National Academies Press: OpenBook

Forecasting Statewide Freight Toolkit (2008)

Chapter: Chapter 6 - Forecasting Models

« Previous: Chapter 5 - Data Sources
Page 27
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 27
Page 28
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 28
Page 29
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 29
Page 30
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 30
Page 31
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 31
Page 32
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 32
Page 33
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 33
Page 34
Suggested Citation:"Chapter 6 - Forecasting Models." National Academies of Sciences, Engineering, and Medicine. 2008. Forecasting Statewide Freight Toolkit. Washington, DC: The National Academies Press. doi: 10.17226/14133.
×
Page 34

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

27 This Toolkit focuses on the five model classes for statewide freight forecasting listed in Section 4.0: the flow factoring method, the O-D factoring method, the truck model, the four-step commodity model, and the economic activity model. These model classes share many of the same components, differing from each other primarily in their organization and use of these components. The key differ- ences between the five classes also are described in this section. 6.1 The Direct Facility Flow Factoring Method Description As shown in Figure 6.1, the direct facility flow factoring method provides freight volumes on transportation system links such as roads, railroad tracks, and ports. The method re- quires information about the facility itself and some forecasts of the factors affecting the facility. Although flow factoring is often used in individual project planning, it neither provides overall system forecasts nor con- siders many factors important in freight forecasting. How- ever, the method may be appropriate for developing forecasts for special generators, such as ports, within a more complex model. The facility flow factoring method is used to rapidly apply existing data to determine one or several forecast volumes. Usually, the method is intended for short-term forecasts; many assumptions are needed to make it work effectively and its range of applicability is limited. Flow factoring is relatively simple, however, and commonly used by state departments of transportation across the United States. The method can be divided into two general classes: one that produces future estimates of flow on a facility based on applying growth fac- tors to the flow on that facility, and one that produces esti- mates of flow on a facility based on applying factors that account for the diversion of flow from that facility to other routes or modes. The flow factoring method relies on regression equations, which may be based on two methods: time series analysis and economic analysis. Time series analysis involves an examina- tion of the historic flows on a transportation facility, with only time as an indicator variable. Economic analysis uses economic variables as indicator variables to explain the his- torical facility flows. Both methods are described below. Time Series Analysis Time series analysis is a means of understanding data vari- ability over time. Because a time series model exclusively represents past events and relationships, it can be used to forecast the future as long as the future is expected to behave like the past. Time series analysis is particularly appropriate when the forecast is short term and insufficient time and resources exist to build and calibrate a behavioral model. Time series models can be used for modal, policy, and data considerations. A simple time series analysis fits a straight line to a series of annual observations of freight flows, such as annual tons shipped through a port. Many statistical software packages, such as SAS or SPSS, or even spreadsheet programs such as Microsoft Excel have regression features to develop equations that can be used to forecast future freight flows based on the observed data. Economic Analysis Economic analysis can be used to forecast changes in freight demand due to changes in the level of economic activity or related factors. Forecasting based on growth in economic factors is useful because it recognizes the fact that demand for freight transportation is derived from underlying economic activities. The economic analysis method relies on C H A P T E R 6 Forecasting Models

forecasts of changes in economic variables to estimate the corresponding changes in freight traffic. To simplify the approach for deriving forecasts of future freight traffic from economic forecasts, the demand for trans- port of a specific commodity is assumed to be directly pro- portional to an economic indicator variable that measures output or demand for the commodity. Consequently, growth factors for economic indicator variables, which represent the ratios of their forecast year values to base year values, can be used as the growth factors for freight traffic. Economic Analysis Process Economic analysis requires data or estimates of freight traffic by commodity type for a reasonably normal base year, as well as base year and forecast year values for the corre- sponding economic indicator variables. The basic steps involved in the process are as follows: 1. Select the commodity or industry groups that will be used in the analysis. This choice is usually dictated by the avail- ability of forecasts of economic indicator variables. Much of the available forecasts are by SIC code. 2. Obtain or estimate the distribution of base year freight traffic by commodity or industry group. If actual data on the distribution are not available, state or national sources may be used to estimate this distribution. For example, the U.S. Census Bureau’s VIUS provides information on the distribution of truck vehicle-miles traveled by commodity carried and industry group. 3. Determine the annual growth factor (AGF) for each com- modity or industry group as follows: AGF = (I2/I1)1/(Y2−Y1) 28 where I1 is the value of the economic indicator in year Y1 and I2 is the value of the economic indicator in year Y2. 4. Using the annual growth factor and base year traffic, cal- culate forecast year traffic for each commodity or indus- try groups as follows: Tf = Tb AGFn where n is the number of years in the forecast period. 5. Aggregate the forecasts across commodity or industry groups to produce the forecast of total freight demand. The most desirable indicator variables are those that meas- ure goods output or demand in physical units (tons, cubic feet, etc.). However, forecasts of such variables frequently are not available. More commonly available are constant-dollar measures of output or demand, employment, or, for certain commodity groups, population or real personal income. The following subsection describes the data sources for forecasts of some of these economic indicator variables. Data Sources of Economic Forecasts Analysts at state departments of transportation, MPOs, and other planning agencies may use several sources to obtain estimates of growth in economic activity, by geographic area and industry or commodity type. Many states fund research groups that monitor the state’s economy and forecast changes. For example, the Center for the Continuing Study of the California Economy develops 20-year forecasts of the value of California products by two- digit SIC code. The Texas Comptroller of Public Accounts develops 20-year forecasts of population for 10 substate regions and 20-year forecasts of output and employment by one-digit SIC code and substate region. A private firm pro- duces 20-year forecasts of output and employment in Texas by three-digit SIC code. Long-term economic forecasts also are available from two Federal agencies. At two-and-one-half-year intervals, the Bureau of Labor Statistics (BLS) publishes low, medium, and high 12- to 15-year forecasts of several economic variables – including real domestic output, real exports and imports, and employment – for each of 226 sectors generally correspon- ding to groups of three-digit SIC industries. Also, at five-year intervals, the Bureau of Economic Analysis (BEA) develops 50-year regional projections of population and personal income as well as employment and earnings by industry sec- tor. The BEA forecasts are published by state for 57 industries, and by metropolitan statistical area and BEA economic area for 14 industry groups. Short- and long-term economic forecasts are available from several private sources as well. The private firms use government and industry data to develop their own models Figure 6.1. The flow factoring method.

and analyses. One of the better known firms is Global Insights, formerly DRI/WEFA. Global Insights provides national, regional, state, Metropolitan Statistical Area (MSA), and county-level macroeconomic forecasts on a contract or subscription basis. Variables forecast include gross domestic product, employment, imports, exports, and interest rates. Global Insights also produces short-term (two-and-one-half to three-year) and long-term (20- to 25-year) industrial input and output forecasts for 250 industries (two-, three-, or four- digit SIC code). Industrial inputs include employment, en- ergy, and materials used in production. These input/output forecasts are updated semiannually. Price and wage indices also are forecast for 650 different industries. Case Studies and References Two case studies demonstrate the truck model: the Min- nesota Trunk Highway 10 Truck Trip Forecasting Model and the Heavy Truck Freight Model for Florida Ports. These are described in Sections 8.2 and 8.3, respectively. The Guidebook on Statewide Travel Forecasting discusses time series methods for direct forecasts of vehicular volumes on highway and for forecasting the inputs to four-step mod- els.5 Major emphasis is on ARIMA models and on growth factor methods. Examples are primarily for passenger car forecasting, but the methods are equally applicable to truck forecasting. The Guidebook also describes a linear regression model to forecast truck volumes on I-40 in New Mexico. Commercial truck traffic was found to be a linear function of the year, the U.S. disposable income, U.S. gasoline costs, and the New Mexico cost of residential construction. 6.2 The Origin-Destination Factoring Method Description As shown in Figure 6.2, the O-D factoring method can use the conventional mode split and assignment model compo- nents. The O-D factoring method uses an existing and factored O-D table of freight as input to mode split and assignment, rather than a table prepared by trip generation and trip distri- bution model components. The acquisition and factoring of commodity O-D tables is widespread. States and the FAF have generally used Reebie Associates’ TRANSEARCH database as the source O-D table. While some efforts have been made to use an O-D Matrix Estimation process, the required observations on links of exclusively freight vehicles are rare and the tables produced by this method are usually exclusively based on truck counts that include freight and nonfreight trucks. The Bureau of Transportation Statistics’ CFS, while available publicly, does not provide geography below the BEA’s Economic Areas, which are insufficiently detailed for statewide freight fore- casting. Based on certain economic indicators, some states have successfully disaggregated Commodity Flow Survey data to the county level, but this process is costly and time con- suming. By using the TRANSEARCH database and its rout- ing options, states can develop freight network assignments even in the absence of an existing state model network. This is particularly useful given the lack of rail assignment models. Growth rates applied to the existing O-D tables can be based on economic, employment, or other indicators of growth at the zonal level and are often developed by using 29 Figure 6.2. The original destination factoring method.

simple economic models. These zonal growth factors are usu- ally applied to tables through an iterative proportional fitting technique that balances production and attraction growth rates. This technique, known as “Fratar factoring,” is usually available in travel demand model packages (TP+, EMME/2, TransCAD). Many states and the FAF have purchased freight forecasts directly from consultants who produce the eco- nomic forecast and also update the O-D commodity tables. In most applications different growth factors are applied for each commodity. The choice of mode split depends on the availability of variables. Because mode split is usually most pronounced for distances over 500 miles, a source of impedances outside the state is needed. If a national network is used, the utility of travel between zones, such as times and costs, can be obtained from the model. Otherwise the national utilities must be acquired from other sources. The network assignment component depends on the avail- ability of other data and is not limited by the O-D factoring models. O-D factoring models can always use a rules-based assignment component, depending on the ability to convert a tonnage table to truck trips. They can also use a modal net- work assignment process that either excludes other automo- bile or nonfreight traffic or recognizes this traffic only as preloaded volumes. Depending on the availability of com- plete auto and nonfreight truck trip tables within a complete statewide model, the commodity freight trucks can be assigned simultaneously with these other tables to allow the analysis of congestion. Obtaining a Current O-D Freight Table In order to factor an O-D table of freight flows, there must be an existing table of freight flows. There are three means of obtaining existing freight O-D tables: • Acquire a trip table from a public or commercial source; • Develop a trip table from a survey of freight shippers, receivers, and/or carriers; or • Estimate a trip table from observed freight flows. In practice, acquiring a trip table from public sources, such as the CFS, or from private sources, such as TRANSEARCH, are the most practical options. Given the diversity of geographic and commodity cover- age, the cost for a state to conduct sufficient surveys to develop a statistically reliable and sufficiently detailed O-D table would be prohibitive. Such surveys more often are con- ducted to develop the parameters in other model steps. Estimating a trip table from observed freight flows involves the use of O-D Matrix Estimation techniques.10 The observed freight flows in most applications are truck volumes. Truck volumes rarely provide information on the contents of the truck or purpose of its trip. Only in instances in which 1) all trucks can be assumed to be carrying freight, and 2) a break- down by commodity is not desired can the method be used. In urban and suburban areas, freight trucks are only a portion of all observed trucks. According to the Federal Highway Administration’s FAF web site, the freight truck percentage of VMT varies from 1% to 6% by urban area, and the total truck percentage (including nonfreight trucks) ranges from 5% to 18% by urban area.11 Due to the limitations of surveys and Origin Destination Matrix Estimation techniques, most statewide O-D factoring methods acquire existing trip tables from public or private sources. Factoring the O-D Freight Table The existing O-D freight trip table can be assigned to trans- portation networks to produce estimates of existing facility flows. To produce future flows it is necessary to factor the table to obtain an estimate of O-D freight flows in a future year. The factoring of O-D tables through an Iterative Pro- portional Fitting or Fratar process is an established practice in transportation planning. In this class of models the differ- ence is the source of the growth factors and the party that does the factoring. A state transportation agency that has obtained economic growth factors that apply to specific industry or commodity origins and destinations may choose to factor the table itself. However, an agency that has obtained the factors from an economic model provided by a private firm may find it advantageous for that firm to factor the table as part of the economic model. Common Model Components Mode Split The O-D freight table can be processed by a mode split equation as described in Section 4.4. If network information is available to provide utilities for movement between origins and destination by mode, and the coefficients of a sophisti- cated mode split model have been developed or transferred from another setting, then that mode split model can be used as part of the O-D factoring class of models. Network Assignment The assignment of a trip table in an O-D factoring model can use a variety of options as described in Section 4.5. Rail, water, and air assignments typically follow the rules-based assignment process. The assignment of truck freight depends 30

on the availability of a highway network model from other sources. If no statewide highway network or other vehicle trip tables are available from a statewide travel demand model or other source, a rules-based assignment model is used. If a highway network is available or can be developed, a freight truck only assignment can be used. If a highway network and other vehicle trip tables are available, a multiclass assignment can be used. Case Studies and References Two case studies demonstrate the O-D factoring method: the Ohio Freight Model Case Study and the Freight Analysis Framework Case Study. These are described in Sections 8.4 and 8.5, respectively. Oklahoma Model – This model and forecast system were developed in 2000 by TranSystems Corporation. It is a con- ventional model based on Reebie TRANSEARCH data. Kentucky Corridor Model – This model was developed in 1997 by Wilbur Smith Associates.12 The network and base data were updated in 2001 by Wilbur Smith, without chang- ing the model methodology.13 6.3 The Truck Model Description As shown in Figure 6.3, truck models use the trip genera- tion and distribution model components to produce a table of truck trips and uses assignment model components to assign that table of truck trips. As truck models address only the single mode of trucks, they do not require a mode split component. Truck models usually attempt to account for all shipments of goods, including local delivery. Freight truck volumes, as freight is defined in most data sources such as the CFS and TRANSEARCH, dominate in rural areas between distant cities. Truck models that include local delivery are more use- ful for states with closely spaced or contiguous urban areas. For this reason, the sole example of a state truck model identified for inclusion in this Toolkit was developed for New Jersey, the most densely populated state in the nation. However the nation’s largest metropolitan planning organi- zation, the Southern California Association of Governments, is included as an example of a statewide truck model because it uses the same techniques and the region’s geographic and population size is greater than that of many states. Truck models are more commonly a component of urban travel forecasting models. Truck models obviously cannot analyze shifts between modes, since by definition they include only the truck freight mode. They are usually part of a comprehensive model which forecasts both passenger and goods movement and conse- quently use a simultaneous assignment of truck trips with automobile trips. Truck models follow a three-step process of trip genera- tion, trip distribution, and traffic assignment. The truck types often considered in a truck model are broadly classified into light, medium, and heavy trucks based on gross vehicle weight (GVW) ratings. Although weight-based, these classi- fications are loosely correlated to other defining characteris- tics of trucks, which are described in the Quick Response Freight Manual.3 • Light trucks are defined as vehicles with four or more tires and two axles, with a GVW of less than 16,000 pounds. • Medium trucks are defined as single-unit vehicles with six or more tires and two to four axles, with a GVW of 16,000 to 52,000 pounds. • Heavy trucks are defined as double-unit, triple-unit, or combination vehicles with five or more axles, with a GVW greater than 52,000 pounds. Using these definitions, medium trucks directly correlate to single-unit trucks collected in truck surveys and heavy trucks directly correlate to double- and triple-unit trucks. The truck counts do not usually separate light trucks from passenger cars and are sometimes estimated as part of pas- senger vehicle travel. 31 Figure 6.3. The truck model.

Common Model Components Trip Generation Trip generation components will produce daily truck pro- ductions and attractions using equations whose coefficients were developed based on local surveys or using parameters borrowed from other sources such as the Quick Response Freight Manual. Trip distribution is accomplished using a gravity model that recognizes that the friction factors from internal-internal and external-internal/external-internal trips will vary, reflecting the difference in average trip length between these types of trips. External-external trips are established based on surveys and factored independently. Trip Distribution In truck models, the trip distribution component follows the process described in Section 4.3. The geographic scope of the model area typically requires that external trips be dis- tributed differently than internal-internal trips. Light, medium, and heavy trucks are distributed from origins to destinations using the gravity model technique. This is the same distribution method used in any typical auto passenger model. The friction factors in the gravity model can be devel- oped from surveys or borrowed from other sources such as the Quick Response Freight Manual. Network Assignment Network assignment of the truck trips is based on the mul- ticlass equilibrium highway assignment described in Section 4.5. Multiclass assignment is possible because truck models almost always are used as part of a complete travel demand forecasting process. Case Studies Two case studies demonstrate the truck model: the New Jersey Truck Model Case Study and the SCAG Heavy Duty Truck Model Case Study. These are described in Sections 8.6 and 8.7, respectively. 6.4 The Four-Step Commodity Model Description As shown in Figure 6.4, the four-step commodity model most closely resembles the four-step urban travel demand model for passengers; both use the trip generation, trip dis- tribution, mode split, and assignment model components. The economic forecasts that serve as the basis for the four- step commodity model are not modified in response to the results of the model. Four-step commodity models and the more familiar four- step passenger models both require the development of a statewide network and zone structure. If a statewide passenger model exists, it is often used to provide the zone and network structure within the state. Since trip distribution and mode split for freight typically involves average distances of hundreds of miles, a skeletal freight network is typically appended to that statewide highway network. While commodity models can 32 Figure 6.4. The four step commodity model.

analyze the impact of changes in employment, modal utility, trip patterns, and network infrastructure, they usually do not account for increases in labor productivity, or the interaction between industries. If the commodity model is integrated with highway passenger trip tables in assignment, the different rout- ing procedures for large freight trucks can be accounted for by the use of passenger car equivalencies and separate volume- delay functions. Common Model Components Trip Generation The four-step commodity model includes a set of annual or daily trip generation rates or equations by commodity pro- viding annual or daily flows as functions of TAZ or county population and disaggregated employment data. Trip Distribution Four-step commodity models typically use gravity models for trip distribution. The commodity groups serve as trip purposes and are distributed separately. The unit of flow in the distribution table is typically annual tons shipped, inde- pendent of mode. The distribution of freight is to a national system of zones, recognizing the large average trip lengths that govern the development of friction factors. Mode Split Four-step commodity models may use any of the mode split models, developing highway modal utility information from their highway component and using this information to approximate the utilities by other modes even in the absence of other modal networks. Because developing mode split models for commodity models is very complex, a simple application of existing mode share or qualitative adjustments of mode shares using market segmentation or other ap- proaches may be used. If a mode split model is developed, it typically uses an incremental or pivot point method to vary existing mode shares. Commodity truck tonnage is converted to daily freight truck trips by applying payload factors. Commodity flow tonnage is converted to vehicles based on commodity-specific factors (tons per truck) developed from state-specific sources or from the national VIUS database. While conversion of rail freight to carloads is not commonly done, there are exceptions. Network Assignment There are a number of options for assigning a trip table. The rail, water, and air assignments typically follow the rules-based assignment process. The assignment of truck freight typically will use either a freight truck only or multiclass assignment model. Rules-based assignments typically are not used for freight trucks since a highway network must already be avail- able to create the zone-to-zone impedances needed for trip distribution and mode split. If highway network is available or can be developed but there are no passenger vehicle tables, a freight truck only assignment can be used. If a highway net- work and other vehicle trip tables are available, a multiclass assignment can be used. Case Studies and References Two case studies demonstrate the truck model: the Florida Freight Model Case Study and the Indiana Freight Model Case Study. These are described in Sections 8.9 and 8.8, respectively. The Wisconsin Freight Model is a four-step freight forecast- ing model. The purpose of the latest Wisconsin effort in freight forecasting was to determine the impact of new rail/truck in- termodal facilities on highway truck volumes and on railroad tonnage.14 The Wisconsin Department of Transportation prin- cipally used the method of factoring trip tables and traffic assignment to accomplish its impact analysis. Cambridge Systematics developed a complete freight fore- casting model as part of the Vermont Statewide Freight Study.15 This model follows a variation of the classic four-step model. 6.5 The Economic Activity Model Description As shown in Figure 6.5, economic activity models use the trip generation, trip distribution, mode split, and assignment model components to produce freight forecasts for trans- portation facilities. The economic forecasts that serve as input to economic activity models are modified as a result of the per- formance determined by the model. Since the performance of the highway/truck freight system depends on the demand and usage of passenger autos, freight economic activity models are usually integrated with passenger forecasting models. Economic activity models are the freight equivalent of integrated land use-transportation models used in urban pas- senger travel; both use an economic/land use model as a step before the traditional four steps. Economic activity models require special data concerning the availability of land and the rules governing the development and location of certain industries, and an understanding of the interdependencies of industries. This information is often unavailable to a state department of transportation, and is usually obtained in part- nership with a state economic development agency that can explicitly account for changes in labor productivity. 33

Economic activity models formulate the flows of com- modities between economic sectors and between zones. The key assumptions in the economic activity models are that the zonal employment or economic activity is not directly supplied to the model but is created by applying an eco- nomic/land use model. Economic activity models use a modeling technique known as a spatial I-O model. An I-O model distributes household and economic activity across zones, uses links and nodes of a transportation network to connect the zones and model the transportation system, and calculates transportation flows on the network. The spatial I-O model uses a land use component to generate and distrib- ute trips and a transport component to generate mode split and network assignments. The two sides of the model inform each other, resulting in a dynamic model, as shown in Figure 6.6. Economic activity models typically use an I-O structure to simulate economic transactions that generate transportation activity, identifying economic relationships between origins and destinations in the corridor. In future years, the spatial allocation of economic activity, and thus trip flows, is influ- enced by the attributes of the transport network in previous years. Thus, the model is dynamic both with respect to land use and transportation. The economic activity class of models differs from the four-step commodity class of models in that the former uses an economic/land use model to forecast zonal employment or economic activity prior to the trip generation step and do not change those forecasts as a result of the forecast perfor- mance of the transportation facilities. Case Studies Two case studies demonstrate the truck model: the Oregon Economic Activity Model Case Study and the Cross-Cascades Economic Activity Model Case Study. These are described in Sections 8.11 and 8.10, respectively. 34 Figure 6.5. The economic activity model. Economy and Land Use Transportation Component Structure of the economy Location of the activity Trip Generation Distribution Network Costs Mode split Network assignment Figure 6.6. Integrated model interaction.

Next: Chapter 7 - Performance Measures »
Forecasting Statewide Freight Toolkit Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB's National Cooperative Highway Research Program (NCHRP) Report 606: Forecasting Statewide Freight Toolkit explores an analytical framework for forecasting freight movements at the state level.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!