Citation
Hurricane Preparedness of Homeowners in the Southeast United States

Material Information

Title:
Hurricane Preparedness of Homeowners in the Southeast United States
Creator:
Sewell, Charles B
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Family, Youth and Community Sciences
Committee Chair:
CANTRELL,RANDALL ALAN
Committee Co-Chair:
SPRANGER,MICHAEL S
Committee Members:
JONES,PIERCE H
SILVER,CHRISTOPHER
KRAFT,JOHN

Subjects

Subjects / Keywords:
disaster-preparedness
homeowner
hurricane
preparedness
preparedness-knowledge
southeast-us
trust

Notes

General Note:
The United States is experiencing the longest drought from major hurricanes on record since 1851. This 10-year absence from major hurricanes has lulled many homeowners into a false sense of security, allowing preparedness to become a waning priority in their minds. Emergency managers across Southeastern Georgia, Florida, and the Gulf Coast are tasked with keeping the public and policymakers prepared at all times for such events. This quantitative study explored the relationships between demographic characteristics (i.e. location, education, income, and age) and two study-generated measures; the Hurricane-Preparedness Knowledge Scale and the Trust in Support Entities Scale. Through the discussion of several hypotheses, the strength of some indicators was observed while possible reasons for the non (statistical) significance of others was discussed. The implications of this study will add to the body of knowledge on disaster preparedness as well as providing insights into demographic indicators of trust in aid providers. Findings from this study and subsequent research resulting from it will aid emergency managers and policymakers in better understanding what areas need more attention while reinforcing those areas that are understood well enough to probe more deeply via advanced discussions regarding preparedness. The findings about support entities will provide insight to what populations may benefit from outreach programs to advance understanding of the aid available to homeowners if needed. A more knowledgeable public that has an increased confidence in the available support will also allow for a more efficient allocation of resources during times of need caused by major hurricanes.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2017

Downloads

This item has the following downloads:


Full Text

PAGE 1

HURRICANE PREPAREDNESS OF HOMEOWNERS IN THE SOUTHEAST UNITED STATES By CHARLES BRADFORD SEWELL A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2015

PAGE 2

2015 Charles Bradford Sewell

PAGE 3

To my girls, Megan and Brianna

PAGE 4

4 ACKNOWLEDGMENTS I want to express my sincere gratitude and respect f or my committee chair, Dr. Randall Cantrell. His mentorship, guidance and friendship have been invaluable to my successful navigation through graduate school. His gener osi ty in including me in his Early Career Grant afforded me the opportunity to really immerse myself in this research and achieve more meaningful conclusions I truly look forward to our future collaborations. I would also like to acknowledge Dr. Michael Spranger, my internal FYC S committee member for encouraging me to reach ever so slightly further to make my class work assignments better as well as this thesis. His excitement and passion for s been a source of motivation for me. My external committee members : Dr. Pierce Jones, Dr. Christopher Silver, and Dr. John Kraft have all provide d different perspectives that have added depth and value to my work as well. I owe all of these gentlemen a de bt of gratitude for the active role they have taken as committee members and it is my hope that they are all proud of this thesis I would be remiss if I did not mention Professor Albert Dambrose and Professor Harold Van Boven of Florida Southwestern Colle ge, as well as Dr. Robert Stwalley of Ivy Tech State College and Purdue University, who all encouraged me to pursue my education at the university level I would also like to thank my family for their continued love and support through my six year return t o school. It has not been the easiest journey, but we made it together.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURE S ................................ ................................ ................................ ........ 12 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 17 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 19 Rationale for Study ................................ ................................ ................................ 22 Purpose of Study ................................ ................................ ................................ .... 23 Significance of Study ................................ ................................ .............................. 24 Definition of Terms ................................ ................................ ................................ .. 25 2 LITERATURE REVIEW ................................ ................................ .......................... 29 Theoretical Framework ................................ ................................ ........................... 29 ........................ 34 Properties of the Bioecological Model ................................ ................................ ..... 35 Applications of the Bioecological Theory of Human Development .......................... 40 Bioecological Theory in Emergency Preparedness ................................ ................ 42 Strengths of the Bioecological Theory of Human Development .............................. 45 Weaknesses of the Bioecological Theory of Human Development ......................... 46 Review of Current Literature ................................ ................................ ................... 46 Preventive Behaviors ................................ ................................ ....................... 47 Risk Perception ................................ ................................ ................................ 48 Knowledge/Awareness ................................ ................................ ..................... 50 Other Fa ctors of Preparedness ................................ ................................ ........ 51 3 METHODOLOGY ................................ ................................ ................................ ... 58 Research Questions and Hypotheses ................................ ................................ ..... 58 Research Design ................................ ................................ ................................ .... 61 Population/S ampling ................................ ................................ ............................... 62 Instrumentation ................................ ................................ ................................ ....... 63 Data Analysis ................................ ................................ ................................ .......... 64 Principal Components Analysis ................................ ................................ ........ 64 Exploring Differences Between Groups ................................ ............................ 65 Multiple Regression Modeling ................................ ................................ .......... 67

PAGE 6

6 Limitations ................................ ................................ ................................ ............... 68 4 RESULTS AND ANALYSIS OF DATA ................................ ................................ .... 71 Overview ................................ ................................ ................................ ................. 71 Principal Components Analysis ................................ ................................ .............. 71 Exploring Differences Between Groups ................................ ................................ .. 73 Research Question 1: Hurricane Preparedness Knowledge ............................ 73 Location of home ................................ ................................ ....................... 73 Highest educ ation level of homeowner ................................ ...................... 76 Household income ................................ ................................ ..................... 78 Age of homeowner ................................ ................................ ..................... 82 Research Question 2: Trust in Support Entities ................................ ................ 84 Location of home ................................ ................................ ....................... 84 Highest education level of homeowner ................................ ...................... 87 Household income ................................ ................................ ..................... 89 Age of homeowner ................................ ................................ ..................... 92 Multiple Reg ression Modeling ................................ ................................ ................. 95 Hurricane Preparedness Knowledge Scale ................................ ...................... 95 Florida regions ................................ ................................ ........................... 95 Study area ................................ ................................ ................................ .. 97 Trust in Support Entities Scale ................................ ................................ ......... 98 Florida regions ................................ ................................ ........................... 98 Study area ................................ ................................ ................................ .. 98 5 DISCUSSION ................................ ................................ ................................ ....... 131 Overview ................................ ................................ ................................ ............... 131 Hypotheses Implica tions ................................ ................................ ....................... 131 Research Question 1: Hurricane Preparedness Knowledge .......................... 131 Location of home ................................ ................................ ..................... 132 Highest education level of homeowner ................................ .................... 134 Household income ................................ ................................ ................... 135 Age of homeowner ................................ ................................ ................... 136 Research Question 2: Trust in Support Entities ................................ .............. 138 Location of home ................................ ................................ ..................... 138 Highest education level of homeowner ................................ .................... 140 Household income ................................ ................................ ................... 141 Age of homeowner ................................ ................................ ................... 142 Multiple Regression Analyses Implications ................................ ........................... 143 Hurricane Preparedness Knowledge Scale (HPKS) ................................ ....... 143 Florida regions ................................ ................................ ......................... 143 Study area ................................ ................................ ................................ 144 Trust in Support Entities Scale (TSES) ................................ .......................... 144 Florida regions ................................ ................................ ......................... 144 Study area ................................ ................................ ................................ 145 Descriptive Implications ................................ ................................ ........................ 146

PAGE 7

7 Summary of Implications ................................ ................................ ....................... 146 Future Research ................................ ................................ ............................. 148 Time living in community ................................ ................................ .......... 149 Mi nors present in household ................................ ................................ .... 149 Policy Implications ................................ ................................ .......................... 150 Further Limitations ................................ ................................ ................................ 151 Conclusions ................................ ................................ ................................ .......... 151 APPENDIX A DESCRIPTIVE STATISTICS OF THE SAMPLE ................................ ................... 160 B DETAIL OF REVISED PRINCIPAL COMPONENTS ANALYSIS .......................... 162 Component 1: Knowledge ................................ ................................ ..................... 163 Component 2: ................................ ................................ ................................ ....... 164 Component 3: T rust in Support Entities ................................ ................................ 165 Component 4: ................................ ................................ ................................ ....... 165 Component 5: ................................ ................................ ................................ ....... 166 Component 6: ................................ ................................ ................................ ....... 166 C DETAIL OF OUTLIER LABELING RULE CALCULATIONS ................................ 168 D DETAIL OF HPKS SCORES BY TIME LIVING IN COMMUNITY ......................... 172 E DETAIL OF HPKS SCORES BY MINORS IN HOUSEHOLD ............................... 175 F RESEARCH ROADMAP ................................ ................................ ....................... 177 REFERENC ES ................................ ................................ ................................ ............ 183 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 191

PAGE 8

8 LIST OF TABLES Table page 1 1 Saffir Simps on Hurricane Wind Scale (SSHWS) ................................ ................ 28 3 1 Data cleaning process ................................ ................................ ........................ 70 4 1 Initial PCA total variance explained ................................ ................................ .... 99 4 2 Revised PCA total variance explained ................................ ................................ 99 4 3 Split sample validation comparison ................................ ................................ .... 99 4 4 Rotated component matrix ................................ ................................ ............... 100 4 5 Descriptive statistics for the HPKS and TSES ................................ .................. 100 4 6 Descriptive statistics of HPKS scores by Region of Florida .............................. 10 0 4 7 HPKS scores by R egion of Florida ................................ ... 101 4 8 Descriptive statistics of HPKS scores by States ................................ ............... 102 4 9 HPKS scores by States ................................ .................... 102 4 10 Descriptive statistics of HPKS scores by Education Levels (Florida) ................ 103 4 11 ANOVA of HPKS scores by Educational Levels (Florida) ................................ 103 4 12 Descriptive statistics of HPKS scores by Education Levels .............................. 103 4 13 ANOVA of HPKS scores by Educational Levels ................................ ............... 104 4 14 Descriptive statistics of HPKS scores by Household Incomes (Florida) ........... 104 4 15 HPKS scores by Household Incomes (Florida) ................. 105 4 16 Descriptive statistics of HPKS scores by Household Incomes .......................... 106 4 17 HPKS scores by Household Incomes ............................... 107 4 18 Descriptive statistics of HPKS Age (Florida) .............. 108 4 19 ANOVA of HPKS Age (Florida) ................................ .. 109 4 20 Descriptive statistics of HPKS Age ............................ 109 4 21 ANOVA of HPKS Age ................................ ................ 109

PAGE 9

9 4 22 Descriptive statistics of TSES scores by Florida Region ................................ .. 109 4 23 TSES scores by Florida Regions ................................ ...... 110 4 24 Descriptive statistics of TSES scores by State ................................ ................. 111 4 25 TSES scores by State ................................ ....................... 111 4 26 Descriptive statistics of TSES scores by Education Levels (Florida) ................ 112 4 27 ANOVA of TSES scores by Education Levels (Florida) ................................ .... 112 4 28 Descriptive statistics of TSES scores by Education Levels .............................. 112 4 29 ANOVA of TSES scores by Education Levels ................................ .................. 112 4 30 Descriptive statistics of TSES scores by Household Incomes (Florida) ............ 113 4 31 ANOVA of TSES scores by House hold Incomes (Florida) ............................... 113 4 32 Descriptive statistics of TSES scores by Household Incomes .......................... 113 4 33 ANOVA of TSES scores by Household Incomes ................................ .............. 114 4 34 Descriptive statistics of TSES scores Age (Florida) ............... 114 4 35 TSES Age (Florida) .................... 115 4 36 Descriptive statistics of TSES Age ............................. 115 4 37 Games Howell post hoc of TSES Age ....................... 116 4 38 Correlations of HPKS and Demographic Characteristics ................................ .. 116 4 39 Model summary of HPKS model 1 (Florida) ................................ ..................... 116 4 40 Regression coefficients of HPKS model 1 (Florida) ................................ .......... 117 4 41 Model summary of HPKS model 2 ................................ ................................ .... 117 4 42 Regression coefficients of HPKS model 2 ................................ ........................ 117 4 43 Model summary of TSES model 1 (Florida) ................................ ...................... 117 4 44 Regression coefficients of TSES model 1 (Florida) ................................ .......... 117 4 45 Model summary of TSES model 2 ................................ ................................ .... 118 4 46 Regression coefficients of TSES model 2 ................................ ........................ 118

PAGE 10

10 5 1 Summary of hypotheses conclusions ................................ ............................... 155 5 2 Values assigned to Income and Florida Region variables for HPKS model 1 .. 155 5 3 Values assigned to Income and States variables for HPKS model 2 ................ 156 5 4 Values assigned to Florida Region Age, and Education for TSES model 1 ..... 156 5 5 Values assigned to Age and State variables for TSES Model 2 ....................... 156 5 6 Descriptive generalizations ................................ ................................ ............... 156 A 1 Gender distribution of sample ................................ ................................ ........... 160 A 2 Age distribution of sample ................................ ................................ ................ 160 A 3 Education distribution o f sample ................................ ................................ ....... 160 A 4 Income distribution of sample ................................ ................................ ........... 160 A 5 Distribution of sample by state ................................ ................................ .......... 161 B 1 KMO and Bartlett's Test ................................ ................................ .................... 162 B 2 Total variance explained by PCA ................................ ................................ ...... 162 B 3 Rotated Component Matrix ................................ ................................ ............... 163 B 4 Component 1: Inter item correlation matrix ................................ ...................... 164 B 5 Component 1: Reliability statistics ................................ ................................ .... 164 B 6 Component 2: Inter item correlation matrix ................................ ...................... 164 B 7 Component 2: Reliability statistics ................................ ................................ .... 165 B 8 Component 3: Inter item correlation matrix ................................ ...................... 165 B 9 Component 3: Reliability statistics ................................ ................................ .... 165 B 10 Component 4: Inter item correlation matrix ................................ ...................... 165 B 11 Component 4: Reliability statistics ................................ ................................ .... 166 B 12 Component 5: Inter item correlation matrix ................................ ...................... 166 B 13 Component 5: Reliability statistics ................................ ................................ .... 166 B 14 Component 6: Inter item correlation matrix ................................ ...................... 166

PAGE 11

11 B 15 Component 6: Reliability statistics ................................ ................................ .... 167 C 1 Hypothesis H1 A1 : HPKS by Florida Region ................................ ....................... 168 C 2 Hypothesis H1 A2 : HPKS by State ................................ ................................ ..... 168 C 3 Hypothesis H1 B1 : HPKS by Education (Florida) ................................ ................ 168 C 4 Hypothesis H1 B2 : HPKS by Education ................................ .............................. 168 C 5 Hypothesis H1 C1 : HPKS by Income (Florida) ................................ .................... 169 C 6 Hypothesis H1 C2 : HPKS by Income ................................ ................................ .. 169 C 7 Hypothesis H1 D1 : HPKS by Age (Florida) ................................ ......................... 169 C 8 Hypothesis H1 D2 : HPKS by Age ................................ ................................ ....... 169 C 9 Hypothesis H2 A1 : TSES by Florida Region ................................ ....................... 170 C 10 Hypothesis H2 A2 : TSES by State ................................ ................................ ...... 170 C 11 Hypothesis H2 B1 : TSES by Education (Florida) ................................ ................ 170 C 12 Hypothesis H2 B2 : TSES by Education ................................ .............................. 170 C 13 Hypothesis H2 C1 : TSES by Income (Florida) ................................ .................... 171 C 14 Hypothesis H2 C2 : TSES by Income ................................ ................................ .. 171 C 15 Hypothesis H2 D1 : TSES by Age (Florida) ................................ ......................... 171 C 16 Hypothesis H2 D2 : TSES by Age ................................ ................................ ........ 171 D 1 Descriptive statistics of HPKS scores by Time Lived in Community ................. 172 D 2 HPKS scores by Time Lived in Community ...................... 174 E 1 Descriptive statistics of HPKS by Minors in household ................................ ..... 175 E 2 Independent samples test of HPKS by Minors ................................ ................. 176

PAGE 12

12 LIST OF FIGURES Figure page 1 1 Billion $ US Weather/Climate Event Frequency ................................ ................. 28 2 1 Social Cognitive Theory ................................ ................................ ...................... 53 2 2 Protection Motivation Theory ................................ ................................ .............. 53 2 3 Extended Parallel Process Model ................................ ................................ ....... 53 2 4 Communication Infrastructure Theory ................................ ................................ 54 2 5 Proximal Processes adapted from Bronfenbrenner ................................ ............ 54 2 6 Person Characteristics adapted from Bronfenbrenner ................................ ........ 55 2 7 Ecological Systems Theory Model ................................ ................................ ...... 55 2 8 Process Person Context Time Model adapted from Bronfenbrenner ................. 56 2 9 .................... 57 4 1 Boxplots of HPKS scores by Florida Regions ................................ ................... 119 4 2 Means of HPKS sc ores by Education Levels (Florida) ................................ ..... 119 4 3 Boxplots of HPKS scores by Education Levels (Florida) ................................ .. 120 4 4 Means of HPKS scores by Education Levels ................................ ................... 120 4 5 Means of HPKS scores by Household Incomes (Flo rida) ................................ 121 4 6 Boxplots of HPKS scores by Household Incomes (Florida) .............................. 121 4 7 Means of HPKS scores by Household Incomes ................................ ............... 122 4 8 Means of HPKS Age (Flo rida) ................................ .... 122 4 9 Boxplots of HPKS Age (Florida) ................................ 123 4 10 Means of HPKS Age ................................ .................. 123 4 11 Boxplots of TSES scores by Florida Regions ................................ ................... 124 4 12 Boxplots of TSES scores by State ................................ ................................ .... 124 4 13 Means of TSES scores by Education Levels (Florida) ................................ ...... 125

PAGE 13

13 4 14 Boxplots of TSES scores by Education Levels (Florida) ................................ ... 125 4 15 Means of TSES scores by Education Levels ................................ .................... 126 4 16 Boxplots of TSES scores by Education Levels ................................ ................. 126 4 17 Means of TSES scores by Household Incomes (Florida) ................................ 127 4 18 Boxplots of TSES scores by Household Incomes (Florida) .............................. 127 4 19 Means of TSES scores by Household Incomes ................................ ................ 128 4 20 Means of TSES Age (Florida) ................................ .... 128 4 21 Means of TSES Age ................................ .................. 129 4 22 Scatterplot of studentized residuals by standardized predicted values ............. 129 4 23 Partial regression plots of HPKS versus Demographic Characteristics ............ 130 4 24 Histogram of HPKS score frequency versus regression standardized residual 130 5 1 Hurricane return period ................................ ................................ ..................... 157 5 2 Mean HPKS score by Household Income ................................ ........................ 157 5 3 Mean of HPKS scores by Age ................................ ................................ .......... 158 5 4 Mean of TSES scores by Region of Florida ................................ ...................... 158 5 5 Mean HPKS score by Time Living in Community ................................ ............. 159 B 1 Screeplot of PCA component eigenvalues ................................ ....................... 162 D 1 Mean HPKS score by Time Lived in Community ................................ .............. 173

PAGE 14

14 LIST OF ABBREVIATIONS ASCE The American Society of Civil Engineers (ASCE) established in 1868, strives to propose solutions to maintaining and updating deteriorating infrastructure, pushes for increases in educational requirements of engineering licensure, and encourages the contributions of civil engineers to a sustainable world. ASCE also CE, n.d.) CIT The Communication Infrastructure Theory (CIT) categorizes communication into neighborhood storytelling networks and the community action context that those networks operate within. Participation in this network and context influences civic e ngagement and influences individual behavior (Kim and Ball Rokeach, 2006). EPPM The Extended Parallel Process Model (EPPM) is based on fear appeal theory; persuasive messaging intended to scare individuals into changing their behavior or suffer terrible consequences. The EPPM posits that individuals process these messages and proceed in three ways: dismiss the threat and do nothing, take the threat seriously and make adaptive changes to control danger, or take the threat serious and make maladaptive chang es to control fear (Witte, 1992). EST Ecological Systems Theory (EST). Bronfenbrenner included EST in his early work. It focused on the environment and external factors of the developing individual and saw the first iteration of the micro meso exo a nd macro system model. This theory lacked any of the biological factors that are internal to individuals and the time aspect later referred to as the chrono system (Bronfenbrenner, 1979). FEMA The Federal Emergency Management Agency (FEMA) is the US agen cy whose mission is to support citizens and first responders to ensure that as a nation we work together to build, sustain and improve our capability to prepare for, protect against, respond to, recover from and mitigate all hazards (FEMA, 2015). GIS Geog raphic Information Systems (GIS) are tools for working with spatial data that relates to geographic space. Some of the applications of GIS include urban planning, biology, epidemiology, forestry, and natural hazard analyses (Huisman and By, 2009).

PAGE 15

15 GOES The Geostationary Operational Environmental Satellite (GOES) is 30+ year joint mission between NOAA and NASA that remains stationary over the equator in the western hemisphere. The purpose of GOES is to monitor weather and other environmental conditions f or the NESDIS (GOES Project, 2014). HPKS The Hurricane Preparedness Knowledge Scale (HPKS) is the name given to the first component that resulted from a Principal NESDIS The National Environmental Satellite, Data, and Information Service (NESDIS) acquires and manages operational environmental satellites and the NOAA National Data Centers. This data is used by various government users from meteorological forecasters to mili tary applications requiring real time observations (NESDIS, 2014). NHC The National Hurricane Center (NHC) is a part of NOAA that specializes in hurricane watching, tracking, and predicting the paths of these disturbances. The NHC is responsible for iss uing coastal tropical cyclone watches and warnings for the United States and its Caribbean territories (National Hurricane Center, 2014a). NASA The National Aeronautics and Space Administration (NASA) is focused on four principal directorates: Aeronautic s, Human Exploration and Operations, Science, and Space Technology. This scope places much more than space exploration and astronauts on Administration, 2015). NOAA The National Oceanic and Atmosphe ric Administration (NOAA) oversee daily weather forecasts, climate monitoring, fisheries management, coastal restoration, and supports marine commerce. Their scope of services touches some aspect of more than a third of the United States gross domestic pro duct (National Oceanic and Atmospheric Administration, n.d.a) PMT The Protective Motivation Theory (PMT) was created to help explain fear appeals by reducing them to four factors of the appeal: the perceived threat, the perceived probability or vulnerabil ity, the efficacy of the preventative behavior, and the self efficacy of the individual (Rogers, 1975). PPCT The Process Person Context Time (PPCT) model that emerged as Bronfenbrenner developed the EST to give more importance to biological factors and proximal processes (Bronfenbrenner, 2005).

PAGE 16

16 SCT Social Cognitive Theory (SCT) is based on the belief that knowl edge acquisition is directly related to observations and the context in which they occur. Self efficacy is also a strong factor for learning in SCT as well (Bandura, 1986). SSHWS The Saffir Simpson Hurricane Wind Scale (SSHWS) is the five category scale that hurricanes are measured by. It is based thresholds for one minute sustained wind speed at 10m over unobstructed exposure (NHC, 2014b). TSES The Trust in Support Entities Scale (TSES) is the name given to the third component that resulted from a Princ ipal Components Analysis

PAGE 17

17 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science HURRICANE PREPAREDNESS OF HOMEOWNERS IN THE SOUTHEAST UNITED STATES By Charles Bradford Sewell December 2015 Chair: Randall A. Cantrell Major: Family, Youth and Community Sciences The U nited S tates is experiencing the long est drought from major hurricanes on record since 1851 This 10 year absence from major hurricane s has lulled many homeowners into a false sense of security allowing preparedness to become a waning priority in th eir minds. Emergency managers across Southeastern Georgia, Florida, and the Gulf Coast are tasked with keeping th e public and policymakers prepared at all times for such events. T his quantitative study explored the relationships between demographic characteri stics ( i.e. l ocation, e ducation, i ncome, and a ge) and two study generated measures; the Hurricane Preparedness Knowledge Scale and the Trust in Supp ort Entities Scale. Through the discussion of several hypotheses, the strength of some indicators was observ ed while possible reasons for the non (statistical) significance of others was discussed. The implications of this study will add to the body of knowledge on disaster preparedness as well as providing insights into demographic indica tors of trust in aid pr oviders. Findings from this study and subsequent research resulting from it will aid emergency managers and policymakers in better understanding what areas need more

PAGE 18

18 attention while reinforcing those areas that are understood well enough to probe more deep ly via advanced discussions regarding preparedness. The findings about support entities will provide insight to what populations may benefit from outreach programs to advance understanding of the aid available to homeowners if needed. A more knowledgeable public that has an increased confidence in the available support will also allow for a more efficient allocation of resources during times of need caused by major hurricanes.

PAGE 19

19 CHAPTER 1 INTRODUCTION Natural hazards are present throughout the world. From tsunamis to earthquakes, forest fires to tornadoes wherever people live; they are not beyond the reach of these natural hazards. Hurricanes, also known as typhoons or cyclones in other parts of the world, are different from other natural hazards in that their presence can be detected earlier than these other hazards and their predicted paths can give several hours and in some cases, days of warning to populations that may be affected Through modern meteorolog ical technology, hurricanes can be tracked in the Atlantic Ocean from the ir inceptions as atmospheric disturbances off of the coast of the African continent and either weaken or strengthen as they move w est towards the Caribbean islands, Central America, a nd the Gulf and Atlantic coasts of the United States (US) ( N ational O ceanic and A tmospheric A dministration 2012). F rom early formation to dissipation t his westerly progress can be tracked by the US National Operational Environmental Satellite (GOES). It is possible for a weather disturbance to be tracked, monitored, and evalu ated for periods as long as two weeks before landfall (N ational O ceanic and A tmospheric A dministration n.d. b ). When observing patterns where hurricanes have made US landfall since 1900, Southeast Florida and the Florida Keys have the highest number of str ikes, followed by Louisiana, Mississippi, Texas, and North Carolina (Landsea, 2015). Despite these areas being more prone to hurricanes, preparedness projects are often neglected by federal, state, and local governments due to cost or low priorit ization. F or example, the lack of levee upkeep undeniably exacerbated the effects of Hurricane Katrina on New Orleans

PAGE 20

20 (Meyer, 2012). The shortsightedness in avoiding the initial economic expenditures of preparedness is also a gamble against the much larger expenditu res repairing, or often times replacing, unfortified infrastructure. Others have suggested that politics impact decisions about emergency preparedness as much as economics Healy and Malhorta (2009) and Gasper and Reeves (2011) provided evidence that polit icians underinvest in preparation projects because the electoral payoffs are higher for bringing disaster funding to afflicted areas post disaster. The Federal Emergency Management Agency ( FEMA ) recently changed eligibility for federal disaster preparednes s funds and these changes will go into effect March 2016 These changes will requir e environmental or climate conditions that may affect and influence the long term into their emergency preparedness decisions (Boyer, 2012). This required acknowledgment of changing climates has unfortunately been politicized and would therefore necessitate changes in the current views of the Governor s offices of Texas, Louisiana and Flori da to alter their emergency preparedness decisions ( Satija, 2014; Alpert, 2015 ; Korten, 2015 ). Unfortunately, it is not just government entities that are unready and ill prepared Many residents still do very little to fortify their homes for a hurricane or prepare their families for an evacuation. The Readiness Quotient Public Opinion Survey conducted by the Council for Excellence in Government found that the American public was ill prepared for emergencies of any sort whether natural or man made ( T he Council for Excellence in Government 2007).

PAGE 21

21 Currently, there is added urgency to learning about and overcoming perceived barriers to hurricane preparedness. This is due to the fact that the continental US is currently in the longest running period of time without landfall of a major ( C lass 3 or greater) hurricane in recorded storm history (Hall, 2015). Based on more than 150 years of historical hurricane data, researchers estimate th at a nine year drought of major hurricanes only statistically occurs every 177 years (Hall and Hereid 2015). Other studies have shown that risk perceptions of natural hazards deteriorate over time, so it is reasonable to assume that Floridians are less pr epared today than they have been in National Hurricane Center (Leger, 2012) With the gr owing prevalence of Geographic Information Systems ( GIS ) being utilized by those modeling risk assessments, more is being understood about the vulnerabilities of communities (Taramelli, Valentini, and Sterlacchini, 2014). GIS is not only being used to visu ally represent data, but advancements are allowing the actual analysis of data to be performed within the GIS software, through geospatial analysis (Taramelli et al., 2014). Others advocate for emergency planners and policy makers to use community vulnerab ility maps to identify and work with high risk areas for disaster preparation and response (Bergstrand, Mayer, Brumback, and Zhang, 2015). In response to this, t he Florida Division of Emergency Management uses GIS technology to display a wide array of info rmation for the public ranging from evacuation routes to the results of

PAGE 22

22 studies that show projected sea level rise maps (Florida Division of Emergency Management, n.d.). Rationale for Study The National Oceanic and A program, the Insurance Services Office/Pr operty Claims Service, and other national and regional sources (Blake and Gibney, 2011). The most costly of this data is aggregated US Billion d ollar Weather/Climate Disaster R eport (Blake and Gibney, 2011) This report only i ncludes events that surpass $1 b illion in economic impact. This includes data of various weather/climate hazards including: drought, flooding, freeze, winter storm, severe storm, wildfires and tropical cyclone (hurricane). By categorizing the frequency of these e vents by decade, it is clear to see that the frequency of these events is increasing (Figure 1 1 ). Furth er, it is alarming that the 49 b illion dollar events from 2010 2014 only account for half of this decade (Blake and Gibney, 2011) This trend shows an i ncreased frequency of all weather and climate events exceeding this $1 billion threshold. Experts posit that the events specifically occurring along the coast will continue to experience increases in their economic impacts. This is due to the increased dev elopmental investments in the coastal plains, the increased population living near the coa st, and the gradual rise of sea level (Hinkel, et al., 2013). This means that future floodin g and hurricane events will surpass the billion dollar threshold with increasing frequency in the future.

PAGE 23

23 Although the frequency of these events is alarming, the loss of life is even more tragic. During the course of the 178 events included in this chart, 9,179 lives were lost. The largest death tolls occurre d rought/ h eat d rought/ h eat wave (502 deaths) (NOAA NCEI, n.d.) Of the type of events recorded in the US Billion D ollar Weather/Climate Disaster R eport hurricanes ar e commonly tracked and new technolog ies provide information that enhance warning to coastal residents as they traverse westward. This journey often meanders through the Caribbean past ocean based observation posts that are able to help predict their severi ty and direction. By the time these storms make landfall i n the U S, they often have been monitored for more than a week by satellite, oceangoing vessels, radar, and specialized aircraft (Smith and Matthews, 2015). Of the hurricanes recorded since 1851, Bla ke et al. (2007) estimate that 40 percent of hurricanes that affect the US make landfall in Florida. In addition, approximately 60 percent of all Category 4 or higher US hurricanes strike Florida or Texas (Blake, Landsea, and Gibney, 2007). Peacock, Brody, and Highfield (2005) suggest that further education of the public is needed so it can more accurately assess its hurricane risk. Purpose of Study The purpose of this study is to examine the associations that socio demographic characteristics and homeowner perceptions have with hurricane preparedness in Southeastern Georgia Florida and t he Gulf Coast. The findings will add to what is disasters, specifically hurricanes. focus o n hurricanes, the targeted

PAGE 24

24 population will be homeowners living in Southeast Georgia Florida, as well as in counties near the Gulf Coast of Texas, Louis iana, Mississippi, and Alabama. towards preparedness, socio demographic characteristics will be explored first and then compared to the findings of previous studies. Next the perceptions of these homeowners will be examined for possible relationships to their views regarding their house hold level of preparedness. Their views will be mapped to show any geospatial patterns that may be present. This visual representation generated through GIS techniques will increase the understanding of the data as opposed to being solely represented by tables and spreadsheets this study will explore the associations that socio demographic characteristics have with the level of trust in the ability of their communi ty, non profit organizations an d federal government to assist during a crisis. Again, these findings will be graphically represented on maps created using GIS techniques. This increased understanding of motivations or barriers to hurricane preparedness will provide valuable insight to g overnmental agencies, policy makers, and insurance organizations as they strive to better prepare homeowners and communities for natural disasters. Findings from this study will enable program managers to develop targeted messaging with tailored informatio n that increase s homeowner and community resiliency designed to lessen recovery time and/or damage incurred by future storms and hurricanes. Significance of Study Although the demographic characteristics of hazard preparedness have been explored by many re searchers (Baker, 2011; Becker et al., 2012; Donahue et al., 2014;

PAGE 25

25 Meyer et al., 2015), this only explains a portion of what influences the preparedness decision making process. Other researchers have explored other factors such as social vulnerability (Be rgstrand et al., 2015), risk perception (Peacock et al., 2005; Meyer et al., 2015), and myopic behavior (Kunreuther et al., 2012). Due to inherent differences in individuals (e.g., where they were raised, their occupation, whether they have children polit ical party affiliati on), statistical decision making models are not intended to explain any process completely. The best that researchers can hope to achieve is to explain more reasons for variation in a particular phenomenon than what current models are c apable of accounting for Given the abundance of decision making models that exist for the adoption of healthy behaviors, medical decisio ns, and even military decisions, it is obvious there is no single, best approach to developing these models. By researc hing more a bout the intent of this study is to explain more about why people vary in their preparation for and resilience to hurricanes than has been documented previously. This study also seeks to make sp atial patterns in this data more easily understood by mapping its findings. Morrow (1999) advocated that mapping areas that are more vulnerable to disasters would allow emergency planners and policy makers to work with these high risk areas to increase the ir resilienc e There currently exists a gap between data that is available and data that is easily understood and clearly represented; this study will narrow that gap. Definition of Terms Disaster A serious disruption affecting a community or population, c ausing deaths, injuries, or damage to property, livelihoods, or the environment, that exceeds the ability of the affected

PAGE 26

26 community to cope using its own resources (UN/ISDR, 2004: 17) Disaster Preparedness The International Federation of Red Cross and Red Crescent prepare for and reduce the effects of disasters. That is, to predict and where possible, prevent disasters, mitigate their impact on vulnerable populations, and respond to and effec tively cope with Disaster Resilience A outcome to ensure benign or small scale negative consequences. Indeed, the goal of disaster risk management is to guarante e minimal loss of life and livelihoods and to allow the Emergency Management The practice of identifying, anticipating, and responding to the ris ks of catastrophic events in order to reduce to more acceptable levels the probability of their occurrence or the magnitude and duration of their social impacts (Lindell and Perry, 2004). Hazards Physical activities, phenomena, or human activities having potential to cause injury, loss of life, damage to property, economic and social disruption, or environmental degradation (Kapucu and Ozerdem, 2013; Makoka and Kaplan, 2005). Hurricane/Typhoon A tropical cyclone in which the maximum sustained surf ace wind (using the US 1 minute average) is 64 kt (74 mph or 119 km/hr) tropical cyclones east of the International Dateline to the tro pical cyclones in the Northern Hemisphere, west of the International Dateline (NHC, 2014 b ). Individual Resilience The ability of an individual to maintain healthy psychological and physical well being despite exposure to adversity (Bonanno, 2004) Indivi dual resilience is partly trait and partly dynamic process that are promoted by two groups of generic factors: 1. Personal attributes such as social competence, problem solving, autonomy, self efficacy and sense of future and purpose; 2. Contextual, environmenta l influences such as peers, family, work, school and local community (Boon et al., 2012)

PAGE 27

27 Landfall The intersection of the surface center of a tropical cyclone with a coastline (NHC, 2014). Major Hurricane A hurricane that is classified as Category 3 or higher on the Saffir Simpson Hurricane Wind Scale (SSHWS) (NHC, 2014 b ). Man made or Technological Hazards These include but are not limited to, industrial accidents, chemical spills, explosions, acts of terrorism, and fires that are started by anthropological sources (Boon et al., 2012). Natural Disasters An event can only be called a disaster if (a) it is triggered by the combination of a na tural hazard (or several hazards) and vulnerable local conditions that (b) results in a disruption of the functioning of individuals, a community or society, and (c) requires external assistance for the subsequent impacts to be adequately dealt with (Wa m sler, 2014) Natural Hazards These include hurricanes, earthquakes, tsunamis, floods, droughts, windstorms, famine, epidemics, and wildfires caused by lightening. (Boon et al., 2012, 383; Rivera and Kapucu, 2015) These hazards cannot be prevented from oc curring but are often able to be predict ed (Rivera and Kapucu, 2015) Preventative Behaviors Any activity undertaken by individuals to prevent a disaster or mitigate the damage done by a disaster (Savoia, Lin, and Viswanath, 2013). Resilience, Community The ability to adapt through the redevelopment of the community evolving understanding of external forces with which it must contend (Kapucu et al., 2013, p. 357) Resilience process linking a set of adaptive capacities to a positive (Norris et al., 2008, p. 130). Risk Perception for the society or p. 176). Saffir Simpson Hurricane Wind Scale (SSHWS) intensity at the indicated time ( Table 1 1 ) The scale provides examples of the type of dama ge and impacts in the United States associated with winds of the indicated intensity.

PAGE 28

28 Table 1 1. Saffir Simpson Hurricane Wind Scale (SSHWS) (NHC, 2014) Category Wind Speed (mph) Damage 1 74 95 Very dangerous winds will produce some damage 2 96 110 Extremely dangerous winds will cause extensive damage 3 111 129 Devastating damage will occur 4 130 156 Catastrophic damage will occur 5 > 156 Catastrophic damage will occur Figure 1 1. Billion $ US Weather/Climate Event Frequency

PAGE 29

29 CHAPTER 2 LITERATURE REVIEW Theoretical Framework To examine the interactions that socio demographic characteristics and homeowner perceptions have on hurricane preparedness, it is imperative to begin with a solid theoretical foundation. A foundation in ex tant research helps to provide focus throughout this study Also, understanding what other researchers have empirically tested in the past helps to focus and those conducted previously. In this way, gaps for future resea rch can be identified by discover ing new connections that have not been explored in the past (Tudge, Mokrova, Hatfield, and Karnik, 2009). In reviewing previous studies regarding disaster preparedness, common themes emerged. One of these themes was that di saster preparedness is considered by many to be a knowledge gaining process that is greatly influenced by contexts of place, personal characteristics, and interactions with environmental factors. Further review revealed components of several theories and m odels that were prominent and recurring across disciplines. A second common theme was that many of these factors were aligned with various propositions of B ronfenbrenner (2005) b ioecological theory of human development This theory and its propositio ns w ill be further explained later in this chapter. T he bioecological theory alignment with these themes and its ability to offer a direction for future research were both compelling reasons for choosing it, rather than the many other possible theories revie wed. Savoia, Lin, and Viswanath (2013) conducted a systematic review of articles published in 2009 addressing emergency preparedness in public health In contrast to

PAGE 30

30 the understood importance of the expansion of knowledge and theory in existing literature, they found that less than 10 percent of empirical studies published made use of a theoretical framework. The se studies included theories that are among some of the most prolifically mentioned in the field of disaster preparedness. These included social co gnitive theory, protection motivation theory, extended parallel process model, and the communication infrastructure theory (Savoia et al., 2013). These four theories are important to this current study because they all have elements that are shared with Br development -the theory that was ultimately chosen to serve as the framework for this study. By briefly discussing these four theories t he reader can better understand research conducted by others and make conn ections that to help clari f y the process that eventually led to the bioecological theory The s ocial cognitive theory (SCT) i s a learning theory based on the notion that individuals knowledge about a subject is directly related to their observations of others and the context in which those observations occur (Bandura, 1986). SCT has been simplified and represented graphically (Figure 2 1) as three categories of factors: environmental cognitive, and behavior. The environmental factors are those that are perceived as either barriers or encouragements to the successful completion of a behavior. The cognitive factors are individuals way of thinking and their mindset often including the ir level of self efficacy produce results) towards a behavior (Bandura, 1977 ). Finally, the behavioral factors are the positive or negative reinforcements that an individual receives after completing an

PAGE 31

31 action (Bandura, 1986). Bandura (1986) also states that l earning and knowledge gain do not imply that behavior change will necessarily follow. The environmental factors are very similar to what Bronfenbrenner refers to as context in the Process Person Context Time (PPCT) Model proposition of the the PPCT in that thinking, mindset and self efficacy are internal to the individual and are The protection motivation theory (PMT) as developed by Rogers (1975), lent understanding to the effectiveness of using fear as a motivation to change attitudes and ultimately behavior. A simplified representation of the PMT (F igure 2 2) includes perceived severity, perceived probability/perceived susceptibility and efficacy of motivation to engage in preven tive behavior (Rogers, 1975). Perceived severity can be defined as the g ravity of the possible hazard. Perceived probability/perceived susceptibility is the likelihood of the hazard affecting the individual. It is important to differentiate this as a perceived probability and not an actual probability. If individual s feel they are at a high or low risk regardless of their actual susceptibility their belief is what they act on (Rogers, 1975). Perceived efficacy of response can be described as the indi preven tive action will lessen or mitigate the affects of the hazard. Again this is based on CT model. Perceived self

PAGE 32

32 severity, probability/susceptibility, and response efficacy can be viewed as the culmination of their experiences combined with their intelligence and lo gic capabilities. the PPCT also are involved. These processes rely on repeated increasingly complex interactions of individuals with their environments to facilitate learn ing (Bronfenbrenner, 2005). The extended parallel process model (EPPM) was based on the PMT and other fear appeal theories (Witte, 1992). It is more complex than the previously mentioned theories but still easily understood. The big gest departures from oth er fear appeal theories are that the EPPM re introduced fear as a variable and that it sought to explain why fear appeals are less effective when applied to some individuals (Witte, 1992). As one starts at the left of the graphical EPPM (Figure 2 3), the d ecision making process begins with the external stimuli. These stimuli consist of the message components, which are the fear appeals themselves as they are presented to the individual. Once received, the individual begins to process these messages. First, individual s evaluate their perceived threat. If the individual determines there is little to no threat the process stops and goes no further. If a threat is perceived as valid then individual s assess their perceived self efficacy and the efficacy of poss ible responses to the hazard (Witte, 1992). Witte (1992) suggested fear was elicited if the individual perceived the threat to be moderate or high. The process that individual s take is dependent on their perceived efficacy. If the individual believes effic acy is high, then fear is managed and the danger control process

PAGE 33

33 is initiated. If the perceived efficacy of the individual or response is low, then fear pushes the outcomes to follow the fear control process. Perceptions of the same threat can be different depending on factors of individuals. Many of these factors are part of the person, process, and context components of the PPCT model, as mentioned with the PMT. The EPPM supports that the threat assessment and decision to proceed based on danger or fear i s greatly motivations. This further supports the argument for using the bioecological theory of human development in the study of hurricane preparedness. Finally, t he communication infrastructure theory (CIT) is an ecological approach to explore the importance of storytelling to civic engagement (Kim and Ball Rokeach, method of understanding an ecological relationship between a co mmunication environment and communicative actions by articulating and empirically unveiling the communication infrastructures of diverse urban residential Rokeach, 2006, p 176). The CIT consists of two major components: the neig hborhood storytelling network and the communication action context. The neighborhood storytelling network consists of formal and informal communication ranging from conversations between neighbors to advertisement fliers from local non profit organizations The communication action context s are the resources that promote that communication such as libraries, schools, churches, and parks (Figure 2 4) early ecological systems theo ry representation but also shares many of the factors as

PAGE 34

34 well, which is somewhat expected given that the CIT also is an ecological model (Kim and Ball Rokeach, 2006) The neighborhood storytelling network involves many entities that are considered parts of context proposition of the PPCT. The communication action context has multiple factors Components of several theories indica te that individuals are, in many ways, cognitive aptitude and characteristics. The idea that behavior can be learned through modeling and repetition is implicitly present thro theory of human development. Bronfenbrenner (2005) also believed there existed and sometimes negative ways. These unintended influen ces are part of the complexity that renders the study of human behavior so challenging. To more fully understand bioecological theory of human development, it is helpful to study its evolution. The components of the e cological f ramework for h uman d evelopment can be traced back to his 1943 doctoral dissertation at the University of Michigan (Bronfenbrenner, 1999). In 1979, he formally intr oduced this model by publishing Ecological Framework for Human Development: Experiments by nature and design (Bronfenbrenner, 1979). From that earliest version, Bronfenbrenner continuously developed, refined, and added to his theory of human development until his death in 2005 (Tudge et al., 2009) Bronfenbrenner addressed this assessing,

PAGE 35

35 revising, and extending as well as regretting and even renouncing some of the 1989, p. 187). This authors cite the Ecological Systems Theory (EST) and then pick and choose what they need from more than 30 years of progressive versions, including concepts that no longer are supported by Bronfenbrenner (Tudge et al., 2009). This current research project u s es la t ter work of 1999 to 2005 (Boon et al., 2012; Bronfenbrenner, 2005; Tudge et al., 2009) Properties of the Bioecological Model an acknowledgement that biology, in the form of genetics, impacts development. This addition is an advancement from the original ecological model that focused solely on external influences on the individual. This change brought with it mor e focus on the processes of human development and its pivotal concept in the context of this hurricane preparedness research study (Tudge et al., 2009). Person Context Time (PPCT) Mode l as the foundation of his theory (Tudge et al., 2009). Within s and their environment. Bronfenbrenner began to describe proximal processes as a key factor in human development (Figure 2 5) (Bronf enbrenner, 1999). These proximal processes can be

PAGE 36

36 2005, p. 6) Some examples of these processes are playing with children, reading, bedtime routines for children, and personal hygiene habits. In the contexts of hurricane preparedness, this could range from as simple as discussing what the dangers of hurricanes are with children to evacuatio n drills or practice of installing window protection. s and their biological characteristics (age, gender, motivation, and intelligence) (Bronfenbrenner, 2005). These characteristics were subdivided into three types : demand, resource, and force characteristics (Figure 2 6) Demand characteristics are those attributes readily assessed by others. This includes age, gender, and physical appearance (Tudge et al., 2009). Resource characteristics are not readily apparent to others but can sometimes be induced from the demand characteristics. These include intelligence, past experience, as well as social and material resources (Tudge et al., 2009). The last type of characteristic is force characteristics; these pertain to int ernal drive, motivation, tenacity, and temperament. These force characteristics can explain much of the variances in development by individuals with similar demand and resource characteristics (Bronfenbrenner, 2005). An example of force characteristics i s when an entire neighborhood of homes all have about the same susceptibility to hurricane damage, yet its homeowners vary widely in their inclination to prepare for an approaching storm. e original e cological m odel of h uman d evelopment. Context can be represented as the four concentric circles of the origi nal ecological model (Figure 2 7 ). The circles radiate

PAGE 37

37 out wards from the individual s located at the center. Those closest to the individ ual s exert the most influence on their development. Examples of the microsystem are family, peers, school, and church. This basically can be any environment with which individual s spend a fair amount of time participating in activities and interacting (Bro nfenbrenner, 2005; Tudge et al., 2009). It follows directly that these groups can vastly influence The mesosystem represents the interactions of the groups of the microsystem with each oth er and with the entities of the exosystem These are interactions that occur between the microsystem group that the individual most associates with and other entities in the microsystem (Tudge et al, 2009). The individual is not directly involved in these interactions, yet the outcomes still affect the ind ividual (Bronfenbrenner, 2005). In the context of hurricane preparedness, this could be observed in the interaction between a program instructor and the entity that facilitates the program and how this aff ects the individual. workplace being an exosystem for her child. The child does not interact with the workplace, but if the mother is in a bad mood because of the workplace it affects the child when the m other returns home (Tudge et al., 2009). The exosystem is where the mass media, local politics, and social services interact with the individual indirectly. They influence the microsystem and individual but to a much lesser degree to the individual than th at of the microsystem (Bronfenbrenner, exosystem refers to the structures in which the entities of the microsystems exist. Local

PAGE 38

38 emergency management and state building code enforcement are examples of entities The macrosystem is the outermost ring and this is where social and societal norms are located (Bronfenbrenner, 2005). They represent some of the weakest influences on t he individual other systems are influenced by the macrosystem and all the other systems affect the macrosystem in turn (Tudge et al., 2009). The degree to which society views the efficacy of hurricane preparedness in general can affect individuals from the macrosystem. Finally, the chronosystem is the effect of time on all influences. An example would be the influence of experiencing tragedy such as Hurricane Katrina, which is overshadowin g of all other factors immediately after occurring, and then its influence diminishes with the passing of time (Bronfenbrenner, 2005). In the later iterations of the PPCT model the chronosystem is broken down into the subfactors of micro time, meso time, and macro time to correspond with whatever system was directly influenced ( Bronfenbrenner and Morris, 1998 ). Throughout the review of literature making use of the PPCT model, a graphical representation was unable to be located. Thus, in this current study, a simplified graphical representation has been created and is offered based on the explanations 2005. This is not a comprehensive model of the propositions from the bioecological theory; however, it in cludes many propositions on which Bronfenbrenner plac ed the most focus (Figure 2 8). Bronfenbrenner listed distinctive defining properties of the Bioecological Model in the

PAGE 39

39 form of propositions. S everal of those propositions already have been discussed as components of the PPCT model The remaining four are experience, strong mutual emotional attachment, internalization and third party role (Bronfenbrenner, 2005). Experience is a critical element of the model because merely being in an environment will not lead to development. The individual must interact with the environment and the persons comprising that environment to experience development al growth (Bronfenbrenner, 2005). The next proposition for human development is strong mutual emotional attachment. Development will occur faster and is more likely to be long er lasting if individual s care about the pe ople (e.g., parents, program leaders) with whom they are sharing proximal processes It is also important that these people reciprocate these feelings tow ards the individuals as well (Bronfenbrenner, 2005). A good example of this is comparing the development of small children and the ir differences in progress based on whether they like their teacher or not. Strong mutual emotional attachment leads to the ne xt proposition of internalization. Internalization is when individual s observe the actions of others in their microsystem and are interested and motivated to engage in related activities (Bronfenbrenner, 2005). The final proposition is the role of third pa rties, which e ssentially is microsystem individual. Bronfenbrenner (2005) also mentioned that it is helpful, but not essential for this third party to be of the opposite sex of contact. It is suggested that this helps to broaden the experiences and activities of the ind ividual (Bronfenbrenner, 2005).

PAGE 40

40 Applications of the Bioecological Theory of Human Development As more is learn ed about Urie Bronfe nbrenner, beyond that of his theories and models, especially his role in the development of the federal Head Start program in 1964, it is not surprising that the EST was initially used in the field of early childhood mostly parents and teachers in the early days of the EST (Bronfenbrenner Center for Translational Research, 2014). During the p Development: have used the EST as a framework as originally intended. Researchers rapidly extended the applicability of the EST from childhood development to also include education studies. Some cont emporary examples of education applications include Maynard, of middle school and high school students. The bioecological model was a good fit because of the complexity of in fluencing factors that affect engagement. Further, the use of the bioecological model allowed their study to include biological (internal) factors in addition to ecological (external) factors that influence school engagement (Maynard, Beaver, Vaug hn, DeLis i, and Roberts, 2014). Another logical area of research that stemmed from child development and and progress. Harding, Morris, and Hughes (2015) focused on maternal educ ation and

PAGE 41

41 parenting activities as they fit into the microsystem, exosystem, and mesos ystem (Hardi ng, Morris, and Hughes, 2015). Gonzalez and Barnett (2014) further explored familial and parents interactions on children by studying maternal psychological distress among Mexican origin families. The subset of this population that Gonzalez and Barnett (2014) focused on was families with non biological father figures and the affect that their support had on the mothers. Gonzalez and Barnett made use of the PPCT model but were unable to truly utilize the sectional res of the model helped to examine the family structure of each participant family. As the EST continued to be refined and more widely used within the previously mentioned fields, its visibility among researchers in other fields also continued to grow. This encouraged propagation into additional fields within the context of human development where a multitude of influences all act on indi viduals, causing different decisions and rates of development to occur. The sport s sciences are one of these fields that have widely used bioecological theory to explore youth sports (Domingues and Goncalves, 2014). The interdisciplinary and integrative fo cus on youth development has spurred the acceptance as a theoretical framework for youth sports. Bioecological theory also has been used to examine the complex influences on barriers and supports for female coaches. Mahoney, Gucciardi, Mallett, and Ntouman is (2014) focused on the demand, resource, and force characteristics from the PPCT model to discuss mental toughness among adolescents who were top performers in sports, academia, and music. This focus group

PAGE 42

42 study was conducted by sports psychologists and explored the possibility of shared characteristics that led to mentally tough adolescents. The microsystems of coaches, parents, and peers also have been explored in sport s sciences (Domingues and Goncalves, 2014). Bioecological Theory in Emergency Prepare dness To explore the application of the bioecologic al theory of human development to disaster (hurricane) preparedness One way this can be done is through observing its usage in the works of other researc hers who also have been concerned with preparedness. Another method in assessing a processes. Hurricane preparedness is not easily compartmentalized into a few simp le environmental factors or interactions. These complex interactions that involve many levels of functioning (e.g., individual to microsystem, microsystem to microsystem) serve as a web within which resilience and development emerge These thoughts were re flected in by Masten and Obradovic (2008) when they stated, development, is said to arise from processes of interaction across multiple levels of r, 2012, p. 389). The comparison of the processes that lead to development and resiliency serves as support that similarities exist between the complexities of the interactions that lead to these two concepts. Thus, there exist theories that are appropriate perspectives to observe developmen t as well as resiliency. This study suggests that the bioecological theory of huma n development is such a theory. The contextual components of the PPCT discuss how an individual continually interacts with other individuals, organizations, and information ( Bronfenbrenner, 2005).

PAGE 43

43 Boon et al. (2012) represented some of these entities that can interact with an individual and influence their feelings and beliefs towards em ergency preparedness (Figure 2 9 ). These interactions and experiences can be directly in co ntact with the individual or they can occur as proximal processes predispositions (Bronfenbrenner, 2005; Boon et al., 2012). Another aspect of bioecological theory that directly coincides with contempor ary understanding of emergency preparedness is that the beliefs and values of individual s cannot be separated from the environment in which they are located (Bronfenbrenner, 2005; Boon et al., 2012). Thus, even a complete understanding of the individual or the environment will still only represent a partial explanation of the motivati ons for development. In tracing the history of the t to ecology have been echoed by others who have studied emergency preparedness (Harney, 2007; Boon et al., 2012; van Kessel Gibbs, and MacDougall, 2015). (2005) thoughts about the future perspectives of bioecological theory generally work well in the context of hurricane preparedness and emergency preparedness with respect to his following three propositions The first proposition is that development is in fluenced by the actions of the child on the parents as the child passes through adolescence and into adulthood (Bronfenbrenner, 2005). As young individual s become increasingly aware of the importance of em ergency preparedness and properly preparing for natural disasters,

PAGE 44

44 how are those who are encouraging this development affected? Nidus and Sadder (2009) suggest ed that the most effective teachers learn from their students and improve their delivery of material based on observing student development. Therefore it follows that i ndividuals who deliver emergency preparedness programs would become mo st effective and improve their delivery of material by observing participants develop the desired outcomes of the training The second proposition is that o f a role reversal where children benefit in their elderly years the This child/parent relationship role reversal can be translated into emergency preparedness education through the participant/instructor relationship. Those program administrators who act as microsystems that encourage individual participants to learn more about preparedness are committ ed to these participants early in this process. Over time a percentage of participants will become intensely interested in learning as much as they can about preparedness and could feasibly return to assist or further the knowledge of those program adminis trators. Thus closing the role reversal loop proffered by Bronfenbrenner ( 2005). The third proposition of future interest mentioned by Bronfenbrenner (2005) is that of replication. I f an assessment of outcomes during an extended period of time detects chan ge, it warrants the replication of the process, program, or experiment to verify that the changes in outcomes are truly due to the intervention (Bronfenbrenner, 2005) De spite the importance of this verification step, academia rewards originality, thus rep lication is quite rare in the social sciences (Bryman, 2012).

PAGE 45

45 The translation of this replication into emergency preparedness would involve assessing participant levels of preparedness before and after any intervention (program, media, etc.) aimed at incre asing the level of preparedness. This also would involve assessing participants from more than one application of the intervention. Strengths of the Bioecological Theory of Human Development There are several strengths associated with the application of bi oecological theory to human development. Arguably what distinguishes many competing theories is the focus on context and proximal processes. This focus allows for the inclu sion of individuals gen etic characteristics as well as e nvironmental factors affecting the development of individual s ete story. Another strength of b ioecological theory is its very systematic approach to the examination of the e environment to be categorized and organized by direct and indirect contact with the individual as well as the amount of influence that each entity possesses. The bioecological theory also acknowledges that individual s are active participant s in their own development rather than uninvolved participants. This active involvement allows for the fact that individual s influence their environment as it simulta neously affects them. The final strength of the bioecological theory is that its evolution did not limit its scope to child development. The concepts of context and proximal processes are applicable to many disciplines concerned with the development of individuals, or even organizations and communities.

PAGE 46

46 Weaknesses of the Bioecological Theory of Human Develop ment The biggest weaknesses of bioecological theory are related to its strengths. Due to its broad applicability, it is difficult to determine exactly how the theory translates into facets of interaction between an individual and his or he r environment but this same inclusionary complexity requires a research design as nearly complex to address each aspect. T hus, t o account for all of the components of bioecological theory, it is necessary to address the PPCT model, multiple genetic attrib utes, three aspects of person al characteristics, four levels of environmental factors, three levels of time, while testing all of the entities comprising individuals environment. This complexity is what has led to many cross sectional studies of micro and macrosystems claiming bioecological theory as a theoretical framework (Tudge et al., 2009). Review of Current Literature N umer ous academic journal articles and stud ies have addressed various aspects of emergency or disaster preparedness in recent years. Many were in response to pandemics (e.g., i nfluenza, Ebola, and AIDS), man made or technological disasters (e.g. chemical spills, oil spills, and damaged nuclear fac ilities), and natural hazards (e.g. tsunamis, earthquakes, tornadoes and hurricanes/typhoons). Given that this current study addresses Southeastern Georgia, Florida and the Gulf Coast, its focus is on natural hazards with an emphasis on hurricanes. Savoi a, Lin, and Viswanath (2013) undertook a systematic review of articles in five peer reviewed journals published to the MEDLINE database in 2009. They searched for articles that included aspects of communications in public health emergency preparedness (Sav oia, Lin, and Viswanath, 2013). After three rounds of

PAGE 47

47 structured review, the final sample for this review included 131 articles and addressed emergencies ranging from infectious disease outbreaks (55%) to terrorism and bioterrorism (17%) to natural disaste rs (8%) (Savoia et al., 2013). Of the articles in this review, 52 were population based, empirical studies (Savoia et al., 2013). A majority of these studies were focused on attitudes, beliefs, socio demographic factors and their as sociation with emergenc y preparedness outcomes. Although the majority of emergencies addressed were health related, the three most frequently studied preparedness outcomes also are prevalent in the current natur al disaster literature. Preven tive Behaviors The preparedness outcom e that occurred with the highest frequency was preventive behaviors ; and it was present in 65 percent of the population based, empirical studie s (Savoia et al., 2013). Preven activity undertaken by individuals to prevent a disease or limit contagion to other contagious disease can be applied to disasters and the mitigation of damage incurred by these disasters. This new definition for pre ven tive behaviors would be: Any activity undertaken by individuals to prevent a disaster or mitigate the damage done by a disaster. Meyer, Baker, Broad, Czajkowski, and Orlove (2014) asked respondents that were in the paths of Hurricane s Isaac and Sandy ab out any protective actions they had taken The Hurricane Isaac study collected data from southeast Louisiana to the two and the Hurricane Sandy study collected data from Virginia to northeastern New Jersey (Meye r, Baker, Broad, Czajkowski, and Orlove, 2014). They found that even though the vast majority (94%) of their sample

PAGE 48

48 made preparatory actions and believed they were adequately prepared to endure the storms, further questioning revealed that only 55 percent of the sample that owned window protection had installed it at the six hour mark before the storms made landfall. In addition, only 25 percent of those sampled had plans to evacuate if ordered to do so (Meyer et al., 2014). This led researchers to determin e that residents were not as prepared as perceived. Risk Perception Risk perception was found in 54 percent of the population based, empirical studies and can be defined as a personal judgment about the severity of a risk and its effects on the society or in dividual (Savoia et al., 2013). hurricane risk perceptions were associated with the physical location of their homes in respect to wind vulnerability zones. These zones were establishe d based on ASCE 7 98 wind contours, which were established by the American Society of Civil Engineers Standards 50 100 year peak gusts (Peacock, Brody, and Highfield, 2005). Peacock et al. (2005) found that homeowners that lived in the highest wind contour s perceived the greatest hurricane risk This finding is positive because it indicated that Florida nsistent with those of experts. Peacock et al. (2005) also found inconsistencies between the implementation of state more stringent standards and instead opts to enforce the state code only on bu ildings within a mile of the coast (Peacock et al., 2005). This exception was heavily lobbied for by the homebuilders in the area in 2000 when it was approved by state

PAGE 49

49 lawmakers (Dunkleberger, 2005) In 2005, following a busy hurricane season, the exception was unsuccessfully c hallenged by insurance lobbyists and the Florida Building Commission. The premise for its existence is that the ASCE 7 98 wind contours are incorrect for homes in the panhandle area because they are protected by trees that ser ve as a windbreak. This is disconcerting for a few different reasons. First, the panhandle is within the same wind vulnerability zo the risk of wind damage is not lessened in the panhandle, which does not lend support to thi s provision. Second, homeowners in the panhandle may be unaware that their homes are not as resilient to high winds as homes built throughout the rest of the state. This would provide these homeowners a false sense of security regarding the durability of t heir home while decreasing their perceived hurricane risk, thus increasing the gap between their perceived risk and their a ctual hurricane risk severity. Incongruences also were found by Meyer, Baker, Broad, Czajkowski, and Orlove (2014) as they also explo red hurricane risk perceptions. They collected data via telephone interviews from individuals that were preparing for Hurricane Isaac and Hurricane Sandy in 2012. The perceptions of the respondents were found to be inconsistent with the actual severity of the wind, storm surge and flooding that would accompany the two storms (Meyer, Baker, Broad, Czajkowski, and Orlove, 2014). There was a tendency for an overestimation of the intensity of the winds and a significant underestimation of the impact of the stor m surge and flooding (Peacock et al., 2005; Meyer et al., 2014).

PAGE 50

50 Trumbo, Meyer, Marlatt, Peek and Morrissey (2014) studied changes in risk percept ion and optimistic bias in Gulf Coast residents during the quiescent two years subsequent to the destructive 2004 and 2005 Atlantic hurricane seasons. They found the change in risk perception was significant towards a more optimistic outlook. This more optimistic outlook which was clearly reflected in the resident responses indicating a significantly lower proba bility of a forced evacuation in the coming year (Trumbo, Meyer, Marl att, Peek and Morrissey, 2014). Trumbo et al. (2014) found that age and past hurricane experience were consistent predictors of risk perception; with older, more experienced respondents i ndicating lower perceived risk. Another finding of this study suggested that risk perception diminished over time and absence of immediate threat (Trumbo et al., 2014). This study also found the change in risk perception increased with lower incomes. Th us, participants with the lowest incomes experienced the largest decrease in ri sk perception from 2006 to 2008 while th ose with the highest incomes experienced a much smaller difference (Trumbo et al., 2014). Knowledge/Awareness Savoia et al. (2013) found the third most frequently occurring preparedness outcome to be knowledge/awareness This outcome was represented in 48 percent of the population based, empirical studies and can be defined as knowledge of specific threats and the behaviors that are adopted to prevent them (Savoia et al., 2013). Lindell, Tierney, and Perry (2001) found several predictors of preparedness that are directly related to knowledge and awareness. These predictors included: attentiveness to the media, higher levels of education, perso nal knowledge of the respective hazard, and high levels of social involvement (Lindell, Tierney, and Perry,

PAGE 51

51 2001). Social involvement is related to knowledge and awareness because those who interact in virtual and face to face social networks have the abil ity to access knowledge indirectly through other members of their network. Older residents with higher incomes and higher levels of education also were found to possess an increased perception of hazard knowledge (Ge, Peacock, a nd Lindell, 2011). Other Fac tors of Preparedness Education 2006 study (Baker, 2011). He found that college graduates reported the highest levels of preparedness and those without a high school diploma or GED scored t he lowest. Baker (2011) also found that higher incomes those between the ages of 40 and 70 years and locations nearest the coast all were related to higher levels of preparedness Residents of the northwest coastal region (Panhandle) had the highest levels of preparedness and residents in the inland counties reported the lowest levels of preparedness (Baker, 2011). Other researchers examined hurricane hazard mitigation incentives One such study split incentives into economic and non economic program s as the researchers explored the effectiveness of each program on the installation of hurricane shutters (Ge et al., 2011). These programs ranged from loans, in surance discounts, and property tax reductions to hurricane preparednes s inspections. Access to evacuation orders or other hurricane information has also been studied by previous research. Taylor Clark, Viswanath, and Blendon (2010) found that individuals who were unemployed as well as those with limited or nonexistent social networks were the least likely to receive evacuation orders. This same study found homeowners and those more than 55 years of age tended to underestimate the severity

PAGE 52

52 of hurricanes whereas females who rented their homes were the most likely to comply with evacuation orders they were aware of (Taylor Clark et al., 2010). trust in government and non profit entities is not a direct affects their sense of security and their p erception of the availability of assistance when needed. Reinhardt (2015) stated that findings in trust at all levels of government were lower following Hurricane Katrina and the 2010 BP oil spill. This is indicative of the ling of these events by emergency management. trust of federal, state, and local governments. Individuals who have positive experiences are less swayed by negative med ia exposure than those unaffected by the hazard (Reinhardt, 2015).

PAGE 53

53 Figure 2 1. Social Cognitive Theory (Bandura, 1986) Figure 2 2. Protection Motivation Theory (Rogers, 1975) Figure 2 3. Extended Parallel Process Model (Witte, 1992)

PAGE 54

54 Figure 2 4. Communication Infrastructure Theory (Kim and Ball Rokeach, 2006) Figure 2 5. Proximal Processes adapted from Bronfenbrenner, 2005

PAGE 55

55 Figure 2 6. Person Characteristics adapted from Bronfenbrenner, 2005 Figure 2 7. Ecological Systems Theory Mod el (Bronfenbrenner, 2005)

PAGE 56

56 Figure 2 8. Process Person Context Time Model adapted from Bronfenbrenner, 2005

PAGE 57

57 Figure 2 al., 2012)

PAGE 58

58 CHAPTER 3 METHODOLOGY Research Questions and Hypotheses This study is guided by two main research questions. The first involves specific location, education, income and age ) associat ion with hurricane preparedness knowledge and the second involv es the asso ciations between the same local non these research questions has four sub questions (repre senting the four demographic characteristics) and each of those has two hypotheses that focus on Florida specifically and then the entire study area. RQ1: To what extent can demographic characteristics ( location, education, income and age ) predict homeow ne H urricane P reparedness K nowledge S cale (HPKS) ? RQ1 A : To what extent can the location of homeowner s affect their self reported score on H urricane P reparedness Knowledge S cale (HPKS) ? H1 A1 : Homeowners located in the counties of Panhandle will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other regions within the state H1 A2 : Homeowners located in Georgia will self report lower scores on this study H urricane P reparedness Knowledge S cale (HPKS) than other S tates within the study area. RQ1 B : To what extent can homeowners E d ucation Level affect thei r self reported score H urricane P reparedness Knowledge S cale (HPKS) ?

PAGE 59

59 H1 B1 : Florida homeowners attaining less than a b achelor s degree will self report lower scores on H urricane P reparedness Knowledge S cale (HPKS) than other homeowners within the state H1 B2 : Homeowners from across the study area attaining less than a b degree will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners RQ1 C : To what exten t can household income level affect homeowners reported H urricane P reparedness Knowledge S cale (HPKS) ? H1 C1 : Florida homeowners with lower levels of household income will self report lower scores H urricane P reparedness Knowledge S cale (HPKS) than other homeowners with in the state H1 C2 : Homeowners from a cross the study area with lower levels of household income will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners RQ1 D : To what extent can homeowners age affect the ir self reported score on this H urricane P reparedness Knowledge S cale (HPKS) ? H1 D1 : Florida homeowners who are younger will self report lower scores on this H urricane P reparedness Knowledge S cale (HPKS) than other homeowners within the sta te. H1 D2 : Homeowners from across the study area who are younger will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners RQ2: To what extent can demographic characteristics ( location, education, i ncome and age ) affect self reported T rust in Support E ntities S cale (TSES) ? RQ2 A : To what extent can the location of homeowners affect their self reported score on T rust in Support E ntities S cale (TSES) ?

PAGE 60

60 H2 A1 : Homeowners located in the counties of P anhandle will self report l owe r T rust in Support E ntities S cale (TSES) than other regions within the state H2 A2 : Homeowners located in Louisiana will self report lower scores on this T rust in Support E ntities S cale (TSES) than other states within the study area. RQ2 B : To what extent can homeowners level of educational attainment affect their self T rust in Support E ntities S cale ( TSES) ? H2 B1 : Florida homeowners holding will self report lower T rust in Support E ntities S cale (TSES) than other homeowners within the state. H2 B2 : Homeowners from across the study area holding less th degree will self report lower T rust in Support E ntities S cale (TSES) RQ2 C : To what extent can Household I ncome affect self reported score T rust in Support E ntities S cale (TSES) ? H2 C1 : Florida homeowners with lower levels of household income will self report lower T rust in Support E ntities S cale (TSES) than other homeowners within the state. H2 C2 : Homeowners from across the study area with lower levels of househol d income will self report lower T rust in Support E ntities S cale (TSES) than other homeowners RQ2 D : To what extent can homeowner s A ge affect their self reported score on this T rust in Support E ntities S cale (TSES) ?

PAGE 61

61 H2 D1 : Florida homeowners who are younger will self report lower scores on this T rust in Support E ntities S cale (TSES) than other homeowners within the state H2 D2 : Homeowners from across the study area who are younger will self report lower T rust in Support E ntities S cale (TSES) than other homeowners Research Design As previously mentioned the purpose of this study is to examine the associations that specific demographic characteristics ( location, education, income and age ) have with hurricane preparedness in Southeastern Georgia, Florida, and the Gulf Coast. Given this intention, along with controlling costs of data collection, a cross sectional research design seem ed warranted (Gorard, 2013). A cross sectional res earch design in order to collect a body of quantitative or quantifiable data in connection with two or more variables, which are then examined to detect patterns of a 2012, p. 58) izes four key elements of cross sectional research design that some definitions lack. The first two elements, definitions. Collecting data from multiple cases allows for exploration in variation between the cases whether grouped by demographic variables, geographical location, or any number of ways the cases can be categorized. Collecting data at a single point in time does not necessarily mean all the data is collected simultaneously, although it could be, but rather it means that data is collected only once from each case. The data collection of this research design usually occurs during a relatively short period of time,

PAGE 62

62 especially when compared to the typical time required for data collection period s in other types of research designs which may take months or even multiple years (Bryman, 2012). often implied in other definitions, whereas it is explicit in the definition chosen by Bryman (2012). This element is extremely important because it specif i es that the data is implied but lacking in other definitions. This is important because it not only describes what can be gleaned from cross s ectional data but what cannot be determined a s well. Due to the lack of time ordering of variables, determinations cannot be made regarding the directionality of causal influence. T hus associations between multiple variables may be found, yet with a true cross sectional design, cause and effect relationships cannot be determined (Bryman, 2012). Cross many of the studies utilizing this design are in fact surveys, there are additional forms of data collection that fit into cross sectional research design. This research design can also include structured observations, diaries, content analysis, and official statistics (Bryman, 2012). Population/Sampling us on disaster preparedness, specifically hurricane preparedness, a population that was in an area prone to hurricanes was desirable. In the US that area contains the Atlantic coast Florida, and the Gulf Coast (Landsea, 2015). In an effort to make the pop ulation of interest more homogenous and to focus on

PAGE 63

63 those with the most personal value to lose by not preparing for hurricanes, the decision to focus on homeowners was made. The population of interest for the data set was homeowners, aged 25 to 75, living in 106 counties comprising Southeast Georgia Florida and the coastal counties of Texas, Louisiana, Mississippi, and Alabama Due to limitations in participant availability, 64 counties were added to the study area in order to collect the desired sample q uantities. Survey Sampling International (SSI) was contracted to collect the sample from its existing membership panels located in the counties of interest. The 2,769 collected responses were cleaned to 1,943 usable responses based on the explanations in Table 3 1 (Descriptive statistics of the Clean Responses are available in Appendix A) Instrumentation This study used a dataset from a multi item instrument that was developed by Dr. Randall A. Cantrell and C. Bradford Sewell of the University of Florida during the s pring semester of 2014. It was the third and final instrument used as part of the overall Decision (Cantrell and Sewell, 2015) research project data collection process. In the interest of furthering the Decision gment homeowners aged 25 to 75 years this online delivered instrumen t explored the themes of energy co nservation, community resilience disaster preparedness, and attitudes regarding relation ships The multi item measure s consisted of six Decision ems, 24 energy c onservation items, 50 resilience and preparedness items, 20 items related to attitudes regarding relationships and 10 socio demographic items for a total of 110 mostly Likert style response items. The instrument was pretested during the summer of

PAGE 64

64 2014 which was when t he finalized data collection occurred as well. The trimmed mean completion time was 18 minutes. Data Analysis Principal Components Analysis This study used Principal Components Analysis (PCA) as a dimension reduction technique to produce scales that were appropriate continuous variables to serve as dependent variables (DV) for analyses of differences between groups The PCA was performed on a section of 53 items from the survey instrument that focused on disaster preparedness and community resiliency. These items were subjected to PCA performed version 22. Prior to performing the PCA, the suitability of data for factor analysis was assessed. This included an i nspection of the correlation matrix with attention given to any coefficients of .3 and greater. The Kaiser Meyer Olkin measure of sampling adequacy was used to determine if there were linear relationships between the items and thus whether it would be appropriate to continue with the PCA. Once the KMO was performed the KMO value was confirm ed to exceed the recommended value of .6 (Kaiser, 1970, 1974) T Sphericity (Bartlett, 1954) was verified to be was used to confirm that the correlation matrix was indeed an identity matrix. This is essential to the PCA because if it were not an identity matrix this would have meant that the re were no correlations between any of the items Once the PCA was performed, in accordance to the eigenvalue one criterion components with eigenvalues greater than 1.0 were of interest (Kaiser, 1960) This 1.0 threshold is of value because any eigenvalue s less than 1.0 would indicate that a component explained less variance than an item would alone. The next steps focused

PAGE 65

65 on increasing the total variance explained through simplifying the PCA and essentially or inte rference to the components. By increasing the total variance explained while decreasing the number of components, the result is a more efficient set of components. Once the revised PCA was settled on, a split sample validation was performed to deter mine generalizability of the PC A components beyond the sample. Exploring Differences Between Groups The two research questions relied on separate DVs. The first DV had eight separate difference between groups analyses performed; one ana lysis for Florida and one for the study area for each of the fo ur demographic characteristics ( location, education, income and age ) This process was repeated for the second DV for a total of 16 differences between groups analyses. Th e same protocol was followed for each analysis. After the initial descriptive statistics were calculated for the IV of interest ( location, education, income and age ) the basic requirements for a one way ANOVA were explored. There were six assumptions tha t were considered in determin ing if a one way ANOVA was appro priate or if another method would have been more appropriate. The first three of these assumptions relate to the study design and the type of data that was collected. These three assumptions wer e; a) the DV is continuous data b) the IV is categorical with two or more independent groups and c) the observations are independent. The first assumption was met for all analyses because the two DVs used were continuous data that us ed scales that were created through the PCA process The second assumption held true for all the analyses because the IVs ( location, education, income and age ) had between five and eight categories that were independent of the other categories.

PAGE 66

66 The last of these assumptions was met by simply using a design that used a survey instrument to collect data from different homeowners. The remaining three assumptions required the output of statistical tests from evaluate. T hese assumptions are as follows: a) there should be no significant outliers in the IVs, b) the DV should be approximately normally distributed within the IV categories, and c) homogeneity of variance should exist in the DV between the IV categories. There has been some di scussion in statistics about what constitutes an outlier, and for this study an outlier will be based on the outlier labeling rule of Hoaglin and Iglewicz (1987). Tukey (1977) originally thought that an outlier could be defined as any point more than 1.5 t imes the value of the interquartile range (IQR ) below or above the IQR. This multiplier is known as g in the literature reviewed for this study. Tukey refuted this in later work with Hoaglin and Iglewicz; suggesting that a g value of 2 .2 was a better multi plier than the original value of 1.5 (Hoaglin, Iglewicz, and Tukey, 1987). Box plots generated by SPSS were used to identify outliers in the IVs, yet these boxplots used g = 1.5 as the multiplier thus providing a more conservative windo w of accepted values Any outliers outside of the range created by the g = 2.2 multiplier will be reported in the text T he normality of the DV was confirmed through normal Q Q plots, histograms, and stem and leaf plots. Finally the homogeneity of the variances were evaluated If the assumptions were verified, the ANOVA was then performed. If the ANOVA consulted to indicate between wh ich le vels of the IVs these differences occurred.

PAGE 67

67 If the assumptions were violated, then alternative non parametric analysis methods such as the Welch test, were used to indicate if significant differences were present. If these non parametric tests indicated s ignificance then the Games Howell post hoc was used to identify w here the differences occurred. Multiple Regression Modeling Multiple linear regression analysis was then conducted to analyze the socio demographic variables predicti ng scores on the H ur rican e P reparedness K nowledge Scale (HPKS) and scores on the T rust of Support Entities Scale (TSES) As with the ANOVA process, there were assumptions to be addressed Multiple regression modeling requires six assumptions to be assessed to be valid. The first of these assumptions is that of an independence of errors/residuals. made it very unlikely that homeowner responses would be related, the Durbin Watson test was still used to ve rify this assumption. A Durbin Watson statistic value of approximately 2 is t he accepted value of this test. The second assumption is that the IVs ( location, education, income and age ) are linear ly related to the DV ( HPKS or TSES ) This was assessed by ob serving a scatter plot of the studentized residual by the unstandardized predicted value, then looking for a linear pattern of distribution of points. The same scatter plot was used to check the third assumption of homo scedasticity. If the data points are evenly distributed across the predicted values this represents homoscedasticity. The fourth assumption is that no multicollinearity exists in the data. This was verified through examining the correlation table of the DV and IVs for any correlations above 0 .7. After this, the collinearity statistics were consulted to verify that none of the tolerance numbers for the IVs are less than 0.1. The fifth assumption is that of no

PAGE 68

68 existence of significant outliers or points of influence. The casewise diagnostics tab le produced by SPSS provided standard residual values; this output was analyzed for any values greater than 3 standard deviations. The last assumption to be verified was that the residuals (errors) were normally distributed. The first step was to assess the leverage values created by SPSS for each response. Once these values were sorted, any values above 0.2 were considered risky, with values over 0.5 being considered dangerous After verifying the residuals, the next step of verifying residual normality of greater than 1.0 warranted further investigation. The final step of this assumption was to verify that the distribution thro ugh a histogram with a normal curve and a P P plot Once all of these assumptions were verified the actual multiple regression analysis reporting occurred. This process included model fit (R 2 ), the model significance ( p value), the individual IV significa nce, and the summary of multiple regression analysis tables Limitations As with all studies, t his one has its limitations O ne of the most difficult barriers to overcome was the physical distribution and availability of participants. Due to the method in which SSI recruits participants for its membership panels, the panels follow trends in population A higher population density yields a larger pool of possible participants in that area. This works well if a study is focused only in populous areas. However for this study, it was difficult to get a large enough sample to legitimately perform certain types of statistical tests. In fu ture analysis, alternative data collection methods should be explored to collect data in sparsely populated areas. Another

PAGE 69

69 poss ible limitation of this sample is gender bias; it has 60 percent female participation. This can be dealt with by weighting the responses during the preparation of the data. This sample was not representative of t he population in the study area, so this mig ht marginally impact the generalizability A limitation of the instrument used is that it co mprised a wide variety of topics thus r isk ing possible consternation for participants. The instrument also was comprehensive, consisting of 110 items and this may have caused some degree of response fatigue in participants to wards the end of the survey. Another limitation of the instrument was the lack of ethnicity or race items that c ould be useful in comparing findings with previous studies. could also be seen as a limitation. If there were items that verified that the homeowners were reading the questions and were not merely responding randomly this would possibly help answer some questions about the reasonin g for the existence of some outlying responses. The cross sectional research design was used due to advantages in cost and appropriateness of analysis, but this design limits the testing of the time aspect of the PPCT model. Without this aspect it is impo ssible to measure development and therefore progress towards increased preparedness

PAGE 70

70 Table 3 1. Data cleaning process Responses Explanations 2,769 Total r aw r esponses 29 Screened out by a ge (under 25 or over 75) 550 Screened out by homeowner status (non homeowners) 11 Dropped due to duplicates (single participant submitted two responses) 236 Dropped for incompletes 1,943 Clean r esponses

PAGE 71

71 CHAPTER 4 RESULTS AND ANALYSIS OF DATA Overview This chapter will serve the purpose of presenting the results of the various analyses laid forth in Chapter 3 M ethodology was twofold, 1) to explore associations between specific demographic characteristics of homeowners (location, education, income, and age) and their level o f hurricane preparedness knowledge; and 2) to explore associations between these specific homeowner demographic characteristics and their level of trust in specific support entities (other community members, local emergency management, non profit organizat ions, and Federal agencies such as FEMA). Principal Components Analysis Of the 53 items from the d isaster p reparedness and c ommunity resilience section of the 1,943 homeowner data set, 45 items were suitable to be used for dimension reduction These items were subjected to principal components analysis (PCA) version 22 Prior to performing the PCA, the suitability of data for factor analysis was assessed. Inspection of the correlatio n matrix revealed that all items had at least one correlation coefficient of .3 or greater The Kaiser Meyer Olkin value was .918, which clearly exceed ed the minimum re commended value of .6 (Kaiser, 1970, 1974) In fact, a KMO greater than 0.9 measure values. T p <.001 significance level, supporting the factorability of the correlation matrix.

PAGE 72

72 This initial P CA revealed the presence of eight factors with eigenvalues exceeding 1.0, explaining 56.08% of the variance collectively (Table 4 1 ) An inspection of the rotated component matrix revealed several variables with factor loadings less than .500 these variables were removed and the revised PCA was performed again T he 29 remaining variables from the c ommunity resilience and d isaster p reparedness section of the data set were again subjected to PCA Prior to performing the PCA, the suitability of data for factor a nalysis was again assessed. Inspection of the correlation matrix revealed m any coefficients of .3 and greater The Kaiser Meyer Olkin value was 878 still exceeding the recommended value of .6 (Kaiser, 1970, 1974) and rtlett, 1954) remained significant at the p <.001 significance level, supporting the factorability of the correlation matrix. This revised PCA revealed the presence of six components explaining between 4.9 1 % to 21.98% of the variance of this component individually and 61.62% of the variance collectively (Table 4 2 ). This parsimonious outcome is much preferred over the initial PCA in which more components explained less variance. The variables that make up each of the components factor loadings, inter item correlation matri x, and reliability measures are fully described in Appendix B In the interest of testing the generalizability of the findings of this study, a split sample validation was performed. The split sample validation method was used as a co st effective alternative to collecting another sample from the population of interest. The sample was randomly split into two halves and the PCA was performed on the halves to confirm the communalities and that the variables split into the same

PAGE 73

73 components with roughly the same fac tor loadings. Table 4 3 compared the two halves with the revised PCA. This research project had an interest in furthering the understanding of refore, the first compone nt that will be referred to as the Hurricane Preparedness K nowledge Scale (HPKS) component was of interest. Another interest of this research was in presenting findings to policy makers that would assist in determining if demographic factors affected homeowners trust in support entities during a time of crisis. This resulted in focus on the third component which will be referred to as the Support En tities Scale (TSES), to a lso be included in this study The factor loadings of the two compo nents used are listed in Table 4 4 and descriptive statistics of the two components of interest are in Table 4 5 ( see full details of the PCA process in Appendix B) Exploring Differences Between Groups Research Question 1 : Hurricane Preparedness Knowledge o what extent demographic characteristics ( location, education, income and age ) could predict homeowne Hurricane P reparedness Knowledge S cale (HPKS) The e ffect of each of these demo graphic characteristics will first be explored within the state of Florida, then once again across the study area. Location of home H1 A1 : Homeowners located in the counties of Panhandle will self report lower scor H urricane Preparedness K nowledge S cale (HPKS) than other regions within the state

PAGE 74

74 The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) by R egions in Florida are shown in Table 4 6. For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the location represented by the regions of Florida Homeowner responses were classified into six groups: North Florida ( n = 163), West Coast ( n = 412), Panhandle ( n = 46), East Coast ( n = 350), South Florida ( n = 305), and Northwest ( n = 56). The HPKS scores increased from North Florida ( M = 38.96, SD = 8.71 ) to the West Coast of Florida ( M = 39.22, SD = 8. 94), to the Panhandle of Florida ( M = 40.48, SD = 7.50), to the East Coast of Florida ( M = 41.47 SD = 8.80 ) to South Florida ( M = 42.71, SD = 8.47), to Northwest Florida ( M = 44.25 SD = 8.07 ) in that order (Table 4 6) For the one way ANOVA some assumptions must first be verified. As men tioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analys e s. The last three assumptions are cal Package for Social Science (SPSS) Statistics, version 22 to perform these tests The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida only one outlier w a s identified by the boxplot created by SPSS and the outlier labeling rule ( Figure 4 1 ) The next assumption is that e) the DV should be approximately normally distr ibuted in each of the IV groups, however t he one way ANOVA is considered robust to violation assumed. This was confirmed through n ormal Q Q p lots, stem & leaf plots, and histograms of each region. Finally, the last assumption was that of f) homogenous variances between

PAGE 75

75 groups. I for equality of variances ( p = 719 ) The HPKS score was significantly different for the Regions of Florida F (5,1326) = 9.34, p < .001 Tukey post hoc analysis (Table 4 7) revealed that the mean HPKS scores in Northwest Florida ( M = 44.25, SD = 8.06) was significantly higher than North Florida ( M = 38.96, SD = 8.71) and the West Coast of Florida ( M = 39.22, SD = 8.94). The Tukey post hoc also revealed that North Florida ( M = 38.96, SD = 8.71) and the West Coast of Florida ( M = 39.22, SD = 8.94 ) HPK S scores were significantly lower than the East Coast ( M = 41.47 SD = 8.80 ) and South Florida ( M = 42.71 SD = 8.47 ) scores. The HPKS scores from the Panhandle of Florida were not significantly different than any other regions of Florida H1 A2 : Homeowners located in the state of Georgia will self report lower scores on this H urricane P reparedness K nowledge S cale (HPKS) than other states within t he study area. The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) by States are shown in Table 4 8. For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the location represented by the States Homeowner responses were classified into five groups: Southeastern Georgia ( n = 106), Florida ( n = 1,332), and the gulf coast counties of Texas ( n = 329), the combined area of Alabama & Mississippi ( n = 55) and Louisiana ( n = 121). The HPKS scores increased from Georgia ( M = 37.32 SD = 9.44 ), to Florida ( M = 40.83 SD = 8.82 ), to Texas ( M = 41.13 SD = 8.74 ), to the combined area of Alabama & Mississippi ( M = 42.07 SD = 8.87 ), to Louisiana ( M = 42.69 SD = 8.14) in that order (Table 4 8 ).

PAGE 76

76 A one way ANOVA was conducted to determine if the HPKS score was different for homeowners in different S tates As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument desi gn held true for all of the analyses. As the remaining three assumptions were address ed, t here was only one outlier observed out of the sample of 1,943, as assessed by percentile output and the outlier labeling rule The DV data was normally distributed within each group and was explored by n ormal Q Q p lots, h istograms, and stem & leaf plots. T here was homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 785 ). The mean HPKS score s were significantly different between the States, F (4, 1938 ) = 5.93 p < .001. Tukey post hoc analysis (Table 4 9 ) revealed that the mean HPKS score in Georgia ( M = 37.32 SD = 9.44 ) was significantly lower than all other states, Florida ( M = 40.83 SD = 8.82 ) Texas ( M = 41.13, SD = 41.13), Alabama & Mississippi ( M = 42.07 SD = 8.87 ) and Louisiana ( M = 42.69, SD = 8.14 ) The HPKS scores from the other S tates were not significantly different. Highest education level of homeowner H1 B1 : Florida homeowners attaining degree will self report lower scores on Hurricane Preparedness Knowledge S cale (HPKS) than other homeowners within the state The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) by educational levels i n Florida are shown in Table 4 10 For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the educational level attained by the homeowner Homeowner responses were classified into five groups: HS Diploma or less ( n = 211), Some college up to AA/AS Degree ( n = 512), ( n = 385),

PAGE 77

77 Degree ( n = 190), and Doctoral Degrees ( n = 34). The H PKS scores increased from Floridian homeowners with HS Diploma or less ( M = 39.68 SD = 9.00 ), to Some college up to AA/AS D egree ( M = 40.83 SD = 9.11 ), to D egree ( M = 40. 95 SD = 8.33 ), to Doctoral D egree ( M = 41.35 SD = 9.41), to egree ( M = 41.37 SD = 8.50 ) in that order ( Figure 4 2 ). For the one way ANOVA, some assumptions must first be verified. As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confir med through statistical tests. The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only one outlier w as found. This outlier can be observed in the boxpl ots of HPKS scores by Education L evels in Figure 4 3 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed th rough the boxplots of Figure 4 3 supplemented by n ormal Q Q p lots, stem & leaf plots, and histograms of each of the E ducational L evel s Finally, the last assumption was that of f) homogenous variances between groups. There was homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 454 ). The HPKS scores were found to not vary significantly between the different E d ucational L evels in Florida, F (4, 1327) = 1.30, p = .270. The full ANOVA table is available as Table 4 11. H1 B2 : Homeowners fr om across the study area attaining lower levels of education will self report lower H ur ricane P reparedness K nowledge S cale (HPKS) than other homeowners

PAGE 78

78 The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) scores by Education Level s are shown in Table 4 12. For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the Education L evel The HPKS scores increased from homeowners holding a H S D iploma or less ( n = 308, M = 39.75 SD = 9. 22 ), to those holding Doctoral D egrees ( n = 56, M = 40. 50 SD = 9.12 ), to those attending S ome college up to AA/AS D egrees ( n = 730, M = 40.70 SD = 9.13 ), to those holding egrees ( n = 567, M = 41.35 SD = 8.52 ), to those holding s D egrees ( n = 282, M = 41.46 SD = 8.19) in that order ( Figure 4 4 ). A one way ANOVA was conducted to determine if the HPKS score was different for homeowners with different E ducational L evels As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confirmed through statistical tests. In the first remaining assumption of no outliers, t here was only one outlier out of the sampl e of 1,943, as assessed by percentile output and the outlier labeling rule ; data was normally distributed for each group, as assessed by normal Q Q plots, h istograms, and stem & leaf plots. There was homogeneity of variances, as assessed by Levene's test f or equality of variances ( p = 189 ). The HPKS score s were found to not vary significantly between the different E d ucational L evels F (4, 1938 ) = 2.07 p = .082 The full ANOVA table is available as Table 4 13. Household income H1 C1 : Florida homeowners with lower levels of household income will self report lower scores H urricane Preparedness Knowledge S cale (HPKS) than other homeowners with in the state

PAGE 79

79 The descriptive statistics of the Hurricane Preparedness Knowledge S cale (HPKS) by F loridian H ousehold I ncome s are shown in Table 4 14 For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the Household I ncome s Floridian h omeowner responses were classified into eight income categories : Up to $14,999 ( n = 59 ), $15,000 $24,999 ( n = 95 ), $25,000 $34,999 ( n = 157 ), $35,000 $49,999 ( n = 192 ) $50,000 $74,999 ( n = 353 ), $75,000 $99,999 ( n = 243 ), $100,000 $149,999 ( n = 160 ), and $150,000 or more ( n = 73 ). The HPKS scor es increased from Up to $14,999 ( M = 36.37 SD = 9.13 ), to $15,000 $24,999 ( M = 39.53, SD = 8.84), to $25,000 $34,999 ( M = 39.57 SD = 8.79 ), to $35,000 $49,999 ( M = 40.59 SD = 9.15 ), to $50,000 $74,999 ( M = 40.60 SD = 8.60 ), to $100,000 $149,999 ( M = 41.58 SD = 8.60 ), to $75,000 $99,999 ( M = 42.64 SD = 8.62 ), to $150,000 or more ( M = 43.01 SD = 8.07 ) in that order ( Figure 4 5 ). For the one way ANOVA, some assumptions must first be verified. As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confirmed through statistical tests. All of these assumptions held tru e for this analysis. The last three assumptions are confirmed through statistical tests. The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only two ou tliers were found. These outliers can be seen in the boxplots of the HPK S scores by Household I ncomes in Figure 4 6 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the boxplots of Figure

PAGE 80

80 4 6 supported by n ormal Q Q p lots, stem & leaf plots, and histograms of each H ousehold I ncome category Finally, the last assumption was that of f) homogenous variances between groups. There was also homogeneity of variances, as assesse d by Levene's test for equality of variances ( p = 939 ). The HPKS score was significantly different by Household In comes F (7, 1324) = 5.35, p < .001. Tukey post hoc analysis (Table 4 15) revealed that the mean HPKS score for Floridian homeowners with Household I ncomes of Up to $14,999 ( M = 36.37 SD = 9.13 ) was significantly lower than Floridian homeowners with H ousehold I ncomes of $35,000 $49,999 ( M = 40.59, SD = 9.15 ), $50,000 $74,999 ( M = 40.60 SD = 8.60 ), $75,000 $99,999 ( M = 42.64 SD = 8.62 ), $1 00,000 $149,999 ( M = 41.58 SD = 8.60 ), and $150,000 or more ( M = 43.01 SD = 8.07 ). The mean HPKS score of Floridian homeowners with Household I ncomes of $75,000 $99,999 ( M = 42.64 SD = 8.62 ) was significantly higher than homeowners in the $25,000 $34,999 ( M = 39.57 SD = 8.79 ) Household I ncome category (Table 4 15). All other differences in mean HPKS scores by Floridian Household I ncome s were not found to be significant. H1 C2 : Homeowners from across the study area with lower levels of H ouse hold I ncome will self report lower Hurricane Preparedness Knowledge S cale (HPKS) than other homeowners. The descriptive statistics of the mean Hurricane Preparedness Knowledge Scale (HPKS) scores by Household Income ar e shown in Table 4 16 For the test of this hypothesis, the dependent variable (DV) is the mean HPKS score s and the independent variable (IV) is the Household I ncome The HPKS scores increased from Up to $14,999 ( n = 82, M = 37.16 SD = 9.71 ), to $25,000 $34,999 ( n = 219, M = 39.94 SD = 9.09 ), to $15,000 $24,999 ( n = 137, M = 39.99 SD = 8.60 ), to $35,000 $49,999 ( n = 264, M =

PAGE 81

81 40.20 SD = 8.99 ), to $50,000 $74,999 ( n = 496, M = 40.66 SD = 8.72 ) to $100,000 $149,999 ( n = 260, M = 41.22, SD = 8.84), to $75,000 $99,999 ( n = 355, M = 42.43, SD = 8.58), to $150,000 or more ( n = 130, M = 42.53, SD = 8.03 ) in that order ( Figure 4 7 ). A one way ANOVA was conducted to determine if the HPKS score was different for homeowners by Household I ncome As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The remaining three assumptions were confirmed through statistical tests. There was only one outlier o ut of the sample of 1,943, as assessed by the outlier labeling rule and percentile output ; data was normally distributed for each group, as assessed by n ormal Q Q p lots, h istograms, and stem & leaf plots. There was also homogeneity of variances, as assesse d by Levene's test for equality of variances ( p = 810 ). The mean HPKS score s were significantly different by Household I ncomes F ( 7 193 5 ) = 5.23 p < .001. Tukey post hoc analysis (Table 4 1 7 ) revealed that the mean HPKS score for homeowners with Household I ncomes of Up to $14,999 ( M = 37.16 SD = 9.71 ) was significantly lower than Household I ncomes of $50,000 $74,999 ( M = 40.66 SD = 8.72 ), $75,000 $99,999 ( M = 42.43 SD = 8.58 ), $100,000 $149,999 ( M = 41.22 SD = 8.84 ), and $150,000 or more ( M = 42. 53 SD = 8.03 ). The mean HPKS score from homeowners with Household I ncomes of $75,000 $99,999 ( M = 42.43, SD = 8.58) was significantly higher than homeowners in the $25,000 $34,999 ( M = 39.94 SD = 9.09 ) and $35,000 $49,999 ( M = 40.20, SD = 8.99) Household I ncome categories (Table 4 1 7 ). All other differences in mean HPKS scores by H ousehold I ncome s were not found to be signifi cant.

PAGE 82

82 Age of homeowner H1 D1 : Florida homeowners who are younger will self report lower H urricane Preparedness Knowledge S cale (HPKS) than other homeowners within the state. The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) A ge is shown in Table 4 18. For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the Floridian h A ge Homeowner responses were classified into five Age categories : 25 34 years old ( n = 162), 35 44 years old ( n = 199), 45 54 years old ( n = 261), to 55 64 years old ( n = 372), to 65 75 years old ( n = 338). The HPKS scores increased from Floridian homeowners 65 75 years old ( n = 338 M = 40.22 SD = 8.83 ), to 45 54 years old ( n = 261 M = 40. 40 SD = 8.97 ), to 35 44 years old ( n = 199 M = 40.64 SD = 8.99 ), to 55 64 years old ( n = 372 M = 41.17 SD = 8.41 ), to 25 34 years old ( n = 162 M = 42.28 SD = 9.21 ) in that order ( Figure 4 8 ). For the one way ANOVA, some assumptions must first be verified. As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confirmed through statistical tests. The last three assumptions are confirmed through statistical tests. The first of these remain ing assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only one outlier was found. This outlier can be seen in the b oxplots of HPK S scores by h A ge categories in Figure 4 9 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the n ormal Q Q p lots, stem & leaf plots, and histograms of each Age category Finally, the last

PAGE 83

83 assumption was that of f) homogenous variances between groups. There was also homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 480 ). The HPKS scores were found to not vary significantly between the different A g e catego ries of Floridian Homeowners, F (4, 1327 ) = 1.82, p = .123. The full ANOVA table is available as Table 4 19. H1 D2 : Homeowners from across the study area who are younger will self report lower Hurricane P reparedness K nowledge S cale (HPKS) than other homeowners The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) A ge is shown in Table 4 20 For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent va riable (IV) is the h A ge The HPKS scores increased from homeowners 65 75 years old ( n = 452, M = 40.22, SD = 8.83), to 45 54 years old ( n = 403, M = 40.40, SD = 8.97), to 35 44 years old ( n = 305 M = 40.64, SD = 8.99), to 55 64 years old ( n = 534, M = 41.17, SD = 8.41), to 25 34 years old ( n = 249, M = 42.28, SD = 9.21) in that order ( Figure 4 10 ). For the one way ANOVA, some assumptions must first be verified. As mentioned in the Data Analysis section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confirmed through statistical tests. The last three assumptions are confirmed through statistical tests. The first of these remaining assumptions was d) no significant outliers in t he IV groups in terms of the DV The re were no outliers in this analysis as confirmed by the outlier labeling rule.

PAGE 84

84 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the normal Q Q p lots, stem & leaf plots, and histograms of each Age category Finally, the last assumption was that of f) homogenous variances between groups. There was also homogeneity of variances, as assessed by Le vene's test for equality of variances ( p = 261 ). The HPKS scores were found to vary significantly between the different A ge categories of h omeowners, F (4, 1938 ) = 2.52 p = 039 Despite the ANOVA finding s ignificant differences in means of HPKS scores between A ge post hoc analysis was unable to distinguish which categories were significantly different. The full ANOVA table is available as Table 4 21 Research Question 2 : Trust in Support Entities rch question sought to quantify t o what extent the same demographic characteristics ( location, education, income and age ) could predict homeowne Trust in Support Entities Scale (TSES) The affect of each of these demographic characteristics will first be explored within the state of Florida, then once again across the study area. Location of home H2 A1 : Homeowners located in the counties of P anhandle will self report lower T rust in S upport E ntities S cale (TSES) than other Re gions within the state. The descriptive statistics of the Trust in Support Entities Scale (TSES) by Florida Region are shown in Table 4 22. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independent variable (IV) is the location represented by the Florida Region Homeowner responses were classified into six groups: North Florida

PAGE 85

85 ( n = 163), West Coast ( n = 412), Panhandle ( n = 46), East Coast ( n = 350), South Florida ( n = 305) and Northwest ( n = 56). The TSES scores increased from North Florida ( n =163, M = 31.44, SD = 8.08), to the West Coast of Florida ( n =412, M = 31.90 SD = 7.87 ), to the Panhandle of Florida ( n =46, M = 32.09 SD = 7.88 ), to the East Coast of Florida ( n = 350, M = 32.99 SD = 7.65 ), to South Florida ( n =305, M = 33.82 SD = 7.83 ), to Northwest Florida ( n =56, M = 34.34 SD = 8.58 ) in that order (Table 4 22 ). For the one way ANOVA, some assumptions must first be verified. As mentioned in the Data Analysi s section of the Methodology chapter, the three assumptions based on instrument design held true for all of the analyses. The last three assumptions were confirmed through statistical tests. The last three assumptions are confirmed through statistical test s The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only six total outliers were found. These outliers can be seen in the boxplots of TSES scores by Florida R egion in Figure 4 1 1 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups, however the one way ANOVA is considered robust This was confirmed through the normal Q Q p lots, stem & leaf plots, and histograms of each R egion Finally, the last assumption was that of f) homogenous variances between for eq uality of variances ( p = .766 ).

PAGE 86

86 The mean TSES score s were significantly different by the Regions of Florida, F (5,1326) = 3.56, p = .003. Tukey post hoc analysis (Table 4 23 ) revealed that the mean TSES score of North Florida ( M = 31.44 SD = 8.08 ) and of the West Coast of Florida ( M = 31.90 SD = 7.87 ) were significantly lower than the mean score of homeowners in South Florida ( M = 33.82 SD = 7.83 ). The TSE S scores from the Northwest, East Coast and Panhandle of Florida were not signific antly di fferent than any other R egions of Florida. H2 A2 : Homeowners located in Louisiana will self report lower T rust in S upport Entities S cale than other S tates within the study area. The descriptive statistics of the Trust in Support Entities Scale (TSES) by State are shown in Table 4 24. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independent variable (IV) is the location represented by the States As previously mentioned, there was a low sam ple collected in the smaller areas of Alabama and Mississippi, and for this reason, these two states were combined to form a single area. Homeowner responses were classified into five groups: Southeastern Georgia ( n = 106), Florida ( n = 1,332), and the gul f coast counties of Texas ( n = 329), the combined area of Alabama & Mississippi ( n = 55) and Louisiana ( n = 121). The TSES scores increased from Georgia ( M = 29.39 SD = 7.97 ), to the Texas ( M = 31.58 SD = 7.79 ), to Louisiana ( M = 31.93 SD = 7.84 ), to Florida ( M = 32.68 SD = 7.90 ), to the combined area of Alabama & Mississippi ( M = 34.00 SD = 8.84 ) in that order (Table 4 2 4 ). All of the assumptions based on instrument design held true for this analysis. The last remaining assumptions were a) no signi ficant outliers in the IV groups in terms of

PAGE 87

87 the DV. Of the 1, 943 home owner responses, only six outliers were found. These outliers can be seen in the boxplots of TSES scores by State in Figure 4 1 2 The next assumption is that b) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the n ormal Q Q p lots, stem & leaf plots, and histograms of each State Finally, the last assumption was that of c) homog enous variances between groups. In this case, there was homogeneity of p = .922). The mean TSES score s were significantly different by the States F (4,1938) = 5.74 p < .00 1 oc analysis (Table 4 2 5 ) revealed that the mean TSES score of Georgia ( M = 29.39 SD = 7.97 ) was significantly lower than the mean score of homeowners in Florida ( M = 32.68 SD = 7. 90 ) and the combined area of Alabama & Mississippi ( M = 34.00 SD = 8.84 ) The mean TSE S scores from Louisiana and Texas were not significantly different than any other S tates Highest education level of homeowner H2 B1 : Florida homeowners attaining will self report lower Trust in S upport Entities S cale (TSES) than other homeowners within the state. The descriptive statistics of the Trust of Support Entities Scale (TSES) by E ducation L evels in Florida are shown in Table 4 26. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independent variable (IV) is the Education L evel s attained by the homeowner. Homeowner responses were classified into five groups: H S Diploma or less ( n = 211), Some college up to AA/AS Degree ( n = 512), ( n = 385), ( n = 190), and Doctoral Degrees ( n = 34). The TSES scores increased from Floridian homeowners with a Doctoral D egree ( M = 31.29 SD = 8.67 ), to D egree ( M = 31.58 SD =

PAGE 88

88 7.86 ), to Some college up to AA/AS D egree ( M = 32.52 SD = 7.91 ), to D egree ( M = 33.25 SD = 7.75 ), to HS Diploma or less ( M = 33.25 SD = 7.96 ) in that order (Figure 4 13 ). For the one way ANOVA, some ass umptions must first be verified. The three assumptions based on instrument design held true for this analysis, as mentioned in the Data Analysis section of the Methodology chapter. The last three assumptions are confirmed through statistical tests. The fir st of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only six outliers were found. These outliers can be seen in the boxplots of HPKS scores by E ducation L evels in Figure 4 14 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the n ormal Q Q p lots, stem & leaf plots, and histograms of each of the Education L evel s Finally, the l ast assumption was that of f) homogenous variances between groups. There was also homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 997 ). The TSES scores were found to not vary significantly between the different Educat ion L evels in Florida, F (4, 1327) = 2.02, p = .090. The full ANOVA table is available as Table 4 27. H2 B2 : Homeowners from across the study area attaining will self report lower T rust in Support Entities S cale (TSES) The descriptive statistics of the Trust in Support Entities Scale (TSES) by Education Level are shown in Table 4 28. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independ ent variables (IV) are the Education

PAGE 89

89 L evel s The TSES scores increased from homeowners holding a Doctoral D egree ( n = 56, M = 31.23, SD = 7.71 ), to those holding egrees ( n = 282, M = 31.35 SD = 8.16 ), to those attending S ome college up to AA/AS D egrees ( n = 730, M = 32.17 SD = 7.96 ), to those holding D egrees ( n = 567, M = 32.75 SD = 7.65 ), to those holding a HS D iploma or less ( n = 308, M = 32.87 SD = 8.24 ) in that order (Figure 4 1 5 ). For the one way ANOVA, some assumptions must first be verified. The three assumptions based on instrument design held true for this analysis, as mentioned in the Data Analysis section of the Methodology chapter. The last three assumptions are confirmed through statistical tests. The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,943 homeowner responses, seven outliers were found. These outliers can be seen in the boxplots of HPKS scores by Education L evels in Figure 4 16 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the n ormal Q Q p lots, stem & leaf plots, and histograms of each Education L evel Finally, the last assumption was that o f f) homogenous variances between groups. There was also homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 912 ). The TSES scores were found to not vary significantly between the different Education L evels F (4, 1 938 ) = 2. 17 p = .0 7 0. The full ANOVA table is available as Table 4 2 9 Household income H2 C1 : Florida homeowners with lower levels of Household I ncome will self report lower Trust in Support Entities S cal e (TSES) than other homeowners within the state.

PAGE 90

90 The descriptive statistics of the Trust of Support Entities Scale (TSES) by Household I ncome s are shown in Table 4 30 For the test of this hypothesis, the dependent variable (DV) is the TSES s core and the independent variable (IV) is the H ousehold I ncome Floridian h omeowner responses were classified into eight income categories : Up to $14,999 ( n = 59 ), $15,000 $24,999 ( n = 95 ), $25,000 $34,999 ( n = 157 ), $35,000 $49,999 ( n = 192 ) $50,000 $74,999 ( n = 353), $75,000 $99,999 ( n = 243), $100,000 $149,999 ( n = 160), and $150,000 or more ( n = 73). The TSES scores increased from Up to $14,999 ( M = 32.32 SD = 9.24 ), to $50,000 $74,999 ( M = 32.50, SD = 7.74), to $35,000 $49,999 ( M = 32.58 SD = 8.12 ), to $100,000 $149,999 ( M = 32.69, SD = 7.95), to $15,000 $24,999 ( M = 32.83 SD = 7.05 ), to $150,000 or more ( M = 32.84 SD = 7.91 ) to $25,000 $34,999 ( M = 32.85 SD = 7.73 ), to $75,000 $99,999 ( M = 32.88 SD = 8 .07), in that order (Figure 4 17 ). For the one way ANOVA, some assumptions must first be verified. The three assumptions based on instrument design held true for this analysis, as mentioned in the Data Analysis section of the Methodology chapter. The last three assumptions are confir med through statistical tests. The first of these remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,332 homeowner responses from Florida, only six outliers were found. These outliers can be seen in the boxp lots of the HPKS scores by Household I ncomes in Figure 4 18 The next assumption is that e) the DV should be approximately normally distributed in each of the IV groups. This was confirmed through the normal Q Q p lots, stem & leaf plots, and histograms of each Household I ncome category Finally, the last assumption was that of f) homogenous variances between groups. There was also

PAGE 91

9 1 homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 725 ). The TSES scores were found to not vary significantly between the different Household Income categories of Floridian homeowners, F (7, 1324) = 0.087 p = 999 The full ANOVA table is available as Table 4 31 H2 C2 : Homeowners from across the study area with lowe r levels of H ousehold I ncome will self report lower Trust in Support Entities S cale (TSES) than other homeowners The descriptive statistics of the mean Trust in Support Entities Scale (TSES) scores by Household Income are shown in Table 4 32 For the test of this hypothesis, the dependent variable (DV) is the TSES score s and the independent variable (IV) is the Household I ncome s The TSES scores increased from Up to $14,999 ( n = 82, M = 31.40 SD = 9 .77 ), to $100,000 $ 149,999 ( n = 260, M = 31.85, SD = 7.82 ), to $50,000 $74,999 ( n = 496, M = 32.04 SD = 7.94 ), to $150,000 or more ( n = 130, M = 32.12 SD = 7.58 ) to $35,000 $49,999 ( n = 264, M = 32.45 SD = 7.94 ), to $75,000 $99,999 ( n = 355, M = 32.52 SD = 7.94 ), to $15,000 $24,999 ( n = 137, M = 32.69 SD = 7.13 ), to $25,000 $34,999 ( n = 219, M = 33.12 SD = 8.12) in that order (Figure 4 19 ). A one way ANOVA was conducted to determine if the TSES score was different for homeowners by H ousehold I ncome The three assumptions based on instrument design held true for this analysis, as mentioned in the Data Analysis section of the Methodology chapter. The last three assumptions are confirmed through statistical tests. There were only seven outliers out of th e sample of 1,943, as assessed by outlier labeling rule ; data was normally distributed for each group, as assessed by n ormal Q Q

PAGE 92

92 p lots, h istograms, and stem & leaf plots. There was also homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 235 ). The TSES scores were found to not vary significantly between the different Household Incom e categories of homeowners, F (7, 1935) = 0.790, p = .596. The full ANOVA table is available as Table 4 33. Age of homeowner H2 D1 : Florida homeowners who are younger will self report lower Trust in S upport E ntities S cal e (TSES) than other homeowners within the state The descriptive statistics of the Trust in Support Entities Scale (TSES) by s shown in Table 4 34. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independent variable (IV) is the Floridian h Age Homeowner responses were classified into five Age categories : 25 34 years old ( n = 162), 35 44 years old ( n = 199), 45 54 years old ( n = 261), to 55 64 years old ( n = 372), to 65 75 years old ( n = 338). The TSES scores increased from Floridian homeowners 45 54 years old ( M = 31.38 SD = 8.44 ), to 65 75 years old ( M = 31.77 SD = 7.52 ), to 55 64 years old ( M = 32.96 SD = 7.48 ), to 3 5 44 years old ( M = 33.76 SD = 7.44 ), to 25 34 years old ( M = 34.70 SD = 8.66 ) in that order (Figure 4 20 ). For the one way ANOVA, some assumptions must first be verified. The first three assumptions are based on the design of the study instrument. These assumptions are a) the dependent variable (DV) is continuous, b) the independent variable (IV) is categorical with two or more independent groups, and c) the cases (homeowner responses) are independent. All of these assumptions held true for this analysis. The last three assumptions are confirmed through statistical tests. The first of these

PAGE 93

93 remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. There were no outliers observed by comparing the extreme values to the range calculated by the outlier labeling rule The next assumption is that e) the DV should be approximately normall y distributed in each of the IV groups. This was confirmed by observing n ormal Q Q p lots, stem & leaf plots, and histograms of each A ge category Finally, the last assumption was that of f) homogenous variances between groups. There was also homogeneity of variances, as assessed by Levene's test for equality of variances ( p = 060 ). The TSES score was significantly different for the different categories of Floridian A ges F (4,132 7 ) = 6.71 p < .00 1 analysis (Table 4 35 ) revealed that the mean TSES score of homeowners 45 54 years old ( M = 31.38 SD = 8.44 ) and 65 75 years old ( M = 31. 77 SD = 7. 52 ) were significantly lower than the mean score s of Floridian homeowners a ge d 35 44 years old ( M = 33.76 SD = 7.44 ) and 25 34 years old ( M = 34.70, SD = 8.66) The TSES scores from Flor idians a ged 55 64 years old ( M = 32.96, SD = 7.48) was not significantly different than any other A ge categories in Florida. H2 D2 : Homeowners from across the study area who are younger will self report lower T rust in S upport Entities S cale (TSES) than other homeowners The descriptive statistics of the Trust in Support Entities Scale (TSES) by A ge is shown in Table 4 36. For the test of this hypothesis, the dependent variable (DV) is the TSES score and the independent variable (IV) is the Age The TSES increased from homeowners 45 54 years old ( n = 403 M = 30.86 SD =

PAGE 94

94 8.39 ), to 65 75 y ears old ( n = 452 M = 31.65 SD = 7.49 ), to 55 64 years old ( n = 534 M = 32.66 SD = 7.59 ), to 35 44 years old ( n = 305 M = 33.20 SD = 7.62 ), to 25 34 years old ( n = 249 M = 34.00 SD = 8.66) in that order (Figure 4 21 ). For the one way ANOVA, some assumptions must first be verified. The first three assumptions are based on the design of the study instrument. These assumptions are a) the dependent variable (DV) is continuous, b) the independent variable (IV) is categorical with two or more independen t groups, and c) the cases (homeowner responses) are independent. All of these assumptions held true for this analysis. The last three assumptions are confirmed through statistical tests. The first of these remaining assumptions was d) no significant outli ers in the IV groups in terms of the DV. Of the 1, 943 homeowner responses, seven outliers were found by comparing the extreme values of each category to the range produced by the outlier labeling rule The next assumption is that e) the DV should be approx imately normally distributed in each of the IV groups. This was confirmed by observing normal Q Q p lots, stem & leaf plots, and histograms of each A ge category Finally, the last assumption was that of f) homogenous variances between groups. Th e assumption of homogeneity of var iances was violated, as assessed by Levene's test for equality of v ariances ( p = 035 ). Due to the assumptions not holding true, a Welch ANOVA was performed The TSES score was significantly different for the different categories of A ges F (4, 855.14 ) = 7.54 p < .001. Games Howell post hoc analysis (Table 4 37 ) revealed that the mean TSES score of homeowners 45 54 years old ( M = 30.86 SD = 8.39 ) w ere significantly lower than the mean scores of homeowners 55 64 years old ( M

PAGE 95

95 = 32.66 SD = 7.59 ), 35 44 years old ( M = 33.20 SD = 7.62 ) and 25 34 years old ( M = 34.00 SD = 8.66 ). The mean TSE Scale score from homeowners 65 75 years old ( M = 31.65 SD = 7.49 ) was significantly lower than homeowners 35 44 years old ( M = 33.20 SD = 7.62 ), and 25 34 years old ( M = 34.00 SD = 8.66) Multiple Regression Modeling M ultiple regression modeling was used to determine if the four demographic characteristics ( loc ation, education, income and age ) could be used to predict a Hurricane Preparedness Knowledge Scale (HPKS) The protocol of the s e analys e s followed a similar pattern as established by the previous Exploring Differences Between Groups section. There were four independent multiple regression models explored; HPKS in Florida then study area, followed by TSES in Florida then study area. Hurricane Preparedness Knowledge Scale Florida regions This analysis explored the predic tive ability that the demographic characteristics ( location, education, income and age ) had in relation to the Hurricane Preparedness Knowledge Scale (HPKS) scores of Floridian homeowners As explained in the Multiple Regression Modeling section of the Me thodology chapter assumptions needed to be verified before any regression models could be further explored. There was independence of residual, as assessed by a Durbin Watson statistic of 2.028 The s catterplots of the studentized residuals against the s tandardized predicted values were examined to verify that overall linear relationship s exist as well as the existence of homoscedasticity (Figure 4 22). No indication of non linearity was identified however in the interest of being comprehensive, the individual partial

PAGE 96

96 regression plots of the independent variables were explored as well (Figure 4 23) Note that education and age were excluded due to not being significant. In respect to homoscedasticity, the scatterplots were examined for the typical patterns that violate this assumption, such as a distribution that was not approximately equal for all values of the predicted dependent variable (Figure 4 22). The next assumption was checking that multicollinearity did not exist bet ween independent var iables. The verification of this assumption entailed two steps. First, the correlation table was examined for any values greater than 0.7 (Table 4 38). The highest correlation was only 0.37 thus this step was verified. The second step required data from t he coefficients table. The tolerance values for each IV from the Coefficients of HPKS table, Table 4 40 was verified to be greater than 0.1. This satisfied the assumption that no multicollineari ty existed in the data. The next assumption involved checkin g for outliers in the data. This was verified through the standard residual values from the casewise diagnostics table provided by SPSS. This produced only three outliers among the 1,332 Floridian homeowner responses. Due to their very small influence on t he analysis these outliers were left in the dataset. The leverage values that were produced during the multiple regression analysis were examined for any risky (0.20 to 0.49) or dangerous (0.50 or above) values. The highest leverage value in Florida was 0. 010 thus satisfying this assumption. c ked to be less than 1.0. e Floridian homeowners was 0.014 thus indicating no highly influential points.

PAGE 97

97 The last assumption to be confirm ed was that the residuals were normally distributed. This was done through a histogram of the HPKS score frequency against the regression standardized residuals (Figure 4 24) As well as visual verification of normality, the mean and standard deviation wer e verified as very near 0 and 1 respectively. One last confirmatory check of normality was to check the normal P P plot for adherence to the diagonal line produced. With all of the assumptions verified, the interpretation of the output was allowed to conti nue. By interpreting the individual significance of the IVs from t he coefficient table (Table 4 40 ) it was observed that Education ( p = .624) and Age ( p = .276) were not significant. Model 1 consisted of two of the original four IVs. By interpreting the m odel summary, Table 4 39 it was determined that this model explained 5 1 % of the variance of the HPKS scores. Flori da Region and Income significantly predict HPKS score s, F ( 2,1329 ) = 36.688 p < .001, adj. R 2 = .0 51 Study area This analysis explored the predictive ability that the demographic characteristics ( location, education, income and age ) had in relation to the Hurricane Preparedness Knowledge Scale (HPKS) scores of homeowners across the study area. There was independence of residual, as assessed by a Durbin Watson statistic of 1. 918 The assumptions of linearity and homoscedasticity were verified through the scatterplots. There only existed six outliers from the 1,943 homeowner responses of the data set, they were retained in the analysis. The nor mality of the residuals as sumption was confirmed as well. In this analysis, Location was represented by the State variable. As in the Florida model, t he Education ( p = .787) and Age ( p = 942 ) variables were both found to not

PAGE 98

98 significantly contribute to the model The State and Income variable s were found to significantly predict HPKS scores, F (2 194 0 ) = 19.4119 p < .001, adj. R 2 = .01 9 (Table s 4 4 1 & 4 42 ). Trust in Support Entities Scale Florida regions This analysis explored the predictive ability th at the demographic characteristics ( location, education, income and age ) had in relation to the Trust in Support Entities Scale (TSES ) scores of Floridian homeowners. The assumptions of linearity, independence of errors, homoscedasticity, and normality of residuals were all met. There were seven outliers identified of the 1,332 homeowner responses. The variable of Income ( p = 670 ) was found to not significantly add to the model. The variable s of Florida Region ( p < .001), Age ( p < .00 5 ) and Education ( p < .05) statistically significantly predicted TSES scores of Floridian homeowners, F ( 3 13 28 ) = 10.892 p < .001, adj. R 2 = 022 (Tables 4 43 & 4 4 4 ) Study area This analysis explored the predictive ability that the demographic characteristics ( location, e ducation, income and age ) had in relation to the Trust in Support Entities Scale (TSES ) scores of homeowners across the study area The assumptions of linearity, independence of errors, homoscedasticity, and normality of residuals were all met. There were eight outliers identified of the 1, 943 homeowner responses. The variables of Education ( p = 165 ), and Income ( p = 544 ) were both found to not significantly add to the model. The variable s of State ( p < .001) and Age ( p < .005 ) statistically significantly predicted TSES scores of homeowners, F (2 194 0 ) = 15.833 p < .00 1 adj. R 2 = 015 (Tables 4 4 5 & 4 4 6 ).

PAGE 99

99 Table 4 1. Initial PCA total variance explained Component Initial e igenvalues Rotation s ums of s quared l oadings Total % of Variance Cumulative % Total % of Variance Cumulative % 1 9.277 20.616 20.616 5.315 11.811 11.811 2 4.178 9.285 29.901 4.246 9.436 21.247 3 3.486 7.747 37.648 3.603 8.008 29.255 4 2.38 0 5.289 42.937 3.186 7.079 36.334 5 1.973 4.385 47.323 2.745 6.101 42.435 6 1.788 3.974 51.297 2.337 5.193 47.628 7 1.122 2.492 53.789 1.96 0 4.356 51.984 8 1.032 2.293 56.082 1.844 4.099 56.082 Table 4 2. Revised PCA total variance explained Component Initial e igenvalues Rotation sums of s quare d l oadings Total % of v ariance Cumulative % Total % of v ariance Cumulative % 1 6.373 21.975 21.975 4.139 14.272 14.272 2 3.678 12.681 34.657 4.028 13.89 0 28.161 3 2.79 0 9.621 44.278 2.793 9.632 37.794 4 1.959 6.757 51.035 2.597 8.955 46.749 5 1.647 5.68 0 56.714 2.573 8.872 55.621 6 1.423 4.907 61.621 1.74 0 6 .000 61.621 Table 4 3. Split sample validation comparison KMO measure of sampling a dequacy Bartlett's test of s phericity Number of c omponents % of variance e xplained % in same c omponents % same factor loading o rder Whole s ample 0.878 p < .001 6 61.62% 100% 100 .00 % Rnd h alf 1 0.874 p < .001 6 62.06% 100% 79.31% Rnd h alf 2 0.873 p < .001 6 61.94% 100% 72.41%

PAGE 100

100 Table 4 4. Rotated component matrix HPKS TSES Q4.52 0 .812 Q4.39 0 .794 Q4.37 0 .793 Q4.30 0 .767 Q4.22 0 .689 Q4.45 0 .591 Q4.15 0 .782 Q4.27 0 .756 Q4.14 0 .733 Q4.16 0 .632 Q4.26 .422 0 .627 Table 4 5. Descriptive statistics for the HPKS and TSES No. of items M (SD) Skewness Kurtosis HPKS 6 40.84 ( 8.85 ) 0. 11 0.0 6 0 .8 8 TSES 5 32.31 (7.95) 0.24 0.41 0 .8 0 Table 4 6. Descriptive statistics of HPKS s cores by Region of Florida N Mean Std. d eviation Std. e rror 95% confidence interval for m ean Minimum Maximum Lower b ound Upper b ound North Florida 163 38.96 8.71 .68 37.61 40.30 19.00 60.00 West Coast 412 39.22 8.94 .44 38.36 40.09 6.00 60.00 Panhandle 46 40.48 7.50 1.11 38.25 42.70 24.00 59.00 East Coast 350 41.47 8.80 .47 40.55 4 2.40 9.00 60.00 South Florida 305 42.71 8.47 .48 41.76 43.67 22.00 60.00 Northwest 56 44.25 8.07 1.08 42.09 46.41 31.00 60.00 Total 1332 40.83 8.82 .24 40.36 41.31 6.00 60.00

PAGE 101

101 Table 4 HPK S s cores by Region of Florida 95% confidence i nterval (I) FL_Reg (J) FL_Reg Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Northwest Panhandle 3.77 1.73 0 .25 1.16 8.71 North Florida 5.29 1.35 0 .00 1.45 9.13 West Coast 5.03 1.24 0 .00 1.50 8.56 East Coast 2.78 1.25 0 .23 0.79 6.34 South Florida 1.54 1.26 0 .83 2.07 5.14 Panhandle Northwest 3.77 1.73 0 .25 8.71 1.16 North Florida 1.52 1.45 0 .90 2.62 5.66 West Coast 1.26 1.35 0 .94 2.60 5.11 East Coast 1.00 1.36 0 .98 4.88 2.89 South Florida 2.23 1.37 0 .58 6.15 1.69 North Florida Northwest 5.29 1.35 0 .00 9.13 1.45 Panhandle 1.52 1.45 0 .90 5.66 2.62 West Coast 0.26 0.80 0 .00 2.56 2.03 East Coast 2.52 0.82 0 .03 4.87 0.17 South Florida 3.75 0.84 0 .00 6.16 1.35 West Coast Northwest 5.03 1.24 0 .00 8.56 1.50 Panhandle 1.26 1.35 0 .94 5.11 2.60 North Florida 0.26 0.80 1 .00 2.03 2.56 East Coast 2.25 0.63 0 .01 4.06 0.45 South Florida 3.49 0.66 0 .00 5.36 1.62 East Coast Northwest 2.78 1.25 0.23 6.34 0.79 Panhandle 1.00 1.36 0.98 2.89 4.88 North Florida *2.52 0.82 0.03 0.17 4.87 West Coast 2.25 0.63 0.01 0.45 4.06 South Florida 1.24 0.68 0.45 3.18 0.70 South Florida Northwest 1.54 1.26 0.83 5.14 2.07 Panhandle 2.23 1.37 0.58 1.69 6.15 North Florida 3.75 0.84 0.00 1.35 6.16 West Coast 3.49 0.66 0.00 1.62 5.36 East Coast 1.24 0.68 0.45 0.70 3.18 The mean difference is significant at the 0.05 level.

PAGE 102

102 Table 4 8. Descriptive statistics of HPKS scores by States 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Georgia 106 37.32 9.44 0.92 35.50 39.14 12 58 Florida 1332 40.83 8.82 0.24 40.36 41.31 6 60 Texas 329 41.13 8.74 0.48 40.19 42.08 14 60 Alabama & Mississippi 55 42.07 8.87 1.20 39.68 44.47 18 57 Louisiana 121 42.69 8.14 0.74 41.22 44.15 24 60 Total 1943 40.84 8.85 0.20 40.45 41.24 6 60 Table 4 HPKS scores by States 95% confidence i nterval (I) States (J) States Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Georgia Florida 3.51 0.89 0.00 5.94 1.09 Alabama & Mississippi 4.75 1.46 0.01 8.75 0.76 Louisiana 5.37 1.17 0.00 8.56 2.17 Texas 3.81 0.98 0.00 6.50 1.13 Florida Georgia *3.51 0.89 0.00 1.09 5.94 Alabama & Mississippi 1.24 1.21 0.85 4.55 2.07 Louisiana 1.85 0.84 0.18 4.13 0.43 Texas 0.30 0.54 0.98 1.78 1.18 Alabama & Mississippi Georgia *4.75 1.46 0.01 0.76 8.75 Florida 1.24 1.21 0.85 2.07 4.55 Louisiana 0.61 1.43 0.99 4.52 3.30 Texas 0.94 1.28 0.95 2.56 4.44 Louisiana Georgia *5.37 1.17 0.00 2.17 8.56 Florida 1.85 0.84 0.18 0.43 4.13 Alabama & Mississippi 0.61 1.43 0.99 3.30 4.52 Texas 1.55 0.94 0.46 1.00 4.11 Texas Georgia *3.81 0.98 0.00 1.13 6.50 Florida 0.30 0.54 0.98 1.18 1.78 Alabama & Mississippi 0.94 1.28 0.95 4.44 2.56 Louisiana 1.55 0.94 0.46 4.11 1.00 The mean difference is significant at the 0.05 level.

PAGE 103

103 Table 4 10. Descriptive statistics of HPKS scores by Education Levels (Florida) 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum HS Diploma or less 211 39.68 9.00 0.62 38.46 40.90 18 60 /AS Degree 512 40.83 9.11 0.40 40.04 41.63 6 60 Bachelor's Degree 385 41.37 8.50 0.43 40.52 42.22 12 60 Master's Degree 190 40.95 8.33 0.60 39.76 42.14 22 60 Doctoral Degree 34 41.35 9.41 1.61 38.07 44.64 19 60 Total 1332 40.83 8.82 0.24 40.36 41.31 6 60 Table 4 11. ANOVA of HPKS scores by Educational Levels (Florida) Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 402.99 4 100.75 1.30 0.270 Linear t erm Unweighted 83.68 1 83.68 1.08 0.300 Weighted 222.94 1 222.94 2.87 0.091 Deviation 180.05 3 60.02 0.77 0.510 Within g roups 103200.67 1327 77.77 Total 103603.66 1331 Table 4 12. Descriptive statistics of HPKS scores by Education Levels 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum HS Diploma or less 308 39.75 9.22 0.53 38.71 40.78 12 60 /AS 730 40.70 9.13 0.34 40.04 41.36 6 60 Bachelor's Degree 567 41.35 8.52 0.36 40.65 42.06 12 60 Master's Degree 282 41.46 8.19 0.49 40.50 42.42 21 60 Doctoral Degree 56 40.50 9.12 1.22 38.06 42.94 18 60 Total 1943 40.84 8.85 0.20 40.45 41.24 6 60

PAGE 104

104 Table 4 13. ANOVA of HPKS scores by Educational Levels Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 647.99 4 162.00 2.07 0.082 Linear t erm Unweighted 57.63 1 57.63 0.74 0.391 Weighted 420.61 1 420.61 5.38 0.02 0 Deviation 227.37 3 75.79 0.97 0.406 Within g roups 151407.76 1938 78.13 Total 152055.75 1942 Table 4 14. Descriptive statistics of HPKS scores by Household Incomes (Florida) 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower bo und Upper b ound Minimum Maximum Up to $14,999 59 36.37 9.13 1.19 33.99 38.75 21 60 $15,000 $24,999 95 39.53 8.84 0.91 37.73 41.33 19 60 $25,000 $34,999 157 39.57 8.79 0.70 38.18 40.95 19 60 $35,000 $49,999 192 40.59 9.15 0.66 39.29 41.89 9 60 $50,000 $74,999 353 40.60 8.60 0.46 39.70 41.50 18 60 $75,000 $99,999 243 42.64 8.62 0.55 41.55 43.73 17 60 $100,000 $149,999 160 41.58 8.60 0.68 40.24 42.92 6 60 $150,000 or more 73 43.01 8.07 0.94 41.13 44.90 25 60 Total 1332 40.83 8.82 0.24 40.36 41.31 6 60

PAGE 105

105 Table 4 HPKS scores by Household Incomes (Florida) 95% confidence i nterval (I) Income (J) Income Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Up to $14,999 $15,000 $24,999 3.15 1.45 0.364 7.54 1.24 $25,000 $34,999 3.19 1.33 0.243 7.24 0.85 $35,000 $49,999 4.22 1.30 0.026 8.16 0.27 $50,000 $74,999 4.22 1.23 0.014 7.95 0.50 $75,000 $99,999 6.27 1.27 0.000 10.11 2.43 $100,000 $149,999 5.21 1.33 0.002 9.24 1.17 $150,000 or more 6.64 1.53 0.000 11.28 2.00 $15,000 $24,999 Up to $14,999 3.15 1.45 0.364 1.24 7.54 $25,000 $34,999 0.04 1.13 1.000 3.48 3.40 $35,000 $49,999 1.06 1.09 0.978 4.38 2.26 $50,000 $74,999 1.07 1.01 0.964 4.13 1.99 $75,000 $99,999 3.12 1.06 0.064 6.32 0.09 $100,000 $149,999 2.05 1.13 0.607 5.49 1.38 $150,000 or more 3.49 1.36 0.168 7.61 0.63 $25,000 $34,999 Up to $14,999 3.19 1.33 0.243 0.85 7.24 $15,000 $24,999 0.04 1.13 1.000 3.40 3.48 $35,000 $49,999 1.02 0.94 0.959 3.87 1.83 $50,000 $74,999 1.03 0.84 0.922 3.57 1.51 $75,000 $99,999 3.08 0.89 0.014 5.79 0.36 $100,000 $149,999 2.01 0.98 0.445 4.99 0.96 $150,000 or more 3.45 1.24 0.099 7.20 0.30 $35,000 $49,999 Up to $14,999 *4.22 1.30 0.026 0.27 8.16 $15,000 $24,999 1.06 1.09 0.978 2.26 4.38 $25,000 $34,999 1.02 0.94 0.959 1.83 3.87 $50,000 $74,999 0.01 0.78 1.000 2.38 2.37 $75,000 $99,999 2.05 0.84 0.224 4.61 0.50 $100,000 $149,999 0.99 0.93 0.964 3.83 1.84 $150,000 or more 2.43 1.20 0.467 6.07 1.22 $50,000 $74,999 Up to $14,999 *4.22 1.23 0.014 0.50 7.95 $15,000 $24,999 1.07 1.01 0.964 1.99 4.13 $25,000 $34,999 1.03 0.84 0.922 1.51 3.57 $35,000 $49,999 0.01 0.78 1.000 2.37 2.38 $75,000 $99,999 2.04 0.73 0.093 4.25 0.16 $100,000 $149,999 0.98 0.83 0.937 3.51 1.54 $150,000 or more 2.42 1.12 0.381 5.82 0.99

PAGE 106

106 Table 4 15. Continued 95% confidence i nterval (I) Income (J) Income Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound $75,000 $99,999 Up to $14,999 *6.27 1.27 0 .000 2.43 10.11 $15,000 $24,999 3.12 1.06 0 .064 0.09 6.32 $25,000 $34,999 *3.08 0.89 0 .014 0.36 5.79 $35,000 $49,999 2.05 0.84 0 .224 0.50 4.61 $50,000 $74,999 2.04 0.73 0 .093 0.16 4.25 $100,000 $149,999 1.06 0.89 0 .934 1.64 3.76 $150,000 or more 0.37 1.16 1.000 3.91 3.16 $100,000 $149,999 Up to $14,999 *5.21 1.33 0 .002 1.17 9.24 $15,000 $24,999 2.05 1.13 0 .607 1.38 5.49 $25,000 $34,999 2.01 0.98 0 .445 0.96 4.99 $35,000 $49,999 0.99 0.93 0 .964 1.84 3.83 $50,000 $74,999 0.98 0.83 0 .937 1.54 3.51 $75,000 $99,999 1.06 0.89 0 .934 3.76 1.64 $150,000 or more 1.43 1.23 0 .942 5.17 2.31 $150,000 or more Up to $14,999 *6.64 1.53 0 .000 2.00 11.28 $15,000 $24,999 3.49 1.36 0 .168 0.63 7.61 $25,000 $34,999 3.45 1.24 0 .099 0.30 7.20 $35,000 $49,999 2.43 1.20 0 .467 1.22 6.07 $50,000 $74,999 2.42 1.12 0 .381 0.99 5.82 $75,000 $99,999 0.37 1.16 1.000 3.16 3.91 $100,000 $149,999 1.43 1.23 0 .942 2.31 5.17 The mean difference is significant at the 0.05 level. Table 4 16. Descriptive statistics of HPKS scores by Household Incomes 95% c onfidence i nterval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Up to $14,999 82 37.16 9.71 1.07 35.02 39.29 13 60 $15,000 $24,999 137 39.99 8.60 0.74 38.54 41.45 19 60 $25,000 $34,999 219 39.94 9.09 0.61 38.73 41.15 13 60 $35,000 $49,999 264 40.20 8.99 0.55 39.11 41.29 9 60 $50,000 $74,999 496 40.66 8.72 0.39 39.89 41.43 12 60 $75,000 $99,999 355 42.43 8.58 0.46 41.54 43.33 17 60 $100,000 $149,999 260 41.22 8.84 0.55 40.14 42.30 6 60 $150,000 or more 130 42.53 8.03 0.70 41.14 43.92 18 60 Total 1943 40.84 8.85 0.20 40.45 41.24 6 60

PAGE 107

107 Table 4 HPKS scores by Household Incomes 95% confidence i nterval (I) Income (J) Income Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Up to $14,999 $15,000 $24,999 2.83 1.23 0.288 6.55 0.89 $25,000 $34,999 2.78 1.14 0.221 6.23 0.67 $35,000 $49,999 3.04 1.11 0.112 6.41 0.33 $50,000 $74,999 3.50 1.05 0.019 6.68 0.32 $75,000 $99,999 5.27 1.08 0.000 8.54 2.01 $100,000 $149,999 4.06 1.11 0.007 7.44 0.69 $150,000 or more 5.37 1.24 0.000 9.13 1.61 $15,000 $24,999 Up to $14,999 2.83 1.23 0.288 0.89 6.55 $25,000 $34,999 0.06 0.96 1.000 2.85 2.96 $35,000 $49,999 0.20 0.92 1.000 3.01 2.60 $50,000 $74,999 0.67 0.85 0.994 3.24 1.91 $75,000 $99,999 2.44 0.88 0.106 5.12 0.24 $100,000 $149,999 1.23 0.93 0.890 4.04 1.59 $150,000 or more 2.54 1.08 0.262 5.80 0.72 $25,000 $34,999 Up to $14,999 2.78 1.14 0.221 0.67 6.23 $15,000 $24,999 0.06 0.96 1.000 2.96 2.85 $35,000 $49,999 0.26 0.80 1.000 2.70 2.17 $50,000 $74,999 0.72 0.71 0.972 2.89 1.44 $75,000 $99,999 2.49 0.75 0.022 4.78 0.21 $100,000 $149,999 1.28 0.81 0.755 3.73 1.16 $150,000 or more 2.59 0.97 0.133 5.54 0.36 $35,000 $49,999 Up to $14,999 3.04 1.11 0.112 0.33 6.41 $15,000 $24,999 0.20 0.92 1.000 2.60 3.01 $25,000 $34,999 0.26 0.80 1.000 2.17 2.70 $50,000 $74,999 0.46 0.67 0.997 2.49 1.57 $75,000 $99,999 2.23 0.71 0.038 4.40 0.07 $100,000 $149,999 1.02 0.77 0.887 3.35 1.31 $150,000 or more 2.33 0.94 0.204 5.19 0.52 $50,000 $74,999 Up to $14,999 *3.50 1.05 0.019 0.32 6.68 $15,000 $24,999 0.67 0.85 0.994 1.91 3.24 $25,000 $34,999 0.72 0.71 0.972 1.44 2.89 $35,000 $49,999 0.46 0.67 0.997 1.57 2.49 $75,000 $99,999 1.77 0.61 0.073 3.62 0.08 $100,000 $149,999 0.56 0.67 0.991 2.60 1.48 $150,000 or more 1.87 0.87 0.375 4.50 0.75

PAGE 108

108 Table 4 17. Continued 95% confidence i nterval (I) Income (J) Income Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound $75,000 $99,999 Up to $14,999 *5.27 1.08 0.000 2.01 8.54 $15,000 $24,999 2.44 0.88 0.106 0.24 5.12 $25,000 $34,999 *2.49 0.75 0.022 0.21 4.78 $35,000 $49,999 *2.23 0.71 0.038 0.07 4.40 $50,000 $74,999 1.77 0.61 0.073 0.08 3.62 $100,000 $149,999 1.21 0.72 0.694 0.96 3.39 $150,000 or more 0.10 0.90 1.000 2.83 2.63 $100,000 $149,999 Up to $14,999 *4.06 1.11 0.007 0.69 7.44 $15,000 $24,999 1.23 0.93 0.890 1.59 4.04 $25,000 $34,999 1.28 0.81 0.755 1.16 3.73 $35,000 $49,999 1.02 0.77 0.887 1.31 3.35 $50,000 $74,999 0.56 0.67 0.991 1.48 2.60 $75,000 $99,999 1.21 0.72 0.694 3.39 0.96 $150,000 or more 1.31 0.94 0.862 4.17 1.55 $150,000 or more Up to $14,999 *5.37 1.24 0.000 1.61 9.13 $15,000 $24,999 2.54 1.08 0.262 0.72 5.80 $25,000 $34,999 2.59 0.97 0.133 0.36 5.54 $35,000 $49,999 2.33 0.94 0.204 0.52 5.19 $50,000 $74,999 1.87 0.87 0.375 0.75 4.50 $75,000 $99,999 0.10 0.90 1.000 2.63 2.83 $100,000 $149,999 1.31 0.94 0.862 1.55 4.17 The mean difference is significant at the 0.05 level. Table 4 18. Descriptive statistics of HPKS Age (Florida) 95% confidence i nterv al for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum 25 34 years 162 42.28 9.21 0.72 40.86 43.71 21 60 35 44 years 199 40.64 8.99 0.64 39.39 41.90 19 60 45 54 years 261 40.40 8.97 0.55 39.31 41.50 6 60 55 64 years 372 41.17 8.41 0.44 40.31 42.02 18 60 65 75 years 338 40.22 8.83 0.48 39.28 41.17 9 59 Total 1332 40.83 8.82 0.24 40.36 41.31 6 60

PAGE 109

109 Table 4 19. ANOVA of HPKS Age (Florida) Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 564.27 4 141.07 1.82 0.123 Linear t erm Unweighted 293.06 1 293.06 3.77 0.052 Weighted 220.68 1 220.68 2.84 0.092 Deviation 343.60 3 114.53 1.48 0.220 Within g roups 103039.39 1327 77.65 Total 103603.66 1331 Table 4 20. Descriptive statistics of HPKS Age 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum 25 34 years 249 41.82 9.49 0.60 40.63 43.00 12 60 35 44 years 305 40.30 9.09 0.52 39.27 41.32 13 60 45 54 years 403 40.21 8.99 0.45 39.33 41.09 6 60 55 64 years 534 41.51 8.37 0.36 40.79 42.22 18 60 65 75 years 452 40.46 8.68 0.41 39.66 41.26 9 59 Total 1943 40.84 8.85 0.20 40.45 41.24 6 60 Table 4 21. ANOVA of HPKS Age Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 787.39 4 196.85 2.52 0.039 Linear t erm Unweighted 74.80 1 74.80 0.96 0.328 Weighted 27.97 1 27.97 0.36 0.550 Deviation 759.42 3 253.14 3.24 0.021 Within g roups 151268.36 1938 78.05 Total 152055.75 1942 Table 4 22. Descriptive statistics of TSES scores by Florida Region 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum North Florida 163 31.44 8.08 0.63 30.19 32.69 7 50 West Coast 412 31.90 7.87 0.39 31.14 32.67 5 50 Panhandle 46 32.09 7.88 1.16 29.75 34.43 16 50 East Coast 350 32.99 7.65 0.41 32.19 33.80 5 50 South Florida 305 33.82 7.83 0.45 32.94 34.70 11 50 Northwest 56 34.34 8.58 1.15 32.04 36.64 11 50 Total 1332 32.68 7.90 0.22 32.26 33.10 5 50

PAGE 110

110 Table 4 TSES scores by Florida Regions 95% confidence i nterval (I) Florida Region (J) Florida Region Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Northwest Panhandle 2.25 1.56 0.702 2.21 6.72 North Florida 2.90 1.22 0.164 0.58 6.37 West Coast 2.44 1.12 0.250 0.76 5.63 East Coast 1.35 1.13 0.841 1.88 4.58 South Florida 0.52 1.14 0.998 2.74 3.78 Panhandle Northwest 2.25 1.56 0.702 6.72 2.21 North Florida 0.65 1.31 0.996 3.10 4.39 West Coast 0.18 1.22 1.000 3.30 3.67 East Coast 0.90 1.23 0.978 4.42 2.61 South Florida 1.73 1.24 0.731 5.28 1.82 North Florida Northwest 2.90 1.22 0.164 6.37 0.58 Panhandle 0.65 1.31 0.996 4.39 3.10 West Coast 0.46 0.73 0.988 2.54 1.61 East Coast 1.55 0.75 0.299 3.68 0.58 South Florida 2.38 0.76 0.023 4.55 0.20 West Coast Northwest 2.44 1.12 0.250 5.63 0.76 Panhandle 0.18 1.22 1.000 3.67 3.30 North Florida 0.46 0.73 0.988 1.61 2.54 East Coast 1.09 0.57 0.399 2.72 0.54 South Florida 1.92 0.59 0.016 3.61 0.22 East Coast Northwest 1.35 1.13 0.841 4.58 1.88 Panhandle 0.90 1.23 0.978 2.61 4.42 North Florida 1.55 0.75 0.299 0.58 3.68 West Coast 1.09 0.57 0.399 0.54 2.72 South Florida 0.83 0.62 0.760 2.59 0.93 South Florida Northwest 0.52 1.14 0.998 3.78 2.74 Panhandle 1.73 1.24 0.731 1.82 5.28 North Florida *2.38 0.76 0.023 0.20 4.55 West Coast *1.92 0.59 0.016 0.22 3.61 East Coast 0.83 0.62 0.760 0.93 2.59 The mean difference is significant at the 0.05 level.

PAGE 111

111 Table 4 24. Descriptive statistics of TSES scores by State 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Georgia 106 29.39 7.97 0.77 27.85 30.92 5 48 Texas 329 31.58 7.79 0.43 30.74 32.43 8 50 Louisiana 121 31.93 7.84 0.71 30.52 33.34 13 50 Florida 1332 32.68 7.90 0.22 32.26 33.10 5 50 Alabama & Mississippi 55 34.00 8.84 1.19 31.61 36.39 10 49 Total 1943 32.31 7.95 0.18 31.95 32.66 5 50 Table 4 TSES scores by State 95% confidence i nterval (I) States (J) States Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Georgia Florida 3.29 0.80 0.000 5.47 1.11 Alabama & Mississippi 4.61 1.31 0.004 8.20 1.02 Louisiana 2.55 1.05 0.110 5.42 0.33 Texas 2.20 0.88 0.094 4.61 0.21 Florida Georgia *3.29 0.80 0.000 1.11 5.47 Alabama & Mississippi 1.32 1.09 0.744 4.29 1.65 Louisiana 0.75 0.75 0.858 1.30 2.80 Texas 1.10 0.49 0.161 0.23 2.43 Alabama & Mississippi Georgia *4.61 1.31 0.004 1.02 8.20 Florida 1.32 1.09 0.744 1.65 4.29 Louisiana 2.07 1.29 0.493 1.45 5.58 Texas 2.42 1.15 0.221 0.73 5.56 Louisiana Georgia 2.55 1.05 0.110 0.33 5.42 Florida 0.75 0.75 0.858 2.80 1.30 Alabama & Mississippi 2.07 1.29 0.493 5.58 1.45 Texas 0.35 0.84 0.994 1.95 2.65 Texas Georgia 2.20 0.88 0.094 0.21 4.61 Florida 1.10 0.49 0.161 2.43 0.23 Alabama & Mississippi 2.42 1.15 0.221 5.56 0.73 Louisiana 0.35 0.84 0.994 2.65 1.95 The mean difference is significant at the 0.05 level.

PAGE 112

112 Table 4 26. Descriptive statistics of TSES scores by Education Levels (Florida) 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum HS Diploma or less 211 33.25 7.96 0.55 32.17 34.33 5 50 /AS 512 32.52 7.91 0.35 31.83 33.21 5 50 Bachelor's Degree 385 33.25 7.75 0.40 32.47 34.02 5 50 Master's Degree 190 31.58 7.86 0.57 30.45 32.70 7 50 Doctoral Degree 34 31.29 8.67 1.49 28.27 34.32 12 50 Total 1332 32.68 7.90 0.22 32.26 33.10 5 50 Table 4 27. ANOVA of TSES scores by Education Levels (Florida) Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 501.332 4 125.333 2.015 0.09 0 Linear t erm Unweighted 163.873 1 163.873 2.635 0.105 Weighted 159.551 1 159.551 2.566 0.109 Deviation 341.781 3 113.927 1.832 0.139 Within g roups 82522.425 1327 62.187 Total 83023.757 1331 Table 4 28. Descriptive statistics of TSES scores by Education Levels 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum HS Diploma or less 308 32.87 8.24 0.47 31.95 33.80 5 50 /AS 730 32.17 7.96 0.29 31.59 32.75 5 50 Bachelor's Degree 567 32.75 7.65 0.32 32.12 33.38 5 50 Master's Degree 282 31.35 8.16 0.49 30.40 32.31 5 50 Doctoral Degree 56 31.23 7.71 1.03 29.17 33.30 12 50 Total 1943 32.31 7.95 0.18 31.95 32.66 5 50 Table 4 29. ANOVA of TSES scores by Education Levels Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 546.59 4 136.65 2.17 0.070 Linear t erm Unweighted 188.42 1 188.42 2.99 0.084 Weighted 205.26 1 205.26 3.26 0.071 Deviation 341.33 3 113.78 1.81 0.144 Within g roups 122099.82 1938 63.00 Total 122646.41 1942

PAGE 113

113 Table 4 30. Descriptive statistics of TSES scores by Household Incomes (Florida) 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Up to $14,999 59 32.32 9.24 1.20 29.91 34.73 9 50 $15,000 $24,999 95 32.83 7.05 0.72 31.40 34.27 13 50 $25,000 $34,999 157 32.85 7.73 0.62 31.63 34.07 10 50 $35,000 $49,999 192 32.58 8.12 0.59 31.43 33.74 5 50 $50,000 $74,999 353 32.50 7.74 0.41 31.69 33.31 7 50 $75,000 $99,999 243 32.88 8.07 0.52 31.86 33.90 7 50 $100,000 $149,999 160 32.69 7.95 0.63 31.45 33.93 5 50 $150,000 or more 73 32.84 7.91 0.93 30.99 34.68 15 50 Total 1332 32.68 7.90 0.22 32.26 33.10 5 50 Table 4 31. ANOVA of TSES scores by Household Incomes (Florida) Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 38.39 7 5.48 0.09 0.999 Linear t erm Unweighted 4.15 1 4.15 0.07 0.797 Weighted 2.09 1 2.09 0.03 0.855 Deviation 36.30 6 6.05 0.10 0.997 Within g roups 82985.37 1324 62.68 Total 83023.76 1331 Table 4 32. Descriptive statistics of TSES scores by Household Incomes 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Up to $14,999 82 31.40 9.77 1.08 29.26 33.55 5 50 $15,000 $24,999 137 32.69 7.13 0.61 31.49 33.90 11 50 $25,000 $34,999 219 33.12 8.12 0.55 32.04 34.20 10 50 $35,000 $49,999 264 32.45 7.94 0.49 31.49 33.41 5 50 $50,000 $74,999 496 32.04 7.94 0.36 31.34 32.74 7 50 $75,000 $99,999 355 32.52 7.94 0.42 31.70 33.35 7 50 $100,000 $149,999 260 31.85 7.82 0.48 30.90 32.80 5 50 $150,000 or more 130 32.12 7.58 0.66 30.80 33.43 15 50 Total 1943 32.31 7.95 0.18 31.95 32.66 5 50

PAGE 114

114 Table 4 33. ANOVA of TSES scores by Household Incomes Sum of s quares df Mean s quare F Sig. Between g roups (Combined) 349.46 7 49.92 0.79 0.596 Linear t erm Unweighted 1.55 1 1.55 0.03 0.876 Weighted 39.02 1 39.02 0.62 0.432 Deviation 310.44 6 51.74 0.82 0.555 Within g roups 122296.95 1935 63.20 Total 122646.41 1942 Table 4 34. Descriptive statistics of TSES Age (Florida) 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum 25 34 years 162 34.70 8.66 0.68 33.35 36.04 7 50 35 44 years 199 33.76 7.44 0.53 32.72 34.80 15 50 45 54 years 261 31.38 8.44 0.52 30.35 32.40 5 50 55 64 years 372 32.96 7.48 0.39 32.20 33.72 11 50 65 75 years 338 31.77 7.52 0.41 30.97 32.58 9 50 Total 1332 32.68 7.90 0.22 32.26 33.10 5 50

PAGE 115

115 Table 4 TSES Age (Florida) 95% confidence in terval (I) Age (J) Age Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD 25 34 years 35 44 years 0.93 0.83 0.792 1.33 3.20 45 54 years *3.32 0.78 0.000 1.18 5.46 55 64 years 1.74 0.74 0.129 0.28 3.75 65 75 years *2.93 0.75 0.001 0.88 4.97 35 44 years 25 34 years 0.93 0.83 0.792 3.20 1.33 45 54 years *2.39 0.74 0.011 0.38 4.40 55 64 years 0.80 0.69 0.771 1.08 2.68 65 75 years *1.99 0.70 0.036 0.08 3.90 45 54 years 25 34 years 3.32 0.78 0.000 5.46 1.18 35 44 years 2.39 0.74 0.011 4.40 0.38 55 64 years 1.59 0.63 0.089 3.31 0.14 65 75 years 0.40 0.65 0.973 2.16 1.37 55 64 years 25 34 years 1.74 0.74 0.129 3.75 0.28 35 44 years 0.80 0.69 0.771 2.68 1.08 45 54 years 1.59 0.63 0.089 0.14 3.31 65 75 years 1.19 0.59 0.256 0.42 2.80 65 75 years 25 34 years 2.93 0.75 0.001 4.97 0.88 35 44 years 1.99 0.70 0.036 3.90 0.08 45 54 years 0.40 0.65 0.973 1.37 2.16 55 64 years 1.19 0.59 0.256 2.80 0.42 The mean difference is significant at the 0.05 level. Table 4 36. Descriptive statistics of TSES Age 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum 25 34 years 249 34.00 8.66 0.55 32.91 35.08 7 50 35 44 years 305 33.20 7.62 0.44 32.34 34.06 13 50 45 54 years 403 30.86 8.39 0.42 30.04 31.68 5 50 55 64 years 534 32.66 7.59 0.33 32.01 33.30 5 50 65 75 years 452 31.65 7.49 0.35 30.96 32.34 5 50 Total 1943 32.31 7.95 0.18 31.95 32.66 5 50

PAGE 116

116 Table 4 37. Games Howell post hoc of TSES Age 95% confidence i nterval (I) Age (J) Age Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Games Howell 25 34 years 35 44 years 0.80 0.70 0.785 1.12 2.72 45 54 years *3.14 0.69 0.000 1.25 5.03 55 64 years 1.34 0.64 0.224 0.41 3.09 65 75 years *2.35 0.65 0.003 0.56 4.13 35 44 years 25 34 years 0.80 0.70 0.785 2.72 1.12 45 54 years *2.34 0.60 0.001 0.69 3.99 55 64 years 0.54 0.55 0.859 0.95 2.04 65 75 years *1.55 0.56 0.047 0.01 3.08 45 54 years 25 34 years 3.14 0.69 0.000 5.03 1.25 35 44 years 2.34 0.60 0.001 3.99 0.69 55 64 years 1.80 0.53 0.007 3.25 0.34 65 75 years 0.79 0.55 0.596 2.29 0.70 55 64 years 25 34 years 1.34 0.64 0.224 3.09 0.41 35 44 years 0.54 0.55 0.859 2.04 0.95 45 54 years *1.80 0.53 0.007 0.34 3.25 65 75 years 1.00 0.48 0.227 0.31 2.32 65 75 years 25 34 years 2.35 0.65 0.003 4.13 0.56 35 44 years 1.55 0.56 0.047 3.08 0.01 45 54 years 0.79 0.55 0.596 0.70 2.29 55 64 years 1.00 0.48 0.227 2.32 0.31 The mean difference is significant at the 0.05 level. Table 4 38. Correlations of HPKS and Demographic Characteristics HPKS Florida Region Education Income Age Pearson c orrelation HPKS 1.00 Florida Region 0.18 1.00 Education 0.04 0.08 1.00 Income 0.15 0.04 0.37 1.00 Age 0.05 0.07 0.07 0.11 1.00 Table 4 39. Model summary of HPKS model 1 (Florida) Model R R 2 Adjusted R 2 Std. error of the e stimate Durbin Watson 1 0.229a 0.052 0.051 8.59519 2.028 a Predictors: (Constant), F lorida Regions, Income

PAGE 117

117 Table 4 40. Regression coefficients of HPKS model 1 (Florida) Unstandardized Coefficients Standardized Coefficients 95.0% Confidence Interval for B Collinearity Statistics Model B Std. Error Beta t Sig. Lower Bound Upper Bound Tol. VIF 1 (Constant) 35.12 0.76 46.0 8 0.000 33.63 36.61 FL_Region 1.029 0.16 0.176 6.58 0.000 0.72 1.34 0.998 1.002 Income 0.697 0.13 0.139 5.19 0.000 0.43 0.96 0.998 1.002 Table 4 41. Model summary of HPKS model 2 Model R R 2 Adjusted R 2 Std. error of the e stimate Durbin Watson 2 .140a 0.02 0.019 8.7659 1.918 a Predictors: (Constant), Income, States Table 4 42. Regression coefficients of HPKS model 2 Unstandardized c oefficients Standardized c oefficients 95.0% confidence i nterval for B Collinearity s tatistics Model B Std. e rror Beta t Sig. Lower b ound Upper b ound Tol. VIF 2 (Constant) 37.036 0.65 57.24 0 .000 35.77 38.31 Income 0.553 0.11 0.112 4.97 0 .000 0.34 0.77 0.998 1.002 States 0.806 0.23 0.08 0 3.55 0 .00 0 0.36 1.25 0.998 1.002 Table 4 43. Model summary of TSES model 1 (Florida) Model R R 2 Adjusted R 2 Std. error of the e stimate Durbin Watson 1 .155 a 0.024 0.022 7.81131 2.045 a Predictors: (Constant), Florida Region, Age, Education Table 4 44. Regression coefficients of TSES model 1 (Florida) Model Unstandardized Coefficients Standardized Coefficients 95.0% Confidence Interval for B Collinearity Statistics B Std. Error Beta t Sig. Lower Bound Upper Bound Tol. VIF 1 (Constant) 34.36 0.89 38.78 .000 32.63 36.10 Florida Region 0.58 0.14 0.11 4.07 .000 0.30 0.86 0.99 0 1.01 0 Age 0.54 0.16 0.092 3.368 .001 0.859 0.227 0.994 1.006 Education 0. 42 0.21 0.054 1.977 .048 0.843 0.003 0.995 1.005

PAGE 118

118 Table 4 45. Model summary of TSES model 2 Model R R 2 Adjusted R 2 Std. error of the e stimate Durbin Watson 2 .127 a 0. 016 0. 015 7. 88698 2.018 a Predictors: (Constant), States, Age Table 4 46. Regression coefficients of TSES model 2 Unstand ardized c oefficients Standardized c oefficients 95.0% confidence i nterval for B Collinearity s tatistics Model B Std. e rror Beta t Sig Lower b ound Upper b ound Tol. VIF 2 (Constant) 32 23 0. 749 43 .0 5 0 .000 30.76 33 70 States .83 0.182 0.103 4.57 0 .000 0.47 1.19 1.00 1.00 Age .46 0.135 0.07 6 3.38 0 .001 0.72 0.19 1.00 1.00

PAGE 119

119 Figure 4 1. Boxplots of HPKS scores by Florida Regions Figure 4 2. Means of HPKS scores by Education Levels (Florida)

PAGE 120

120 Figure 4 3. Boxplots of HPKS scores by Education Levels (Florida) Figure 4 4. Means of HPKS scores by Education Levels

PAGE 121

121 Figure 4 5. Means of HPKS scores by Household Incomes (Florida) Figure 4 6. Boxplots of HPKS scores by Household Incomes (Florida)

PAGE 122

122 Figure 4 7. Means of HPKS scores by Household Incomes Figure 4 8. Means of HPKS Age (Florid a)

PAGE 123

123 Figure 4 9. Boxplots of HPKS Age (Florida) Figure 4 10. Means of HPKS Age

PAGE 124

124 Figure 4 11. Boxplots of TSES scores by Florida Regions Figure 4 12. Boxplots of TSES scores by State

PAGE 125

125 Figure 4 13. Means of TSES scores by Education Levels (Florida) Figure 4 14. Boxplots of TSES scores by Education Levels (Florida)

PAGE 126

126 Figure 4 15. Means of TSES scores by Education Levels Figure 4 16. Boxplots of TSES scores by Education Levels

PAGE 127

127 Figure 4 17. Means of TSES scores by Household Incomes (Florida) Figure 4 18. Boxplots of TSES scores by Household Incomes (Florida)

PAGE 128

128 Figure 4 19. Means of TSES scores by Household Incomes Figure 4 20. Means of TSES Age (Florida)

PAGE 129

129 Figure 4 21. Means of TSES Age Figure 4 22. Scatterplot of studentized residuals by standardized predicted values

PAGE 130

130 Figure 4 23. Partial regression plots of HPKS versus Demographic Characteristics Figure 4 24. Histogram of HPKS score frequency versus regression standardized residual

PAGE 131

131 CHAPTER 5 DISCUSSION Overview This chapter begin s by exploring the results of the prior chapter s as they support or do not support the hypotheses that were developed out of the literature review. Then the discussion move s to possible insights that may have been illuminated by the analys e s, demonstrating results that were a li g ne d with the expectations or contradict ed Possible explanations for these results wi ll be then be discussed followed by possible implications that may affect how hurricane preparedness information is disseminated, future research, and government policy Finally, any additional limitations that may have emerged during the analyse s are addr essed and the thesis proper ends with the C onclusions section Hypotheses Implications The layout of this section is the same as the Exploring Differences Between Groups section of the previous chapter. The first eight hypotheses all focus on differences in the mean HPKS scores grouped by the demographic characteristics ( location, education, income and age ) The second eight hypotheses all focus on the differences in the mean TSES scores grouped by the same demographic characteristics. Research Question 1: Hurricane Preparedness Knowledge To recap o what extent demographic characteristics ( location, education, income and age ) could predict homeowne reported score on this Hurricane P reparedness Knowledge S cale (HPKS) The e ffect of each of these demographic characteristics will first be

PAGE 132

132 explored within the state of Florida, then once again across the study area. After this there wil l be a brief discussion of the findings of the analysis and possible explanations of these findings. Location of home H1 A1 : Homeowners located in the counties of Panhandle will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other regions within the state Despite the existence of statistically significant differences of mean HPKS scores between Florida Regions ( p < .001), there existed no difference s Panhandle and other regions. Therefore, there is not support for the hypothesis that Panhandle would report lower HPKS scores than other R egions within the state. The existence of relaxed building codes in the panhandle of Florida contributed to the assumption that hurricanes were not taken as seriously as in other regions of Florida. This hypothesis of lower HPKS scores in the panhandle was thought to possibly be a supportin g reason for the relaxed wind speed building codes, with the economic burden of local bu ilders being the primary motivation ( Peacock et al. 2005) Another basis for the hypothesis was the predicted vulnerability of landfall locations based on past hurricanes Figure 5 1 depicts the p redicted frequency of hurricanes across the study area. The Panhandle West coast and North Fl orida areas all exhibited lower risk due to reduced frequency of hurricane strikes. The analysis shows that although this hypothesis was not supported the inc lusion of the hurricane return period in the process was warranted analysis indicated that homeowners in North Florida ( n = 163, M = 38.96, SD = 8.71 ) and the West Coast of Florida ( n = 412, M = 39.22, SD = 8.94) reported significantly

PAGE 133

133 lower mean HPKS score s than the homeowners in the East Coast ( n = 350, M = 41.47, SD = 8.80 ), South Florida ( n = 305 M = 42.71 SD = 8.47 ), and Northwest Florida ( n = 56 M = 44.25 SD = 8.82 ). Initially the findings that homeowners in t he East Coast and Sout h Florida possessed a higher hurricane preparedness knowledge was more easily understood than those in Northwest Florida After a deeper review of the literature about hurricanes and the Northwest area of Florida, this was slightly better understood. Soli s, Thomas, and Letson ( 2010) explored They compared homeowners in the Northwest Panhandle to those living in the Southeast Peninsula of Florida who had experienced the Atlantic Hurricane Season of 2005 Solis, Thomas, and Letson (2010) found that these regional differences were significant and that homeowners in the Northwest Panhandle were more likely to evacuate when ordered. This finding might crossover to an increased knowledge of hurricane preparedness and warrants further exploration. For this study, the Northwest Florida region only consisted of Escambia, Santa Rosa, and Okaloosa counties and slightly more than four percent of the Floridian responses. Further exploration of this region could justifiably be an area of future interest to further the understanding of these recent results. H1 A2 : Homeowners located in Georgia will self report lower H urricane P reparedness Kno wledge S cale (HPKS) than other S tates within the study area. There existed statistically significant difference s of mean HPKS scores between States ( p < .001). The mean HPKS score of h omeowners in the state of Georgia was significantly lower than the other five states ; therefore the hypothesis of lower HPKS

PAGE 134

134 scores in Georgia was supported. Referring back to Figure 5 1, it is readily see n that this area of Southeastern Georgia is the least threatened coast of the entire study area. The study area also stretched inland the furthest in Georgia in order to collect a larger samp le. It stands to reason that perceived h urricane risk would decrease as the distance from the coast increases. The connection of perceived risk to motivation of increase d knowledge h as been made by the Protection Motivation Theory ( Rogers, 1975) ; t herefore, this further supported the hypothesis of lower HPKS scores due to decreased hurricane risk to homeowners in the Georgia sample. Highest education level of homeowner H1 B1 : Florida homeowners attaining will self report lower scores on H urricane P reparedness Knowledge S cale (HPKS) than other homeowners within the state There were no statistically significa nt differences of Floridian mean HPKS scores between the different levels of E ducation ( p = 270). Therefore the hypothesis of Floridian homeowners attaining Degree reporting lower HPKS scores than other Floridian homeowners was not supported H1 B2 : Homeowners from across the study area attaining will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners mean HPKS scores between the different levels of E ducation ( p = 082 ). Therefore the hypothesis of homeowners attaining reporting lower HPKS scores than other homeowners was not supported The lack of significant differences between education levels was surprising. This hypothesis was based on other studies linking education al attainment to higher motivation to seek knowledge in different areas and more specif ically to aspects of

PAGE 135

135 hurr icane and hazard knowledge Tierney, Lindell, and Perry (2001) listed education u ring the review of literature, there were several studies in which education was not a significant factor as well. This may indicate that the lack of education being a significant factor in previous risk perception (Donahue et al. 2014 ) and optimistic bias (Trumbo et al. 2014) shared similar ities to this stu HPKS measure In turn, this also suggested that studies where education was a factor, which measured areas such as perceived haz a rd knowledge (Ge et al. 2011) or even other measures of hurricane preparedness knowledge (Baker, 2011) were dissimilar f HPKS measure. Household income H1 C1 : Florida homeowners with lower levels of household income will self report lower scores H urricane P reparedness Knowledge S cale (HPKS) than other homeowners with in the state There existed statistically significant differences of mean HPKS scores between levels of Household Income ( p < .001). The mean HPKS score of Floridian households claiming Up to $14,999 was significantly lower than the five highest Fl oridian household income levels of $35,000 or more ; therefore, the hypothesis of lower HPKS scores being reported by lower I ncomes was supported. H1 C2 : Homeowners from across the study area with lower levels of household income will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners HPKS scores between levels of Household Income ( p < .001). The mean HPKS score of househo lds claiming Up to $14,999 was significantly lower than the four highest

PAGE 136

136 household income levels of $ 50 ,000 or more ; therefore, the hypothesis of lower HPKS scores being reported by lower I ncomes was supported. This study considered increased income as a mitigating factor in the amount of damage from hurricanes. This was based on the premise that homeowners with increased availability to financial resources would have better access to products and home features that could potentially lessen damage ca used by hurricanes to their property. Hurricane preparedness knowledge was thought of in a similar manner : homeowners with increased knowledge could mitigate some damages incurred from hurricanes. This rationale was supported by previous studies as resear chers found that higher levels of household income was a contributing factor to lowered social vulnerability and increased community resilience ( Bergstrand et al. 2015) increased household hurricane preparedness ( Baker 2011) and increased hurricane ris k perception ( Trumbo et al. 2014) This current study found a strong trend of increasing HPKS scores wit h increasing household incomes within both the Florida sample and the study area as a whole (Figure 5 2) The Florida sample indicated the lowest income level of Up to $14,999 was significant ly different from the top five income levels beginning at $35,000 The entire study area sample indicated significant differences between the l owest income level of Up to $14,999 beginning at $50,000 and continuing through the highest levels (Figure 5 2). Age of homeowner H1 D1 : Florida homeowners who are younger will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners within the state.

PAGE 137

137 HPKS scores between the different levels of Age ( p = 123 ). Therefore the hypothesis of Floridian homeowners who are Younger reporting lower HPKS scores than other Floridian homeowners was not supported H1 D2 : Homeowners from across the study area who are younger will self report lower H urricane P reparedness Knowledge S cale (HPKS) than other homeowners Despite t he existence of HPKS scores between the different levels of Age ( p = .039) ; t he post hoc analysis was unable to determine between which A ge g roups these differences existed. T he plot of means indicated that the 25 34 year old homeowners reported the highest mean score compared to the other age groups (Figure 5 3) Based on this reasoning, the hypothesis of homeowners who are Younger reporting lower HPKS scores than other homeowners was not supported The finding of Age not being a significant factor of HPKS in Florida was surprising. Many studies from the literature review found Age to be a significant factor. Trumbo et al. (2014) found Age to be the only demographic characteristic that was a semi consistent predictor of hurricane risk perception and optimistic bias. Trumbo et al. (2014) also posited that those with the most experience were the most prepared. In some of the most convincing evidence, Sattler, Kaiser, and Hittner (2000) found gen der, age and income to be significant factors that could explain more than 2 0% of the variance of hurricane preparedness. Ge et al. (2011) reported that younger respondents tended to report a lower perceived hazard knowledge. This added to the thought that younger ages would report lower scores on the HPKS Baker (2011) also found Age to

PAGE 138

138 be a significant factor in hurricane preparedness and found that residents aged 30 to 45 to be the most prepared. This current study took this into consideration and incorr ectly linked age to experience. The finding of : 1) no significant differences between ages and HPKS scores in Florida and 2) significant differences in the study area, even though these differences were small enough to be undetectable by the post hoc anal ysis was confounding given the strong evidence from other studies. A ssumption s in this study might have be en incorrect based on the location of the study area; many homeowners move to the coastal areas of the US after spending years in other regions. Moreover as previously mentioned, the US is currently in a so there is a possibility that a large percentage of this sample has never personally experienced a tropic al storm or hurricane If the younger homeowners were life long residents of the area, this might also explain their reported high level of knowledge. This is briefly explored later in the Future Research section. Research Question 2: Trust in Support Entities To recap t to quantify t o what extent the same demographic characteristics ( Location, Education, I ncome and A ge ) could predict homeowne Trust in Support Entities Scale (TSES) The affect of each of these demographic character istics will first be explored within the state of Florida, then once again across the study area. Location of home H2 A1 : Homeowners located in the counties of P anhandle will self report lower T rust in Support E ntities S cale (TSES) than other regions within the state.

PAGE 139

139 Despite the existence of statistically significant differences of mean TSES scores between Florida Regions ( p = .003 Panhandle and other regions. Therefore, there was not support for the hypothesis that Panhandle would report lower TSES scores than other Regions within the state. This study found that South Florida ( n = 305, M = 33.82, SD = 7.83) reported significantly higher scores on the T SES than North Florida ( n = 163, M = 31.44, SD = 8.08) and the West Coast of Florida ( n = 412, M = 31.90, SD = 7.87) It is worth mentioning that Northwest Florida reported an even higher mean TSES score than test did not report this as significant. A little further research suggested that this might be due to the fewer number of responses in the Northwest region ( n = 56) compar ed to the other regions. There is a large military presence in Northwest Florida and thi s may have an influence on reported TSES scores in this area. Another factor that might influence a higher mean TSES score in Northwest Florida could be past experiences with support entities after past tropical storms and hurricanes. Both of these possibl e explanations attitudes towards support entities. H2 A2 : Homeowners located in Louisiana will self report lower T rust in Support E ntities S cal e (TSES) than other states within the study area. Despite the existence of statistically significant differences of mean TSES scores between States ( p < .001), there existed no difference between Louisiana and other states. Therefore, there was not support for the hypothesis that homeowners in Louisiana would report lower TSES scores than other States within the study area.

PAGE 140

140 Sterett (2015) discussed the work of NGO staff and volunteers post Hurricane Katrina in Louisiana She made mention of the suspi cion and distrust that the se case workers felt when suspected of working directly for FEMA. hypothesis about Louisiana reporting low scores on the TSES was largely based on their experience after Katrina. This current study found that homeowne rs in Southeastern Georgia ( n = 106 M = 29.39 SD = 7.9 7) reported significantly lower scores than Florida ( n = 1332, M = 32.68, SD = 7.90) and the combined area of coastal Alabama & Mississippi ( n = 55 M = 34.00 SD = 8.84 ) This migh t be partially explained using similar rationale as applied towards HPKS score by state If Georgia has experienced fewer hurricanes in the past, there may be fewer opportunities for homeowners to have positive or negative experiences with support entities. Conversely, A labama & Mississippi might have had positive experiences after Hurricane Katrina. Other factors that might influence the TSES scores could be political climate, transparency of local government, or reputation of support entities operating in each state Th e reputation of Craig Fugate (past Director and current head of FEMA) and then Bryan Koon (Current Director) at the helm of the Florida Division of Emergency Management might positively influence TSES scores in Florida Highest education level of homeowner H2 B1 : Florida homeowners attaining will self report lower T rust in Support E ntities S cale (TSES) than other homeowners within the state. TSES scores between the different levels of Education ( p = .090) Therefore the

PAGE 141

141 hypothesis of Floridian homeowners who attain reporting lower TS ES scores than other Floridian homeowners was not supported H2 B2 : Homeowners from across the study area attaining will self report lower T rust in Support E ntities S cale (TSES) TSES scores between the different levels of Education ( p = .070) Therefore the hypothesis of homeowners who attain reporting lower TSES scores than other homeowners was not supported increased education was related to increased trust in Federal and Local governments, yet decreased trust in State governments influenced these hypotheses. In asking about ederal government (FEMA), local government, and community; the decision to view increasing education as an indicator of increasing trust was made. The relationship between education and trust was also partially based on studies that have li nked perceived c ustomer trust to increased customer knowledge. For this study it appears that too much emphasis was placed on increased knowledge of a subject and education being synonymous. In retrospect, it is understandable how a formal education is not equivalent to increased subject knowledge in the operation and function of support entities. Household income H2 C1 : Florida homeowners with lower levels of household income will self report lower T rust in Support E ntities S cale (TSES) than other homeowners within the state. There were no statistically significant differences of Floridian TSES scores between the different levels of Income ( p = .999) Therefore the

PAGE 142

142 hypothesis of Floridian homeowners claiming lower Incomes reporting lower TSES scores than other homeowners was not supported H2 C2 : Homeowners from across the study area with lower levels of household income will self report lower T rust in Support E ntities S cale (TSES) than other homeowners There TSES scores between the different levels of Income ( p = 596 ) Therefore the hypothesis of homeowners of lower Incomes reporting lower TSES scores than other homeowners was not supported The rationale for generating these hypotheses was based on work such as that of Meyer et al. (2012) in the demographic indicators of trust in government health policy makers. Trust in government was found to be diminished in areas of lower income. Further review of literature found a study by Kim (2015) that suggested that trust in Federal and State governments decreased slightly with increasing income. This contradiction may be the result of slightly different composition in the various types of trust bein g studied. This current study found that income was clearly not an indicator of TSES score, especially in Florida ( p = .999) This finding was surprising at first, yet in finding conflicting support afterwards makes this easier to understand. This is indic ative of income not being related to TSES scores at all. Age of homeowner H2 D1 : Florida homeowners who are younger will self report lower T rust in Support E ntities S cale (TSES) than other homeowners within the state Despite the existence of statistically significant differences of Floridian mean TSES scores between levels of Age ( p < .001), the analysis

PAGE 143

143 indicated that the youngest group, 25 34 years, reported higher TSES scores than two of the older groups Therefore, there was not support for the hypothesis that Floridian homeowners who are Y ounger would report lower TSES scores than other Age Groups within Florida H2 D2 : Homeowners from across the study area who are younger will self report lower T rust in Support E ntities S cale (TSES) than other homeowners Despite the existence of statistically significant differences of mean TSES scores between Age Groups ( p < .001), the analysis indicated that the youngest gr oup, 25 34 years, reported higher TSES scores than two of the older groups. Therefore, there was not support for the hypothesis that homeowners who are Y ounger would report lower TSES scores than other Age Groups within the study area The premise for the se hypotheses came from studies suggesting the important influence that past experiences have on feelings towards an individual, organization, or entity. One explanation for this could be the combination of the study area being a desirable location to retire to and the years that have passed since a major hurricane Both of these could essentially negate the influence that previous experience may have on learned trust of support entities. Another issue of having a higher proportion of homeow ners moving to this area could be that they bring negative, or positive experiences with them from the locations which they arrived. Multiple Regression Analyses Implications Hurricane Preparedness Knowledge Scale (HPKS) Florida regions A m ultiple linear regress ion was calculated to predict Floridian HPKS score based on R egion of Florida and I ncome A significant regression equation

PAGE 144

144 was found ( F ( 2 1329 ) = 36.688 p < .001), with an adjusted R 2 of 051 Floridian homeowners predicted HPKS score s were equal to 35.12 0 + ( 1.029 [ REGION OF FLORIDA ]) + (0 697 [ INCOME ] ) where Region of Florida and Income values corresponded to Table 5 2 Therefore, t he base HPKS score was 35.12 0 and increased by a multiple of 1.029 depending upon the Region of Fl orida and increased 0 .697 for each increase in Household Income level Both variables were significant predictors of HPKS score. Study a rea HPKS score based on State and Income A significant regression equation was found ( F (2, 1940) = 19.419, p < .001), with an adjusted R 2 HPKS scores were equal to 37.036 + (0 .806 [ STATES ] ) + (0 .553 [ INCOME ] ) where States and Incomes values corresponded to Table 5 3. Therefore, the base HPKS score was 37.036 and increased by a multiple of 0.806 depending on the State and increased 0.553 for each increase in the Income level Both the States and Income variable s were significant predictor s of HPKS score. Trust in Support Entities Scale (TSES) Florida regions TSES score based on Age A significant regression equation was found ( F ( 3,1328 ) = 1 0 892 p < .001), with an adjusted R 2 of 0 22 Flo ridian homeowners predicted TSES scores were equal to 3 4 36 4 + (0 .58 1 [REGION] ) (0.54 3 [AGE]) (0.42 3 [EDUCATION]) where the Region Age and Education values corresponde d to Table 5 4 Therefore, the base TSES score was 34.36 4 and increased 0.58 1 for each

PAGE 145

145 increase in Florida Region decreased by 0.54 3 for each level increase in Age and decreased by 0.423 each level increase in Education The Florida Region, Age and Education variable s w ere significant predictor s of Floridian TSES score. Stu dy area TSES score based on State and Age A significant regression equation was found ( F (2, 1940) = 15.833, p < .001), with an adjusted R 2 TSES scores were e qual to 32.230 + (0 .830 [ STATE ] ) (0 455 [ AGE ] ), where States and Age values corresponded to Table 5 5 Therefore, the base TSES score was 32.230 and increased by 0.830 for each increasing State value and decreased by 0.455 for each increasing value of Age Both the State and Age variables were significant predictors of TSES score. provided as many questions as answers in this study. Were the demographic characteristics of this s tudy merely the wrong demographics for predicting TSES scores or are demographics in general not an indicator of these scores? Clearly Household Income did not lend any aid in predicting TSES scores, but perhaps race/ethnicity or marital status may provide insights. There also exists the possibility that demographics simply were not component s of the TSES scores. there are often difficulties in quantifying these affects, which leads to even more complication when determ ining predictors. V ariables that are tied to mental states, perceptions, or other intangible items may provide more definitive results.

PAGE 146

146 Descriptive Implications Through the analysis process the differences between groups were explored as well as the presence of predictors of the two scales developed ( HPKS and TSES ). Not all differences were found to be statistically significant These differences are still useful in discussing descriptive differences and may not be general ized beyond this sample. The differences that were found to be significant may be generalized to the target population. Tab le 5 6 describes generalizations of the homeowners scoring in the two extremes of the HPKS and TSES The areas that are appropriate t o generalize outside of the sample are identified. Summary of Implications An underpinning of the HPKS aspect of this study was the understanding that before homeowners can consider making a behavior change to be more prepared for natural disasters such as hurricanes, there must exist an awareness about preparedness. This skills and a spirations concerning hurricane preparedness. The knowledge aspect of awareness is simply learning more about hur ricanes as a weather disturbance, what types of preparedness products exist, and what preparedness best practices (behaviors) exist for homeowners. In addition to becoming more aware of these, homeowners can learn how to evaluate how effective each product or behavior could be in their specific circumstance. For example, learning ways to mitigate damage from storm surge is more of a priority to a homeowner living in low lying areas near a body of water than homeowners in other situations. des must involve a degree of positive self efficacy towards preparedness before any behavior changes are initiated as well. In other words, if

PAGE 147

147 homeowners do not understand and then also believe that they are capable of successfully making a change towards becoming better prepared for a natural disaster, then the process will never begin in earnest and they are destined to fail before they have begun. The skills that must be learned by homeowners as they become more prepared for hurricanes involve a compone nt of the knowledge aspect mentioned earlier. Homeowners learning that a specific preparedness behavior exists is the first step, then becoming more confident in the effectiveness of the behavior results from observing the behavior. Only after these steps can the homeowners then practice or role play the behavior to learn the skill. Sometimes all of these steps overlap and occur quickly such as during training events or at home shows demonstrating new products and behaviors In these situations it is often difficult for homeowners to determine when and how they successfully gained knowledge and confidence in the effectiveness of a product or behavior or even when they themselves gained a higher level of self efficacy in demonstrating the produc t or behavior themselves, family members, and/ or their homes. Without these aspiration s behavior change s can not occur. It has been argued that this prerequisite of behavior change is the most critical because no matter how educated on a subject an individual becomes, without a desire to change there will be no change. This remains true for hurricane preparedness. The Trust in Support Entities Scale ( TSES ) was of interest to explore because this level of trust affects the influence that these entities exert on homeowners. This

PAGE 148

148 trust may be expresse and thus was a consideration when the scale was named. By bet ter understanding what factors influence and in turn aid in predicting areas or populations about diminished trust these support entities can more effectively target areas/populations for outreach programs. The areas/populations with high levels of trust might also serve as a targeted audience from which these entities can solicit donations, monetary or time volunteering. The aphorism tide lifts all boats may be applied towards programs that may work towards increasing HPKS scores and TSES scores in areas. It stands to reason that areas of high HPKS scores and high TSES scores pose the least resistance to efforts to increase the physical preparedness of homeowners as well as supporting efforts to increase the resilience of their community t o natural disasters. Future Research The hypotheses of this study, while not predominately supported provided many additional opportunities for future research This future research could attempt to increase understanding of why the study area and scales did not provide the support anticipated by the hypotheses or explore additional factors to increase the variance explained in the scale scores. Specific research questions may discuss w hat aspects of the HPKS measure caused education to not be significan t when it has been in past studies (Gee et al., 2011; Baker, 2011) A brief, additional literature review was performed that focused on possible explanations for the findings of several hypotheses not being supported During this literature review two ad ditional factors of interest were found in several prior studies concerned with natural disaster preparedness knowledge T he increased length of time that homeowners had lived in their community and the presence of minors were both

PAGE 149

149 suggested as indicators of increased disaster preparedness knowledge (Solis, Thomas & Letson, 2010) Both of these factors could be valid foci for future research in the area of hurricane preparedness knowledge. Further inspection of the survey instrument revealed that these fact ors were addressed during the data c ollection process. Time living in community The analyses of differences in HPKS scores based on Time Living in Community can be found in Appendix D. The analysis indicated that homeowners HPKS had indeed increased base d on the increase of Time Living in Community (Figure 5 5). The HPKS scores of h omeowners living in their community for fewer than 10 years were significantly lower than those living there 20 years or more. Coincidently, these findings aligned with the 10 year drought from major hurricanes. In other words, homeowners who lived in the study area for at least 10 years before the last major hurricanes reported significantly higher scores on the HPKS. T hese findings support the explanation for why Age was not f ound to be a significant indicator of HPKS scores in this current sample Minors present in household The analyses of differences in HPKS scores based on Minors present in HPKS did indeed increase based on the presence of Minors in the household. The mean HPKS score of households with Minors P resent was 1.06, 95% CI [0.18 to 1.94] higher than the mean HPKS score of households with No Minors This difference was statistically sig nificant, t (1046) = 2.36, p = .018.

PAGE 150

150 Policy Implications By quantifying hurricane preparedness knowledge in the HPKS scores, this study sought to learn more about what areas and populations were lagging in this area of awareness. Based on the findings of this current study low income areas of N orth Florida appear to be the popula tion that would most benefit from hurricane preparedness education programs. This finding would also suggest that this area might not be willing to embrace hurricane preparedness behavior change due to this lack of awareness. Much like North Florida was indicated as an area that would benefit from hurricane preparedness education programs, Southeastern Georgia appears to possess that same need. The proximity of these two areas would allow for a regional focus whether through Federal coordination or cooperatively between the state of Florida and Georgia Given the results of the analyses of the additional variables of Time Living in Community and Minors it wa s suggested that homeowners that are new to the study area, especially those without minors in their household would benefit the most from hurricane preparedness education courses. As previously mentioned, the populations reporting higher HPKS scores could be a beneficial resource for working with others. Soliciting residents to assist with delivering these programs that have been in the area more than 20 years who have minors living in their household could be the most effective. The results of the analys TSES scores would show entities such as FEMA, Red Cross, county e mergency management, and local governments where and who to target for outreach programs. These areas or

PAGE 151

151 populations of diminished support could be the result of a myriad of factors ranging from perceptions of these entiti es in general to past negative interactions with the entities whether personally experienced by the homeowners or by their friends or family members. Another benefit of learning more about areas an d populations of low or high HPKS and TSES scores could allow emergency management personnel to more efficiently deploy resources in times of crisis. Based on the findings of this current study, the low income areas of Southeastern Georgia would be an area of above average need if ever hit by a major hurricane. Therefore resources could be relocated from areas of lesser need depending on the predicted trajectory of an approaching hurricane. Further Limitations During the interpretation of the findings of this current study, it became clear that exploration of additional variables would be needed to explain the variance in HPKS and TSES scores to an acceptable level. The multiple linear regression models only used one to two of the four IVs and those variab les used produced an adjusted R 2 of 5.1% in the best model ( HPKS score of Floridian Homeowners). Conclusions The re is no denying the necessity of hurricane preparedness in the Southeastern US Even after more than a decade ha s passed the mention of Hurricane s Katrina Andrew, Ike, Wilma, Rita, Charley, Irene, Hugo, and Ivan evokes memories and images of the magnitude of destruction and devastation that these storms brought to coastal communities Public safety is the responsibility of emergency manag ers who are tasked with always maintaining a level of preparedness, even during time s when storms have

PAGE 152

152 The motiv ation for this study preparedness as well as their feeling s concerning their trust or confi dence in the entities that exist to pro vide support in times of crisis This will assist thes e emergency managers, homeowners, and government officials in maintain ing the priority of preparedness T he two scales introduced by this study: the Hurricane Preparedness Knowledge Scale (HPKS) and the Trust in Support Entities Scale (TSES) propose slight variations from previous scales in this field of research. Due to these differences and that they are previously untested scales, there is much to learn about them before they are fully understood. The usefulness of the HPKS and TSES is the ir ability to be similar to prior research, yet just different enough to provide a slight twist in perspective to the existing body of knowledge What has been learned in this study and what can be learned by further explor ing these scales adds depth and d iversity to wha t is currently understood which further exploration of existing scales could not provide. HPKS scores between groups segmented by the demographic characteristics of location education, income and age were all explored. R esponses across th e entire study area as well as the sub sample in Florida showed that HPKS scores differ significantly by Location and Household Income Among homeowners in the entire sample, HPKS scores were also observed to differ significantly by Age TSES scores between groups segmented by the demographic characteristics of location education, income and age were also explored.

PAGE 153

153 Homeowners in the study area as well as Floridian ho meowners detected statistically significant differences in their TSES scores by L ocation as well as Age T he intent of t to determin e if the selected HPKS scores. Thus, it was observed that the Region of Florida and Household I ncome were highly significant indicators of Floridian HPKS score s Despite their high level of significance, these indicators could only expla in a small percentage of the variance in HPKS scores in Florida That is to say, approximately 95% of the HPKS score s can be attributed to factors outside of this study. Even mirrored the Flori da analysis as much as Location and Household Income were highly significant indicators, they too could only explain a small amount of the variance in HPKS scores of homeowners from across the study area. These low percentage s of variance explained demonstrated the need for further exploration of additional indicators HPKS scores in the future. The second research question of this study explored the feasibility of the demographic characteristics as predictors of TSES scores This scale sought to quantify trust or confidence that homeowners held for support entities that provide aid during and after hurrica nes and other natural disasters. The analysis of Floridian homeowners Region of Florida Ag e and Education were all significant indicators of their TSES scores. As in the previous small portion TSES scores. When including

PAGE 154

154 the responses of homeowners in the entire study area Education was no longer a significant indicator, which left State and Age as the only indicators remaining in the reg ression model. These indicators explain ed less than 2 % of the varianc e in TSES scores. This low percentage of variance explained suggests that a large number of predictors remain unidentified. By finding the differences in and predictors of HPKS and TSES scores these results can be generalized to the homeowner s across the study area allowing policymakers and support entities to better target specific populations with tailored outreach programs. The tailoring of information deliv ery and marketing to targeted audience s has been widely studied and has been accepte d as more efficient and effective versus mass marketing campaigns. Specifically advancing the understanding in the areas of hurricane preparedness knowledge and trust in support entities, policymakers and support providers will be able to more effectively conduct public outreach to home owners. Homeowners who possess relatively high levels of knowledge and trust will be more open to not only increasing backing community measures to better protect the entire community.

PAGE 155

155 Table 5 1. Summary of hypotheses conclusions Hypothesis Abbreviated d escription Significance Determination H1A1 Florida Panhandle HPKS < Other Regions HPKS p < .001 Not Supported H1A2 Georgia HPKS < Other States HPKS p < .001 Supported H1B1 Floridian less than Bachelor's Degree HPKS < Floridian Bachelor's Degree or higher HPKS p = .270 Not Supported H1B2 Less than Bachelor's Degree HPKS < Bachelor's Degree or higher HPKS p = .082 Not Supported H1C1 Floridian lower income HPKS < Floridian upper Income HPKS p < .001 Supported H1C2 Lower income HPKS < Upper Income HPKS p < .001 Supported H1D1 Floridian younger HPKS < Floridian older HPKS p = .123 Not Supported H1D2 Younger HPKS < Older HPKS p = .039 Not Supported H2A1 Florida Panhandle TSES < Other Regions TSES p = .003 Not Supported H2A2 Louisiana TSES < Other States TSES p < .001 Not Supported H2B1 Floridian less than Bachelor's Degree TSES < Floridian Bachelor's Degree or higher TSES p = .090 Not Supported H2B2 Less than Bachelor's Degree TSES < Bachelor's Degree or higher TSES p = .070 Not Supported H2C1 Floridian lower income TSES < Floridian upper Income TSES p = .999 Not Supported H2C2 Lower income TSES < Upper Income TSES p = .596 Not Supported H2D1 Floridian younger TSES < Floridian older TSES p < .001 Not Supported H2D2 Younger TSES < Older TSES p < .001 Not Supported Table 5 2. Values assigned to Income and Florida Region variables for HPKS model 1 Income Value Florida Region Up to $14,999 0 North Florida $15,000 $24,999 1 West Coast $25,000 $34,999 2 Panhandle $35,000 $49,999 3 East Coast $50,000 $74,999 4 South Florida $75,000 $99,999 5 Northwest $100,000 $149,999 6 $150,000 or more 7

PAGE 156

156 Table 5 3. Values assigned to Income and States variables for HPKS model 2 Income Value States Up to $14,999 0 Georgia $15,000 $24,999 1 Florida $25,000 $34,999 2 Texas $35,000 $49,999 3 Alabama & Mississippi $50,000 $74,999 4 Louisiana $75,000 $99,999 5 $100,000 $149,999 6 $150,000 or more 7 Table 5 4. Values assigned to Florida Region Age, and Education for TSES model 1 Florida Region Value Age Education North Florida 0 25 34 years HS Diploma or less West Coast 1 35 44 years Some college up to AA/AS Panhandle 2 45 54 years East Coast 3 55 64 years South Florida 4 65 75 years Doctoral Degree Northwest 5 Table 5 5. Values assigned to Age and State variables for TSES Model 2 Age Value State 25 34 years 0 Georgia 35 44 years 1 Texas 45 54 years 2 Louisiana 55 64 years 3 Florida 65 75 years 4 Alabama & Mississippi Table 5 6. Descriptive generalizations HPKS s cores TSES s cores Indicators Lowest Highest Lowest Highest Florida Region North Florida* Northwest Florida* North Florida* Northwest Florida* State Georgia* Louisiana Georgia* Alabama & Mississippi Education HS Diploma or less Master's Degree Doctoral Degree HS Diploma or less Income Up to $14,999* $150,000 or more* Up to $14,999 $25,000 $34,999 Age 45 54 years 25 34 years 45 54 years 25 34 years Years in Community Less than 6 months* 20 years or more* Minors in Household No* Yes* indicates p < .05

PAGE 157

157 Figure 5 1. Hurricane return period (EPA, 2010) Figure 5 2. Mean HPKS score by Household Income

PAGE 158

158 Figure 5 3. Mean of HPKS scores by Age Figure 5 4. Mean of TSES scores by Region of Florida

PAGE 159

159 Figure 5 5. Mean HPKS score by Time Living in Community

PAGE 160

160 APPENDIX A DESCRIPTIVE STATISTICS OF THE SAMPLE Table A 1. Gender distribution of sample Frequency Percent Valid Male 771 39.7 Female 1,172 60.3 Total 1,943 100.0 Table A 2. Age distribution of sample Frequency Percent Cumulative Percent Valid 25 34 years old 249 12.8 12.8 35 44 years old 305 15.7 28.5 45 54 years old 403 20.7 49.3 55 64 years old 534 27.5 76.7 65 75 years old 452 23.3 100.0 Total 1,943 100.0 Table A 3. Education distribution of sample Frequency Percent Cumulative Percent Valid Did not complete High School 29 1.5 1.5 GED or High School Diploma 279 14.4 15.9 Some College 411 21.2 37.0 Technical Certificate 103 5.3 42.3 Associate Degree 216 11.1 53.4 Bachelor Degree 567 29.2 82.6 Master Degree 282 14.5 97.1 Doctoral Degree 56 2.9 100.0 Total 1,943 100.0 Table A 4. Income distribution of sample Frequency Percent Cumulative Percent Valid Up to $14,999 82 4.2 4.2 $15,000 $24,999 137 7.1 11.3 $25,000 $34,999 219 11.3 22.5 $35,000 $49,999 264 13.6 36.1 $50,000 $74,999 496 25.5 61.7 $75,000 $99,999 355 18.3 79.9 $100,000 $149,999 260 13.4 93.3 $150,000 or more 130 6.7 100.0 Total 1943 100.0

PAGE 161

161 Table A 5. Distribution of sample by state Frequency Percent Cumulative Percent Valid 1 .1 .1 Alabama 37 1.9 2.0 Florida 1,331 68.5 70.5 Georgia 106 5.5 75.9 Louisiana 121 6.2 82.1 Mississippi 18 .9 83.1 Texas 329 16.9 100.0 Total 1,943 100.0

PAGE 162

162 APPENDIX B DETAIL OF REVISED PRINCIPAL COMPONENTS ANALYSIS Table B 1. KMO and Bartlett's Test Kaiser Meyer Olkin Measure of Sampling Adequacy. .878 Bartlett's Test of Sphericity Approx. Chi Square 24579.652 df 406 Sig. .000 Table B 2. Total v ariance e xplained by PCA Component Initial Eigenvalues Rotation Sums of Squared Loadings Total % of Variance Cumulative % Total % of Variance Cumulative % 1 6.373 21.975 21.975 4.139 14.272 14.272 2 3.678 12.681 34.657 4.028 13.890 28.161 3 2.790 9.621 44.278 2.793 9.632 37.794 4 1.959 6.757 51.035 2.597 8.955 46.749 5 1.647 5.680 56.714 2.573 8.872 55.621 6 1.423 4.907 61.621 1.740 6.000 61.621 Extraction Method: Principal Component Analysis. Figure B 1. Screeplot of PCA component eigenvalues

PAGE 163

163 Table B 3. Rotated Component Matrix Component 1 2 3 4 5 6 Q4.52 .812 Q4.39 .794 Q4.37 .793 Q4.30 .767 Q4.22 .689 .323 Q4.45 .591 Q4.41 .754 Q4.38 .685 Q4.42 .322 .670 Q4.44 .663 Q4.49 .650 .328 Q4.13 .648 Q4.47 .647 Q4.12 .512 Q4.15 .782 Q4.27 .756 Q4.14 .733 Q4.16 .632 Q4.26 .422 .627 Q4.6 .908 Q4.7 .899 Q4.8 .882 Q4.56 .734 Q4.53 .664 Q4.46 .648 Q4.10 .590 Q4.48 .334 .517 Q4.1 .926 Q4.4 .911 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Component 1: Knowledge Q4.52. I know a lot about disaster planning and ways to prepare my home for natural disasters. Q4.39. How prepared for a natural disaster do you feel that your home is? Q4.37. How would you rate your knowledge of disaster preparedness? Q4.30. I have adequate information on what to do in the event of a natural disaster. Q4.22. I make time to plan and make physical preparations for natural disasters. Q4.45. What affect have on your future quality of life?

PAGE 164

164 Table B 4. Component 1: Inter item c orrelation m atrix Q4.52 Q4.39 Q4.37 Q4.30 Q4.22 Q4.45 Q4.52 1.000 Q4.39 .567 1.000 Q4.37 .657 .560 1.000 Q4.30 .683 .567 .593 1.000 Q4.22 .591 .512 .503 .534 1.000 Q4.45 .446 .506 .398 .461 .468 1.000 Table B 5 Component 1: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .875 .874 6 Component 2: Q4.41. How interested are you in making your home more prepared for a natural disaster? Q4.38. How important do you feel it is to prepare for disasters? Q4.42. How important is it to you to maintain the level of disaster preparedness of you r home? Q4.44. How do you feel that the quality of your life would be affected if your home was more prepared for a natural disaster? Q4.49. I wish I could financially afford to make my home more prepared for a natural disaster. Q4.13. I often think abo ut what might happen in the future. Q4.47. How would you feel if your home was more prepared for a natural disaster than Q4.12. I often think about what has happened in the past. Table B 6. Component 2: Inter item correlation m at rix Q4.41 Q4.38 Q4.42 Q4.44 Q4.49 Q4.13 Q4.47 Q4.12 Q4.41 1.000 Q4.38 .568 1.000 Q4.42 .552 .629 1.000 Q4.44 .467 .451 .478 1.000 Q4.49 .390 .260 .240 .307 1.000 Q4.13 .394 .354 .383 .359 .363 1.000 Q4.47 .469 .447 .441 .445 .306 .299 1.000 Q4.12 .314 .273 .299 .298 .248 .529 .223 1.000

PAGE 165

165 Table B 7. Component 2: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .821 .834 8 Component 3: Trust in Support Entities Q4.15 The local government (ex. County Emergency Management) can be depended on to assist during a crisis. Q4.27 My local government is prepared to handle a natural disaster. Q4.14 The federal government (ex. FEMA) can be depended on to assist during a cr isis. Q4.16 Non profit organizations (ex. Red Cross) can be depended on to assist during a crisis. Q4.26 My community has adequate resources to handle a natural disaster. Table B 8. Component 3: Inter item correlation m atrix Q4.15 Q4.27 Q4.14 Q4.16 Q4.26 Q4.15 1.000 Q4.27 .621 1.000 Q4.14 .493 .439 1.000 Q4.16 .406 .355 .409 1.000 Q4.26 .503 .618 .356 .296 1.000 Table B 9. Component 3: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .800 .803 5 Component 4: Q4.6 The gulf/ocean waters seem the same to me as they always have. Q4.7 The gulf/ocean shores seem the same to me as they always have. Q4.8 The gulf/ocean marine life seem the same to me as they always have. Table B 10. Component 4: Inter item c orrelation m atrix Q4.6 Q4.7 Q4.8 Q4.6 1.000 Q4.7 .809 1.000 Q4.8 .764 .773 1.000

PAGE 166

166 Table B 11. Component 4: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .915 .915 3 Component 5: Q4.56 I do not have the time to prepare my home for natural disasters. Q4. 53 Other members of my household are not interested in preparing our home for natural disasters. Q4.46 How would you feel if your home was less prepared for a natu ral disaster than Q4.10 I have traded goods/services for things that I need. Q4.48 Disaster preparation products are too expensive. Table B 12. Component 5: Inter i tem c orrelation m atrix Q4.56 Q4.53 Q4.46 Q4.10 Q4.48 Q4.56 1.000 Q4.53 .453 1.000 Q4.46 .435 .411 1.000 Q4.10 .286 .235 .266 1.000 Q4.48 .322 .283 .220 .167 1.000 Table B 13. Component 5: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .683 .690 5 Component 6: Q4.1 How long have you lived in this community? Q4.4 How long has your family (immediate or extended), lived in this community? Table B 14. Component 6: Inter i tem c orrelation m atrix Q4.1 Q4.4 Q4.1 1.000 Q4.4 .724 1.000

PAGE 167

167 Table B 15. Component 6: Reliability s tatistics Cronbach's Alpha Cronbach's Alpha Based on Standardized Items N of Items .814 .840 2

PAGE 168

168 APPENDIX C DETAIL OF OUTLIER LABELING RULE CALCULATIONS Table C 1. Hypothesis H1 A1 : HPKS by Florida R egion Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Northwest 37.00 44.00 49.75 12.75 2.20 8.95 77.80 Panhandle 35.00 40.00 47.00 12.00 2.20 8.60 73.40 North 33.00 38.00 45.00 12.00 2.20 6.60 71.40 West 34.00 38.00 46.00 12.00 2.20 7.60 72.40 East 35.00 41.50 48.00 13.00 2.20 6.40 76.60 South 37.00 43.00 49.00 12.00 2.20 10.60 75.40 Table C 2. Hypothesis H1 A2 : HPKS by State Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Georgia 32.00 36.00 44.25 12.25 2.20 5.05 71.20 Florida 35.00 40.00 47.00 12.00 2.20 8.60 73.40 Alabama & Mississippi 36.00 43.00 48.00 12.00 2.20 9.60 74.40 Louisiana 36.00 43.00 49.00 13.00 2.20 7.40 77.60 Texas 35.00 41.00 47.00 12.00 2.20 8.60 73.40 Table C 3. Hypothesis H1 B1 : HPKS by Education (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound HS Diploma or less 34.00 39.00 46.00 12.00 2.20 7.60 72.40 Some college up to AA/AS 35.00 41.00 47.00 12.00 2.20 8.60 73.40 Bachelor's Degree 35.00 41.00 48.00 13.00 2.20 6.40 76.60 Master's Degree 35.00 41.00 47.00 12.00 2.20 8.60 73.40 Doctoral Degree 34.00 40.00 48.00 14.00 2.20 3.20 78.80 Table C 4. Hypothesis H1 B2 : HPKS by Education Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound HS Diploma or less 34.00 39.00 46.00 12.00 2.20 7.60 72.40 Some college up to AA/AS 35.00 41.00 47.00 12.00 2.20 8.60 73.40 Bachelor's Degree 35.00 41.00 48.00 13.00 2.20 6.40 76.60 Master's Degree 36.00 41.00 48.00 12.00 2.20 9.60 74.40 Doctoral Degree 34.00 39.50 47.00 13.00 2.20 5.40 75.60

PAGE 169

169 Table C 5. Hypothesis H1 C 1 : HPKS by Income (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Up to $14,999 31.00 35.00 40.00 9.00 2.20 11.200 59.800 $15,000 $24,999 34.00 39.00 46.00 12.00 2.20 7.600 72.400 $25,000 $34,999 33.00 40.00 45.00 12.00 2.20 6.600 71.400 $35,000 $49,999 35.00 40.00 47.00 12.00 2.20 8.600 73.400 $50,000 $74,999 35.00 39.00 47.00 12.00 2.20 8.600 73.400 $75,000 $99,999 36.00 43.00 49.00 13.00 2.20 7.400 77.600 $100,000 $149,999 35.00 41.50 48.00 13.00 2.20 6.400 76.600 $150,000 or more 37.00 42.00 48.50 11.50 2.20 11.700 73.800 Table C 6. Hypothesis H1 C2 : HPKS by Income Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Up to $14,999 32.00 35.50 42.00 10.00 2.20 10.00 64.00 $15,000 $24,999 34.00 40.00 46.00 12.00 2.20 7.60 72.40 $25,000 $34,999 33.00 40.00 47.00 14.00 2.20 2.20 77.80 $35,000 $49,999 34.25 40.00 47.00 12.75 2.20 6.20 75.05 $50,000 $74,999 35.00 40.00 47.00 12.00 2.20 8.60 73.40 $75,000 $99,999 36.00 43.00 49.00 13.00 2.20 7.40 77.60 $100,000 $149,999 35.00 41.00 47.00 12.00 2.20 8.60 73.40 $150,000 or more 36.75 42.00 48.25 11.50 2.20 11.45 73.55 Table C 7. Hypothesis H1 D1 : HPKS by Age (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound 25 34 years 36.00 42.00 48.25 12.25 2.20 9.05 75.20 35 44 years 34.00 40.00 48.00 14.00 2.20 3.20 78.80 45 54 years 35.00 40.00 47.00 12.00 2.20 8.60 73.40 55 64 years 35.00 41.00 47.00 12.00 2.20 8.60 73.40 65 75 years 34.00 40.00 47.00 13.00 2.20 5.40 75.60 Table C 8. Hypothesis H1 D2 : HPKS by Age Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound 25 34 years 36.00 42.00 48.00 12.00 2.20 9.60 74.40 35 44 years 34.00 40.00 47.00 13.00 2.20 5.40 75.60 45 54 years 34.00 40.00 47.00 13.00 2.20 5.40 75.60 55 64 years 36.00 41.00 47.00 11.00 2.20 11.80 71.20 65 75 years 34.00 40.00 47.00 13.00 2.20 5.40 75.60

PAGE 170

170 Table C 9. Hypothesis H2 A1 : TSES by Florida Region Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Northwest 29.25 34.00 41.00 11.75 2.20 3.40 66.85 Panhandle 29.00 31.50 37.25 8.25 2.20 10.85 55.40 North 27.00 32.00 37.00 10.00 2.20 5.00 59.00 West 28.00 32.00 37.00 9.00 2.20 8.20 56.80 East 29.00 33.00 38.00 9.00 2.20 9.20 57.80 South 29.00 33.00 39.50 10.50 2.20 5.90 62.60 Table C 10. Hypothesis H2 A2 : TSES by State Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Georgia 25.00 30.00 34.00 9.00 2.20 5.20 53.80 Florida 28.00 32.00 38.00 10.00 2.20 6.00 60.00 Alabama & Mississippi 30.00 34.00 39.00 9.00 2.20 10.20 58.80 Louisiana 27.50 32.00 36.00 8.50 2.20 8.80 54.70 Texas 27.00 31.00 37.00 10.00 2.20 5.00 59.00 Table C 11. Hypothesis H2 B1 : TSES by Education (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound HS Diploma or less 29.00 33.00 38.00 9.00 2.20 9.20 57.80 Some college up to AA/AS 28.00 32.00 37.00 9.00 2.20 8.20 56.80 Bachelor's Degree 29.00 33.00 39.00 10.00 2.20 7.00 61.00 Master's Degree 27.75 31.00 37.00 9.25 2.20 7.40 57.35 Doctoral Degree 27.75 32.00 37.00 9.25 2.20 7.40 57.35 Table C 12. Hypothesis H2 B2 : TSES by Education Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound HS Diploma or less 28.25 32.00 38.00 9.75 2.20 6.80 59.45 Some college up to AA/AS 28.00 32.00 37.00 9.00 2.20 8.20 56.80 Bachelor's Degree 28.00 32.00 38.00 10.00 2.20 6.00 60.00 Master's Degree 27.00 31.00 36.25 9.25 2.20 6.65 56.60 Doctoral Degree 28.00 31.50 36.75 8.75 2.20 8.75 56.00

PAGE 171

171 Table C 13. Hypothesis H2 C1 : TSES by Income (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Up to $14,999 27.00 32.00 39.00 12.00 2.20 0.60 65.40 $15,000 $24,999 29.00 32.00 37.00 8.00 2.20 11.40 54.60 $25,000 $34,999 29.00 33.00 38.00 9.00 2.20 9.20 57.80 $35,000 $49,999 29.00 33.00 37.00 8.00 2.20 11.40 54.60 $50,000 $74,999 29.00 32.00 38.00 9.00 2.20 9.20 57.80 $75,000 $99,999 28.00 32.00 38.00 10.00 2.20 6.00 60.00 $100,000 $149,999 28.00 33.00 38.00 10.00 2.20 6.00 60.00 $150,000 or more 28.00 33.00 38.50 10.50 2.20 4.90 61.60 Table C 14. Hypothesis H2 C2 : TSES by Income Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound Up to $14,999 26.00 31.00 39.00 13.00 2.20 2.60 67.60 $15,000 $24,999 29.00 32.00 37.00 8.00 2.20 11.40 54.60 $25,000 $34,999 29.00 33.00 38.00 9.00 2.20 9.20 57.80 $35,000 $49,999 28.00 33.00 37.00 9.00 2.20 8.20 56.80 $50,000 $74,999 28.00 32.00 37.00 9.00 2.20 8.20 56.80 $75,000 $99,999 28.00 32.00 38.00 10.00 2.20 6.00 60.00 $100,000 $149,999 28.00 32.00 37.00 9.00 2.20 8.20 56.80 $150,000 or more 28.00 32.00 37.00 9.00 2.20 8.20 56.80 Table C 15. Hypothesis H2 D1 : TSES by Age (Florida) Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound 25 34 years 30.00 34.50 41.00 11.00 2.20 5.80 65.20 35 44 years 30.00 33.00 39.00 9.00 2.20 10.20 58.80 45 54 years 27.00 32.00 37.00 10.00 2.20 5.00 59.00 55 64 years 29.00 32.50 38.00 9.00 2.20 9.20 57.80 65 75 years 28.00 32.00 37.00 9.00 2.20 8.20 56.80 Table C 16. Hypothesis H2 D2 : TSES by Age Outlier Labeling Rule Q1 Mean Q 3 Range g Lower Bound Upper Bound 25 34 years 29.00 33.00 40.00 11.00 2.20 4.80 64.20 35 44 years 29.00 33.00 38.00 9.00 2.20 9.20 57.80 45 54 years 27.00 31.00 35.00 8.00 2.20 9.40 52.60 55 64 years 28.00 32.00 37.25 9.25 2.20 7.65 57.60 65 75 years 28.00 31.00 37.00 9.00 2.20 8.20 56.80

PAGE 172

172 APPENDIX D DETAIL OF HPKS SCORES BY TIME LIVING IN COMMUNITY Table D 1 Descriptive s tatistics of HPKS s cores by Time Lived in Community 95% confidence interval for m ean N Mean Std. d eviation Std. e rror Lower b ound Upper b ound Minimum Maximum Less than 6 months 25 36.92 12.17 2.43 31.90 41.94 6 57 6 months to 2 years 108 39.03 8.13 0.78 37.48 40.58 17 60 2 10 years 549 39.72 8.98 0.38 38.97 40.47 9 60 10 20 years 551 41.09 8.58 0.37 40.38 41.81 12 60 20 30 years 355 41.90 8.38 0.44 41.02 42.77 13 60 More than 30 years 355 41.97 9.14 0.48 41.02 42.92 12 60 Total 1943 40.84 8.85 0.20 40.45 41.24 6 60 The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) by Time L ived in C ommunity is shown in Table D 1 For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the in dependent variable (IV) is the Time L ived in C ommunity Homeowner responses were classified into six Time Lived in C ommunity categories and the mean HPKS scores increased in order from : Less than 6 months ( n = 25, M = 36.92, SD = 12.17), to 6 months to 2 years ( n = 108, M = 39.03, SD = 8.13), to 2 10 years ( n = 549, M = 39.72, SD = 8.98), to 10 20 years ( n = 551, M = 41.09, SD = 8.58), to 20 30 years ( n = 355, M = 41.90, SD = 8.38), to More than 30 years ( n = 355, M = 41.97, SD = 9.14) (Figure D 1 ).

PAGE 173

173 Figure D 1 Mean HPKS s core by Time Lived in Community For the one way ANOVA, some assumptions must first be verified. The first three assumptions are based on the design of the study instrument. These assumptions are a) the dependent variable (DV) is continuous, b) the independent vari able (IV) is categorical with two or more independent groups, and c) the cases (homeowner responses) are independent. All of these assumptions held true for this analysis. The last three assumptions are confirmed through statistical tests. The first of the se remaining assumptions was d) no significant outliers in the IV groups in terms of the DV. Of the 1,943 homeowner responses, seven outliers were found by comparing the extreme values of each category to the range produced by the outlier labeling rule. Th e HPKS score was significantly different for the different length of time in community categories, F (5,1937) = 5.987, p D

PAGE 174

174 2 ) revealed that the mean HPKS score of homeowners living in the community 6 months to 2 year s ( M = 39.03, SD = 8.13) and 2 10 years ( M = 39.72, SD = 8.98) were significantly lower than the mean scores of homeowners living in the community 20 30 years ( M = 33.76, SD = 7.44) and More than 30 years ( M = 34.70, SD = 8.66). Table D 2 oc of HPKS s cores by Time Lived in Community 95% confidence i nterval (I) Time Lived (J) Time Lived Mean d ifference (I J) Std. e rror Sig. Lower b ound Upper b ound Tukey HSD Less than 6 months 6 months to 2 years 2.11 1.95 0.889 7.67 3.46 2 10 years 2.80 1.80 0.627 7.93 2.33 10 20 years 4.17 1.80 0.186 9.30 0.95 20 30 years 4.98 1.82 0.069 10.17 0.21 More than 30 years 5.05 1.82 0.062 10.24 0.14 6 months to 2 years Less than 6 months 2.11 1.95 0.889 3.46 7.67 2 10 years 0.69 0.93 0.976 3.33 1.95 10 20 years 2.07 0.93 0.223 4.71 0.57 20 30 years 2.87 0.97 0.036 5.63 0.11 More than 30 years 2.94 0.97 0.029 5.70 0.19 2 10 years Less than 6 months 2.80 1.80 0.627 2.33 7.93 6 months to 2 years 0.69 0.93 0.976 1.95 3.33 10 20 years 1.37 0.53 0.099 2.89 0.14 20 30 years 2.18 0.60 0.004 3.89 0.47 More than 30 years 2.25 0.60 0.002 3.96 0.54 10 20 years Less than 6 months 4.17 1.80 0.186 0.95 9.30 6 months to 2 years 2.07 0.93 0.223 0.57 4.71 2 10 years 1.37 0.53 0.099 0.14 2.89 20 30 years 0.80 0.60 0.760 2.51 0.90 More than 30 years 0.87 0.60 0.689 2.58 0.83 20 30 years Less than 6 months 4.98 1.82 0.069 0.21 10.17 6 months to 2 years *2.87 0.97 0.036 0.11 5.63 2 10 years *2.18 0.60 0.004 0.47 3.89 10 20 years 0.80 0.60 0.760 0.90 2.51 More than 30 years 0.07 0.66 1.000 1.95 1.81 More than 30 years Less than 6 months 5.05 1.82 0.062 0.14 10.24 6 months to 2 years *2.94 0.97 0.029 0.19 5.70 2 10 years *2.25 0.60 0.002 0.54 3.96 10 20 years 0.87 0.60 0.689 0.83 2.58 20 30 years 0.07 0.66 1.000 1.81 1.95 The mean difference is significant at the 0.05 level.

PAGE 175

175 APPENDIX E DETAIL OF HPKS SCORES BY MINORS IN HOUSEHOLD Table E 1. Descriptive s tatistics of HPKS by Minors in h ousehold N Mean Std. d eviation Std. error m ean HPK S No Minors 1355 40.52 8.63 0.23 Minors Present 588 41.58 9.29 0.38 Total 1943 40.84 8.85 0.20 The descriptive statistics of the Hurricane Preparedness Knowledge Scale (HPKS) by Minors in Household are shown in Table E 1. For the test of this hypothesis, the dependent variable (DV) is the HPKS score and the independent variable (IV) is the Minors in Household variable Homeowner res ponses were classified into two responses: No Minors ( n = 1355 M = 40.52 SD = 8.63) and Minors Present ( n = 588 M = 41.58 SD = 9.29 ) and their mean scores are in that order. Independent samples t tests have similar assumptions to the one way ANOVA that were detailed in the Results and Analysis of Data chapter, but with a few differences. The first assumption is that the DV is a continuous variable; HPKS is indeed continuous. The second assumption is that the IV is dichotomous. This analysis used th e presence of minors in the household as the I V and this variable was coded as No Minors and Minors P resent The next assumption is that of independent observations and the data collection method of the dataset used meets this assumption. The next three assumptions relied on output from SPSS (v.22). Freedom from outliers was the next assumption and based on the outlier labeling rule there only existed one outlier. The next assumption was that the DV should be approximately normally distributed in each gro up of the IV. The Shapiro Wilk test for normality was significant at the p < .001 value for both groups, thus indicating normality. The last

PAGE 176

176 assumption is homogeneity of variance between groups. The assumption of homogeneity of variances was violated, as a est for equality of v ariances ( p = .013). Given this violation of assumption, the t test for equality of means values were gathered from the second row of values, or the values for equal variances not assumed. The mean HPKS score of h ouseholds with Minors Present was 1.06 95% CI [ 0.18 to 1.94] higher than the mean HPKS score of households with No Minors This difference was statistically significant, t (1046) = 2.36, p = .018. Table E 2. Independent samples t est of HPKS by Minors Levene's test for equality of v ariances t test for equality of m eans 95% confidence interval of the d ifference F Sig. t df Sig. (2 tailed) Mean d ifference Std. error d ifference Lower Upper HPK S Equal variances assumed 6.23 0.013 2.43 1941.00 0.015 1.06 0.44 0.20 1.92 Equal variances not assumed 2.36 1046.00 0.018 1.06 0.45 0.18 1.94

PAGE 177

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 177 APPENDIX F RESEARCH ROADMAP Research Questions Hypotheses Items Type of Analysis Variables RQ1: To what extent can demographic characteristics ( location, education, income and age reported score on this Hurricane Preparedness Knowledge Scale (HPKS) ? RQ1 A : To what extent can the location of homeowners affect their self Hurricane Preparedness Knowledge Scale (HPKS) ? H1 A1 : Homeowners located in the counties of Panhandle will self report lower Hurricane Preparedness Knowledge Sc ale (HPKS) than other regions within the state. FLReg 1 ( Northwest, Panhandle, North, East Coast, West Coast, and South Florida ) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA or test IV: FLReg 1 DV: HPKS H1 A2 : Homeowners located in Georgia will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other States within the study area. States_v2 2 (Alabama and Mississippi are combined due to low responses) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA or test IV: States_v2 2 DV: HPKS RQ1 B : To what extent can Education Level affect their self reported score on this Hurricane Preparedness Knowledge Scale (HPKS) ? H1 B1 : Florida homeowners attaining less will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners within the state. Education (limited to Florida) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA or test IV: Education DV: HPKS Florida, East Coast of Florida, West Coast of Florida, & South Florida. (Community Collaborative Rain, Hail & Snow Network, 2014) e xas, Louisiana, Georgia, Florida, and Alabama/Mississippi.

PAGE 178

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 178 Research Questions Hypotheses Items Type of Analysis Variables H1 B2 : Homeowners from across the study area attaining will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners Education 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA test IV: Education DV: HPKS RQ1 C : To what extent can household income Hurricane Preparedness Knowledge Scale (HPKS) ? H1 C1 : Florida homeowners with lower levels of household income will self report lower scores Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners within the state. Income (limited to Florida) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA test IV: Income DV: HPKS H1 C2 : Homeowners from across the study area with lower levels of household income will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners Income 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA test IV: Income DV: HPKS RQ1 D : To what extent can age affect their self Hurricane Preparedness Knowledge Scale (HPKS) ? H1 D1 : Florida homeowners who are younger will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners within the state. Age (limited to Florida) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA o test IV: Age DV: HPKS

PAGE 179

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 179 Research Questions Hypotheses Items Type of Analysis Variables H1 D2 : Homeowners from across the study area who are younger will self report lower Hurricane Preparedness Knowledge Scale (HPKS) than other homeowners Age 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) One way between groups ANOVA test IV: Age DV: HPKS RQ1: To what extent can demographic characteristics ( location, education, income and age ) predict reported score on Hurricane Preparedness Knowledge Scale (HPKS) ? Florida Specific Multiple Regression Model FLReg, Education, Income, Age (limited to Florida) 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) Multiple Regression Modeling IV: FLReg 1 Education, Income, Age DV: HPKS RQ1: To what extent can demographic characteristics ( location, education, income and age ) predict reported score on Hurricane Preparedness Knowledge Scale (HPKS) ? Study Area Multiple Regression Model States_v2, Education, Income, Age 4.22, 4.23, 4.30, 4.39, 4.40, 4.45, 4.52 (HPKS variables) Multiple Regression Modeling IV: States_v2, Education, Income, Age DV: HPKS a, East Coast of Florida, West Coast of Florida, & South Fl orida. (Community Collaborative Rain, Hail & Snow Network, 2014)

PAGE 180

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 180 Research Questions Hypotheses Items Type of Analysis Variables RQ2: To what extent can demographic characteristics ( location, education, income and age reported score on this Trust in Support Entities Scale (TSES) ? RQ2 A : To what extent can the location of homeowners affect their self Trust in Support Entities Scale (TSES) ? H2 A1 : Homeowners located in the counties of will self report lower Trust in Support Entities Scale (TSES) than other regions within the state. FLReg 1 ( Northwest, Panhandle, North, East Coast, West Coast, and South Flor ida ) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: FLReg 1 DV: TSES H2 A2 : Homeowners located in Louisiana will self report lower Trust in Support Entities Scale (TSES) than other states within the study area. States_v2 2 (Alabama and Mississippi are combined due to low responses) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: States_v2 2 DV: TSES RQ2 B : To what extent can educational attainment affect their self reported Trust in Support Entities Scale (TSES) ? H2 B1 : Florida homeowners holding less than will self report lower scores on this st Trust in Support Entities Scale (TSES) than other homeowners within the state. Education (limited to Florida) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: E ducation DV: TSES H2 B2 : Homeowners from across the study area holding will self report lower Trust in Support Entities Scale (TSES) Education 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA or test IV:Education DV: TSES a, East Coast of Florida, West Coast of Florida, & South Florida. (Community Collaborative Rain, Hail & Snow Network, 2014) ppi.

PAGE 181

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 181 Research Questions Hypotheses Items Type of Analysis Variables RQ2 C : To what extent can Household Income Trust in Support Entities Scale (TSES) ? H2 C1 : Florida homeowners with lower levels of household income will self report lower scores Trust in Support Entities Scale (TS ES) than other homeowners within the state. Income (limited to Florida) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: Income DV: TSES H2 C2 : Homeowners from across the study area with lower levels of household income will self report lower Trust in Support Entities Scale (TSES) than other homeowners. Income 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: Income DV: TSES RQ 2 D : To what extent can Age affect their self Trust in Support Entities Scale (TSES) ? H2 D1 : Florida homeowners who are younger will self report lower Trust in Support Entities Scale (TSES) than other homeowners within the state. Age (limited to Florida) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: Age DVs: TSES H1 D2 : Homeowners from across the study area who are younger will self report lower Trust in Support Entities Scale (TSES) than other homeowners. Age 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables One way between groups ANOVA test IV: Age DVs: TSES

PAGE 182

An Equal Opportunity Institution Barnett, R. V., Research Roadmap 101915 182 Research Questions Hypotheses Items Type of Analysis Variables RQ1: To what extent can demographic characteristics ( location, education, income and age ) affect reported score on Trust in Support Entities Scale (TSES) ? Florida Specific Multiple Regression Model FLReg, Education, Income, Age (limited to Florida) 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables Multiple Regression Modeling IV: FLReg, Education, Income, Age DV: TSES RQ1: To what extent can demographic characteristics ( location, education, income and age ) affect reported score on Trust in Support Entities Scale (TSES) ? Study Area Multiple Regression Model States_v2, Education, Income, Age 4.15, 4.26, 4.27, 4.14, & 4.16 TSES variables Mu ltiple Regression Modeling IV: States_v2, Education, Income, Age DV: TSES

PAGE 183

183 REFERENCES Alexander, D. E. (2013) Resilience and disaster risk reduction: an etymological journey. Natural Hazards and Earth System Sciences, 13 2707 2716. doi: 10.5194/nhess 13 2707 2013 hange during New Orleans visit. The Times Picayune Retrieved from http://www.nola.com/katrina/index.ssf/2015/08/jindal_tells_obama_its_a_mista.ht ml ASCE. (n.d.) Codes & Standards. American Society of Civil Engineers website. Retrieved from http://www.asce.org/codes and standards/codes and standards/ Baker, E. J. (2011) Household preparedness for the aftermath of hurricanes in Florida. Applied Geography, 31 46 52. doi:10.1016/j.apgeog.2010.05.002 Bandura, A. (1977). Self efficacy: toward a unifying theory of behavioral change. Psychological Review, 84 (2), 191 215. doi:10.1037/0033 295x.84.2.191 Bandura, A. (1986) Social foundations of thought and action: a social cognitive theory. Prentice Hall, Inc. Upper Saddle River, NJ. Journal of the Royal Statistical Society. Series B (Methodological), 16 (2), 296 298. Becker, J. S., Paton, D., Johnston, D. M., & Ronan, K R. (2012) A model of household preparedness for earthquakes: how individuals make meaning of earthquake information and how this influences preparedness. Natural Hazards 64 107 137. doi:10.1007/s11069 012 0238 x Bergstrand, K., Mayer, B., Brumback, B., & Zhang, Y. (2015) Assessing the relationship between social vulnerability and community resilience to hazards. Social Indicators Research, 122 (2), 391 409. doi:10.1007/s11205 014 0698 3 Blake, E. S., & Gibney, E. J. (2011) The deadliest, costliest and most intense United States tropical cyclones from 1851 2010 (and other frequently requested hurricane facts). NOAA/National Weather Service, National Centers for Environmental Prediction, National Hurricane Center. Retrieved from http://www.nhc.noaa.go v/pdf/nws nhc 6.pdf Blake, E. S., Landsea, C., & Gibney, E. J. (2007). The deadliest, costliest, and most intense United States tropical cyclones from 1851 to 2006 (and other frequently requested hurricane facts). NOAA/National Weather Service, National Ce nters for Environmental Prediction, National Hurricane Center.

PAGE 184

184 Bonanno, G. A. (2004) Loss, trauma, and human resilience: have we underestimated the human capacity to thrive after extremely aversive events? American Psychology, 59 20 28 Boon, H. J., Cott rell, A., King, D., Stevenson, R. B., & Millar, J. (2012) bioecological theory for modeling community resilience to natural disasters. Natural Hazards 60 381 408. doi:10.1007/s11069 011 0021 4 Boon, H. (2014) Investigation rural commun ity communication for flood and bushfire preparedness. Australian Journal of Emergency Management, 29 (4), 17 25 Boyer, D. (2015, March 23). FEMA targets climate change skeptic governors, could withhold funding. The Washington Times. Retrieved from http:/ /www.washingtontimes.com/news/2015/mar/23/fema targets climate change denier governors could/ Bronfenbrenner Center for Translational Research. (2014) Urie Bronfennbrenner. Cornell University College of Human Ecology. Retrieved from http://www.bctr.cornell.edu/about us/urie bronfenbrenner/ Bronfenbrenner, U. (1979). The Ecology of Human Development: Experiments by Nature and Design Harvard University Press. Cambridge, MA. Bronfenbrenner, U. (1989). Ecological systems theory. In R. Vasta (Ed.), Annals of child development (Vol. 6 pp. 187 249). Greenwich, CT: JAI Press. Bronfenbrenner, U. (1999) Environments in developmental perspective: theoretical and operational models. In S. L. Friedman & T. D. Wachs (Eds.), Measuring environment across the life span: emerging methods and concepts (pp. 3 28). Washington, DC: American Psychological Association Press. Bronfenbrenner, U. (2005) Making human beings human: bioecological per spectives on human development. Thousand Oaks, CA : Sage Brofenbrenner, U., & Morris, P. A. (1998). The ecology of developmental processes. In W. Damon & R. M. Lerner (Eds.), Handbook of Child Psychology, Vol. 1: Theoretical models of human development (6 th ed., pp. 793 828). New York: Wiley. Bryman, A. (2012) Social Research Methods, Fourth Edition. Oxford University Press. Oxford, UK. Cantrell, R. & Sewell, C. (2015). Decision Ade : An innovative process for segmenting U.S. homeowners by utility bill "bo theredness" and budget constraints Journal of Architectural Engineering doi:10.1061/(ASCE)AE.1943 5568.0000176

PAGE 185

185 Committee on Increasing National Resilience to Hazards and Disasters, Committee on Science, Engineering, and Public Policy, & The National Acad emies. (2012). Disaster Resilience: A National Imperative National Academies Press. Cutter, S. L., Ahearn, J. A., Amadei, B., Crawford, P., Eide, E. A., Galloway, G. E., ... & Zoback, M. L. (2013). Disaster resilience: A national imperative. Environment: Science and Policy for Sustainable Development 55 (2), 25 29. Cutter, S. L., Barnes, L., Berry, M., Burton, C., Evans, E., Tate, E., & Webb, J. (2008) A place based model for understanding community resilience to natural disasters. Global Environmental Ch ange 18, 598 606 Domingues, M., & Goncalves, C. E. B. (2014) Systematic review of the bioecological theory in sport sciences. Baltic Journal of Health and Physical Activity, 6 (2), 142 153 doi:10.2478/bjha 2014 0014 Donahue, A. K., Eckel, C. C., & Wilson R. K. (2014) Ready or not? How citizens and public officials perceive risk and preparedness. American Review of Public Administration, 44 (4S), 89S 111S. doi:10.1177/0275074013506517 Dunkleberger, L. (2005, August 6) Hurricanes put focus on Panhandle building exemption. The Gainesville Sun. Retrieved from http://www.gainesville.com/article/20050806/LOCAL/208060332 Environmental Protection Agency. (2010). Storm surge inundation and hur ricane strike frequency map. Water: Climate webpage. Retrieved from http://water.epa.gov/infrastructure/watersecurity/climate/stormsurge.cfm Federal Emergency Management Agency. (2015) About the Agency. Retrieved from https://www.fema.gov/about agency Florida Division of Emergency Management. Hazards Page, Hurricanes. FloridaDisaster.org. (n.d.) Retrieved from http://www.floridadisaster.org/EMTOOLS/Severe/hurricanes.htm Gasper, J. T., & Reeves, A. (2011). Make it rain? Retrospection and the attentive electorate in the context of natural disasters. American Journal of Political Science, 55 340 355. Glik, D. C. Eisenman, D. P., Zhou, Q., Tseng, C., & Asch, S. M. (2014) Using the Precaution Adoption Process model to describe a disaster preparedness intervention among low income Latinos. Health Education Research, 29 (2), 272 283. doi:10.1093/her/cyt109. GOES Pro ject. (2014). GOES Project FAQ. NASA GOES webpage. Retrieved from http://goes.gsfc.nasa.gov/text/goesfaq.html

PAGE 186

186 Gonzalez, H. & Barnett, M. A. (2014) Romantic partner and biological father support : associations with maternal distress in low income Mexican origin families. Family Relations, 63 371 383 doi:10.1111/fare.12070 Gorard, S. (2013) Research d esign: Creating robust approaches for the social sciences. Thousand Oaks, California : Sage Hal l, T., & Hereid, K. (2015) The freq uency and duration of US hurricane droughts Geophysical Research Letters 42 3482 3485 doi:10.1002/2015GL063652. Harding, J. F., Morris, P. A., & Hughes, D. (2015) The relationship between maternal Journal of Marriage and Family, 77 60 76 doi:10.1111/jomf.12156 Healy, A. J., & Malhorta, N. (2009). Myopic voters and natural disaster policy. Americ an Political Science Review, 103 387 406. (2013) Coastal flood damage and adaptation costs under 21 st century sea level rise. Proceedings of the National Academy of Sciences of the United States of America, 111 (9), 3292 3297 doi:10.1073/pnas.1222469111 Hoaglin, D. C., Iglewicz, B., & Tukey, J.W. (1986). Performance of some resistant rules for outlier labeling. Journal of American Statistical Association 81, 991 999. Hoaglin, D. C., & Iglewicz, B. (1987), Fine tuning some resistant rules for outlier labeling. Journal of American Statistical Association 82, 1147 1149. Huisman, O., & de By, R. A. (2009). Principles of geographic information systems: An introductory tex tbook. International Institute for Geo Information Science and Earth Observation Enschede, The Netherlands. International Federation of Red Cross and Red Crescent Societies. (n.d.) Preparing for disasters. Retrieved from http://www.ifrc.org/en/what we do/disaster management/preparing for disaster/ Kaiser, H. F. (1960). The application of electronic computers to factor analysis. Educational and Psychology Measur ement 20 141 151. Kaiser, H. (1970). A second generation little jiffy. Psychometrika, 35 (4), 401 415 doi:10.1007/BF02291817 Kaiser, H. (1974). An index of factorial simplicity. Psychometrika, 3 9 (1), 31 36 doi:10.1007/BF02291575 Kapucu, N., & zerdem, A. (2013) Managing emergencies and crises Jones & Bartlett Publishers : Boston

PAGE 187

187 Kim, Y. & Ball Rokeach, S. J. (2006) Civic engagement from a communication infrastructure perspective. Communication Theory, 16 (2), 173 197 doi:10.1111/j.1468 2885.2006.0026 7.x Kim, Y. & Kang, J. (2010) Communication, neighborhood belonging and household hurricane preparedness. Disasters, 34 (2), 470 488. doi: 10.1111/j.0361 3666.209.01138.x Miami He rald. Retrieved from http://www.miamiherald.com/news/state/florida/article12983720.html Kunreuther, H., Meyer, R., & Michel Kerjan, E. (2012) Overcoming decision biases to reduce losses from natural catastrophes. In Behavioral Foundations of Policy. Princ eton, NJ: Princeton University Press. Landsea, C. (2015) How many direct hits by hurricanes of various categories have Oceanographic & Meteorological Laboratory. Retrieved from http://www.aoml.noaa.gov/hrd/tcfaq/E19.html Leger D. (2012 August 21 ) Hurricane warning: Is USA becoming complacent? USA Today Retrieved from http://usatoday30.usatoday.com/weather/hurricane/story/20 12 08 14/hurricane prepared katrina/57192442/1 Lindell, M. K. & Perry, R. W. (2004) Communicating environmental risk in multiethnic communities. Thousand Oaks, CA : Sage Lindell, M. K., and Perry, R. W. (2012) The protective action decision model: theo retical modifications and additional evidence. Risk Analysis 32 (4) 616 632. doi:10.1111/j.1539 6924.2011.01647.x Mahoney, J. W., Gucciardi, D. F., Mallett, C. J., & Ntoumanis, N. (2014) Adolescent elopment: The utility of the bioecological model. The Sport Psychologist, 28, 233 244 doi:10.1123/tsp.2013 0050 Makoka, D. & Kaplan, M. (2005) Poverty and vulnerability. An interdisciplinary approach. Bonn : Universitat Bonn Manyena, S. B. (2006) The concept of resilience revisited. Disasters, 30 (4), 433 450. Masten, A. S., & Obradovic, J. (2008) Disaster preparation and recovery: lessons from research on resilience in human development. Ecology and Society, 13 (1). Retrieved from http://www.ecolo gyandsociety.org/vol13/iss1/art9/

PAGE 188

188 Maynard, B. R., Beaver, K. M., Vaughn, M. G., Delisi, M., & Roberts, G. (2014) Toward a bioecological model of school engagement: A biometric analysis of gene and environmental factors. Social Work Research, Advance Acces s, 1 13 doi: 10.1093/swr.svu018 Meyer, R. J. (2012) Failing to learn from experience about catastrophes: The case of hurricane preparedness. Journal of Risk and Uncertainty, 45, 25 50 doi: 10.1007/s11166 012 9146 4. Meyer, R. J., Baker, J., Broad, K., Czajkowski, J., & Orlove, B. (2014) The dynamics of hurricane risk perception: real time evidence from the 2012 Atlantic hurricane season. American Meteorological Society doi:10.1175/BAMS D 12 00218.1 Meyer, S. B., Mamerow, L., Taylor, A. W., Henderson, J., Ward, P. R., & Coveney, J. (2012). Demographic indicators of trust in federal, state and local government: implications for Australian health policy makers. Journal of the Australian Healthcare & Hospitals Association, 37 (1), 11 18. doi: 10.1071/AH1107 3 Miller, C. H., Adame, B. J., Moore, S. D. (2013) Vested interest theory and disaster preparedness. Disasters, 37 (1), 1 27. doi: 10.1111/j.1467 7717.2012.01290.x Morrow, B. H. (1999) Identifying and mapping community vulnerability. Disasters 23, 1 18. doi:10.1111/1467 7717.00102 National Aeronautics and Space Administration. (2015). What does NASA do?. Retrieved from http://www.nasa.gov/about/highlights/what_does_nasa_do.html N ational Hurricane Center. (2014). About the National Hurricane Center. Retrieved from http://www.nhc.noaa.gov/aboutintro.shtml National Hurricane Center (2014) Glossary of NHC t erms. National Ocean ic and Atmospheric Administration. Retrieved from http://www.nhc.noaa.gov/aboutgloss.shtml National Oceanic and Atmospheric Administration. (n.d.) About NOAA. NOAA US Department of Commerce. Retrie ved from http://www.noaa.gov/about noaa.html National Oceanic and Atmospheric Administration. (n.d.) NOAA Geostat ionary Satellite Server. NOAA US Department of Commerce. Retrieved from http://www.goes.noaa.gov/index.html National Oceanic and Atmospheric Administration. (2012) NOAA know s hurricane forecasting. NOAA US Department of Commerce. Retrieved from http://www.noaa.gov/factsheets/new%20version/hurricane_forecasting.pdf NESDIS. (2014). About NESDIS. NOAA Satellite and Information Service. Retrieved from http://www.nesdis.noaa.gov/about_nesdis.html

PAGE 189

189 Nidus, G. & Sadder, M. (2009) Learning from student work. Educational Leadership, 66 (5) Retrieved from http://www.ascd.org/publications/educational leadership/feb09/vol66/num05/Learning from Student Work.aspx Norris, F. H., Stevens, S. P., Pfefferbaum, B., Wyche, K. F., & Pfefferbaum, R. L. (2008) Community resilience as a metaphor, theory, set of capacities, and strategy for disaster readiness. American Journal of Community Psychology 41 127 150. Paek, H., Hilyard, K., Freimuth, V., Barge, J. K., & Mindlin, M. (2010) Theory based approaches to understanding public eme rgency preparedness: implications for effective health and risk communication. Journal of Health Communication: International Perspectives, 15 (4), 428 444. doi:10.1080/10810731003753083 Peacock, W. G., Brody, S. D., & Highfield, W. (2005) Hurricane risk perceptions among family homeowners. Landscape and Urban Planning, 73 120 135 doi:10.1016/j.landurbplan.2004.11.004 Reinhardt, G. Y. (2015). First hand experience and second hand information: Changing trust across three levels of governm ent. Review of Policy Research, 32 (3), 345 364. doi: 10.1111/ropr.12123 Rivera, F. & Kapucu, N. (2015). Hazards (pp. 43 55) Springer International Publishing. doi:10.1007/978 3 319 16453 3_4 Rogers, R. W. (1975) A protection motivation theory of fear appe als and attitude change. The Journal of Psychology: Interdisciplinary and Applied, 91 (1), 93 114 doi:10.1080/00223980.1975.9915803 Romo Murphy, E., & Vos, M. (2014) The role of broadcast media in disaster preparedness education: Lessons learned in the sc ientific literature 2002 2012. Media Asia, 41 (1), 71 85. Satija, N. (2014, July 14). Texas scientists bothered by climate change denying politicians. Texas Tribune. Retrieved from http://www.governing.com/news/headlines/gov texas scientists bothered by cli mate deniers.html Sattler, D. N., Kaiser, C. F., & Hittner, J. B. (2000). Disaster preparedness: Relationships among prior experience, personal characteristics, and distress. Journal of Applied Social Psychology, 30 (7), 1396 1420. Savoia, E., Lin, L., & Viswanath, K. (2013) Communication in public health emergency preparedness: a systematic review of the literature. Biosecurity and Bioterrorism: Biodefense Strategy, Practice, and Science. 11 (3), 170 184 doi:10.1089/bsp.2013.0038

PAGE 190

190 Smith, A. B., & Matthews J. L. (2015) Quantifying uncertainty and variable sensitivity within the US billion dollar weather and climate disaster cost estimates. Natural Hazards, 77 (3), 1829 1851 doi:10.1007/s11069 015 1678 x Solis, D., Thomas, M., & Letson, D. (2010). An empirical evaluation of the determinants of household hurricane evacuation choice. Journal of Development and Agricultural Economics, 2 (3), 188 196. Sterett, S. M. (2015). Disaster, displacement, and casework: Uncertainty and assista nce after Hurricane Katrina. Law & Policy, 37 (1 2), 61 92. doi:10.1111/lapo.12029 Taramelli, A., Valentini, E., & Sterlacchini, S. (2014) A GIS based approach for hurricane hazard and vulnerability assessment in the Cayman Islands. Ocean & Coastal Managem ent, 108, 116 130 doi:10.1016/j.ocecoaman.2014.07.021 Taylor Clark, K. A., Viswanath, K., & Blendon, R. J. (2010) Communication inequalities Health Communication, 25, 221 229. doi:10.1080/10410231003698895 T he Council for Excellence in Government. (2007) Are we ready? Introducing the public readiness index: A survey based tool to measure the preparedness of individuals, families and communities WhatsYourRQ.org, excelgov.org. Washington, DC Retrieved from http://www.fema.gov/media library data/20130726 1910 25045 5489/public_readiness_index.pdf Tierney, K., Lindell, M., & Perry, R. (2001). Facing the unexpected: Disaster preparedness and response in the United States. Washington, DC: Joseph Henry Press. T udge, J. R. H., Mokrova, I., Hatfield, B. E., & Karnik, R. (2009) Uses and misuses of Journal of Family Theory & Review, 1 (4), 198 210. doi:10.1111/j.1756.2589.2009.00026.x Tukey, J.W. (1977). Ex ploratory Data Analysis. Reading, MA: Addison Wesley. UN/ISDR (2004) Living with risk: a global review of disaster reduction initiatives. United Nations Inter Agency Secretariat of the International Strategy for Disaster Reduction, Geneva. Van Kessel, G. Gibbs, L., & MacDougall, C. (2015) Strategies to enhance resilience post natural disaster: a qualitative study of experiences with Australian floods and fires. Journal of Public Health, 37 (2), 328 336 doi:10.1093/pubmed/fdu051 Wamler, C. (2014). Cities disaster risk and adaptation. Routledge, New York, NY. Witte, K. (1992) Putting the fear back into fear appeals: the extended parallel process model. Communication Monographs, 59 (4), 329 349 doi:10.1080/03637759209376276

PAGE 191

191 BIOGRAPHICAL SKETCH Charles Bra dford Sewell attended the public schools of New Castle, Indiana. Immediately following high school, Brad attended Purdue University majoring in electrical engineering technology. After two unsuccessful semesters, he left Purdue and entered the workforce. After a few general labor jobs, Brad was hired as a retail assistant manager. At this time, he registered at Ivy Tech Community College of Indiana in Lafayette, Indiana and majored in computer aided design Family obligations did not allow for the management. After 13 years in management Brad was laid off and made the decision to return to Edison State College (currently Southwestern Florida Stat e College) in the spring of 2010 to complete his education. While at Edison, Brad earned an Associate of Arts degree and an Associate of Science in design and drafting. In the fall of 2011, Brad transferred to the University of Florida mic accomplishments grade point average, and honors thesis, he was selected as one of two outstanding two year scholars for the fall 2013 commenc ement. At this time Brad graduated Summa Cum Laude with a Bachelor of Science in sustainability and the built environment with a minor in urban and regional planning. As a graduate student, Brad studied community development and his research interest in homeowner perceptions and behavior led to his co developing of Decision Ade a process for segmenting US homeow ners In December of 2015 Brad earned a Master of Science in family, youth and community science s with a minor in entrepreneurship and a graduate certificate in geospatial analysis.