Citation
Modeling Vehicle Purchase Decisions

Material Information

Title:
Modeling Vehicle Purchase Decisions The Choice of Vehicle Type and Age
Creator:
Ankomah, Marian
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (45 p.)

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Civil Engineering
Civil and Coastal Engineering
Committee Chair:
SRINIVASAN,SIVARAMAKRISHNAN
Committee Co-Chair:
STEINER,RUTH LORRAINE
Committee Members:
STEINER,RUTH LORRAINE
Graduation Date:
8/8/2015

Subjects

Subjects / Keywords:
Adults ( jstor )
Automobiles ( jstor )
Children ( jstor )
Households ( jstor )
Modeling ( jstor )
Pickup trucks ( jstor )
Transportation ( jstor )
Travel ( jstor )
Trucks ( jstor )
Vehicles ( jstor )
Civil and Coastal Engineering -- Dissertations, Academic -- UF
age -- air -- emission -- quality -- road -- transport -- type -- vehicle
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Civil Engineering thesis, M.S.

Notes

Abstract:
This study examines the factors that influence the purchase of vehicles in the United States and its impact to the road transportation system. Road transportation is one of the major sources of air pollution in the United States and the world at large. The high emission of toxic substances by vehicles into the environment has a tremendous effect on the environment and the lives of people as well. In this paper, the data from the 2009 National household Travel Survey was studied to determine a households' vehicle type and age purchase decision. The MNL model and linear regression models were used to model the vehicle type and age purchase decision respectively. The results from the study showed that households with one other older automobile have a higher propensity to purchase an older automobile than households with multiple vehicles having at least one older automobile. The household income, life cycle classification of the household and the census location of the household by region all had a significant impact of the age of the various vehicle types. This finding is however, of great importance to help reduce emission by improving or amending the standards set for vehicle usage. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (M.S.)--University of Florida, 2015.
Local:
Adviser: SRINIVASAN,SIVARAMAKRISHNAN.
Local:
Co-adviser: STEINER,RUTH LORRAINE.
Statement of Responsibility:
by Marian Ankomah.

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Ankomah, Marian. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Classification:
LD1780 2015 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

MODELING VEHICLE PUR CHASE DECISIONS: THE CHOICE OF VEHICLE TY PE AND AGE By MARIAN OWIREDUA ANKOMAH A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE D EGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2015

PAGE 2

© 2015 Marian Owiredua Ankomah

PAGE 3

To my family

PAGE 4

4 ACKNOWLEDGMENTS I wish to acknowledge my advisor, Dr. Siva Srinivasan for his encouragement and suppor t, and for being a great mentor to me. I also thank my other committee member, Dr. Ruth Steiner and the entire transportation faculty at University of Florida. I am very grateful to my parents and brothers for their endless support, prayers, encouragement

PAGE 5

5 TABLE OF CONTENTS ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 6 LIST OF FIGURES ................................ ................................ ................................ .......... 7 ABSTRACT ................................ ................................ ................................ ..................... 8 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ...... 9 Motivation ................................ ................................ ................................ ................. 9 Safety ................................ ................................ ................................ ...................... 10 Emissions ................................ ................................ ................................ ............... 11 2 LI TERATURE REVIEW ................................ ................................ .......................... 13 Vehicle Type Models ................................ ................................ ............................... 13 Vehicle Age ................................ ................................ ................................ ............. 14 Summary and Conclusions ................................ ................................ ..................... 15 3 DATA ................................ ................................ ................................ ...................... 21 Data Assembly ................................ ................................ ................................ ........ 21 Data Descriptives ................................ ................................ ................................ .... 23 4 MODELING METHODOLOGY AND ANALYSIS ................................ .................... 31 Vehicle Type Purchase Decision Model ................................ ................................ .. 31 Vehicle Age Model ................................ ................................ ................................ .. 32 Variable Effects on the vehicle purchase model ................................ ..................... 36 Effect of Socio Economic Variables ................................ ................................ . 36 Residential Location ................................ ................................ ......................... 37 Vehicle Ownership ................................ ................................ ........................... 38 Variable Effects on the vehicle age model ................................ .............................. 38 Effect of Socio Economic Variables ................................ ................................ . 38 Residential location ................................ ................................ .......................... 40 Vehicle Ownership ................................ ................................ ........................... 40 5 SUMMARY AND CONCLUSION ................................ ................................ ............ 41 LIST OF REFERENCES ................................ ................................ ............................... 43 BIOGRAPHICAL SKETCH ................................ ................................ ............................ 45

PAGE 6

6 LIST OF TABLES Table page 2 1 Classification of vehicles over the years ................................ ............................. 16 2 2 Summary of Previous Studies on Vehicle Type Choice ................................ ...... 17 3 1 Characteristics of Households ................................ ................................ ............ 26 4 1 MNL model on vehicle purchase ................................ ................................ ........ 33 4 2 Vehicle Age Regression models ................................ ................................ ......... 34

PAGE 7

7 LIST OF FIGURES Figure page 2 1 Share of Vehicles , by Age, 2001 2012 ................................ ............................. 20 3 1 General Modeling Structure ................................ ................................ ................ 28 3 2 Age Distributio n of Vehicles Purchased in Last Year ................................ .......... 29 3 3 Conventional retail gasoline prices ................................ ................................ ..... 30 4 1 MNL structure for vehicle purchase model ................................ ......................... 35 4 2 Linear Regression structure for vehicle age model ................................ ............. 35

PAGE 8

8 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 MODELING VEHICLE PUR CHASE DECISIONS: THE CHOICE OF VEHICLE TY PE AND AGE By Marian Ankomah August 2015 Chair: Siva Srinivasan Major: Civil Engineering This study examines the factors that influence the purchase of vehicles in the United States and its impact to the road transportation system. Road transportation is one of the major sources of air pollution in the United States and the world at large. The high emission of to xic substances by vehicles into the environment has a tremendous effect on the environment and the lives of people as well. In this paper, the data from the 2009 National household Travel Survey was stu vehicle type and age p urchase decision . The MNL model and linear regression models were used to model the vehicle type and age purchase decision respectively. The results from the study showed that households with one other older automobile have a higher propensity to purchase an older automobile than households with multiple vehicles having at least one older automobile. The household income, life cycle classification of the household and the census location of the household by region all had a significant impact of the age of the various vehicle types. This finding is however, of great importance to help reduce emission by improving or amending the standards set for vehicle usage.

PAGE 9

9 CHAPTER 1 INTRODUCTION Motivation Vehicles play an important role in the life of Americans as it serves as a medium for commuting to work, school and other recreational purposes. In the US after the great recession, there has been a tremendous increase in the sales of vehicles. According to ca rs dealership published by the National Automobiles Dealers Association (NADA), about thirty three million cars both new and old were purchased in 2013 (NADA, 2014). Not surprisingly, household vehicle ownership is routinely used as a key predictor variab le in practically all travel forecasting models. However, substantial focus is placed on the total number of vehicles owned by the household with limited emphasis on the various attributes characterizing these vehicles. More recently, there has been increa sed interest in determining the number of vehicles by type (such as sedans, SUV, pick ups, and vans). To date, to our knowledge, there have been very minimal efforts in However, un derstanding the age and type of the vehicles that households choose to buy is important from the perspectives of improved forecasting of travel demand, safety, and emissions. Imani et al . (2014) show that there is a difference in how vehicles of different types and ages are used for different travel purposes. Some statistics from their paper show that, for individuals with cars who participated in shopping activity alone, the ratio of non workers to workers is 28.63% to 20.94% respectively. Also, for the w orkers, 4.63% of the individuals with cars participated in shopping activity with only family

PAGE 10

10 members with cars. Both workers and non workers participate in shopping activities alone by car. Lim (2012) has developed models of vehicle allocation which also show that, for households with only male workers, the male is more likely to be the main driver of automobiles whereas for households with only female workers, the males is more likely to be the main driver of vans. In addition, their models showed that females with children are more likely to be the main driver of Vans and SUVs. Also, female adults are more likely to be allocated to vehicles with higher miles per gallon. In general vehicles with larger annual mileage is more likely to be assigned to the male. These results underscore the need to be able to predict the choice of vehicle type and age as a precursor to predicting daily travel patterns. Safety Vehicle crashes still remains one of the leading causes of death in the United States and the wo rld at large. Past research shows that the model year and type of the vehicle has an impact on the severity of the crash even after controlling for other contributing factors. (NHTSA, 2013; Toy and Hammitt, 2003) Over the years, there have been many safety improvements such as automatic and anti lock braking systems, seat belts, airbags, backup camera, and night vision assistance and collision warning for different types of vehicles, which has steadily reduced the injury and fatality rate. It has been estim ated that the probability of crashing in 100,000 miles of driving has decreased from 30 percent in a model year 2000 car to 25 percent in a model year 2008 improvements between the 2000 and 2008 car fleets, the probability of escaping a vehicle crash without being injured has improved from 79 to 82 percent. According to the NHTSA report (NHTSA, 2013), the probability that a driver was fatally injured

PAGE 11

11 increased (1) as the age of the vehicle being driven increased, and (2) in vehicles of earlier model year. Further, several studies have shown that the type of vehicles people drive has than cars. (Toy and Hammitt, 2003; Gayer, 2004; Anderson, 2008) From the paper by significantly different from that of automobiles. However, a driver faces a higher risk of se rious injury if the other vehicle is a van or pickup instead of a car. In addition, in terms Truett (2000), SUVs generally have larger sizes and improved visibility a nd as such might be able to provide greater safety for its occupants than small cars but they also have higher center of gravity and so they are more likely to rollover than automobiles. Also, from 1980 through 1998, the proportion of rollover fatalities f or occupants of non SUV vehicles involved in single vehicle crashes is much lower (only about 45%) than that of small SUVs. In addition, a rollover type crash is one of the most severe type of crashes and light duty trucks (e.g. SUVs) are fourteen times mo re likely to rollover when struck in the side than automobiles. (Kockelman et al., 2002) Generally, both the vehicle type and age have a significant on safety in transportation. Emissions Although emissions from on road motor vehicles in the United States have decreased significantly over the past four decades even with increase in traffic volume (Vijayaraghavan, Krish, et al., 2012), road transport still tends to be one of the major sources of air pollution especially in the cities. Vehicle engines emit m any types of pollutants such as nitrogen oxides, carbon monoxide (CO), carbon dioxide (CO2),

PAGE 12

12 volatile organic compounds (VOCs), sulfur dioxide (SO2) and lead. The MOVES (motor vehicle emissions simulator) model (US EPA, 2010), which is used for estimating emissions, uses vehicle age as one of the major inputs of the emission model. Also older vehicles are known to emit more pollutants and greenhouse gases. (Harrington and McConnell, 2003) The EPA provides default age distribution for MOVES based on each veh each year therefore the MOVES calculate the age distribution in each year by using the age distribution of the base year in addition to the sales and scrappage information. Mov es covers a range of ages up to 31 years, with vehicles older than 30 years grouped in the last category for each of the 13 source types (US EPA, 2010). In summary, there is a critical need to forecast vehicle ownership decisions considering the type of v intage (age) of the vehicles purchases and not simply in terms of the total vehicle holdings of the household as it is being predominantly done today. This study contributes towards developing a vehicle purchase model considering both the type and age of t he vehicle using recent national level data from the United States. The rest of this document is organized as follows. Chapter 2 presents a synthesis of literature. Chapter 3 describes the data assembly process and Chapter 4 presents a short summary of nex t steps.

PAGE 13

13 CHAPTER 2 LITERATURE REVIEW This chapter gives a summary of previous studies on the modeling of vehicle type and age. It is useful to acknowledge that there is also a very vast body of literature on the modeling of total household vehicle owner ship (see for instance, Mannering and Winston (1985), Kitamura and Bunch (1990), Bhat and Purugurta (1998), and Potoglou and Susilo (2008)). We do not present a detailed discussion of these studies to focus our attention of the choices of vehicle type and age. In general, these studies report that the structure and dwelling size of the household in addition to its life cycle, the vehicle ownership of a household. Vehicle Ty pe Models Table 2 2 gives a summary of the vehicle type choice models It is interesting to note that while some of the earliest studies on vehicle type dates back to the 1970s, there has been a resurgence of ineptest in this topic only very recently. From the very early models, Lave and Train (1979) till now, the vehicle price as well as the income of the household have been the most common significant variable across all models. However, in recent years, the study by Choo and Mokhtarian (2004) showed how t he vehicle type choices. Table 2 2, which summarizes the vehicle types studied, contains the following information on each study; the data set used for the study, the methodology adopted and the results. The vehicle type classification adopted appears to be different across the studies although most of them distinguish at least by size (sedans, SUVs, Vans, etc.) Table 2 1

PAGE 14

14 shows the different classification of vehicle types from various disciplines (engineering, environmental, social sciences etc.). The NHTS classified vehicles into nine categories (automobile, van, SUV, pickup trucks, recreational vehicles, motorcycles, golf cart and others. EPA (2012) classified their vehicles into nineteen categories solely based on the prestige, compactness and various sizes of the cars, SUVs, Vans and pickup trucks . In this study, however, the choice variables for vehicle types will be based on the vehicle type coding from the 2009 NHTS vehicle file since the data used is from the NHTS. To a large extent, most of the models from the studies summarized in Table 2 1that were developed were MNL type models and their predicted variables are largely similar. Throughout all the studies, t he household income has a very significant impact on the choice of vehicle types. High income households have a higher propensity for SUVs and luxury vehicles. Another common finding is that households with children are more likely to choose larger vehicle s like van. Finally it is useful to note that all the models presented, model the overall vehicle fleet composition of the household and not specifically the purchase decision. Vehicle Age Disaggregate studies on the age of the vehicles purchases are ext remely few. A study by Bhat et al. (2009) classified vehicles based on two age categories: new vehicles (vehicles at least five years old) and old vehicles (vehicles older than five years) apart from classifying them by type. From this study, it is eviden t that medium and high income households are more likely to choose new SUVs, and less likely to choose old vans. Also, larger households prefer old vehicle types and employed household members have high preference for new vehicle types and low preference for

PAGE 15

15 large sedans and minivan. An interesting finding from this study is that seniors are more likely to use old station wagons, old vans. However, aggregate statistics on vehicle age show interesting trends worth further examination. According to the US Bureau of Statistics, From 2001 2012, the average age of vehicles sold were between 11 and 20 years. Also from Figure 2 1, it can be seen clearly that in recent years, there is a higher purchase of older vehicles to newer vehicles. From 2007 2012, the shar e of newer vehicles dropped by approximately 33% whiles that of vehicles that were 11 20 years old increased by 25%. Although household income was a high factor contributing to the purchasing power of different households, there could be more factors to be developed based on the personal, vehicle and household characteristics. In addition, the statistics provided by or newer vehicles based by all the different types of old and new vehicle classes. There are also a couple of other sources that give the general trend of vehicle age based on the various vehicle fleet but this study is not about the frequency distribution of the vehicle age. Summary and Conclusions The synt hesis of the pertinent research cited in this chapter shows how the ownership of the household and the vehicle type choice. Overall, there have been numerous studies o n vehicle type but extremely few on vehicle age. Therefore, this vehicle. This study uses the 2009 NHTS data to model the purchase decision for vehicle age and type.

PAGE 16

16 Table 2 1. Classification of vehicles over the years Reference Study Vehicle Classification/Sample Distribution EPA Consumer Vehicle Choice Model Documentation (2012) Nested logit used Prestige Two Seaters, Prestige Subcompact Cars, Prestige Compact cars a nd Small Station Wagons, Prestige Midsize Cars and Station Wagons, Prestige Large Cars, Two Seater, Subcompact Cars, Compact Cars and Small Station Wagons, Midsize Cars and Station Wagons, Large Cars, M inivans, Cargo/Large Passenger Vans, Small Pickup Trucks, Standard Pickup Trucks, Ultra Prestige Vehicles NHTS Vehicle File Public Use Codebook (2009) Automobile/car/station/wagon, Van (mini, cargo, passenger), SUV, Pickup Trucks, Other Trucks, RV, Motorc ycle, Golf Cart, Others Engineering Household Vehicle Type Holdings and Usage: An Application of the Multiple Discrete Continuous Extreme Value (MDCEV) Model (2006) Passenger Car, Sports Utility Vehicles (SUV), Pickup Truck, Minivan, Van School of Social Science and ITS A Disaggregate Model of Auto Type Choice Charles A. Lave and Kenneth Train (1979) Subcompact, Sports Cars, Subcompact A, Subcompact B, Compact A, Compact B, Intermediate, Standard A, Standard B, Luxury

PAGE 17

17 T able 2 2. Summary of Previous Studies on Vehicle Type Choice Reference Lave and Train (1979) Choo and Mokhtarian (2004) Study A Disaggregate Model of Auto Type Choice Charles A. Lave and Kenneth Train (1979) What type of vehicle do people drive? The role of attitude and lifestyl e in influencing vehicle type choice (2004) A Discrete Choice Model of Car Type Choice Data Set Household: random sample individual who bought new cars in seven cities in 1976 Vehicle: 1976 Automotive News Market Data 1904 residents in three neig hborhoods in San Francisco Web based survey research of 1622 consumers from an European metropolitan area Methodology Multinomial logit model with ten vehicle type choice classes Multinomial logit model Multinomial Choice Model with 12 vehicle classes. Model focuses on vehicle preferences Results Households are less likely to choose an auto class whose initial cost is high Increase in car weight has a much stronger positive influence on older people than y ounger ones Increase in the price of gasoline causes the profile of car choices to shift downward. Larger sizes are more likely to choose sub subcompact and subcompact A cars. Small: Stronger pro high density and weaker travelling freedom attitudes, loners and non workaholics Midsized: Those more organized or with a higher household income are more likely to drive mid sized cars. Large cars: Attitude, Perso nality, lifestyle, mobility, and travel liking characteristics are insignificant. Luxury: Stronger travel dislike, pro high densities are significant. Travel by air a lot is significant. Sports: Younger, Not workaholics, power hungers (sign ificant) high density attitude (significant) frustrated (insignificant) Minivan: calm attitude (significant) Sports/SUV's: age ( ) Large Cars: education ( ) Luxury/SUV's: Household Income (+) Results are similar to Choo and Mokhtaria paper. In addition to the driver and vehicle characteristics and demographics, the pre purchase information source, consumer Involvement and attachment usage showed some significance importance to certain vehicle types.

PAGE 18

18 Table 2 2 . Continued . Reference Bhat and Sen (2006) Bhat et al. (2009) Eluru et al. (2010) Study Household Vehicle Type Holdings and Usage: An Application of the Multiple Discrete Continuous Extreme Value (MDCEV) Model (2006) The impact of demogr aphics, built environment attributes, vehicle characteristics, and gasoline prices on household vehicle holdings and use (2009) A joint flexible econometric model system of household residential location and vehicle fleet composition/usage choices (2010) Data Set 2000 San Francisco Bay Area Travel Survey (BATS). Fuel economy guide from the US EPA and DOE 5 vehicle types: PC, SUV, Pickup Truck, Minivan, Van 2000 San Francisco Bay Area Travel Survey (BATS) .The EPA Fuel Economy guide, Consumer Guide 10 vehicle types categories 2 vehicle age categories: new vehicles (<=5 years) and old vehicles (> 5 years) 2000 San Francisco Bay Area Travel Survey (BATS) Methodology Multiple Discrete Continuous Extreme Value model. It models households with different vehicle types jointly with the annual miles of each vehicle type Joint model of Multiple Discrete Continuous Extreme Val ue model and Multinomial Logit Model (MDCEV MNL). This joint model, modeled the vehicle type/vintage, vehicle make/model and vehicle usage Joint GEV based logit Three different models are developed: 1) Residential location neighborhood type mo del, 2) vehicle type model 3) vehicle usage models Results High Income households have a higher propensity Minivan Children have a significant effect on the type of vehicle: Households with children<4 have a prefere Minivan 5 15years: Same results but such households prefer minivan more Households with young adults: Strongest desire for Minivan The more the number of people in a househ old the higher the propensity for Minivans Households in high dense areas usually Pickup Trucks Mostly Low Income households prefer vehicle types with low operating cos t MDCEV Model Results: Medium and high income households have higher preference for new SUVs, and low preference for old vans Households with children less than 16 prefer minivan. Households with children between 16 and 17 are less likely to use old vans. Seniors: more likely to use c ompact, midsize, and large sedans relative to coupe and subcompact sedans. More likely to use old station wagons, old vans and travel more by non motorized forms Household size: Larger households prefer old vehicle types. Employment count: more employed household members have high preference for new vehicle types and low preference for large sedans and minivan MNL model resul ts for vehicle make/model choice: Cost: Averagely, households prefer to purchase and use vehicle make/model that are less expensive. Households are less likely to use vehicles with high amounts of greenhouse gas emissions Vehic le type Model Results: In general, households are more likely to prefer sedans to other vehicle types and least likely to prefer vans Larger households are likely to acquire larger vehicles or choose not to acquire a v ehicle Households with children prefer larger vehicles Households with more workers are more likely to choose pickup trucks High Income Househ olds are more likely to choose SUV and coupe compared to the other vehicle types As land use density and land use mix increases, the likelihood of acquiring pickup trucks decreases Households in neighborhoods with enhanced tran sit accessibility are likely to acquire compact and large sedans compared to pickups, SUVs or vans

PAGE 19

19 Table 2 2 . Continued . Reference Konduri et al. (2011) Garikapati et al. (2014) Study A Joint Tour Based Model of Vehicle Type Choice and Tour Length (201 0) Characterizing Household Vehicle Fleet composition and Count by Type in Integrated Modeling Framework (2014) Data Set 2009 NHTS data 2008 2009 National Household Travel Survey Methodology Multinomial probit model for vehicle type choice Log linear regression for tour length Other simulation methods Two model specification were assumed: Tour length affects vehicle type choice and vice versa Joint model of Multip le Discrete Continuous Probit (MDCP) model and a Multivariate Count (MC) model Results Tour length affects vehicle type choice results: Individuals determines the distance to travel, and then chooses the type of vehicle dependent on the distance. P ickup Trucks is most preferred by males and vans are least preferred Older individuals prefer vans Households with more children are more like to use vans than cars Households in non urban areas are less likely to use large vehicle types Households with children were less likely to own multiple cars or multiple pickup trucks. High income households were more likely to own areas were more likely to own multiple pickup trucks. Households in rural areas were more likely to own multiple pickup trucks Households with more than two workers were more likely to own multiple cars. One person households were more lik ely to own more than one vehicle of any type and retired households were less likely to own multiple SUVs.

PAGE 20

20 Figure 2 1. Share of Vehicles , by Age, 2001 2012

PAGE 21

21 CHAPTER 3 DATA This chapter describes the data sources, the steps taken to prepar e the data for analysis and a statistical summary of data assembled. Data Assembly The data used in this study come from the 2009 National Household Travel Survey (NHTS), which was administered by the Federal Highway Administration (FHWA). The overall dat a assembly procedure is presented schematically in Figure 3 1. During the survey period that is from April 2008 to April 2009 the nation experienced a wide range in gas prices from 4.1 dollars per gallon to 1.6 dollars per gallon across all states. The che apest gas price the nation ever had since 2002 was the 1.6 dollars per gallon in December 2008. This significant decrease in the price of the fuel during the survey period could account for the reason why about one fourth of the households bought a vehicle within that year. A graph of the conventional gas prices over the years is shown in Figure 3 3. The 2009 NHTS dataset contains the household, person, vehicle, and daily trip data for 150,147 households interviewed (box 1). Fourteen states and six other re gional planning agencies purchased additional samples to help them develop models that accurately capture the behavior of the households at the regional dataset of 25,702 hous contains several socio economic characteristics of the household (such as size and composition, income, housing unit type) and variables describing the location of the household (such as the census region classification, population density).

PAGE 22

22 The vehicle file provides detailed characteristics (such as model year, type, purchase date, and fuel) of all vehicles in all the households surveyed (box 4). The percentages of vehicles that are motorcyc les, other trucks, recreational vehicles and golf cart were relatively minimal; therefore they were excluded from the data. Thus the vehicle types considered in this study are Automobile, Van, Sports Utility Vehicle (SUV) and Pickup Truck (box 5). The age of the vehicle was also retained from the survey. The focus of this analysis is on vehicle purchase decisions. The survey explicitly collected data on how long the vehicle was owned by the household. The, vehicles that were purchased within one year of t he survey were identified (box 6). For these households, the socio economic characteristics at the time of vehicle purchase are likely to be very similar to the socio economic characteristics at the time of the survey. Therefore, the data from the survey c an be used as predictor variables to model the purchase decision. The rest of the vehicles (i.e., those that were not bought in the last year retained in a separate file (box 7) . Few households that bought multiple vehicles within the last year of the survey were excluded. It is useful to note that 28% of all households in the survey sample bought a vehicle in the one year period before the travel survey. The vehicles that were n ot bought in the last year were aggregated to the household newest and oldest vehicles owned (box 8). We hypothesize that the characteristics of have an impact on the characteristics of the vehicle purchased in the past year.

PAGE 23

23 The characteristics of the chosen vehicle (box 6) and the characteristics of the usef ul to note that the 25,702 households in the national sample include 1591 household with no vehicles (box 10). The person file provides detailed characteristics (age, gender, education, employment, US born, and race) of all household members surveyed. Sin ce the vehicle purchase decisions are assumed to be made at the household level, certain person level variables were aggregated to the household level to determine new variables describing the household. Specifically, the education levels of all persons we re used to determine the overall highest level of education attained by anyone in the household. A outside the US. The household was classified into Caucasian if any member in the household was Caucasian. The aggregate person information (box 9) was also added into the household file (box 10). The aggregated person information also includes indivi The household with the aggregated vehicle and person information was subject to additional cleaning and consistency checks leading to a cleaned sample of 20,996 households. Data Descriptives A full descriptive of the explanatory variables available in the final file is presented in Table 3 1. This table presents descriptive on the life cycle composition of the households, the income distribution and the highest level of education attained by anyone in the household.

PAGE 24

24 O n average, there are 2.28 members per household and there is about one worker per household on average. Table 3 1 presents descriptive on the life cycle composition of the households and the income distribution. Table 3 1 also presents the details on the h ighest level of education attained by anyone in the household. In terms of dwelling units, a larger part (78.4%) of the households lived in single family dwellings and of these 96% owned their homes and the rest were renting. About 21.5% of the population lived in multi family housing units (62% owned and 38% rented). The remaining proportion of households (0.1%) lived in other housing units including mobile homes and dormitories. About two thirds of the households lived in urbanized areas. About 14% of th e households have adult non drivers. About 97% of the households had all members born in the United States and 90% of all households are Caucasian. Table 3 The average age of one othe r vehicles is about 7.7 years whereas the average minimum age of multiple other vehicles is about 5.8 years with its average maximum being 14.5 years. 5989 of the 20,996 households purchased a vehicle in the past year. Of these 3136 were automobiles, 1422 were SUVs, 409 were vans and the rest were pickups. The descriptive statistics of all the categorical variables for the four types of the vehicles are also shown in Table 3 1. For the continuous variables, the average population density is 2686 and the ho usehold sizes are 2.59, 3.32, 2.73, and 2.64 for the households that purchased auto, van, SUV, and pickup trucks respectively within the

PAGE 25

25 past year of survey. On average, there are 1.24, 1.15, 1.28 and 1.32 workers per household with an average vehicle age of 5.8, 5.79, 4.07, and 7.85 for the households that purchased auto, van, SUV and Pickup Trucks. Of these, the maximum age of the other vehicles are 9.38, 10.19, 7.90, and 8.98 years. The distribution of the age of the purchased vehicles by type is shown i n Figure 3 2. The figure shows that a significant percent of the vehicles purchased within the last 12 months of the survey were between the ages of 1 and 2 years. Many households purchased vehicles that were between the ages of 1 and 7. On average, househ olds purchased fairly new vehicles.

PAGE 26

26 Table 3 1. Characteristics of Households Full Sample Households bought a vehicle within last year All Auto Van SUV Pickup Characteristic Mean Mean Mean Mean Mean Life Cycle Classification One adult, no chi ldren 11% 8% 5% 6% 5% 2+ adults, no children 23% 26% 19% 27% 34% Adults with children age 5 or under 9% 9% 29% 15% 10% Adults with children age 6 15 11% 15% 18% 17% 15% Adults with children age 16 21 5% 11% 4% 7% 8% One adult retired, no children 14% 7% 3% 4% 4% 2+ adults, retired, no children 28% 24% 22% 23% 25% Adult non drivers 14% 6% 7% 4% 5% Household Income and Highest Education HH with income less than $25,000 22% 15% 17% 9% 13% HH with income between $25,000 and $49,999 28% 25% 29% 22% 30% HH with income between $50,000 and $74,999 18% 18% 19% 19% 22% HH with income between $75,000 and $99,999 14% 16% 17% 20% 17% Less than high school graduate 4% 2% 3% 2% 3% High school graduate 23% 21% 22% 17% 25% Some college or associate deg ree 29% 29% 31% 32% 34% Bachelor's degree 23% 26% 22% 26% 24% Graduate or Professional Degree Housing Type and Ethnicity 21% 22% 21% 24% 14% Single Family Housing Unit owned 75% 77% 78% 82% 81% Single Family Housing Unit rent 3% 4% 4% 3% 4% Multi Fam ily housing unit owned 13% 12% 12% 11% 12% Multi Family housing unit rented 8% 7% 6% 4% 3% HH ethnicity 90% 91% 93% 93% 95% HH members were born in U.S. 97% 96% 96% 97% 98% Residential location Household in urban/rural area 31% 32% 38% 35% 50% Northeast census region 19% 21% 15% 22% 15% Midwest census region 23% 25% 33% 24% 26% South census region 35% 34% 33% 32% 36% West census region 23% 21% 19% 22% 22%

PAGE 27

27 Table 3 1. Continued Full Sample Households bought a vehicle within last year All A uto Van SUV Pickup Characteristic Mean Mean Mean Mean Mean Other vehicles HH with zero vehicles 8% HH with zero other vehicles (ZeroOtherVeh) 4% 17% 11% 13% 5% HH has one other auto (OneOtherAuto) 23% 17% 25% 20% 23% HH has one other v an (OneOtherVan) 3% 4% 4% 3% 3% HH has one other SUV (OneOtherSUV) 6% 8% 10% 11% 14% HH has one other pickup truck (OneOtherPickup) 5% 11% 11% 14% 6% Households with multiple other vehicles, at least one is an auto (MultiOtherAuto) 37% 26% 19% 19% 27% Households with multiple other vehicles, at least one is a van (MultiOtherVan) 9% 6% 8% 5% 6% Households with multiple other vehicles, at least one is an SUV (MultiOtherSUV) 19% 14% 8% 12% 18% Households with multiple other vehicles, at least one is a pi ckup (MultiOtherPickup) 24% 18% 16% 16% 21%

PAGE 28

28 Figure 3 1. General Modeling Structure

PAGE 29

29 Figure 3 2. Age Distribution of Vehicles Purchased in Last Year

PAGE 30

30 Figure 3 3. Conventional retail gasoline prices

PAGE 31

31 CHAPTER 4 MODELING METHODOLOGY AND ANALY SIS In Chapter 3, we described the data sources, the cleaning process of the data, the variables chosen for this analysis and the data statistical summary. In this chapter, we further discuss the modeling methodology used for analysis. Two modeling approac hes are adopted: the multinomial logit (MNL) for the vehicle type purchase decision modeling and Linear Regression for the vehicle age modeling. Vehicle Type Purchase Decision Model An MNL model was developed to determine the factors that affect a househo choice of purchasing a vehicle or not purchasing a vehicle and the structure of this model is shown in Figure 4 1. For the households that were more likely to buy a vehicle, we wanted to model for the type of vehicle they would most likely buy. The NL OGIT discrete choice modeling software was used for this analysis and a step wise approach was used to obtain our best models. All the variables were considered as explanatory variables for the initial model specification with the exception of the choice v ariables. In the step wise approach, a model was developed with every single explanatory variable against the choice variables first to determine the significance of those variables, the insignificant variables in each utility function of the model was rem oved and the models were re run again. However there were a few exceptions, the explanatory variables whose t statistic were close enough to the t critical value at 95% confidence were kept to be tested for during the intermediate model specification. Th is was done until all the models had significant variables, that is, there was a strong correlation between the explanatory variables and the choice variables of the utility functions. The final model

PAGE 32

32 shown in Table 4 1 captures only the significant explan atory variables from the demographics and other factors of the household. Vehicle Age Model A linear regression model was developed for the vehicle age modeling. We pa rticular vehicle age after the household had chosen the type of vehicle. Also, some new interaction variables were created for the linear regression models by multiplying inter OneOtherAuto_MaxAge and the maximum age of households with one other automobile. In general, t he step wise approach was also used for this linear regression modeling and the structure of the vehicle a ge modeling is shown in Figure 4 2 with the results of the model shown in Table 4 2.

PAGE 33

33 Table 4 1. MNL model on vehicle purchase Variables Did not buy Auto Van Suv Pickup Coeff. T stat Coeff. T stat Coeff. T stat Coeff. T stat Coeff. T stat Cons tant 3.771 28.645 2.905 19.427 0.734 5.874 0.48 4.761 One adult, no children 1.043 11.683 0.224 2.036 2+ adults, no children 0.349 5.267 Adults with children age 5 or under 1.5 13.051 Adults with ch ildren age 6 15 0.242 4.707 Adults with children age 16 21 1.091 13.909 One adult, retired, no children 1.423 13.329 0.551 4.23 Adult non drivers 1.407 17.166 HH with income between $50,00 0 and $74,999 0.207 4.332 HH with income between $75,000 and $99,999 0.623 8.301 0.349 4.068 0.323 2.988 HH income greater than $100,000 0.991 12.798 0.287 3.441 0.689 6.473 Number of workers 0.451 15.902 0.135 3.915 0.564 3.882 Less than high school graduate 0.29 4.553 High School Graduate 0.324 4.236 0.335 3.862 0.343 3.507 Some college or associate degree 0.291 5.271 0.303 4.729 Profession al or Graduate Degree 0.225 4.599 Multi Family housing unit rented 0.328 2.242 HH is Caucasian 0.553 6.417 0.473 4.542 Household in rural area 0.495 10.943 0.516 9.103 South census region 0.402 4.079 West census region 0.124 2.869 0.482 4.295 Midwest Census region 0.401 3.858 OneOtherAuto 1.042 17.39 OneOtherVan 0.392 4.204 OneOtherSUV 0.211 2.97 0 .695 6.762 MultiOtherAuto 1.372 26.28 MultiOtherVan 1.182 17.171 0.373 1.925 MultiOtherSUV 1.184 22.358 0.234 2.435 MultiOtherPickup 0.548 6.781 Number of observations 20996 Log Likelihood, constant only 19530.584 Log Likelihood at convergence 17524.235

PAGE 34

34 Table 4 2. Vehicle Age Regression models Variables Auto Van SUV Pickup Truck Coeff. T stat Coeff. T stat Coeff. T stat Coeff. T stat Constant 0.367 0.74 5.923 11.036 0.428 1.126 1.302 2.549 Count of household members 0.247 2.576 One adult, no children 2.189 4.955 2+ adults, no children 1.49 5.088 Adults with children age 5 or under 2.663 6.624 Adults with children age 6 15 1.998 5.763 Adults with children age 16 21 2.822 7.388 1.229 2.684 Adults with children 1.605 3.283 2+ adults, retired, no children 1.611 2.668 1.06 3.621 HH with income le ss than $25,000 4.902 12.198 4.618 10.198 4.002 5.094 HH with income between $25,000 and $49,999 2.851 8.98 2.335 6.863 1.528 2.739 HH with income between $50,000 and $74,999 1.661 5.126 2.675 3.978 1.507 4.389 HH with income between $75,0 00 and $99,999 1.249 3.735 3.35 4.74 0.675 2.015 HH income greater than $100,000 5.083 7.138 1.358 2.121 Less than high school graduate 2.541 3.608 3.363 2.394 2.739 2.06 Single Family Housing Unit rent 1.747 2.99 Mult i Family housing unit owned 0.636 1.939 Multi Family housing unit rented 1.298 2.909 2.958 2.838 1.738 2.756 2.849 2.114 HH is Caucasian 0.84 2.227 Northeast census region 0.696 2.38 West census region 1.034 3.538 0.841 2.997 Midwest Census region 0.714 2.582 ZeroOtherVeh 0.93 2.416 2.24 1.986 OneOtherAuto_MaxAge 0.25 11.261 0.161 6.736 OneOtherSUV_MaxAge 0.331 9.145 0.162 3.958 OneOtherPickup_MaxAge 0.166 6.76 0.158 5.994 OneOtherVeh_MaxAge 0.214 6.314 0.398 12.123 MultiOtherAuto_MaxAge 0.171 11.64 0.038 2.17 MultiOtherVan_MaxAge 0.092 3.518 0.112 3.965 MultiOtherSUV_MaxAge 0.125 7.375 0.153 7.398 MultiOtherPickup_MaxAge 0.106 6.787 0.083 4.533 MultiOtherVeh_MaxAge 0.149 5.309 0.383 16.637 Sample size 3136 409 1422 1022 R square 0.244 0.270 0.226 0.294 Adjusted R square 0.238 0.256 0.217 0.287

PAGE 35

35 Figure 4 1. MNL structure for vehicle purch ase model Figure 4 2 . Linear Regression structure for vehicle age model

PAGE 36

36 Variable Effects on the vehicle purchase model Effect of Socio Economic Variables An adult whether retired or not that has no children in the household is more likely to not b uy a vehicle than an adult with children between the ages of 6 and 15 years. This may be that, adults with children within those age ranges are more concerned about the safety of their children as such believe that having a vehicle of their own would provi de the maximum safety for their children. On the other hand, households with adults and no children usually do not travel much therefore they are more likely to not buy a car if they already have one. In addition, households with children 16 to 21 years ar e more likely to purchase an automobile than households without children. Usually, children within those age ranges are considered as young adults and are eligible to even drive therefore households with such children are more likely to purchase a vehicle most probably an automobile since automobiles are usually cheaper than SUVs, vans and pickup trucks. Furthermore, households with an increase in the number of children under 5 years have a higher propensity to purchase a van. Vans are usually larger to con tain household with many children and are also fuel efficient compared to pickup trucks and SUVs and this could be a possible explanation to that effect. This result is also consistent with the findings by Eluru et al. (2010) Households with income between $50,000 and $75,00 0 are more likely to not buy a vehicle than households with income greater than $75, 000. Usually the higher income, the higher the propensity of the household to purchase a vehicle. Also, households with income more than $100,000 are mo re likely to purchase an automobile and less likely to buy pickup trucks than households with lesser income. A possible explanation to this is that households with such high income are more geared towards

PAGE 37

37 purchasing luxurious vehicles and automobiles have many luxurious vehicle options compared to pickup trucks, thus are more likely to purchase automobiles. households with the highest education being a professional degree is less l ikely to buy a vehicle compared to households with the highest form of education being an associate degree or high school graduate. This result was not consistent with my expectation and a possible reason for this effect could be that households with a pro fessional degree may have multiple vehicles as such may not have a desire to purchase another. There the highest education of an associate degree have a similar propensity to purchase an automobile like households with a high school graduate as the highest education. The racial background of a household motivates their choice in the purchase of vehicles. Caucasians are more like to purchase vehicles than households of other raci al groups. This finding could be as a result of the bias in the sample, that is the sample has about 90 percent Caucasians compared to other races. Residential Location Househol ds in rural areas are less likely to purchase automobiles compared to pickup trucks than households in urban areas. It was also interesting to find out that households in rural areas are less likely to not purchase a vehicle than households in urban areas. Generally one would expect households living in the urban areas to have access to public transportation easily than those in the rural areas, as such those households would not be keen on purchasing vehicles and will more likely have a lower desire to pur chase a vehicle. In addition, households in the south, Midwest and west

PAGE 38

38 regions of the U.S are more likely to purchase pickup trucks than those in the northeast region. Vehicle Ownership Households that already have one SUV are less likely to buy another vehicle compared to households that already have a van. Also, there is no significant difference households with multiple vehicles, at least one being a van or an SUV have a similar desire to not purchase another vehicle. If the household should purchase a vehicle, then households with at least one automobile have a lower propensity to purchase another automobile. It was interesting to find out that households that have mult iple vehicles at least one being a van are more likely to buy another van. Also, households with multiple other vehicles, at least one SUV are less likely to purchase another SUV compared to households with only one vehicle, which is an SUV. A possible exp lanation could be that households with multiple vehicles of different types prefer other vehicle types other than SUVs that they own. In general households with multiple other vehicles at least one pickup truck are less likely to purchase another pickup tr uck. A plausible reason for this is that households purchase pickup trucks usually for a specific function; therefore they would be less likely to buy another vehicle of the same time to perform the same function. Also pickup trucks are more expensive comp ared to many other vehicle types. Variable Effects on the vehicle age model Effect of Socio Economic Variables The positive coefficient of the household size variable on the households that want to purchase an SUV shows that larger households are more li kely to purchase older SUV. This result is consistent with the initial assumption made. All other things

PAGE 39

39 being equal, larger households usually have bigger responsibilities in meeting the basic needs of its members than smaller household sizes. Therefore those large household for basic household expenditures. The large positive coefficients on the life cycle variables for adults with no children and adults with children f or households that want to purchase automobiles, SUVs and vans show that those households are more likely to purchase older vehicles. It was interesting to see that the presence of children in a household played a significant role in determining if a hous ehold would buy a van newer van. Also, household with more than one retired adults with no children were more likely to purchase newer vans and SUVs. Income played a significant factor in deciding the age of the vehicle to buy. For the vehicle age of automobiles, SUVs and pickup trucks, although the coefficients of the income categories were positive, their parameter effects show that households with higher income are more likely to purchase newer vehicles than households with lower income. A possible reason for this is that, the more income a household has, the easier it is for that particular household to meet the basic needs of the family, and as such the more funds need, once the physiological, safety, and sense of belongingness needs are met, usually, households with higher income desire to boost their esteem and status as such, those househo lds are likely to purchase newer automobiles, SUVs or pickup trucks. On the other hand, low income households are more likely to purchase older automobiles, SUVs or pickup trucks due to their tight budget and basic household responsibilities. It

PAGE 40

40 was intere sting to find out that, households with income greater than $75,000 are more likely to buy old vans compared to households with income between $50,000 and $75,000. The high positive co automobiles, va ns and pickup trucks indicate that households with the highest education being less than a high school graduate are more likely to purchase older automobiles, vans and pickup trucks. This could be because, it is expected that that households with lowest ed ucation; less than a high school graduate are more likely to be on the low income spectrum therefore more likely to make the same decisions as the low income households. Households that rent single family homes have a higher propensity to purchase older au tomobiles. In addition, households that rent multi family types of homes are more likely to purchase older automobiles than households that own the homes. Residential location propensity to purchase newer automobiles than households in the west census region. Vehicle Ownership Generally, households with one other older vehicle type are more li kely to purchase an older type of that vehicle than households with multiple vehicles having at least one older type. For example, households with one other older automobile have a higher propensity to purchase an older automobile than households with mult iple vehicles having at least one older automobile.

PAGE 41

41 CHAPTER 5 SUMMARY AND CONCLUSION There is a vast body of literature on household vehicle ownership in terms of the total number of vehicles owned by a household and the ir vehicle type s but extremely few on vehicle age. Understanding the age and type of vehicles that households chose to buy is important for improving travel demand forecasting, safety and emissions thus the need to forecast vehicle ownership d ecisions considering the type or vintage (age) o f the vehicles purchases and not simply in terms of the total vehicle holdings . In this paper, a purchase decision model for vehicle type and age were dev eloped using the on samples) househo lds wit h household, personal an d vehicle characteristics. In this study, the vehicle types The study developed and MNL model to determine the various factors that affect sing a certain type of vehicle or not purchasing a vehicle and linear regression was used to model the age of the v arious types of vehicles chosen . The study established the relation between many explanatory variables that affect the type and age of a vehi cle. The most interested explanatory variables were the other vehicles in a household would affect the purchase decision of the households. Some notable results shows tha t households with multiple vehicles, at least one being a van or an SUV have a similar desire to not purchase another vehicle. Also, if the household should purchase a vehicle, then households with at least one automobile have a lower propensity to purchas e another automobile.

PAGE 42

42 Interestingly, the variable representing the racial background of the household was a significant factor in motivating the choice of a household to purchase a vehicle. Caucasians were more likely to purchase vehicles than households of other racial groups. Also, the racial background of a household could determine whether the household is more likely to purchase a newer automobile or older one. In addition, income, the life cycle classification of the household and the residential loc ation of the household had strong impacts on the choice of vehicle type and age. This study is however important in helping to determine what affects a choice in purchasing a vehicle and help make more stringent standards for vehicle usage in t he United States in order to help increase safety, reduce the rate of emissions and improve the air quality and the health of the people in general. Future research in vehicle purchase modeling should consider a wide fleet of vehicle types such as recreat ional vehicles and even electric and/or hybrid vehicles as the data in this study did not have a large sample of those vehicle types.

PAGE 43

43 LIST OF REFERENCES Anderson, Michael. "Safety for whom? The effects of light trucks on traffic fatalities." Journal of H ealth Economics 27.4 (2008): 973 989. Bhat, Chandra R., and Sudeshna Sen. "Household Vehicle Type Holdings and Usage: An Application of the Multiple Discrete Continuous Extreme Value (MDCEV) Model." Transportation Research Part B 40.1 (2006): 35 53. Bhat , Chandra R., Sudeshna Sen, and Naveen Eluru. "The impact of demographics, built environment attributes, vehicle characteristics, and gasoline prices on household vehicle holdings and use." Transportation Research Part B: Methodological 43.1 (2009): 1 18. Caserini, S., et al. "Impact of the Dropping Activity with Vehicle Age on Air Pollutant Emissions." Atmospheric Pollution Research 4.3 (2013): 282 9. Choo, Sangho, and Patricia L. Mokhtarian. "What Type of Vehicle do People Drive? the Role of Attitude an d Lifestyle in Influencing Vehicle Type Choice." Transportation Research Part A 38.3 (2004): 201 22. Consumer Vehicle Choice Model Documentation. Washington, DC: U.S. Environmental Protection Agency, Office of Transportation and Air Quality, Assessment an d Standards Division, 2012. Davis, Stacy C., and Lorena F. Truett. An Analysis of the Impact of Sport Utility Vehicles in the United States. No. ORNL/TM 2000 147. Oak Ridge National Laboratory, 2000. Eluru, Naveen, et al. "A joint flexible econometric mo del system of household residential location and vehicle fleet composition/usage choices." Transportation 37.4 (2010): 603 626. Gayer, Ted. "The Fatality Risks of Sport Utility Vehicles, Vans, and Pickups Relative to Cars." Journal of Risk and Uncertainty 28.2 (2004): 103 33. Garikapati, Venu M., et al. "Characterizing Household Vehicle Fleet Composition and Count by Type in Integrated Modeling Framework." Transportation Research Record: Journal of the Transportation Research Board 2429.1 (2014): 129 137. Glassbrenner, Donna. An Analysis of Recent Improvements to Vehicle Safety . No. DOT HS 811 572. 2012. Harrington, Winston, and Virginia D. McConnell. Motor vehicles and the environment . Washington, DC: Resources for the Future, 2003.

PAGE 44

44 Imani, Ahmadreza Fa ghih, et al. "A multiple discrete approach for examining vehicle type use for daily activity participation decisions." Transportation Letters 6.1 (2014): 1 13. Lim, Kwangkyun. Intra household interactions in social recreational activities and travel. Univ ersity of Florida, 2012. Kockelman, Kara Maria, and Young Jun Kweon. "Driver injury severity: an application of ordered probit models." Accident Analysis & Prevention 34.3 (2002): 313 321. Konduri, KC, et al. "Joint Model of Vehicle Type Choice and Tour Length." Transportation Research Record 2.2255 (2011): 28 37. Lave, Charles A., and Kenneth Train. "A Disaggregate Model of Auto Type Choice." Transportation Research Part A: General 13.1 (1979): 1 9. NADA Data Report, 2014. National Automobile Dealershi p Association. Web. Nov. 2014 National Highway Traffic Safety Administration. How Vehicle Age and Model Year Relate to Driver Injury Severity in Fatal Crashes. Traffic Safety Facts Research Notes : Traffic Safety Facts Research Notes;2013 ASI 7766 17.186; DOT HS 811 825., 2013. November, 2014) Pubic Use Codebook, Version 2.1. National Household Trave l Survey, 2009. Web. Nov. 2014. Ryan Pfirrmann Beyond the Numbers: Prices and Spending , vol. 3, no. 9 (U.S. Bureau of Labor Statistics, May 2014) Technical Guidance on the Use of MOVES2010 for Emission Inventory Preparati on in State Implementation Plans and Transportation Conformity. Washington, D.C.: U.S. Environmental Protection Agency, Office of Transportation and Air Quality, Transportation and Regional Programs Division, 2010. Toy, Edmond L., and James K. Hammitt. "S afety Impacts of SUVs, Vans, and Pickup Trucks in Two Vehicle Crashes." Risk Analysis 23.4 (2003): 641 650. Vijayaraghavan, Krish, et al. "Effects of Light Duty Gasoline Vehicle Emission Standards in the United States on Ozone and Particulate Matter." Atm ospheric Environment 60 (2012): 109 20.

PAGE 45

45 BIOGRAPHICAL SKETCH Marian Ankomah earned her Bachelor of Science degree in c i vil e ngineering from Youngstown State University in 2014. She received h er Master of Science degree in civil e ngineering in 2015 from t he University of Florida.