The Nature and Causes of the 1982 Traffic Accident Casualty Reductions in British Columbia

Transcription

1 Transportation Researh Reord The Nature and Causes of the 1982 Traffi Aident Casualty Redutions in British Columbia PETER J. COOPER ABSTRACT The extent of the 1982 traffi aident asualty redution in British Columbia has been evaluated and found to be beyond what ould be expeted from hane flutuation alone. The greatest hanges were found in asualty-produing aidents involving young drivers, espeially during the nighttime. These aidents were seemingly independent of alohol use. Given this initial determination, the possible external auses for the derease have been examined with the result that eonomi fators alone an be shown to aount for a large part of the effet, although for a signifiant portion of the hange, there is as yet no ready explanation. The most important onlusion arising from the analysis is that should eonomi onditions improve, a onomitant rise in traffi asualties an be expeted unless mitigated by signifiant safety program interventions. A omparison of 1982 statistis on traffi aidents in British Columbia with those for 1981 indiates a dramati and welome hange. Total aidents dereased approximately 12 perent while injuries and fatalities dereased 23 and 3 perent, respetively. A similar effet was observed throughout Canada but British Columbia showed the greatest redution, partially as a onsequene of its high 1981 ranking. The question, of ourse, is why? What happened during 1982 to ause the aident and asualty redutions- was it a hange in travel patterns or a hange in safety attitudes? In this paper, an attempt is made to ast some light on the ontroversy. THE NATURE OF THE EFFECT The overall aident statistis (,!,ll for 1981 and 1982 are as follows: Fatal aidents Injury aidents 3,626 23,662 Property-damage-only (PDO) aidents 96,444 87,682 Deaths Injuries 44,123 33,87 Change _(%_) Beause the greatest hanges are apparent in injury and fatal aidents or asualties, they form the primary fous for this paper. Casualties have been hosen as the best historial index beause they represent a somewhat greater sample size than asualty-produing aidents and form one of the statistis normally employed in epidemiologial studies. Figure 1 shows the 1982 asualty redutions set against the monthly totals for previous years. A fairly onsistent seasonal flutuation is evident 5 average seasonal flutuation in relation to the linear trend line - \,'-"\,',_, ', ---"\ --,r-, I,, I ' FIGURE 1 Historial traffi aident asualty trends in British Columbia

2 2 Transportation Researh Reord 147 from the data and this has been superimposed on the upward trend line established by using the 1979 and 198 information. The first issue to address is whether or not the 1982 redution is signifiant and when the redution atually ommened. This an be aomplished fairly simply. An examination of Figure l would lead one to suspet that the deline in asualties atually ommened about halfway through Using the Box-Jenkins tehnique to projet expeted values for 1982 based on the time series for the previous 5 years, there was a marginally signifiant differene found between atual and projeted values (Xi probability of 8) The time series before 1982 is not entirely devoid of other events, however, suh as periodi seat belt wearing ampaigns and drivingwhile-intoxiated (DWI) blitzes, whih may affet the signifiane assessment of the redution. In using analysis-of-variane tehniques to test the asualty data, whih is grouped by half-year periods, it was learned that the redution ourring between the first half of 1982 and Anova Summary Table Sour of Variation DF Sum of Squ r '1 an Squar F Si9nlfian L v l : Within Qroups Total ~ :1.926 Group St tisti Group Codes a. Labels N Me.an SD Group :17 Group ~ Group Group 4 19-::: Group ~ 2~-3C) : : Group 6 '.ll ~.ooo Group :1.891 Group Group 9 49-~ :148. ::S3::S T-Teat e.t.. n Group '1 n - <Value of p r for a Not : Stati tia ar only printed if p 1 l than t 2.:192 p.2e b1-tail d teat> or equal to.:1 Group ::s Group s t 4.73~ p..1 Group s Group 7 t p..1 Group Group II FIGURE 2 Comparison of half-year period asualty totals,

3 ooper 3 the first half of 1981, and between the seond half of 1982 and the seond half of 1981, were statistially signifiant at the.1 level (see Figure 2). The 1982 asualty derease is thus a real effet that annot be attributed to hane alone and, from a visual examination of the monthly figures in omparison with the seasonal flutuations and the trend line established from the data, it an be postulated that this derease atually ommened at approximately the midpoint of ASSESSMENT OF POSSIBLE CAUSES There have been a number of suggested ontributory auses for the effets that have just been observed. Some of these, together with an assessment of their appliability, are as follows: 1. A Change in Vehile Oupany--For 1981 and 1982, we an alulate both fatality-to-fatal aident and injury-to-injury aident ratios. These are 1.13 and 1.44, respetively, for 1981, and 1.15 and 1.43, respetively, for Beause no major vehile safety design advanes ourred during this period, it must be onluded that vehile oupanies as refleted in the asualty statistis were the same in 1982 as in An Overall Travel Redution--It is ommon pratie to assess overall travel hanges through fuel sales. Indeed, suh an exerise indiates that a 1 perent travel redution did our in 1982 (as opposed to 1981 levels) and this losely mathes the total aident rate redution of approximately 12 perent but it does not, by itself, aount for the muh larger redutions in injuries and fatalities. 3. A Derease in the Number of Vehiles--Dur ing 1982, the number of new passenger ar sales ertainly suffered a signifiant downturn but the number of registered vehiles in British Columbia still inreased marginally. If anything, the result should have been toward less safety beause there evidently were more older ars on the road. 4. A Derease in the Number of Drivers--Atually, the total number of liensed motor vehile operators inreased by just under 2 perent from 1981 to In spite of this, however, there was a marked redution in new lienses issued to young drivers. This will be disussed in more detail later in the paper. 5. Improved Safety Attitudes of Dr ivers--if something along these lines had ourred, it should be refleted in other statistis suh as seat belt wearing rates. In fat, the average surveyed level of seat belt wearing among the general driver population in 1982 was approximately 55 perent for all oupants, whih is not onsidered to be muh of an improvement over 1981 when the survey results were, unfortunately, not as reliable. In spite of this, there does seem to be some indiation of inreased use in the fat that rural fatalities for unrestrained oupants apparently dereased at a higher rate than that for all oupants. Also, aident reports for injury-produing rashes suggest a 5 perent restraint-use inrease in 1982 over 1981 for oupants of vehiles involved. This indiates that the various seat belt eduation and enforement efforts were at least partially effetive. Figure 3 shows a summary of fatality redutions from 1981 to 1982 for some important aident and vitim ategories. Beause fatalities and fatal aidents showed the greatest overall hange, any asual effets should beome evident within this subset. Aident ategories have been seleted so as to highlight the three most important areas of safety at present--age, alohol use, and seat belt wearing. In terms of traffi law violations, there is no indiation that 1982 was any better than In fat, for young drivers, who aumulate a substantial Total Fatalities - All Ages Total Fatalities - Ages 16-2 Passenger Fatalities- All Ages Passenger Fatalities- Ages 16-2 Driver Fatalities - All Ages Driver Fatalities - Ages 16-2 Drinking Driver Fatalities- All Ages Drinking Driver Fatalities- Ages 16-2 All Fatal Aidents Single Vehile Fatal Hwy. Aidents All Fa tali ties Rural Nighttime M.V. Fatalities All Fatalities to Rural M.V. Oupants Fatalities to Unbelted Rural Oupants % REDUCTION FIGURE 3 Changes in seleted fatal aident ategories,

4 4 Transportation Researh Reord all drivers 5% redution _from '" ~ CP6. 'ti. '"..,., "' / t'-... Qj O> '" ~ 4 / u,,/', ~9 % i:edution.,, from 19~. '.. - j uvenile drf vers. al2 o"' ""'""'------~ Year FIGURE 4 DWI harges laid per liensed driver. number of these onvitions, the number of violations per liensed driver rose by 13 perent in 1982, For the major Criminal Code onvition (DWI offenses), Figure 4 shows that there indeed were redutions from 1981 to 1982 but it is doubtful if these ould be regarded as signifiant in light of previous flutuations. Alohol and age are, to a great extent, intertwined, but the fat that two ommonly utilized surrogates for alohol-involved aidents--rural single vehile and nighttime fatal aidents or fatalities--dereased by more than their respetive parent populations, suggests that less drinking-anddriving ourred during 1982, Beause alohol sales (inluding beer) showed a slight overall inrease between 1981 and 1982, the redution was apparently in driving not drinking. This is supported by the similar levels of derease for alohol-involved driver fatalities as for all driver fatalities and by the fat that the proportions of asualty-produing aidents involving alohol were almost idential for 1982 and 1981 in eah age group. The greatest level of over-involvement in alohol-related fatal aidents has traditionally been with the 16-2 age group and Figure 2 onfirms that the most dramati effets are evident here, speifially among drivers. Figure 5 shows that the redution in asualty-produing aidents was most pronouned for young drivers at nighttime and that this redution has ontinued through The question still remains, however, as to what ould have triggered suh apparent hanges. 6, The Eonomy--There is little argument that North Ameria in general and Canada in partiular have been in the grips of a severe reession. It has previously been observed that the 1982 asualty redutions were signifiant and apparently had ommened following the first half of Various eonomi indiators that have been published may now be examined to determine whether or not there is any orrespondene in timing. Figure 6 illustrates the major eonomi hanges. It is evident that the beginning of the downturn ame lose to the middle of 1981 and thus one might be tempted to onlude that the major ause of the redution was the eonomy. This an only be onfirmed, however, if a fairly reliable eonomi preditor or preditors an be identified that will reprodue the flutuations over a number of previous years and then "predit" the 1982 redution. The most likely andidate for a general trend preditor is the unemployment rate, whih Eshler (3) reported following an examination of annual u. s. fatal aident statistis between 1949 and 1973, and whih was onfirmed bv Partvka's work 141 usina annual U. S. fatality st~tisti~ from 196 th.rough l982. It might thus be antiipated that unemployment (or employment) would be a useful indiator, espeially if ombined with a fator representing travel exposure to help aount for the month-to-month variations that Eshler and Partyka did not have to ontend with. By grouping the monthly asualty data by gasoline sales and employment, it ould be shown (through analysis-of-variane tehniques) that both variables signifiantly differentiated the dependent variable groups (p <,5). Gasoline sales are ertainly an important aspet of onsumer spending but they also relate diretly to travel, and therefore aident, exposure and have traditionally shown a seasonal flutuation. Figure 7 shows how major flutuations in employment rate alone orrespond with major hanges in monthly asualties. Leastsquares trend lines have been fitted for the periods January 1979 to June 1981 and July 1981 to Deember The question posed by the apparent orrespondene between traffi aident asualties and unemployment is, of ourse, what will happen to traffi safety in the event of an eonomi reovery? Eshler <.~.> found that fatal aidents invariably inreased following periods of reession and, while the expeted reovery has been somewhat delayed, it an be expeted that

5 Cooper 5. "' '1l > '1l.. 1 CD 2 "'. '1l > : 3 u 'tl "'1l day-all drivers -- day-young dr i vers nigh all drivers nightyoung drivers 5 '-~~~~~~~~~~~~~,~~~~~~~~~~~~ Year FIGURE 5 Redution in asualty aident involvement. the various eonomi indiators will eventually rise again. A number of different models were tested using variations of asualty rate and employment statistis ombined with gasoline sales and other fators (i.e., reported seat belt use in asualty-produing aidents that were pereived to have hanged between 1981 and 1982). In terms of prediting atual asualty numbers, only employment and gasoline sales were found to be useful although seat belt-wearing rates were signifiantly related to asual ties per billion vehile kilometer (BVK) of estimated travel at a low level (r z =.134). Employment for young males was examined separately from total employment as it was believed that this group would have been hardest hit by the reession. In relation to all asualties, there was no differene between the orrelation using total employment and that using young male employment, suggesting that there is a onsistent relationship between these two independent variables. In relation to asualties resulting from young driver aidents, however, the use of young male employment in ombination with gasoline sales aounted for 8 perent of the variane as opposed to 72 perent of all asualty variane explained when all employment and gasoline sales were used. The best model for total monthly asualties and the atual asualty flutuations are shown in Figure 8 and there is obviously a reasonable orrespondene not only before 1982 but also during that year and 1983 as well. In other words, simply by realizing the trend in employment and fuel sales, one ould have predited the approximate nature and magnitude of the 1982 asualty derease. The fat that this simple model only aounts for 72 perent of the asualty population variane is not surprising sine many other fators are at work. Weather effets for instane, aounted for the peaks in the last quarter of eah year and gasoline sales, when onsidered alone (as a surrogate for travel), produed a orreted rz of.59. Fatalities and property-damage-only (PDO) aidents were also found to orrelate signifiantly with eonomi fators although at a redued level of assoiation. This was undoubtedly beause (a) fatalities in British Columbia are relatively small in number and thus subjet to onsiderable random flutuation and (b) PDO aidents are not as reliably reported as those where injuries are involved. (PDO aidents are also highly weather related). It an thus be onfirmed that the major part of the 1982 redution in the number traffi aident asualties was an expeted result of the eonomi downturn, beause a simple eonomi model ould have predited it without having to onsider any noneonomi fators. Also, this same model does reasonably well in aounting for the month-to-month and yearly variations before It should be noted, however, that just beause 72 perent of the historial asualty variane an be explained by suh means, it annot be said that the potential exists to redue asual ties by no more than 28 perent no matter what safety efforts are undertaken. The eonomi fators are superimposed on a general limate of safety onsiousness or behavior (and vie versa), whih, between 1978 and 1981 in British Columbia, has been onsidered relatively onsistent. It is still theoretially possible to lower this baseline aident level through energeti and judiious appliation of safety-related regulations and programs. A POSSIBLE MECHANISM FOR CASUALTY REDUCTION IN BRITISH COLUMBIA As seen ear lier, the most notieable redution in fatalities during was for young drivers espeially where nighttime driving was involved. This suggests a possible, eonomially motivated redution in high-risk driving exposure by young adults. During this period, the number of aloholrelated fatalities and DWI harges for young drivers also fell dramatially although the proportion of alohol-related asualty aident involvement by age group remained quite stable.

6 6 Hou1inp St:.aru and Compli=tiani (Thousands of unin. 1euonolly.Sjuatld 11 annual rates) 225 Tehnial Servie Counil Job Vaanies for Prafe:nlonab v1 lnduflrial Produtron Soure: Canada Mortgage and Housing Corporations; The Conferene Board of Canada. 5 4 J ~ti:&11 W11ga.1 and lnomn (Perentage hange from same quarter a year earlier) Real sable. no~,' ~, :.,/ I 25 x w 225! > (.) 2 z < (.) < > 175 "' ~ u "' - TSC..!!:! '.'.".C.".NCY INDEX 15 INDUSTRIAL -PRODUCTION INOE The BC PROFESSIONAL ENGINEER Marh z ~ (.) 13 ::> "' a. 12 ~ "' t; 11 ::>!!!: u.. 1 x w!!!: Soure: Statistis Canada; The Conferene Board of Canada. F1GURE 6 Performane of eonomi indiators, : <11 ~. E <11 >, ;:: : E 'O <11 Ul :::>., 'O "' : :::> Ul <11. "' 5 " Ul u "' >, 1% ;:: 4 : E 95% 9% 3 "'...~-C- - - r I' employment 85% 1 8% traffi aident FIGURE 7 The oinidene of employment and traffi asualty trends in British Columbia. 1983

7 Cooper 4 3 /, model,/~ -1 I '.f' \ I \ I I \ ' l\ /~ 11~ * I I 1,o\ 1,_:;., / I ~ ",'/* ~"~- J ~ 1 i'x. \ 1 1 I,' \ \ l I Monthly Unadjusted rasoline j 'f' j I /"\ 75.8 Perent Sale for 624 Casualties = Emplo yment Road Use [ in mill ions of litres J 1 /~. / \ (employment and gas sales are moderately related \ ' ' "' with r =.58) '.! - '- ',\ /\ I~... ~.; "'(;. I ;,..I \'.\ A~ \ ; ::.~ j 1 orreted r 2.72,,' \;"' \ ' '\!~' \:~ \ observed ~/ F i( '- ',)t,,,. I rt I (signifiant at a:<.1) \v~. ~,' I 7 2 J ' M J J A~.. bl.j f' M. I.J A, t. N 1!>1.J,.. &...J.J A & t.:i Ol.J, M J J..!... l>l.j, M.J j " FIGURE 8 Predition of total mont hly traffi asualties in employment and gasoline sales year old 16 year-old Year Age Liensed Drivers Nev Liensee Traffi Offenses Speeding All Ka le A.U fl.ale All Iner lie. I Cr:>er lie. I (.139) (-38.5\) (-36.9\) all (.123) H (+14. \) (+14. 2\) all (.14) H (+14.\) (+12.\) all (. 167) all different reporting proedure all ontains some estimation for ages between 17 and 2 prior to 1981 sine the oaplete age break-down available within this group was only available following the 198 alendar year expeted no. of new lienses 34588, (aale & total respetively) atual no. of new lienses 1865, the differene la s1gn1f1ant (X2) at p<.1 liensed dr1ves \ n eased 1.7\ (+. 9\ for aales) but new lienses Issued to 16-2 year-old drivers de r eased 36.9\ (-38. 5\ for aalea) FIGURE 9 Changes in new lienses and traffi violations (. 231) 428 (.24) 3527 (. 221) 3638 (.249) different reporting proedure There are two possible explanations for redued travel or exposure: less driving by liensed drivers or fewer liensed drivers. The latter possibility an be readily tested by examining the driver liense reords from 1978 to 1982 and estimating from these the historial trend in new lienses issued annually to young (16-2 year-old) drivers. Although from 1978 to 1981 the number of new lienses rose by approximately 14 perent per year as shown in Figure 9, it dropped by 37 perent in The ombination of both fators an be investigated by testing the hypothesis that redued travel was the sole ause of all the asualty aident redutions, using an inferential type of approah to the data analysis. Begin by assuming that between 1981 and 1982, there was no driver attitude toward behavioral hanges, that is, the relative risk of aident involvement given a ertain level of exposure for eah age group did not hange from 1981 to This is akin to saying that 1 perent of the aident redution resulted from hanges in driving exposure. Based on fuel sales in British Columbia (~) and fuel surveys from Statistis Canada (6), it an be estimated that travel was approximately 26.8 and 24. BVI< in 1981 and 1982, respetively, whih represents a 1 perent redution. For 1983, the estimate was 22.8 BVI<,

8 8 Transportation Researh Reord 147 whih represents a 5 perent redution from Using the estimated 24-hr driving exposure in kilometers of travel based on the 1979 National Driving Survey results for British Columbia (7), and asualty aident proportions by age (take~ from unpublished data on the "Counterattak" program, whih was supplied by the British Columbia Ministry of the Attorney General), statistis for 1981 are given in Table 1. TABLE 1 Casualty Aident Proportion8 by Age for 1981 Drivers Involved in Total Casualty-Produing Casualty Aidents Aident Likely Driving Involvement Age Grouo Exposure (BVK) Total Perenhl g_p. f!p.t RVK % x 26.8 = , , % x 26.8 = , , % x 26.8 = , , % x 26.8 = , % x 26.8 = , ~ Total 47, ,268 TABLE 3 Total Travel Value 1981 Casualty 1982 Adjusted Aident Casualty Aident 1982 Revised Involvement Involvement per Estimated Driving Age Group per BVK BVK Exposure (BVK) ,473 4, ,278 2, ,517 1, ~ 1, Total 11,268 1, TABLE4 Risk Redution Adjustments, nn1,,, \..,d.'.:iuiuly 1982 AUju~t~u Aident Casualty Aident 1982 Revised Age Group Involvement Involvement per Estimated Driving per BVK BVK Exposure (BVK) ,473 3, ,278 2, ,517 1, ~ 1,9 1.6 Total 11,268 9, TABLE 2 Casualty Aident Proportions by Age for 1982 Drivers Involved in Casualty-Produing Aidents Casualty Aident Involvements per Estimated Driving Age Group Total Perentage BVK (from 1981) Exposure (BVK) , l 4, , , , , , , ~ 1.49 Total 36, , Using the 1981 ratio of asualty aident proportion to exposure, it is possible to (a) ombine this with the 1982 asualty aident proportion and then tu ib) work bakwards to obtain what would have been the exposure perentages had no hange in the ratio of asualty aident involvements per BVK of travel ourred. These statistis are given in Table 2. It is known, however, that total travel in 1982 was approximately 24 BVK--not the 21.6 BVK arrived at previously. The only solution to the disrepany is to postulate a derease in the driving risk, that is, a redution in the asualty aident involvements per BVK. The mehanism for applying suh a driving risk redution aross the various age groups is not known but it an be assumed that the range of eventualities will be overed by the following ases: i. Distributing the derease in risk in proportion to the size of the 1981 risk levels (aidents per BVK), whih amounts to an equal distribution of perentage risk redution aross all age groups; or 11. Distributing the derease in risk in proportion to the level of asualty aident redution between 1981 and 1982 in eah age group. This amounts to a differential perent risk derease by age group with the effet assumed greatest for those age groups having the highest levels of aident redution. For Case i it is found that approximately a 1 perent risk redution applied equally to all age groups will result in bringing the total travel value up to the orret level, as given in Table 3. TABLE 5 Comparison of Travel Changes with Redution in Aidents Based on Travel Derease Alone Redution in Apparent Change Aidents Based 1981 Casualty in Travel from 1981 on Travel Derease Aident to 1982 (BVK) Alone(BVK) Involvement Age Group perbvk Case i Case ii Case i Case ii , ,98 2, , ,278 1, , I.II 1,517 1, , Total 11, ,353 5,667 Similarly, when the level of risk redution is adjusted to reflet the perentage of the asualty aident derease in eah age ategory as ompared to the overall perentage derease, the result would be that the proposed risk redution would range from a high of approximately 17 perent for 16 to 2 year-olds down to approximately 7 perent for those age 66 and older. The results of applying these dereases are given in Table 4. To assess the degree to whih redutions in driving exposure have influened the hange in aident ourrene, as opposed to any postulated hanges in risk-taking behavior, the 1981 risk levels an be applied to the apparent travel hanges (in BVK) and the results ompared to the atual hange that ourred, as given in Table 5. Beause the total reorded redution of asualty aident involvement for drivers was 1,456 (47,388-36,932), it is apparent that an overall redution in travel alone an aount for, at most, 61 perent (Case i) and perhaps only 54 perent (Case ii). The regression model mentioned earlier showed that 59 perent of the monthly asualty variane ould be explained by gasoline sales (travel) alone. It should be noted here that even though a net travel redution an aount for, at most, 61 perent of the observed aident redution, a substantial portion of the postulated derease in risk of aident involvement ould ome about through a

9 Cooper 9 swith from nighttime to daytime driving. Figure 5 lends support to the possibility that nighttime driving may have dereased onsiderably more than daytime driving and, beause the former is assoiated more with nonessential soial ativities, it is logial to assume that the eonomi reession would initially affet it the most although as job-hunting ativity dereases, it might be expeted that the differene would beome less signifiant. The expeted redution, based solely on a onsideration of net travel derease, thus only aounts for at most, 61 perent of the atual hange. (Within this 61 perent, it an be further estimated that approximately 7 perent is due to the fewer number of liensed young drivers as mentioned earlier.) In other words, redued travel annot have been the only fator. Even seat belt usage would only add, at most, 2 perent to the redution explained, and thus, a signifiant perentage is left that must somehow be related to other hanges in driving behavior. In summary, it may be unreasonable to suggest that, of the signifiant asualty aident dereases that ourred between 1981 and 1982, a maximum of approximately 63 perent ould possibly be explained by an eonomially motivated redution in travel (espeially among young drivers during night hours) and the small inrease in restraint use. The remaining 37 perent or so is as yet unexplained and ould be assoiated with a hange in driver risk-taking behavior. It should be noted, however, that even suh postulated behavioral hanges as these ould be linked to the depressed eonomi situation and would thus be reversible in the event of an eventual improvement. CONCLUSIONS The major onlusions arising out of this investigation are as follows: 1. The 1982 redution in traffi aident asualties in British Columbia was a real and signifiant event that annot be attributed to hane alone. 2. The major part of this redution an be attributed to the effets of the eonomi reession. Redued travel was undoubtedly a signifiant fator, but there still remains a signifiant portion that annot as yet be explained in suh simple terms and that will require further researh to unover. 3. When the provine reovers from the reession, there is every indiation that traffi aident asualties will rise againi although the exat speed and extent of the rise are at present unertain, the asualties should follow the general trend in employment levels. REFERENCES 1. Annual Report Motor Vehile Department, British Columbia Ministry of Transportation and Highways, Vitoria, British Columbia, Canada. 2. Annual Report Motor Vehile Department, British Columbia Ministry of Transportation and Highways, Vitoria, British Columbia, Canada. 3. J.M. Eshler. Filtering of Fatal-Aident Rates. In Transportation Researh Reord 643, TRB, National Researh Counil, Washington, D.C., 1977 I P s.. Partyka. Simple Models of Fatality Trends using Employment and Population Data. Aident Analysis and Prevention, Vol. 16, No. 3, Gross Monthly Sales of Gasoline by Provine and Territory for Road Use, 1979 to Statistis Canada, Ottawa, Ontario, Canada. 6. Passenger Car Fuel Consumption Survey, 1981 to Statistis Canada, Ottawa, Ontario, Canada. 7. J.J. Lawson and D.E. Stewart. National Driving Survey Annual Conferene of Canadian Assoiation of Applied Soial Researh, Halifax, Nova Sotia, May Publiation of the paper sponsored by Committee on Traffi Reords and Aidents Analysis.