xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader ntm 22 Ka 4500
controlfield tag 008 s flunnn| ||||ineng
datafield ind1 8 ind2 024
subfield code a C01-00102
Factors that influence the non-discretionary work trip by bicycle using MSA-level data from the 1990 U.S. Census
b Center for Urban Transportation Research (CUTR)
Bicycle commuting--United States
University of South Florida. Center for Urban Transportation Research
Baltes, Michael R.
t Center for Urban Transportation Research Publications [USF].
Paper No. 96-0069 REPRINT Duplication of this reprint for publication or sale is strictly prohibited without prior written permission of the Transportation Research Board Title: Factors that Influence the Non-Discret ionary Work Trip by Bicycle Us ing MSA-Level Data from the 1990 U.S. Census Author(s ): Michael R. Baltes, Research Associate, Center for Urban Transportation Research (CUTR} University of South Florida, Tampa, Florida Transportation Research Board 75th Annual Meeting January 7-11, 1996 Washington, D.C.
ABSTRACT r ACTORS THAT INFLUENCE THE NON-DISCRETIONARY WORK TRIP BY BICYCLE USING MSA-LEVEL DATA FROM TH:: 1 990 U S. CENSUS Mic h ael R. Ba l tes Research :::enter ior U rban Tra n soonaiio n Research ( C U TR ) University of South Florida, Tampa, Florida Phone: ( 813) 974-9843 :::mai l : b a lt es@cutr eng.usi .edu THIS p aper explores an extens ive r ange of fact ors which might have some contributory i nfluence on the select i on of the b i cycle as the mode choice for the trip to work using MSA-Ieve l data extracted from the 1990 U S. Census The sample included 100 percent oitbe MSA S in the U S Regre ss ion a n alysis was employed t o provide insig h t i n t o tne con tributory factors. The dependent variable us e d i n t he analysis was the percentage of the modal split captured by bicycle for the trip t o work in each M S A. INTRODUCTION HISTORICALLY, in the U.S., little regard has been given to the bicycle-to-work element of the peak-period commute This in part, may be due to our on-going infatuation with the automobile which, in tum, nourishes our unwillingness to embrace other travel modes such as public wa l king and BICYCLING. In addition, i t is often pointed out that high l evels of automobile commuting can be explained by low fue l prices nominal cost or free park i ng at the trip end, and evol utionary land-use development o r iented towards travel by automobile. CONCOMITANT w i th increases i n veh icle travel demands, roads have become more congested (1) and energy usage (2) continues to increase Transportation planners and enginee r s a r e searching for new and i nnovative ways to combat t hese serious problems The spectrum of travel is very broad, however, and, unf ortunately, one particular commute mode B ICYCLING has received relatively l itt le prior i ty from transportat ion planners and engineers, and policy and decis i on-makers. CoMMUTING b y bicycling constitutes a small port ion of t h e total trips made in the U.S Komanoff and Roelofs estimate that, in the U.S. in 1990-1991, bicycl ists rode 5 8 to 21.3 b i llion miles annually, representing 0 .28 to 1.03 percent of tota l vehicle miles TEMPORAL data from the Nationwide Personal Transportation Survey (NPTS) may shed some l i ght on the Mure of b icycling as a commute alternative.(ID According to data from the NPT S, the perce n tage of indiv i duals in the Un i ted States who bicycled to work decreased from 0 .75 percent in 1983 to 0.72 percent i n 1990 This slight decrease suggests that the viability of the bicycle as a commute alterna tive has not lost ground with regard to its share of the moda l split that has long been dominated by the privately owned vehi c le (POV) Moreover according to NP T S data for this same t ime period, the
FACTORS THAT INFLUENCE BICYCLING To WORK oercentage oi indi viduals using a POV to get to work increased from 81.78 oercent ic 37.35 oercent. a n increase of 5.57 percent. It is imoortant to point out that. even in light of the percentage increase in POV use over this same time irame the modal spiit share of trips by bicycle remained relatively sta t ic. MosT persons agree that it is desirable to encourage individuals to travel by bicycle more often, no matter what the trip purpose. It is believed that this would diminish various auto related problems such as congestion, pollution, and energy consumption ;. lessen the orospect of infrastructure expenses fo r auto-related facilities like roads and parking lots ; reduce persona l expendit l,!r es on auto-re l ated tr ansportation such as insurance, fuel, parking, and maintenance; and improve personal health and well-being. Cleariy commuting by bicycling to work, or for any other travel purpose, is regarded as an admirable and health-conscious pursuit.() LITERATUR: Review of bicycle-related literature indicates tha t there has been some heed given to the factors that may or may not influence individuals to commute to work by bicycle rather than by POV or public transit. As is well known, commuting by bicycling or any other mode to work, or for any other trip purpose, is a matter of personal choice and is dependent on many tangential factors that are both within and outside an individual's control. Although it appears that there has not been an examination at the metropolitan level (macro-level) of what factors may influence commuting by bicycle using 1990 Census data, there have been a few sp ecialized studies (micro level) that suggest some of the factors that influence an individual's choice to commute by bicycle IN a study of employees at six employment sites in the greater Seattle area Badgett, Niemeier and Rutherford found that excessive distance; unsafe streets; lack of sidewalks; inadequate trip end facilities such as showers and bicycle racks; the convenience, speed, and low cost of driving; the need to make multiple trips during the day; and a public perception that biking is not fashionable were all factors that deterred individuals from commuting by bicycle in the greater Seattle area.(Z) IN a study that looked at bicycle ownership and use in Amsterdam, Holland, Beck asked bicycle owners what their reasons were for choosing and not choosing to use their bicycle as a commuting option.@) Beck found that, in most cases, the three main r easons for choosing to use a bicycle were because it is faster, riders don't have to rely on public transit, and it is healthier to bicycle. Some of the main reasons noted by Beck for not commuting by bicycle were trip distance, uncomfort, inability to travel with other people, and I can t carry bags (cargo). IN a similar vein Ohrn makes assumptions regarding factors that might be related to commuting by bicycle in Minneapolis-St. Paul, Minnesota These factors included flexibility of schedule, average trip length, age of trip maker, availability and cost of automobile
FACTORS THAT INFLUENCE BICYCLING TO WORK storage. cargo needs of trio street congestion quality oi the pedestrian system and the availability of oublic transi t. ill) HANSON compared the affects oi daily weather data on discretionary and non-d i scretionary travel by bicycle in Uppsala, Sweden.(.1Q) Unlike the non-discretionary work trip, travel for discretionary reasons can be postponed or not undertaken if adverse weather exists However, due to the inflexibility of the time in which the work trip must be made, Hanson points out that "the only travel decision [that an individual) is really free to make (aside from the route to be taken ) is the mode of travel to be used."(lQ) Specifically with regard to the e ffects of weathe r on bicycling to work Hanson found that daily tr avel to work .by bicyc l e was sensitive to temperature and amount of cloud coverage at 7 a m She fdund that t he percentage of dai ly trips t o work by bicycle increased as a function of increasing temperature and decreased as a function of greater morning cloud coverage. SIMILARLY, Ashley and Banister conducted a study that considered an extensive array of factors at the ward level in Greater Manchester, England, which might influence an individual's choice to uti li ze the bicycle for the work trip.(11) They found that trip distance, car ownership, tran sit (bus) availability, rainfall traffic, h illiness and soc ial class of head of household were all factois that influenced bicycle commuting. EVERETI and Hirst, in two analogous studies, assumed that the essential ingredients for explaining an individual's choice not to commute to work by bicycle were based solely on economic grounds.(12)(ll) In their ana l yses they noted that nominal trip costs and travel time favor the choice of the automobile over the bicycle as a commut in g option. GOLDSMITH foun d that, in addition to the many factors listed above, family circumstances, personal habits, and topography a l so affect an individual s decision t o commute by b i cyc le.W) OBJECTIVE ABSENT from the literature is an examination at an aggregate level that focuses on factors that influe nce travel to work by bicycle. This article addresses this dearth through an empirical investigation that explores some of the factors that influence the nondiscretionary work trip by bicycle at the metropolitan level utilizing data extracted from the 1990 U.S. Census. The aim of this paper, therefore, is to develop a predictive model using regression analysis which might suitably depict the contribution of the factors that influence the use of the bicycle for the trip to work at the metropolitan area levei.U2) SAMPLE THE U S Census Bureau (USCB) conducts periodic and special studies to describe the characteristics of the American people, their governments and their businesses. The USCB aggregates and r eleases data from the Decennia l Census by political and statistical areas.
FACTORS T HAT INFLUENCE BiCYCLING TO WORK CENSUS data include a comprehensive codification of possible travel modes available for analysis. The trave l modes codified by the Census include less traditional modes such as bicycling and walking as well as the more traditional modes of POV and public transit. The Cen sus contains information pertaining only to the non-discretionary work trip: discretionary trip purposes are not included. THE data used in the analysis included factors from a sample of 284 Metropolitan Statistica l Areas (MSA) in the United States, including both Hawaii and Alaska, and exc lud ing Puerto R i co. This sample represents 100 percent of the MSAS in the U S For reference, the 1990 Census file and tab l e sources and universes are iden t ified in Table I for ai i variab les used in this paper; universe refers to the segment of the population to which the respective variables refer. IT would be hard to model i nd ividual ized travel behavior without first gathering primary data. On the other hand, secondary data, such as the Decennial Census, are of h ig h quality, readily obtainable, and easily analyzed. However, relying on aggregate and summarized data such as Census data limit the specificity of the analysis that can be performed. The acquisition of individualized records is required for maximum analytical flexibility. Therefore, due to the aggregate and summarized nature of Census data, its use will not fully expose the micro-level determinants that influence bicycling to work. FACTORS Due to a lack of previous research, judgement was utilized in selecting the variables used in the analysis. This resulted in the selection of 27 variables. They are listed and briefly explained in Table II. METHOD To analyze the sample of 284 MSAs, multiple regression analysis was performed utilizing the stepwise regression option set at the 0.05 significance level. This small significance level of 0.05 was chosen to guard against the inclusion of any independent variables that may not contribute to the overall predictive power of the model. The stepwise option performs regression in "steps," i.e., it substitutes each indepe ndent variable one at a time either through backward elimination or through forward selection in an attempt to find the best model. Stepwise regression analysis is a useful and valid statistical procedure, particularly when attempting to identify a few significant variables that should be included in the model from a multitude of independent variable s SOMETIMES in postulating a regression model the independent variables are often highly related to each other. T his problem is termed multicollinearity When two independent variables are highly correlated the regression equation cannot accurately estimate their independen t effects on the dependent variable. Optimally, a correlation of zero is preferred. Unfortunately, however, when performing regression analysis, multicollinearity is usually present.
rACTORS THAT I N F LU ENCE B I CYCUNG TO WORK TABL;t 1 990 CENSUS FILE A N D TABLE SOUR CES AND U NIVERSE VARI.BUl SOUltCS F lu; Teu Utd\I'EIUf T OTAl. Poi>UtJ.n::-: SiF1C ""'' DetlOI\S I.AW:lAAU S7F1 C n" AG:a S Tf1C POt>1 3 Etr.r...Ar iCfli.MI. <: ... t .. l" -' .) ... ? 0&1 persons a9e STFlC 9!! ;ge 3 -OiNSiiY STF1C Ps <111 pet$0rl$ Q$1101N STF 1 C cf Hispai'IC orip1.1 M:OV.N HOuS&IO'.Ob.'CCf>IE. smc POIIOA e l l tlou:s.ehOX!s Pol/aRTY L\ia.$T A TI)$ STF:l C ? 121 &eleae
FACTORS THAT INFLUENCE BICYCLING To WORK TABLE II lABEl AND ExPLANATION OF INDEPENDENT VARIABLES PEIU62S Pcr.;e n : of IX!OuL;!)On betWeen tne ape of 1 6 29 Pert$nt of 11\at i$ Percent 01 age 1 6 ano twer :.n 1ne wonc force Me:Ula:K:n trial s or H 1 saant: cnplll Perce., t of in Percent of not."Seho'
FACTORS THAT INFLU ENCE BICYCLING To WORK multip l e correlation coefficients of 0 .821 0 .831 an d 0 764. meaning that the equations explained approximately 67, 77. and 58 percent of the variation i n bicycling to work i n each region. resoectiveiy. F our variables ent ered the eoua i ion for the WEST. The WEST equation generated a multiple correlat ion coefficient of 0. 797. meaning that neariv 64 oercent of t h e variation in PERBIKE was explained TABLE IV USCB GEOGRAPHI C REGIONS GEOCIW>I: CiNM REOIO't CAsES Peooem Wf.S'1 n=
FACTORS T HAT INFL U ENCE B ICYCUN G T o W ORK substit u ted wit h the p e r centage of work trips made by walkin g ( labe l ed :: ::RWAL K ) i n 1 992 in each MSA. A regression run was performed using PERWALK as the dependent variable utilizing the same stepwise opti ons ior backward elimi n at i on and iorward selection TABLEV CORREL.AnON BETWEEN FACTORS A N O PERBIKE LAS:l. COfiAELJo LAea P : R \ 629 0.357 1-"''VPOP 0 014 P1'WS1t.t: 0 .253 P R'NOC 0 .081 P:RFEW-0 074 l NWI'I\'!H::>. 0 .2 7 3 P !tOIIT 0.062 Q J;14 PEk.'>"h'r O .OOi HO"N:ll\1<. 0 .023 H!:3HSCH 0.1 7 PUt.COUG 0.$12 IW.V.HO "'""""' 0 .08S PER!N 0 0 7 PtR:PO'o' 0 .051$ TIME10 0 128 . ...,.. 0 267 P eRNAw 0 353 ....... -0.100 USING PERWALK as the dependent variable, a total of 11 independent variables entered the equ ation for ALL MSAS and had a multip l e correlation coeffic i ent of 0 .856, meaning that approximately 73 percent of the variation in PERWALK was accounted f o r by the equation. As hypothesized, both PERML TY and PERCOLLG entered in t o the equation and were sign i ficant at the 1 percent leve l when PERWALK was used as the dependent variable Therefore, t h is result lends further evi dence to support the supposition t hat metropolitan areas that contain communities with unique housing characterist i cs and economic strata promote the use of non-motorized trans p ortation modes. FoR ALL MSAs, a str ong inverse relationship appeared between not having a vehic l e available (NOVEHCL) and bicycling to work This variable was sign i ficant at the 1 percent level. This parti cu lar var iable, however, did n o t appear in any of the regional equations Explanation of this is somewhat problematic One possible reason for these regiona l differences might be that persons t oo poor to own and operate a POV are forced to c ommute to work by b i cycle or some other non-POV mode. However, variables such as median family income (labeled MEDINC) did not appear in any of the reg i o nal e q uations and the percentage of famili e s that are poor (labeled PERPOV) only ente red int o the region al equation for the WEST. However, the percentage of the popu l ation that is unemployed (labeled NOWORK) did enter into the equatio n for A L L MSAs, the WEST, and the SOUTH. Another possib l e exp l anation is that workers who live so near to the i r pl a ce o f employment
FACTORS THAT INFLUENCE BICYCLING To WORK :io not nee d to own a POV. If this wer e the case it would be logical to expect that variables such as POPDENS. P:Rwcc. and PERPCC would have entered in to one or more of !::e rag i onal eouat io ns TABLE VI VARIABLES THAT ENTERED THE STEPWISE REGRESSION EQUA T IONS WE" NOit'I'M Cr:Nnw. NOitnMf Sv.n>< \'.a Co!n TV.&W "" T\'<14\IE COUF. r\I..Wt COUI. T\'ALUC coc" !I ';,9)S' '0<>' 0,61$ P(AAc.lli : z.ore-o.n: 2021' Zto:: 7 ,i7i o m 4 ,&$6.' .,. o""" .:.<:22' 0.2'59 4.34f 1 $ 1 3 031' .o.sn 2 .$4-... .......,.,. .... 2 c.;,. 0 489 S n S .t.Gt-7' ...... 0 .435 J.:?.t;f P!;ltNOWI .113 2 .49'2-PeA. 0.1$1 9.21g' "'"""'v -0.224 .66$" Ct"Sf-4'1': O .Ot 1' 0.005' 0.00006- 0.00 7 o .on 0.101 .,., .,. 0 ,815 0.754 "' o .soo 0.635 0 ,$74 0 7<0 .., NJ.I. ...... .... 0 66 1 (1,748 ..,. 'v"'" !14. 395' 18.12&' 49.029" 42 .65'1 Sl. r .e: s ign ifitant attl't! 1 pereenl k:W!I. -sigt'lilie#nl31lhe S percent level. at lhe 10 l evel As mentioned, there appears to be some relationship between income and the propensity to bicycle to work since NOWORK, a strong indicator of i ncome entered into three of t h e five equations Contrary to expectation, however, MEDtNC did not enter into any o f the equa t ions and PERPOV entered only into the equation for the WEST, but was inversely re l ated The variable OWNER, another strong indicator of income, ente r ed the equation for A L L MSAS w i th a negative sign, thus, reinforcing the assumption that income may have had some effect on PERBIKE as well. FOR all MSAS, except those in New England which are an amalgamation of cities and towns are composed of entire counties and i nclude some land designated as agricu l tural.(15) The percentage of persons employed in agriculture (labeled PERAGRI) appeared in the equation for ALL MSAS and for the WEST. In both of these equations, PERAGRt was significant at the 5 percent level. Somewhat unexpectedly, however, PERAGR I did not enter the equation for the SOUTH. An interpretation for this result could not be inferred. Moreover
i=ACTORS THAT INFLUENCE BICYCLING TO WORK 1 0 ?::RAGRI revealed itself as the dominant variable in the eauation for the NORTHEAST and was sig nificant at the 1 percent level. FIGURE I ?IVE STRONGEST SIMPLE CORRELATIONS WITH PERBIKE PtRCOLLG -. -: : 0 .. : c . : oc 0 -. NOVEIICL PRl629 . . ' THE percentage of the population employed in manufacturing (labeled PERMANU) and the percentage ofthe population age 16 to 29 (labeled PER1629) entered only int o the equation for the NORTHEAST region; these variables did not enter into the equation for ALL MSAs nor the other regional equations. These two variables were significant at the 5 percent and 1 p e rcent significance levels, respectively. Several variables related to minority status did not enter into any of the equations. T he variables labeled as PERFEMA (percentage offemales age 16 and over in the work force) PERBLACK (percentage of population that is Black), and PERHISP (percentage of population that is of Hispanic origin) were included as variables i n order to test the supposition that minority status is related to commuting to work by bicycle. Surpris ingly, none of these variables entered into any of the equations. However, the variable PERASIAN or the percentage of the population that is Asian entered into the equations for ALL MSAS, NORTH CENTRAL, and for the SOUTH. In the equation for ALL MSAs, PERASIAN was significant at the 5 percent leve l and for the NORTH CENTRAL and souTH equations, it was significant at th e 1 percent level. The interpretation is that minority status, Asian in this case, does have a small effect on choosing the bicycle as a commute alternative for making the trip to work. This finding may also suggest that certain minorities may have less opportunity in securing a residence that is in close proximity to their location of employment due to a host of mitigating factors such as low income and housing segregation.
rACTORS T HAT INFLUENCE B ICYCLING TO WORK , As pointed out by Hanson (1 0), there appears to be some correlation between weathe r condit i ons and the leve l of bicycling Testi ng thi s hypothesis in this paper i s impossible due to the fact thai weather i nformation is not gathere d as part of the Decennia l Census data collect ion function However based on h isto r ica l t r ends i t i s possible i o draw some inferences pertaining t o regional weather patterns and each region's level of bicycling The reg io nal means f o r PERBIKE were cal culated as : WEST 1.01; NORTH CENTRAL 0 364; NORTHEAST, 0 .269; and souTH 0.321. Not surprising the highest leve l s o f b icycling oc c urred i n t he WES T which typically has good ye a r-round w eather a n d th e l owes t levels oc:urre d i n ine "ORTH!:'I
FACTORS THAT INF LUENCE BICYCUNG TO WORK 1 2 th e absolute numoer oi universities wou ld be viewed as an inauspicious policy, at bes t. However. heed should be given to the lessons learned from the unique MSAs that enjoy high l evels of bicycling. Wrth the prooer combination of education, publicity, incentives. and planning, bicycle usage will likely spill over into other metropolitan areas and, thus. gain a greater share of the national modal split. As Everett makes clear, careful planning and establishment o f bike-routes where they will enjoy the greatest usage is essential for the long-run viability oi bicycle transportation."(.ll) IN the absence of similar studies, a set of fac tors have been presented in this paper that ha,ve been l argely neglected. Planners and eng ineer s are well aware that practica l solutions for inc reasing the use of the bicycle are not often straightforward. Hopefully, others will fo llow suit with studies of this nature that utilize a host of data ranging from futur e Censuses, NP TS trave l surveys, and individualized case studies. These studies will be a welcome addendum to these pre lim inary ACKNOWLEDGMENTS THE author is grateful to Mitchell York, Research Associate, Center for Urban Transportation Research, for providing the data for this paper. In addition, the author would like to sincerely thank Patricia Henderson, Joel Rey, Dennis Hinebaugh, and Steve Polzin for their editorial expertise and valuable comment s regarding the content of this paper.
ENDNOTES Acco rding to data compi1eo oy Gross and Feldm an in Table 5 o n pa ge 62 of Nationai T ransportaiion Statist ics 1995 Bureau ofTransporta t i on Statist ics. USDOT November 1994. vehic l e miles for passenger cars have inc r eased from 588 billion in 1960 to 1 .59 trillion i n 1992 an increas e of approx1mately 171 percent. 2. Gross. Marilyn and Richard Feldman National Transportation Statistics 1995 Bureau of Transportation Statistics USDOT, November 1994 pp. 170 3 To con vert miles to k ilom eters multiply the numbe r of miles by 1.61 4 Kom anoff Charles and C ora Roelofs The Environmental Benefits of Bicycl ing and Wa lking, Nat i ona l Bicycling and Wa l king Study Case Stuqy 15 USDOT J anuary 1993. FHWA-PD -93 015 pp. 3 . 5. Vincen t Mary Jayne et al., NPTS Urban Travel Paltems: 1 990 Nationwid e Personal Transp ortation Survey Office of Highway I nforma tion Management, FHWA, H P M-40, 1994 page 4-4. 6 For a thorough treatment of the health benefits of bicycling see: Burke Ed mund R., Benefits of Bicycli ng and Walk ing to Health, Natio nal Bicycling and Walking S tudy Case Study 14, USDOT, June 1992 FHWA-PD-93-025. 7. Badgett, Shauna 1., Debbie A. Nie m eier and G Scott Rutherford Bicycling Commuting Deterrents and Incentives : A survey of Selected Companies in the Greater Seattle Area; Paper 940802 Presented at the 73rd Annual Meeting of the Tran sportation Research Board January 9-13 1994 8 Beck, Michie! J H Bicycling Owne rsh ip and Use in Amsterdam," Paper 940948. Presented at the 73rd Annual M eeting of the Tra nsporta tion Research Board, January 9-13, 1994 9 Ohm Carl E., "Pred icting the Type and Volume of Purpose ful Bicycle Trip s," Tran spo rtation Research Record 570, 1976, pp 14-18. 10 Hanso n, Susan 'Evaluating the Impact of Weather on Bicycle Use." Transportation Research Recor d 629 1 977 pp 43-48 11. Ashley, Carol A. and Chris Banister. cycling to W ork From Wards In a Metropolitan Area : Factors Influencing Cycl ing to Work ; Traffte Engineering and Control, Vol. 30, Issue No. 6 June 1989 pp. 297-302. 12 Eve rett, M i chael "Commuter Demand for Bicycle T ransportation in the States; Tr affic Quarterly Vol. 28, No. 4 1974, pp. 585-601 13 Hirst, E Bicycles, Cars and Energy," Tr affic Quarterly, Vol 28, No.4, 1974, pp. 573-584. 14 Gol dsmith, Stewart A., Reasons IM!y Bicycling And W alking Are And Are Not Being Used More Extensively As Travel Modes National Bicy c ling an d Walking Study Case Study No. 1, USDOT, January 1993, FHWAPD 92-041. 15 The general concept of a METROPOUTAN AREA ( MA) Is one of a l arge populat ion nucleus, tog ether w ith adjacent ocmmunities t hat have a high degree of economic and social integration with that nucleus Some MAs are defined around two or more nuclei. Each MA must contain a place with a minimum population of 50, 000 or a Census Bureau-defined u rbanized area and a total MA popul a tion of at least 100 000 (75 000 in New Eng land). An MA comprises one or more cent ral counties An MA also may includ e one or more outlying counties that have close economic and socia l relationships with the central county. An outlying county must have a specified leve l of commu ting to the centra l counties and also must meet ce rtain standards regarding metropolrtan olharacter, such as populat ion density. urban population, and population growth. I n New Eng l and, MAs are composed of cities and to wns rather than whole ocunties.