1Truth in Transportation PlanningDONALD C. SHOUPUniversity of California, Los AngelesABSTRACTTransportation engineers and urban planners oftenreport uncertain estimates as precise numbers, andunwarranted trust in the accuracy of these precisenumbers can lead to bad transportation and land-use policies. This paper presents data on parkingand trip generation rates to illustrate the misuse ofprecise numbers to report statistically insignificantestimates. Beyond the problem of statistical insignif-icance, parking and trip generation rates typicallyreport the parking demand and vehicle tripsobserved at suburban sites with ample free parkingand no public transit. When decisionmakers usethese parking and trip generation rates for city plan-ning, they create a city where everyone drives totheir destinations and parks free when they getthere.Beware of certainty where none exists.DANIEL PATRICK MOYNIHANINTRODUCTIONHow far is it from San Diego to San Francisco? Anestimate of 632.125 miles is precise but not accu-rate. An estimate of somewhere between 400 and500 miles is less precise but more accurate, becauseKEYWORDS: parking, regression analysis, urban planning.2 JOURNAL OF TRANSPORTATION AND STATISTICS V6/N1 2003the correct answer is 460 miles.1 Nevertheless, ifyou did not know the distance from San Diego toSan Francisco, whom would you believe: someonewho confidently says 632.125 miles or someonewho tentatively says somewhere between 400 and500 miles? You would probably believe the onewho says 632.125 miles, because precision createsthe impression of accuracy.Although reporting estimates with extreme preci-sion suggests confidence in their accuracy, transpor-tation engineers and urban planners often useprecise numbers to report uncertain estimates. Asexamples of this practice, I will use two manualspublished by the Institute of Transportation Engi-neers (ITE): Parking Generation (ITE 1987a) andTrip Generation (ITE 1987b, 1991, 1997). Thesemanuals have enormous practical consequences fortransportation and land use. Urban planners rely onparking generation rates to establish off-street park-ing requirements, and transportation planners relyon trip generation rates to predict the traffic impactsof development proposals. Yet a close look at theparking and trip generation data shows that placingunwarranted trust in these precise but uncertainestimates of travel behavior leads to bad transporta-tion and land-use policies.TRIP GENERATIONTrip Generation reports the number of vehicletrips as a function of land use. Transportationengineers survey the number of vehicle trips to andfrom a variety of locations, and for each land usethe ITE reports a trip generation rate that relatesthe number of vehicle trips to a characteristic ofthe land use, such as the floor area or number ofemployees at a site. The sixth (and most recent)edition of Trip Generation (ITE 1997, vol. 3, pp.ix and 1) describes the data used to estimate tripgeneration rates as follows:This document is based on more than 3,750 tripgeneration studies submitted to the Institute bypublic agencies, developers, consulting firms,and associations. . . . Data were primarily col-lected at suburban localities with little or notransit service, nearby pedestrian amenities, ortravel demand management programs.ITE says nothing about the price of parking at thestudy sites, but since parking is free for 99% ofvehicle trips in the United States, most of the studysites probably offer free parking.2 Trip Generationuses these 3,750 studies to estimate 1,515 trip gen-eration rates, one for each type of land use. Half the1,515 reported trip generation rates are based onfive or fewer studies, and 23% are based on a singlestudy.3 The trip generation rates thus typically mea-sure the number of vehicle trips observed at a fewsuburban sites with free parking but little or no pub-lic transit service, pedestrian amenities, or traveldemand management (TDM) programs. Urbanplanners who rely on these trip generation rates asguides to design the transportation system are there-fore planning an automobile-dependent city.Figure 1 shows a typical page from the fourthedition of Trip Generation (ITE 1987b).4 It reportsthe number of vehicle trips to and from fast foodrestaurants on a weekday. Each point in the figurerepresents one of the eight studies and shows thenumber of vehicle trips per day and the floor area ata restaurant. Dividing the number of vehicle trips bythe floor area at that restaurant gives the trip genera-tion rate at that restaurant. A glance at the figuresuggests that vehicle trips are unrelated to floor areain this sample. The extremely low R2 of 0.069 forthe fitted curve (regression) equation confirms this1The airline distance between San Diego and San Fran-cisco is calculated from the latitudes and longitudes of thetwo cities. See “How far is it?” at http://www.indo.com/distance/. “Accurate” implies fidelity to fact and freedomfrom error, while “precise” implies exactness.2The U.S. Department of Transportation’s 1990 Nation-wide Personal Transportation Survey (NPTS) askedrespondents, “Did you pay for parking during any part ofthis trip?” for all automobile trips made on the previousday. Of the responses to this question, 99% were “no.”The NPTS asked the “did you pay for parking” questionfor all vehicle trips except trips that ended at the respon-dents’ homes, thus free parking at home does not explainthis high percentage.3This refers to the sixth edition of Trip Generation (ITE1997). The ITE Trip Generation Handbook (ITE 2001, p.10) notes that the warning “Caution—Use Carefully—Small Sample Size” is placed on each trip generationreport if the sample includes five or fewer sites. At mostsites, vehicle trips are observed during the course of onlyone day.4The fourth edition (ITE 1987b) is shown because this isthe date of the most recent edition of Parking Generation,to which Trip Generation will be compared. Vehicle tripswere surveyed at McDonald’s, Dunkin Donuts, BurgerChef, and similar fast food restaurants.SHOUP 3impression.5 Nevertheless, ITE reports the sample’saverage trip generation rate—which urban plannersnormally interpret as the