiate 1996-97
convicted snowmobile law violators and other snowmobilers, only two were included in
the LR model. First, as noted in the descriptive analysis, 1996-97 convicted snowmobile
law violators tended to spend more for snowmobiling equipment, maintenance, insurance,
and storage costs than other snowmobilers (W2k= 7.5300, p= 0.0061). The second
signiﬁcant variable was the proportion of a household that completed a snowmobiling
safety class (W2k= 5.1859, p= 0.0228). As speculated, a smaller proportion of 1996-97
convicted snowmobile law violator household members had taken a snowmobile safety
class than those in other snowmobiler households.

Altogether, the variables selected using the backwards stepwise approach
produced a statistically signiﬁcant model (Model X2= 152.516, d% 8, p= 0.0000).
However, the proportional reduction in the model chi-square (R2L= 0.1037) suggests only
a slight association between the combination of independent variables and the dependent
variable distinguishing 1996-97 convicted snowmobile law violators from other
snowmobilers. This weak association suggests that additional independent variables are
necessary to reduce the badness of ﬁt, or simply that the variation between the
respondents in each group is in fact small. This latter explanation is perhaps more
plausible since group membership over time may shiﬁ. Other snowmobilers, while
characterized as non-violators, cannot be regarded has never having violated snowmobile
laws. It is also likely that some of the 1996-97 convicted snowmobile law violators
disobeyed the law unknowingly. Lastly, the R2L statistic should be analyzed with caution,
since the focus of this part of the LR analysis was to identify distinguishing variables.
The adequacy of the model in distinguishing 1996-97 convicted snowmobile law
violators and other snowmobilers is also a measure of goodness of ﬁt of the model and

will be further addressed.

99

Single Block Eng Model
To substantiate the signiﬁcance of the independent variables selected in the

backwards stepwise logistic regression model, a second LR analysis was conducted using
a single block entry procedure. This procedure enters the selected independent variables
at the same time in the order in which they emerged in the backwards stepwise logistic
regression model. This second model is used to validate the signiﬁcance of each
independent variable distinguishing 1996-97 convicted snowmobile law violators from
other snowmobilers (Hosmer & Lemeshow, 1989). This validation is accomplished
through a comparison of regression coefﬁcients and Wald statistics from the original LR
model to the LR model using single block entry.

The logistic regression statistics for independent variables are displayed in Table
34. While all the independent variables exhibited some variation in the regression
coefﬁcients and Wald statistic, four in particular changed noticeably.

The regression coefﬁcient for the proportion of the household with a snowmobile
safety class changed ﬁ'om -0.5450 to -0.4740, while the Wald statistic decreased from
5.1859 to 4.1228. However, these changes did little to affect the statistical signiﬁcance of
the variable. Likewise, the danger associated with the public trail conditions regression
coefﬁcient decreased from 0.2187 to 0.1916 and the Wald statistic decreased from
12.9278 to 10.4223 without affecting the signiﬁcance of it.

The other two variables did have signiﬁcance levels that change markedly from
the backwards stepwise model to the single block entry model. The statistical signiﬁcance
of the variable gauging the danger associated with the speed of snowmobiles increased
ﬁ'om 0.0371 to 0.0772. In this case, the regression coefﬁcient when from -0.1335 to -
0.1105 and the Wald statistic decreased from 4.3436 to 3.1221. Lastly, the statistical
signiﬁcance level of the average engine size of household snowmobiles increased from
0.0541 to 0.0664. However, little change in the regression coefﬁcient and Wald statistic

was noted.

100

.80 mg a 85000000 ..

.3000— 00053 w:_0:0 :0»_:0=2 E 000: 0230:0305 000:0 :0 0000: 003020. .:

4:060:08 500—00000 00:00: N 0 00:85:?”

n 0 .800000 .00 0000m00 n 00 600080 0.03 n .03 06065000 :0mmm00w00 .00 00000 0000500 ".m.m £56508 :000000m00 n m .0

 

000.0 30.000 2308305 0

 

 

N586- .. 33.0 _ 32.: mmmmd 9.3.0- :00: 00:0 0:0 0000.» N: 20:00.00: 0:: .00 5000000:
. . . . . 0w0000m 0:0 00:00:05 60:38:02: 4:083:00
38 o .. goo o _ mm: n :58 0 880 o w:=_:0:030:0 00.: 8030:0900 20:00.00: 030,—.
3.00043
008.0 0000.0 _ moomd good m Sod .0 0.0—30.00305 20:00:00: .00 A08 00:0 3&5 0w000><
. . . . . .0 2508305 20:00:00: 00:
53 o .. wooo o _ 0mmm : 3.8 c R S o 930 30.508305 0000:0000 .00 00:80: 0w000><
003.040
:86- «Rod : :NN— .m Rood m3 :6. 0230:0305. .00 000% 05 :03 0030800 00w:0Q
N306 .. «Sod _ mmmvd— named 03— .o 0:020:00 00.: 0::0: 0:0 :00: 00.060000 00w:0Q
. - . . . . 0:002:30 0::05305
3N: o .. coco o _ memo cm 33 c m EN 0. $0000 00 808000005 0 $8005 0008 00.0 000096
:2 .o- _.. ooood : wamwm N806 named- 0%: 0.300330%
m.||m.|l:mmo 0000:
0 a 00 «a 0.0 m 2.005
00000.:

 

. 002505305 0050 0:0 00000—00, 30. 2508305. 00:03:00 3.03. :00300: $0205 :0_mm00w00

000%2 :0 9:0 :02: 0_w:_0 3%: 003053 w:m000.0::00 05 $5305 0500000000 00.: 0000080 :000000w00 000mg: .3“ 030,—.

101

The variation of the regression coefﬁcients and the Wald statistic between these
two entry procedures is principally related to the statistical algorithms used in calculating
them (Hosmer & Lemeshow, 1989). Nevertheless, it does highlight the potential
uncertainty about the relative importance of the variables in the overall logistic regression
model. As a result, a comparison of the logistic regression model with and without these
non-signiﬁcant variables was performed.

The logistic regression model with all eight independent variables entered as a
single block produced a statistically signiﬁcant Model X2 of 149.122 with a proportional
reduction model chi-square of 0.097 (Table 35). Excluding the independent variables
measuring the respondent’s opinions about the danger associated with the speed of
snowmobiles reduced the Model X2 to 145.989 and the R2L to 0.094. In doing so, the
signiﬁcance level of the average engine size of household snowmobiles increases from p=
0.0664 to p= 0.1340, reducing its overall signiﬁcance.

Table 35. Logistic regression model statistics with, without, and combined non-

signiﬁcant variables identiﬁed in the single block entry in logistic regression analysis
between 1996-97 convicted snowmobile law violators and other snowmobilers. '

 

 

Logistic Regression Model 5 Model X2 df R2L
With both variables included 149. 122 8 0.097
Without the danger associated with the speed of

snowmobiles variable 145989 7 0’094
Without the average engine size (cc) of household

snowmobiles variable 149641 7 0'095
Without either variable 147.762 6 0.093
Combining both variables ° 147.818 7 0.097

 

 

a. Mode 72 = model chi-square, df = dimes ofﬁeedom, and R‘L = proportion reduction
in the model chi-square.

b. LR model includes independent variables selected in the backwards stepwise logistic
regression analysis.

c. SPSS automatically calculates and creates new variable for those speciﬁed independent
variables believed to be interactive terms.

102

The same situation arises when the independent variables measuring the average
engine size of household snowmobiles are excluded. While the Model X2 actually
increases by 0.519, the R2L decreases slightly by 0.002 from the LR model with all eight
variables, and the danger associated with the speed of snowmobiles becomes
insigniﬁcant.

When both independent variables are excluded the resulting Model X2 equaled
147.762 and the R2L equaled 0.093. These results suggest that both the danger associated
with the speed of snowmobiles and the average engine size of household snowmobiles
add only slightly to the overall model. Moreover, both independent variables appear to be
mutually related to one another since the exclusion of either from the model affects the
other.

Since the speed of snowmobiles is partially related to the average engine size of
household snowmobiles, these two variables were multiplied together to form a cross-
product or interaction term (Hosmer Lemeshow, 1989). This term was then entered in the
logistic regression model along with the other signiﬁcant independent variables. The
change in the model chi-square and proportional reduction in the model chi-square from
the original LR model to the new LR model, with the interactive term as well as the
statistical signiﬁcance of the new independent variable, exhibited some signiﬁcance
(Hosmer Lemeshow, 1989) (Table 35). The results yielded a Model X2 of 147.818 and
R2L of 0.097. Moreover, this new variable produced a statistically signiﬁcant (p= 0.0431)
coefﬁcient of -0.0002 and Wald statistic of 4.40908. However, its signiﬁcance is
marginal and its overall inclusion does little to improve the model. Furthermore, it would
be difﬁcult to draw concise conclusions about this variable, so it was eliminated from the

model.

103

The logistic regression statistics for the single block entry model without the
cross-product between the average engine size of household snowmobiles and danger
associated with the speed of snowmobiles included is displayed in Table 36. Both the
respondent’s age and the variable measuring support for more intensive enforcement of
existing snowmobile regulations had Wald statistic that increased signiﬁcantly. For
instance, the Wald statistic from age increased from 28.3198 in Table 35 to 30.4656.
Support for more intensive law enforcement increased from 26.0803 in Table 35 to
37.2407. Overall, these six independent variables produced a statistically signiﬁcant
Model X2 of 147.762 with a proportional reduction model chi-square of 0.093.

104

.80 00 a 8:800:00 ..

.3000: 00053 w:_0:0 53:22 :2 000: 0230:0305 000:0 :0 0000: 00w000>< .:

3208000 :00200000 3000: .I. 0 .00:005:w_0

u 0 .800000 .00 0000w00 u .00 6000000 0:03 H .03 03208000 :20000w00 3 00000 0000530 ".m.m 03208000 :20000w00 u m .0

00:0 30.000 2308305 0

 

 

vovod- .. 38.0 2 800.0 SmNd N098- :: 3 000.0 0:0 0000 x N: 03:88: 0:: .00 5:00:00:
. . . . . 0w00000 0:0 00:00:03 60:32:08 30:00:60
003 o .. ~08 o _ :80 0 508 o 880 o w::30:030:0 00.: 0003005900 03:83: 300,—.
000.000
. . . . . 0 2308380 03:00:00: 00:
some o .. Boo o : wnmm m: 030 o 9.5 o 0 >00 w::30:030:0 0000:0000 .«0 03:00: 0w000><
m:::0:m
9.3.0 .. Seed _ mmmwd «03.0 2.3: .o 0:030:00 =00 23:0 05 :23 0000000000 0m§Q
. - . . . . 0:002:30 2308380
003 o .. 88 o _ 80m 0m 83 0 Sen 0. @0008 3 308000030 0 30:00:: 0008 00.0 000096
mmm _.o- a. good 2 030.9” 0086 2.8.0- 0w0 0350:0003:
3
0 a .00 003 .m.m m 03200>
00000.:

 

. 002305380 0:00 0:0 00000—0? 30_ 2305308 00:03:00 3.003 :00300:
20%—0:0 :20000w00 000:»0. :_ 300:0 2003 2w£0 m:_0: 00>0:000 0230:0305 .00 000% 0:0 :23 0000000000 00m:00 0:0 0230:0305
03:00:00: .00 020 0:_w:0 0w000>0 0:0 :23 0030:0003 €300,330 0:0 3:305 3500:2005 00.0 00000000 :20000wo0 000304 .3 030:.

105

Initial Logistic Regression Model Diagnostics

To assess nonlinearity in the logit, a Box-Tidwell transformation was made of the
six independent variables, which were then added to the single block entry logistic
regression model (Hosmer Lemeshow, 1989; Menard, 1995). Any transfonned variable
that was statistically signiﬁcant (p= <0.05) indicated nonlinearity (Menard, 1995). The
results of this transfonned model showed that the average number of estimated days each
household snowmobile was used during the winter of 1996-97 and the total household
expenditures on snowmobiling that same winter were signiﬁcant (Table 37). Contending
with this issue can simply be a matter of rescaling the values of variables (Hosmer &
Lemeshow, 1989). Consequently, the grouped values form of these independent variables
presented in Tables 19 and 23, were transformed using the Box-Tidwell procedure and
reapplied to the model. The grouping of the data follows the conﬁguration used by Rowe
Table 37 . Nonlinearity in the logit results using the Box and Tidwell transformation of

the relevant predisposing, enabling, and reinforcing variables distinguishing 1996-97
convicted snowmobile law violators and other snowmobilers. "

Factor
Variable B p

Predisposing
Respondent’s age 0.0285 0.4361

Support for more intensive enforcement of existing
snowmobile regulations '0-3164 01507

Danger associated with the public trail conditions 0.0687 0.7730

 

Enabling

Average number of estimated snowmobiling days
per household snowmobile " ‘0-0141 0-0011 *

Reinforcing
Total household expenditures for snowmobiling 410001 0.0489 ..
equipment, maintenance, insurance, and storage
Proportion of the household 12 years and older With 0.6241 03443
a snowmobile safety class

a. B = regression coefﬁcients and p = sigm:ﬁcance. .
b. Averaged based on those snowmobiles used in Michigan during the 1996-97 wmter.

"' Signiﬁcance level of p5 0.05 indicates nonlinerarity of the logit.

 

106

and associates (1993). The results of this analysis are displayed in Table 38. In both
cases, the grouped version of the total household expenditures independent variable (B= -
0.2778, p= 0.2385) and the average number of estimated snowmobiling days per
household snowmobile (B= -0.4958, p= 0.1137) reduced the nonlinearity.

Table 38. Nonlinearity in the logit results using the Box and Tidwell transformation of
relevant and reconﬁgured predisposing, enabling, and reinforcing variables distinguishing
1996-97 convicted snowmobile law violators and other snowmobilers. '
Factor

Variable B p

Predisposing
Respondent’s age - 0.0293 0.4207

Support for more intensive enforcement of existing
snowmobile regulations '0-3229 0.1433

Danger associated with the public trail conditions 0.0893 0.7078

 

Enabling
Grouped average number of estimated snowmobiling
days per household snowmobile ” ‘0-4958 (“137

Reinforcing
Grouped household expenditures for snowmobiling _
equipment, maintenance, insurance, and storage ° 0'2778 02385
Proportion of the household 12 years and older with
a snowmobile safety class 05237 0'3665
a. B = regression coefﬁcients and p = signiﬁcance.
b. Average days categories: l= $5, 2= 6-10, 3= 11-20, 4= 21-30, and 5= 231.
c. Expenditures categories: 1 = S $500, 2= $501-$2,000, 3= $2,001-$5,000, 4= $5,001-
$10,000, 5= 2310,001.

 

When both of these grouped variables are included in a logistic regression model
using a single block entry procedure, they were determined to be signiﬁcant factors
differentiating 1996-97 convicted snowmobile law violators from other snowmobilers
(Table 39). In this model, the Wald statistic for the grouped variable of the average
number of estimated snowmobiling days per household snowmobile increased to 23.7754

from a value of 12.2278 from the previous model in Table 36. The Wald statistic for the

107

.000 00 0 808000000 ..

08.2% "0 08.000-08.00 n0 08.00-0800 um 0800-500 "0 .800 m u 0 “88038 023005900 .0
.SN "0 05 .020 "0 .00-: um 00-0 um .00 n: ”88028 0000 000022 .0

002000000 000200000 :00000 .I- 0 0000000000

u 0 800000.: :0 0000000 .I- :0 .0::0::0:0 0:03 n 00>» 082000000 50000000 .:0 00000 00000000 n.m.m 082000000 :0:0000w00 u m .0

 

000:0 30:00 23083000 0

 

0000.0- ... 2.8.0 : 00.2.0 003.0 0000.0- :0 3 00:0 000 0000» N: 03:33: 0:: .:0 000000000
. . . . . 0 000020 0:0 00000008 6000:2808 00080300
0000 0 _.. 0000 0 : 2mm N: 300 0 :00: 0 0003083000 00.: 0000000000000 0:0:0000: 0000000
00300040000:
. . . . . .0 23083000 0:0:0000: 000 00:00
00:: 0 _.. 0000 0 : 00: mm 0000 0 000m 0 080308308 00:08:00 .:0 00:80: 00000 >0 0000000
0308'”:
0000.0 0 0.0000 : 020.:- 00000 20:0 0:200:00 000: 0:300 0:: :03 00:0:00000 00:09
. - . . . . 00002300 2308380
N00: 0 .. 0000 0 : wmw: mm 0000 0 0:0m 0- 0:00:00 .:0 8080000.:00 0:00:08: 0008 00.: :00000m
00~:.0- _._ 0000.0 : mmmmﬁm 0000.0 N080- 0w0 00000000000:
wag-:3“
0 0 00 .03 0.0 0 2000;
0.01:.00. 0:

 

. 002308308 0050 0:0 000:0:0:> 30: 23083000 00:03:00 00-000: 000300: 0:00:80 020000000 2:200: :: x0000 0:003 200:0
00:00 0030:00> 000000-8000 000 083000 0800000000 00000000000 0:: :03 :0008 :000 00.: 0000080 :20000000 0:00:00: .0m 0300.

108

grouped total household expenditures also increased over its former version from 7.4831
to 12.3312. The inclusion of both these new variables resulted in much better model
overall as the Mode X2 increased from 147.762 to 165.378, while the R2L also increased
from 0.093 to 0.104.

Collinearity between independent variables was assessed using the linear
regression function of SPSS to calculate the tolerance statistic (Menard, 1995). Those
variables with values less than 0.20 indicate a problem of collinearity (Menard, 1995). As
indicated in Table 40, all of the independent variables had a tolerance statistic of 0.85 or
greater. Thus, collinearity is not an issue with these data.

Table 40. Colinearity results for the reconﬁgured predisposing, enabling, and reinforcing

variables distinguishing 1996-97 convicted snowmobile law violators and other
snowmobilers.

Factor
Variable Tolerance

 

Predisposing
Respondent’s age 0.913

Support for more intensive enforcement of existing 0 845
snowmobile regulations '

Danger associated with the public trail conditions 0.977

Enabling
Grouped average number of estimated snowmobiling O 949
days per household snowmobile ‘ '

Reinforcing
Grouped household expenditures for snowmobiling 0 962
equipment, maintenance, insurance, and storage " ‘

Proportion of the household 12 years and older with O 986
a snowmobile safety class ’

a. Average days categories: 1= SS, 2= 6-10, 3= 11-20, 4= 21-30, and 5= 231.
b. Expenditures categories: 1 = SSSOO, 2= $501-$2,000, 3= $2,001-$5,000, 4= $5,001-
$10,000, 5= 2510,001.

 

109

 

 

 

The review of Studentized residuals for the seven independent variables revealed
45 cases with values that ranged ﬁ'om 2.01 to 2.58. While values greater than £3.00 are

considered a major concern, values greater than i 2.00 also should be reviewed for
problems (Menard, 1995). Of the 45 cases, all were 1996-97 convicted snowmobile law
violators who were identiﬁed in the logistic regression model as an other snowmobiler. A
comparison of the means and percentage for the six independent variables between the 45
cases with Studentized residuals values greater than 2.00, all 1996-97 convicted
snowmobile law violators, and other snowmobilers is displayed in Tables 41 - 43. As
shown, the average age of the identiﬁed cases was 10 years older than the 1996-97
convicted snowmobile law violators and 6 years older than other snowmobilers. They
also had slightly fewer household members take a safety class. These 45 respondents
Table 41. Mean age and proportion of the household completing a snowmobile safety

class between cases with Studentized residual values greater than 2.00, 1996-97
convicted snowmobile law violators, and other snowmobilers.

Identiﬁed 1996-97

 

M cases Violators Others
Variable (N= 45) (N= 360) (N=l,348)
Predisposing 52 SE >‘<
Respondent’s age 44.5 34.2 40.0
Reinforcing

Proportion of the household 12 years and older 0 13

with a snowmobile safety class 0'14 0-17

 

were also much more likely to support more intensive law enforcement efforts and
believe that the danger associated with the trail conditions is less serious than the other
two groups. Moreover, this group also rode their household snowmobiles much less and
spent less than either the 1996-97 convicted snowmobile law violators or other

snowmobilers.

110

Table 42. Percentage of support for more intensive enforcement of existing snowmobile
regulations and rating of the danger associated with the public trail conditions between
cases with Studentized residual values greater than 2.00, 1996-97 convicted snowmobile
law violators, and other snowmobilers.

Identiﬁed 1996-97

 

Factor cases Violators Others
Variable (N= 45) (N= 360) (N=l,348)
Predisposing % % %

Support for more intensive enforcement of
existing snowmobile regulations

Strongly supports 40.0 16.3 33.6
Moderately supports 28.9 18.0 25.5
Neutral 28.9 34.4 25.2
Moderately oppose 0.0 12.4 7.3
Strongly oppose 2.2 18.9 8.4
Danger associated with the public tr. conditions
Extremely dangerous 0.0 10.9 4.5
Highly dangerous 6.7 10.9 9.7
Moderately dangerous 33.3 31.4 31.1
Slightly dangerous 44.4 30.0 32.9

Not dangerous 15.6 16.9 21.8

 

111

Table 43. Percentage of grouped average number of estimated snowmobiling days per
household snowmobile and household expenditures for snowmobiling equipment,
maintenance, insurance, and storage between cases with Studentized residual values
greater than 2.00, 1996-97 convicted snowmobile law violators, and other snowmobilers.

Identiﬁed 1996-97

 

E99191 cases Violators Others
Variable (N=45) (N=360) (N=l,348)
Enabling % % %

Grouped average number of estimated
snowmobiling days per household snowmobile

55 37.8 17.8 34.9
6-10 35.6 23.5 26.5
11-20 22.2 31.2 23.2
21-30 2.2 11.5 7.9
231 2.2 16.0 7.5
Reinforcing
Grouped household expenditures '
55500 55.6 23.4 36.2
$501-$2,000 15.6 16.4 19.7
2,001-$5,000 17.8 16.1 21.1
5,001-$10,000 11.1 30.7 22.3
2$10,001 0.0 13.5 9.7

 

a. Expenditures mcludes snowmobiling eqqument, maintenance, insurance, and storage.

As noted previously, it is expected that a number of cases will not ﬁt the model.
This is because the deﬁnition of a 1996-97 convicted snowmobile law violator is of
limited time duration. A convicted snowmobile law violator one year will not necessarily
be a violator in another year. An individual may also have been cited and convicted of a
violation simply as the result of unwitting action, not a deliberate attempt to break the
law. The 45 cases appear to be a more conservative group than either the 1996-97
convicted snowmobile law violators or other snowmobilers. Hence, while these cases do
not illustrate a problem with the data and thus are retained in this analysis, they do

indicate a theoretical challenge for ﬁmher research.

112

The analysis of leverage values, which indicate cases that may unduly inﬂuence
the parameters of the model, highlights the variability of the respondents in each group
(Hosmer & Lemeshow, 1989). In this analysis, 73 cases were found to have leverage
values of greater than 0.012 (Menard, 1995). These 73 cases were almost equally split
between the 1996-97 convicted snowmobile law violators and other snowmobilers. When
comparing the means and percentages of the independent variables, some interesting
patterns emerge (Tables 44 - 46). For instance, the cases with leverage values over 0.012
were younger, averaging 30.4 years old compared to 34.2 years for 1996-97 convicted
snowmobile law violators and 40.0 years for the other snowmobilers. These 73 cases also
had households with a much greater proportion completing a snowmobiling safety class.
Table 44. Means for the general predisposing, enabling, and reinforcing variables

between cases with leverage values greater than 0.012, 1996-97 convicted snowmobile
law violators, and other snowmobilers.

Identiﬁed 1996-97

 

£10104 cases Violators Others
Variable (N= 73) (N= 360) (N=l,348)
Predisposing )’< >7 >‘<
Respondent’s age 30.4 34.2 40.0
Reinforcing
Proportion of the household 12 years and 0 43 0 14 0 17

older with a snowmobile safety class

 

This younger group also had vastly differing views. They had little support for more
intensive law enforcement, while their opinions about the danger of public trail
conditions were much stronger than either the 1996-97 convicted snowmobile law
violators or other snowmobilers. These 73 cases also had households with more days per
snowmobile and higher annually non-trip related expenditures on snowmobiling.

While these residual values point to cases that may adversely affect the result of
the logistic regression model, it suggests that there is wide variation within each group.

This implies overlap in group membership.

113

Table 45. Percentage of support for more intensive enforcement of existing snowmobile
regulations and rating of the danger associated with the public trail conditions between
cases with leverage values greater than 0.012, 1996-97 convicted snowmobile law
violators, and other snowmobilers.

Identiﬁed 1996-97

 

5.3—CM cases Violators Others
Variable (N= 73) (N= 360) (N=l,348)
Predisposing % % %

Support for more intensive enforcement of
exrsting snowmobile regulations

Strongly supports 12.3 16.3 33.6
Moderately supports 1 1.0 18.0 25.5
Neutral 24.7 34.4 25.2
Moderately oppose 13.7 12.4 7.3
Strongly Oppose 38.4 18.9 8.4
Danger associated with the public tr. conditions
Extremely dangerous 43.8 10.9 4.5
Highly dangerous 8.2 10.9 9.7
Moderately dangerous 15.1 31.4 31.1
Slightly dangerous 11.0 30.0 32.9
Not dangerous 21.9 16.9 21.8

 

114

Table 46. Percentage of grouped average number of estimated snowmobiling days per
household snowmobile and household expenditures for snowmobiling equipment,
maintenance, insurance, and storage between cases with leverage values greater than
0.012, 1996-97 convicted snowmobile law violators, and other snowmobilers.

Identiﬁed 1996-97

 

15% cases Violators Others
Variable (N= 73) (N= 360) (N=l,348)
Enabling °/o °/o %

Grouped average number of estimated
snowmobiling days per household snowmobile

$5 13.7 17.8 34.9
6-10 2.7 23.5 26.5
11-20 16.4 31.2 23.2
21-30 16.4 11.5 7.9
231 50.7 16.0 7 .5
Reinforcing
Grouped household expenditures ’
55500 20.5 23.4 36.2
$501-$2,000 13.7 16.4 19.7
2,001-85,000 11.0 16.1 21.1
5,001-810,000 31.5 30.7 22.3
2810,00] 23.3 13.5 9.7

 

a. Expenditures includes snowmobﬂing equipment, maintenance, insurance, and storage.

The analysis of the dbeta values, which also indicates the inﬂuences of a

particular case, did not identify observations with a value more than one (Menard, 1995).

Initial Logistic Regression Analysis gummy

The initial logistic regression analysis to identify those independent variables that
best differentiate 1996-97 convicted snowmobile law violators from other snowmobilers
encompassed two steps. In the ﬁrst step, the backwards likelihood-ratio stepwise variable
selection approach was employed with all cases and the ﬁfteen statistically signiﬁcant
variables identiﬁed in the univariate analysis. To substantiate those independent variables
selected in the backwards stepwise logistic regression model, a second model was

computed using a single block entry procedure.

115

From the backwards likelihood-ratio stepwise variable selection model, eight
independent variables differentiated 1996-97 convicted, snowmobile law violators ﬁom
other snowmobilers. Of these, age was the best discriminator, followed by the variable
deﬁning the respondent’s support for more law enforcement. The average number of
estimated snowmobiling days per household snowmobile and the assessment of the
danger associated with trail conditions were the next best at discriminating these two
groups. In addition to these independent variables, the total household expenditures for
snowmobiling equipment, maintenance, insurance, and storage as well as the proportion
of the household 12 years and older with a snowmobile safety class were also identiﬁed.
Finally, the assessment of the danger associated with the speed of snowmobiles and
average engine size of household snowmobiles were also slightly signiﬁcant in the
backwards stepwise logistic regression model. However, both of these variables were
found to be insigniﬁcant discriminators in the single block entry logistic regression
model. Yet when multiplied together and incorporated as an interacting term in the
logistic regression model, they were found to be statistically signiﬁcant. In spite of this,
the interacting term did little to improve the overall model and thus was eliminated.

The single block entry logistic regression model and six independent variables
differentiating 1996-97 convicted snowmobile law violators from other snowmobilers
were further scrutinized in a series of diagnostic procedures. In this phase, both the
average number of estimated snowmobiling days per household snowmobile and the total
household expenditures for snowmobiling equipment, maintenance, insurance, and
storage were detected as having a nonlinear relationship with the dependent variable, but
this was remedied by rescaling them to ﬁve ordinal categories and reentering them in the

single block entry LR model (Table 39).

116

Of these six ﬁnal independent variables, support for more intensive enforcement
of existing snowmobiles regulations (W2k= 35.1838, p= 0.0000) was best at
differentiating 1996-97 convicted snowmobile law violators from other snowmobilers. As
hypothesized, 1996-97 convicted snowmobile law violators were less supportive of
enforcement than other snowmobilers. Second most signiﬁcant was respondent’s age
(W21,== 27.3353, p= 0.0000). Here 1996-97 convicted snowmobile law violators were
younger than their counterparts. The grouped average number of estimated days each
household snowmobile was used (W2,= 23.7754, p= 0.0000); spending on snowmobiling
equipment, maintenance, insurance, and storage costs (W2k= 12.3312, p= 0.0004); the
danger of public trail conditions (W2k= 7.6335, p= 0.0057); and the proportion of the
household completing a snowmobile safety course (W 2,= 4.3370, p= 0.0373) round out
the rest of the independent variables. Convicted snowmobile law violators are likely to
snowmobile more, spend more on snowmobiling, rate public trail conditions as more
dangerous, and be least likely to have someone in their household with a snowmobile
safety course than other snowmobilers.

Analysis of the residuals from the single block entry LR model identiﬁed
relatively signiﬁcant numbers of high Studentized residuals and leverage values. This
suggests that a notable proportion of the 1996-97 convicted snowmobile law violators
and other snowmobilers, while deﬁned as members of each exclusive group, could be
characterized as a member of the other group in a future or past year. A more detailed
analysis of the model’s ability to discriminate between the groups will be conducted in a

subsequent section.

117

Phase 3: Logistic Regression Analysis: Model Discrimination

The performance of the model is ﬁrrther assessed for accuracy in distinguishing
1996-97 convicted snowmobile law violators and other snowmobilers by examining
predicted and observed group membership. However, since there are many other
snowmobilers versus fewer 1996-97 convicted snowmobile law violators, the
classiﬁcation outcome is potentially compromised as most individuals will be classiﬁed
in the larger group (Morrison, 1969). As a result, the total number of 1996-97 convicted
snowmobile law violators cases used in testing the model was applied to derive the
appropriate sample size for each group. All 1996-97 convicted snowmobile law violators
were included, while a random selection procedure was used to generate a comparably
sized sample of other snowmobilers. This resulted in a sample of 321 cases for each
group.

The model’s ability to discriminate was evaluated using a classiﬁcation table, a
histogram of estimated probabilities, and the c statistic. These were computed or derived
from the results of the single block entry logistic regression model with the ﬁnal
independent variables identiﬁed in Table 39.

The classiﬁcation table distinguishes both the observed and predicted group
membership. A predicted probability value of 0.5 or greater was used to classify the case
as a snowmobile law violator, while less than 0.5 indicated an other snowmobiler. A
histogram of predicted probability is used to show where the cases fall along the
distribution.

Finally, the c statistic is used to assess the signiﬁcance of the model’s capability
in distinguishing between the two groups. In order to calculate the c statistic, the
predicted probabilities (ranging from 0 to 1) are divided by 0.00005 to create a new
variable that groups the predicted probabilities into a large number of distinct groups
(Norusis, 1997). This is then used in a cross-tabulation with the actual group membership

(dependent variable) to calculate a Somers’ (1 statistic. Somers’ d is then divided by two

118

and 0.5 is then added to the quotient to derive the c statistic (Norusis, 1997). Its value
ranges from 0.5 to 1.0. A value of 0.5 suggests the model has no predictive ability and a
value of 1.0 that is 100% accurate in the model assigns higher probabilities to cases with

the event (being a violator) than without (Norusis, 1997).

Model Discrimination

The classiﬁcation table for the ﬁnal logistic regression model between 1996-97
convicted snowmobile law violators and other snowmobilers shows approximately two-
thirds of the cases correctly classiﬁed (Table 47). The model classiﬁed slightly more
other snowmobilers correctly (220 or 68.5%) than 1996-97 convicted snowmobile law
violators (212 or 66.0%).

Table 47. Classiﬁcation table for the ﬁnal independent variables logistic regression model

using single block entry between 1996-97 convicted snowmobile law violators and other
snowmobilers. '

 

 

 

Predicted
1996-97 Violators Others
Observed (N= 321) (N= 321)
N % N %
1996-97 Violators 212 66.0 109 34.0
Others 101 31.5 220 68.5

 

a. Cut-oil value rs 0.50.

Figure 3 is a histogram of predicted probabilities comparing the distribution of the
observed and predicted group membership. Nearly half of the other snowmobilers
incorrectly classiﬁed as snowmobile law violators fell between a predicted probability of
0.5 to 0.59, while less than 10% of other snowmobilers have predicted probabilities of 0.7
or better (Table 48). There is a sirrrilar pattern for 1996-97 convicted snowmobile law
violators as nearly 80% of misidentiﬁed convicted snowmobile law violators had
predicted probabilities between 0.30 to 0.48 with about equal proportions classiﬁed
between 0.30-0.39 and 0.40-0.48. This distribution confrrms the ﬁndings from the

119

+HHH+HHH+HHH+HHH

82335 833 Be 9.80 8288 do Sameam .m 2&3

.880 mm; ﬁgmoaom 32.—Sm nonm—
ﬁoﬁ—oE 33 3305305 38:68 363— I >
ﬁBEoEBoE .850 I Z ”235%
.2. m_ .8333 32 ozpoEBoam 33:68 3-32 a ma QEEonEoE tom—338a 38:65 05 SH 33> been". on...

unsouo
H m. m. b. m. m. o. n. N. H. o "noun
.......... +---------+------u--+------r--+---------+---------+---------+---------+---------+--------- couuHooum
z 2 H
2 H
g>>§>§§>§z§§zz>zzz§>zz>>zzzz>zz§z>zz§zzzzzzzzzzzzz>zzzzz H
>>>>> >z>>>>>z>>>>>>z>zzzz>>zz>>z>>z>>2222>zzzz>zzzzz>zz>zzzzzz>zz +m
>>>> >>>>>>>>>>>>>>>>zz>>z>>z>>z>>zzzz>zz>z>zz§>zz>zz>zzz>zz H Hr
>>> >>>>>>>>>>>>>>zz>>z>>>>z>>>>>z>z>>>zz>z>zz>z>zz > H 0
>> >> >> >>>>>>z>>z>>> >>>>>>>>>>>>z> >2 H z
> >> >>>>> >>>> >>>>> >>>>z> z +OH m
> >>>>> >> >>>> >>z> > H D
> >>>>> >> > >> H O
> > > H m
> > +mH m
> H m
> H
H
+ on

120

analysis of the residuals, which suggested that a notable proportion of the 1996-97
convicted snowmobile law violators and other snowmobilers, while deﬁned as a member
of the group, can be characterized as a member of the opposite group.

Table 48. Distribution of predicted probabilities for the ﬁnal independent variables

logistic regression model using single block entry between 1996-97 convicted
snowmobile law violators and other snowmobilers.

 

 

1996-97 Violators Others

Predicted Probabilities (N= 321) (N= 321)

N % N %
0.0 - 0.09 0 0.0 2 0.6
010-019 4 1.2 33 10.3
0.20 - 0.29 19 5.9 64 19.9
0.30 - 0.39 39 12.1 61 19.0
0.40 - 0.49 42 13.1 57 17.8
0.50 - 0.59 59 18.4 47 14.6
0.60 - 0.69 52 16.2 27 8.4
0.70 - 0.79 51 15.9 20 6.3
0.80 - 0.89 49 15.3 8 2.5
0.90 — 1.00 6 1.9 2 0.6

 

In all, the c statistic indicates that in 74% of all possible pairs of cases, the ﬁve
independent variables assign a higher probability of being a 1996-97 convicted
snowmobile law violator to cases that are in fact violators. In other words, the model is

moderately better than chance at distinguishing between cases in these two groups.

Model Discrimination Summa_ry
With two-thirds of the cases correctly classiﬁed, the model is considered

moderately successful in discriminating between 1996-97 convicted snowmobile law
violators and other snowmobilers. However, a sizable proportion of the cases were
incorrectly classiﬁed. This suggests that while a given respondent may be deﬁned as
either a 1996-97 convicted snowmobile law violator or other snowmobiler, they may
characterized as a member of the opposite group at another time or in different

circumstances.

121

This may be due to a number of factors. First, by deﬁnition, 1996-97 convicted
snowmobile law violators were only deﬁned based on one winter’s behavior and their
apprehension and conviction by enforcement and judicial ofﬁcials. Hence, some
proportion who violate the law are not caught in a given year. Convicted snowmobile law
violators also range from deliberate (conscious and lacking respect for authority), to
situational (would not normally violate but circumstances inﬂuenced it), to unwitting
(would not knowingly violate). This conclusion corroborates the results of the analysis of
Studentized residuals and leverage values. A subsequent comparison of those cases with
extreme values to their respective group found many differences (Tables 41- 46). For
instance, those cases with high Studentized residuals were all 1996-97 convicted
snowmobile law violators classiﬁed as an other snowmobiler since they were older,
snowmobiled less, and were more likely to support increased law enforcement. These
present a classic picture of the unwitting violator. As for those cases with high leverage
values, these respondents were on average younger, snowmobiled more, and were less
supportive of more law enforcement than either group. These present a proﬁle of

violators, some of whom were cited and convicted in winter 1996-97.

122

CHAPTER SEVEN
SUMMARY, IMPLICATIONS, AND CONCLUSION

Stud Back und

In the 1990’s, more snowmobiling and snowmobile accidents, coupled with
expanded trail facilities, renewed snowmobiler safety concerns that had declined during
the slump in snowmobiling in the 1980’s. Attempts to curb accidents through the
institution and enforcement of safety regulations have only had moderate success, due to
limited law enforcement personnel, budget restrictions, and snowmobiler apathy toward
the regulations. The situation is further complicated by a lack of knowledge about
inﬂuences and circumstances that predispose, enable, and reinforce risky and illegal
snowmobiling behaviors. Typically, the research concerning the safety of snowmobilers
has been limited to descriptions of accident circumstances. Those few studies designed to
distinguish underlying factors that inﬂuence aberrant snowmobile behaviors are outdated
and restricted to comparisons between snowmobile accident and non-accident victims.
Accordingly, this study attempts to identify salient variables that discriminate between
known convicted snowmobile law violators and other snowmobilers to create a proﬁle of
snowmobilers who are known to violate snowmobile laws.

To accomplished this, 983 individuals cited by the DNR-LED in winter 1996-97
and convicted for a snowmobile safety violation and a sample of 3,325 other
snowmobilers who purchased a 1996-97 Michigan Snowmobile Trail Permit and were
not cited by the DNR-LED during winter 1996-97 were compared using a comprehensive
mail questionnaire. Of those surveyed, 39.8% of the 1996-97 convicted snowmobile law

violators and 48.7% of the other snowmobilers responded to the questionnaire.

123

Summm of Univariate Analysis Findings
The factors hypothesized to differentiate 1996-97 convicted snowmobile law

violators from other snowmobilers were conceptualized using the PRECEDE Model,
which postulates that our behaviors are inﬂuenced by predisposing, enabling, and
reinforcing factors Nullen et al., 1987). These potential factors were derived from
studies examining snowmobiling accidents and folk crime research comparing trafﬁc and
ﬁsh and game law violators and non-violators. Initial descriptive and likelihood ratio

statistics on the variables identiﬁed a number of distinctions among the two groups.

Predisposing Factors
In the PRECEDE model, predisposing factors refer to what an individual brings to

an experience. These would include age, knowledge, attitudes, beliefs, and perceptions
that either support or inhibit a given behavior. In this study, 1996-97 convicted
snowmobile law violators were found to be on average, nearly six years younger than
other snowmobilers. This is consistent with previous snowmobile, trafﬁc safety, ﬁsh, and
wildlife research (Dewar, 1973; Anderson, 1987; James et al., 1991; Elander et al., 1993).
Convicted snowmobile law violators also had differing perceptions of danger
relating to risky behaviors and situations. For instance, they regarded trail conditions and
the presence of non-motorized users on the trails as more dangerous than did other
snowmobiling, snowmobiling on roadways, and the presence of other vehicles on
seasonal roads as less dangerous. Such perceptions are consistent with the behaviors of
injured snowmobilers reported in past studies. They tended to operate their snowmobiles
at greater speeds and consume more alcohol while snowmobiling than those not involved
in accidents (Waller & Lambom, 1975; Rowe et al., 1993). Similar perceptions and

behaviors of violators have been noted in studies comparing trafﬁc law violators and non-

124

violators as well as those involved and not involved in an automobile accident (Evans &
Wasielewski, 1982; Donovan et al., 1983; Mortimer, 1988; Dejoy, 1992).

Along with perceptions of risk that differed ﬁom other snowmobilers, 1996-97
convicted snowmobile law violators were less supportive of selected proposed initiatives
to reduce fatalities. This was especially true for increased enforcement of existing
snowmobile regulations. This is similar to the results of Andersen’s (1987) work with
deer poachers and Cox and White’s (1988) work with trafﬁc law violators. In general,
most folk crime law violators oppose management actions, law enforcement, and statutes

regulating behaviors (Melynk, 1977; Mayer & Treat, 1977; Kanellaidis et al., 1995).

Enabling Factors

Enabling factors are depicted as objective characteristics, such as level of
participation and snowmobile attributes, that provide a potential means to deviant
behaviors. Results showed that 1996-97 convicted snowmobile law violators
snowmobiled more than other snowmobilers. This was also true in studies comparing
injured and uninjured snowmobilers (Waller & Lambom, 1975; Rowe et al., 1993).
Moreover, Gray and Kaminski’s (1994) comparative study of hunters found that violators
of wildlife laws hunted more days than non-violators. Likewise, comparative studies of
trafﬁc law violators and non-violators have often found that violators tend to drive more
miles in a given year (Reason et al., 1990; Kanellaidis et al., 1995).

While there is considerable debate about the role high performance snowmobiles
play in contributing to deviant behaviors and accident proneness, these data indicate that
1996-97 convicted violators own and use more powerful snowmobiles than their
counterparts. While higher horsepower snowmobiles differentiated injured and non-
injured snowmobilers in Waller and Larnbom’s (1975) study, Rowe and associates (1993)

did not ﬁnd a signiﬁcant difference. Higher performance being associated with accidents

125

also has some support in automobile (Rothengatter, 1988) and motorcycle (Jonah et al.,

1981) literature.

Reinforcing Factors

Reinforcing factors are those that stimulate behavior, whether as an incentive or
pmrishment. These might include personal support networks, devotion to the activity, and
education related to the activity. Of the four variables hypothesized to differ among
groups, two were deemed insigniﬁcant. Curcione (1992) and Forsyth and Marckese
(1993) found that hunting violators were not only introduced to the activity, but exposed
to deviant values and situations, by family members. Consequently, the proportion of the
household 12 years and older that snowmobiled during the winter of 1996-97 was
hypothesized to be greater among convicted snowmobile law violators than other
snowmobilers. However, no statistical difference was found between the groups.
Secondly, while there are conﬂicting conclusions among researchers, wildlife law
violators are considered to be more devoted to their sport than non-violators since they
practiced more and utilized more tools and information in hunting (Jackson et al., 1979).
The total number of snowmobiles owned by the household and total annual household
expenditures on snowmobile equipment, maintenance, insurance, and storage were used
to represent devotion to the sport of snowmobiling. Of these, total number of household
snowmobiles was similar among the groups and thus statistically insigniﬁcant. However,
1996-97 snowmobile law violators were found to have signiﬁcantly higher total annual
household expenditures than other snowmobilers. This contrasts with Shafer and his
colleagues’ (1972) ﬁndings of no difference in hunting equipment investments. In
addition to the signiﬁcant difference in spending, 1996-97 snowmobile law violators had

fewer household members completing a snowmobiling safety class.

126

Summgy of Logistic Regression Analysis Findings
To fulﬁll the ﬁrst two objectives of this study, a series of logistic regression

models and diagnostics calculations were conducted. The regressions used the signiﬁcant
hypothesized variables identiﬁed in the univariate analysis to ascertain which best
differentiated 1996-97 convicted snowmobile law violators from other snowmobilers and
predicted their group membership. Based upon this analysis, a total of six independent
variables were found to best distinguish 1996-97 convicted snowmobile law violators
ﬁom other snowmobilers.

Given its signiﬁcance in the descriptive analysis, the best discriminator among
these six variables was that measuring support for more enforcement of existing
regulations. Convicted snowmobile law violators were much less likely than other
snowmobilers to support increased law enforcement as a means to help reduce incidents
of fatal accidents. This coincides with the claims of other researchers that citizens who
have had contact with police and/or received a citation generally have a more unfavorable
attitude toward police ofﬁcers than those who have not had contact (Cox & White, 1988).
This may be especially relevant in this study, since over a third of 1996-97 convicted
snowmobile law violators who provided a rationale for their support or the lack of it for
more snowmobile law enforcement considered the ofﬁcer to be rude and overzealous
(Nelson et al., 1998).

Secondly, as one might expect based upon folk crime violator and non-violator
comparative studies, the next best discriminator was the age of the respondent. In this
study, 1996-97 convicted snowmobile law violators were younger than their counterparts.
Some researchers have suggested that it is not so much the age of the individual that is
the cause, but rather some combination of the younger adult’s tendency to engage in risky
and illegal behaviors and lack of experience (W allach & Kogan, 1961; Jonah, 1986). Yet,
the focus of this assertion generally revolves around the 16-25 year olds. In this

snowmobile study, the average age of the 1996-97 convicted snowmobile law violators

127

was 34 years old, 6 years younger than the other snowmobilers. Presumably this group
should be less apt to engage in such risky activities. Yet, those aged 14-25 accounted for
only 18% of the 1996-97 convicted snowmobile law violator respondents, while 26-35
years old accounted for 42% of this group. It is likely that snowmobile use and ownership
is skewed to older individuals than are driving a car or violating games laws.

Next, 1996-97 convicted snowmobile law violator households rode their
snowmobiles more often than other snowmobilers. This enabling characterization is
consistent with other comparative studies between folk crime violators and non-violators
(Waller & Lambom, 1975; Reason et al., 1990; Rowe et al., 1993; Gray & Kaminski,
1994; Kanellaidis et al., 1995).

The fourth most signiﬁcant discriminator was total household expenditures for
snowmobile equipment, maintenance, insurance and storage. Convicted snowmobile law
violators spend nearly a third more than other snowmobilers.

Fifth, 1996-97 convicted snowmobile law violators rated trail conditions as more
dangerous than did other snowmobilers. Interestingly, in the initial backward variable
selection LR model, the speed of snowmobiles was another signiﬁcant variable that was
considered less dangerous by 1996-97 convicted snowmobile law violators. While this
variable was found to be an insigniﬁcant discriminator in the single block entry model,
one could argue that 1996-97 convicted snowmobile law violators perceptions about the
speed of snowmobiles could dictate to some degree their perception of trail conditions.
For instance, the greater the speed of the snowmobile, the more likely that imperfect
conditions will manifest themselves leading to the appearance of poor trail conditions.
Moreover, this perspective suggests that 1996-97 convicted snowmobile law violators
view accident factors beyond the operator’s control (e.g., trail conditions) as more
meaningful than those factors that result from controllable behavior choices (e.g., speed),
which is consistent with the conclusion reached by Donovan and others (1983) in a

synthesis of trafﬁc safety research.

128

Lastly, the least signiﬁcant independent variable that differentiated these two
groups was the proportion of the household completing a snowmobile safety class. In this
case, 1996-97 convicted snowmobile law violator households had a smaller proportion
completing a snowmobile safety class.

Overall, based upon the LR model, the 1996-97 convicted snowmobile law
violators were younger, snowmobiled more, spent more on snowmobiling, were more
critical of law enforcement and trail management, were more likely to perceive the
dangers of snowmobiling as someone else’s failure rather than theirs, and household
members were less likely to have taken a safety class. The model correctly classiﬁed
about two-thirds of 1996-97 convicted snowmobile law violators and other
snowmobilers. Hence, while this successful classiﬁcation rate was signiﬁcantly better

than chance, it was a less than perfect classiﬁcation tool.

Study Implications and Recommendations

Two fundamental implications emerge from the univariate and LR modeling
analyses differentiating 1996-97 convicted snowmobile law violators from other
snowmobilers. First, many of the same variables that were signiﬁcant discriminators of
these two groups parallel those distinguishing snowmobilers involved in snowmobiling
accidents from those not involved (Waller & Lambom, 1975; Rowe etal., 1993). This
study suggests a link between many snowmobiling accidents and risky and illegal
behaviors (Rowe et al., 1993). Furthermore, these factors appear to mirror many of the
characteristics differentiating other folk crime law violators from non-violators
suggesting that similar inﬂuences and circumstances may be at the foundation of deviant

behavior.

129

Secondly, this study also provides vital and timely knowledge for those
administering snowmobile programs. The distinguishing variables, particularly the six
variables gleaned from the LR modeling analysis that best differentiated these two groups
of snowmobilers, provide invaluable insight into the fundamental inﬂuences and
circumstances snowmobile program administrators face in safeguarding snowmobilers,
bystanders, and the environment. As a rule, neither of the six variables necessarily
improves law enforcement’s ability to identify potential snowmobile law violators in ﬁeld
settings, however they still present administrators with some useful information in
developing intervention strategies that may help mitigate risky and illegal snowmobile
behaviors.

Just as this research has shown that a snowmobiler’s choice to engage in risky and
illegal behaviors arises from a diversity of interconnected factors, any intervention
strategies devised to alleviate such behaviors must not only be integrative and
comprehensive, but must also mirror these interactive factors (Rowe et al., 1993).
However, a few points regarding these convicted violators need to be clariﬁed.

First, snowmobile law violators are a minority of total snowmobilers. Based on
the number of citations issued, most snowmobilers conduct themselves in an appropriate
manner while snowmobiling. Secondly, the deﬁnition used to delineate 1996-97
convicted snowmobile law violators from the other snowmobilers was based upon one
winter’s behavior as well as their apprehension by Conservation Ofﬁcers and conviction
by judicial ofﬁcials. Given the constraints facing law enforcement, data for a single
season cannot ‘catch’ all the law violators. Consequently, those other snowmobilers,
while depicted as ‘non-violators,’ were characterized in the LR modeling process as
convicted snowmobile law violators and thus may represent ‘potential violators’ that
were not caught in winter 1996-97. Lastly, research from other folk crimes has revealed,
as data from this research suggest, that the act of violating arises from a host of

underlying motives and situational circumstances (Gibbions, 1983; Gramann & Vander

130

Stoep, 1987; Reason et al., 1990; Goldstein, 1996). As a result, there is a wide variety of
circumstances associated with violators. For instance, some violations may arise because
the snowmobiler is unaware of the regulations, such as operating a snowmobile in a
restricted area that is not adequately posted. Moreover, there are others that may be aware
of the regulations but unwittingly deviated ﬁ'om them, such as operating with a tail light
burned out. Still further, there are those that willingly violate regulations, but rationalize
it as necessary because they saw no other alternative. An example would be
snowmobiling on a public road, even if illegal, in order to obtain gasoline. Such sorts of
violations might explain some of the 1996-97 convicted snowmobile law violators that
did not ﬁt the model well, particularly the 45 cases with high Studentized residuals.
These 1996-97 convicted violators were older, less active in snowmobiling, and were
more likely to support increased law enforcement. In addition to these violators, there are
those that are clearly deliberately intentional, such as racing down a road merely to see
who can go the fastest. Such violators appear to correspond with those 1996-97 convicted
snowmobile law violators with high leverage values. They were on average younger,
more active in snowmobiling, and were less supportive of more law enforcement than
1996-97 convicted snowmobile law violators as a whole. As a result of these different
types of violations, intervention strategies must cover the gamut, targeting the unwitting
violators, those that violate only by “necessity” (situational), and the willful violator
(Rowe et al., 1993). In addition, these intervention strategies must be integrative so that a
number of tactics are being used concurrently. Such intervention strategies might include
dissemination of information, education, law enforcement, legislation, and management

actions.

131

Dissemination of Information

Communicating information regarding snowmobile safety regulations, the
consequences of risky and illegal snowmobiling activities, and the details of snowmobile
accidents would be useful to both unwitting and situational snowmobile violators to help
acquaint them with the hazards of snowmobiling and perhaps stimulate them to engage in
more acceptable legal and safe snowmobiling behaviors. This is because 1996-97
convicted snowmobiler law violators generally had inaccurate perceptions about the
dangers of risky snowmobiling behaviors and situations. Moreover, it appears that they
are not necessarily so cognizance or adept as other snowmobilers at recognizing the
inherent consequences of their actions. Which may perhaps in part be related to a lack of
knowledge about the situation.

There are a number of mechanisms to disseminate such information. For instance,
in order to operate on public lands and frozen waters in Michigan, all snowmobiles must
have an annual Michigan Snowmobiling Trail Permit. The point of sale for these permits
provides a good opportunity to distribute a concise outline of snowmobiling regulations
and other pertinent information, such as a summary of fatal accident circumstances from
the pervious year. Presently, no such information is provided. It could be modeled after
hunting and ﬁshing pamphlets. This information booklet should also be provided with
any snowmobile registration. The booklets should also be made available for distribution
to business establishments, such as restaurants, sporting good shops, hotels, gas stations,
etc.

Since many younger snowmobilers have access to and readily utilize the intemet,
information such as the consequences of risky and illegal snowmobiling activities and the
details of snowmobile accidents can be published on web pages. Another underutilized
means of getting the message out could be through the use of billboards and other
promotional materials made available to establishments that cater to snowmobilers. The

International Snowmobile Manufacturers Association (ISMA) has developed a safety and

132

rider responsibility campaign using posters and other literature. These are made available
to the public free of charge. Regulatory agencies and the manufacturing industry could
combine efforts and jointly publish such information with other snowmobile
organizations and clubs. The ISMA has also developed a short safety video entitled ‘Safe
Ridersl,’ that provides a basic overview of riding responsibilities and proper techniques.
Given the number of snowmobile clubs and organizations around the state that could gain
access to schools and civic organizations, this could provide another positive means of
getting the message out, especially if the message was tailored to Michigan
snowmobiling.

The DNR-LED has also undertaken an initiative to improve snowmobile accident
reporting so that more detailed and timely information is gathered. All available efforts
should be made to summarize this information and make it more readily available,
especially to the media including newspapers, magazines, and television. Other
opportunities to promote safe snowmobiling exist through television weather forecasts
providing a snowmobile safety tip or a public service announcement at the conclusion of
their weather forecasts. While this is not an exhaustive list of mechanisms to disseminate
information, it offers a set of ideas that could be incorporated relatively inexpensively.
The key is to explore and develop partnerships with communicators of information to

effectively disseminate information to the snowmobiling public.

Education

Besides disseminating information, a formal education program can also serve as
a mechanism to positively inﬂuence and perhaps modify predisposing factors, especially
for unwitting and situational violators. This is especially relevant since the LR model
found that 1996-97 convicted snowmobile law violators were generally younger and had
fewer household members who had completed a snowmobile safety course. Moreover, a

majority of 1996-97 convicted snowmobile law violators (64%) and other snowmobilers

133

(72%) supported mandatory snowmobile safety education and certiﬁcation. While the
State of Michigan currently offers a snowmobile education program, it is voluntary and
targeted at snowmobilers under 18 years old. Given that Michigan requires all individuals
to pass an education/training course to operate an automobile and to hunt, it seems
reasonable that similar requirements be extended to all ﬁrst time snowmobilers regardless
of age. Hunter education, which is given partial credit for reducing hunting accidents
state and nationwide, provides both information and positive reinforcement to encourage

good or proper behavior. The snowmobile program should be modeled after this program.

Law Enforcement

Enforcement of regulations is intended to control and eliminate illegal behaviors
by encouraging individuals to act responsibly, while making explicit the prohibitions
against and the consequence of undesirable behaviors (Anderson et al., 1998). Because a
uniformed law enforcement ofﬁcer is easily recognizable in recreation settings, they are
generally believed to be effective deterrents to non-compliant behaviors (Rowe et al.,
1993; Swearingen & Johnson, 1995; Anderson et al., 1998; Duda et al., 1998). As a
result, enforcement of safety regulations is often one of the most popular and common
responses to modifying risky and illegal behaviors in recreation settings (Rowe et al.,
1993). In fact, when the other snowmobilers were asked to rate possible initiatives to
reduce snowmobiling fatalities, 60% were strongly or moderately supportive of more
intensive enforcement of existing snowmobile regulations. Other snowmobile studies
have noted the same desires (Rowe et al., 1993; McElvany, 1994) as have studies of
hunters and ﬁshermen (Purol etal., 1978; Responsive Management, 1993; Gray &
Kaminski, 1994).

134

In many regards, law enforcement is an effective approach for the unwitting and
situational violators as well as the willful violator. Studies of trafﬁc violations have
shown that the visible presence of a law enforcement ofﬁcer reduces incidences of
violations, but as visibility is reduced violations increase to their former levels (Shinar &
McKnight, 1985). The study of Michigan snowmobiling during the winter of 1996-97
found that nearly a third of the households reported that at least one member was stopped
by a law enforcement ofﬁcer, while over half of the households had at least someone who
saw an ofﬁcer while snowmobiling (Nelson et al., 1998).

In spite of their apparent visibility, there are still challenges for snowmobile law
enforcement. For example, limited budgets and personnel, coupled with a massive land
base, limits the time that any given trail or frozen lake can be patrolled. In 1999-2000,
increased funding for the Michigan Snowmobile Enforcement Grant Program, which
provides matching monies to county sheriff, city, village, and township police
departments to assist their enforcement efforts, increased local law enforcement presence.
Perhaps some of this money should be used to develop memorandums of agreements or
understandings with these other jurisdictions to create joint data bases and sharing
information.

In other recreation settings, non-enforcement trained volunteer safety patrols have
been used effectively to aid in controlling and eliminating non-compliant behaviors (East
Bay Regional Parks, 1991; Moore, 1994). In Michigan, a volunteer Off-Road Vehicle
(ORV) Safety Patrol has been established for some ORV trails. This trained group has no
law enforcement authority, but does act as law enforcement eyes and ears with the ability
to contact ofﬁcers and provide on-site information and education resources. This program
could be adapted to snowmobiling. In some cases, these programs do provide targeted,
limited law enforcement authority, such as the Snowmobile Trail Ofﬁcer Program
(STOP) in Sudbury, Ontario, Canada. In this program, individuals are given extensive

135

training and are designated as special constables and Provincial Offenses Ofﬁcers to
assist law enforcement ofﬁcers on the snowmobile trails (Rowe et al., 1993).

In spite of visible enforcement efforts, law enforcement alone may not necessarily
be effective with the willful violator group. They were much less supportive of law
enforcement than others and tended to avoid times or areas where law enforcement
ofﬁcers are patrolling. Consequently, low proﬁle enforcement, such as undercover
operations where ofﬁcers patrol without uniforms or marked snowmobiles, may be
necessary. Additionally, ofﬁcers may be able to use surveillance tactics in known trouble
spots to apprehend willful violators.

Moreover, one could argue that some of this disrespect towards law enforcement
may be related to violators’ treatment by ofﬁcers. As noted previously, over a third of
1996-97 convicted snowmobile law violators who provided a rationale for their rating of
current snowmobile law enforcement considered the ofﬁcer who contacted them to be
rude and overzealous (Nelson et al., 1998). The image and effectiveness of law
enforcement go hand in hand. As a result, law enforcement ofﬁcers should make every
possible effort to develop a positive approach to contacts rather than alienate
snowmobilers.

They can also use positive reinforcing tactics. For instance, law enforcement
ofﬁcers in Sudbury, Ontario rewarded non-drinking snowmobilers with license holders
bearing the message “thanks for being a safe and sober snowmobiler” (Rowe et al., 1993).
Similar means can also be applied with those who are stopped and found to have been
obeying all laws. Additionally, expanding the efforts of law enforcement should be tied to
a safety promotion campaign. For example, changes in Michigan’s drinking and driving
laws in 2000 were accompanied with an extensive public announcement program. There
have been similar updates in Michigan’s snowmobiling law, including points added to the
driver’s license and/or the conﬁscation of driver’s license for those snowmobilers

convicted of operating under the inﬂuence of alcohol.

136

Legislation and Regulations

An important component to ensuring the safety of snowmobilers has been and
will continue to be the institution and improvement of operation and safety regulations
(U SDI, 1971). Yet, merely imposing regulations does not guarantee their success in
modifying risky and illegal behaviors. Pertinent regulations, particularly those that are
linked to direct beneﬁts, such as increased snowmobiler safety or the elimination of a
particular behavior that interferes with the freedom and enjoyment of others, would be
more accepted (Lucas, 1983). Communicating the rationale and informing all of the
regulations, especially new ones, is also paramount in helping gain compliance
(Anderson et al., 1998). Consequently, imposing new regulations should be part of an
overall comprehensive intervention strategy that needs to include an information
campaign and consistent enforcement. Finally, it also helps if the regulation is targeted at
a particular group that is properly deﬁned. For example, a speed limit regulation may
target a designated trail system and those snowmobilers who engage in racing there.
Possible new legislation and regulations might include tougher penalties for selected
violations, requiring personal liability insurance, and imposing speed limits at deﬁned
locations that would be applicable for all violators, but especially the willful violator.

One of the most frequently cited factors associated with snowmobiling accidents
has been alcohol use (Wenzel & Peters, 1986; James et al., 1991; Rowe et al., 1992;
Rowe et al., 1993; Avis, 1994; Rowe et al., 1994; Nelson et al., 1998). Based upon the
data in this study, it appears that most snowmobilers are aware of the dangers of
operating a snowmobile while intoxicated. Eighty-six percent of the 1996-97 convicted
snowmobile law violators and 89% of the other snowmobilers rated it as extremely or
highly dangerous. Moreover, 30% of the 1996-97 convicted snowmobile law violators
and 38% of other snowmobilers also considered the operation of a snowmobile by an
individual who has been drinking, but was not intoxicated, as extremely or highly

dangerous.

137

Rowe and colleagues (1993, p. 22) assert that snowmobilers “have a fair grasp of
alcohol effects and generally are aware of the risks associated with snowmobiling and
drinking.” However, they go on to conclude that this apparent knowledge does not
necessarily translate into appropriate behaviors and that simply educating snowmobilers
about the effects of alcohol use will not necessarily bring about a behavior change (Rowe
et al., 1993). Many of the other snowmobilers in the current Michigan study suggested
imposing tougher penalties, such as adding points to the driver’s license and conﬁscation
of driver’s license or snowmobile for those convicted of operating under the inﬂuence of
alcohol (Nelson et al., 1998). Imposing stricter penalties has been found to be effective in
deterring drinking and driving behaviors (Rowe et al., 1993) and with serious hunting
violations (Hall, 1992). As a result, Michigan Public Act 451 (Natural Resources and
Environment Protection Act), was amended by Public Act 43 of 1999, to allow law
enforcement ofﬁcers to conﬁscate a snowmobiler’s driver’s license if they were operating
a machine under the inﬂuence of alcohol. In addition, Michigan Public Act 300
(Michigan Vehicle Code Act) was amended by Public Act 21 of 1999 to add points to a
snowmobiler’s driver’s license for operating under the inﬂuence of alcohol while
snowmobiling. While such regulations are an improvement, perhaps they should be
extended to other moving violations, such as operating on a roadway, or speciﬁc road
right-of-way violations, such as operating against ﬂow of trafﬁc, not in a single ﬁle, not
at the extreme right, etc.

In cases where the adoption of a new regulation is necessary to help modify risky
and illegal behaviors, it is suggested that the regulations’ effect be broadly targeted at
those who regularly violate the law (Anderson et al., 1998). For instance, penalties added
to driving records would not only affect an individual’s ability to drive, but they would
also affect their insurance rates. Allowing the market forces in the insurance industry to
ﬁnancially discourage bad behavior may help alleviate problems. Many consider this a

viable means to achieving an end, as 45% of the 1996-97 convicted snowmobile law

138

violators and 59% of the other snowmobilers strongly or moderately supported
mandatory personal liability insurance for all who snowmobile in Michigan.

Another frequently cited, but less understood, factor associated with
snowmobiling accidents is the role of speed. (Waller & Lambom, 1975; James et al.,
1991; Rowe et al., 1993). In this study, the danger associated with the speed of
snowmobiles was noted as being extremely or highly dangerous by 34% of the 1996-97
convicted snowmobile law violators and 48% of the other snowmobilers. Furthermore,
36% of the 1996-97 convicted snowmobile law violators and 65% of the other
snowmobilers noted speed as a factor contributing to a snowmobile accident that they
knew about. However, the challenge of adopting speed reducing regulations was made
clear when snowmobilers were asked about their support for establishing an enforced
snowmobile speed limit on public land, trails, and waterways. Only 25% of the 1996-97
convicted snowmobile law violators and 41% of the other snowmobilers strongly or
moderately support the speed limits. Nevertheless, establishing a speed limit on
snowmobile trails, especially in accident prone locations, may be necessary (Hermance,
1995). This topic is currently under review by Michigan’s DNR and legislative bodies
(Lansing State Journal, 2000).

Management

As in any recreation setting, proper design and maintenance of facilities are
essential for safety and resource protection. Yet, these activities can also help minimize
risky and illegal behaviors through limiting or directing behavioral options. Such
activities would be especially useful for targeting the unwitting and situational violators.

Of the types of moving citations issued to 1996-97 snowmobile law violators,
seven out of ten were roadway related infractions, such as operating on a public roadway,
operating against ﬂow of trafﬁc, not in a single ﬁle, or failing to be to the extreme right of

the roadway going with the ﬂow of trafﬁc. In addition, when asked about the dangers of

139

operating on state or county roads, only 10% from both groups considered it to be
extremely or highly dangerous. This is similar to rider perspectives about the
circumstances related to a snowmobile accident. Convicted 1996-97 snowmobile law
violators cited road riding as a factor in 15% of the accidents of which they were aware
and other snowmobilers cited it in about 18% of accidents. These perspectives are counter
to the details contained in 146 police reports of Michigan snowmobile fatalities from
1993-1997. Review of these reports found that nearly half of the accidents began or
occurred on a public roadway (Nelson et al., 1998). This clearly suggests a false sense of
security among snowmobilers as few perceive danger in operating on roadways.

As of 1999, Michigan laws governing snowmobiles permit them to operate on the
right-of-way of a plowed state highway, except along limited access highways. In
addition, counties or cities may designate rights-of-ways on plowed roadways for use by
snowmobilers. Based upon the analysis of the 146 Michigan fatal accident reports, it
appears that the hazards pertaining to roadways are related to their inherent characteristics
(Nelson et al., 1998). For example, Michigan roadways are relatively ﬂat, straight, wide-
open places that provide an opportunity for a snowmobiler to rapidly attain high levels of
speed. In contrast, most snowmobile trails are narrower, follow the contour of the land,
and change direction often. They thereby discourage snowmobilers from attaining high
speeds. The exception to this is rail trails, which often mimic the characteristics of roads
without the automobile trafﬁc. Much of Michigan’s designated and groomed snowmobile
trail system relies on unplowed state and national forest roads. However, plowed county
roadway shoulders are often used to create connections to other snowmobile trails and
towns allowing access to food, fuel, and lodging services. With the improved comforts of
snowmobiles and an established network of trails, long distance, multi-day snowmobile
trips have become an increasingly typical pattem. This is likely to continue in the future.
Development of connecting trails to towns and other trail systems is also likely to

increase as communities recognize the ﬁnancial beneﬁts of providing services (Stynes et

140

al., 1998). Yet, since new trail construction costs are high and trails cross both public and
private lands, development of new connecting trails has lagged behind snowmobiling
activity (Nelson et al., 1998). The draft Michigan Snowmobile Plan advocates additional
safer snowmobile trail connectors to services and towns (MDNR, 1998). This includes
construction of additional trail connectors on public lands and securing easements where
necessary to cross private property.

Managers also have some control over trail conditions. While trail conditions
depend upon many factors, such as snowfall and weather, grooming is also a strong
inﬂuence. Based upon the LR model, rider assessment of trail conditions was one aspect
that differentiated 1996-97 convicted snowmobile law violators from other
snowmobilers. The convicted violators rated trail conditions as more dangerous than
other snowmobilers. In addition, they considered the trail conditions more likely to be a
factor associated with accidents than other snowmobilers. While improving trail
conditions may help reduce the risk of accidents, it could also be counterproductive since
well groomed, wide, straight trails may also contribute to increases in speed and accident
risk. Hence, proper trail grooming that provides safe, enjoyable riding conditions while
limiting speed is desirable.

Research Recommendations

As of the late 1990’s, research concerning the safety of snowmobilers has been
almost exclusively focused on describing the nature of snowmobile injuries and the
circumstances surrounding snowmobiling accidents (Rowe et al., 1993). Several studies,
however, have considered how factors, such as a snowmobiler’s attitudes, perceptions,
and other characteristics, are related to the likelihood of one being involved in an accident
(Waller & Lambom, 1975; Rowe et al., 1993). These studies have entailed comparing
those involved in a snowmobile accident with those not involved. While such

comparative studies have begun to identify factors that inﬂuence behaviors contributing

141

to a snowmobiler’s risk of an accident, they are limited and all have applied univariate
analysis between factors. Comprehensive comparative studies between those who violate
folk crime laws and non-violators have revealed a host of distinguishing factors
associated with aberrant behaviors and accident involvement. Accordingly, this research
aimed to identify differences between 1996-97 convicted snowmobile law violators and
non-violators using a comprehensive conceptual model and logistic regression analysis.
This type of analysis allows researchers to ﬁnd the most parsimonious model to describe
the relationship between a dependent variable and a set of independent variables.

In this study, the PRECEDE Model (Green et al., 1980) was used to ascertain
differences that predispose (e.g., attitudes), enable (e. g., snowmobile characteristics), and
reinforce (e.g., support networks) those that violate safety regulations. In this study, the
PRECEDE model and analysis proved moderately effective in identifying key inﬂuences
and circumstances that contribute to risky and illegal snowmobiling behaviors.
Consequently, future endeavors aimed at identifying distinctions between accident
victims or violators and their counterparts might ﬁnd this model useful. Yet, as with any
research, there is room to improve methodology and answer new questions brought about
by data analysis. For example, examining different ways to delimit violators and non-
violators, segmenting different types of violators, and gathering additional information

distinguishing snowmobile law violators and other snowmobilers.

Delimiting Violators and Non-Violators and Characterizing Different Violators
One of the challenges of this research pertained to the limited deﬁnition used to

segment 1996-97 convicted snowmobile law violators and other snowmobilers. In this
study, 1996-97 convicted snowmobile law violators were deﬁned as those who were cited
by Michigan DNR-LED Conservation Ofﬁcers during the winter season of 1996-97 and

convicted of violating operation or safety regulations. Conversely, other snowmobilers

142

were all those who purchased a Michigan Snowmobile Trail Permit and were not cited by
Conservation Ofﬁcers for a snowmobile safety violation during the winter 1996-97.

Applying such a segmentation scheme creates a few research challenges. First,
these 1996-97 convicted snowmobile law violators include only those that were caught
and cited by Michigan DNR Conservation Ofﬁcers and later convicted. Those cited by
County Sheriffs and Michigan State Police were not included as there is no central data
base of all these citations. Secondly, the 1996-97 convicted snowmobile law violators
were only deﬁned based upon one winter. Intuitively, some proportion of snowmobilers
who violate the law are not caught in a given year. This means that some of the other
snowmobilers, while characterized as non-violators during winter 1996-97, violated, but
were not cited. This may explain why some of the other snowmobilers were
characteristically similar to the 1996-97 convicted snowmobile law violators. Thirdly,
and perhaps somewhat more challenging, characterizing snowmobiler law violators based
on a single conviction does not necessarily constitute an individual as a high risk or
regular willful law violator.

There is a wide range of reasons and situations leading to law violation. For
instance, there are willful violators (conscious and lacking respect for authority),
situational violators (would not normally violate but circumstances inﬂuenced it),
unwitting violators (would not knowingly violate) and those who do not violate. These
differences in violators can and do, as this research indicates, affect the distinguishing
variables identiﬁed and the results of the logistic regression analysis. For example,
comparison between those cases with extreme residual values within the groups found
many differences. For instance, those cases with high Studentized residuals were all
1996-97 convicted snowmobile law violators classiﬁed as an other snowmobiler. They
were older, snowmobiled less, and were more likely to support increased law
enforcement and, thus, they are likely unwitting violators. As for those with high leverage

values, these respondents were on average younger, snowmobiled more, and were very

143

unlikely to support more law enforcement. As a result, they are considered to be likely
willful violators who where either caught or avoided a citation by LED Ofﬁcers in winter
1 996-97.

These segmentation scheme issues highlight two areas of opportunity for further
research work. First, applying other approaches to characterize violators from non-
violators could provide a better portrait of those who are likely to engage in risky and
illegal snowmobile activities. Typically, other folk crime researchers, particularly in
trafﬁc safety, deﬁne law violators based upon a speciﬁed quantity of law convictions or
the seriousness of their violation(s). A deﬁnition involving quantity is used because
traﬂic safety researchers generally feel that some driving violations are unintentional
errors and this is one way to lessen their inﬂuence (Reason et al., 1990). Perhaps other
natural resources related citations (e.g., ORV, hunting, ﬁshing) or driving citations could
be used in conjunction with the snowmobile violation conviction to deﬁne violators.

This presents the challenge of developing a data base of citations for natural
resource based recreation violations. While all the citations issued by DNR-LED
Conservation Ofﬁcers are kept on ﬁle in one location, they are not catalogued in an
electronic data base. Moreover, no attempt is made to link the DNR citations to driving
records. However, with the recent mandate adding points to driver’s licenses for
snowmobile law infractions, this connection is slowly being made. Besides this, the
driver’s license point system used by the Secretary of State to evaluate driver records can
also be used to develop a better means of segmenting violators from non-violators.

Lastly, another means trafﬁc safety researchers have used to classify violators
from non-violators has been through a series of questions where respondents rate their
probability of engaging in a set of trafﬁc offenses and deviant behaviors (Lawton et al.,
1997a; Lawton et al., 1997b). These questionnaires have proven useful in segmenting
drivers. The point system and questionnaire of trafﬁc offenses and deviant behaviors also

have been used to segment different types of violators (Reason et al., 1990; Meadows et

144

al., 1998). In such cases, quantity and seriousness of various trafﬁc offenses are used to
develop a composite picture of various segments of violators. These scales are then
applied to models to develop distinct proﬁles of the different types of violators. Using
such information, when available, to segment snowmobile law violators into different
types of violators would identify pertinent segments useful to snowmobile law

enforcement and managers.

Additional Information

While this study attempted to elicit a gamut of inﬂuences associated with aberrant
snowmobiling behaviors, more information could provide more accurate and detailed
segmentation of the snowmobile law violator. For instance, trafﬁc safety researchers have
explored the role of personalities in the tendency to violate trafﬁc laws (Elander et al.,
1993). Using psychometric measurements, such as sensation seeking, irnpulsiveness, and
hostility, may also help differentiate violators from non-violators. While at a ﬁeld
enforcement level this has little practical applications in a safety training program, it may
help target efforts and key messages to potential problem riders.

Another potential area of research might include understanding snowmobilers’
knowledge about general snowmobile operation and safety regulations. This could prove
useful to understanding unwitting violators. This information could also be used to
improve safety education. Such a study might also involve following individuals who
completed a snowmobile safety course for a period of time to answer the question of
whether those who complete safety training are less likely to violate or have an accident.
While this has been determined for hunter safety, it has yet to be studied for snowmobile
safety as mandates for safety education are more recent or, in some jurisdictions, non-

existent.

145

In addition to this, few data in this study were collected about speciﬁc riding
habits of snowmobile law violators, such as primary riding party composition, location
where they typically snowmobile, places they frequent, time of day and week they ride,
other activities they engage in while riding, etc. Such information could help law
enforcement better focus their patrols.

Other useful data that could aid in developing better promotion campaigns would
be information about the types and sources of information snowmobile law violators use.
Intervention messages could then be targeted at these channels.

Finally, another aspect that trafﬁc safety researchers have recently begun to
identify as they develop proﬁles is correlating violations with a variety other risky
deviant behaviors, such as stealing, ﬁghting, etc. (West et al., 1993; Lawton et al.,
1997a). Criminal checks for warrants, automobile driving records, or other activities such
as hunting or off road vehicle violations could prove useful in understanding the

relationship of snowmobile law violators to other illegal behaviors.

9mm

Snowmobiling has changed considerably since its inception in the 19605,
explosive growth in the 19708, decline in the 19808, and rebound in the 19905. The
development of more powerful snowmobiles and better suspension has given riders
quicker acceleration and improved the ease of handling, stability, and comfort. Besides
improvements to snowmobiles, designated trail opportunities also have increased,
providing an extensive network connecting snowmobilers to communities providing
services. Increasingly, snowmobilers are traveling further over several days in family and
other groups. Yet, along with the improvements in snowmobile technology, increased
participation, and more designated opportunities, an increasing number of states have
reported increases in the number of snowmobiling accidents, particularly fatalities.

Curtailing such snowmobile accidents is a fonnidable task, given the magnitude of the

146

enforcement situation and the perspective some individuals take toward risky
snowmobiling behaviors and the regulations governing them. It is imperative that all
involved in alleviating risky and illegal behaviors which lead to accidents, including
snowmobilers themselves, have an adequate knowledge of the key factors affecting
snowmobilers’ risky and illegal behaviors. While this research provides a baseline of the
inﬂuences driving risky snowmobiling behaviors and circumstances, much more needs to
be done. As snowmobiling moves into its ﬁfth decade, it is likely to continue to evolve.
As such, it will be critical for trail managers and law enforcement personnel to regularly
monitor changes in snowmobilers, machines, and legal and illegal activity patterns and
seek ways to enhance rider safety, reduce conﬂicts, and protect the natural and social

environment.

147

APPENDICES

148

APPENDIX A

MICHIGAN SNOWMOBILE QUESTIONNAIRE

149

 

\Ill lllgnll \mmnmlilllllg I‘llitNliHllllJll't‘

Please complete the following table to describe the people in your household and their involvement in snowmobiling,

If there are no people in a certain category, please use 0 for that category.

Number in Num who snowmobiled Num. who snowmobiled in Num. who completed
Household winter of 1996-97 MI winter of 1996-97 snowmobile safety class

 

Adults 18 or older

 

Children 12-17

 

 

Children li-younger

 

 

 

 

 

 

 

[If no one from your household snowmobi_led in Michigan in winter 1Wease skip to QUESTION 16.
2.

 

 

Model Year

For each snowmobile owned in you household. please complete the table below.

CC of 1996-97? sticker?

            

  

 

Please esnmte, using the table below, how my MACHINE DAYS during the winter of 1996-97 you
and household members operated your snowmobile(s) (in question #2) in these ﬁve regions of Michigan.

Please consider each day, or part of a day, a machine was operated as l MACHINE DAY.

ugu s when you were snowmobiling vs. using the snowmobile to support activities like ice ﬁshing.

For Examples, if household members with 2 machines snowmobiled for 3 days in the Western UP, and 1
machine was also used for I day to support ice ﬁshing, you would write 6 under Western UP
snowmobiling days and I under Western UP support days. If you didn 't snowmobile or use the machine
to support another activity in a region, please leave it blank

Snowmobile
Region # Days

Used during winter Had 1997 M1 trail Michigan Approx. # Miles driven in
MI winter 1996.97

   

Support
it Days

 

1 Western UP (West of Marquette/Escanaba)

 

2. Eastern UP (East of Marquette/Escanaba)

 

3 Northwest LP (N of Bay City/Muskegon &. W of 175/23)

 

4 Northeast LP (N of Bay City/M uskegon & E of 175/23)

 

5 Southern LP (South of City/Muskegon)

 

TOTAL MACHINE DAYS

 

 

 

 

If you had any SNOWMOBILE DAYS in Region 3 (Northwest LP), how many days were near one or
more of the towns of Traverse City, Kalkaska. or Cadillac? ....................... it days

Of the total number of SNOWMOBILE DAYS (lst colunm in Question 3 above). how many were:

 

Days involving overnight stays from your permanent home ....................... # days
Days not involving stays away from home but traveled

100 or more miles one way to snowmobile ........................... # days
Days not involving overnight stays and less than 100 miles from home .............. # days
TOTAL MICHIGAN SNOWMOBILE DAYS (same it as Q3 1st column) .......... # days
How many of the TOTAL SNOWMOBILE DAYS (same it as Q3 1st column) were mostly spent on the
public snowmobile trail system (trails marked with orange diamonds)? ............ # days

How many gallons of gasoline did your household use in all of its snowmobiles in Michigan during winter

of 1996-97?
it gallons

150

 

 

 

 

way from home or snowmobiled more than i an Ice (one way) from your permanent home. please answer

you reported a winteWMlchlggn snowmobile outing in response to question 4 when you stayed overnight
questions 7-13 about your MOST RECENT outing. If not, pleas skip to QUESTION 16.

 

 

 

8. What were the date(s) of the MOST RECENT outing? ................. From To
9. In what Michigan County (or near what town) was it prirmrily focused? . . . .
How many miles was this from your principle home? ................... miles
10. How many people from your household .............................. it
11. How many of your snowmobiles it were involved during that MOST RECENT outing?

12. How many MACHINE DAYS were the snowmobiles used for snowmobiling (not as support for fishing)
during that MOST RECENT outing? ................................

13. Was there an overnight stay away from your permanent home? ........... Yes No

if you stayed overnight away from your home. where did you stay? Please check all that apply.

 

Own Second Home Motel / hotel / rental cottage or cabin
Camped At friend’s lrelatives
Other: Please explain

 

14. How much money did your household spend on the entire trip during that MOST RECENT outing?
Please complete the table below for spending at home in preparation, traveling to and from the snowmobile
area, and in the local area where you rode. If you spent nothing on a item, please leave it blank.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

At Home En Route Local Area
Grocery and convenience store food and drink 5 00 S .00 S .00
Tow vehicle expenses (gasoline, repairs, etc.) 5 .00 S .00 S .00
Snowmobile expenses (gasoline. repairs, etc. 5 00 S .00 S .00
Restaurant and bar (meals and drinks) g: ,‘j‘j :‘Q'Mﬁ‘j. :‘3 S .00 S .00
Sporting goods (bait, ﬁshing tackle. etc.) 5 .00 S .00 S .00
Lodging (motel, campground, rental cabin, etc.) ﬁgmlgy S .00 S .00
All other item (ﬁlm, souvenirs, etc.) 8 .00 S .00 S .00

15. Whom did these expenditures cover? _ Your household only OR Your household + _ # others?

16. What were your household’s snowmobiling related expenses in the past l2 months in the following categories?
Purchase of snowmobile equipment (snowmobile, trailer. etc.) .......................... S .00
Snowmobile repair /. maintenance not done during a Ml snowmobile trip ................. S .00
Insurance on your snowmobile(s) ................................................ S .00
Off-season storage costs ........................................................ S .00

 

151

 

 

I Now I’d like to ask some questions about your perceptions oTM'lchlgan snowmoblTlng & its management. ]

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

17. Please rate the following services / situations regarding the Michigan DNR’s snowmobile program.
Use a scale of l to 5, with 5 as very good, 4 as good, etc. Pleas circle the appropriate number for each item.
Very Very
Services I Situations Good Good OK Poor Poor Why your ratting?
_Publlc trail groomlng 5 4 3 2 I
DNR maps of snowmobile trails 5 4 3 2 I
Public trail design 5 4 3 2 i
Trailhead parking / staging areas 5 4 3 2 i
Snowmobile law enforcement 5 4 3 2 l
Snowmobile safety education 5 4 3 2 1
18. Is the public snowmobile trail system in Michigan overcrowded? Yes No
19. Please list ONE IMPROVEMENT you would most like to see in the Michigan snowmobile trail system
20. Would you be willing to support an increase in the cost of the state snowmobile trail sticker to pay for this improvement
if it involved increased costs?
Yes No __ Undecided
21. Over winter 1996-97. more than 40 snowmobilers were killed during Michigan Snowmobiling. The DNR
is considering potential ways to reduce fatalities. Please circle your level of support for each of these on
a scale of] to 5. with 5 strongly support, 4 moderately support. etc.
Strongly Moderately Moderately Strongly
Fatality Reduction Initiative Support Support Neutral Oppose Oppose
Enforce snowmobile speed limit on public 5 4 3 2 1
lands. trails, and water
Mandatory training required for all ﬁrst year 5 4 3 2 l
snowmobile operators regardless of age
Licensing required of all snowmobile
tors 5 4 3 2 I
Mandatory personal liability insurance for 5 4 3 2 1
all snowmobile operators
More intensive enforcement of existing 5 4 3 2 l
snowmobiling regulations
22. Please list any other ideas you have about ways to reduce fatalities?
23. if implementing fatality reduction initiatives you supporwd in question 19, or your own suggestion, results in additional

costs, would you support an increase in the Michigan snowmobile trail sticker to cover that cost?

Yes No __ Undecided

152

24.

25.

26.

27.

28.

29.

30.

32.

How many times during the winter of 1996-97 in Michigan were you or members of your household checked or
stopped by a law enforcement ofﬁcer while snowmobiling? it

Did you or members of your household see, but were not stopped or checked by a law enforcement officer while
snowmobiling in Michigan during the winter of 1996.97 Yes No

Please rate the level of danger to snowmobilers from the following, based on your snowmobiling experiences.
Use a rating scale of l to 5, with 5 being extretnely dangerous, 4 highly dangerous, etc. and circle the
appropriate number for each.

 

 

 

 

 

 

 

 

 

Extremely Highly Moderately Slightly Not
Behavior / Situation Dangerous Dangerous Dangerous Dangerous Dangerous
Operation of snowmobﬂes by persons
who have been drinking, but are not 5 4 3 2 l
intoxicated
Operation ﬁncwmobﬂes by
intoxicated persons 5 4 3 2 1
Speed of snowmobiles 5 4 3 2 1
Drivers lack of skill in operating their
machines 5 ‘ 3 2 1
Cars / Trucks on seasonal roads 5 4 3 2 1
Public trail conditions 5 4 3 2 1
Public trail design 5 4 3 2 1
Other uses of snowmobiles trails (e.g.
dog sledding, X-C ski) 5 2 ‘
Snowmobiling on county / state roads 5 4 3 2 1

 

List other behaviors / situations that you feel are extremely or highly dangerous?

 

 

Do you know the circumstances of a snowmobile accident that involved an injury or fatality? Yes No

If Yes to question 26, please check each of the following factors that contributed to the accident

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Alcohol ‘Wrn-snowmobﬂer on trail

Speed Fatigue

Lack of driver skill Visibility

Snowmobile on roadway lce Conditions

Trail conditions Other (Explain)

What is your 5 digit home zipcode? 31. What is your age? years
At what age did you begin snowmobiling? years

If you have any other comments about Michigan Snowmobiling, please write them here. Thanks for your input.

 

 

 

 

Please min the completed questionnaire back to tne in the postage paid envelope provided. Thanks for you help and your time.

Chuck Nelson, Assistant Professor

Department of Park, Recreation. and Tourism Resources
131 Natural Resources Building, Michigan State University
East Lansing, Michigan 48824

(517) 353-5190

153

APPENDIX B

OTHER SNOWMOBILER COVER LETTER

154

June 3, 1997

Dear Michigan Snowmobile Trail Sticker Purchaser:

The Michigan Department of Natural Resources (DNR) and Michigan State University
are cooperating to better serve the snowmobiling public. Enclosed is a brief questionnaire
asking about your recent Michigan snowmobiling experiences, your assessment of
current snowmobiling programs, opinions about future program options and basic
information about your household.

Your responses will be critical in shaping the DNR’s future snowmobiling program. In
particular, the results will be used by trail managers, housed in the Forest Management
Division and conservation ofﬁcers, housed in the Law Enforcement Division.

Your name was chosen from DNR records of snowmobile trail stickers sold for winter
1995-96. This was the last winter for which complete information was available.
Snowmobile trail sticker sales were chosen rather than snowmobile registrations, as all
who snowmobile on public lands in Michigan, residents and visitors, are included.

Please take the 15 minutes or so necessary to complete the questionnaire. Your
participation is voluntary and you indicate your willingness to participate by completing
and returning this questionnaire. This will allow your opinions to be heard by the DNR
and other policy makers concerning future snowmobile programs. When you have
completed the questionnaire, please mail it back to me in the postage paid enveloped
provided. Your responses will be kept conﬁdential and you name will not be associated
with any of the results.

Thanks for you help. If you have any questions or concerns about this survey, please
contact me at the phone or fax number listed alongside. If you wish to contact the DNR

snowmobile program managers, for trail maintenance and design matters contact (517)
373-1275 and for law enforcement matters contact (517) 373-1230.

Sincerely,

Chuck Nelson,
Assith Professor

CDC.

155

APPENDIX C

1996-97 CONVICTED SNOWMOBILER LAW VIOLATOR COVER LETTER

156

June 12, 1997

Dear Snowmobiler:

The Michigan Department of Natural Resources (DNR) and Michigan State University
are cooperating to better serve the snowmobiling public. Enclosed is a brief questionnaire
asking about your recent Michigan snowmobiling experiences, your assessment of
current snowmobiling programs, opinions about future program options and basic
information about your household.

Your responses will be critical in shaping the DNR’s future snowmobiling program. In
particular, the results will be used by trail managers, housed in the Forest Management
Division and conservation ofﬁcers, housed in the Law Enforcement Division.

Your name was chosen from DNR records of snowmobilers issued a citation for safety
related rule violations during 1996-97. This was done to understand the perspective of
those who have had contact with law enforcement personnel while snowmobiling.

Please take the 15 minutes or so necessary to complete the questionnaire. Your
participation is voluntary and you indicate your willingness to participate by completing
and returning this questionnaire. This will allow your opinions to be heard by the DNR
and other policy makers concerning future snowmobile programs. When you have
completed the questionnaire, please mail it back to me in the postage paid enveloped
provided. Your responses will be kept conﬁdential and you name will not be associated
with any of the results.

Thanks for you help. If you have any questions or concerns about this survey, please
contact me at the phone or fax number listed alongside. If you wish to contact the DNR

snowmobile program managers, for trail maintenance and design matters contact (517)
373-1275 and for law enforcement matters contact (517) 373-1230.

Sincerely,

Chuck Nelson,
Assistant Professor

CDC.

157

APPENDIX D

OTHER SNOWMOBILER FOLLOW-UP COVER LETTER

158

July 10, 1997

Dear Michigan Snowmobile Trail Sticker Purchaser:

About a month ago I mailed you a questionnaire concerning Michigan snowmobiling. At
this time, I have not received your response. If you have already replied and our letters
cross in the mail, thank you for responding and please accept my apologies for this extra
contact. I have sent this ﬁnal questionnaire to you by certiﬁed mail to emphasize its
importance and to ensure it delivery in case you did not receive the ﬁrst one or have
misplaced it. For a few who did respond but destroyed the code number stamped on the
survey, you are being surveyed again because I was unable to determine that you
responded the ﬁrst time.

While your participation is voluntary, it provides an excellent opportunity for your
opinions to be heard. Policy makers will use the results of this study when considering
snowmobile regulations. law enforcement, fees and trail design and maintenance. Please
take the 15 minutes or so necessary to complete it and join with the many other
snowmobilers who have let their opinions be heard. Remember even those of you who
did not snowmobile in Michigan during the winter 1996-97, there are portions of the
questionnaire still directed at you since you have snowmobiled here in the past.

When you have completed the questionnaire, please mail it back to me in the postage paid
enveloped provided. Your responses will be kept conﬁdential and you name will not be
associated with any of the results.

Thanks for you help. If you have any questions or concerns about this survey, please
contact me at the phone or fax number listed alongside. If you wish to contact the DNR

snowmobile program managers, for trail maintenance and design matters contact (517)
373-1275 and for law enforcement matters contact (517) 373-1230.

Sincerely,

Chuck Nelson,
Assistant Professor

159

APPENDIX E

1996-97 CONVICTED SNOWMOBILER LAW VIOLATOR
FOLLOW-UP COVER LETTER

160

July 17, 1997

Dear Snowmobiler:

About a month ago I mailed you a questionnaire concerning Michigan snowmobiling. At
this time, I have not received your response. If you have already replied and our letters
cross in the mail, thank you for responding and please accept my apologies for this extra
contact. I have sent this ﬁnal questionnaire to you by certiﬁed mail to emphasize its
importance and to ensure it delivery in case you did not receive the ﬁrst one or have
misplaced it. For a few who did respond but destroyed the code number stamped on the
survey, you are being surveyed again because I was unable to determine that you
responded the ﬁrst time.

While your participation is voluntary, it provides an excellent opportunity for your
opinions to be heard. Policy makers will use the results of this study when considering
snowmobile regulations. law enforcement, fees and trail design and maintenance. Please
take the 15 minutes or so necessary to complete it and join with the many other
snowmobilers who have let their opinions be heard. Remember even those of you who
did not snowmobile in Michigan during the winter 1996-97, there are portions of the
questionnaire still directed at you since you have snowmobiled here in the past.

When you have completed the questionnaire, please mail it back to me in the postage paid
enveloped provided. Your responses will be kept conﬁdential and you name will not be
associated with any of the results.

Thanks for you help. If you have any questions or concerns about this survey, please
contact me at the phone or fax number listed alongside. If you wish to contact the DNR

snowmobile program managers, for trail maintenance and design matters contact (517)
373-1275 and for law enforcement matters contact (517) 373-1230.

Sincerely,

Chuck Nelson,
Assistant Professor

161

APPENDIX F

MICHIGAN SNOWMOBILING NON-RESPONDENT TELEPHONE SURVEY

162

Michigan Snowmobiling Non-Respondent Telephone Follow-up Phone #

 

Hello, my name is from Michigan State University. I’m calling about his/her
Michigan Snowmobiling. (if not individual, say “May I speak to him/her?”)

, l have a few questions about your household's snowmobiling this past winter. We mailed you a questionnaire
earlier this year, but we didn’t receive a completed one back. To best represent the opinions of snowmobilers to the DNR, I would
appreciate just 5 minutes of your time to answer a few key questions. Would you be willing to help? (Wait for a response. if no, say
thank you and good bye. If yes, say thank you.)

First I’d like to get some information about your household and their snowmobiling.

1. Including you, how many people are in your household? #
2. How may snowmobiled in winter of 1996-97 #
3. How may snowmobiled in Michigan in winter of 1996-97 #
4. Please tell me about the model year and CC level of each of the snowmobiles your household owns.

(DO A through D ONE AT A TIME AND FILL IN THE CHART BELOW)

a. Which machines were used during winter of 1996-97?
b. Which had Michigan trail stickers in winter of 1996-97?
c. Approximately, how many miles was each driven in Michigan during winter of 1996-97?
d. How may different days was each machine operated in Michigan for snowmobiling.
This does not include it use for support of other activities such as ice ﬁshing.

Used during winter Had 1997 M1 Approx. # Miles driven in Num. of Michigan
Model Year CC of 1996-97? trail sticker? Ml winter 1996-97 snowmobile

 

5. Of the # (Total from chart above) MI snowmobile days, how may were in the UP? # days in UP

6. How many were in the NLP (N. of Bay City/ Mount Pleasant! Muskegon)? # days in NLP

8. SKIP - Don’t ask. just subtract Southern Lower Peninsula? # days in SLP

9. Of these # (Total from chart above) MI snowmobiling days, how many were on trips more than 100 miles from
home or where you stayed away from home one or more nights? # days ovemight/ 100 miles away

10. Were you or members of your household ever stopped or checked by a law enforcement officer while snowmobiling in
Michigan during winter 1996-97? Yes or No

163

Now I’d like to ask you some questions about Michigan snowmobiling and its management.

ll.

l3.
14.

Please rate the following services and situations on a scale of very good to very poor, with the choices being
very good (5), good (4), OK (3), poor (2), and very poor (1 ). First would you rate:

a. Public trail grooming # Why this rating?

 

b. Snowmobile law enforcement # Why this rating?

 

Over winter l996«97, more than 40 snowmobilers were killed during Michigan Snowmobiling. The DNR is considering
potential ways to reduce fatalities. Please rate your level of support for each of these on a scale of l to 5. with 5 strongly
support, 4 moderately support, 3 neutral, 2 moderately oppose. and 1 strongly oppose. How would you rate:

 

 

 

 

 

 

 

Strongly Moderately Moderately Strongly

Fatality Reduction Initiative Support Support Neutral Oppose Oppose

An enforce snowmobﬂe speed limit

on public lands, trails, and water 5 ‘ 3 2 I
Tandatory training requrredTor all

ﬁrst year snowmobile operators 5 4 3 2 I

regardless of age
Wensrng required 0T all snowmobile

operators 5 4 3 2 i
Mandatory personaTﬁaFiITfy

insurance for all snowmobile 5 ‘ 3 2 I

operators

More intensive e'r'i'i'ércement of

existing snowmobiling regulations 5 4 3 2 1

I'd like to ﬁnish up by asking your current age and at what age you began snowmobiling

SKIP — Fill in state after interview

 

That’s it. Thanks for your help in Improving Michigan Snowmobiling. Have a good evening/morning/alternoon.

164

REFERENCES

165

REFERENCES

Ajzen, I. 1985. From intentions to actions: A theory of planned behavior. In Kuhl, J. and

Beckrnan, J. (eds), Action Control: From Cogm'tion to Behavior.
Springer-Verlag, Berlin, Germany. 11-38.

Akers, R. L. 1994. Criminologjcal Theories: Introduction and Evaluation. Roxbury
Publishing Co., Los Angeles, CA.

Anderson, D. H., Lime, D. W., and Wang, T. L. 1998. Maintaining the Quality of Park
Resources m Visitor Experiences: A Handbook for Managers. Tourism Center,
University of Minnesota Extension Service, St. Paul, MN.

Andersen, T. A. 1987. Human Dimensions of Michigan Deer Law Violations Attitudes

and Demoggphic Variables: A Multivariate Analysis and Description of
Correlations. MS. Thesis. Department of Sociology, Eastern Michigan

University. Ypsilanti, MI.
Avis, S. P. 1994. Snowmobile fatalities. Journal of Forensic Sciences, 39(5). 1161-1167.
Babbie, E. 1995. The Practice of Social Research (7th ed.). Wadsworth, Belemont, CA.

Brown, T. L., Decker, D. J ., and Connelly, N. A. 1989. Response to mail surveys on
resource-based recreation topics: A behavioral model and an empirical analysis.
Leisure Sciences, 11. 99-110.

Bulter, R. W. 1974. How to control 1,000,000 snowmobiles? Canadian Geoggphical
Journal, 88(3). 4-13.

 

Carlson, W. L. and Klein, D. 1971. Behavioral patterns of snowmobile operators --
A preliminary report. Journal of Safeg Research, 3(4). 150-156.

CDC. 1995. Injuries associated with use of snowmobiles -- New Hampshire, 1989-1992.
Morbidity and Mortalig Weekly Remit, 44(1). 1-3.

CDC. 1997. Injuries and deaths associated with use of snowmobiles - Maine, 1991-1996.
Morbidig and Mortalig Weekly Report, 46(1). 1-4.

Cooper, P. J. 1997. The relationship between speeding behavior (as measured by
violation convictions) and crash involvement, Journal of Safety Research, 28(2).
83-95.

166

Cox , T. C. and White, M. F. 1988. Trafﬁc citations and student attitudes towards the
police: An examination of selected interaction dynamics. Journal of Police

Science and Administration, 16(2). 105-121.

Crompton, J. L. and Tian-Cole, S. 1999. What response rate can be expected from
questionnaire surveys that address park and recreation issues. Journal of Park and
Recreation Administration, 17(1). 60-72.

Cullen, F. T. and Agnew, R. 1999. Criminologjcal Theory: Past to Presept.
Roxbury Publishing Co., Los Angeles, CA.

Curcione, N. 1992. Deviance as delight: Party-boat poaching in southern California.
Deviant Behavior, 13. 33-57.

DeJoy, D. M. 1992. An examination of gender differences in trafﬁc accident risk
perception. Accident Analysis and Prevention, 24(3). 237-246.

Demaris, A. 1992. Logit Modeling: Practical Applications. Sage University Paper Series
on Quantitative Applications in the Social Sciences, 07-86. Sage Publications,
Newbury Parks, CA.

Dewar, R. E. 1973. A review and critique of snowmobile accidents reports. Journal of

Safeg Research, 5(1). 36-44.

Donovan, D. M., Marlatt, G. A., and Salzberg, M. P. 1983. Drinking behavior,
personality factors and high-risk driving. Journal of Studies on Alcohol, 44(3).
395-428.

Douglas, M and Wildavsky, A. 1982. Risk and Culture. University of California Press,
Berkeley, CA.

Duda, M. D., Bissell, S. J ., and Young, K. C. 1998. Wildlife and the American Mind:

Public Opinion on and Attitudes Towards Fish and Wildlife Management.
Responsive Management, Harrisonburg, VA.

East Bay Regional Parks, 1991. Volunteer Bicycle and Mounted Patrol Bylaws. East Bay
Regional Parks, Oakland, CA.

Einstadter, W. and Henay, S. 1995. Qriminological Theory: An analysis of Underlying
Assumptions. Harcourt Brace College Publishers, Forth Worth, IN.

167

Elander, J ., West, R., and French, D. 1993. Behavioral correlates of individual differences
in road-trafﬁc crash risk: An examination of methods and ﬁndings. Psychological
Bulletin, 113(2). 279-294.

Evans, L. and Wasielewski, P. 1982. Do accident involved drivers exhibit riskier
everyday driving behavior. Accident Analysis and Prevention, 14(1). 57-64.

Fishbein, M and Ajzen, I. 1975. Belief, Attitude Intention and Behavior:
An Inuoduction to Theog and Research. Addison-Wesley, Reading, MA.

 

Forsyth, C. J. and Marckese, T. A. 1993. Thrills and skills: A sociological analysis of
poaching. Deviant Behavior. 14. 157-172.

Forsyth, C. J ., Grarnling, R. and Wooddell, G. 1998. The game of poaching: Folk crimes
in southwest Louisiana, Socieg and Natural Resources, 11. 25-38.

Gabany, S. G., Plummer, P., and Grigg, P. 1997. Why drivers speed: The speeding
perception inventory. Journal of Safcg Research, 28(1). 29—36.

Gabor, T. 1986. The Prediction of Criminal Behavior: Statistical Approaches.

University of Toronto Press, Toronto, Canada.

Garretson, M. R. and Peck, C. 1982. Factors associated with fatal accident involvement
among California drivers. Journal of Safety Research, 13. 141-156.

Gibbons, D. C. 1983. Mundane crime. Crime and Delinguency. 213-227.
Goldstein, A. P. 1996. The Psychology of Vandalism. Plenum Press, New York, NY

Gramann, J. B. and Vander Stoep, G. A. 1987. Prosocial behavior theory and natural
resources protection: A conceptual synthesis. Journal of Environmental
Management, 24. 247-257.

Gray, B. T. and Kaminski, R. M. 1994. Illegal waterfowl hunting in the Mississippi
flyway and recommendations for alleviation. Wildlife Monogtaphs, 127. 1-60.

Green, L. W., Kreuter, M. W., Deeds, S. G., and Partridge, K. B. 1980. Health Education
Planning: A Diagpostic Approach. Mayﬁeld, Palo Alto, CA.

168

Guppy, A. 1993. Subjective probability of accident and apprehension in relation to

self-other bias, age, and reported behavior. Accident Analysis and Prevention,
25(4). 375-382.

Hall, D. L. 1992. Hunter compliance in North America: Past, present, and future.
In Beattie, K. H. (tech. ed.) Proceedings International Conference on Improving
Hunter Compliance with Wildlife Laws, College of Natural Resources, University
of Wisconsin Steven Point, Steven Point, WI. 21-35.

Healey, J. F. 1993. Statistics: A Tool for Social Research, 3rd ed. Wadsworth Inc.,
Belmont, CA.

Hermance, R. 1995. Snowmobile Accident Reconstruction: A Technical and Legal
Guide. Lawyers and Judges Publishing Co., Inc. Tucson, AZ.

Homey, R. L. 1970. Snowmobiling Guidelines, Environmental Impact, Legislation, and
Lrogpan—rm—ing. NRPA Management Aid Bulletin No. 89. National Recreation and

Park Association, Washington, D. C.

Hosmer, D. W. and Lemeshow, S. 1989. lied Lo 'stic Re ession. John Wiley and
Sons, New York, NY.

Jackson, R. 1980. Models and inﬂuences shaping hunter attitudes and behaviors. Paper
Presented at 42nd Midwest Fish and Wildlife Conference. St. Paul, MN.

Jackson, R., Norton, R., and Anderson, R. 1979. Improving ethical behavior in hunters.

Transactions North American Wildlife and Natural Resources gonference, 44.
306-3 1 8.

James, E. C., Lenz, J. O., Swenson, W. M., Cooley, A. M., Gomez, Y. L., and
Antonenko, D. R. 1991. Snowmobile trauma: An eleven-year experience.

The American Surgeon, 57(6). 349-353.

Jessor, R. and Jessor, S. L. 1977. Problem Behavior and Psychosocial Development: A
Longitudinal Study of Youth. Academic Press, New York, NY.

Jonah, B. A., Dawson, N. E., and Bragg, B. W. 1981. Predicting accident involvement
with the motorcycle operator skill test. Accident Analysis and Prevention, 9(4).
307-3 1 8.

Jonah, B. A. 1986. Accident risk and risk-taking behavior among young drivers.
Accident Analysis and Prevention, 18(4). 255-271.

169

Kanellaidis, G., Golias, J ., and Zarifopoulos, K. 1995. A survey of drivers’ attitudes
towards speed limit violations. Journal of Safety Research, 26 (1). 31-40.

Kescl, J. A. 1974. Some of the Qharacteristics and Attitpdes of Michigan Deer Hunting
Violators. Masters Thesis. Michigan State University, East Lansing, Michigan.

Kleinbaum, D. G. 1994. Logistic Regtession: A Self-Learning Text. Springer-Verlag,
New York, NY.

Klim, E. 1996. Survival of the ﬁttest and the “funniest” The evolution of the
snowmobile. International Snowmobile Manufacturers Association (ISMA) Press
Release. Haslet, MI.

Klirn, E. 1997. Snowmobiles are gaining in popularity. International Snowmobile
Manufacturers Association (ISMA) Press Release. Haslet, MI.

Lanier, L. 1974. An Exploratog Study of Use Patterns and User Characteristics of
Michigan Snowmobile Owners. Ph.D. Dissertation. Department of Resource

Development, Michigan State University. East Lansing, MI.
Lansing State Journal. 2000. Capitol highlights. February 20, 2000. BB.

Lawton, R., Parker, D., Stradling, S. G., and Manstead, A. S. R. 1997a. Predicting road
trafﬁc accidents: The role of social deviance and violations. British Journal of
Psychology, 88. 249-262.

Lawton, R., Parker, D., Manstead, A. S. R. and Stradling, S. G. 1997b. The role of
affective rn predicting social behaviors: The case of road trafﬁc violations.

Journal of Applied Social chhology, 27(14). 1258-1276.

Lucas, R. C. 1983. The role of regulations in recreation management. Western Wildlands
9(2): 6-1 0.

Lynch, J. A. and Nelson, C. M. 1998a. Snowmobilers’ perceptions of danger and
relationship to fatal accidents. In: Proceedirga of the 1998 Northeast Recreation
Research Symposium, April 5-7, 1998, Bolton Landing, NY. GTR NE-255.
USDA Forest Service. Northeastern Forest Experiment Station.

Lynch, J. A. and Nelson, C. M. 1998b. Proﬁle of Snowmobilers Cited by Michigan
Dppartrnent of Natural Resources Ofﬁcers During Winter 1996- 97 and Convicted
of Safeg Violations. Report to the Michigan Department of Natural Resources,
Lansing, Michigan. Department of Park Recreation, and Tourism Resources,
Michigan State University, East Lansing, MI.

170

Mangione, T. W. 1995. Mail Surveys: Improving the Qualig. Applied Social Resource
Methods Series, 40. Sage Publications. Thousand Oaks, CA.

Mayer R. E. and Treat, J. R. 1977. Psychological, social, and cognitive characteristics of
high- risk drivers; A pilot study. Accident Analysis and Prevention, 9. 1-8.

McElvany, N. D. 1994. Snowmobiling in Vermont: An Economic Impact Study and
Snowmobiler User Survey. Report submitted to Vermont Association of Snow
Travelers, Montpelier, VT. Department of Business and Economics, Johnson
State College, Johnson, VT.

McMillen, D. L., Pang, M. G, Wells-Parker, E., and Anderson, B. J. 1992. Alcohol,
personality traits, and high risk driving: A comparison of young, drinking driving
groups. Addictive Behaviors, 17. 525- 532.

Meadows, M. L., Stradling, S. G., and Lawson, S. 1998. The role of social deviance and
violations in predicting road trafﬁc accidents in a sample of young offenders.
British Journal of Psychology, 89. 417-431.

Melynk, M. J. 1977. Hunters Attitudes Towards Wildlife Laws and Wildlife Ofﬁcers: An

Analysis of Formational meets. MS. Thesis. University of Alberta, Edmonton,
Canada.

Menard, S. 1995. Applied Logistic Regtession Analysis. Sage University Paper Series on
Quantitative Applications in the Social Sciences, 07-106. Sage Publications,
Thousand Oaks, CA.

Mickey, J. and Greenland, S. 1989. A study of the impact of confounder-selection criteria
on effect estimation. American Journal of Epidemiology, 129. 125-137.

Miether, T. and Meier, R. 1994. Qrimes and its Social Context: Toward an Intemted

Theogt of Offenders, Victims, and Situations. State University of New York
Press, Albany, NY.

Morrison, D. 1969. On the interpretation of discriminant analysis. Journal of Marketing
Research, 6. 156-63.

Moore, R. L. 1994. Conﬂicts on Multi-Use Trails: Smthesis of Literature and State of
Practice. Report to Federal Highway Administration, Washington DC.

Mortimer, R. G. 1988. A ﬁuther evaluation of the motorcycle rider course. Journal of
Safety Research, 19. 187-196.

171

Muth, R. M. and Bowe, J. F. 1998. Illegal harvest of renewable natural resources in North
America: Towards a typology of the motivations for poaching. Society and
Natural Resources, 11. 9-24.

Mullen, P. D., Hersey, J. C., and Iverson, D. C. 1987. Health behaviors models
compared. Social Science Medicine, 24(11). 973-981.

Nelson, C. M. and Lynch, J. A. 1994. Estimating dispersed recreational use on
Michigan’s state and national forests. In: Vander Stoep, G. A. (ed.) Proceeding of

the 1994 Northeastern Recreation Research Symposium GTE NE-198. April
10-12, 1994. Saratoga Springs, New York. USDA, Forest Service, Northeastern

Forest Experiment Station. Gen., Tech. Rpt. NE 198. Randor, PA. 13-16.

Nelson, C. M. and Lynch, J. A. 1995. Diapersed and Developed Recreation on the
Hiawatha National and Lake Superior State Forests. Report to the USDA Forest
Service, Hiawatha National Forest and Michigan Department of Natural
Resources. Department of Park Recreation, and Tourism Resources, Michigan
State University, East Lansing, MI.

Nelson, C. M., Lynch, J. A., and Stynes, D. 1998. An Assessment of Snowmobiling in
Michigan by Snowmobilers with Michigan Trail Permit_s_. Report to the Michigan
Department of Natural Resources, Lansing, Michigan. Department of Park
Recreation, and Tourism Resources, Michigan State University, East Lansing, MI.

Norusis, M. J. 1997. SPSS Professional Statistics 7.5. SPSS Inc. Chicago, IL.

Parker, D., Manstead, A. S. R., Stradling, S. 6., Reason, J. T., and Baxtrer, J. S. 1992.
Intention to commit driving violations: An application of the theory of planned

behavior. Journal of Applied Psychology, 77(1). 94-101.

Parker, D., Stradling, S. G., and Manstead, A. S. R. 1996. Modifying beliefs and attitudes
to exceeding the speed limit: An intervention study based on the theory of

planned behavior. Journal of Applied Social Psychology, 26(1). 1-19.

Press, J. and Wilson, S. 1978. Choosing between logistic regression and discriminant

analysis. Journal of the American Statistical Association, 73(364). 699-705.

Purol, D. A., Hill, H. R., Hussain, N. G., and Stoll, G. D. 1978. The Percpptions and

Attitudes of Michigan Deer Hunters Towards an Accaptable Rate of Compliance.
Michigan Department of Natural Resources, Law Enforcement Division,

Lansing, MI.

172

Reason, J ., Manstead, A., Stradling, S., Baxter, J ., and Campbell, K. 1990. Errors and
violations on the roads: A real distinction? Ergonomics, 33(10/11). 1315-1332.

Responsive Management. 1993. Arizona Residents’ Knowledge, Opinions, and

Attitudes Towards the Arizona Game and Fish Dapartments’ Law Enforcement
Activities. Report prepared for the Arizona Game and Fish Department.

Responsive Management, Harrison, VA.

Robertson, R. A. 1996. Assessment of Snowmobiling in New Hampshire: Summgy and
Recommendations. Department of Resource Economics and Development,

University of New Hampshire, Durham, NH.

 

Rothe, J. P. 1994. Beyond Trafﬁc Safety. Transaction Publishers, New Brunswick, NJ.

Rothengatter, T. 1988. Risk and the absence of pleasure: A motivational approach to
modeling road user behavior. Ergonomics, 31(4). 599-607.

Ross, H. L. 1961. Trafﬁc law violation: A folk crime. Social Problems, Winter. 231-241.
Ross, H. L. 1973. Folk crimes revisited. Qriminology, 5. 71-85.

Rowe, B., Milner, R., Johnson, C., and Bota, G. 1992. Snowmobiling-related deaths in

Ontario: A 5-year review. Journal of Qanadian Medical Association, 146(2).
147-1 52.

Rowe, B., Caverson, R., R., Therrien, S., Bota, G., Giesbrecht, N., and Smythe, C. 1993.
Snowmobilers in a Northeastern Ontario Community: A Survey of

Qharacteristics, Injtgt Proﬁles, and Strategies for Injgy Prevention. Sudbury
General Hospitable of the Immaculate Heart of Mary and Addiction Research

Foundation, Ontario, Canada.

Rowe, B., Milner, R., Johnson, C., and Bota, G. 1994. The association of alcohol and
night driving with fatal snowmobile trauma: A case-control study.
Annals of Emergency. Medicine, 24. 842-848.

Shafer, E. L., Amidon, P. H., and Severinghaus, C. W. 1972. A comparison of violators
and nonviolators of New York’s deer-hunting laws. Journal of Wildlife
Management, 36(3). 933-939.

Sheridan, D. 1979. Off-Road Vehicles on Public Land. Council on Environmental
Quality, Washington D. C.

173

Shinar, D. and McKnight, A. J. 1985. The effects of enforcement and public information
on compliance. In Evans, L. and Schwing, R. C. (eds.) Human Behavior and
Trafﬁc Safety. Plenum, New York, NY.

Shoham, S. G., Rahav, G., Markovski, R., Chard, F., and Baruch, I. 1984. “Anxious” and
“reckless” drivers. Deviant Behavior, 5. 181-191.

Stoll, G. D. 1975. The Attitudes of Recreational Resource Usera Arre§ted by Michiga_n
Conservation Ofﬁcers. MS. Thesis. Department of Park and Recreation
Resources, Michigan State University, East Lansing, MI.

Stynes, D., Nelson, C. M., and Lynch, J. A. 1998. State and Regional Economic Impacts
of Snowmobiling in Michiga_n. Report to the Michigan Department of Natural
Resources, Lansing Michigan. Department of Park, Recreation and Tourism
Resources, Michigan State University, East Lansing, MI.

Swearingen, T. C. and Johnson, D. R. 1995. Visitors’ responses to uniformed park
employees. Journal of Park and Rgreation Administratiop, 13(1): 17-19.

Tihnan, W. A. and Hobbs, G. E. 1949. The accident-prone automobile driver. American
Journal of Psychiag, 106. 321-331.

Tull, D. S. and Hawkins, D. I. 1993. Marketing Research: Measurement and Method.
Macmillan Publishing Co., New York, NY.

Wallach, M., and Kogan, N. 1961. Aspects of judgments and decision making:
Interrelationships and change with age. Behavior Science, 6. 23-26.

Waller, Julian A. 1975. A survey of snowmobiling in Vermont. Accident Analysis and
Prevention, 7. 207-212

Waller, J. A. and Lambom, K. R. 1975. Snowmobiling: characteristics of owners,
patterns of use and injuries. Accident. Analysis and Prevention, 7. 213-223.

Wenzel, F. J ., Peters, R. A., Hintz, C. S., and Olsen, T. W. 1973. Snowmobile accidents
in central Wisconsin. Wisconsin Medical Journal, 72(2). 89-72.

Wenzel, F. J. and Peters, R. A. 1986. A ten-year survey of snowmobile accidents,
injuries, and fatalities in Wisconsin. The Physician and Sportrncdicine, 14(1).
140-149.

174

 

West, R., Elander, J ., and French, D. 1993. Mild social deviance, type-a behavior pattern
and decision-making style as predictors of self-reported driving style and traffic
accident risk. British Journal of Psychology, 84. 207-219.

Willard, L. 1998. Department of Natural Resources, Law Enforcement Division. Lansing,
MI. personal communication July, 1998.

Williams, F. P. and McShane, M. D. 1999. Criminological Theog. Prentice Hall,
Upper Saddle River, NJ.

Wilson, J. 1992. Convicted impaired drivers and high-risk drivers: How similar are they?
Journal of Studies on Alcohol, 35(4). 335-344.

Wilson, M. 1983. Folk crimes: Patterns of accommodation. Deviant Behavior, 4.
123-140.

Withington, R. L. and Hall, L. W. 1970. Snowmobile accidents: A review of injuries
sustained in the use of snowmobiles in northern New England during the 1986-69

season. Journal of Trauma, 10(9). 760-763.

Zesiger, S. 1997. Zero to 60 in three second (cold). Fortune. 178-179.

 

GOVERNMENT REFERENCES

IL DC. 1996. Snowmobile Accident Rama: 1995-1996 Season. Illinois Department of

Natural Resources Division of Education.

M DNR. 1998. Draft Michigan Dppartment of Natural Resources Snowmobile Trail
System Master Plan. Michigan Department of Natural Resources, Forest
Management Division, Lansing, Michigan.

MN DNR. 1970. Minnesota Snowmobile Survey 1970. Minnesota Department of
Conservation, Bureau of Planning. St. Paul, MN.

MN DNR. 1997. 1997 Snowmobile Accident Summm. Minnesota Department of
Natural Resources, Division of Enforcement. St. Paul, Minnesota.

WA-SPRC. 1988. 1988 Snowmobiler Study. Washington State Parks and Recreation
Commission. Olympia, WA.

175

 

WI DNR. 1997. 1996-97 Snowmobile Accident Rama. Wisconsin Department of
Natural Resources, Law Enforcement Division. PUB-LE-402-97.

USDI. 1971. ORRV: Off Road Recreation Vehicles. US. Department of Interior Task
Force Study. Washington D. C.

176

MICHIGAN sran UNIV. LIBRQRIES
lllIllllllIllIllllllllllllllllllllllllllllllll
31293020485415