Chapter 4

Analysis of Police Crash Report Narratives

 

Chapter 4 Table of Contents

4.1 Introduction

4.2 Purpose

4.3 Method

4.4 Results

4.5 Discussion

4.6 Conclusions

 

4.1 Introduction

 

Given the nature and extent of cellular telephone use in the automobile today, the potential implications for safety on the road are obvious. The predicted growth in cellular telephone use along with the implementation of increasingly complex functionality (e.g., e-mail, paging, access to the WWW) heightens the importance of understanding the potential implications for safety as well as the nature of causal factors associated with any relevant crashes. Such information could be invaluable to both system designers and users.

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives

 

4.2 Purpose

 

This research was undertaken to supplement existing data and provide a somewhat different perspective on the relationship between cellular telephone use and safety. In addressing this type of issue, it is common practice for researchers to examine crash records, histories, or data bases as a means of determining common trends, contributing factors, and causes. When crash data bases are studied, information is usually gleaned from information filled in by investigating officers. Examples of information gathered in this manner include alcohol involvement, character of the crash, and whether or not seat belts were used. Such information is usually contained in specific check-off or fill-in "boxes" on crash reporting forms.

When dealing with potential sources or contributing factors, the fill-in and check off approach appears to work well for conventional causes of crashes. However, when the potential source is unusual or relatively new, performing searches on categorized or "boxed" information may not uncover the true influence of the potential source on number of crashes. In such cases, a different approach must be used.

Many crash reporting forms contain so-called narratives in which the reporting officer describes in a few sentences how the crash occurred and what the contributing factors were. These narratives usually allow greater freedom in describing the crash and therefore may contain more detailed information on contributing factors or causes. To take advantage of these narratives, they must be entered into a data base and then they must be retrievable by keywords.

The State of North Carolina has had an ongoing project for many years that is being carried out jointly by the State's Department of Motor Vehicles (DMV) and the Highway Safety Research Center (HSRC) at the University of North Carolina, Chapel Hill. DMV enters the narratives as they become available, and HSRC uses computer search programs that find, retrieve, and print narratives containing keywords. The printed narratives can then be screened by researchers to determine whether they do indeed involve the target influence or cause, or are spurious.

This chapter describes an attempt to use the keyword-narrative search approach to determine the extent to which cellular telephone usage in vehicles is contributing to crashes.

This chapter describes an attempt to use the keyword-narrative search approach to determine the extent to which cellular telephone usage in vehicles is contributing to crashes. Because cellular telephones are a relatively recent technology, more conventional approaches to crash database searches are not likely to provide accurate information. It appeared that the keyword-narrative search approach would be more likely to produce accurate and meaningful results.

Of course, there are other research approaches to understanding the effects of cellular telephone usage on driver behavior and driver workload, many of which have been discussed in Chapter 3 of this report. A bibliography is also provided as Appendix E, referencing the variety of methods that can be used. However, none of these methods directly assesses the effects of cellular telephone usage on the number of crashes per se. Rather, the connection is implied through such measures of cognitive load as eyes-off-road time, lane deviations, and missed detections of targets. It thus appears that this study is the first attempt to assess the occurrence and nature of cellular telephone related crashes in a crash data base.

Previous searches using the narrative approach have worked reasonably well. Perel (1976, 1988) has used the approach with reasonable success to determine driver-vehicle interaction problems, particularly those involving hand and foot controls. More recently, Wierwille and Tijerina (1995) have used the approach successfully to show that increased allocation of vision to interior sources of the vehicle is associated with increased crash incidence.

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives

4.3 Method

 

To conduct the search, a set of keywords and combinations had to be constructed. This list was developed based on words that reporting officers might employ to describe cellular telephones or their use. Such terms as "car phone" and "talk ... on" were included.

The list is shown in Table 4-1. It should be noted that the list is written using a coding scheme. A comma means that the stem word may have any ending. For example "phon," will retrieve any narrative containing any of the words phone, phoning, phoned, phones, etc. An asterisk between words indicates that the two words need not appear consecutively. For example "speak,*on" will retrieve a narrative with the words "speaking to his wife on." Thus, the list was constructed to retrieve as many relevant citations as possible. Of course, the list was expected to provide a large number of spurious citations, which had to be screened by direct reading.

The list shown in Table 4-1 also contains a number of computer-related terms. At the time that the study was planned, there were anecdotal indications that in-vehicle use of computers, facsimile machines, and data terminals might be causing some crashes. Since such devices are sometimes connected through modems to cellular telephones, it seemed prudent to perform the searches simultaneously. In that way, it might be possible to uncover behaviors involving cellular telephones as part of a system in which information is being transferred with computers.

Table 4-1. Keyword list used for the data base search

answer,

comput,

laptop,

phon,

auto,*dial,

convers,*on

listen,*on

port,*comput,

beep,

dial,

mac,

port,*phon,

call,*on

facsi,

microcomp,

powerbook,

carfon,

fasci,

minicomp,

receiv,

carphon,

fax,

mob,*phon,

reciev,

carteleph,

fon,

modem

ring,

cell,

handset

modum

speak,*on

celphon,

hang,*up

notebook,

talk,*on

celul,

headset

PC,

teleph,

com,*link,

keyboard

P.C.,

tel,*number

Personnel at HSRC were able to obtain large data bases for the years 1989, 1992, 1993, 1994, and for the first part of 1995. Databases of sufficient size were not available for the years 1990 and 1991. Search programs were prepared for each of the latest five available years, they were run, and the narratives containing keywords were retrieved and printed. The printed narratives were then transferred to the author for careful examination. The main reason for studying the most recent five years of data was to determine if there were trends occurring.

Following receipt of the results, all narratives were carefully read to determine whether cellular telephones or connections to cellular telephones (such as when using a fax) were primary contributors to crashes. If so, the narratives were saved. If not, they were discarded. Thereafter, a classification scheme was developed and saved narratives were categorized using the classification scheme.

 

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives

 

4.4 Results

 

Tables 4-2 through 4-6 (below) contain the results for each of the search years. Included for each year are: the size of the database, the number of computer retrievals or hits, the number of saved narratives, and the number of saved narratives by category. Each table contains both the actual number of saved narratives and the "adjusted number," which assumes a data base size of 200,000.

The adjusted number is obtained by multiplying the number of saved narratives by 200,000, then dividing by the actual data base size, and then rounding to the nearest tenth. The adjusted number is used to compare crash occurrences across years. (The reason for using the number 200,000 is that databases for each given year tend to fluctuate around this number. Since number of entries is not necessarily an indication of the total number of crashes, adjustment must be performed.)

It should be mentioned that, in a few of the saved narratives it could not be determined whether the driver or the front seat passenger was involved with the cellular telephone at the time of the crash. These few cases have been entered as one-half of a crash under the assumption that the driver and passenger would be equally likely to be distracted by the cellular telephone. This assumption is conservative, because it is probably more likely that the driver would be using the cellular telephone than the passenger.

Table 4-2. Cellular Telephone Related Crashes by Category for the Year 1995.

Data Base Size: 127,328 - Initial Number of Retrievals: 522

Actual Number

No. Adjusted to 200,000

Category

1

1.6

Looking at cellular telephone to determine status or connecting cellular telephone to vehicle

1

1.6

Answering cellular telephone or distracted by ringing cellular telephone

0

0

Dialing cellular telephone

7

11

Using cellular telephone

1

1.6

Hanging up cellular telephone (not dropped)

4

6.3

Reaching for cellular telephone (not dropped)

4

6.3

Picking up dropped cellular telephone

1

1.6

Pulling off road to use cellular telephone or moving vehicle for better reception of cellular telephone

19

30

TOTAL CELLULAR TELEPHONE RELATED

0

0

Looking at computer screen or mobile data terminal

1

1.6

Distracted by beeper (pager)

Table 4-3. Cellular Telephone Related Crashes by Category for the Year 1994

Data Base Size: 209347 - Initial Number of Retrievals: 782

Actual Number

No. Adjusted to 200,000

Category

0

0.0

Looking at cellular telephone to determine status or connecting cellular telephone to vehicle

1

1.0

Answering cellular telephone or distracted by ringing cellular telephone

3

2.9

Dialing cellular telephone

12.5

11.9

Using cellular telephone

0

0.0

Hanging up cellular telephone (not dropped)

0

0.0

Reaching for cellular telephone (not dropped)

3

2.9

Picking up dropped cellular telephone

1.5

1.4

Pulling off road to use cellular telephone or moving vehicle for better reception of cellular telephone

21

20.1

TOTAL CELLULAR TELEPHONE RELATED

2

1.9

Looking at computer screen or mobile data terminal

1

1.0

Distracted by beeper (pager)

Table 4-4. Cellular Telephone Related Crashes by Category for the Year 1993.

Data Base Size: 192140 - Initial Number of Retrievals: 637

Actual Number

No. Adjusted to 200,000

Category

2

2.1

Looking at cellular telephone to determine status or connecting cellular telephone to vehicle

3

3.1

Answering cellular telephone or distracted by ringing cellular telephone

3

3.1

Dialing cellular telephone

5

5.2

Using cellular telephone

3

3.1

Hanging up cellular telephone (not dropped)

4

4.2

Reaching for cellular telephone (not dropped)

2

2.1

Picking up dropped cellular telephone

Pulling off road to use cellular telephone or moving vehicle for better reception of cellular telephone

22

22.9

TOTAL CELLULAR TELEPHONE RELATED

0

0.0

Looking at computer screen or mobile data terminal

0

0.0

Distracted by beeper (pager)

Table 4-5. Cellular Telephone Related Crashes by Category for the Year 1992.

Data Base Size: 175178 - Initial Number of Retrievals: 644

Actual Number

No. Adjusted to 200,000

Category

0

0.0

Looking at cellular telephone to determine status or connecting cellular telephone to vehicle

3

3.4

Answering cellular telephone or distracted by ringing cellular telephone

0

0.0

Dialing cellular telephone

7

8.0

Using cellular telephone

1

1.1

Hanging up cellular telephone (not dropped)

2

2.3

Reaching for cellular telephone (not dropped)

1

1.1

Picking up dropped cellular telephone

0

0.0

Pulling off road to use cellular telephone or moving vehicle for better reception of cellular telephone

14

15.9

TOTAL CELLULAR TELEPHONE RELATED

0

0.0

Looking at computer screen or mobile data terminal

0

0.0

Distracted by beeper (pager)

Table 4-6. Cellular Telephone Related Crashes by Category for the Year 1989.

Data Base Size: 189464 - Initial Number of Retrievals: 1307

Actual Number

No. Adjusted to 200,000

Category

0.5

0.5

Looking at cellular telephone to determine status or connecting cellular telephone to vehicle

2

2.1

Answering cellular telephone or distracted by ringing cellular telephone

1

1.1

Dialing cellular telephone

6

6.3

Using cellular telephone

2

1.1

Hanging up cellular telephone (not dropped)

1

1.1

Reaching for cellular telephone (not dropped)

0

0.0

Picking up dropped cellular telephone

0

0.0

Pulling off road to use cellular telephone or moving vehicle for better reception of cellular telephone

12

13.2

TOTAL CELLULAR TELEPHONE RELATED

0

0.0

Looking at computer screen or mobile data terminal

0

0.0

Distracted by beeper (pager)

It should also be mentioned that the categories selected do, in some cases, overlap one another. For example, the "answering the cellular telephone" category overlaps the "using the cellular telephone" category. However, each saved narrative has been placed in only one category, based largely on the wording appearing in the reporting officer's narrative. Thus, the categories provided are not necessarily mutually exclusive, but are logically derived from the narratives.

Further analysis was carried out by comparing total adjusted number of cellular telephone related crashes across years. Figure 4-1 shows a plot of the data by year. Also included in the figure is a parabolic curve that has been hand-fitted to the data. The curve was obtained by forcing a fit through a value of 13 for 1989, a value of 20 for 1993, and a value of 28 for 1995. These three points then define the parabola whose equation appears in the figure. (It should be emphasized that this is a "fitted" curve, not one obtained by mathematical or statistical optimization.)

Table 4-7 summarizes the actual adjusted number of cellular telephone related crashes by year and then uses the fitted parabolic curve to estimate the number of crashes for future years. A regression line fit, to be described subsequently, has also been used to estimate crashes in future years. Of course, projecting to future years in this way is highly speculative, but it does help to illustrate the likely range of adjusted crashes as cellular telephones become more prevalent.

As just indicated, regression using a linear model was performed on the available data. The results of the regression are presented in Table 4-8 and Figure 4-2. The table shows that the slope of the line, strictly speaking, is not significant (a= 0.05), p= 0.056, assuming a two-tailed test.

Figure 4-1 Plot of Adjusted Number of Cellular Telephone Crashes by Year with Fitted Parabolic Curve.

Table 4-7 Actual and Projected Numbers of Adjusted Cellular Telephone Related Crashes by Year.

ACTUAL

YEAR

DATA

x

1989

- - - -

1992

1993

1994

1995

13.2

- - - -

15.9

22.9

20.1

30.0

x

 

PROJECTED

YEAR

Parabolic Fit

Linear Regression

x

 

1996

1997

1998

1999

2000

33.1

39.0

45.6

53.0

61.1

28.5

31.0

33.5

35.9

38.4

However, with the accelerated increase in availability of cellular telephones and the consequent influence their use has on visual allocation, as demonstrated in the review of research (see Chapter 5), it is reasonable to assume that an increase in such use will be associated with an increase in related crashes. Thus, if the prior hypothesis is assumed to be that cellular telephone use (over years) would be associated with increased crashes, then a one-tailed (directional) test would be appropriate. For this case tcritical is t.05, 3df = 2.353, and the slope is then significantly different from zero (p< 0.05). Figure 4-2 shows the plotted data with the regression line superimposed. Also shown are the 95% confidence limits on the regression line. If one chooses to use the regression line to estimate the adjusted number of crashes, its equation is given at the top of the figure and its values are included in Table 4-7.

Another way to view the total adjusted number of cellular telephone related crashes is in comparison with the number of cellular telephones in use (in general) during the same period of time. Such a view gives a more direct indication of whether or not crashes can be expected to increase with cellular telephone prevalence. The Cellular Telecommunications Industry Association (CTIA) provided the following information on the number of cellular telephones in use in the U.S. by year:

1989 3,508,944

1992 11,032,753

1993 16,009,461

1994 24,134,421

Additional information from CTIA indicated that between 1994 and 1995, the number of cellular telephones increased by approximately 9.6 million, leading to an estimate of 33,734,421 for 1995.

Figure 4-2 Plot of Adjusted Number of Cellular Telephone Related Crashes by Year with Regression Line and 95% Confidence Limits on Regression Line

Year vs. Crashes, Crashes = - 4890 + 2.4642 * Year, Correlation: r = 0.86868

Table 4-8 Linear Regression Summary Performed on the Adjusted Number of Crashes by Year.

Regression Summary

R = 0.86868235 R2 = 0.75460903 Adjusted R2 = 0.67281204

F(1, 3) = 9.2254 p< 0.05599 Std. Error of estimate: 3.7354

x N=5

Fitted Parameter

Estimate B

St. Err. of B

t(3)

p-level

Intercpt

&endash;4889.65

1616.573

&endash;3.02470

0.056548

Slope

2.46

0.811

3.03733

0.055985

Again, using a linear model, a regression analysis was carried out with the adjusted number of cellular telephone related crashes in North Carolina as the dependent variable and the number of cellular telephones in use in the U.S. as the independent variable. The results are presented in Table 4-9 and are plotted with 95% confidence limits in Figure 4-3. The figure also designates each data point by its year and provides the regression equation above the plot. The table shows that the slope of the regression line is significant and the R-value is relatively high.

These results show that the number of cellular telephone related crashes is increasing reliably with their prevalence. However, in observing figure 4-3, if the number of crashes per year is divided by the number of cellular telephones in use during that year, the crashes per cellular telephone in use is seen to be decreasing. (This result is discussed later in the chapter.)

A regression analysis was also carried out with the intercept forced through zero. (It could reasonably be assumed that if there were no cellular telephones in use in the U.S., there would be no crashes resulting from them.) The value of the slope under these conditions was 0.988 North Carolina crashes per million cellular telephones in use in the U.S., and the slope was significant t(4) = 6.501, p< 0.01.

The results show that the number of cellular telephone related crashes is increasing with their prevalence.

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives

4.5 Discussion

 

The results of the database search and analysis have provided useful information on cellular telephone related crashes. The results suggest that the number of cellular telephone related crashes is increasing with the growing number of cellular telephones in use. However, given the small sample size and large standard errors of estimate, the results must be considered exploratory rather than confirmatory in nature.

The study also provides clues as to what specific cellular telephone activities are most likely to cause trouble. For example, reaching for cellular telephones or picking up dropped cellular telephones is a major contributor among relevant crashes. Wierwille and Tijerina (1995) observed a similar trend in an earlier study on in-vehicle visual allocation (see Figure 6-1, Chapter 6). In particular, reaching for and picking up dropped items resulted in a large number of crashes. They hypothesized that drivers felt compelled to retrieve the dropped items, because of a perceived sense of urgency to do so.

 

The study also provides clues as to what specific cellular telephone activities are most likely to cause trouble.

The ratio of cellular telephone-related crashes to cellular telephone subscribers is non-constant over the years. In general, it appears to be decreasing in this data set. Due to the sample size, this may simply be due to random variation; the logic of a straight-line fit assumes this. On the other hand, over the period of time covered in the data, there has been a progressive evolution toward hand-held phones, which now constitute the majority of cellular telephones in use. The hand-helds are less likely to be reported than mobile phone (installed) units since they are less likely to be seen by police investigating a crash. This would explain the downward trend in the crash involvement ratio that may be derived from the data. Hence, the only legitimate way to understand the true ratio is to secure the cooperation of cellular telephone carriers and check for cellular telephone use during the pre-crash period of every crash that occurs.

Furthermore, a conclusion that cellular telephones are getting safer over time is not warranted based on the currently available data. We know that hand-held cellular telephones represent more than 70% of current cellular telephone phone sales. An examination of the crash data and the case studies shows that nearly all the crashes included in this report involved phones of the hand-held type. This over representation compared to the proportion of hand-held units sold would suggest that drivers using hand-held units may experience a greater risk of crash involvement. The risk may be associated with specific design factors such as the flip top design or the smaller keypad architecture, or it may be related to mounting and accessibility issues. The preponderance of cellular telephone related crashes reported to be associated with responding to a call in the Japanese data, where the use of handhelds is even more widespread than in the United States, lends further credence to this argument.

As an aside, the present study did not uncover a large number of computer-related crashes. There were only two instances of such crashes, both occurring in 1994. Additional instances, including a crash that involved the use of a fax machine, were noted earlier in the discussion of the FARS and NASS data (see Section 3). While the incidence of such crashes is small, it does indicate the willingness of some drivers to use such devices in vehicles while driving.

To the extent that the computer and fax functionality is now being incorporated directly into cellular telephone architectures, there is some concern that such expanded and convenient use beyond voice communication may further compromise safety when these functions are accessed from a moving vehicle. Given the expense of these systems, however, it may be some time before they are generally available at affordable prices. As the availability of laptop computers and various cellular interfaces become more prevalent, however, the likelihood that they will appear as a contributing factor in crash databases might increase. It may be a bit too early for these devices to appear frequently as a contributing factor in the databases at this time.

This study uncovered relatively low numbers of crashes resulting from cellular telephone usage. However, through other, more comprehensive crash analyses it has generally been recognized that driver inattention/distraction is associated with between 30 and 50 percent of crashes (Sussman, Bishop, et al, 1985).

In the Wierwille and Tijerina (1995) study, based on an analysis of the same database, driver inattention/distraction was associated with only about 1.5 percent of crashes. This finding suggests that the reporting of crashes as inattention/distraction related on the North Carolina crash report form greatly underestimates the true incidence of this type of crash. In fact, in addition to the Sussman study cited above, several other sources of data (e.g., Treat, et al, 1977; NASS CDS) suggest that the frequency of such crashes should be substantially greater than was actually found. It is important therefore to review possible reasons why such crashes may be under-reported (or over-reported).

Some of the possible reasons for under-reporting might include:

1. Drivers may attempt to hide their use of cellular telephones at the time that the crash occurred.

2. Drivers may not consider the use of their cellular telephones as relevant and therefore may not mention their use.

3. Investigating officers may not have asked about cellular telephone use or may not have considered cellular telephone use as relevant in crashes.

4. North Carolina does not have a major metropolitan area such as Los Angeles, Atlanta, or Chicago, and therefore the number of cellular telephones in the driving population may have been below average compared with other states. On the other hand, North Carolina does have several medium size metropolitan areas including Charlotte, Winston-Salem, Greensboro, Research Triangle Area, and Wilmington.

5. North Carolina has a large population with incomes that are below average. Those having such incomes may have been less likely to have been cellular telephone users. This might have caused the state to have fewer cellular telephones in use when compared with other states.

6. It is estimated that only about one in two crashes in North Carolina is reported to authorities. The remainder are handled privately and therefore would not appear in crash data bases.

7. Even though the market share for installed car phones has decreased, the absolute numbers of such phones being sold continue to increase each year. This is a consequence of the increased availability of such installations as new car options. Typically, these cellular telephones include hands-free features. It is possible that the incidence of crashes related to installed car phone use is not increasing due to the evolution in installed car phones toward hands free models as well as other improvements to design and installation (e.g., more convenient locations, larger, more readable displays).

8. The greatest increase in cellular telephone sales can be attributed to the introduction of the portable, hand-held (e.g., flip-phone) models. Because these phones can easily be displaced or concealed following a crash, it may be more difficult for police officers to detect the presence of cellular telephones and their possible use as an antecedent to crashes.

 

Of course there are also reasons why cellular telephone related crashes might be over-reported. They include:

1. Drivers might have attributed crashes to cellular telephone use when in fact there was another cause, such as speeding.

2. Investigating officers might have jumped to the conclusion that cellular telephones were the primary contributor, when some other factor was the primary contributor.

3. As investigating officers may have become more aware that cellular telephones could be causing crashes, they may have over-emphasized them as a cause in their reporting.

Figure 4-3 Adjusted Number of Crashes in North Carolina vs. Number of Cellular Telephones in Use in the U.S. (millions), by Year.

Crashes = 11.373+0.512* Number

Table 4-8 Linear Regression Summary Performed on the Adjusted Number of Crashes by Year.

x

Regression Summary

R = 0.91630320 R2 = 0.83961156 Adjusted R2 = 0.78614875

F(1, 3) = 15.705 p< 0.02870 Std. Error of estimate: 3.0200

x N=5

Fitted Parameter

Estimate B

St. Err. of B

t(3)

p-level

Intercpt

11.37265

2.652577

4.287398

0.023322

Slope

0.51161

0.129100

3.962902

0.028699

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives

4.6 Conclusions

This study has been successful in demonstrating the usefulness of the keyword-narrative search approach as a means of studying crash frequency and type related to cellular telephone use. The results demonstrate that there is an increasing trend in these crashes, and the results also provide information on the types of specific activities that are causing the crashes.

The study did find that the number of reported cellular telephone related crashes is relatively small, considering what might be expected based on anecdotal reporting. In an earlier study using this same database, Wierwille and Tijerina (1995) also found a relatively low number of reported crashes as being inattention/distraction related (as derived from police crash reports).

These findings are in sharp contrast with what would be expected on the basis of detailed crash investigations (1.5 percent vs. 30-50 percent).

The number of reported cellular telephone related crashes is relatively small, considering what might be expected based on anecdotal reporting.

Whether the reported number of crashes is in fact small or is a result of under-reporting remains to be determined.

In 1995 NHTSA's National Automotive Sampling System - Crashworthiness Data System (NASS CDS) began collecting data on precrash inattention/distraction related factors. The 1995 findings indicate that inattention/distraction related crashes account for about 26 percent of tow-away crashes with 0.1 percent of all CDS tow-away crashes attributable to cellular telephones (Wang, Goodman and Knipling, 1996). Although the actual number of relevant crashes for this period is relatively small, the findings are consistent with other data and suggest that under-reporting is the likely explanation for the low numbers in the North Carolina data.

In summary, the findings indicate relatively few cellular telephone related crashes in North Carolina during the period from 1989 through 1995. It has been argued that these data may substantially underestimate the true incidence of these crashes, based on other research that suggests that attention related crashes should occur more frequently than found in the North Carolina data. This suggests the need for improved reporting techniques to better identify and categorize these crashes.

The data indicate the wide range of crash causal factors associated with cellular telephone use.

In addition, the findings suggest an increase in cellular telephone related crash frequency as more cellular telephones become available. Furthermore, as the functionality of cellular telephones is expanded to include more "demanding" tasks (e.g., access to the internet, email, faxing, etc.), there is concern that there will be an associated increase in risk where these services are accessed from a moving vehicle. Finally, the data indicate the wide range of crash causal factors associated with cellular telephone use. While this information may be useful to designers of cellular telephone systems, it also highlights the relative importance of conversation itself as an important causal factor.

Chapter 4 - Table of Contents

Document Table of Contents

An Investigation of the Safety Implications of Wireless Communications in Vehicles Chapter 4: Analysis of Police Crash Report Narratives