DEVELOPMENT AND T E S T I N G OF TECH111 gUES FOR INCREAS I WG THE COI:.ISPICUITY OF IIOTORCYCLES AlilD MOTORCYCLE D R I V E R S

P . L . O lson , R . Ha l s tead -Nuss loch , M. S i v a k

C o n t r a c t DOT-HS-6-01459

Highway S a f e t y Research I n s t i t u t e U n i v e r s i t y o f M i c h i g a n

Huron Parkway and B a x t e r Road Ann A r b o r , M i c h i g a n 48109

Oc tobe r 1979

FINAL REPORT

Prepared f o r

NATIONAL HIGHWAY TRAFFIC SAFETY ADMINISTRATION U.S. Department o f T r a n s p o r t a t i o n

Washington , 0. C . 20591

"Prepared f o r t h e Department o f T r a n s p o r t a t i o n , N a t i o n a l Highway T r a f f i c S a f e t y A d m i n i s t r a t i o n under C o n t r a c t No. : DOT-HS-6-01459. The o p i n i o n s , f i n d i n g s , and c o n c l u s i o n s expressed i n t h i s pub1 i ca- t i o n a r e those o f t h e au tho rs and n o t n e c e s s a r i l y t hose o f t h e i

N a t i o n a l Highway T r a f f i c S a f e t y A d m i n i s t r a t i o n . "

Tuhnical 2- DMulmtaiiom Page

I

4. Tit le ad Subtitle / 5. Rcport Dot* 1

1. Report No. 2. Gov*mamc Accession No.

Development a n d Testing of Techniques for Increasing the Conspicui ty of

3. R u ~ p ~ m t ' r Catalog No.

October 5, 1979 I 6. Puforning O ~ ~ ~ . . I I O I I Code

Motorcycles and ~o to r c ) c l e k i v e r s

7. Au*Iads) 1 Olson, P . L. , Halstead-Nussloch, R . , & Sivak, M . 9. f w h i n g Or9mtao(~on N w e atd Address

12. Spmsuing Agmcy N w e md Add**%*

National Highway Traffic Safety Administration 1 Final - 8-16-76

6-30-79 U.S. Department of Transportation Washington, D . C . 20590 14. bonsorlng Agency Cod.

015016 8 . Pwfom~ng O r g m l z O t l ~ R a w t NO.

UM-HSRI-79-76 9

10. Vorb Untt No. (TRAIS)

Highway Safety Research Ins t i tu te University of Michigan Ann Arbor, Michigan 48109

r

I

I S . S u ~ l n m t o r y Notas

11. Controct or Grant No.

, DOT-HS-6-01459 13. TYPO of R w w t ond Period Co*er.d

This p r o j e c t was i n i t i a t e d t o determine whether i t might be poss ib le t o reduce the incidence o f mu1 t i - v e h i c l e motorcycle crashes by improving the conspicui t y of the motorcyc le-dr i ver u n i t . S p e c i f i c a l l y , the f o l l o w i n g tasks were c a r r i e d ou t :

1. A review o f motorcycle acc ident data. 2 . Development o f p o t e n t i a l consp i cu i t y t reatments. 3. F i e l d eva lua t i on o f a se lec ted sample o f these treatments. 4. Determinat ion o f cos t bene f i t s .

A crash data ana l ys i s c a r r i e d o u t by H S R I , together w i t h several o the r publ ished s tud ies , makes i t c l e a r t h a t motorcycles are over invo lved ( r e l a t i v e t o cars and t rucks) i n accidents i n which the o the r veh i c l e i s execut ing a maneuver (gene ra l l y a l e f t t u r n ) across t h e i r path.

I More than t h i r t y consp i cu i t y t reatments were developed and demonstrated f o r sub jec t i ve app ra i sa l . A sample o f these was se lec ted f o r f i e l d eva luat ion .

The countermeasures were evaluated by means o f a gap-acceptance procedure. The study was run i n normal t r a f f i c , a n d measures were taken on d r i v e r s who were not aware they were invo lved i n a study.

The r e s u l t s i n d i c a t e t h a t daytime consp i cu i t y can most e f f e c t i v e l y be improved by use o f f l uo rescen t garments o r steady o r modulat ing l i g h t s . N ight t ime consp i cu i t y seems t o be aided by use o f r e t r o r e f l e c t i v e garments and runn ing l i g h t s .

The cos t -bene f i t ana l ys i s i n d i c a t e d t h a t a l l o f the items found t o produce s i g n i f i c a n t consp i cu i t y e f f e c t s are a l so c o s t - b e n e f i c i a l .

17. Key Wuds 1.3. Oistvibrtiac S to inm, Motorcycle , moped, conspi - 1

cui ty , gap-acceptance , motorcycle accidents, v i s i b i l i t y a ids , 1 ighting re t roref lec tor iza t ion , f l uorescence 1

I 19. k n t y CIesul. (ef h i s ~t 2. kry r t ty Clessif. (of p . ~ . ) 21. No. a l Pogea 22. Price 4

I

Unclassified Unclassified 133

DEPARTMENT OF TRANSPORTATION N A T I O N A L H I G H W A Y T R A F F I C S A F E T Y A D M I N I S T R A T I O N

REPORT A U T H O R ( S ) Olson, P . L . , Halstead-Nussloch, R . , & Sivak, M .

TECHNICAL SUMMARY

-- - - -- - - --

Motorcycles are a t once a bane and a blessing. They are a fuel- stingy form of transportation and recreation for mi 11 ions of Americans. They are also a re la t ively dangerous vehicle to r ide , with thousands of persons being ki 1 led and hundreds of thousands being injured annually. As ways were sought to reduce t h i s t o l l , i t began to appear tha t a s ignif icant proportion of motorcycle-car crashes came about because of problems the car driver had i n seeing and/or properly identifying motorcycles in certain types of potential confl i c t s i tua t ions . I t was thought that means for enhancing the conspicuity of motorcycles and/or motorcycl i s t s might prove an effect ive accident countermeasure.

CONTRACTOR The Regents of the University of Michigan Highway Safety Research Ins t i tu te

REPORT T I T L E Development and Testing of Techniques fo r Increasing the Conspi cui ty of Motorcycles and Motorcycl e Drivers

This project was i n i t i a t ed in 1976 by NHTSA t o develop and t e s t conspicuity treatments for both day and n i g h t use. Specif ica l ly , the following steps were undertaken:

CONTRACT NUMBER

DOT-HS-6-01459

REPORT DATE

October 5 , 1979

1. A review of motorcycle accidents. 2 . Development of potential conspicui ty treatments, 3. Field evaluation of the effectiveness of the treatments. 4 . Analysis, in te rp re ta t ion , and calculat ion of cost-

effectiveness.

The accident analysis was based on a sample of about 10,000 motor- cycle crashes which occurred in Texas i n 1975. These were compared with a like-sized sample of car accidents from the same period. The most important finding was t ha t certain types of pre-crash orientat ions are much more prevalent in motorcycle than in car crashes. The most common configuration for a motorcycle accident involves a s t r a i gh t traveling bike and l e f t turning car . While a l e f t - turn maneuver i s often d i f f i c u l t , the overrepresentation of lef t - turning cars in the motorcycle accident data suggests t ha t there i s indeed a conspicui ty problem.

Other points which emerged from the accident analysis were the following:

1. Persons wearing bright colored clothing seem under- represented.

2 . There i s some reason t o believe the conspicuity problem may not be a serious a$. niyQtt a g e s )

? c o n t i n u e on a d ltlo a

!!PREPARED FOR THE DEPARTMENT OF TRANSPORTATION, N A T I O N A L HIGHWAY T R A F F l C S A F E T Y ADMl N l S T R A T l ON UNDER C O N T R A C T NO.: DOT - HS - - 6 01459 . THE O P I N I O N S , F INDINGS, AND CONCLUSIONS EXPRESSED I N T H I S P U B L I C A T I O N ARE THOSE OF THE AUTHORS AND NOT N E C E S S A R I L Y THOSE OF THE N A T I O Y A L HIGHWAY T R A F F l C SAFETY A D M I N I S T R A T I O N . "

HS Form 321 J u l y 1 9 7 4

These data are in general agreement w i t h the resul ts of the in-depth motorcycle accident study carried out by Hurt and his colleagues for NHTSA.

Using various l ight ing techniques and high-vis ib i l i ty materials , more than 30 conspicuity treatments were developed for both day and night use. A sample of the most promising ones was selected for eval uation.

A gap-acceptance methodology was used in the evaluation stage. To do t h i s , instrumented motorcycles were placed in the t r a f f i c stream. The r iders attempted t o open a space (gap) between themselves and * leading t r a f f i c large enough t o tempt other vehicles waiting to cross the lane t o undertake the maneuver (accept the gap). Measures were made of gap s ize (both time and d i s tance) , whether the gap was accepted or not , and the type-of maneuver involved. T h e data were analyzed t o determine the probabil i ty of re la t ive ly shor t gaps being accepted. I t was anticipated t ha t improving conspicui t y should reduce the probabil i ty of very short gaps being accepted. The method worked very well .

The r e su l t s of the study indicate tha t daytime conspicuity can be s ign i f i can t ly improved by:

1. Causing the headlamp to modulate a t a r a t e of about 3 h z . 2 . Riding with the headlamp on. 3. Wearing high visi bi l i t y ( f luorescent) garments.

I t was especial ly in teres t ing t o note t ha t the same high-visi- b i l i t y materials , when attached to the bike, were not as e f f ec t i ve ,

The nighttime resul ts were not as c lea r cut . However, i t appears tha t the use of running l i gh t s and re t ro re f lec t ive garments may be beneficial . Again, the same re t ro re f lec t ive materials attached t o the bike were not as e f fec t ive .

Another part of the study examined lane position ( r i g h t , l e f t , or center) as a fac tor in gap acceptance. The probabil i ty of short gaps being accepted was lowest fo r the center lane posit ion, and next lowest for the l e f t lane posit ion.

A t e s t was also run t o examine the e f f e c t of having a car follow- ing behind the motorcycle. The resu l t s suggest t ha t the presence of a car close behind a motorcycle (one second headway in t h i s case) may reduce the probabil i ty of shor t gaps in f ront of the motorcycle being accepted fo r one of the maneuvers studied.

Limited t e s t s were a lso run w i t h a moped, using a s ingle t r e a t - ment. While the resu l t s suggest a beneficial e f f ec t associated w i t h the use of a f luorescent f l ag , the response charac te r i s t i c s o f the automobile drivers were so much d i f fe ren t t o the mopeds than t o the

moto rcyc les t h a t i t i s f e l t t h e method may n o t be a p p r o p r i a t e f o r such s low moving v e h i c l e s .

The r e s u l t s o f t h i s s t u d y i n d i c a t e t h a t m o t o r c y c l e c o n s p i c u i t y can be improved i n a c o s t - b e n e f i c i a l way u s i n g any o f a v a r i e t y o f t echn iques .

ACKNOWLEDGEMENTS

This was a complex study, the smooth and successful completion

of which was possible only with the cooperation of a large number of

people. I n par t icular , the authors would l ike t o express their great

appreciation t o the Contract Technical Monitor, Dr. Robert Henderson,

who provided valuable guidance and assistance throughout the project. '

The authors would also 1 ike t o express their appreciation t o the young men and women w h o assisted in data collection. Special thanks

go to Michael and Patrick Barton and Michael Gladden, w h o not only

aided in data collection, b u t recruited and trained other riders and

helped t o maintain the bikes.

Finally, we were greatly aided by Mr. Robert Vanstrum of 3M

Company, who supplied most of the materials used in the t e s t , as well as a great deal o f technical information.

TABLE OF CONTENTS

1.0 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 Background . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 The Presen t Study . . . . . . . . . . . . . . . . . . . . 1

. . . . 1.3 P rev ious I n v e s t i g a t i o n s o f M o t o r c y c l e C o n s p i c u i t y

2.0 A REVIEW OF MOTORCYCLE ACCIDENT DATA . . . . . . . . . . . . . 7

2 .1 Q u e s t i o n s t o be Addressed . . . . . . . . . . . . . . . . 7

2 .2 P rev ious M o t o r c y c l e A c c i d e n t Analyses . . . . . . . . . . 8

2.3 HSRI M o t o r c y c l e A c c i d e n t A n a l y s i s . . . . . . . . . . . . 15

2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 26

3.0 SEEING AND REACTING TO MOTORCYCLES . . . . . . . . . . . . . . 27

3 .1 I n t r o d u c t i o n . . . . . . . . . . . . . . . . . . . . . . 27

3.2 The F u n c t i o n a l I m p l i c a t i o n s o f t h e S t r u c t u r e of t h e R e t i n a : Foveal vs . P e r i p h e r a l V i s i o n . . . . . . . . 28

3 .3 C r i t i c a l Cues . . . . . . . . . . . . . . . . . . . . . . 35

3.4 O the r D r i v e r Dec is ion-Mak ing about Mo to rcyc les . . . . . 44

3.5 Conc lus ions . . . . . . . . . . . . . . . . . . . . . . . 45

. . . . . . . . . . . . . . . . . . . . 4.0 FIELD TEST METHODOLOGY 47

. . . . . . . . . . . . . . . . . . . . . . 4.1 I n t r o d u c t i o n 47

4.2 C r i t e r i o n . . . . . . . . . . . . . . . . . . . . . . . . 47

4.3 Equipment . . . . . . . . . . . . . . . . . . . . . . . . 50

4.4 Data Reduct ion Procedure . . . . . . . . . . . . . . . . 65

4.5 Data A n a l y s i s Procedure . . . . . . . . . . . . . . . . . 67

5.0 RESULTS AND DISCUSSION . . . . . . . . . . . . . . . . . . . . 71

5 .1 Daytime Treatments . . . . . . . . . . . . . . . . . . . 7 1

5.2 N i g h t t i m e Treatments . . . . . . . . . . . . . . . . . . 83

5.3 M o t o r c y c l e L a t e r a l P o s i t i o n . . . . . . . . . . . . . . . 87

5.4 E f f e c t o f F o l l o w i n g Veh ic les . . . . . . . . . . . . . . 9 1

5.5 Moped Consp icu i t y . . . . . . . . . . . . . . . . . . . . 104

5 .6 D iscuss ion . . . . . . . . . . . . . . . . . . . . . . . 104

TABLE OF CONTENTS

6.0 COST-BENEFIT ANALYSIS . . . . . . . . . . . . . . . . . . . 115

6.1 Assumptions . . . . . . . . . . . . . . . . . . . . . 115

6.2 Cost -Benef i t Anal. y s i s Procedure . . . . . . . . . . . 115

6.3 Cos t -Bene f i t A n a l y s i s R e s u l t s . . . . . . . . . . . . 118 *

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

v i i i

LIST OF TABLES

2-1. Percentages of mu1 t ip1 e-vehicl e coll is ion configurations observed in Pol anis ' data on motorcycle and a1 l crashes in Pennsylvania during 1975 . . . . . . . . . . . . . . . . . 10

2-2. Polanis ' analysis of day and non-day motorcycle . . . . . . . . . . . . . . . . . . col l i s ion configurations 12 *

2-3. Pre-crash maneuvers in 6647 motorcycle-car crashes--Texas 1975. Frequencies are above percentages in each cel l . . . . 16

2-4. A cornp-arison of marginal percentages of motorcycle and car pre-crash maneuvers with the car-car s i tua t ion . . . . . . 18

2-5. Re1 a t ive travel direction in 1975, Texas motorcycle-car accidents involving on vehicle going s t r a i g h t , the other turning l e f t . X5 = 130.6, df = 2 , p < .001. Frequencies are above percentages in each cel l . . . . . . . 19

2-6. Re1 a t i ve travel di rection in 1975, Texas car-truck accidents involving one vehicle going s t r a i gh t and the other turning l e f t . ~2 = .818, df = 2, p > .5. Frequencies are above percentages in each cel l . . . . . . . 19

2-7 . Time-of-day versus number of vehicles in 1975, Texas . . . . . . . . . . . . . . . . . . . . motorcycle accidents 21

2-8. Light conditions in motorcycle and car co l l i s i ons , Texas motorcycle accidents, 1975 . . . . . . . . . . . . . . 22

2-9. Daytime l e f t - turn motorcycle-car coll i s ions, Texas motorcycle accidents, 1975 . . . . . . . . . . . . . . . . . 23

2-10. Non-day 1 ef t - turn motorcycle-car coll is ions , Texas . . . . . . . . . . . . . . . . . motorcycle accidents, 1975 23

2-11. Color of upper garment worn by 1975 Texas motorcyclists involved in single-vehicle crashes, c lass i f i ed by time-

. . . . . . . . . . . . . . . . . . . . . of-day of accident 24

2-12. Color of upper garment worn by 1975 Texas motorcyclists involved in two-vehicle crashes, c lass i f i ed by time-

. . . . . . . . . . . . . . . . . . . . . of-day of accident 25

3-1. Types of c r i t i c a l cues fo r identifying a motorcycle . . . . . 36

LIST OF TABLES

3-2. Retinal image s ize changes from a fu l l - s i z e automobile and motorcycle c los ing, head-on from 500 f e e t a t 30 mph. (44 f t / s e c . ) . The automobile image i s measured in horizontal plane a n d i s based on an 80" width. The motorcycle image i s measured in the

. . . . . ver t ica l plane and i s based on a 5 ' height 4

4-1. Coding functions f o r t r ans la t ing d a t a from motor- c y c l i s t ' s action t o t r i a l codes . . . . . . . . . . . . . . . 66

5-1. Rejected-gap analysis fo r r igh t - cross o r l e f t turn . . . . . . . . . . . . . . . . maneuvers--daytime condi t i ons 74

5-2. Rejected-gap analysis fo r center - l e f t turn maneuver--daytime conditions . . . . . . . . . . . . . . . . 79

5-3. Rejected-gap analysis f o r r igh t - r igh t turn . . . . . . . . . . . . . . . . . maneuver--daytime conditions 82

5-4. Rejected-gap analysis fo r r i gh t . - cross or l e f t turn--nighttime conditions . . . . . . . . . . . . . . . . . 85

5-5. Rejected-gap analysis f o r the center - l e f t turn--nighttime conditions . . . . . . . . . . . . . . . . . 88

5-6. Rejected-gap analysis fo r r i gh t - r igh t turn--nighttime . . . . . . . . . . . . . . . . . . . . . . . . . conditions 90

5-7. Rejected-gap analysis fo r r i gh t - cross or l e f t turn maneuver--1 ateral- lane-posi t ion analysis . . . . . . . . . . 95

5-8. Rejected-gap analysis for center - 1e.ft turn maneuver--1 ateral- lane-posi t ion analysis . . . . . . . . . . 96

5-9. Rejected-gap analysis fo r r i gh t - r i gh t turn maneuver--lateral-lane-position analysis . . . . . . . . . . 97

5-10. Rejected-gap analysis f o r r igh t - cross or l e f t turn . . . . . . . . . . . . . . maneuver--car-foll owing analysis 101

5-11. Rejected-gap analysis fo r center - l e f t turn maneuver--car-following analysis . . . . . . . . . . . . . . 102

5-12. Rejected-gap analysis f o r r igh t - r i gh t turn . . . . . . . . . . . . . . maneuver--car- fol lowing analysis 103

5-13. Rejected-gap analysis f o r r igh t - cross o r l e f t turn maneuvers--moped study . . . . . . . . . . . . . . . . . . . 108

LIST OF TABLES

5-14. Rejected-gap analysis for center - l e f t turn maneuvers--moped study . . . . . . . . . . . . . . . . . 109

5-15. Rejected-gap analysis fo r r ight - r igh t turn maneuvers--moped study . . . . . . . . . . . . . . , . . 110

6-1. Probability of par t icular types o f in ju r ies occurring as r e su l t o f a motorcycle accident . . . . . . 117 '

6-2. Total average losses fo r motorcycle accidents . . . . . 117

6-3. Expected crash costs . . . . . . . . . . . , . . . . . . 119

6-4. Operating costs o f cons pi cui ty enhancers . . . . . . . . 120

6-5. Benefits per conf l i c t of conspicuity treatments . . . . 121

LIST OF FIGURES

4.1 . Schematic of t yp i ca l gap s i t u a t i o n employed in t e s t . . . . . 48

4.2 . Hypothet ical gap acceptance d i s t r i b u t i o n . . . . . . . . . . 49

4.3 . Fa i r i ng w i t h f l u o r e s c e n t cover i n p lace . . . . . . . . . . 51

4.4 . Motorcycles used i n t e s t program . . . . . . . . . . . . . . 5%

. . . . . . . . . . . . . . 4.5 . Magnetic s enso r and wheel marker 56

4.6 . Photograph of c o n t r o l s a r e a . showing push bu t tons used f o r coding da ta . . . . . . . . . . . . . . . . . . . . 58

4.7 . Schematics o f the t h r e e maneuvers i n v e s t i g a t e d i n . . . . . . . . . . . . . . . . . . . the p r e sen t s tudy ; . . 59

4.8 . Rear o f motorcycle s add l e . shuwi ng r eco rde r . . . . . . . . . . . . . . . . . . . . . i n po s i t i on . . . . 61

. . . . . . . . . . . . . . . . . . . . . . 4.9 Sample da t a t r a c e s 62

5.1 . Diagram of a r i g h t . c ro s s o r l e f t turn maneuver . . . . . . 72

5.2 . Probi t r e s u l t s f o r daytime conditions.. r i g h t . c r o s s or l e f t turn maneuver . . . . . . . . . . . . . . . . . . . 73

. . . . . . . . . . 5.3 . Diagram o f a c e n t e r . l e f t turn maneuver 76

5.4 . Probi t r e s u l t s f o r daytime conditions.. c e n t e r . . . . . . . . . . . . . . . . . . . . . l e f t turn maneuver 77

. . . . . . . . . . 5.5 . Diagram of a r i q h t . r i g h t turn maneuver 80

5.6 . Probi t resu l t s f o r daytime condi t i ens.. r i gh t . . . . . . . . . . . . . . . . . . . . . . r i g h t turn maneuver 81

5.7 . Probi t r e s u l t s f o r n igh t t ime conditions.. r i g h t . . . . . . . . . . . . . . . . . . c r o s s o r l e f t turn maneuver 84

5.8 . Probi t r e s u l t s f o r n igh t t ime conditions.. c e n t e r . . . . . . . . . . . . . . . . . . . . . . l e f t turn maneuver 86

5.9 . Probi t r e s u l t s f o r n igh t t ime conditions.. r i g h t . . . . . . . . . . . . . . . . . . . . . . r i g h t turn maneuver 89

5.10 . P r o b i t r e s u l t s f o r l a t e r a l - l a n e - p o s i t i o n . . . . . . . . analysis.. r i g h t . c ro s s o r l e f t turn maneuver 92

LIST OF FIGURES

5-11. Probi t resul t s for 1 ateral -1 ane-posi tion analysis--center - l e f t turn maneuver . . . . . . . . . . . . 93

5-12. Probi t results for lateral -lane-posi tion analysis--right - right turn maneuver . . . . . . . . . . . . 94

5-13. Probi t results for fol 1 owing-car analysi s--right - I

cross or l e f t turn maneuver . . . . . . . . . . . . . . . . . 98

5-14. Probi t results for fol lowing-car analysi s--center - l e f t turn maneuver . . . . . . . . . . . . . . . . . . . . . 99

5-15. Probi t results for fol 1 owi ng-car analysi s--ri g h t - right turn maneuver . . . . . . . . . . . . . . . . . . . . . 100

5-16. Probi t resul t s foi moped analysis--ri g h t - cross or l e f t turn maneuver . . . . . . . . . . . . . . . . . . . . 105

5-17. Probit results for moped analysis--center - l e f t turn maneuver . . . . . . . . . . . . . . . . . . . . . . . . 106

5-18. Probit results for moped analysis--right - right turn maneuver . . . . . . . . . . . . . . . . . . . . . . . . 107

1.0 INTRODUCTION

1.1 Background

Motorcycles have recently enjoyed phenomenal growth both as

forms of basic transportat ion and in recreation. Since 1961 motor-

cycle regis t ra t ions have increased a t nearly four times the ra te of a1 1 other vehicles. In 1977 U.S. motorcycle regis t ra t ions passed the' f ive mil 1 ion mark. Also in 1977, nearly ten percent of these motor-

cycles were involved in accidents. These mishaps caused about 4,000

deaths and about 400,000 in ju r ies (Carraro, 1979).

The principal: safety problem w i t h motorcycles i s the vul ner- a b i l i t y of the persons w h o r ide them. When a motorcycle becomes

involved in a crash, i t s occupants are about ten times more l ikely to

suffer injury than i f they were in a car (Carraro, 1979).

The engineering solutions which have he1 ped reduce crash-re1 ated

dangers in automobiles ( e . g . , sea t be l t s ) cannot be applied to motor-

cycles without s ignif icant ly changing t h e i r nature. Thus, motorcycle

r iders wi 11 remain vulnerable. Consequently, the injury ra te per

crash will l ike ly remain high for the foreseeable future.

Some trends in the motorcycle accident data suggest means to reduce the frequency of motorcycle crashes. For example, there i s reason to believe t ha t a s ignif icant number of motorcycle-car crashes are conspicuity re la ted ; i . e . , car d r ive rs , for whatever reason, often

behave as though they do not see motorcycles. I f motorcycles are hard to see , making them more conspicuous should reduce the incidence of

such crashes.

1.2 The Present Study

The National Highway Traffic Safety Administration (NHTSA)

recognized the safe ty potential of making motorcycles, motorcycl i s t s , and mopeds more conspicuous. In 1976, NHTSA awarded a contract t o the Highway Safety Research In s t i t u t e (HSRI) of the University of Michigan

to discover and t e s t ways of making motorcycles and motorcycl i s t s more

conspicuous. Specif ica l ly , the original project aimed t o complete the following s i x a c t i v i t i e s :

I . Examine motorcycle accident data to determine:

a the importance of conspicuity problems, and

a potential means fo r making motorcycles more conspicuous.

2. Review past s tudies of motorcycle conspicuity and conspicuity in general .

3. Survey relevant basic and applied research re la ted t o visual perception.

4. Recommend nighttime and daytime conspicuity treatments tha t might be appl ied t o the bi ke and/or r ide r .

5 . Eva1 uate the proposed treatments under real i s t i c f ie1 d

condi t ions . 6 . Analyze a n d i n t e rp r e t the data resul t ing from s tep 5 ,

including consideration o f costs and benef i ts .

I n 1978, the project was expanded t o cover additional conspicuity treatments, motorcycle lane positioning e f f e c t s , and moped conspicui ty

in the f i e l d t e s t ing a c t i v i t i e s .

1.3 Previous Investigations of Motorcycle Conspicui ty

Several other investigations of motorcycle conspicui t y preceded t h i s one.

Reiss and Haley (1968) f i r s t iden t i f i ed conspicuity as a s igni- f i can t problem area in motorcycle-automobil e coll i s ions . They s t a t e ( p p . 2-3) :

"A 1 arge number of motorcycl e-automobi 1 e coll i s ions are re la ted to the f a c t t ha t the motorist does not see the motorcycle unti l too l a t e . . . "

Their analysis of the potential cost-effectiveness of motorcycle countermeasures shows t ha t those dealing w i t h conspicuity rank t h i rd

behind motorcycl i s t 1 icensing and helmet use. Subsequent problem ident i f ica t ion projects (e .g . , Reiss, Berger, & Val l e t t e , 1974) pro- duced s imi lar conclusions.

Wol tman and Austin (1973) carried out a systematic analysis aimed a t understanding the cons p i cui ty problem. They compared the s ize of the surface area of motorcycles with other vehicles and pedestrians. Their analysis suggests the conspicui ty problem stems from the re la t ive ly small s ize of motorcycles in comparison with other vehicles.

One simple way of making motorcycles more conspicuous during the day i s to turn on the headlamp. In the l a t e 19601s, many s t a t e s

passed "1 ights-on" laws, mandating the daytime use of headlights on motorcycles, and several s tudies tested the e f fec t of t h i s counter-

measure. The Frank1 in I n s t i t u t e (Janoff , Cassel , Fartner, and Smierciak, 1970; Janoff & Cassel , 1971; Janoff , 1973) evaluated how effect ively daytime headl i g h t use increased motorcycle conspicui t y . They examined accident data , comparing s t a t e s with and without 1 ights- on laws, and took f i e l d measurements of motorcycle conspicuity, as

reported by other motorists. The resu l t s of both types of studies indicate a posit ive conspicui ty benefit for daytime headl ight use.

Subsequent analyses of accident experience in s t a t e s with and without lights-on laws (Waller & Gri f f in , 1977), and the daytime conspicui ty o f a motorcycle with i t s headl i g h t on ' ( ~ a m s e ~ & Brinkley, 1977, Will iam & Hoffman, 1977) provide additional evidence t ha t 1 ights- on enhances conspi cui ty .

Turn signals m i g h t be used as running lamps t o improve conspicuity as we1 1 . Some evidence, reported by Bart01 e t a1 . (1975), indicates running l igh t s increase nighttime conspicuity. A recent t e s t of the daytime conspicuity of running l i g h t s , by a panel of SAE experts, indicates potential benefi ts , b u t the evidence favoring th i s counter- measure i s ra ther weak.

A r ecen t s tudy by Ramsey & B r i n k l e y (1977) examined the consp i -

c u i t y o f s t r o b e and f l a s h i n g l i g h t s . The researchers found t h a t

o t h e r m o t o r i s t s , when stopped and asked whether they saw a moto rcyc le

they had j u s t passed, n o t i c e d motorcyc les equipped w i t h a medium- t o

h i g h - i n t e n s i t y f l a s h i n g l i g h t more o f t e n than motorcyc les w i t h no o r

l o w - i n t e n s i ty 1 i g h t s . *

I n a d d i t i o n t o l i g h t s , o t h e r means a re a v a i l a b l e t o improve

c o n s p i c u i t y . For example, Wol tman & A u s t i n (1973) suggested

t h a t f l uorescent and r e t r o r e f l e c t i ve garments and accessor ies m igh t

inc rease moto rcyc le /moto rcyc l i s t consp icu i ty . Stud ies performed by

Burg and Beers (1976), W i l l i ams and Hoffman (1977), and Bar t01

e t a l . (1975) demonstrated t h a t d e t e c t i o n and i d e n t i f i c a t i o n d i s -

tances inc rease w i t h s p e c i a l f l u o r e s c e n t and r e t r o r e f l e c t i v e p a i n t s

and garments . K i r k b y and St roud (1978) used a gap-acceptance methodology t o

eva lua te t he e f f e c t i v e n e s s o f l i g h t s - o n and h i g h v i s i b i l i t y c l o t h i n g

on t h e behav io r of m o t o r i s t s . The s tudy was s e t up on a t r a f f i c

c i r c l e and da ta taken on v ideotape from a c e n t r a l l o c a t i o n t o measure

changes i n pe rcen t acceptance of d i f f e r e n t gaps between t he t e s t

moto rcyc le and a l ead c a r . The au thors found no d i f f e r e n c e s between

the t r e a t e d and c o n t r o l c o n f i g u r a t i o n s . Wi l l i ams and Hoffman (1977) compared t he consp icu i t y o f h i g h -

v i s i b i l i t y m o t o r c y c l i s t c l o t h i n g , f a i r i n g s , and headl i g h t s . They

found, i n a s e r i e s o f l a b o r a t o r y s t ud ies us i ng s l i d e s o f road scenes

w i t h o r w i t h o u t a moto rcyc le p resen t , t h a t headlamp use, a f l uo rescen t

j a c k e t , and f a i r i n g inc reased moto rcyc le d e t e c t a b i l i ty over an

un t rea ted c o n t r o l .

I n sum, data f rom a number o f s t u d i e s , us ing a v a r i e t y o f tech-

n iques, suggest severa l means f o r improv ing moto rcyc le consp icu i t y .

These are:

e dayt ime use o f headl i g h t s

e runn ing l i g h t s

0 f lashing l i gh t s e fluorescent garments or paints

re t ro re f l e c t i ve devices and garments 9 fa i r ings

Only the daytime use of headlights can be said to have been extensively tes ted under real-worl d conditions . One of the problems

I

w i t h the other studies stems from an apparent assumption tha t poor conspicui ty equals " fa i lu re to see ." Actually the trends in the crash data which will be reviewed in the next section could be explained by f a i l u r e to see and any or a1 1 of the following:

0 misidentif icat ion as a low-performance vehicle

0 er ro rs in speed-spacing judgments

e del ibera te ly perverse behavior on the part of the car dr iver .

I t i s not l ike ly tha t the primary problem ( i f there i s a primary problem) can be ident i f ied in a single study. However, the work to be described examines the problem from several points of view and

offers some in teres t ing insights regarding possible explanations.

The r e s t of t h i s report i s directed to a description of the methods, f indings, conclusions, and recommendations of t h i s project .

2.0 A REVIEW OF MOTORCYCLE ACCIDENT DATA

2.1 Questions t o be Addressed

For operational purposes 1 e t us define the motorcycle conspicui t y problem as occurring :

. . .whenever another motorist f a i 1s to s ee , iden t i fy , @

or appropriately react to a legal ly used motorcycle, o r , in the case of coll iding with a motorcycle, claims such a f a i l u r e resulted in a crash.

Given t h i s statement, the primary question which i t i s hoped motorcycle crash data might answer i s the following:

1 Do differences between patterns of motorcycle-car and car-car crashes suggest other drivers sometimes f a i l t o see o r ident i fy motorcycles?

I f the crash data suggest motorcycles are d i f f i c u l t to see and ident i fy , four additional questions become meaningful :

1. Do data suggest a day-night difference in motorcycle conspicui ty?

This question stems from an in tu i t ive ly expected difference. During the daytime, drivers look for and see other vehicles d i rec t ly . B u t , in the nighttime, headlamps and t a i l l i g h t s a re the dominant cues 'for seeing and identifying other vehicles. Thus, motorcycles should be re la t ive ly more v i s ib le a t night.

2 . Do data suggest crash involvement varies with the conspicuity of the motorcycl i s t ?

This question a r i ses from the expectation tha t more cons pi cuous motorcycles should be involved in re la t ively fewer crashes. I n search- ing for an answer t o t h i s question, we examined the color of clothes crash-invol ved motorcyclists were wearing and s imi lar variables.

'we wi 11 use the term "car" t o re fe r to autornobi 1 es , t rucks , e t c , , and "motorcycles" to re fe r to motorcycl es and mopeds.

3. What are the most prevalent motorcycle-car crash

configurations?

An answer to this question will aid in tracing the roots of the conspicui ty problem. Finding that specific crash configurations are more prevalent in motorcycle-car than in car-car crashes might suggest angles a t which motorists have diff icul ty in seeing motor- cycles, and other driver errors motorcyclists find d i f f icu l t t o defend against.

4 . How severe i s the conspi cui ty problem; i .e. , how many motor- --.

cycle crashes involve the conspicui ty problem?

This final question provides a measure of the extent of the conspicuity problem. We need an estimate of how prevalent the problem i s , a t least in crashes, in order t o effectively analyze the costs and benefits o f potential countermeasures.

2 . 2 Previous Motorcycle Accident Analyses

There have been several analyses of motorcycle accidents pub1 ished. These are reviewed below.

2.2.1 Analyses 3f U.S. Motorcycle Accidents. In the U.S., motorcycle accident analysis started in the la te 1960's. The work has covered specific issues (e .g . , the work of Waller and Griff in , 1977, on lights-on laws), as well as providing detailed, in-depth and broad spectrum analyses of motorcycle accidents (such as the work currently being carried out by Hurt and his colleagues a t the University of Southern Cal i fornia) . Many o f these analyses provide information useful for answering the questions of current interest .

2 .2 .1 .1 Differences Between Motorcycl e-Car and Car-Car Crashes. We intuit ively expect that automobiles and trucks are more conspicuous than motorcycles. I f so, we should observe differences between motorcycle-car and car-car crashes.

Two prior analyses compared motorcycle-car and car-car accidents. Harano and Peck (1968) compared car-car and motorcycle-car crashes

occurring in Cal i fornia during 1966. Polanis (1979) compared

motorcycle-car crashes with car-car crashes occurring i n Pennsylvania in 1975.

Harano and Peck f i r s t noticed differences between motorcycl e-car and car-car crashes and found tha t mu1 t i -vehicl e motorcycle accidents occur more often a t intersections than do accidents invol ving only

t

cars .

Polanis' data provide a more detai led look a t the differences in mu1 t i p l e vehicle coll i s ions . Table 2-1 contains the col l i s ion

configuration percentages he observed. This indicates motorcycle-

car col l i s ions are more often head-on or angle col l i s ions in compari-

son to car-car co l l i s ions .

In sum, these crash data show two differences between motorcycle- car and car-car crashes. F i r s t , motorcycle-car crashes tend to occur

a t in tersect ions more so than car-car crashes. Second, motorcycles

are more l ike ly involved in angle col l i s ions and head-on co l l i s i ons , which implies the other driver primarily viewed the motorcycle's f ront

s i lhoue t te . The f ront s i lhouet te i s small and harder to see than the

s ide p rof i l e . These differences between car-car and motorcycle-car coll is ions suggest tha t there i s a conspicuity problem.

2 . 2 . 1 . 2 Day-Night Differences in Conspicui t y . In tu i t ive ly ,

motorcycles should be more conspicuous in the nighttime than in the daytime, since other drivers will be looking for vehicle l i gh t s . I f

so , conspicui ty-related crashes should decrease in the nighttime.

Harano and Peck (1968) found a larger percentage of motorcycle accidents occur i n the daytime than non-motorcycl e accidents ( 7 2 %

motorcycle accidents versus 66% non-motorcycl e accidents) , A higher percentage of daytime motorcycle (or motorcycl e-car) accidents in

comparison to daytime car (or car-car) accidents also emerges in

studies by Griff in (1974) and Waller & Griffin (1977).

The more deta i led analysis of Polanis provides an indication that motorcycles are less conspicuous in the day than in the night , Table

TABLE 2-1. Percentages of mu1 t ip le-vehic le coll is ion configurations observed in Polanis ' d a t a on motorcycle and a l l crashes in Pennsyl vania duri ng 1975.

t % of motorcycle accidents in t h i s

COLLISION CONFIGURATION 1

I Head-on , Angle

configuration I I

Backing Up

1.2

2.3

- 5 2

1.8

% of a l l accidents I 1.0 in t h i s configura- / t ion

5 5 . 3

40.2

Side Swipe

1.4

2.8

. 5

I Rear-end 1

9 . 2

I

19.9 1 1

-46 !

I i

1 1.8 1

Percentage r a t i o , i . e . , motorcycle-car/ I car-car I

1.38

2-2 shows that the angle collision i s overrepresented in the daytime data. All the res t are virtually even odds or associated with the nighttime.

These analyses of day-night differences in crash involvement pro- vide some evidence that motorcycles are less conspicuous in the day- time. A larger percentage of motorcycle accidents occur in the daytime than non-motorcycle accidents. Motorcycle col 1 i s ions, where the othe;

motorist views the motorcycle head-on or from an angle occur predomi- nantly in the daytime. Unfortunately, these data are n o t as definit ive as they might appear, because exposure data are 1 acking .

2.2.1.3 Does Crash Involvement Vary with the Conspicui t y

of the Motorcyclist? Very few data have been collected on the conspicui ty of accident involved motorcyclists. The only previous examination i s reported by Tratner (1978) in an interview with Hurt, the principal investigator of the USC motorcycle accident investiga- tion project. I n the interview, Hurt points o u t that ( p g . 43) :

" . . . there i s a spectacular absence of high v is ib i l i ty upper torso garments in a11 the accidents we tabulated. I n the intersection accident, where conspicui ty i s c r i t i c a l , we found lack of conspicuous upper torso garments.. . . You don't see the guy riding down the s t r ee t in the bright orange or day-glow orange or yellow jacket involved in an accident. I t i s the guy in the army surplus jacket."

Although one might argue that only the more cautious motor- cyclists will wear highly conspicuous garments of their own volit ion, the finding indicates a strong possibil i t y t h a t conspicuous motor- cyclists are less apt t o be involved in a collision because other drivers can more easily see them.

2.2.1.4 Motorcycle-Car Crash Configurations. Two studies have examined the configuration of motorcycl e-car crashes. Griffin (1974) examined the pre-crash maneuvers of b o t h vehicles in 1 ,267

motorcycle-car crashes. Reiss, Berger, a n d Val l e t t e (1974) examined 100 Maryland motorcycl e-car crashes occurring a t urban intersections.

TABLE 2-2. Po lan is ' ana lys is o f day and non-day motorcyc le c o l l i s i o n con f i gu ra t i ons .

Both studies obtained similar resul t s . The motorcycle most often travels s t raight (78% in Gr i f f in ' s data and 87% in the data of Reiss

e t a1 . ) and the car i s e i ther going s t raight (39% in both data bases) or turning l e f t (44% i n Griff in 's data and 51% in the data of Reiss e t a1 . ) . Both analyses show the other motorist i s maneuvering most of the time (about 60%) while the motorcycle i s traveling s t raight most of the time (about 80%). t

The crash configuration of a motorcycle traveling s t raight and a car maneuvering provides additional evidence t h a t a conspicui ty problem exis t s . If no conspicui ty problem existed, we would expect l i t t l e difference in pre-crash direction of travel between cars and motorcycles. Furthermore, the evidence suggests that special e f for t should be directed a t the conspicuity of motorcycles t o the front .

2.2.1.5 Culpability in Motorcycle Crashes. Both Waller (1972) and Reiss, Berger, and Vallette (1974) examined other motorist culpabil i ty in motorcycle-car crashes. Culpabil i ty can serve as a surrogate measure of conspicui t y as i t implies a fa i lure t o see,

identify, or respond appropriately t o a motorcycle. Wal l e r ' s analysis o f the 630 mu1 tivehicl e motorcycle accidents reported in North Carolina in 1968 concluded that 62.2% were caused by the other driver. Reiss, Berger, and Vallet te 's analysis of a sample of 400 1973 Mary1 and m u 1 tivehicle motorcycle accidents concluded that the other driver was culpable in 61.1% o f these accidents,

2 . 2 . 2 Foreign Motorcycle Accident Experience. Foreign motor- cycle analyses have primarily examined configurations of motorcycle- car crashes. Of these, we have chosen one from Australia (Smith,

1975) and one from Japan (Nagayama, e t a l . , 1979) because of the detail they provide.

Srni t h examined configurations of motorcycle-car crashes in Australia along with the errors made by the motorcycl i s t and car operator. His da ta indicate the car driver most often erred i n turn-

ing r ight (Australians drive on the l e f t ) or in f a i l ing t o yield a t an intersection. These results closely corresbond t o those of Griff in , Berger, e t a1 . , and Pol an i s for the U .S.

A s imi lar p rof i l e i s observed i n the Japanese data of Nagayama, Mori t u , Miura, Watanabem, and Murakami (1979). Japanese motorcycle- car crashes involve s t r a i gh t traveling motorcycles (86%) and r igh t turning (Japanese drive on the l e f t as well) (20.9%) or s t r a i gh t traveling (14.3%) cars .

2.2.3 Summary and Conclusions. Both U.S. and foreign motorcycle i

accident analyses point to conspicuity as a real problem. A compari- son of motorcycle-car and car-car crashes shows differences suggesting other drivers have d i f f i cu l ty seeing and identifying motorcycles. Since simi 1 a r patterns are observed worl d-wi de, the conspicui ty problem 1 i kely does not stem from lack of experience in deal ing with motorcycles on the part of U.S. d r ive rs ,

Specif ica l ly , these patterns emerge from the data:

e A higher percentage of motorcycle-car crashes occur a t in tersect ions than car-car crashes.

e A higher percentage of motorcycle-car crashes are angle or head-on col 1 is ions than car-car crashes.

e Motorcycle-car accidents tend t o occur during the day.

Motorcycle-car accidents, where the vehicles col 1 i de a t an angle, occur predominately in daytime.

e The most prevalent motorcycle-car crash confi gura- t ions involve a s t r a i gh t traveling motorcycle and a car which i s

- turning l e f t , or

- t raveling s t r a i gh t

e The car driver i s culpable in about 60% of a l l car- motorcycle crashes.

9 Motorcycl i s t s wearing highly conspicuous upper garments are underrepresented in the accident population.

2.3 HSRI Motorcycle Accident Analysis

This analysis was carried o u t using computerized data f i l e s a t

HSRI. The aim was to secure additional detailed information which

might aid in identifying c r i t i ca l si tuations a n d possibly counter- measures.

2.3.1 Method. A1 1 motorcycle crashes occurring in Texas during,

1975 were selected for examination. This sample contains about

10,000 crashes and represents about 10% of a l l U . S . motorcycle crashes

in 1975. A second sample of 5% of a l l motor vehicle crashes occurring

in Texas in 1975 was used in selected comparisons

A two-stage analysis procedure was employed: in the f i r s t

stage, percentage profi 1 es of motorcycle accidents were developed,

e.g. , how many occurred in the daytime versus the nighttime. The

second stage selectively compared the motorcycle crash profi 1 es where

the comparison would provide information on conspicui ty . 2.3.2 Results.

2.3.2.1 Motorcycl e-Car Crashes Compared with Car-Car

Crashes. Table 2-3 breaks down the pre-crash vehicle maneuvers in

the 6,467 motorcycle-car crashes in Texas during 1975. ("Car"

includes b o t h automobiles and trucks.) This matrix i s very similar

to Griff in 's data (1974). The marginal percentages show that rnotor- cycles are traveling s t raight in about 87% of these crashes whereas

other vehicles were traveling s t raight in only about 46% of the

crashes. [Griffin reports figures of about 78% and 39%, respec- t ively. )

Two ce l l s dominate this analysis, I n b o t h of these ce l l s the

motorcycle i s traveling s t raight and in one the car i s also traveling

s t raight and in the other i t i s turning l e f t . Left turns or s t raight

crossings can be d i f f i cu l t maneuvers, providing an opportunity for

conflict with other t r a f f i c in one or b o t h directions.

TABLE 2-3. Pre-crash maneuvers in 6647 motorcycle-car crashes--Texas 1975. Frequencies are above percentages i n each cel l .

I

I Right Turn 430 468 1

I , I

i Motorcycle ' s Maneuver

I Left Turn 15 ; 3 2 1 0 8 1 0 / 2443 35.9 0 . 2 : 0 . 5 i 0 .0 0 . 0 0 . 0 ! 36.7 I 2388 1 i

I

Car's Maneuver

1 Straight

Straight j 2378

Right / Left 1

i I I

I Turn / Turn Back / Stopped Parking T o t a l

I 130 1 347 4 216 ; 4 , 3 0 8 d

--

Stopped

I 35.8 2 . 0 5 .4 0.0 i 3 .2 0 . 0 1 46 .4 i

Back 88 1 1 .3

Parking

0 O / O I 25 1 0 1 113 1

0 . 0 , 0 . 0 0.0 0 .4 i 0 . 0 1 1 . 7 1 ,

i I I i Column Total 5794 i 185 407 ' 4 253 1 4 i 6647 j 87.2 2 . 8 1 6 .2 , 0 .0 3 .7 0 . 0 9 5 . 9 1

Table 2-4 compares marginal percentages of the pre-crash maneu- vers in motorcycle-car crashes with the car-car s i tuat ion. The car pre-crash maneuvers are very similar in both s i tuat ions, except for the left-turns and stopped pre-crash s ta tes : Cars are more l ikely t o

be turning 1 e f t in motorcycl e-car crashes than in car-car crashes ;

cars are more l ikely to be stopped in car-car crashes than in motor- cycle-car crashes.

#

Table 2-5 shows the relat ive direction of travel in motorcycle-

car crashes involving a l e f t turn. A chi-square t e s t indicates a dependency <i thin the matrix, which, by inspection, stems from the large number of crashes involving the car turning l e f t a n d the vehi-

cles travel ing in the opposite direction.

Table 2-6 shows a similar analysis of 1974 Texas automobile- truck crashes involving a l e f t turn. A chi-square t e s t shows no dependency in this table.

A comparison of Tables 2-5 and 2-6 suggests that a major differ- ence between car-motorcycle and car-car accidents i s in those cases where a car i s making a l e f t turn in front of an approaching motor- cycle. This argues against perversity as an explanation, since i t does n o t seem reasonable that drivers would behave perversely in only one maneuver.

2 .3 - 2 . 2 Summary. Our comparison of motorcycl e-car and car-car crashes has shown these points:

0 I n the prepondence of motorcycle-car crashes:

- the motorcycle i s traveling s t raight - the car i s maneuvering, most often turning l e f t

e Cars are more l ikely t o be maneuvering in motorcycle- car crashes than in car-car crashes,

These accident profiles provide strong support for concluding that motorcycle conspicuity i s a serious problem.

TABLE 2-4. A comparison o f marginal percentages of motorcycle and car pre-crash maneuvers with the car-car situation.

L

Motorcycle N=6647 Texas 1975 Crashes

Car N=6647 Texas 1975 Crashes

Car-Car N=16353 Texas 1975 Crashes

Straight

87.2

46.4

39.7

Right Turn

2.8

7.0

7.6

Left Turn

6.2

36.7

22.9

Back

0

1.7

2.7

Stopped

3.4

Park

0

lS2 I - 9 i I

24.4

T A B L E 2-5. Re1 a t i ve travel direction in 1975, Texas motorcycle- car accidents involving one vehicle going s t r a i g h t , the other turning l e f t . x 2 = 130.6, df = 2, p < .001. Frequencies are above percentages in each c e l l .

T A B L E 2 - 6 . Relative travel direction in 1975, Texas car-truck accidents involving one vehicle going s t r a i gh t and the other turning l e f t . X2 = ,818, df = 2 , P > . 5 . Frequencies a re above percentages in each c e l l .

Vehicle

-

1 Re1 a t i ve Travel Direction

Vehicle Turning Left

I Turning Left

I I I

I

Relative Travel Direction I

144 126 407 1 i 18.6 I 16.3 I 52.6

128 105 1 367 i 16.5 j 13.6 47 .4 1 I

Truck

Angl e

137 17.7

Angle

I Row Same / Opposite Total

Car 134 1 17.3

Same 1 Opposite / Row Total 1

142 1 Motorcycle 1 5.2 122 4.5

I 83 347 1 3.0 I 12.7 1

Car i

549 I 485 j zO.l 1 17.7 1354 2388 ! 49.5

I 87.3 I

2.3.2.3 Day-Night Differences in Motorcycle Crashes and

Conspicui ty . Table 2-7 presents the time-of-day breakdown of sing1 e-

and two-vehicle 1975 Texas motorcycle accidents. The table shows t ha t motorcycle accidents tend t o occur in the day (723). Two-vehicle

accidents are more l ike ly t o occur in the day than are single-vehicle

accidents (76.9% vs. 59.4$), while single-vehicle accidents occur

twice as often as two-vehicle accidents a t night (28.6% vs. 1 4 . 7 % ) . j

Table 2-8 examines the types of motorcycle and car crashes t ha t

occur during day1 ight and non-day1 ight hours. For both motorcycles

and ca rs , there i s a higher percentage of single-vehicle crashes dur-

ing non-day1 ight hours. In con t ras t , during the day, mu1 tiple-vehi-

c le crashes are predominant fo r both types of vehicle, In comparison

w i t h ca r s , motorcycles have more daytime single-vehicle crashes, and,

fo r b o t h day and non-day, more crashes where the other vehicle i s

turning. Cars, in comparison to motorcycles, are involved in more

mu1 tiple-vehicle coll is ions in the daytime with both vehicles

traveling forward and in both daytime and non-daytime with one vehicle

stopped, In general though, Table 2-8 shows a s imi lar breakdown of

crashes by l i gh t condition for both motorcycles and cars .

Table 2-9 and 2-10 provide a c loser examination of the l igh t -

condition differences observed fo r motorcycles in Tab1 e 2-8. Di s-

playing motorcycle-car data , both tab1 es contras t the pre-crash maneuvers of the motorcycle with tha t of the ca r . Table 2-9 i s for

daytime data ; Table 2-10 i s fo r non-daytime data , Although fewer

mu1 t i -vehic le motorcycle crashes occur in non-day1 ight hours, as indi-

cated in Table 2-8, the pattern of motorcycle-car crashes remains

constant across 1 i g h t conditions . Tables 2-11 and 2-12 show t h a t some daytime versus nighttime

differences in motorcycle accidents appear to be dependent on the color of clothes worn by motorcyclists. Table 2-11 i s for single-

vehicle motorcycle crashes, Table 2-12 i s for motorcycle-car crashes. Darker colors tend t o be involved w i t h nighttime crashes, especial ly

in the two-vehicle groups (Table 2 - 1 2 ) .

TABLE 2-7. Time-of-day versus number of v e h i c l e s i n 1975, Texas motorcycle a c c i d e n t s .

, Day

1 / Night wi th 1 ! 1 Row , N i g h t , S t r e e t L i g h t s , Dawn/ Dusk ! T o t a l s

I I I I

I I i 808 I , 258 I 14 I 67 1 2823 28.6 1 9.1 I 0 . 5 2 . 4 ' 100.0 43.0 1 37.4 / 42.4 30.3 28.0

8 . 0 / 2 . 6 j 0.1 0 . 7 28.0 I

1 1072 1 431 1 I 20 154 7268 14.7 5 .9 1 0 . 3 2.1 100.0 57.0 1 62.6 l 5 7 . 6 I 69 .7 72 . O 10 .6 1 4 . 3 0 .2 , 1 . 5 72 . O

i i

1880 1 689 34 221 10091 18 .6 1 6 . 8 0 . 3 2.2 100.0

100.0 1 100.0 100.0 100.0 100.0 18 .6 / 6 . 8 0 . 3 2.2 100.0

I

I S i n g l e

I Frequency 1676 ROW % I 59.4 Col %

I Tot %

/ Motorcycle-Car Frequency Row % 1 C o l %

23.1 16.6

5591 76.9 76.9

1 Tot % 1 55.4 1 Column T o t a l s

! Frequency , 7267 Row % 1 72.0 1 C o l % ! 100.0

I Tot % 1 72.0

I

T A B L E 2.8 . L i g h t c o n d i t i o n s in motorcycle and c a r c o l l i s i o n s , Texas motorcycle a c c i d e n t s , 1975.

MOVEMENTS OF THE VEHICLE

S i n g l e Forward

S i n g l e Turn

S i n g l e Back

Both Forward

ForwardlTurn

Forward/Back

Forward/S t o p

Turn/Turn

Turn/Back

DAY 1 NON -DAY

C A R / C A R

43.6

3 .9

C Y C L E T C A R j CAR / C A R / / C Y C L E / C A R

1.1

.3

1 . 2

.2

. 2

Turn/Stop 1 - 4

i I i 40.8

1 ' i 4 . 3

23.6

2.7

.9

.1

.4 Back/Back

Back/Stop

14.0

2.2

0

I 2 7 3 6 . 6

1

7.6

.9

0

0

. 3

2.9

32.6

23.7

2.2

1 8

1 . 7

0

I 0 I 2.8 , - -

20.6

26.3

.7

6 . 1

. 7

21.9

14.7

. 8

9 .9

1

LEFT TURN ANALYSIS

TABLE 2-9. Daytime l e f t - t u r n motorcycle-car co l l i s i o n s , Texas motorcycle a c c i d e n t s , 1975.

TABLE 2-10. Non-day l e f t - t u r n motorcycle-car col 1 i s ions , Texas motorcycle a c c i d e n t s , 1975.

DAY

Motorcycle

Car

NON-DAYTIME

Motorcycle

Car

Opposite

62

2%

1045

48%

Angle

13 7

6%

415

19%

Same

102

4 %

397

18%

Angle

11 2 :4

120

22%

! Same Opposite I

70

3 %

8 1

15%

2 1

3 %

2 7 3

5 1 %

TABLE 2-11. Color of upper garment worn by 1975 Texas m o t o r c y c l i s t s involved in s ing le -veh i c l e c r a s h e s , c l a s s i f i e d by t ime o f day of a c c i d e n t .

I

I I I Day

White I (Frequency)

T o t a l s

298 I

20.1

6 1

Night

7 6 25.5 16.2

5 .1

16

194

Night wi th L igh ts

2 8 9 .4

19 .6 1 . 9

4

(Row % )

26.2 3 .4 1 .1

1 99 31 .3 40.6 12.9

56 54.6 30.6

(Row % ) 65.1 1 22.4 (Tot % ) 1 13.1

Ye1 low i (Frequency) 4 1 (Row % ) 1 67.2 (Col % ) I 4 . 7

14 .8 18 .9

1 . 8

9 10 .3

6 . 3 0 . 6

12 8 . 5 8 . 4 0 . 8

9 9.1

(Tot % )

~1 ue (Frequency) (Row % ) (Col % ) (Tot % )

Brown (Frequency)

12 .4 /

8 7 I

5 .8 1 141

9 . 5

9 9

2.8

364 59.9 42.0 24.6

100

6 . 6 I

6 .6 2 .8 0 .3 1 4 .2

I

(Col % ) ' 11 .5 12 .0

(Tot % ) 1 3 .2

Column T o t a l s

5 4 8 . 9

37.8 3 .7

27

(Tot % )

Black (Frequency) (Row % ) (Col % ) (Tot % )

Green

2 .8 0 . 6 6.3

' 608

41.2

183

1 866 1 468 Frequency) ( ~ o t a l % ) 58.5 I 31.6

6 .8

3 9 44.8

4 .5 2 .6

I

143 9 . 7

3 . 8

3 9 44.8

8 . 3 2 .6

(Frequency) 1 80 (Row % ) I 56.7 (Col % ) 9 .2 (Tot % ) i I 5 .4

Red ( ~ r e q u e n c y ) 1 48 (Row % )

I 48 .5

(Col % ) 5 . 5

49 34.8 10 .5

3 . 3

42

TABLE 2-12. C o l o r o f upper garment worn by 1975 Texas mo to r - c y c l i s t s i n v o l v e d i n t w o - v e h i c l e c rashes, c l a s s i f i e d by t ime-o f -day o f a c c i d e n t .

'

I I I I

1 B1 ay:requency) i ( R O W % ) 1 (Col % )

I ( T o t % )

I Green 1 (F requency) ; (Row % ) I (Co l % ) I

, ( T o t % )

N i g h t 1 I i Day I N i g h t w i t h L i g h t s T o t a l s

Red I I I I

(F requency) I 136 ' 28 ' 15 (Row % )

1 179 76.0 15.6 I 8 .4 I

1 (Co l % ) 5.9 , 5.8 6.9 I

! ( T o t % ) I

I 4 .5 I 0 .9 0.5 I 5.9 1 I

1 Column T o t a l s 1 , (F requency) I I

I I

I I

! I

( T o t a l % ) 1 76.8 1 16.1 7.3 I 100.2 I , 1

1 I I I

2 .4 ! 1 . 2

80 29 16 1 125 64.0 23.2

I I 12.8

3.5 6 .0 7 .3 I

2.7 1 .O 0 .5 1 4.2 1

237

I

/ Wh i te i (Frequency) 1 (ROW %) 1 (Co l % ) I ( T o t % ) I 1 Ye l l ow

(Frequency) I (Row % ) 1 (Col % ) 1 ( T o t % ) 1 B l u e

(Frequency) I (ROW 9 ) i (Co l % ) I ( T o t % ) 1 Brown

(Frequency) I ( R o w % ) I (Co l % )

( ~ o t 7 ; )

I I

38 26 ' 301

539 81.8 23.3 17.9

122 86.5

5.3 4.1

925 75.3 40.1 30.7

270 71.6 11.7

9.0

87 I 33 1 659 13.2 ' 5.0 18.0

1

15.1

13 6 1 141 9.2 4.3 2.7 2.8 1

0 .2 ' I 4 .7

217 1 87 17.7 1 7.1

78.7 12.6 8.6 10.3 1 7.9 1 11 .9

1229

8 . 6 I

7.9 I 1 .3 i 0.9 10.1 , I I

39.9 I

7 2 19.1

3 5 1 377 9 .3

14.9 , 16 .1

2 . 4 Summary

In general , the crash data indicate the following f ive answers

are the best t o the questions posed in Section 2.1.

8 Differences in patterns of motorcycle-car and car-car crashes suggest tha t other drivers do sometimes f a i l t o see o r

ident i fy motorcycles, i . e . , conspicuity appears t o be a fac- , t o r in motorcycle crashes .

9 Crash data are inconclusive on a day-night difference in

motorcycl e conspi cui ty . e Crash data do suggest crash involvement decreases with

increased motorcycl i s t conspicui t y .

e The most prevalent motorcycle-car crash configuration, in

b o t h the day and night , involves a s t r a i gh t traveling

motorcycle, and a lef t - turning car .

Estimates of the number o r percentage of crashes involving conspicuity i s very d i f f i c u l t due t o the necessity of rnaki ng some ra ther tenuous assumptions. The cul pabi 1 i t y

data suggest i t may be a fac to r i n about 10% of a l l

motorcycle-car crashes (about 60% of the time the car driver i s judged a t f a u l t when chance d ic ta tes each should

be culpable 50% of the t ime) .

3.0 SEEING A N D REACTING TO MOTORCYCLES

3.1 Introducti on

Based on analyses of driver performance ( s ee , e . g . , Alexander &

Lunenfel d , 1975; Fell , 1976 ; and Shi nar, 1978), one can postulate three stages of processing in a d r i ve r ' s sensory-cogni t ive response to a moving motorcycle. These stages are :

9

1. Detection

2. Ident i f ica t ion

3. Decision

Detection implies t ha t su f f i c ien t visual stimulus has been pro- vided t o cause the observer t o rea l i ze t ha t "something" i s there . Typically, as will shor t ly be described, detection i s peripheral and an eye movement ca l led saccade i s made t o permit the "something" to be studied in the foveal portion of the eye.

The iden t i f i ca t ion stage involves the acquisi t ion of su f f i c i en t information about the object of concern to be able to make a proper

decision regarding action t o be taken. Thus, i f the object i s a mov- ing vehicle, as in the case of a motorcycle, the driver must, a t a min imum, recognize the object as a vehicle with h i g h performance character is i t c s and make some estimate of i t s distance and speed.

With the ident i f ica t ion process complete, the dr iver must decide an appropriate course of ac t ion. Typically, in the present context , t h i s will be a go or no-go decision. The decision i s influenced in part by factors other than those acquired in the ident i f ica t ion s tage . For example, a dr iver i n a hurry or who has wai ted a 1 ong time may accept a gap in t r a f f i c t ha t would normally be re jec ted. I t could a l so be influenced by factors such as prejudice, conceivably leading some drivers t o del i berately engage i n actions dangerous to motor- c y c l i s t s ,

T h u s , motorcycle crashes which appear t o be conspicuity-related

could actual ly a r i s e from one or more of the following:

Failure t o detect .

Misidentification (e . g . , "bicycle" ra ther t h a n "motorcycle").

Errors in speed-spacing judgments.

Inappropriate decision ( e . g. , based on excessive haste o r 1

prejudice against motorcycles).

Inappropriate decisions based on has te , drugs, o r fac tors other

t h a n prejudice can be ruled o u t as an explanation of the data presented

in Section 2.0, as they should apply equally t o cars and motorcycles.

This s t i l l leaves a minimum of four possible problem areas which may

account, in some combination, fo r the accident patterns noted.

3.2 The Functional Implications of the Structure of the Retina: Foveal vs . Peripheral Vision

From the functional point of view, the human re t ina ( the par t of

the eye where the image becomes focused) can be divided in to two main

regions, the fovea and the periphery. The fovea i s the area of the 0

re t ina corresponding t o the central 1 - 2 of the visual f i e l d (Polyak,

1941) and the periphery corresponds t o the remainder of the visual

f i e l d . Important differences between the roles of the fovea a n d the

periphery in visual information processing are discussed below.

In photopic and mesopic 1 ight conditions , the 1 uminance threshold

i s lowest a t or near the fovea a n d decreases with increasing re t inal

periphery ( A u l horn & Harms, 1972). In scotopic 1 ight conditions the

peak s ens i t i v i t y i s obtained a t 1 5 ~ - 2 0 ~ in the periphery (Aulhorn &

Harms, 1972; Pirenne, 1967). However, most of the driving i s done

under photopic o r mesopic conditions (Cole, 1972; Projector & Cook,

1972; Schmidt, 1 9 6 6 ) . Therefore i t can be argued tha t in most driving

s i tuat ions the 1 uminance threshold i s lowest a t or near the fovea.

The fovea i s t igh t ly packed with wavelength (co lo r ) sens i t ive

cones (Osterberg, 1935) . Since there i s a sharp decrease in the

frequency of the cones outside of the fovea (Osterberg, 1935), color

discrimination i s best in the fovea and decreases with increased

eccentr ic i ty from the fovea (Morel and, 1972; Wall s , 1942).

The threshold fo r motion detection (despite popular be1 ief t o the contrary) i s lowest in the fovea; i . e . , the thresh01 d speed i s

lower a t or near the fovea t h a n in the periphery (Aubert, 1886; t

Gordon, 1947; Klein, 1942; McCol gin, 1960).

Acuity refers t o the ab i l i t y t o resolve f ine de ta i l , and one mani-

fes ta t ion of good acuity i s e f f i c i en t ident i f ica t ion of form. Acuity

(and consequently form perception) i s best a t or near the fovea and

decreases with increasing eccentr ic i ty from the fovea. The superior

acuity of the fovea holds whether the target i s moving in relat ion to

the observer, so-call ed dynamic acuity (Gordon, 1947; Kl ei n , 1947 ;

Low, 1947), or i s s tat ionary in relat ion t o the observer, so-called

s t a t i c acuity (Aulhorn & Harms, 1972; Brown, 1972; Feinberg, 1948;

Green, 1970; He1 d , 1959; Hershenson, 1969; Kerr, 1971; Low, 1951;

Ludwigh, 1941; Mandelbaum & Sloan, 1947; Wertheim, 1894).

As mentioned above, the periphery of the re t ina includes the

whole visual f i e l d except the central 1-2' of the fovea. I n the horizontal meridian the active binocular visual f i e l d ( the area in

which one can detect the presence of an object) extends t o about

175-180'; in the vert ical meridian i t extends t o a b o u t 100-130' (Burg, 1

1968 ; Connol ly , 1966 ; Schmidt , 1966) . Since the periphery compri ses

more than 99% of the active visual f i e l d , the 1 ikelihood of the image

of an object f a l l ing on the periphery as opposed to the fovea i s high.

However, as a l ready noted, acuity and consequent form ident i f ica t ion

i s rather ine f f i c ien t in the periphery. Therefore, most objects have

to be identif ied i n the fovea (Waldram, 1960).

l ~ h e extent of the active visual f i e l d varies with the s ize of the t a rge t to be detected. Furthermore, there i s a general shrinkage of the active visual f i e l d with age (Burg, 1968; Wolf, 1967). The val ues presented here apply for young observers.

3.2.1 The Peripheral F i l t e r . The functional differences between

the fovea a n d the periphery led several researchers t o postulate

d i f fe ren t processing modes of these regions of the ret ina (Boynton,

1960; Mackworth & Bruner, 1970; Mourant & Rockwell, 1970; 1971;

Schiffman, 1972; Treverthen, 1968). These authors argue t h a t the

periphery i s involved primarily in the monitoring and detection pro-

cesses, while the fovea i s involved primarily in the indentif icat ion I

process. However, ident i f ica t ion by the fovea requires sequential

sh i f t ing of the fovea to d i f ferent spat ia l locations, resul t ing in a

generally se r ia l ident i f ica t ion process, while the detection process

within the visual f i e l d i s e ssen t ia l ly pa r a l l e l . Because of the

sequential 1 imitation of the ident i f ica t ion process a n d because of

the re la t ive ly large s ize of the periphery in comparison t o the fovea,

i t i s usually impossible t o identify a l l the targets detected by the

periphery. Therefore, i t i s reasonable t o postulate, as did these a u - thors, t h a t the resu l t s of peripheral processing will se lec t ively deter-

mine where the foveal at tention will be directed. I n other words, a

peripheral f i 1 t e r i s postulated which f i 1 t e r s (gates) the information

i r re levant t o the task in order for the relevant information t o be

more thoroughly inspected by the fovea (Boynton, 1960; Hochberg, 1970;

Howett, Kelly, & Pierce, 1978; Mackworth & Bruner, 1970; Mackworth &

Morandi , 1967; Mourant & Rockwell, 1970; Townsend & Fry, 1960).

The determination of the relevant parameters for the peripheral

f i l t e r comes from several types of investigations: eye or head move-

ment monitoring, conspicui ty evaluation, a n d detection thresh01 d

determination. If subjects are told t o search for specif ied targets

(e.g. , given s ize o r shape), they are less 1 i kely t o f ixate on the

i r re levant targets (Will iams, 1970). This indicates tha t some prel i -

minary categorization i s taking place even i n the periphery. In a

free-search s i tua t ion , where the person i s under no ins t ruc t ion , the

eye-novement studies indicate tha t the fovea is a t t r ac ted towards

high-information points of the display (Mackworth & Morandi, 1967;

Notton & Stark , 1971; Yarbus, 1967), high-contrast areas (Thomas,

1968), f l ickering stimul i (Thomas, 1968; 1969) large-si zed stimul i

(Thomas, 1968) , or moving objects (Thomas, 1968).

Studies monitoring visual searches (eye and head movements) indi - cate tha t there i s an increase in visual searches with an increase in

the amount of t r a f f i c (Robinson, 1972; Robinson, Clark, Erikson, &

Thurston, 1971). These findings suggest tha t the relevant visual tar- get f o r the driver t o de tec t , f i xa t e upon, and ident i fy might be

movi ng objects .

I t i s known (Bart ley, 1938; Gerathewohl , 1953; 1957; Halstead,

1941) t h a t f lashing 1 ights are generally more conspicuous t h a n steady

l i g h t s , whether the conspicuity i s measured in terms of apparent 2 brightness or reaction time. Therefore, i t i s reasonable t o assume

t h a t under cer ta in conditions a f lashing 1 ight will be more 1 i kely t o r e su l t in a f ixat ion than a steady l i g h t .

I t can be assumed t ha t in a free-search s i tua t ion an object t h a t

i s more 1 i kely t o be detected will more 1 i kely be f ixated upon a n d

thereby more l i ke ly iden t i f i ed . I t i s well known t h a t the detect-

a b i l i t y of a t a rge t i s affected by i t s s i z e , luminance, and con t ras t . 3

With an increase in the s ize of a s ta t ionary target there i s a decrease

in the luminance or contras t threshold; conversely, with an increase in

t a rge t luminance or contras t there i s a decrease in the s ize threshold

(Austin, 1951; Bl ackwell , 1946; 1959; Brown, 1947; Graham & B a r t l e t t ,

1939; Graham, Brown & Mote, 1939; Hi1 I s , 1976; Krisstofferson, 1954;

Lamar, Hecht, Schlaer, & Hendley, 1947; Riopelle & Chow, 1953; Taylor,

1964). Therefore, i t can be assumed t h a t 1 a rger , b r igh te r , o r more

contrast ing stimuli wi 11 be more frequently f ixated.

'?his advantage of the f lashing 1 ight i s absent in the presence of other f lashing 1 ights (Crawford, 1962; 1963).

3 ~ h e luminance i s the variable of i n t e r e s t i f the background luminance i s zero; contras t i f the background luminance i s n o t equal t o zero.

3.2.2 Eye Movements and Saccadic Suppression. The change of the

f ixat ion point i s most frequently achieved by a f a s t jumping eye move-

ment, ca l led the saccade. A saccadic eye movement can be e l i c i t e d

voluntari ly or i t can be tr iggered by a peripheral stimulus, whether

visual or auditory. Kestenbaum (1961) refers t o a saccade tr iggered

by the peripheral visual stimulus as the op t ica l ly e l i c i t e d movement

( O E M ) . According t o Kestenbaum, an OEM (a saccade) i s " e l i c i t ed by an

object s i tua ted in any place of the visual f i e l d , except the fovea,

when t h i s object a t t r a c t s the a t t en t ion [ i . e . , i t passes the peripheral

f i l t e r ] " ( p . 300). The saccadic movement " i s semi-reflex, occurring

without conscious vol i t ion b u t depending on a t t en t ion and capable of

suppression (Cole, 1972, p . 108) . " The importance of saccades in the present context 1 i e s in t he i r

e f f ec t on the visual processing. There i s extensive evidence t ha t

during the saccadic eye movements the visual capabil i t i e s a re reduced.

I f the t a rge t i s s t a t ionary , t h i s saccadic suppression i s evident

whether the visual performance i s measured in terms of a decrease in

the probabil i ty of detect ion, in terms of a decrement in detection

s e n s i t i v i t y (as measured fo r example by - d l ) , or in terms of an

increase in the luminance threshold (Beeler, 1967; Latour, 1962;

Mitrani, Yakimoff & Matteeff, 1970; Pearce & Porter , 1970; Richards,

1969; Vol k m a n , 1962; Vol kman, Schick & Riggs, 1968; Zubek & Stark,

1966). If the t a rge t i s moving, the saccadic suppression i s mani-

fested by a decrease in the probabil i ty of the motion detection or decrease in the probabi 1 i ty of correct ly identifying the di rect ion of

the motion (Beeler , 1967; Ditchburn, 1955; Holly, 1975; Sperling &

Speelman, 1965; Wallach & Lewis, 1965) .

3.2.3 Imp1 ica t ions fo r Motorcycles. In order t o i l l us t ra te what

the concepts a n d processes described above mean fo r motorcycle conspi-

cui t y , consider the fol lowing example:

Assume tha t a car i s attempting t o merge in to a primary road from

a secondary road. In the search of the in te r sec t ion , the dr iver of

the car i s l ike ly t o execute f i r s t a voluntary saccade and/or a head

movement of a fixed magnitude, say t o the l e f t . However, the conten-

tion here i s t ha t his fu r the r eye movements ( a t l e a s t in the leftward

d i rec t ion) wil l be guided by the objects passing the peripheral f i l - t e r . In a case of no t r a f f i c near the in te r sec t ion , there might not be any object passing the peripheral f i l t e r and the d r i ve r ' s gaze

would s h i f t to the r igh t of s t r a i gh t ahead and repeat the search in

the rightward di rect ion. Again, i f there i s no t r a f f i c , no objects

wil l pass the peripheral f i l t e r and the driver wil l proceed to the

planned maneuver. (The whole cycle of the search--leftward and rightward--is sometimes repeated several times before making the maneuver, as in the case of a cautious d r i ve r . )

Now assume tha t there i s some t r a f f i c a t the in te r sec t ion , b u t

the t r a f f i c i s l i g h t , say a car and a motorcycle both approaching

from the l e f t . After executing the l e f t (voluntary) saccade, there wil l be only two moving objects ( the car and the motorcycle) competing

fo r foveal a t t en t ion . In such a s i t ua t i on , the peripheral f i l t e r

would l ike ly pass both objects ( sequen t ia l ly ) , resul t ing in a t l e a s t one f ixat ion on each of the two vehicles. Such a pattern of f ixat ions

would r e su l t in high probabil i ty of correct ly identifying both vehi- c les and taking the proper ac t ion.

In a heavy t r a f f i c , however, there might be dozens of vehicles

approaching the in tersect ion from both s ides . The poss ib i l i ty i s t ha t i f there i s a motorcycle within the approaching t r a f f i c , i t will

be l e s s l i ke ly than other vehicles t o pass the peripheral f i l t e r and

thus wil l be l e s s 1 i kely t o t r igger a saccade than the larger-sized

vehicles (cars and especial ly t r ucks ) . The resul t ing lower proba- b i l i t y of f ixat ions on the motorcycles would make them less l ike ly to be ident i f ied because of t h e i r remaining in the periphery and

because of the consequence of saccadic suppression.

In terms of Signal-Detection theory (Green & Swets, 1966) i t could be argued t ha t the amount of t r a f f i c has an e f f e c t on the

decision c r i t e r ion (6) b u t not on the s ens i t i v i t y ( d ' ) . (The c r i t e r ion here i s the c r i t e r ion of the peripheral system for the

i n i t i a t i on of a saccade.) Therefore, i t i s proposed that in conditions

ranging from no t r a f f i c t h r o u g h l i gh t t o moderate t r a f f i c the c r i t e r ion

i s low, result ing in ~"ua t ions where even objects with less favor;-

able physical character is i tcs ( e . g . , smaller s i z e ) would t r igger

saccades and lead t o f ixat ions . Throughout t h i s range of conditions

the c r i t e r ion would remain constant, s ince the foveal capacity f o r the

frequency of f ixat ions (2-4/sec. , Cole, 1972) might be reached only i n

the moderate t r a f f i c . However, w i t h a fur ther increase in the amount of t r a f f i c the c r i t e r ion fo r the i n i t i a t i on of a saccade i s

increased since the 1 imi t of the essen t ia l ly se r ia l ident i f ica t ion

processing by the fovea (frequency of f ixat ions) has been reached. In

heavy t r a f f i c there are many potential peripheral candidates ( f o r the

approximately 2-4 f i xations/sec. ) and therefore the selection process

has t o be more s t r ingen t . These conditions may lead t o the reject ion

of targets which would not be rejected in l igh te r t r a f f i c ( e . g . ,

motorcycles).

3.2.4 Conclusions. On the basis of th i s analys is , i t can be

concluded t ha t motorcycles are l e ss l ike ly t o be detected and to pass

the peripheral f i l t e r a n d thus less 1 ikely t o be ident i f ied because of

the following considerations:

F i r s t , the physical features of motorcycles (especial ly t he i r smaller s i z e ) make them more 1 ikely t o be closer t o ( i f not

below) peripheral thresh01 ds t h a n those same features fo r automobi l e s .

This implies t h a t motorcycles are less l ike ly t o be detected, r esu l t -

ing in a diminished likelihood of t he i r passing the peripheral

f i l t e r and therefore a diminished likelihood of being subjected t o sub

sequent processing ( e . g . , being fixated upon and ident i f ied in the

fovea).

Second, even i f detected, motorcycles are less 1 i kely t o t r igger saccades, which would resu l t in a foveal f ixation on them. That i s ,

the probability tha t motorcycles would occupy the area of the ret ina

most e f f i c i en t a t motion detection, acui ty , and color discrimina-

tion--the fovea--is reduced from tha t of automobiles. The higher

threshold speed fo r motion detection in the periphery would r e su l t in

high probabil i ty of misperceiving such moving objects as being s t a -

t ionary. The poor acuity and consequently inaccurate form perception

would lead t o a low ra te of correct iden t i f i ca t ion of motorcycles. 4

Third, motorcycles will be 1 i kely t o fa1 1 on the fovea only dur-

ing saccades, s ince they are unlikely t o be the terminal points of *

saccades ( the f ixat ion po in t s ) . Therefore, the e f f ec t of the saccadic

suppression will be more detrimental t o the iden t i f i ca t ion of motor-

cycles. In other words, the objects t ha t do not pass the peripheral

f i 1 t e r occupy the fovea only during saccades ; the objects t h a t pass

the peripheral f i l t e r occupy the fovea during f ixat ions as well . This

s i tua t ion r e su l t s in the objects n o t passing the peripheral f i l t e r

having a decreased probabil i ty of correct iden t i f i ca t ion , since the

ident i f ica t ion during saccades i s l e s s e f f i c i e n t t h a n during f ixa-

t ions .

3.3 Cr i t i ca l Cues

Our analysis of the c r i t i c a l cues fo r motorcycle iden t i f i ca t ion

i s based on Henderson and Burg's (1974) analysis of dr iver visual

requirements. Table 3-1 shows a l i s t of c r i t i c a l cues for ident i fy-

ing a motorcycle. Generally speaking, other drivers ident i fy motor-

cycles based on the motorcycle's motion, color , form, shape, s i z e ,

luminance, or some combination of these cues.

Many drivers probably use a special p r i o r i t y ordering or combina-

t ion of these cues t o ident i fy a motorcycle. For example, a North 5

Carolina resident might ident i fy a1 1 moving s ingle headlamps as

4 ~ h e assumption i s t ha t a f ixat ion might be a necessary condi- t ion fo r e f f i c i e n t iden t i f i ca t ion . No claim i s being made t h a t i t i s a lso a su f f i c ien t condition fo r e f f i c i e n t iden t i f i ca t ion , s ince i t i s k n o w n t h a t a f ixat ion on an object does n o t guarantee i t s iden t i f i ca - t ion (Thomas, 1968).

5 ~ o r t h Carol ins currently has a motorcycle 1 ights-on law.

T A B L E 3-1. Types of c r i t i c a l cues f o r identifying a motorcycle.

I 1 Cue 1 Cue Definition

Angul a r movement / Lateral motion vector / component

Color contras t Perceived color d i f f e r - ences between object I and background I

Form I El ernen tary geometric 1 element, e . g . , l i n e , j c i r c l e

Movement in depth

1 Shape

Closure motion vector , component

i Size

Color / Perceived color 1 I I

! Complex geometric ! s t ruc tu re , e . g . , two-

wheel s , hand1 ebars I 1

Retinal image s ize

/ Luminance Perceived brightness

Luminance contras t Perceived brightness I d i fferences between

~ object and background

motorcycles. However, we currently do not know the actual patterns of cues drivers use to ident i fy motorcycles. Thus, we do not know which pattern of cues i s the most c r i t i c a l fo r identifying a motor-

cycle.

Given t h i s knowledge gap, we cannot proceed straightforwardly t o discuss jus t how a motorcycle i s ident i f ied . Instead, we must examine

#

how drivers perceive each of these cues.

In general , perception of the color , 1 uminance, form, and shape cues i s no d i f fe ren t fo r motorcycles than fo r any other vehic le , except tha t smaller s i ze produces d i f f i cu l t y i n perceiving them. A

major d i f f i cu l t y appears in perceiving motion. Two cues are con- sidered here as providing information necessary fo r motion perception. The motion-in-depth type of cue allows the perception of r e l a t i ve

closure. The angular motion type of cue,' allows perception of l a t e r a l movement. Since the motion-in-depth cues are most affected by vehicle s i z e , and because i t i s a type of cue involved in almost a l l motion perception in driving, they will be focused upon in our discussion here.

3.3.1 Motion Perception. Michaels (1963) postulates a two-stage model of the process involved in detecting closure and r e l a t i ve speeds by the dr iver of a car following another ca r . The f i r s t stage applies to those cases where the time r a t e of change in the lead c a r ' s image s ize i s too slow to be perceived d i rec t ly . I n t h i s case detec- t ion i s thought to be accomplished by comparing the present image with a remembered image. A t higher time ra tes of image s ize change, the observer can d i rec t ly sense ra te and l ike ly use t h i s as a basis

fo r judgment. Thus, below the threshold fo r detection of motion, s ens i t i v i t y i s based on the a b i l i t y t o make judgments of s i ze change

on an absolute bas is , Michaels estimates the Weber r a t i o (A image size/image s i z e ) in t h i s task to be about 0 .1 . The threshold fo r detection of r a te of change of the s ize of the image of a lead car i s estimated by Michaels t o be about 6 x rad/sec (about 2 minutes of arc per second).

Hoffman (1968) has elaborated on Michaels ' model , Fol lowing the l i n e of his 1966 paper, Hoffman reanalyzed the data from several

s tudies (such as those described below) t o determine whether per-

formance seemed t o be in accord with predictions based on a two-stage

process. He fee l s t ha t the r e su l t s do support such a model, although

he argues t ha t there i s probably a range of ra tes wherein both abso- a

l u t e judgment and r a t e detection operate. Hoffman a lso reports t ha t

a low (sub-threshold) time r a t e of change in lead car image s i z e

detection was "degraded by the presence of these angular ve loc i t i e s . "

No comprehensive investigations of the val i d i t y of Michaels ' model have been reported. However, Sal vatore (1965) has investigated

s ens i t i v i t y t o the s i z e change cue. O n an oscil1oscope he displayed a bar representing the width of a c a r . By changing the length of the bar he simulated three spacings and presented two ra tes of change in

bar length deemed sub-threshold (34 and 102 seconds of a rc per

second). The dependent variable was response time. Salvatore reports

shor ter response times fo r the higher ra te b u t the Weber r a t i o for

change in visual angle was somewhat l e ss for the lower r a t e . I t was

a l so found t ha t the Weber r a t i o decreased from about 0.13 for a 1.5'

t a rge t t o 0.065 fo r a 6" t a rge t .

Using a d i f fe ren t procedure, Vincent e t a1 . (1969) have a lso studied s ens i t i v i t y t o s i ze change. Their t a r g e t , an Apol l o command

and service module, presented a t a rge t s i ze of a b o u t 13 x 24 f ee t and

was viewed s t a t i c a l l y f o r three seconds, obscured fo r three seconds, and viewed again in a d i f fe ren t s i ze fo r three seconds. They

reported r a t i o s of 0.028 a t 200 f e e t , 0.029 a t 400 f e e t and 0.03 a t 800 f e e t . Beyond 800 f ee t the Weber r a t i o increased and was reported

as 0 .065 a t 12,800 f e e t , the g rea tes t viewing distance used.

A s imi lar procedure was used by Olson (1971) for automobile t a rge t s a t viewing distances u p t o 300 f e e t in both laboratory and f i e l d invest igat ions . The Weber r a t i o was 0.03 fo r a l l condit ions.

There has been i n t e r e s t in measuring thresholds for angular

motion for many years. A number of these investigations have been

summarized by Brown (1960), who points out tha t the l e a s t detectable

difference in speed ( A W ) varies in d i rec t proportion to speed over a range of 0.1' t o 20' visual angle per second. The Weber r a t i o A i d / W

i s equal to 0 . 1 . *

The work to which Brown refers i s based on motion in a plane

normal t o the observer 's l i ne of s igh t . The motion of i n t e r e s t in

t h i s review i s of the edges of the image of a lead vehicle, produced as the image increases or decreases in angular s i z e . I t i s not

obvious t ha t s ens i t i v i t y t o the two types of motion should be d i rec t ly re1 a ted.

The r a t e threshold of 6 x rad/sec assumed by Michaels i s

based on the resu l t s of an investigation of another driving s i t ua -

t ion (Michaels and Cozan, 1963), where an attempt was made to i so la te the variables by means of which the driver locates himself r e la t ive to fixed objects in or near his path. The authors measured l a te ra l

displacement of a vehicle as a function of i t s speed and the position

of a roadside obstacle. They interpreted t he i r r esu l t s t o mean tha t drivers were using as a cue to object position the perceived l a te ra l velocity of the obstacle and deduced the above value as a perceptual

thresh01 d .

Baker and Steedman (1961) have reported data on the s ens i t i v i t y of observers t o movement in depth. They used a luminous disc which

could be moved toward or away from the subject in an otherwise t o t a l l y dark f i e l d . The f i r s t study investigated the e f fec t of t a r -

get luminosity. The t a rge t subtended an angle of 0.0117 radian (equivalent to a six-foot wide car viewed a t 513 f e e t ) and was moved toward or away from the subject to produce an i n i t i a l r a te of change

in s i ze of 1.29 x 1 0 ' ~ rad/sec. Exposure durations were controlled a t s i x levels . The resu l t s indicated improved performance b o t h as t a rge t 1 uminosi ty and exposure durations increased,

A second study reported in the same paper held t a rge t lumi-

nosity constant and varied t a rge t speed and exposure duration. Tar-

get speeds produced i n i t i a l ra tes of change in target image s ize of

0.64, 1.29, 2.57, and 5.5 x radlsec. Exposure durations varied

from 0.15 second t o 13.2 seconds. The r e su l t s indicated improved

performance associated with higher ra tes a n d longer viewing times.

For instance, given a viewing time of 0.9 second, subjects correct ly

ident i f ied the direction of travel 95% of the time a t the highest r a t e ,

85% of the time a t the second highest r a t e and 60% of the time a t the

th i rd highest r a t e .

In another study, the same authors (Steedman and Baker, 19621,

investigated the e f f ec t of stimulus s i z e on the capabi l i ty of subjects

t o detect movement in depth. The apparatus was the same as in the

e a r l i e r study. Six t a rge t s s izes were used, ranging from 0.0176

radian t o 0.00044 radian. Exposure durations ranged from 0.4 to 16.46

seconds. The t a rge t f i r s t became v i s ib le a t a viewing distance of 300

inches and moved a t a constant speed of s i x inches per second. With

target speed f ixed, the i n i t i a l r a te of change in angular s i z e ,

measured in radians per second, was d i rec t ly related t o target diameter.

For instance, the i n i t i a l r a te for the l a rges t t a rge t was 3.52 x radlsec, the i n i t i a l r a te for the next l a rges t t a rge t was half t h i s

value and so o n . The resu l t s show a steady improvement in performance

as larger targets were used. Unfortunately, the design of the study does not permit the separation of t a rge t s i ze from ra te of change in

s i ze .

Granting tha t the stimuli are qui te d i f fe ren t from a car viewed

against a roadway environment, these two s tud ies , pa r t i cu la r ly the

f i r s t one, o f fe r a chance t o check the model proposed by Michaels.

According t o t h i s model, below a r a t e of 6 x rad/sec a given percent change in s i ze should produce performance equivalent t o s t a t i c , absolute judgments (75% correct fo r a 39 change, based on Vincent e t a1 . , 1969)" Indeed, performance a t the lowest r a t e used

1.64 x 10'~ radisec) was about 7 5 % correct a t 3% change, a1 though i t

increases t o 80-85% correct a t higher r a t e s . A t the highest r a te

(5.15 x 1 0 ' ~ radlsec) 3% performance began t o deter iora te r e l a t i ve

to the intermediate ra tes although, as pointed out by the authors,

t h i s may be a t t r ibu tab le t o the very shor t viewing times (approxi-

mately 0 .5 second) i nvol ved,

A1 though the data are sparse , they do lend some support t o the

model. Since a r a t e of 6 x rad/sec was not a t t a ined , one can- 1

not be cer ta in that performance does not s t ab i l i z e above t ha t level

and the gradually increasing performance a t middle ra tes could be a t t r ibu tab le t o the "dual mode" zone proposed by Hoffman. Clearly

more evidence i s required before the model can be regarded as va l id .

An in te res t ing model of motion perception has been proposed by Kinchla and A 1 lan ( 1 9 6 9 ) . According to t h i s model an image of a

t a rge t i s s tored a t time - k . A t time - k + - t , where - t denotes a decay

time, a comparison i s made with the a l t e red image. The probabil i ty

of detecting whether an object i s in motion depends on the decay r a t e and the time between comparisons. In a s i tua t ion where the

conditions a re re la t ive ly constant and the decay r a t e i s fixed the model predicts performance varying as a function of extent of change with shor t - t ' s and a drop t o chance levels of performance as - t becomes 1 ong .

Olson (1971) conducted a se r ies of investigations t o compare pre- dict ions of the two models of motion perception. Using movie presen-

t a t i ons , d i f fe ren t r a t e s , t a rge t s izes and viewing times, he concl uded t ha t the model of Kinchla and Allan was more nearly correct . Olson

found t ha t a t s i ze change ra tes of 3 x rad/sec and above per- formance was reasonably s tab le and dependent upon the extent of

change. For example, a 4% change was correct ly ident i f ied 95% of the time a t these ra tes and a 2?: change 85-90% of the time. A t r a t es

below 3 x loe4 radlsec performance deter iora ted s tead i ly and seemed dependent on r a t e more than extent of change. These data suggest a high degree o f s ens i t i v i t y t o the direction of change ( .e .g. , 95%

accuracy for a 4% change a t 2 f t / s e c a t a 200 foot headway), Similar

levels of s ens i t i v i t y a re reported by Evans and Rothery (1974) based on a f i e l d investigation.

The s tudies reviewed above suggest t h a t , a l l other fac tors being

equal, detection of movement in depth i s eas ie r when the approaching

object i s large . For example Table 3.2 compares the changes in image

s ize over time for a car and motorcycle approaching a t the same speed. t

These studies have been concerned with the question of whether

something i s moving, without regard to d i rect ion. Another question concerns ra te of movement, especial ly fo r the case where the object i s moving toward o r away from the observer. The research in thTs a rea , which i s less well s tudied, has been reviewed by Hoffman ( 1 9 7 4 ) .

I t i s apparent t ha t observers have great d i f f i cu l t y dist inguishing

d i f fe ren t ra tes of closure. However, there i s no d i r ec t evidence to suggest tha t such judgments would be more d i f f i c u l t in the case of

motorcycles than automobiles o r trucks.

Complete ident i f ica t ion of an approaching vehicle f o r purposes of deciding whether to attempt a maneuver requires a reasonably

accurate distance estimate. Size constancy, (basical Iy , the assump- t ion tha t the apparently miniature object seen i s actual 1y a normal

object a t a distance) i s a fac tor in distance judgment where the

motorcycle could be a t a disadvantage. Motorcycles vary in s i z e , as

do cars . B u t many motorcycles have fa i r ings , and fa i r ings vary

greatly in area presented to the f ron t , compounding the problem of estimating the actual s i z e of the bike. The re la t ive ly low famil-

i a r i t y of most drivers with motorcycles, the various s izes and configurations adds t o the problem. Thus, drivers may tend t o e r r

in judging the distance in an approaching motorcycle more than to an approaching ca r . However, there i s no d i r ec t evidence t ha t such i s the case.

These data provide some reason t o believe t ha t speed-spacing decisions may be more d i f f i c u l t in the case of a motorcycle. I f so ,

i t may account for some of the conspicuity.-related crashes.

T A B L E 3-2. Retinal image s i z e changes from a fu l l - s i z e automobile and motorcycle c los ing, head-on from 500 f e e t a t 30 m p h . (44 f t / s e c . ) . The automobile image i s measured in horizontal plane and i s based on an 80" width. The motorcycle image i s measured in the ver t ica l plane a n d i s based on a 5 ' height.

Elapsed Seconds

I ' lotorcycle Image I A Size

1 Full Size Auto- 1 mobile Image 1

A Size I

Distance 1 Degrees 1 Radians Degrees 1 Radians I

3 .4 Other Driver Decision-Making about Motorcycles

We define decision-making here as the process of deciding how to

react t o a detected and ident i f ied motorcycle,

Information, used in the detection and ident i f ica t ion s tages , i s

carried forward in to the decision s tage , For example, the shape of the motorcycle's image t e l l s which direction i t i s t raveling and

thereby determines, in pa r t , i f a potential conf l i c t e x i s t s ; i t s

perceived motion indicates how f a s t the motorcycle i s t raveling with

respect t o the other driver a n d thereby provides another piece of

information about the poss ib i l i ty of a conf l i c t . Other d r i ve r ' s deci sions are therefore pa r t i a l l y based u p o n the same information used in

the detection and ident i f ica t ion stages.

Other drivers also use information from the i r memories in deci- sions about motorcycles. For example, drivers k n o w by memory how much

room they should allow other vehicles, including motorcycles, in

executing a maneuver. Other drivers a l so know the difference between

a motorcycle and car means t ha t any co l l i s ion with motorcycles will

l ike ly not produce much personal loss , thus possibly making them more

l ike ly to r i sk a co l l i s ion with a motorcycle.

In general , during decis ion-making, drivers combine perceived

information from detection and ident i f ica t ion with remembered infor-

mation. Ideal ly , drivers will accurately decide when a threat of a

col l i s ion with a motorcycle ex i s t s and reduce t ha t t h r ea t . Thus, other drivers should make two decisions about detected and ident i -

f ied motorcycl es :

a Is there a th rea t of a conf l i c t?

0 I f so , what can be done t o avoid the conf l i c t?

To determine i f a conf l i c t i s possible w i t h a detected and ident i f ied motorcycle, another driver must determine:

a If the vehicle paths will cross each other.

0 I f the paths will cross, whether they will cross a t the same time.

Since perceiving a motorcycle's distance and speed may be d i f f i c u l t (see Section 3 . 3 ) , determining the timing of path crossing could be

more d i f f i c u l t f o r other dr ivers . Thus, the degraded cues fo r motion

and speed perception provided by motorcycles can lead other drivers t o e r r in deciding how t o r eac t .

However, even i f other drivers can accurately ant ic ipate confl i c t s 8

with motorcycles, they m i g h t decide to increase the r isk o f a conf l i c t because t h e i r personal loss will be l e ss than in the case of a c a r ,

3.5 Concl us i ons

Th.e material presented in t h i s section provides a number of possible explanations fo r the differences in accident trends comparing

cars and motorcycles noted in Section 2 . 0 . Clearly, the material reviewed does not indicate which of the explanations i s most 1 ikely

t rue nor does i t rule out any. Furthermore, i t indicates a motor- cyc le ' s "conspicui ty" involves a complex se r ies of detect ion, i den t i f i -

ca t ion, and decision by the other d r ive r . Further research may c l a r i f y

these issues ; in the meantime i t would be best to assume tha t counter- measures should, as f a r as p r ac t i c a l , be e f fec t ive across as broad a

range of possible deficiences as possible.

4 . 0 FIELD TEST METHODOLOGY

4 . 1 Introduction

The purpose of t h i s study was t o determine whether the conspicuity

of motorcycles/motorcycl i s t s could be improved in a way which would

have a meaningful e f f e c t on multi-vehicle motorcycle crashes. To do

t h i s , various conspicui ty-increasing treatments were fabricated and '

tested using a r e a l i s t i c driving s i tua t ion which involved measuring

the response of naive dr ivers . This section will describe h o w the

t e s t was carr ied o u t .

4 . 2 Cri terion

A gap-acceptance measure was employed. The in ten t was t o build

d is t r ibut ions showing the probabil i ty of a c a r ' s pull ing out in f ront

of a motorcycle as a function of the distance between the motorcycle

and the nearest car in f ront of i t t raveling in the same di rect ion.

A typical s i t ua t i on , and the terminology which shall be used, are

shown in Figure 4-1. The s t ra tegy i s simple. A gap i s created in the

t r a f f i c stream between a lead vehicle and a t e s t vehicle (generally a motorcycle in t h i s s tudy) . The dr iver o f the subject vehicle may

"accept" the g a p , t h a t i s merge with or cross the t r a f f i c stream, o r

" r e j ec t " i t .

Figure 4-2 i s a hypothetical d i s t r ibu t ion showing h o w the resu l t s

of the study might appear. Basically, the probabil i ty of a g a p being

accepted varies as a function of i t s s i z e , Changes in the conspicuity

of the t e s t motorcycle and/or r ide r would be expected t o change the

location of the d i s t r ibu t ion along the abscissa, o r a t l e a s t change

the location of the small gap end.

This measure i s useful i f motorcycle crashes r e su l t a t l e a s t in

part from a f a i r l y general a t t i t ude o r response charac te r i s t i c on the

par t of automobile d r ive rs . If the problem a r i s e s from inappropriate

responses on the part of a very small f rac t ion o f the driving public,

then the data collect ion e f f o r t would become impractically 1 arge, and

SUBJECT

LEAD L VEHICLE """ TEST VEHICLE

- - - - - - - _ I - -

I+-------- GAP

Figure 4-1. Schematic of t y p i c a l g a p s i t u a t i o n employed i n t e s t .

PROBABILITY OF GAP BEING ACCEPTED

the approach described here would not be useful . I n order to construct the required distributions i t was neces-

sary to know gap s ize , whether the subject vehicle accepted the gap or not, and the type of maneuver the subject driver executed or planned to execute (since the type of maneuver probably affects the gap size deemed acceptable). How th i s was accomplished will be des-

1

cribed shortly.

4.3 Equipment

4.3.1 Test Treatments. The study called for testing of t rea t - ments under both day and night conditions. Some were applied t o the motorcycle, others t o the r ider . A demonstration of more than th i r ty possible treatments was staged for NHTSA o f f i c i a l s , w h o selected the most promising ones for t e s t . Certain other treatments (lane posi- tion and following car) were selected t o provide basic information regarding conspicuity in the absence of special treatments,

The treatments were as fol 1 ows :

4,3 ,1 ,1 Motorcycle

x: 1. Car control. A 1969 maroon Plymouth station wagon was

used.

2 . Motorcycle control. A normal motorcycle with no 1 ights was used. The driver wore dark clothing and ei ther a white or dark colored helmet.

3. Orange fluorescent fa i r ing . The bike was equipped w i t h a fa ir ing to increase the frontal area. An orange fluorescent fabric was stretched over the ent i re fa i r ing , including the head1 ight aper twe. (See Figure 4-3 . )

4 . Green fluorescent fair ing. This was the same as described above, except for the color o f the fabric .

5. Headlamp o n , The bike ran with low beam o n .

Figure 4-3. Fairing w i t h f1 uorescent cover i n place.

6 . Modulating headlamp. The high beam filament was modulated

from ful l to low intensity a t about 3 hz . Low beam filament was of f .

7 . Reduced brightness head1 amp. A neutral density f i 1 t e r reduced the intensity of the low beam t o 1110th normal.

8. Orange fluorescent ou t f i t . The same material as used in treatment 3 was made into a vest and helmet cover to be worn by the rider.

9 . Green fluorescent ou t f i t . Same as treatment 8 , using the

green material . 10. Orange vest. Just the vest from treatment 8 was used.

11. Orange cap. Just the helmet cover from treatment 8 was used.

12. Following car. To t e s t the effect a nearby car might have on the gap acceptance behavior of subject vehicles, t es t s were run using the station wagon described in treatment 1 .

I t followed the bike in the same lane a t 50 fee t (about one

second headway) in one condition and 200 fee t (about 4

seconds headway) for the other condition. The motorcycle ran under control (treatment 2 ) conditions .

13. Lane position. A1 1 motorcycle conditions described above were run with the bike in the center of the right lane. Additional data were taken with the control bike in the l e f t

1/3rd of the lane ( l e f t position) and the right 113rd of the lane ( r ight posit ion).

14. Combination. The best r ider treatment (number 8 ) and bike treatment (number 6 ) , based on t e s t resu l t s , were combined.

Night:

1, Car control. The same vehicle was used as in the day condi-

t ion. Low beam headlamps were used.

2. Motorcycle control. This was the same as in the day condi- t ion, except that the low beam headlamp was in use,

3. Retroreflective fair ing. The fairing was covered with a retroreflective fabric , leaving the headlamp aperture open. The low beam headlamp was also on in t h i s , as a l l other night treatments. The specific 1 uminance of th is material ( a t an entrance angle of -4' and an observation angle o f

2 0.2') was about 330 cdl f t c / f t . The material appeared white when viewed with headlamps a t n i g h t . For purposes o f ' comparison, most white retroref lect i ve materials used in road signs have a specific luminance (new) ranging from about 100 to 250 cd/f t c / f t 2 .

Since a retroreflector can only work i f car headlamps are directed toward i t , pre-crash configurations in which the car i s a t r ight angles to the bike should not be affected.

4. Running l ights . The turn signal lamps were on full time (not flashing).

5 . Retroreflective ou t f i t . The same material described in treatment 3 was used t o make a vest and helmet cover.

4.3.1.2 Moped. The moped was tested only during the day. Two treatment levels were used:

1. Control. No l ights or other treatments were applied t o the moped. The rider wore dark clothing and no helmet.

2 . Flag. Condition 1 was supplemented by a triangular orange fluorescent flag mounted on a six foot long flexible mast, This i s a cormon bicycle accessory and was purchased a t a bicycle shop.

Other motorcycle treatments were planned b u t no data were taken. These were as follows:

1. Wheel markers. Retroreflective tabs were attached t o the wheel rim. I t was intended that these would create a flashing ef fec t t o approaching vehicles. However, f l icker

fusion occurred a t normal operating speed and the effect was

los t .

2 . Motion perception aid. Note in Figure 4-4 that one of the

motorcycles has an array of five lamps below the handlebars. These could be caused to flash so that the center one came on f i r s t , was extinguished, and the pair outboard of the

j

center one came o n , and were in turn extinguished with the most outboard lamps coming on. This cycle was repeated a t

about 0.5 h z . The idea was to provide a display which made the bike appear t o grow larger (hence be approaching fas te r ) than i t actually was. However, when used in t r i a l s

the system was apparently confused w i t h an emergency vehi- c le so that lead vehicles pulled to the side of the road.

4.3.2 Vehicles. Two motorcycles were used in the data collection stage. Both were donated by Honda Motor Company. One was a 1974 350 cc 4 cylinder and the other was a 1976 400 cc 4 cylinder. The two bi kes are shown in Figure 4-4.

The moped was a 1973 Ciao. I t was typical of mopeds i n general and had a t o p speed of about 25 mph.

4 .3 .3 Instrumentation. The instrumentation requirements for this project posed rather substantial limitations in that s ize , weight a n d

power consumption h a d to be kept to a minimum. A t the same time, the equipment had t o be accurate and easy t o operate. Finally, since thousands of individual t r i a l s were t o be collected, data reduction had t o be effected in a n economical manner. As f inal ly developed, the equipment met these demands very well.

The equipment carried on the motorcycles consisted of units which generated, conditioned and stored signals. Distance traveled was measured by means of a magnetic sensor mounted on the chain guard as shown in Figure 4-5. Two metal markers attached t o the wheel triggered

Figure 4-4. Motorcycles used i n t e s t program.

Figure 4 - 5 . Magnetic sensor and wheel marker.

the sensor and produced a signal which, when conditioned, resulted in

a spike of fixed duration and amplitude. Since the wheel traveled 75

inches per rev01 ution, each spike occurred a t 37.5 inches. Event

markers in the form of push buttons were mounted on o r near the handlebars for easy access by the r ide r (see Figure 4-6). Pressing

any of these buttons converted the next wheel pulse in to a spike hav- ing a d i s t inc t ive amp1 i tude (e ight levels were possible) . s

The events of i n t e r e s t were:

1. The s t a r t and end of a gap.

2 . Whether the subject vehicle accepted the gap.

3. The type of maneuver the subject driver executed or planned t o execute.

Data were collected as follows:

The t e s t r ide r sought to position the motorcycle t o the rear of a group of cars and open a gap between the rearmost car ( lead vehicle)

and the t e s t motorcycle su f f i c ien t t o produce a reasonable likelihood

of acceptance by vehicles seeking to execute maneuvers of the type of

i n t e r e s t . There were three such maneuvers. These are described below and i l l u s t r a t ed in Figure 4-7.

1. The subject vehicle i s on the r igh t of the t e s t vehicle and i t s driver i s seeking t o cross the road completely o r turn

l e f t and go in the opposite d i rect ion. This shall be referred t o as " r igh t - cross or l e f t t u rn , "

2 . The subject vehicle i s on the r igh t of the t e s t vehicle and i t s driver i s seeking t o turn r igh t and go in the same

di rect ion. This shall be referred t o as " r ight - r ight tu rn . "

3. The subject vehicle i s in the center of the road, facing toward the t e s t vehicle and i t s driver i s seeking t o make a l e f t turn across the path of the t e s t vehicle. This shall be referred to as "center - l e f t turn ."

F i g u r e 4-6. Photograph of c o n t r o l s a r e a , showing push- b u t t o n s used f o r cod ing d a t a .

RIGHT - CROSS OR I$([ ( LEFT TURN

RIGHT - 1 1 RIGHT TURN

- - - - - - - E l l

I CENTER - L 3 LEFT TURN 1A

F i a u r e 4-7. Schematics o f tne t h r e e maneuvers i n v e s t i g a t e d i n :he Presen t s t u d y .

The t e s t riders were instructed that when the lead vehicle passed

a subject vehicle standing in e i ther of the positions shown in Figure

4-7, they should press the "1" b u t t o n located near the l e f t handgrip.

This marked the s t a r t of the gap. If the subject driver accepted the

gap, the rider pressed the " 2 " b u t t o n located near the right handgrip

as soon as the subject vehicle began to move. Then the rider pressed

the " 2 " button again while passing the point where the subject vehicle

was or had been standing t o mark the end of the gap. All that remained

a t t h i s time was t o press one of three buttons located just below the

speedometer to indicate the type of maneuver. These data were - recorded on a small cassette audio recorder which was carried on the

rear of the saddle as shown in Figure 4-8.

The resultant records appeared as shown in Figure 4-9. The t o p

tracing i s an example of a gap which was accepted. The small marks

are from the wheel pulsor, each one representing about three fee t of

t ravel . The gap f i r s t became available t o the subject driver a t the

point marked by the 1 pulse, and the motorcycle passed the place where

the vehicle had been located a t the point marked by the second 2 pulse.

Acceptance i s indicated by the presence of two 2 pulses. The code

pulse ( c ) indicates t h a t t h i s was a type 3 maneuver (center - l e f t turn) as described in Figure 4-7 . Gap s ize i s determined by counting

the number of wheel pulses intervening between the 1 and second 2

pulse and multiplying by 37.5 inches.

The lower trace i s an example of a rejected gap. In this case

the maneuver was a type 2 ( right - right turn) as described in

Figure 4-7.

The accuracy of the system was checked in three different ways.

F i rs t the bike was walked through a series of measured intervals. I t

was then ridden a t 40 mph with the motorcyclist pressing the 1 a n d 2

buttons while passing markers se t a t k n o w n intervals. Some data were

then collected a l o n g with an independent written record of the rough

gap s ize ( large, medium or small) a n d type of maneuver t o verify that the computer data processing system yielded the same resul ts .

Figure 4- 8. Rear of motorcycle s a d d l e , showing recorder i n p o s i t i o n .

The system described worked very we1 1 , a1 though i t required

regular maintenance and occasional repai r . Because much of the night data were collected a t a point remote from HSRI, a system was developed which required less maintenance.

In t h i s system the sensor drove a d igi ta l counter, which was mounted over the bike speedometer. Pressing the "1" b u t t o n rese t and

s tar ted the pulse counter and a timer mounted next to i t . When the ' r ider passed the subject vehicle position he pressed the button again to stop the counters. The information on the counters, together with

the maneuver and accept-reject data were recorded verbally on a small tape recorder.

The main advantage to th i s system was freedom from maintenance. Data reduction was a slow, manual process however.

4 .3 .4 Test s i t e . A s i t e was sought which had a h i g h volume of vehicles attempting the maneuvers of i n t e r e s t . A reasonable volume

of paral lel t r a f f i c was required as well , to provide lead vehicles for the front end of the gap. The s i t e used fo r the day data collec- tion i s a major thoroughfare connecting the two c i t i e s of Ann Arbor

and Ypsilanti, Michigan. In terms of data points per hour, i t seemed f a r the best s i t e of those explored. I t i s f ive lanes wide (center

lane for l e f t turns) and 1 ined for much of i t s length with small businesses of various kinds. Speed 1 imits were 45 mph for about 3/4ths of the four-mile length, 35 mph fo r the r e s t . There are three stop l igh t s in the t e s t section.

Most night data were collected during the winter months in the c i t y of Gainesvil l e , Florida. A major thoroughfare having many of the character is t ics of the northern s i t e was used. Data taken on the

same configurations in both s i t e s did not d i f fe r s t a t i s t i c a l l y .

There were some d i f f i cu l t i e s wi t h both motorcycle s i t e s . Briefly, these were as fol lows :

1. Data could not be taken on vehicles attempting to merge or

cross from the l e f t s ide of the s t r e e t . Due t o the w i d t h

of the s t r e e t , vehicles making a l e f t merge from the l e f t side would n o t have been affected by a single vehicle in the f a r right lane. Full crossing maneuvers rarely occurred except a t signal ized intersections. Thus, the "right-cross or l e f t turn" maneuver was almost always a "right - 1 e f t turn. "

I t was easily possible for other vehicles t o pass the t e s t 3

vehicle. This made i t more d i f f i cu l t t o take data, especially in heavy t r a f f i c . I t also required the t e s t r ider to closely monitor t r a f f i c behind him, to be certain that he was sufficiently isolated so that subject drivers would be responding only t o the t e s t vehicle. I t will be recalled that one of the conditions ("following car") offered an opportunity to examine th is e f f ec t , The resu l t s , as will be noted shortly, provide reason t o believe that the presence of a car close t o the motorcycle might have an effect on the response of merging drivers.

3. There were questions concerning the gap s t a r t point for some maneuvers. For example, i f the rearmost car of the group leading the motorcycle was in the l e f t lane, what effect would this have on a driver making a r ight-right turn maneuver (type 2 in Figure 4 -7 )? No def ini te answer can be provided. Data were taken with the rearmost car in ei ther lane under the assumption that differences, i f any, would apply equally t o a l l conditions. I t was also d i f f i - cul t t o judge the e f fec t of opposing t r a f f i c on drivers

attempting r ight - le f t turn maneuvers (type 1 maneuvers in Figure 7 ) . When facing t r a f f i c from the r ight , most drivers f i r s t turned into the l e f t turn lane, then merged right when the opportunity presented i tsel f . Some drivers may have waited until there was an adequate g a p in both directions. The simplest solution was t o take data as though there was no oncoming t r a f f i c and assume t h a t d i f -

ferences would be distributed equally over a1 1 conditions .

Finding a t e s t s i t e for the moped was very d i f f i c u l t . Because

i t s speed capabi l i t ies were so 1 imited, d a t a could n o t be taken a t the

motorcycle s i t e . An attempt was made t o use some secondary roads in

the area , b u t the t r a f f i c volumes were t o o low t o provide adequate

d a t a . The t e s t was f i na l l y s e t u p a t a shopping mall, The moped

c i rc led the ma1 1 on the i n t e r i o r road, c loses t t o the buildings (a

route a b o u t two miles long) a n d data were taken on cars moving in a n d '

o u t of the parking l o t s .

4 . 4 Data Reduction Procedure

Data reduction was completed in three phases. A computer program,

called PULDIG, digi t ized the analog s ignals . The signals corresponded

t o the "bl ips" i l l u s t r a t ed in Figure 4-9. PULDIG translated the b l ips ,

made by the motorcycle's wheel pulses and buttons, into numbers. The

s ize of the number corresponded t o the height of the bl ip . Since the

pulser and d i f fe ren t but tons produced blips of d i f ferent heights,

PULDIG's numbers coded the se r ies of wheel pulses and b u t t o n pulses.

In the second phase, a second computer program, called AUTRED,

ident i f ied t r i a l s from the se r ies of PULDIG-coded numbers, computed

the time and distance gaps of each t r i a l , a n d coded the maneuver

performed by the subject vehicle. AUTRED ident i f ied t r i a l s by look-

ing for se r ies of PULDIG codes beginning with a No. 1 b u t t o n push

followed by one or two No. 2 b u t t o n push(es) interspersed with wheel

pulses a n d ending with a maneuver code. A U T R E D also ident i f ied non-

sensical se r ies of PULDIG numbers t o provide the experimenters with

feedback on the performance of the motorcycles and instrumentation.

As A U T R E D looked for t r i a l s , i t counted the distance and time between

b u t t o n pushes. The calculat ions were based on the wheel pulses.

AUTRED coded the maneuver based on the numbers representing the b u t t o n pushes.

Table 4-1 presents the coding functions used t o t r ans la te motor-

cycl i s t a n d instrument actions into numbers fo r the computer programs.

The columns represent possible actions by the motorcycle ins trumenta-

t ion (wheel pulses) and experimenter-motorcycl i s t ( b u t t o n pushes).

TABLE 4-1. Coding f unc t i ons f o r t r a n s l a t i n g data f rom m o t o r c y c l i s t ' s a c t i o n t o t r i a l codes.

Trans 1 a t i on

Meaning i n experiment

Vol tage o f b l i p on analog tape

PULDIG code number

Meaning t o AUTRED

I Wheel Pulse

1 Motorcyc le had t r a v e l e d another 37.5 inches

Increment t ime and d is tance counters

A p o t e n t i a l maneuver gap had j u s t opened

EXPERIMENT ACTION

New tri a1 begins

# 1 Bu t ton Push

E i t h e r sub- j e c t v e h i c l e began maneuver, o r motorcyc le passed i n i t i a l veh i - c l e l o c a t i o n

3

Maneuver beg inn ing marker; tri a1 compl e t i on marker

I

#2 But ton Push

Right -cross o r l e f t t u r n

#3 Bu t ton Push

Maneuver Code

1

#4 t #5 But ton Push 1 But ton Push

R i g h t - r i g h t C e n t e r - l e f t t u r n 1 t u r n

I

Maneuver Manuever Code Code

The f i r s t row describes what meaning this action had in the experiment. The second row indicates the voltage (height) of the blip p u t on the tape by each action (see Figure 4-9 for examples). The third row shows P U L D I G code ranges for the voltages. The ranges are much larger than those PULDIG actually p u t out; in general, PULDIG p u t o u t numbers which were the voltages mu1 tip1 ied by 100. The fourth row 1 i s t s each code's meaning t o A U T R E D ; AUTRED re1 ied upon this information t o a

identify t r i a l s .

I n the third and final phase of data reduction, a senior experi- menter transcribed the time and distance gaps of each t r i a l o n t o d a t a - coding sheets. The experimenter recorded counts of the gaps which f e l l in different intervals, and whether they were accepted or rejected. The transcription phase a1 lowed senior s ta f f t o constantly monitor data qua1 i ty . The experimenter discarded the few ( less t h a n 1%) nonsensical t r i a l s during transcription.

I n sum, the three phases of data reduction translated the wheel

p u l se/button push records into frequency distributions of time and

distance gaps. Constant quality control by senior s ta f f ensured the data were reliable and accurate.

As noted, both time and distance data were taken. Both forms showed the same resu l t s ; however, the distance data were noisier t h a n the time data. Therefore, the t ine data were used in the full analysis reported here.

4 . 5 Data Analysis Procedure

The purpose of the s t a t i s t i ca l analyses was t o determine which conspicuity treatments--in comparison t o a no-treatment control condition--served to reduce the acceptance rate of small gaps. Two

separate s t a t i s t i ca l procedures were u t i 1 ized: (1) probit analysis, and ( 2 ) re jected-gap analysis. These analyses serve- t o corroborate

one another.

The pr9bi t analysis technique (e .g . , Finney , 1971) i s specifically designed t o deal with population response frequencies a t varying levels

of an independent variable. The basic data in this study are the

numbers of motorists accepting or rejecting gaps o f various sizes.

I n theory every motorist has a threshold gap size in a given s i tuat ion.

Beneath th is threshold the gap will be rejected, above i t the gap will

be accepted. I t was anticipated that the gap thresholds for a popula-

tion of motorists would be approximately normally distributed across

a logarithmic transformation of the gap s izes . If the exact normal 5

distribution were known, i t would be possible t o predict the popula-

tion acceptance rate for any gap s ize . I t i s the goal of the probit

analysis t o f i t the best normal distribution t o the data.

Since the probi t analysis requires frequencies a t specific g a p

s izes , the data were tabulated into 2 by 10 frequency tables, one for

each maneuver and treatment. The two rows of the tables referred t o accepted and rejected gaps respectively, the ten columns referred t o

gap s ize midpoints, in seconds, ranging from 1 t o 10. These tables were then supplied as input t o an i te ra t ive probit computer program

capable of f i t t i n g a normal distribution to data within a specified error tolerance. The o u t p u t for each treatment was a 1 inear regres-

sion function: normal standard deviates as predicted by (logarithmic)

gap s izes . Thus, for any specified gap s ize , the treatments can be

ranked by simply comparing the predicted proportions of motorist

responding .

There i s one aspect of the probit analysis technique which pre- sents potential diff icul ty with regard t o these data. A normal distribution can be best predicted when most of the data f a l l near

the middle of the distribution. This study necessarily emphasized

data collection a t small gap sizes--where less than 50% of the popu-

lation of motorists respond. I t i s therefore appropriate t o verify the probit analysis results with a technique that i s less sensitive t o emphasis a t particular gap s izes . The second analysis was based on the probability t h a t the subject drivers w o u l d re ject gaps of

various s izes . I t will be referred t o as a "Rejected Gap Analysis."

I n most cases the two forms o f analysis agree. Where they d o

not, i n t e rp re ta t ions a re based on the re jec ted gap analys is , d u e to the possible problem with probit mentioned above.

5.0 RESULTS AND DISCUSSION

The resu l t s of both the probit and rejected gap analyses are pre- sented in t h i s section. The various conspicuity treatments are separated depending on whether they were intended fo r day or night

use. Car-fol lowing, lane-posi t i o n , and the moped r e su l t s , follow the main analyses. d

I t will be recalled tha t measurements were made on three maneu- vers of the subject vehicle. The resul ts from the three maneuvers

-- d i f f e r , so each treatment or c lass of treatments i s considered

separately under each of the maneuvers.

5.1 Daytime Treatments

5.1.1 R i g h t - Cross or Left Turn. Figure 5 .1 i s a diagram of th i s maneuver. Of the three maneuvers, t h i s was probably the most d i f f i -

cu l t fo r the subject motorist , since he/she must be concerned with

t r a f f i c from two di rect ions .

Figure 5-2 plots the probit regression curves fo r the conspicuity treatments, and motorcycle and automobile controls . Values on the

ordinate are proportional to the probability of gap acceptance, so the lower the point of in tercept , the lower the probability o f short gaps being accepted, and the be t t e r the treatment. These data indicate that there was a higher probabil i t y of subject drivers accepting smaller

gaps when confronted w i t h the control motorcycle and dimmed headlamp conditions. The r e s t of the conditions were associated with a dimi-

nished 1 i kel i hood of accepting small gaps.

Table 5-1 presents the rejected gaps analysis fo r the right-cross or l e f t turn maneuver. Three analyses are shown. The leftmost i s fo r gaps u p t o and including three seconds. The second analysis includes the t o t a l s from the f i r s t and adds gaps u p to and including four seconds. Gaps u p t o and including f ive seconds are added in the l a s t

analysis . Data f o r the control motorcycle are provided a t the t o p of

Figure 5-1. Diagram of a right - cross or l e f t turn maneuver.

'I-

u 8

v, v,

2 -3

v,

-g 0 U

Ln

d-

- -- -- -- I

a . aJ 7 N o m 3 ' L II 0 t 'a O " C 1 = 4 - 0 3 O N

L C, II s 0 2 - 0

z . --

N

% 4- -7

u

p.

Z

LO N cn cn r-4 d u2 03 LD co 03 N cn d- cu b m o 6 0 3 ' ,-I N N N N 4 N m m N

-.- - - - . --- - - ;-

- - 7 = -

* * * % % * * * * %

N LO LD m M d- d- 4 cd c3 h 03 r-4 ,-I d b b d- Ln

0 I 4 ,-I N N N Cu m N I I I I I I I

----- - -

N d- * u) U3 Ln Ln h h C\1 0 0 0 0 0 0 0 0 0 0 0

I I I I 1 I I I - ---- N d- u2 a h 03 Cn Ln 0 h h cn 01 cn cn cn cn m m cn 03 0

-- -- - m 0 m 0 u3 M d h 4 d- a d cn cn N m N cn 03 2 '2 4 ,I 4 N N m * c u 4

each group of columns. The meaning of the column headings i s as

fol 1 ows :

N = number of subject vehicles included.

P = percentage of subject vehicles rejecting the g a p .

'di f f = P for control motorcycle minus P for the treatment condition.

Z = Z score, on which s t a t i s t i ca l differences are estimated.

The treatments are grouped within the table as follows:

Automobile control

High v i s ib i l i t y materials applied t o the bike

High v is ib i l i ty materials worn by the rider

Lighting treatments

Combination

I n general, the rejected gap analysis shows the same trends as the probit analysis. For gaps of three seconds or l e s s , any of the tested treatments are better than the control motorcycle. The dimmed headlamp, combination, and automobile control are n o t s t a t i s t i c a l l y

different , however.

The situation changes as larger gaps are considered. Nominally , this i s of l i t t l e concern, since the function of the treatments i s t o protect against acceptance of very short gaps. However, i t i s interesting t o note the continued effect o f the modulated headlamp

even a t gaps as large as five seconds. This may represent a novelty e f fec t . If so, i t would be expected to diminish were the treatment to be commonly employed.

5.1.2 Center - Left Turn. Figure 5-3 i s a diagram of the center- l e f t turn maneuver.

Figure 5-4 plots the probit regression lines for the maneuver. Two groups appear t o be divided by the automobile control within the

4) i - - - - - - - - - - - - - - 4-m Ell 1

CCW

Figure 5-3. Diagram o f a center - l e f t turn maneuver.

graph. The control motorcycle, dimmed head1 amp, 1 ights-on , and treated fairings bunch together with relatively poor performance in the small gap region. In comparison, motorcyclist garments and the modulating headlamp show re1 atively good performance. The automobile control probit f a l l s in between the two clusters , b u t closer to the most effective conditions.

Table 5-2 presents the rejected gap analysis for th is maneuver. ' I n general, the resul ts compare well with those for the right-cross or l e f t turn maneuver. The differences tend n o t t o be as 1 arge for this maneuver, and there are fewer s t a t i s t i c a l l y significant d i f fe r - ences a t the smaller gaps. Interestingly, the differences between t e s t and control conditions become more pronounced as larger gaps are included in the analysis. This may represent a novelty effect : however, the performance of the headlamp on condition argues against this interpretaion. (A1 t h o u g h Michigan does n o t have a 1 ights-on law, many motorcyclists ride with l ights on in daytime. Thus the subject

drivers should have been used t o seeing motorcycles with 1 ights on . )

5.1.3 Right-Right Turn. From the point of view of the subject driver, th i s i s probably the simplest of the three maneuvers. I t i s diagrammed in Figure 5-5.

Figure 5-6 presents the probit traces for the conspicuity t rea t - ments in the right-right turn maneuver. No clear-cut pattern or

profile i s evident in that graph, compared with the previous probit graphs.

Table 5-3 presents the rejected gap analysis for the right-right turn maneuver. The probabilities appear much different than for the other two maneuvers. There are a few s t a t i s t i ca l ly significant differences, b u t these are not consistent and, as a matter of f a c t , occur about as often as would be expected based on chance.

Overall, both the probit and rejected g a p analyses provide no strong indication that any condition i s different from the motorcycle control on the right-right turn maneuver. The apparent reason for the

v, s 0 'I- C, .r w s 0 u

'I- +-'

ri- 0

I

m cn

oZI

* CU m

N a a s ' 7 g 7 0 I+ I d d r-l r-l

I .----.

CJ m N d - N C U M U) + I I 0 0 0 0 0 0 0 0 0

0 0 a a I I I I I I I I

. . a , ~h Ur -

$ 5 ) L II 0

cc b

u

d I d d N C U C r ) C r ) rl . d I I 1 I I I I I I I

U) N ri- d h O g z z Ln CU d- o o o o o o 4

+'a, a I I I I I I I I I I I

Figure 5-5. Diagram of a right - right t u r n maneuver.

w vrl 7 cot

I I

I . . I I r-l I I 4

P a . " I:; o Ten 4 0

vr aJ C, a 0 ' r- -0 s '7

* X

u s 5

m 0

LO aJ C, 5 U .r u s .r

* ,$.

n

1, a 7 aJ s 0 u

a.! U s a U .I-

cc ' r s m

* r LO

+ 0

7

aJ > aJ r-

4

vr a: C, 5 0 .t

-0 s .r

=?2

a m 'Q d 0 a o o c n N r-l h cn a3 cn al a c n c n c o 01 cn S O

lack of differences i s the very high probabil i ty of shor t gaps by the control bike being rejected (0.98), leaving l i t t l e room for improve-

ment. Why the subject drivers behaved more cautiously on t h i s maneu-

ver than on the other two i s not c l e a r , although i t may a r i s e from the f ac t t ha t t h i s maneuver places one in the path of an approaching vehicle much longer than the other two.

In teres t ingly , the r ight- r ight turn maneuver corresponds to the ' maneuver studied by Kirkby & St roud (1978) in t he i r use of the gap

acceptance methodology. They fa i l ed to find s ign i f i can t differences as we1 1.

5.1.4 Summary, Daytime Treatments. These data suggest t ha t the

daytime cons pi cui ty of motorcycles can be improved in several ways

having a meaningful e f f ec t on the drivers of automobiles. Riding

with the headlamp on seems very e f f ec t i ve , although causing the head-

lamp t o modulate from low to h i g h in tensi ty seems t o improve response even more. The wearing of f l uorescent materials a l so seems e f fec t ive , However, applying the same materials t o the motorcycle has a l e s s

pronounced e f f e c t , especi a1 ly fo r the center-1 e f t turn maneuver,

5 . 2 Nighttime Treatments

5 . 2 . 1 Right-Cross o r Left Turn. Figure 5-7 shows the probit

f i t s f o r each of the nighttime conditions. The p rof i l e s are a l l s imi lar except fo r the running 1 ights condition. Running 1 ights

appear t o be somewhat more e f fec t ive than the other condit ions?

Table 5-4 presents the rejected gap analysis f o r the nighttime conditions. Only the running l igh t s treatment appears to o f fe r con- s i s t e n t improvement re1 a t ive to the control bike. However, because

the subject vehicle i s i n i t i a l l y a t r igh t angles t o the approaching t e s t vehicle, r e t ro re f lec t ive treatments would not be expected t o be

e f fec t ive .

5 . 2 . 2 Center-Left Turn. Figure 5-8 presents the probit f i t s fo r t h i s maneuver. The differences are not large: however, the retro- re f l ec t ive treatments appear somewhat be t t e r than the control .

TABLE 5-4. Rejected-gap a n a l y s i s f o r r i g h t - c r o s s o r l e f t tu rn - -n igh t t ime c o n d i t i o n .

# i n d i c a t e s -1 level of s i g n i f i c a n c e (one-way), * i n d i c a t e s .05 and ** i n d i c a t e s -01.

Control Motorcycle ,111111111111- Retroref lect i ve Su i t .-I-.

Y - .

Retroreflect ive Fairing Running Lights

I I I I t I I I I

.I - 2 .3 .4 -5 .6 -7 -8 -9 I

1.0

2 SEC. 3 SEC. 5 SEC. 10 SEC.

LOG SECONDS -1 - L.

*

Figure 5-ti. Pr-obi t r-esul Ls fo r r i i g h t t i ~ ~ ~ e condi t ions--center - 1e- f t t u r n maneuver.

Table 5-5 presents the rejected gap analysis for t h i s maneuver. The re t ro re f lec t ive s u i t , in pa r t i cu la r , appears t o be associated with

s ignif icant ly reduced probabil i t y of short gaps being accepted. As

in the case of the daytime treatments with t h i s maneuver, ef fect ive- ness continues out to gaps as large as f ive seconds.

5.2.3 Right-Right Turn. Figure 5-9 presents the probit f i t s fo r t h i s maneuver. While a l l treatments seems be t t e r than the contra\, the differences are small .

Table 5-6 presents the rejected gap analysis for t h i s maneuver. Only the car control condition appears consistently be t t e r than the

motorcycle control . I t i s t rue tha t the re t ro re f lec t ive fa i r ing

treatment appears marginally be t t e r , however the - n i s small. In

addi t ion, the i n i t i a l r igh t angle orientat ion of the vehicles means the headlamps of the subject vehicle could not have impinged on the motorcycle.

The cautious behavior of the subject drivers noted fo r t h i s maneuver during the day i s not evident in the nighttime. Whether t h i s i s a real difference or r e f l e c t s chance variat ions i s not c lea r .

5.2.4 Summary, Nighttime Treatments. These data suggest t ha t the use of running l igh t s and re t ro re f lec t ive clothing may improve nighttime conspicui t y . Running 1 ights appear t o be l ess e f fec t ive for head-on confrontations than does the re t ro re f l ec t i ve gear. This i s unfortunate, because the re t ro re f lec t ive treatments cannot be of assistance unless the c a r ' s headlamps are pointed toward the bike.

Therefore, the use of both treatments seems advisable,

5.3 Motorcycle Lateral Position

There i s some controversy among motorcyclists as to whether they should r ide in the center , r igh t or l e f t portion of the lane. There are a number of points t o consider, one of which i s the e f f ec t on vehicles seeking to maneuver across the motorcycl e J s path, The present data may provide some insight on t h i s issue.

m o ,

-2 = 4 0 F-d- h al h d u orc a m 07 cn cn aJ L v, C, I I

c m O Z 0

N 0 U) M Z m d h 4

rc d 4

'I- + 'r u s 0 u

rl 0

v, aJ C, a u

-r- u t

'I-

$: -N

u s a

L n 0

v, aJ c, a u .I-

0 c .I-

$:

n

2 Sf aJ s 0 V

aJ U s a u .r (4- .I-

s m ' r- Ln

rc a v aJ > QI v

,-I

v, aJ C, id 0 'T

-0 8 ' r-

=k

t '

v,

dl

v, U s 0 o aJ ul

LO

ul v,

4

v, u C 0 U a v,

I *

ffl cn

oZI

C o u a v,

m

.. aJ -a2

O=? $ 2 L II 0 + a 0 r: *.

0 ~m O M L + I [ , s O Z 0

. . a, 7 -

ucn 2 3

L II 0 ~a 0 E - -

LO -N O N L t' I1 s 0 O Z

. . aJ

- L n ucn

25' L II 0 +P.

n o . ' 0 5 "

CO -CU O d L C , I l - c 0 2 0

N

q- CC -r

a

a

Z

N

4- 4- .r U a

P

Z

N

4- 4- 'r -0

a

a

Z

C, s aJ E C, a aJ L t-

a aJ > > .r .I- - C, 3, + 0,

o I O C 1 0 I= in C.' L O Q J - r OaJ .r C, c -v L - L L T ~ , c c L 4 J C C . I - -9b.I- s 0 - 7 a o aJaJa a J a J J =*I- Z 0 0 Q : L L d L L ? &A

$: % $: m m h m d 0 m cn m rl I 4 I I I

a2 d 7 4 LO 0 0 0 0

I I I I

UY N cn m cn cn a2 cn

m LO m LO LO cn m m CU N CU

% $: =k LO 0 0 N 03 Lo CO N

M I rl d I I I

a2 CU d M 0 0 0 0

I I I I

OI M u7 b- Ul Cn Dl Or

UY * h h d cn a3 d (V rl N

9 9 * N UY 0 a2 d CU

m N 1-4 ,-I

I I I I

LO m ,-I m 0 0 0 0

I I I I

CO 03 o o cn cn l-4 4

V) I4 ul CU LO d 0 0 d rl rl

.

I t should be noted these data were taken in Florida, where a headl ights-on law appl ies . Therefore, the "control " condition was

also with headl ights on.

Figure 5-10 through 5-12 present the probit f i t s fo r the three

maneuvers studied. In a l l cases the center and l e f t lane positions

are associated with the lowest probabiliy of shor t gaps being accepted, *

with the r i gh t lane position being poorest.

The rejected gap analyses f o r the lane positions a re provided in Table 5-7 through 5-9. Such differences as are present favor the

center lane posi t ion, with the exception of the cen te r - l e f t turn

maneuver, where the l e f t lane position i s very c lose . ( I t should be

noted that in t h i s case, where the concern i s with treatments which

are poorer than the con t ro l , a two-tailed t e s t of significance has been used.)

5.4 Effect of Following Vehicles

I t will be recalled t h a t t h i s condition measured the e f f e c t on gap acceptance of having an automobile e i t he r 50 or 200 f e e t behind the motorcycle. These data were compared with the control motorcycle

and automobi 1 e .

Figures 5-13 through 5-15 show the probit r esu l t s fo r the three maneuvers. These suggest possible di fferences favoring the car and

both the 50 and 200 foot following distances fo r the r ight-cross and l e f t turn maneuver and the 200 foot following distance fo r the r ight-

r igh t turn maneuver.

The rejected gap analysis (Tables 10 through 12) indicates marginal s ignificance only fo r the 50 foot condition in the r ight- cross or l e f t turn maneuver. In the cen te r - l e f t turn maneuver there

are s ign i f i can t differences fo r both the car and the 50 foot condition,

b u t only fo r larger gaps. Final ly , the rejected gap analysis does not support the apparent differences in the probit fo r the r ight- r ight turn maneuver. Differences between treatment and control conditions are generally small and in favor of the control .

= - - Center Lane P o s i t i o n ( con t ro l ) - Right Lane Pos i t i on L e f t Lane Pos i t i on

-1 .2 -3 -4 .5 .6 .7

2 SEC. 3 SEC. 5 SEC.

LOG SECONDS -

10 SEC.

f-igure 5-11. Probi t r e s u l t s f o r l a t e r a l - lane-posi t i o n ana lys i s - - cen t e r - l e f t t u r n sraneuver.

N m 'I- . rc

.C

II u II

n

03 m e. m II

Z

Z

4 0

fn aJ C, m U .r 0 s

* %

u 8 m

LO 0

fn aJ C, m U .r- u s .I-

* n

n

2 7 0 3 C, u

aJ U E m U .I- cc 'I-

s tr, .r fn

't 0

C

aJ > a F

4

V1 aJ C, m 0 ' r -0 s ' F

2k

.r- F u

C, c aJ E C, m aJ L

I-

aJ Q1 t s t m c a 0 -10 -1 .I- 'I-

C, t ' C , C.' 'I- I .I- % v I mv, a J O ' ~ 0 .A& eII

u C . 'I-

.I- .r F

0 g z L F - 7 c,

F - OF- L S 0 0 0 0 6 0 Ink Nk 0 0

V) a 4 6 0 . I- u c 'I-

i * l L i , . A I Urn A .

rrJ 7 - 7

E b z 8 5 2: 1 + Ink Nk U O

a -a UCO

3 ' L l l

4

C, E=

E C, re 2

I-

rn m s s -7 '7 C

0 - - F +J g L

- O F L C 0 0 0 0 re0 L n L L N'A u u

5.5 Moped Cons pi cui t y

I n a p i lo t study using the gap-acceptance methodology, a compari- son was made between the conspicuity of an untreated (control) moped and a moped with a n orange bicycle flag. Figures 5-16 through 5-18 show the probit plots for the key maneuvers; Tables 5-13 through 5-15 present the rejected gap analyses. The flag appears effective in the

9

center-left turn maneuver, and for larger gaps in the right-cross or l e f t turn maneuver as well .

I t i s interesting t o note that motorists were much less l ikely t o re ject short time gaps when confronted by the moped than the motor- cycle. Due to speed differences, the moped a t three seconds was typically about 75 fee t away, the motorcycle 150-200 fee t . I t would take the subject vehicle just as long in ei ther case t o accelerate

through and clear the path for the t e s t vehicle. As a matter o f f ac t , given the t ighter maneuvering space and lower speeds in the shopping mall, such a maneuver may have taken longer in front o f the moped.

Lacking car control data, we do n o t know whether these response characteristics are typical under such conditions. However, our riders experienced a great deal more diff icul ty in th is brief study than in the motorcycle study. They were nearly run off the road or had t o take evasive action on a number of occasions. All i n a l l , these results lead us t o believe that the gap acceptance methodology as applied here may n o t be suitable for use with slower moving vehi- cles such as mopeds.

5.6 Discussion

5.6.1 Methodology. Before s tar t ing a discussion of the results of this study, i t seems appropriate t o spend some time considering the methodology which was used t o collect the data.

I n general, the investigators feel that the gap-acceptance pro- cedure was quite successful in the current application. The data seem meaningful ( i . e m , intui t ively related t o the 1 ikelihood of

ffl VI (U - ca V) w s 0 U aJ ffl

m

- -

ln V) aJ F

a V) u is: 0 0 aJ ffl

d-

- -

ffl V)

aJ 7

ca V) -0 s 0 U aJ V)

m

-

vr - '0 0 s L 0 +J U s a 0 vlo

d

accidents), and can be collected quickly, economically , and with re1 a- t ively simple instrumentation.

Our experience suggests that something l ike one thousand data points are required per treatment t o ensure reliable resul ts . This assumes more or less equal distribution across three maneuvers, or about 300-350 data points per maneuver. I t further assumes that the data are concentrated in the short-gap region, i . e . , f ive seconds or * less .

Two points should be made regarding the validity of the gap acceptance method in this application. F i r s t , the relationship of gap acceptance and accidents i s largely unknown a t this time. I t i s true that riding with headlights on seems t o be an effective accident

countermeasure, as we have noted ea r l i e r . The fact that the headlamp on treatment was effective as measured by gap acceptance in this study i s encouraging evidence of validity. To some extent the finding of Hurt, also mentioned ea r l i e r , that riders wearing high v i s ib i l i t y clothing seem underrepresented in the crash data, also indicates the method i s val id , However, further val idation data would be desirable.

The second point i s a limitation already noted i n Section 4 . 2 ;

that i s , the method can only measure a f a i r ly general response characteristic on the part o f automobile drivers. The fact that

differences were found i n this study does not mean that other, less general responses cannot account for a significant portion of the problem. I f t h i s i s the case, i t i s also possible that different countermeasures may be appropriate.

Finally, a word about safety. The investigators were very con- cerned about safety; we were, a f t e r a l l , asking our riders t o del iber- ate1 y recreate pre-crash configurations known t o be overrepresented in the crash s t a t i s t i c s . Our hope was that the high level of atten- tion required t o be able t o take data would reduce the r isk. I n the more than twenty thousand miles that were accumulated during the t e s t our riders experienced one minor crash (the rider suffered a scraped

knee and the bike was damaged about $200 worth) and several near

misses. In teres t ingly , none of these occurred while data were being

taken, b u t a1 1 involved pre-crash configurations of the c lass ic type. Based on t h i s experience, we feel t ha t the method poses no special dangers to the motorcycle r i ders .

5.6 .2 Means fo r Improving Conspicuity. I t appears t ha t there *

are a number of ways t o improve daytime motorcycle-motorcyclist

conspicuity t ha t should have a meaningful e f fec t on the behavior of car dr ivers . The simplest i s t o drive with the headlamp on a t a11 times. The modulating headlamp i s apparently even more e f f ec t i ve , b u t does require some investment on the part of the motorcyclist. High v i s i b i l i t y materials seem quite ef fect ive as wel l , but , fo r some reason, work be t t e r when worn by the r ide r than when f i t t e d to

the bike.

The l a t t e r finding i s somewhat surpr is ing. In the opinion of the investigators , the fluorescent fa i r ing treatment was in tu i t ive ly a much more e f fec t ive a t tent ion-get ter than the f luorescent vest o r

helmet cover. Yet the f i e l d t e s t data indicate the opposite. This

suggests t ha t laboratory studies of motorcycle conspicuity can pro-

duce misleading resu l t s . However, i t i s not c lea r why the resu l t s came about. One possible explanation i s tha t effectiveness i s improved by height. Another i s tha t by emphasizing the r i d e r , speed- spacing judgments are faci 1 i ta ted . This happens because apparent s i ze i s an important distance cue. However, i t i s based on a know- lege of actual s i z e . Most drivers know l e s s about the s ize of motor- cycles, especial ly motorcycle f a i r i ngs , than they do about people.

When considered against a c r i t e r ion of appeal t o motorcycl i s t s , the headlamp on and modulating headlamp treatments seem l i k e l i e s t t o

be voluntari ly adopted. High-visibility treatments on the helmet may

have appeal to some r i de r s , especial ly i f they can be convinced the treatments are of value. High-visibil i t y garments, however, may present a problem in acceptance.

For nighttime riding conditions there may be value in wearing re t ro re f lec t ive garments and using running 1 igh t s . Retroreflect ive treatments applied t o the bike seem l e s s e f fec t ive b u t may be of help.

I t should be pointed out t ha t there are combination fluorescent-

r e t r o r e f l e c t i ve materials available ( the orange and green materials used in t h i s study were of t h i s type) , which can provide day and

night conspicuity in one package. *

Obviously, r e t ro re f lec t ive treatments can only be of value when

a vehic le ' s headlamps are directed toward the motorcycle. Thus, there i s merit in using another treatment, such as the running light, to aid when confronted with vehicles in other or ienta t ions . I t should

a lso be noted t ha t in a real-world s e t t i ng , a re t ro re f lec t ive jacket would a lso a id in other types of potential conf l i c t s i t ua t i ons , such

as overtaking.

While i t seems unlikely t ha t r e t ro re f lec t ive apparel of the usual type would have much appeal to motorcyclists, they may be wi l l -

ing t o accept additional r e t ro re f lec t ive trim on t h e i r bikes o r helmets. I t i s also possible to t r e a t ordinary fabr ics with beads and make them re t ro re f lec t ive without changing t h e i r appearance under

normal viewing conditions. This may have potential for other vehicles where conspicui ty i s a problem as we1 1 .

The resu l t s of the lane--positioning study indicate t h a t , a t

l e a s t insofar as the e f f ec t on maneuvering cars i s concerned, the

motorcyclist i s be t t e r off in the center of the lane.

There i s some evidence from the resu l t s of the following car study that the presence of a car close behind a motorcycle may reduce

the likelihood of short gaps being accepted, a t l e a s t fo r the r ight-

cross and l e f t turn maneuver. While t h i s information i s probably of no practical benefi t as an accident countermeasure, i t does mean tha t

investigators who use gap acceptance in the future must be careful t o keep the t e s t vehicle as i sola ted as possible.

The resu l t s of the moped study suggest that the use of a v i s i - b i l i t y aid such as the fluorescent i l ag may be benef ic ia l . Indeed, our r iders reported they f e l t much sa fe r w i t h the f lag in place and experienced much less trouble with cars . However, for reasons noted e a r l i e r , we are doubtful t ha t the gap acceptance methodology as used here f o r motorcycles i s applicable to slower-moving vehicles such as mopeds. *

6 . 0 COST-BENEFIT ANALYSIS

As a f ina l evaluation of each conspicuity t reatment 's e f fec t ive - ness , a cost-benefi t analysis was car r ied ou t . This sect ion describes the analysis and presents i t s r e s u l t s .

6 . 1 Assumptions I

Three simplifying assumptions were made i n the present analys is :

1. I t i s assumed t h a t in one mile of driving a motorcyclist enters in to each of the three types of conf l i c t s ( the maneuvers studied in the f i e l d t e s t ) once. This assump- t ion i s 1 i kely conservati ve, and therefore the calculated benefi ts are 1 ikely underestimated.

2 . I t i s assumed t h a t an acceptance of a one-second gap

r e s u l t s in a motorcycle-car crash. I t i s not possible t o determine the exact gap tha t wil l lead t o a crash in each maneuver s i t u a t i o n . However, i t i s 1 ikely t h a t a maneuvering vehicle in a one-second-gap s i tua t ion will be

encroaching on the motorcycle's path a t the end of one second, making t h i s c r i t e r ion plausible. Analysis i s fur ther simp1 i f i e d by holding t h i s c r i t e r i o n constant across the three maneuvers.

3 . I t i s assumed t h a t a l l crashes will manifest the same injury-probabi 1 i ty da ta , regardless of which conspicui ty treatment ( i f any) i s used. In o ther words, i t i s assumed t h a t the conspicui ty treatments introduce no biases in to the type of injury resul t ing from a crash,

6 . 2 Cost-Benefi t Analysis Procedure

For each cons pi cui ty treatment the cost-benef i t analysis com- puted the expected accident cost of each c o n f l i c t between a motor- cycle and a c a r , added the cost of acquiring and using the conspicuity enhancer, and subtracted t h i s sum from the expected accident costs of

an untreated (control ) motorcycle. In equation form:

BENEFITS = COSTSC - COST^ + OPERATING C O S T S ~ )

where

BENEFITS are the economic benefi ts to be derived from the use o f a par t i cu la r conspicuity enhancer

9

COSTSC a r e the expected accident costs f o r an untreated (control ) motorcycle

COSTSE are the expected accident costs f o r a t rea ted (experimental ) motorcycle

OPERATING COSTSE are the costs of implementing and operating the par t i cu la r conspicuity enhancer

The following subsections discuss how each term o f the equation 6-1 was computed,

6 . 2 . 1 Computing the Expected Accident Costs of the Control and Treated Motorcycles. The COSTSC and COSTSE terms of the equation

were computed in an analogous manner. The f i r s t s tep was t o determine,

from the probit equation and a table of cumulative normally d i s t r i - buted p robab i l i t i e s , the probabil i ty of a d r i ve r ' s accepting a one-

second gap. This probabi l i ty , ca l led here PROBcrash, r e f l e c t s the l ikelihood of a co l l i s ion between a car and a motorcycle given t ha t treatment.

The second s tep involved estimating the expected costs of each actual crash. The costs of a crash are ref lec ted in the type of

injury which occurs. Five injury c lass i f i ca t ions were used in our analysis :

Fatal i ty

National Safety Council Class i f ica t ions " C"

No Injury--Property Damage Only ( P D O )

Given tha t a crash has ac tual ly occurred, the probabil i ty of

each type of injury i s indicated i n Table 6-1 (based on the 1975

Texas motorcycle da t a ) .

TABLE 6-1. Probabil i ty of pa r t i cu la r types of in ju r ies occurring as resu l t of a motorcycle accident (from 1975 Texas motorcycle accident data) .

Overall I)ay Night

Fatal ,017 .012 ,031

" A " .I64 ,154 .I99

"8" .368 .371 .358

"C" ,114 .I10 .I27

P DO ,337 .353 ,285

Each of the f ive injury types can be assigned a monetary value

corresponding t o the losses incurred. This i s shown in Table 6-2.

T A B L E 6-2. Total average 1 osses fo r motorcycle accidents .

Sources: a U,S. Department of Transportation--NHTSA estimate. 1975 Societal Costs of Motor Vehicle Accidents (December , 1 9 7 6 ) . Page 2 , Table 2 .

b National Safety Council estimates HSRI estimate

The t h i rd and f inal s tep in the process involved putt ing the

above information in to a working equation: The expected costs of an actual crash were derived by multiplying the monetary cost of an

injury type by the probabil i ty of t ha t injury type, a n d by summing

across injury types. By multiplying t h i s summation, i n t u rn , by the

probabil i t y of a crash, we derived the expected accident costs f o r

a pa r t i cu la r treatment (or control ) .

In equation form:

COSTS = C ( ~ ' ~ ' l n j u r y Type X LOSS Injury T Y P ~ ) pRo'~rash Injury Type

Table 6-3 shows the expected crash costs f o r each treatment. I t

i s evident from Table 6-3 t h a t the expected crash costs vary across

the maneuvers. In general , however, they are lowest fo r day t r e a t -

ments 5, 6 , 7 , and 9 , and lowest fo r night treatments 3 and 4.

6.2.2 Computing the Operating Costs fo r Each Conspicui ty Treat- ment. The operating costs of an untreated motorcycle were s e t as - the zero baseline. I t was assumed t ha t a motorcycle las ted f ive

years and ran 4,500 miles ( s p l i t in to 3,510 daytime and 990 nighttime

miles according t o Polanis ' 1978 data) each of the f i ve years ( M V M A

Fact Book, 1976).

In general , the per-mile and hence per-conf l ic t cos t of using a

conspicuity treatment was computed by adding the treatment 's pur-

chase price to i t s energy, maintenance, and replacement costs over

f i ve years , and dividing by the to ta l mileage over f ive years .

Table 6-4 elaborates on the computation fo r each treatment.

6 .3 Cost-Benefit Analysis Results

Table 6-5 shows the per-confl ict benefi ts f o r each conspicuity treatments. A negative benefi t value indicates the treatment i s

expected to cos t more than the benefi ts derived from the use of the treatment. The daytime data shows t h a t wearing a f luorescent vest

y ie lds the highest benef i t . This i s due in pa r t t o i t s absolute

TABLE 6-3. Expected c rash c o s t s ( i n j u r i e s and/or p roper ty damage). All f i g u r e s i n d o l l a r s pe r c o n f l i c t .

I

R-RT C-LT R-CLT

Day

1. Cycle Day Control 6.0543

2. Orange F a i r i n g .0418

3. Green F a i r i n g ,0030

4 . Orange S u i t . 0 148

5. Green S u i t .00017

6 . Orange Vest .000003

7. Orange Helmet .000006 7

1 8 . Lights On

9 . Modulated Headlamp

10. Headlamp l / l O t h I n t .

1 Night I 1. Cycle Night Control .8875 10.0157 .0900

2 . R e t r o r e f l e c t i v e S u i t .0178 1.2678 ,7607

3 . R e t r o r e f l e c t i v e F a i r i n g .0127 1.205 .I242

4 . Running L igh t s .0305 38.033 ,0006

TABLE 6-4. Ooerat ing costs c f c o n s g i c u i t y enhancers.

YTIME - Yea r l y Energy

Cost ( 3 )

Cost Per Year* 1 1 Conspicui t y I Treatment

a g e r i g 150 o r

Green F a i r i n g

10 ( t o rep lace f l u o - rescent m a t e r i a l )

Purchase Cost ( 9 )

6 (uses aporox. 1 g a l l o n o f gas 8 $1 f o r each 15 hours [600 m i l e s l o f ope ra t i on )

Year ly Maintenance Cost ( 3 )

Orange Helmet 65 helmet 5 p a i n t -

70

Orange ' les t l 2 2 (buy one new ves t each y e a r )

Orange S u i t 0 r

Green S u i t

65 he lmet 5 p a i n t 2 v e s t -

72

2 (new ves t each yea r )

Lights-On 7.5 ( rep lace l i g h t )

10 ( rep lace - 17.5 ::J$r{

yea rs )

1 a (uses app rox i - mate ly 1 g a l l o r o f gas B 51 for each 6 hours i240 m i l e s ] of ope ra t i on )

11.38 ( acco rd ing t o manufacturer , m d u l a t e d head- lamp uses 65% o f 20th energ;, and maintenance o f l i g h t s - o n cos t s )

Modulated Head lam

8 .75 (maintenance and energy cos t s com- m t e d 3s 50% o f : iqh ts -on c o s t s )

I n t e n s i t y

R e t r o r e f l e c t i v e S u i t

Y I GHTTIME 65 heirre'. 10 p a i n t 10 ves t - 85

5 ( p a i n t touch-up and ' les t r e p a i r )

-- -

X e t r o r e f l e c t i v e F a i r i n g

22

10 ( t o rep lace r e t r o - r e f l e c t i v e m t e r i a l )

5 ( r ep lace l i g h t s ) 6.6 ( one b a t t e r y

repiacemenc i n 5 yea rs )

L 2unning L i g h t s

, 0222

* Sased on an average o f 3,513 da -n ine rni ies =or eacn motorcyc le i n t he Y . 5 .

*+ 3ased on an average of 390 n ign t c ime m i l e s f o r gacn n o t o r c y c l e i n t he U . S .

! energy c o s t ) i

6 i 46 (uses aoprox i - n a t e l y 1 g a l l o n 1 o f gas 3 51 f o r 1 each 15 hours [600 rn i ies ] 9 f ope ra t i on

7 18 .6 1 (computed 35 S O X I of l j g n t j - o n i

,3465

.3185

TABLE 6-5. B e n e f i t s pe r conf l i c t o f conspicui t y t r e a t m e n t s (a1 1 f i g u r e s i n d o l l a r s ) .

DAYTIME

NIGHTTIME

Conspicui t y Treatment

Orange F a i r i n g

Green F a i r i n g

Orange S u i t

Green S u i t

Orange Vest Orange He1 met Li ghts-On

Modul a t e d Head1 amp

Head1 amp l / l O t h I n t e n s i t y I

R e t r o r e f l e c t i v e S u i t 1 .8475 1 8.7257 -.6929

Expected b e n e f i t s pe r c o n f l i c t by maneuver

Conspicui t y Treatment Enhancer

Right-Right Turn-

-9.0241

,0589

.0049

.0674

,0700

.0622

.0565

.0262

-2.3545

Expected b e n e f i t s pe r conf l i c t by maneuver

R e t r o r e f l e c t i v e F a i r i n g

R u n n i n g l i g h t s

Center-Lef t Turn '

,0743

- . 9682

. I838

. I838

,1878

,1844

. I270

,1789

-7.215

R i g h t - R i g h t Turn

Right-Cross o r L e f t Turn,

5.9994

6.0382

I 6.0349

6.0496

6.0537

6.0503 ' 6.0408

6.0448

Center-Lef t Turn ' Right-Cross Le f t Turn o r

.8283

.8382

I -7.404

I

8.7642

-28.064

- .0807

,0706

percentage of expected crashes (indicating a raised level of conspi- cuity) plus the fac t that these vests are very inexpensive t o the motorcycl i s t . Other treatments that consistently showed strong potential benefits were the green s u i t , orange helmet, 1 ights-on, and modulating head1 amp,

The nighttime data show that the retroreflective s u i t and f a i r - #

ing are cost-effective on the center-1 eft-turn maneuver. A 1 t h o u g h a1 1

treatments seem equally effective on the right-right-turn maneuver, this i s most likely an a r t i f a c t , since the retroref lect ive treatments cannot work in this configuration.

REFERENCES

Alexander, G . J . , and Lunenfeld, H . P o s i t i v e guidance in t r a f f i c

c o n t r o l . FHWA, Of f i c e o f T r a f f i c Opera t ions , April 1975.

Auber t , H. Die Bewegungsempfindung. Archiv f u r d i e gesammte

Phys io log ie , 1886, - 39, 347-370. 9

Aul horn, E . , and Harms, H . Visual Per imet ry . In D. Jameson and

L . M . Hurvich ( E d s . ) , Handbook of sensory physiology, Vol . V I I l 4 ,

v i sua l psychophysi c s , Berl i n : Spri nger-Verl ag , 1972.

Austin, G . A . The e f f e c t o f s t imulus a rea on v i sua l i n t e n s i t y t h r e s -

ho ld . Unpublished doc tora l d i s s e r t a t i o n , The Univers i ty o f

Mi chi gan , 1951.

Baker, C . A . , and Steedman, W . C . Perceived movement i n depth a s a f unc t i on of luminance and v e l o c i t y . Human Fac to r s , 1961, - 3 ,

166-173.

B a r t l e y , S. H . Sub j ec t i ve b r i gh tne s s i n r e l a t i o n t o f l a s h r a t e and

the 1 igh t -dark r a t i o . Journal of Experimental Psycho1 ogy, 1938,

B a r t o l , J . A., L i v e r s , G . D . , & Mienner t , R . Requirements a n a l y s i s

and f e a s i bi 1 i t y s t u d i e s f o r an experimental s a f e t y motorcycle ,

AMF Inc . , Advanced Systems Laboratory, Go le t a , C A , J u ly 1975,

Bee le r , J . W . Visual t h r e sho ld changes r e s u l t i n g from a spontaneous

s accad i c eye movement. Vision Research, 1967, - 7 , 769-775.

Blackwel l , H . R . Con t r a s t t h r e sho ld of the human e y e . Journal of

the Opt ical Soc ie ty o f America, 1946, - 36, 624-643.

Blackwel l , H . R . Development and use of a q u a n t i t a t i v e method f o r

s p e c i f i c a t i o n of i n t e r i o r i l l umina t i on l e v e l s on t he ba s i s o f

performance d a t a . I1 1 uminating Engineer ing, 1959, - 54, 317-353.

Boynton, R . M . Optimal v i sua l sea rch techn iques . Summary and d i s -

cu s s ion , P a r t I . In Visual Search Techniques, National Academy

o f Sciences--National Research Counci 1 , Pub1 i c a t i o n 712, 1960.

Brown, B . Resolut ion t h r e sho ld s f o r moving t a r g e t s a t the fovea and

i n t h e pe r i phe ra l r e t i n a . Vision Research, 1972, - 12 , 293-304.

Brown, R . H . Complete s p a t i a l summation in the pe r i phe ra l r e t i n a . , American Journa l of Psychology, 1947, - 60, 254-259.

Brown, R . H. Weber r a t i o f o r v i sua l d i s c r im ina t i on o f v e l o c i t y .

Sc ience , 1960, - 131, 1809-1810.

Burg, A . La t e r a l v i sua l f i e l d a s r e l a t e d t o age and s ex . Journal o f

Appl i e d Psychology, 1968, - 52 , 10-15.

Burg, A . , and Beers , J . R e f l e c t o r i z a t i o n f o r n igh t t ime consp i cu i t y o f

b i cyc l e s and motorcycles . Journal o f Sa f e ty Research, 1978, - 1 0 , 69-77.

Car ra ro , B. Update: Motorcycle a c c i d e n t s in 1977, T r a f f i c Safety , ,

Vol, 79, No. 2 , February 1979, pp. 8-11 and 29.

Cole , B . L . Visual a s p e c t s of road eng inee r i ng . Aus t r a l i an Road

Research Board, Proceedings of t h e S ix th confe rence , P a r t I ,

Connolly, P . L . Visual cons ide r a t i ons of man, t h e v e h i c l e , and t h e

highway, P a r t I I . Soc i e ty of Automotive Engineers , SP-279, 1966.

Crawford, A . The percep t ion o f l i g h t s i g n a l s : The e f f e c t of the

number of i r r e l e v a n t 1 i gh t s . --- Ergonomics , 1962, - 5 , 417-428.

Crawford, A . The percep t ion of l i g h t s i g n a l s : The e f f e c t o f mixing

f l a s h i n g and s t e ady i r r e l e v a n t ? i g h t s . -..-- Ergonon!ics, 1963, - 6 ,

287-294.

Di tchburn, R . W . Eye movements i n r e l a t i o n t o r e t i n a l a c t i o n .

Opt ica Acta, 1955, - 1, 171-176.

Evans, L . , and Rothery, R . Cetect ion of the s ign of r e l a t i v e motion

when fol lowing a veh i c l e . H~rnan Fac to r s , 16 ( 2 ) , 161-173, 1974.

Feinberg, R . A s tudy of some aspec t s of per ipheral v isual a c u i t y .

P a r t s I , 11, 111, and IV. American Journal o f Optometry, 1948,

26, 49-56; 105-118. - F e l l , J . C . A motor veh ic le acc iden t causal system: the human e l e -

ment. Human Fac to r s , 1976, 18 , 85-94. * -

Finney, D . J . P rob i t Ana lys i s , 3rd e d i t i o n , Cambridge (England) :

Cambridge Univers i ty P r e s s , 1971.

Gerathewohl , S . J . Conspicuity of s teady and f l a sh ing 1 i g h t s i g n a l s : v a r i a t i o n of c o n t r a s t . Journal of t he Optical Soc ie ty of

America, 1953, - 43, 567-571.

Gerathewohl , S. J . Conspicuity o f f l a sh ing l i g h t s i g n a l s : e f f e c t s

of v a r i a t i o n among frequency, du ra t i on , and c o n t r a s t o f t he

s i g n a l s . Journal of the Optical Society of America, 1957, - 47,

27-29.

Gordon, D. A . The r e l a t i o n between t h e th resho ld o f form, motion and

displacement in parafoveal and per iphera l v i s ion a t a sco top ic

1 eve1 of i 11 umi n a t i on . American Journal of Psycho1 ogy , 1947, - 6 0 ,

202-225.

Graham, C . H . , and B a r t l e t t , N . R . The r e l a t i o n of s i z e of s t imulus

and i n t e n s i t y in the human eye . 11. I n t e n s i t y th resho lds f o r

red and v i o l e t 1 i g h t . Journal o f Experimental Psychology, 1939,

24, 574-587. -

Graham, C . H . , Brown, R . H . , and Mote, F . A . The r e l a t i o n of s i z e of

s t imulus and i n t e n s i t y i n the human eye. I . I n t e n s i t y th resho lds

for whi te 1 i g h t . Journal of Experimental Psycho1 ogy , 1939, - 24,

Green, D . G . Regional v a r i a t i o n s in t he visual a c u i t y f o r i n t e r f e r -

ence f r i nges on the r e t i n a . Journal of Physiology, 1970, - 207,

351-356.

Green, D . M . , a n d Swets, J . A . Signal detection theory and psycho-

physics. New York: Wiley, 1966.

Gr i f f in , L . I . Motorcycle accidents: Who, when, where, and why.

University of North Carolina, Highway Safety Research Center,

March, 1974,

Halstead, W. C . A note on the Bartley e f f e c t in the estimation of

equivalent brightness. Journal of Experimental Psychol ogy,

Harano, R . M . , a n d Peck, R . C . The California motorcycle study;

dr iver and accident character is t i c s , Cal i fornia Department of

Motor Vehicles, Division of Administration, 1968.

Held, H . H . Peripheral acui ty . Br i t i sh Journal of Physiological

Optics, 1959, - 16, 126-143.

Henderson, R . L . , and Burg, A . Vision a n d audition in driving. Final

repor t . System Development Corporation, Santa Monica, Ca., 1974.

Hershenson, M . Perception of l e t t e r arrays as a function of absolute

re t ina l focus. Journal of Experimental Psycho1 ogy , 1969, - 80,

201-202.

H i l l s , B . L . V i s i b i l i t y under night driving conditions. Part I .

Laboratory background and theoret ica l considerat ions.

Austral i an Road Research, 1976, - 6 ( 2 ) , 3-8,

Hochberg , J . Components of 1 i teracy : Spaculation and exploration

research. In H . Levin and J .P . Williams (Eds . ) , Basic Studies

on Reading. New York: Basic Books, 1970.

Hoffman, E . R . Detection of vehicle velocity changes in car following,

Proceedings, Austral i an Road Research Board, Fourth Conference,

Hoffman, E. R . Note on detection of vehicle velocity changes. Human

Factors, 1966, - 8, 139-141.

Hoffman, E . R . Percept ion o f r e l a t i v e v e l o c i t y . In R . G . Mortimer

e t a l . , S t u d i e s of automobile and t ruck r e a r l i g h t i n g and

s i g n a l i ng sys terns. HSRI r e p o r t UM-HSRI-74-25, 1974.

Hol ly , F . Saccadic p r e s e n t a t i o n o f a moving t a r g e t . Vision Research,

1975, - 15 , 331-335.

Howett, G . L . , Kel ly , K. L . , and P i e r c e , E . T . Emergency-vehicle 9

warning l i g h t s : S t a t e of the a r t s . NBS Specia l P u b l i c a t i o n ,

480-16, 1978.

J a n o f f , M . S. Motorcycle n o t i c e a b i l i t y and s a f e t y - .. during t h e daytime.

In Proceedings of the 2nd I n t e r n a t i o n a l Congress on Automotive

S a f e t y . Paper No. 73034, 1973.

J a n o f f , M . S . , and Cassel l , A . E f f e c t o f daytime motorcycle headl i g h t

law on motorcycle a c c i d e n t s . Research Record, 1971, $377, 56-63.

J a n o f f , M . S . , C a s s e l , A . , F e r t n e r , K . S . , & Smierciak, E . S. Daytime

motorcycle headl i g h t and t a i l 1 i g h t o p e r a t i o n . Frankl in I n s t i t u t e

Report No. Fc2488, August, 1970, DOT, HS800-321.

Kerr, J . L . Visual r e s o l u t i o n i n t h e pe r iphery . Percept ion and

Psychophysics , 1971, - 9 , 375-378.

Kestenbaum, A . Cl i n i c a l methods o f neuro-ophthalmologic examination

(2nd e d . ) . New York: Grune & S t r a t t o n , 1961.

Kinchla, R . A , , and Al lan , 1. G . The theory of v i sua l movement

p e r c e p t i o n . Psychological Review, 1969, - 76, 537-558.

Kirkby, C . , and S t roud , P . G . E f f e c t s o f m o t o r c y c l i s t s ' h igh-

v i s i bi 1 i t y a i d s on d r i v e r gap-acceptance. T r a f f i c Engineer ing

and Control , Vol . 10, No, 819, August/September 1978, 401-403.

Kle in , G . S . The r e l a t i o n between motion and form a c u i t y in para-

foveal and per iphera l v i s i o n and r e l a t e d phenomena. Archives

o f Psychology, 1942, - 3 9 ( 2 7 5 ) , 1-70.

K r i s t o f f e r s o n , A . B . Foveal i n t e n s i t y d i s c r i m i n a t i o n a s a f u n c t i o n

o f a r e a . Unpubl i shed d o c t o r a l d i s s e r t a t i o n , The U n i v e r s i t y o f

Michigan, 1954.

La tour , P . L . Visual t h r e s h o l d dur ing eye movements. Vis ion Research,

1962, - 2 , 261-262.

Lamar, E . S . , Hecht, S . , S c h l a e r , S . , & Hendley, C . D. S i z e , s h a p e , 9

and c o n t r a s t i n d e t e c t i o n o f t a r g e t s by d a y l i g h t v i s i o n . I. Data and a n a l y t i c a l d e s c r i p t i o n . Journal of t h e Opt ica l S o c i e t y

o f America, 1947, - 37, 531-545.

Low, F . N . P e r i p h e r a l v i s u a l a c u i t y . Archives o f Oph thalrnol ogy , 1951, - 45, 80-99.

Low, F . N . The p e r i p h e r a l motion a c u i t y o f 50 s u b j e c t s . American

Journa l of Phys io logy , 1947, - 148, 124-133.

Ludwigh, E . Extrafoveal v i s u a l a c u i t y a s measured w i t h Snel l e n t e s t -

l e t t e r s . American Journa l o f Ophthalmology, 1941, - 24 , 303-310.

McColgin, F . H . Movement t h r e s h o l d s i n p e r i p h e r a l v i s i o n , Journa l of

t h e Opt ical S o c i e t y o f America, 1960, - 5 0 , 774-779.

Mackworth, N . H. , and Bruner, J . S . How a d u l t s and chi1 dren s e a r c h

and recognize p i c t u r e s . Human Development, 1970, - 1 3 , 149-177,

Mackworth, N . H . , and Morandi, A . J . The gaze s e l e c t s i n f o r m a t i v e

d e t a i l s w i t h i n p i c t u r e s . Pe rcep t ion and Psychophysics , 1967,

Mandel baum, J . , and S l o a n , L . L . P e r i p h e r a l v i s u a l a c u i t y . American

Journa l of Ophthalmology , 1947, 30, 581-588.

Michaels , R . M . Percep tua l f a c t o r s i n c a r fo l lowing . Proceedings

o f t h e Second I n t e r n a t i o n a l Symposium on t h e Theory o f T r a f f i c

Flow. London, 1963, 44-59. - Michaels , R . M . , and Cozan, L . W . Perceptual and f i e l d f a c t o r s caus ing

1 a t e r a l d i sp lacement . Highway Research Record, 1963, - 25, 1-13.

Mit r an i , L . , Yakimoff, N . , & Meteeff , S . Dependence of visual

suppress ion on the angula r s i z e of voluntary saccad ic eye move-

ments. Vision Research, 1970, - 10, 411-415.

Moreland, J . D . Pe r iphera l c o l o r v i s i on . In D . Jameson and C . M .

Hurvich (Eds . ) . Handbook of sensory physiology, VII /4 , Be r l i n :

Springer-Verl ag , 1972. 9

Mourant , R . R . , and Rockwell , T . H . Mapping eye-movement p a t t e r n s t o the v i sua l scene in d r i v i n g : An exp lana tory s tudy . Human

Fac to r s , 1970, - 12 , 81-87.

Mourant, R . R . , and Rockwell , T . H . Visual scan pa t t e rn s of novice

and exper ienced d r i v e r s . In Psycho1 ogical a spec t s of d r i v e r

behavior , vo l . 11, papers presented a t the In t e rna t i ona l Sympo-

sium on Psychological Aspects o f Driver Behavior, held a t

Noordwi j kerhout , The Nether lands , 1971.

Nagayama, Y . , Mori ta , T . , Miura, T . , Watanabem, J . , & Murakami, N .

Mo to rcyc l i s t ' s v isual scanning pa t t e rn in comparison with auto-

mobile d r i v e r s . Presented a t t he Soc ie ty of Automobile Engi - neers Congress and Exposi t ion. Paper No. S A E 790252, 1979.

Notton, D . , and S t a r k , I. Eye movements and visual pe r cep t i on ,

S c i e n t i f i c American, 1971, 224 (6 ) , 34-43.

Olson, P . L . The de t ec t i on of spacing changes i n c a r fo l lowing .

Research r e p o r t TP-25, General Motors Research Labo ra to r i e s ,

September 1971.

Os te rberg , G . A . Topography of the l a y e r of rods and cones i n the

human r e t i n a . Acta Ophthalmologica Supplementum, 1935, - 6 , 1-102.

Pearce , D . , and P o r t e r , E . Changes in visual s e n s i t i v i t y a s soc i a t ed

w i t h voluntary saccades . Psychonomic Sc ience , 1970, - 19 , 225-227.

P i renne , M . H . Vision and the eye , 2nd ed . London: Associated Book

Pub1 i s h e r s , 1967.

Po l an i s , S. F . Motorcycle a cc iden t p a t t e r n s i n Pennsylvania. T r a f f i c

Q u a r t e r l y , 1979, 2, 139-155.

Polyak, S. L . The r e t i n a . Chicago: The Un ive r s i t y o f Chicago

Press, 1941.

P r o j e c t o r , T. H . , and Cook, K . G . Should r e a r l i g h t s of motor vehi-

c l e s be c o l o r coded? Journal of t h e 11 1 umination Engineering

S o c i e t y , 1972, - 1, 135-142.

Ramsey, J . D . , and Br ink ley , W. A . Enhanced motorcycle n o t i c e a b i l i t y 9

through daytime use o f v i sua l s i gna l warning dev i ce s . Journal

of Sa f e ty Research, 1977, - 9 , 77-84.

Re i s s , M . L . , Berger , W . G . , & V a l l e t t e , G . R . Analysis o f motor-

c y c l e a c c i d e n t r e p o r t s and s t a t i s t i c s . BioTechnology, Fa1 1 s

Church, V A . , February 1974.

Re i s s , H . L . , and Haley, J . A . Motorcycle s a f e t y , f i n a l r e p o r t .

Cutler-Hammer, I n c . , Airborne Inst ruments Lab, New York,

May 1968.

Richards , W . Saccad ic supp re s s ion . Journal o f the Opt ica l Soc i e ty

of America, 1969, - 59, 617-623.

R i o p e l l e , A . J . , and Chow, K . L . Sco top ic a r e a - i n t e n s i t y r e l a t i o n s

a t va r ious r e t i n a l 1 o c a t i ons . Journal of Experimental Psycho1 ogy , 1953, 46 , 314-318.

Robinson, G . H . Visual sea rch by automobile d r i v e r s . Human F a c t o r s .

1972, 14 , 314-323.

Robinson, G . H., C l a rk , R . L. , Erickson, D. J , , & Thurs ton , G . L ,

Visual sea rch by automobi 1 e d r i v e r s . In Psychological a s p e c t s

of d r i v e r behav ior , Vol. I , Papers presented a t the I n t e r n a t i o n a l

Symposi urn on Psychological Aspects of Driver Behavior , he1 d a t

Noordwi jkerhout , The Nether1 ands , 1971.

S a l v a t o r e , S. The constancy of t h e weber r a t i o i n c a r - fo l l owing .

Paper p resen ted a t t h e Ohio S t a t e Univers i ty Symposium on c a r - fo l 1 owing , Apri 1 6 , 1965.

Schmidt, I . Visual cons idera t ions of man, the veh i c l e , and t he

highway, P a r t I . Soc ie ty of Automotive Engineers, SP-279, 1966.

Schiffman, H. R . Some components of sensa t ion and percept ion f o r the reading process . Reading Research Q u a r t e r l y , 1972, - 7 , 588-612.

Sh ina r , D. Psychology on t he road: The human f a c t o r i n t r a f f i c

s a f e t y . New York: Wiley, 1978. 9

Smith, D . I . A n i nve s t i ga t i on t o determine whether the daytime usage

o f motorcycle head l igh t s and t a i l l i g h t s should be made compulsory

in Western A u s t r a l i a . Road T r a f f i c Authori ty,-July 1975.

Spe r l i ng , G . , and Speelman, R . Visual s p a t i a l l o c a l i z a t i o n dur ing

o b j e c t mo t ion , apparent o b j e c t motion and image motion produced

by eye movements. Paper presented a t t h e convention of t he

Optical Society of America, 1965. Abstracted in Journal of the Opt ical Soc ie ty of America, 1965, - 55, 1576.

Steedman, W. C . , and Baker, C . A . Perceived movement in depth as a

funct ion of s t imulus s i z e . Human Fac tors , 1962, - 4 , 349-354.

Tay lor , J . H . Use of v i sua l performance da ta in v i s i b i l i t y p r ed i c t i on

Applied Opt ics , 1964, - 3, 562-569.

Thomas, E . I. Movements o f the e,ye. S c i e n t i f i c American, 1968, - 219 (2 ) , 88-95.

Thomas, E . I. Search behavior . The Radiologic C l i n i c s of North

America. 1969, - 7 , 403-417.

Townsend, C . A , , and Fry, G . A . Automatic scanning of a e r i a l p h o t o - graphs. I n Visual Search Techniques, National Academy of

Sciences--National Research Council , Pub l ica t ion 712, 1960,

T ra tne r , A. Accidents and i n j u r i e s : Why they happen. Rider , 1978,

5 , ( 2 ) , 42-43, 74-80, & 84. -

Trever then, C . B . Two mechanism of v i s ion in pr imates . Psycho-

log i sche Forschung, 1968, - 31, 299-337.

Vincent, R . J . , Brown, B . R . , Markley, R . P . , & Arnoul t , M . D . Dis-

tance discrimination in a simulated space environment. Perception

& Psychophysi cs , 1969, - 5 , 235-238.

Vol kman, F. C . Vision during voluntary saccadic eye movements.

Journal of the Optical Society of America, 1962, - 52, 571-578.

Volkman, F . C., Schick, A. M . L . , & Riggs, I. A . Time course of 9

visual inhibi t ion during vol untary saccades . Journal of the

Optical Society of America, 1968, - 58, 562-569.

Wal dram, J . M . Vision a n d eye movements of motor drivers . - New

S c i e n t i s t , 1960, - 8 , 1264-1267.

Wallach, H . , and Lewis, C. The e f f e c t of abnormal displacement of

the re t ina l image during eye movements. Perception and Psycho-

physics, 1965, - 1, 25-29.

Waller, P . F . An analysis of motorcycle accidents with recommendations

fo r 1 icensing a n d operation. Highway Safety Research Center,

University of North Carol i na , Chapel Hi 11 , North Carol i na , 1972.

Waller, P . F . , and Gr i f f in , L . I . The impact of a motorcycle l i gh t s -

on law. In D . F. Hue1 ke ( E d . ) American Association fo r the A u t o - motive Medicine. Proceedings of the 21st Conference, 1977.

Walls, G . L . The vertebrate eye and i t s adaptive radia t ion. Bloomfield

Hi 1 1 s , Michigan : The Cranbrook I n s t i t u t e of Science, 1942.

Wertheim, T . Uber die indirekte Sehscharfe. Zei t s c h r i f t f u r

Psychologie and Physiologie der Sinnesorgane, 1894, 1, 172-187.

Will iams , L . G . Studies of extrafoveal discrimination a n d detect ion.

I n Visual Search, Synposium conducted a t the meeting of the

Commi t t e e on Vision, Division of Behavioral Sciences, National

Research Counci 1 , 1970.

Williams, M . J , , and Hoffman, E . R. The influence of motorcycle v i s i -

bi 1 i ty in t r a f f i c accidents. Me1 bourne University, Department

of Mechanical Engineering, Parkvil l e (Austra l ia) , July 1977.

Wolf, E . S tud ies on shr inkage of t he visual f i e l d with age. Highway

Research Record, No. 164, pp. 1-7, 1967.

Noltman, H . L . , and Aust in , R . L . Some day and night t ime visual

a spec t s of motorcycle s a f e t y . In Proceedings of t he 2nd I n t e r -

na t iona l Congress on Automotive Sa fe ty , Paper No. 73036, 1973.

Yarbus, A . L . Eye movements and v i s i o n . New York: Plenum P re s s , , 1967.

Zubek, B . L . , and S t a r k , L . Saccadic suppress ion: Elevation of

visual thresh01 d a s soc i a t ed with saccad ic eye movements,

Experimental Neurology, 1966, - 16, 65-79.