FRA Dispatchers Report

U.S. Department of
Transportation
Federal Railroad
Administration

Work Schedules and Sleep Patterns of
Railroad Dispatchers

Office of Research
and Development
Washington, DC 20590

DOT/FRA/ORD-07/11

Final Report
April 2007

This document is available to the
U.S. public through the National
Technical Information Service,
Springfield, VA 22161.

NOTICE
This document is disseminated under the sponsorship of the
Department of Transportation in the interest of information
exchange. The United States Government assumes no liability for
its contents or use thereof.

NOTICE
The United States Government does not endorse products or
manufacturers. Trade or manufacturers’ names appear herein
solely because they are considered essential to the objective of this
report.

REPORT DOCUMENTATION PAGE

Form Approved
OMB No. 0704-0188

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources,
gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this
collection of information, including suggestions for reducing this burden, to Washington Headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson
Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188), Washington, DC 20503.
1. AGENCY USE ONLY (Leave blank)

2. REPORT DATE

3. REPORT TYPE AND DATES COVERED

April 2007

Final Report 7/2005-12/2006

4. TITLE AND SUBTITLE

5. FUNDING NUMBERS

Work Schedules and Sleep Patterns of Railroad Dispatchers
6. AUTHOR(S)

Judith Gertler and Alex Viale
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)

8. PERFORMING ORGANIZATION
REPORT NUMBER

Foster-Miller, Inc.
350 Second Avenue
Waltham, MA 02451-1196

DFRA.010350

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)

10. SPONSORING/MONITORING
AGENCY REPORT NUMBER

U.S. Department of Transportation
Federal Railroad Administration
Office of Research and Development
Washington, DC 20590

DOT/FRA/ORD-07/11

11. SUPPLEMENTARY NOTES

COTR: Dr. Thomas Raslear
12a. DISTRIBUTION/AVAILABILITY STATEMENT

12b. DISTRIBUTION CODE

This document is available to the public through the National Technical Information Service,
Springfield, VA 22161 and at www.fra.dot.gov.
13. ABSTRACT (Maximum 200 words)

This report presents the results of a study designed to characterize the work/rest schedules and sleep patterns of U.S. railroad
dispatchers and to examine the relationship between these schedules and levels of alertness of the individuals working the
schedules. The study methodology was a survey of a random sample of currently working U.S. railroad dispatchers who
completed a background survey and kept a daily log for 2 weeks. Railroad dispatchers are a predominantly healthy middle-aged
male population, but 14 percent are women. Dispatchers work as either a trick dispatcher, subject to the limitations of the Hours
of Service Law, or an assistant chief dispatcher who oversees the trick dispatchers. All dispatching jobs have a 40-hour nominal
workweek, but assistant chief dispatchers average more than 40 hours per week. Dispatchers are a shiftwork population. Many
are subject to working nights and a variable work schedule, making it difficult to get adequate quality sleep. Overall, 39 percent
of dispatchers get 6 or fewer hours of sleep while 29 percent of U.S. adults get this amount of sleep. Across all three shifts,
dispatcher alertness on workdays peaked after arrival at work and then declined through the workday. The decline was greatest
for those working third shift.
14. SUBJECT TERMS

15. NUMBER OF PAGES

Railroad dispatcher, fatigue, alertness, shiftwork, extra board, relief job, sleep disorder, assistant
chief dispatcher, trick dispatcher
17. SECURITY CLASSIFICATION
OF REPORT

18. SECURITY CLASSIFICATION
OF THIS PAGE

19. SECURITY CLASSIFICATION
OF ABSTRACT

Unclassified

Unclassified

Unclassified

NSN 7540-01-280-5500

103
16. PRICE CODE

20. LIMITATION OF ABSTRACT

None
Standard Form 298 (Rev. 2-89)
Prescribed by ANSI Std. 239-18
298-102

i

METRIC/ENGLISH CONVERSION FACTORS
ENGLISH TO METRIC
LENGTH

METRIC TO ENGLISH
LENGTH (APPROXIMATE)

(APPROXIMATE)

1 inch (in)

=

2.5 centimeters (cm)

1 millimeter (mm) = 0.04 inch (in)
1 centimeter (cm) = 0.4 inch (in)

1 foot (ft)

=

30 centimeters (cm)

1 yard (yd)

=

0.9 meter (m)

1 mile (mi)

=

1.6 kilometers (km)

1 meter (m) = 3.3 feet (ft)
1 meter (m) = 1.1 yards (yd)
1 kilometer (km) = 0.6 mile (mi)

AREA (APPROXIMATE)
2

1 square inch (sq in, in )

=

2

1 square foot (sq ft, ft )

=

AREA (APPROXIMATE)
2

6.5 square centimeters (cm )

2

2

0.09 square meter (m )

2

1 square meter (m ) = 1.2 square yards (sq yd, yd )

2

=

0.8 square meter (m )

1 square mile (sq mi, mi )

2

=

2.6 square kilometers (km )

1 acre = 0.4 hectare (he)

=

4,000 square meters (m )

1 square yard (sq yd, yd )

2

1 square centimeter (cm ) = 0.16 square inch (sq in, in )

2

2

2

2

1 square kilometer (km ) = 0.4 square mile (sq mi, mi )
2

2

10,000 square meters (m ) = 1 hectare (ha) = 2.5 acres

2

MASS - WEIGHT (APPROXIMATE)
1 ounce (oz)

=

28 grams (gm)

1 pound (lb)

=

0.45 kilogram (kg)

1 short ton = 2,000 pounds
(lb)

=

0.9 tonne (t)

MASS - WEIGHT (APPROXIMATE)
1 gram (gm) = 0.036 ounce (oz)
1 kilogram (kg) = 2.2 pounds (lb)
1 tonne (t) = 1,000 kilograms (kg)
= 1.1 short tons

VOLUME (APPROXIMATE)

VOLUME (APPROXIMATE)

1 teaspoon (tsp)

=

5 milliliters (ml)

1 tablespoon (tbsp)

=

15 milliliters (ml)

1 milliliter (ml) = 0.03 fluid ounce (fl oz)
1 liter (l) = 2.1 pints (pt)

1 fluid ounce (fl oz)

=

30 milliliters (ml)

1 liter (l) = 1.06 quarts (qt)

1 cup (c)

=

0.24 liter (l)

1 liter (l) = 0.26 gallon (gal)

1 pint (pt)

=

0.47 liter (l)

1 quart (qt)

=

0.96 liter (l)

1 gallon (gal)
3

1 cubic foot (cu ft, ft )
3

1 cubic yard (cu yd, yd )

=

3.8 liters (l)

=

0.03 cubic meter (m )

=

3

3

3

1 cubic meter (m ) = 36 cubic feet (cu ft, ft )

3

3

0.76 cubic meter (m )

3

1 cubic meter (m ) = 1.3 cubic yards (cu yd, yd )

TEMPERATURE (EXACT)

TEMPERATURE (EXACT)

[(x-32)(5/9)] °F = y °C

[(9/5) y + 32] °C = x °F

QUICK INCH - CENTIMETER LENGTH CONVERSION
1

0
Inches
Centimeters

0

1

2

2

3

4

3

5

6

7

4

8

9

5

10

11

12

13

QUICK FAHRENHEIT - CELSIUS TEMPERATURE CONVERSION
°F -40° -22°

-4°

14°

32°

50°

68°

86°

104°

122°

°C -40° -30°

-20°

-10°

0°

10°

20°

30°

40°

50°

140° 158° 176° 194° 212°
60°

70°

80°

90° 100°

For more exact and or other conversion factors, see NIST Miscellaneous Publication 286, Units of Weights and
Updated 6/17/98
Measures. Price $2.50 SD Catalog No. C13 10286

ii

Contents
Illustrations ..................................................................................................................................... v
Tables

.................................................................................................................................... vi

Acknowledgements........................................................................................................................ ix
Executive Summary ........................................................................................................................ 1
1.
1.1
1.2
1.3
1.4
1.5

Introduction ................................................................................................................. 7
Nature of the Dispatcher’s Job .................................................................................... 7
Objectives.................................................................................................................. 10
Overall Approach ...................................................................................................... 10
Scope ......................................................................................................................... 11
Organization of the Report ........................................................................................ 11

2.1
2.2
2.3
2.4
2.5

Survey Design ........................................................................................................... 13
Potential Respondent Universe ................................................................................. 13
Survey Instruments.................................................................................................... 14
Data Collection Period .............................................................................................. 14
Sampling Plan............................................................................................................ 14
Procedure................................................................................................................... 15

3.1
3.2
3.3
3.4
3.5
3.6
3.7

Analysis of Survey Data............................................................................................ 19
Survey Response Rate ............................................................................................... 19
Non-Response Bias Study ......................................................................................... 19
Dispatcher Demographic Characteristics .................................................................. 20
Job Characteristics..................................................................................................... 26
Sleep Characteristics ................................................................................................. 36
Alertness.................................................................................................................... 49
Textual Data .............................................................................................................. 52

4.1
4.2
4.3

Findings and Recommendations................................................................................ 61
Key Study Findings ................................................................................................... 61
Recommendations for Improvements in Study Procedures ...................................... 63
Recommendations for Additional Research.............................................................. 64

2.

3.

4.

5.

References ................................................................................................................. 65

Appendix A. Survey Materials .................................................................................................... 67
Appendix B. Supporting Data...................................................................................................... 85
Appendix C. Adjustments to the Data ......................................................................................... 89
Abbreviations and Acronyms ....................................................................................................... 91

iii

Illustrations
Figure 1. Modern dispatching center ............................................................................................. 9
Figure 2. Overall approach .......................................................................................................... 11
Figure 3. Distribution of dispatchers by age group and sex ........................................................ 21
Figure 4. Self-assessment of overall health ................................................................................. 24
Figure 5. Workdays lost due to illness in past year ..................................................................... 24
Figure 6. Relief schedules............................................................................................................ 27
Figure 7. Number of breaks per day by job type ......................................................................... 31
Figure 8. Length of day’s longest break by job type ................................................................... 32
Figure 9. Commute time and workday by job type ..................................................................... 33
Figure 10. Start of commute to work ........................................................................................... 33
Figure 11. Sources and levels of stress by job type ..................................................................... 34
Figure 12. Nighttime sleep on workdays for permanent first and second shift dispatchers versus
U.S. adults................................................................................................................... 41
Figure 13. Sleep in 24 h for dispatchers versus U.S. adults......................................................... 42
Figure 14. Nap start times for workdays...................................................................................... 48
Figure 15. Nap start times for non-workdays .............................................................................. 48

v

Tables
Table 1. Breakdown of survey responses ...................................................................................... 2
Table 2. Estimate of number of actively working U.S. dispatchers by labor organization ......... 13
Table 3. Dispatcher age by sex (yr) ............................................................................................. 21
Table 4. Discrepancies between chronological and perceived age by age group (percent) ........ 22
Table 5. Years of experience as a dispatcher by job type (yr)..................................................... 22
Table 6. Years of experience as a dispatcher with current employer by job type (yr) ................ 22
Table 7. Years of experience as a dispatcher by sex (yr)............................................................. 23
Table 8. Work schedule by type of job and overall (percent)...................................................... 26
Table 9. Work schedule by age and experience (yr).................................................................... 27
Table 10. Actual shifts worked by job schedule (percent)........................................................... 28
Table 11. Shift variability for relief and extra board work schedules (percent) .......................... 29
Table 12. Typical and actual work for 2-week period by job type (h:min) ................................. 29
Table 13. Number and length of longest break per day by job type............................................ 30
Table 14. Breaks by shift worked ................................................................................................ 31
Table 15. Stress ratings by job type ............................................................................................. 36
Table 16. Total sleep by type of day and work schedule (h:min)................................................ 38
Table 17. Total sleep by work schedule versus shiftwork norms for workdays (h:min)............. 39
Table 18. Total and primary sleep on workdays by shift worked versus shiftwork norms (h:min)
..................................................................................................................................... 39
Table 19. Primary sleep by shift worked (h:min) ........................................................................ 40
Table 20. Ratings of primary sleep (mean of means) by work schedule and type of day ........... 43
Table 21. Workday primary sleep period ratings by shift worked .............................................. 44
Table 22. Sleep ratings and duration by sleep disorder status (workdays only).......................... 45
Table 23. Sleep latency by type of day and job type (min) ......................................................... 45
Table 24. Sleep latency by shift worked (min) ............................................................................ 45
Table 25. Number of naps in 2-week period by work schedule .................................................. 46
Table 26. Mean nap length by work schedule and type of day (h:min)....................................... 47
Table 27. Nap length on workdays by shift worked (h:min) ....................................................... 47
Table 28. Alertness at work by job type (percent)....................................................................... 49
Table 29. Mentally drained after work by job type (percent) ...................................................... 50
Table 30. Alertness at work by work schedule (percent)............................................................. 50
vi

Table 31. Mentally drained after work by work schedule (percent)............................................ 51
Table 32. Alertness throughout the day by shift worked ............................................................. 51
Table 33. Linear regression results by shift for alertness and time on shift ................................ 51
Table 34. Alertness and sleep disorders (workdays only) ........................................................... 52
Table 35. Keywords used for each topic area search................................................................... 54
Table 36. Frequency of comments by topic area and source....................................................... 55

vii

Acknowledgements
This report presents the results of a research study designed to characterize the work schedules
and sleep patterns of U.S. railroad dispatchers and to examine the relationship between these
schedules and levels of alertness for the individuals working the schedules. Foster-Miller, Inc.
conducted the work for the Federal Railroad Administration (FRA) under contract DTFR53-01D-00029 with guidance from Dr. Thomas Raslear, Office of Research and Development, Human
Factors Program. The authors worked closely with the American Train Dispatchers Association
(ATDA) in designing and conducting the survey. Special thanks are due to Mr. Leo McCann
and Mr. Gary Melton, both of ATDA, who provided invaluable insight on the job of a railroad
dispatcher and facilitated distribution of the survey materials to their members. Their
participation was a key element in the successful conduct of the study. The authors are
especially appreciative of the efforts of Mr. Robert Brogan, FRA Office of Safety, who
facilitated the Office of Management and Budget approval process for this study.
The authors also wish to thank several individuals at Foster-Miller for their significant
contributions to the study. Ms. Susan McDonough managed the database for tracking and
compensating the survey participants. Ms. Audrey Murray provided invaluable assistance with
the coding and processing of the data in the background surveys and daily logs. Ms. Gayle
Staffiere from Foster-Miller’s Publications Department was responsible for layout of the log
books, and Ms. Eileen Ballard-Sarcone oversaw the printing and mailing of the survey
documents. Ms. Jeanne Seaquist designed the database for the survey data.

ix

Executive Summary
In a continuing effort to improve railroad safety and to reduce the number of injuries and
fatalities to railroad workers, the Federal Railroad Administration (FRA) and the railroad
industry, through the North American Rail Alertness Partnership (NARAP), have focused on the
issue of fatigue among train and engine crew personnel. Because railroading is an around-theclock, 7-days-a-week operation, and because a wide array of workers are needed both to operate
and maintain the Nation’s railroads, other crafts besides train and engine crews can also be
subject to fatigue. The non-operating crafts, which include right-of-way construction and
maintenance, signal system construction and maintenance, and dispatchers, fall into this
category. With all of the non-operating craft groups, staff shortages, expanding territories, call
schedules, and working conditions may contribute to employee fatigue. In 2001, FRA decided to
begin exploring non-operating craft workers’ fatigue by initially focusing on signalmen. A
subsequent study addressed maintenance of way (MOW) workers. This study, similar in
methodology to the two earlier studies, investigated the fatigue issues of railroad dispatchers.
This study had two primary objectives:
•

To design and conduct a survey to collect work schedules and sleep data from railroad
dispatchers.

•

To analyze data to characterize work/sleep patterns and to identify work schedule-related
fatigue issues.

The goal was to characterize U.S. railroad dispatchers as a group, not to characterize dispatchers
on a specific railroad.
The railroad dispatcher is responsible for the safe, efficient, and economical movement of trains
and other railway vehicles over the railroad, as well as for the protection of those who work on
the railroad. Dispatcher jobs fall into two categories, trick dispatcher and assistant chief
dispatcher. Trick dispatchers manage and control access to a specific territory or track. The
assistant chief dispatcher oversees several trick dispatchers and assists them as necessary.
Because an assistant chief dispatcher does not directly control train movements, this position is
not subject to the limitations of the Hours of Service Law.
The work schedules for both types of dispatcher jobs are similar, with shifts that cover 24 hours
(h) of each day, 7 days (d) a week. Three types of work schedules exist for dispatchers: regular
daily shift (first or day, second or evening, third or night), relief, and extra board. Regular jobs
work 5 consecutive d on the same shift followed by 2 consecutive d off. Relief jobs work 5
consecutive d by rotating through the shifts in a pattern such as 2 days, 2 evenings, and 1 night.
While the regular and relief jobs work the same days each week, the extra board jobs do not have
a fixed schedule. The extra board dispatchers fill in for regular and relief dispatchers as
necessary.
The research described in this report had three phases: preparation, field data collection, and
data analysis. Since no existing data source would provide answers to the study’s research
questions, a survey of railroad dispatchers was the only means to obtain the necessary data. The
preparation phase included securing approval from the Office of Management and Budget

1

(OMB) for the survey. Representatives from the American Train Dispatchers Association
(ATDA) worked closely with the researchers throughout the study.
Survey Design
The study used two survey instruments–a background survey and a daily log. Survey
participants used the background survey to provide demographic information, descriptive data
for their job type and work schedule, and a self-assessment of overall health. The daily log
provided a place for recording work periods and sleep on workdays and non-work or rest days.
Dispatchers recorded not only the starting and ending times for each sleep and work period but
also a subjective assessment of their alertness at different times during their workday. Capturing
the week-to-week variability in the work schedules of extra board dispatchers required a 2-week
data collection period.
Researchers drew a random sample of 963 dispatchers from the ATDA database of actively
working U.S. dispatchers. Retirees, full-time union officials, and anyone currently holding a
railroad management position were specifically excluded from the sampling frame.
Determination of the sample size assumed a 95 percent confidence interval on the estimates for
mean sleep time, an error tolerance of 20 percent, and a 40 percent response rate. The method
for sizing the sample assured that the sample could include an adequate number of extra board
dispatchers.
Mailing of the survey materials occurred on April 18, 2006. One month (mo) later, every
dispatcher in the random sample received a reminder postcard encouraging him/her to participate
and to call the researchers if he/she needed additional materials.
Survey Response Rate
The overall response rate for the survey was 46 percent. Table 1 provides a breakdown of the
survey responses.
Table 1. Breakdown of survey responses
Number

Percent

445

46.2

5

.5

20

2.1

No response

493

51.2

Total number of surveys mailed

963

Returned both background survey and daily log
Returned only one survey instrument
Materials undeliverable due to invalid address

Of the 445 complete responses, 2 could not be a part of the analysis because those individuals
were currently working as a chief dispatcher, which is a management job.
The non-response bias study based on age found no difference between survey respondents and
non-respondents.

2

Dispatcher Demographics
The majority of survey respondents (87 percent) held trick dispatcher jobs. Ten percent held an
assistant chief position, and the remaining 3 percent reported their job type as other. Fourteen
percent of the survey respondents were female. Average dispatcher experience was 14 years
(yr), but the average for women was 11 yr while that of their male counterparts was 15 yr.
Nearly half of all dispatchers are 50 yr and older, with male dispatchers being, on average, 3 yr
older than the female dispatchers.
The majority of dispatchers (76 percent) are married, but few have children under the age of 2 yr.
In contrast, 55 percent of U.S. adults age 18 and older are married. Since many railroaders
report that their work schedule strains marital relationships, finding such a high proportion of
married dispatchers was surprising. The survey did not, however, ask how many times the
dispatcher had been married. The lack of young children is consistent with the average age of
this population.
Nearly 85 percent of dispatchers rated themselves in good or excellent health. This selfassessment did vary by work schedule with third shift dispatchers having a lower overall health
self-rating than extra board dispatchers. While the majority of dispatchers rated their health as
good or excellent, this group averaged 5.6 workdays lost due to illness annually. In comparison,
U.S. employed adults with paid sick time average 3.6 d. The higher rate of use of sick time may
be due to the stressful nature of the job or, for those working an irregular schedule, the need to
catch up on sleep.
Approximately 9 percent of the dispatchers reported having a diagnosed sleep disorder, and
7 percent reported having sleep apnea. One-third of these dispatchers have gone without
treatment. The dispatcher rate for sleep apnea is higher than the estimated norm of 4 percent for
U.S. middle-aged working men. Two possible explanations exist for the higher rate among
dispatchers: the sedentary nature of their job which may lead to weight gain which exacerbates
sleep apnea, and an awareness of the symptoms of sleep apnea which led them to seek diagnosis
and treatment.
Job Characteristics
The around-the-clock nature of railroading requires railroads to staff dispatching centers with
three 8-h shifts. Dispatchers working permanent first, second, or third shifts rarely work outside
their typical schedules. The most common relief rotation was to work two first shifts, followed
by two second shifts, and a third shift. Since extra board dispatchers fill in for dispatchers who
cannot work their regular shifts, the expectation was that this group of dispatchers would be split
evenly among the three shifts. This did not occur. Forty-two percent of the shifts worked by an
extra board dispatcher during the study were on first shift. First shift dispatchers are the most
experienced and therefore accrue the largest number of vacation and sick days. Due to the
seniority of first shift dispatchers, vacancies are most likely to occur on first shift and require
assignment to an extra board dispatcher. A third of the extra board dispatchers experienced no
shift variability during the study because they filled in for a regular dispatcher on vacation or
other extended leave.
Both trick dispatchers and assistant chiefs have a nominal 40-h workweek. While trick
dispatchers averaged close to 40 h of work during the study, assistant chiefs averaged
approximately 45 h of work per week. In addition, one-quarter of the assistant chiefs worked a
3

day per week of overtime. Actual work, as reported in the daily logs, was similar to reported
typical workweeks for both groups, suggesting that the time period for this study was a typical
one with respect to dispatcher work schedules.
Railroad dispatchers have no contractual provisions for breaks. Nearly a third of the time, both
groups of dispatchers took no breaks during their shift. Trick dispatchers averaged 2 breaks per
day with the longest break averaging 8 minutes (min). Assistant chiefs averaged 1½ breaks per
shift with the longest break averaging 10 min. Despite the lack of breaks, this was not a primary
source of stress for dispatchers.
Trick dispatchers and assistant chiefs had similar sources of stress. In general, trick dispatchers
had higher stress ratings than assistant chiefs, with ratings on sleep loss, lack of time off,
ambiguous rules, and lack of breaks being significantly higher.
Sleep Characteristics
The study examined primary and total daily sleep, which includes naps. Dispatchers working
permanent first shift jobs get considerably less total sleep on workdays than those with other
work schedules, but these dispatchers appear to make up for their loss of sleep on non-workdays.
In comparison with shiftwork norms, railroad dispatchers get less sleep than other shiftwork
populations. First shift for a railroad dispatcher may begin as early as 6 a.m. This early start
time appears to limit the nighttime sleep for those who work first shift. For total and primary
sleep, all shifts are statistically different from each other.
Because many dispatchers on a permanent third shift use a split sleep strategy on workdays, the
napping rate was highest for this group. The majority of workday naps for all dispatchers
combined began between 4 and 8 p.m. In contrast, naps on non-workdays tended to start
between 12 and 4 p.m., a time that coincides with the afternoon nadir in the circadian cycle.
Comparison of dispatcher sleep with normative data for U.S. adults indicates that, in terms of
average daily sleep in 24 h, railroad dispatchers are getting less sleep than U.S. adults. While
29 percent of U.S. adults get 6 or fewer h of sleep daily, 39 percent of dispatchers get this
amount of sleep.
Overall, dispatchers gave higher ratings to their non-workday sleep than their workday sleep.
Dispatchers’ qualitative ratings of their sleep differed based on the shift actually worked. In
particular, sleep ratings for dispatchers working second shift were statistically different for all
ratings except ease of arising. Correlations between each of the four sleep quality ratings and
length of the primary sleep period were all significant but did not show a strong relationship.
Dispatchers with diagnosed but untreated sleep disorders rated their sleep of lower quality than
those with treated sleep disorders or no sleep disorders, but these differences were not
statistically significant.
Alertness
Data from dispatchers’ daily logs revealed some differences in alertness levels based on the shift
worked. For all shifts, alertness peaked after the commute to work but then declined throughout
the day. The decline in alertness was greatest on third shift.
Dispatchers with diagnosed but untreated sleep disorders had lower alertness ratings throughout
the day than those with treated or no sleep disorders, but these differences were not statistically
significant.
4

Textual Analysis of Log Book Comments
A systematic qualitative analysis of the textual comments in the daily logs provided greater
insight into the concerns of dispatchers and, in many cases, added further insight to the
quantitative survey results. The most frequently mentioned topics were sleep, workload,
alertness/fatigue, and naps. Sleep was in three times as many comments as workload, which was
the second most frequently mentioned topic in the comments.
Findings and Recommendations
The following lists the key findings with respect to the dispatcher’s health, work periods, sleep
patterns, and alertness:
•

Dispatchers average 5.6 workdays lost annually due to illness. In contrast, U.S.
employed adults with paid sick time average 3.6 d per year. The need of this shiftwork
population to catch up on lost sleep and the stressful nature of dispatching may be
responsible for this difference.

•

The 8-h workday of a dispatcher plus commute time allow adequate time for sleep and
personal activities, but varying and unpredictable schedules for some dispatchers and the
need to sleep at times that run counter to human physiology may prevent the dispatcher
from getting adequate rest. For assistant chief dispatchers, the situation has the added
complication that no limitation exists on the number of hours that they may work in a
day, thus allowing 2 shifts in 1 d and backward shift rotation.

•

Although dispatchers do not have any planned or guaranteed breaks during their 8-h shift,
this was not a major source of stress to either trick dispatchers or assistant chiefs, perhaps
because they have become accustomed to their few short breaks and the need to eat meals
at their desk.

•

Overall, dispatchers are a sleep-deprived group. They sleep less than other shiftwork
populations and less than the norm for U.S. adults. Research has shown that sleep
deprivation leads to performance degradation and may also have health consequences.
Many dispatchers, who perform a safety-critical function on the railroad, may be unaware
of their degraded performance.

•

The incidence of sleep apnea among U.S. dispatchers exceeds the U.S. adult norm. Only
two-thirds of those with sleep apnea or another sleep disorder reported being treated. To
encourage these individuals to accept treatment, railroads and unions should continue
their sleep education programs, pointing out the possible performance and health
consequences of untreated sleep disorders. Initial dispatcher training and periodic rules
training provide the opportunity for fatigue education that includes discussion of sleep
disorders, as well as strategies for coping with shiftwork.

Based on the experience of this study, several methodological changes should be a part of any
future studies of this nature. Two of the four recommendations below focus on issues unique to
a shiftwork population. The following lists the recommended changes:
•

Design and test instructions for third shift workers to use in recording their primary sleep
period. Despite instructions specifically directed at third shift dispatchers, some third
shift dispatchers failed to record their primary sleep on the correct day. Improved

5

instructions, inclusion of a hypothetical example in a pilot test, and participation of more
third shift dispatchers in the pilot test for a future study should correct this problem.
•

Separate the question on job type (e.g., trick dispatcher, assistant chief, other) from the
one covering work schedule. A poorly worded question on the background survey
covered both job type and work schedule. Despite this, researchers were able to
effectively categorize each survey respondent’s job type and work schedule. Two
questions, one for each characteristic, will avoid confusion and the need to manually code
these two job characteristics.

•

Include a definition of a work break and guidance on recording break information. The
lack of a definition for a work break led to some ambiguity among survey participants as
to how to respond to this entry in the daily log book.

•

Include a question on the background survey that asks, “How many times have you been
married?” Both shiftwork and railroad careers can lead to marital and family difficulties.
For this reason, an additional question on the number of marriages would be helpful.

The researchers have developed an effective procedure and methodology for characterizing the
work schedules and sleep patterns of railroad workers. Additional studies of other populations
using this methodology would provide a more complete picture of work schedule and fatigue
issues in the industry. Yardmasters, locomotive engineers, and conductors are candidate groups
for study.
A number of biomathematical models exist for predicting human fatigue and alertness. Model
developers could use the data from the dispatcher survey, as well as the data from the two earlier
studies, to further refine their models and predict how the typical railroad worker schedule may
be affecting on-the-job alertness.
Further analysis of the dispatcher survey data is possible. Additional analyses might examine the
relative likelihood (i.e., odds ratio) of getting less than a minimum amount of sleep as a function
of shift worked or work period start time. A similar analysis could estimate the likelihood of a
workday nap as a function of shift worked. Comparisons across the different railroad craft
groups are possible to determine if differences in work schedules and sleep patterns occur when
a group is subject to the work limitations of the Hours of Service Law.

6

1. Introduction
In a continuing effort to improve railroad safety and to reduce the number of injuries and
fatalities to railroad workers, FRA and the railroad industry, through NARAP, have focused on
the issue of fatigue among train and engine crew personnel. Because railroading is a round-theclock, 7-days-a-week operation, and because a wide array of workers are needed to operate and
maintain the Nation’s railroads, other crafts besides train and engine crews can also be subject to
fatigue. The non-operating crafts, which include right-of-way construction and maintenance,
signal system construction and maintenance, and dispatchers, fall into this category. With all of
the non-operating craft groups, staff shortages, expanding territories, call schedules, and working
conditions may contribute to employee fatigue.
In 2001, FRA decided to begin exploring fatigue of the non-operating craft workers by initially
focusing on signalmen. A separate report presents the results of the signalmen study (Gertler &
Viale, 2006a). Subsequent to the signalmen study, FRA sponsored a similar study of MOW
workers (Gertler & Viale, 2006b). The study described in this report, which is similar in scope
and methodology to the two earlier studies, concerns railroad dispatchers. Previous studies of
dispatcher fatigue (e.g., Popkin, Gertler, & Reinach, 2001) used convenience samples of
dispatchers rather than a random or stratified sample based on estimates of accuracy and target
error rate, as was the case for the surveys of signalmen and MOW workers. Unlike the earlier
dispatcher work, this study provides defensible and definitive data on work/rest cycle parameters
and fatigue among dispatchers.
1.1

Nature of the Dispatcher’s Job

The railroad dispatcher is responsible for the safe, efficient, and economical movement of trains
and other railway vehicles over the railroad, as well as for the protection of those who work on
the railroad. The dispatcher’s principle duties include:
•

Monitoring radio and telephone and communicating with other railroad personnel
regarding train and track information.

•

Scheduling the routing and movement of trains to provide for safe meets and passes.

•

Arranging for track use by engineering forces for maintenance activities.

•

Managing unplanned events to protect the safety of the public, railroad employees, and
railroad property.

•

Maintaining records of train movements, track assignments, maintenance activities, and
other events.

The job requires the dispatcher to issue, monitor, and cancel track usage authorizations in
accordance with the railroad’s operating rules and procedures. The dispatcher also operates
signals, switches, and bridges; communicates with train and MOW crews; responds to
emergency events; and performs administrative and clerical duties. The dispatcher spends a
significant portion of his/her day communicating on the radio or telephone.
Every dispatcher is responsible for a predefined territory or portion of the railroad’s network.
One of the dispatcher’s first tasks at the beginning of the shift is to plan the known track moves
7

on the territory for the duration of the shift. This strategy takes into account current traffic,
expected traffic, the physical characteristics of the territory, train priorities (e.g., passenger
versus freight), track and signal maintenance requirements, characteristics of train performance,
and the presence of hazardous materials. Information from MOW crews, yard personnel, and
dispatchers on adjoining territories also contributes to the dispatcher’s action.
Once this plan is in place, dispatchers spend the remainder of the shift trying to keep the trains
moving while adjusting for the inevitable occurrence of delays and unforeseen events. Even
short delays may necessitate the reformulation of the entire plan. Indeed, the dispatcher may
have to reformulate the plan multiple times during the course of the shift as unplanned events
transpire. The dispatcher is also responsible for monitoring train crew hours to ensure
compliance with the Hours of Service Law. Finally, the dispatcher must also assume
responsibility for railroad and non-railroad problems that are phoned in by railroad personnel, as
well as the general public, and must be knowledgeable about proper procedure for notifying the
appropriate authorities in the event of an emergency, such as a hazardous materials spill.
With few exceptions, today’s dispatchers work with computer-based dispatching and
communications technology. Dispatchers for the larger Class I railroads work in shifts around
the clock in large centralized operations. Upwards of 100 dispatchers may work the same shift
in a large centralized operations center. Some may control territories that are located over 1000
miles away. As the cost of computer systems decreases, even the smaller railroads are
abandoning paper forms and radio directives in favor of the computer-based dispatching
technology.
To meet the need for 24-h operation, railroads staff their dispatching center with three 8-h shifts.
Typical shifts are 7 a.m. to 3 p.m. (day), 3 p.m. to 11 p.m. (evening), and 11 p.m. to 7 a.m.
(night). Three categories of jobs exist in all dispatching centers: regular jobs, relief jobs, and
extra board jobs. Regular jobs work 5 consecutive d on the same shift followed by 2 consecutive
d off. Relief jobs work 5 consecutive d by rotating through the shifts, in a pattern such as 2 days,
2 evenings, and 1 night. Occasionally, a relief job will work the same shift each day but will not
be responsible for the same territory each day. While the regular and relief jobs work the same
days each week, the extra board jobs do not have a fixed schedule. The extra board dispatchers
fill in for regular and relief dispatchers during vacations, training, and road days, as well as when
an unplanned absence occurs. On occasion, a regular dispatcher on a rest day may fill a vacancy
if an extra board dispatcher is not available. Most dispatching centers have a guaranteed extra
board. This means that the extra board dispatchers are guaranteed 5 d of work per week, but the
days and shifts that they work may change weekly. In addition, extra board dispatchers usually
do not have 2 consecutive rest days.
All dispatching centers have a chief dispatcher who oversees the entire dispatching operation. In
larger centers, assistant chief dispatchers supervise groups of trick dispatchers. (A trick
dispatcher works as described above. The term dispatcher may refer to an assistant chief or a
trick dispatcher.) The chief and/or the assistant chief provide backup support to the trick
dispatchers as required. Like the majority of trick dispatchers, the chief and assistant chief
positions have an assigned shift. Some assistant chiefs work a relief schedule that involves a
rotating shift pattern. No extra board exists specifically for assistant chief positions. When a
vacancy occurs in an assistant chief position, a trick dispatcher or other assistant chief who is
qualified to work the open position will fill in.

8

As is the case with train and engine crews and signalmen, the Hours of Service Law limits the
length of the dispatcher’s workday. The Hours of Service Law stipulates that the dispatcher may
not remain on duty for more than 9 h, whether consecutive or in the aggregate, in any 24-h
period, in operations that employ 2 or more shifts. This means that once the dispatcher has
worked for 9 h, he/she must have 15 h of rest. Where only 1 shift is employed, the dispatcher
may remain on duty up to 12 h in any 24-h period. During an emergency situation, the law
allows the dispatcher to remain on duty for an additional 4 h in any 24-h period for a maximum
of 3 d over the course of 7 d. This law limits the length of the dispatcher’s shift and provides for
guaranteed time off, but it does not address the number of consecutive days that the dispatcher
may work. Because chiefs and assistant chiefs are typically not directly responsible for
overseeing train movements, the Hours of Service Law does not apply to these positions. As a
result, individuals in these positions may sometimes work for 12 or 16 h to cover a vacancy.
The majority of U.S. dispatchers earn vacation days, sick leave, and personal days, all in
proportion to their years of service with their respective railroad. This is true for dispatcher jobs
covered by a labor agreement, as well as those considered a management position and thus
covered by company policies.
The nature of the dispatcher’s job has remained the same for at least the last 30 yr, but changes in
the technology of dispatching and restructuring of the industry have resulted in a significantly
different work environment. Today’s dispatcher is likely to work in an office environment with
suitable lighting and temperature controls. The dispatcher’s work area consists of several
computer screens, a keyboard, and a mouse. An overhead schematic display of multiple
territories may also be present, as Figure 1 illustrates.

Figure 1. Modern dispatching center
The introduction of computer-aided dispatching has made it possible to control larger territories
from much farther away. Changes in signal technology have reduced the need for tower
operators and other field operations personnel, resulting in more direct dispatcher control over
train movements, an increase in responsibilities, and an increase in the number of individual
tasks involved in carrying out the same responsibilities. While some believe that these changes
in technology have made the dispatcher’s job easier, others argue that they have increased the
dispatcher’s workload and associated job stress and fatigue. Regardless, the dispatcher performs
a safety critical job with many responsibilities. Alertness is key to his/her ability to carry out
9

those responsibilities safely and effectively. Staff shortages, causing dispatchers to work on rest
days, can compromise the dispatcher’s alertness.
1.2

Objectives

This study had two primary objectives:
•

To design and conduct a survey to collect work schedules and sleep data from railroad
dispatchers.

•

To analyze data to characterize work/sleep patterns and to identify work schedule-related
fatigue issues.

The goal was to characterize U.S. railroad dispatchers as a group, not to characterize dispatchers
on a specific railroad.
Specific research issues that the study sought to answer included the following:

1.3

•

What is the distribution of dispatchers among different work schedules?

•

What is average number of hours worked per day? Per week?

•

How does average number of hours worked vary based on type of job?

•

How frequently are dispatchers able to take a break?

•

What are the levels of alertness of dispatchers as a function of work schedule?

•

To what extent is sleep quality a function of age?

•

What is the average number of hours of sleep on workdays? Non-workdays? How does
this compare with U.S. adult norms and other shiftwork populations?

•

Does quality and amount of sleep vary by shift or work schedule?

•

What is the relationship between number of breaks and end of day fatigue?

•

What is the average number of hours that the dispatcher spends commuting to and from
work? Is commute time related to level of alertness?
Overall Approach

Since no existing data sources could provide answers to the above issues, a survey of dispatchers
was the only means to obtain the necessary work schedule and sleep data. The research project
consisted of three phases: preparation, field data collection, and data analysis (see Figure 2).
The preparation phase involved designing the survey methodology and procedures, conducting a
pilot survey to refine the survey instruments and data collection procedure, securing approval
from OMB, and preparing the final survey instruments. (Because this survey involved more than
nine participants, Federal regulations required that OMB approve the overall study design.)
Activities during this phase included discussions with ATDA to assure that the survey
instruments had suitable wording and would collect the data necessary to address the research
issues. A pilot survey, conducted in parallel with the OMB review process, assured that the
survey would capture the data needed to meet the survey objectives.

10

Figure 2. Overall approach
The second phase of the research consisted of distributing the survey materials and collecting the
survey data. Analysis of the survey data was the final phase. A non-response bias study
validated that no difference existed between the survey participants and the non-respondents.
The data analysis methods for the survey data included descriptive statistics, analysis of variance
(ANOVA), correlation analysis, and textual analysis of the log book comments.
1.4

Scope

This research involved railroad dispatchers working in the United States. The study
characterized these workers as a group. It did not attempt to characterize dispatchers working for
specific railroads. Making recommendations regarding fatigue countermeasures was beyond the
scope of the study.
1.5

Organization of the Report

Section 2 describes the overall survey design and procedures. Section 3 provides analysis of the
survey results, and Section 4 contains the findings and recommendations. Appendix A contains
copies of the survey materials. Appendix B contains detailed summaries that support the
statistical analyses of Section 3, and Appendix C describes adjustments to the data that were part
of the analysis process. A list of abbreviations and acronyms used in the report follows the
appendices.
11

2. Survey Design
One of the objectives of this study was to characterize the work schedules and sleep patterns of
U.S. railroad dispatchers. Achieving this objective required a nationwide survey. The only
practical means of reaching these individuals was through either their union or, in the case of
non-union dispatchers, their employing railroad. This section describes the potential respondent
universe, the survey instruments, sampling plan, and procedures that the researchers developed
to survey this population. This methodology is similar to that used in the earlier studies of
railroad signalmen and MOW workers.
2.1

Potential Respondent Universe

The potential respondent universe is the 3000 actively working railroad dispatchers in the United
States. The majority of the dispatchers, two-thirds, are members of ATDA. Nearly 30 percent
are non-union, and the Transportation Communications International Union (TCU) and the
Association of Commuter Rail Employees (ACRE) represent the remainder. Table 2 shows how
the potential respondent universe is split among these organizations.
Table 2. Estimate of number of actively working U.S. dispatchers
by labor organization
Labor Organization

Number of Members

ATDA

2000

TCU–Norfolk Southern Railway

60

TCU–Long Island Railroad

33

ACRE–Metro-North Railroad

61

Union Pacific Railroad (Non-Union)

375

Other Non-Union

471

Total U.S. Dispatchers

3000

Ideally, the survey should have drawn from all of these subgroups, but several factors prevented
this. First, no way existed to identify and contact the non-union dispatchers. Privacy concerns
prevent an employer from providing the names and addresses of their employees to researchers.
The researchers attempted to secure participation of multiple unions in the study; however, this
was not possible. Since each union’s participants must receive a cover letter from their union
official, the administrative cost to do this with multiple unions prevented the inclusion of TCU
and ACRE in the sampling frame. The final respondent universe was the 2000 actively working
ATDA dispatchers.

13

2.2

Survey Instruments

This study used two survey instruments, a background survey and a daily log. (Copies of both
instruments appear in Appendix A.) The background survey gathered demographic information,
descriptive data for the dispatcher’s job type and work schedule, and a self-assessment of overall
health. The purpose for collecting this data was twofold. First, it provided data for
characterizing the U.S. dispatcher population. Second, it provided identifying data that
researchers used in conjunction with the daily log to characterize the work/sleep patterns of the
two major categories of dispatcher jobs: trick dispatcher and chief/assistant chief. This
instrument also asked participants to rate, using a Likert scale of 1 to 4, potential sources of
stress at work. The background survey also included a list of life stress events. In the event that
a participant’s daily log indicated frequent nighttime awakenings or excessive fatigue, the
researchers could use the individual’s response to this section of the background survey to assure
that no non-work circumstances were confounding the survey data. Completion of the
background survey required less than 15 min.
The daily log provided the means for survey participants to record sleep and work periods on
both workdays and non-workdays. Dispatchers recorded not only the starting and ending times
for each sleep and work period but also a self-assessment of alertness at different times during
the day. These subjective assessments used a five-point Likert scale. The daily log included
space to record “Comments on today’s sleep experience” and “Comments on today’s work
experience.” The instructions for the log encouraged participants to use this space to explain
anything unusual about the day’s sleep or work. These comments proved useful in
understanding an irregular work or sleep pattern and, in general, complemented the study’s
quantitative findings. Completion of the daily log required less than a total of 10 min daily.
2.3

Data Collection Period

Examination of the relationship between work schedules and fatigue requires data from each
person that encompasses at least a full work cycle. Fatigue is cumulative, and its effects on the
individual are not readily identified from 1 or 2 d of data. In addition, adequate data must be
available to compare sleep periods from both workdays and non-workdays. The typical
dispatcher work cycle is 7 d. The study used a 2-week data collection period in order to capture
the week-to-week variability in the work schedules of extra board dispatchers.
2.4

Sampling Plan

ATDA maintains a database with the names, mailing addresses, and date of birth for all of its
members. This database does not include the type of job that the dispatcher is currently working
or his/her work schedule. Only actively working ATDA members living in the United States
could be in the sampling frame. Retirees, full-time union officials, and anyone currently holding
a railroad management position were not eligible for the study. The effective sampling frame
was 1,963 after these exclusions. The researchers drew a simple random sample without
replacement from these individuals.
One of the most important issues in conducting this study was determining how large a sample
was necessary for the estimates obtained in the sample survey to be reliable enough to meet the
objectives of the study. In general, the larger the sample, the greater the reliability of the

14

resulting estimates, but this must be traded off against the expense of a larger sample. The first
step in this process was to specify the level of reliability needed for the resulting estimates.
Since the study design includes examining characteristics of subgroups of dispatchers working
different work schedules (regular shift, relief schedule, extra board), the study design must assure
that the subgroups have adequate numbers within the overall sample to support reliable estimates
of their characteristics. One statistic of interest is mean number of hours of sleep per day for
each subgroup. Using the ATDA estimate of the workforce breakdown by type of work
schedule, then approximately 1,080 dispatchers work jobs with regular shifts (55 percent), 393
dispatchers work relief jobs (20 percent), and 491 dispatchers work extra board jobs (25 percent).
The appropriate sample size, n, for estimating the mean daily sleep time can be computed from
the following (Levy and Lemeshow, 1999):
( z 2 NV x2 )
n≥ 2 2
z V x + ( N − 1)ε 2
where z = reliability coefficient (1.96 for 95 percent confidence level)
N = population size (regular = 1,080; relief = 393; extra board = 491)
Vx = unknown population variance (1)
ε = error tolerance (.20)
This estimation for sample size also applies to other mean values, such as work and commute
time, that the study seeks to estimate.
Webb (1992) estimates that the standard deviation for daily sleep for the general population is
1 h (Webb, p. 72). Applying this estimate of standard deviation (and hence Vx, variance) to the
dispatcher population (N) and using an error tolerance of ± 10 percent (ε = .20), the sample must
include 88 dispatchers working regular jobs, 77 working relief jobs, and 80 working extra board
jobs. Since relief dispatchers are estimated to be 20 percent of the dispatcher population, using
the assumptions above, the sample size must be 77/.2 or 385 to be reasonably certain of
including enough relief job dispatchers in the study. This sample size will include more than
enough dispatchers working regular jobs and extra board jobs to meet the target error tolerance
of ± 10 percent. Because it was not possible to know a priori the type of job that each ATDA
member works, a stratified sample was not possible.
Since not every dispatcher who is selected to participate in the study would choose to do so, over
sampling was necessary. The extent of over sampling was a function of the anticipated response
rate. Since the signalmen and MOW surveys achieved response rates of 50 percent and 30
percent, respectively, planning for the dispatcher survey assumed a 40 percent response rate,
which was the average of the 2 prior surveys of this type. If 40 percent of the random sample
chose to participate, then the random sample had to be 963 (385/.4) to yield 385 participants.
2.5

Procedure

In accordance with government regulation, FRA sought approval for the proposed survey from
OMB. OMB approved this collection of information under OMB control number 2130-0570 on
February 23, 2006.

15

Concurrent with submittal of the OMB application, the researchers conducted a 1-week pilot
survey with 8 participants to refine the data collection procedures and survey instruments.
ATDA assisted the researchers in identifying suitable participants for this pilot survey. Since
reporting of third shift work and sleep periods might prove problematic, three third shift
dispatchers participated in the pilot. In addition, two participants worked first and second shifts,
two worked relief jobs, and one worked an extra board job. Pilot study participants completed
the Railroad Dispatcher Background Survey and Dispatcher’s Daily Log, as well as a brief PostSurvey Form to provide feedback on the survey instruments and procedures. Similar to the full
survey, pilot participants received a $75 gift certificate to a national retail establishment. Based
on the experience with the pilot survey, the researchers made several changes in the background
survey. The changes were the following:
•

To allow for comparison of survey results with national data on annual use of sick days,
the question on use of sick days was reworded.

•

The question concerning the Hours of Service Law was re-worded.

•

Two sources of stress at work were added, “Lack of break time” and “Inadequate time
off.”

Because some confusion existed among the third shift dispatchers in the pilot as to where to
record morning sleep after a work period, the final instruction sheet that accompanied the survey
materials contained additional instructions that specifically addressed third shift dispatchers.
Following the pilot survey and before mailing of the survey materials, an article in the March
2006 issue of the ATDA publication, Train Dispatcher Journal, publicized the upcoming survey.
The researchers drew a simple random sample of 963 ATDA members, without replacement,
from the sampling frame derived from the ATDA membership list. The package mailed to each
participant on April 18, 2006, consisted of the following items:
•

Railroad Dispatcher Background Survey in booklet form. Each page was 5.5 x 8.5 in,
printed on white paper with no questions on the cover page.

•

Railroad Dispatcher’s Daily Log in spiral notebook form. Each page was 5.0 x 3.25 in.
The log included 14 sections, one for each day of the data collection period. One of the
introductory pages contained brief instructions on completing the log.

•

Cover letter signed by the President, ATDA. This letter explained the purpose of the
study and encouraged ATDA members to participate.

•

Instructions explaining the survey procedures and how to complete the daily log.

•

Return envelope, postage paid.

•

$5 bill.

Copies of the cover letter and instructions appear in Appendix A along with the survey
instruments.
All materials were printed on high quality paper, and each letter was individually addressed to
the recipient. The instruction sheet was printed on yellow paper to increase the likelihood that
recipients would read it. The mailing label for the survey packet used the ATDA return address,
rather than Foster-Miller, because it would be familiar to recipients. The purpose of the $5 was
16

to encourage participation. Those who returned both the background survey and daily log also
received a $75 gift certificate to a national retail establishment.
The instructions emphasized that (a) a total of 14 consecutive d of data should be provided,
(b) data collection should begin on the first day of the next work cycle, and (c) data should not be
reported during vacation periods. Both the instructions and the log included contact information
for two Foster-Miller research staff members who were available to answer questions regarding
the survey instruments and procedures.
Approximately 6 weeks after mailing of the materials, every survey recipient who had not
returned the survey materials received a reminder postcard encouraging him/her to participate
and to call Foster-Miller if he/she needed additional materials.

17

3. Analysis of Survey Data
This chapter presents the survey findings based on data provided in respondent background
surveys and daily logs. The results are organized into six subtopic headings:
•

Survey response rate

•

Non-response bias study

•

Dispatcher demographic characteristics

•

Job characteristics

•

Sleep characteristics

•

Alertness

A separate section presents the results of the textual analysis of the log book comments.
This study used a confidence interval of 95 percent. The researchers used SPSS 15.0 to analyze
the survey data and ATLAS ti V5.2.8 for textual analysis.
3.1

Survey Response Rate

The survey materials were mailed to 963 actively working dispatchers. A total of 445 people
returned both the background survey and the daily log. Five individuals returned only one of the
survey items, and 20 mailings failed to reach the addressees due to bad addresses. Two
respondents were chief dispatchers. Since chief dispatcher is a management job, the responses
from these two people could not be a part of the analysis. (ATDA was unaware of the change in
status for these people so they were a part of the sampling frame.) The overall response rate was
46 percent. The final analysis used data from 443 dispatchers who returned both data collection
instruments.
3.2

Non-Response Bias Study

OMB requires that a non-response bias study be conducted if the survey response rate is below
75 percent. The purpose of the non-response bias study is to assure that no difference exists in
the characteristics of the survey respondents versus the non-respondents.
Information about non-respondents was limited to information available from the ATDA
membership database. In addition to each member’s address, this database includes birthdate.
Birthdate (or age) is an appropriate variable to use for determining non-response bias. For a
number of reasons, age is an important characteristic for assessing potential bias in this study.
First, human sleep patterns change with age (Van Cauter, Leproult, & Plat, 2000). In addition,
age is highly correlated with years of work experience and seniority. Seniority allows a
dispatcher more opportunity to select work schedules that meet his/her personal needs.
All 445 individuals who returned both the background survey and the daily log were
respondents, and the remaining 518 were non-respondents. Of the 518 non-respondents, ATDA
membership records included birthdates for 470. Analysis of mean age for each of the groups

19

found no significant difference between the respondents and the non-respondents, t(913) = 1.93,
p = .054.
3.3

Dispatcher Demographic Characteristics

This section provides demographics, as well as basic job-, family-, and health-related
information based on responses in the background survey. Where appropriate, the study includes
comparisons of the study results with national norms.
Characterization of dispatchers considered a number of factors. These factors are job type, work
experience, sex and age, marital and family status, overall health, workdays lost due to illness,
incidence of sleep disorders, and consumption of caffeinated beverages. The sections below
discuss each of these elements. A brief summary of this information follows.
3.3.1 Job Type
Respondents reported their job type as trick dispatcher, assistant chief, or other. The majority
(87 percent) was trick dispatchers, 10 percent held assistant chief positions, and the remaining
3 percent reported their job type as other. Review of the other responses to the job type question
allowed classification of these respondents into one of the other two groups based on whether or
not the position was covered by the Hours of Service Law. Those positions covered by this law
became part of the trick dispatcher group, and those not covered were combined with the
assistant chief group.
3.3.2 Sex and Age
Railroad dispatchers are a predominantly male population. ATDA membership in May 2006
was 14.5 percent female. Fourteen percent of the survey participants were female.
The average age for all dispatchers is 46.9 yr, and the median age is 49 yr. Figure 3 displays the
age distribution for dispatchers overall and by sex, based on the survey results. As is typical for
other railroad crafts, this is an aging work force. Nearly half of all dispatchers are 50 yr and
older, and over three-quarters are 40 yr and older. Approximately two-thirds of the female
dispatchers are under 50 yr, but 49 percent of the male dispatchers are in this age group.
Table 3 presents the overall age data by sex. The mean age for male dispatchers was 47.3 yr and
for female dispatchers, 43.9 yr. The average female dispatcher is 3.4 yr younger than the
average male dispatcher, but the median ages for the two groups differ by 6 yr. A statistically
significant difference exists in the age of the two groups, t(441) = 2.68, p < .05.
In the early 1980s, railroads did not hire many dispatchers. By the end of this decade when
railroads began hiring again, an effort to include women and minorities in all crafts, including
dispatchers, occurred. At the same time, changes in communications technology made the
position of tower operator obsolete. Traditionally, a tower operator position was on a career path
to becoming a dispatcher. It was at this time that women began to enter dispatching in larger
numbers than previously. This hiring pattern is likely responsible for the current difference in
both age and experience levels between male and female dispatchers. With retirement of the
senior male dispatchers in the coming years, this difference by sex should become smaller and
perhaps disappear.

20

60

50

Male
Female
Total

48.2
45.6

40

Percent

36.1

29.5

30
26.2

25.4
20.8

19.9

20

26.9

10
3.7

4.9

3.8

2.9

3.3

2.9

0
20–29

30–39

40–49

50–59

60+

Age Group

Figure 3. Distribution of dispatchers by age group and sex

Table 3. Dispatcher age by sex (yr)
Mean

Median

Standard
Deviation

Male

47.3

50

9.3

Female

43.9

44

8.7

All

46.9

49

9.3

Group

Research has found that a higher perceived age, relative to chronological age, can be an indicator
of chronic stress and poor psychological well-being (Barnes-Farrell & Petrowski, 1989, 1991).
Overall dispatchers reported a lower perceived age (43.2 yr) in comparison with their average
chronological age (46.9 yr). As shown in Table 4, the dispatcher population tends to feel
younger as they age. This is the same pattern that Barnes-Farrell and Petrowski found with
permanent day shift workers in a manufacturing plant. Barnes-Farrell and Petrowski point out
that younger people tend to report feeling older to reflect perceived maturity. The dispatchers’
perceived age follows the pattern reported by Barnes-Farrell and Petrowski and, as such, is not
indicative of poor psychological well-being.

21

Table 4. Discrepancies between chronological and perceived age by age group (percent)
Worker Age (yr)

Age
Perception

20–29

30–39

40–49

50–59

60+

Younger

17.6

39.1

54.6

62.4

69.2

Same Age

35.3

19.6

19.3

15.8

15.4

Older

41.2

41.3

18.5

15.3

7.7

3.3.3 Experience
The average dispatcher in the survey had 14.2 yr of experience as a dispatcher. The median level
of experience was 11.4 yr. The lower median value indicates that individuals with years of
experience below the mean dominate the group (see Table 5). Assistant chiefs had more
experience than the trick dispatchers, which is not surprising, since a dispatcher cannot become
an assistant chief without desk dispatching experience.
Table 5. Years of experience as a dispatcher by job type (yr)
Mean

Median

Standard
Deviation

Trick Dispatcher

13.8

11.4

9.5

Assistant Chief

17.8

14.7

11.8

Other

16.0

12.9

12.5

All

14.2

11.8

9.9

Job type

As Table 6 indicates, dispatchers have had nearly all of their dispatching experience with their
current employer. This is true regardless of job type. On average, dispatchers have been with
their current employer for 12 yr of their 14-yr career as a dispatcher. It also appears that, based
on the age of this group of railroad employees, most dispatchers came to dispatching after
working in another job. The average age of a dispatcher is 47 yr, and he/she has been working in
this craft an average of 14 yr. (The survey did not collect information concerning whether or not
the prior job was in the railroad industry.)
Table 6. Years of experience as a dispatcher with current employer by job type (yr)
Mean

Median

Standard
Deviation

Trick Dispatcher

11.9

9.5

8.7

Assistant Chief

14.0

12.7

8.9

Other

13.6

7.0

12.2

All

12.2

10.0

8.9

Job type

22

Males have historically dominated all railroad crafts. In recent years, however, the dispatching
profession has seen an increase in the number of women in its ranks. Table 7 presents years of
dispatching experience by sex. On average, female dispatchers have 4 yr less experience than
their male counterparts. The level of experience by sex is consistent with the difference in age
between the two groups.
Table 7. Years of experience as a dispatcher by sex (yr)
Mean

Median

Standard
Deviation

Male

14.8

12.3

10.1

Female

10.6

9.2

7.3

All

14.2

11.8

9.9

Sex

3.3.4 Marital and Family Status
Most recent statistical data from the U.S. Census indicates that 54.7 percent of the U.S.
population age 18 and older and 56.6 percent of the U.S. male population 18 and older are
married (U.S. Census Bureau, 2005). At the time of the study, 75.6 percent of participants were
married, 12.2 single, 10.4 divorced, 0.5 widowed, and 1.4 categorized themselves as other (these
people were likely separated or living with a partner). Since many railroaders report that their
work schedule strains marital relationships, finding such a high proportion of dispatchers who
are married was surprising. This data, however, does not indicate whether or not the married
individuals were in an initial marriage or one subsequent to a divorce.
The survey asked participants whether or not their family included young children, a factor that
can lead to disrupted sleep. While a large percentage of dispatchers are married, very few have
children under the age of 2 yr (6.1 percent). This finding is not surprising given the average age
of a dispatcher.
3.3.5 Health
Participants rated their health as excellent, good, fair, or poor. Nearly 85 percent of dispatchers
rated themselves in good (65.5 percent) or excellent (19 percent) health (see Figure 4). Analysis
of these self-assessments of health by work schedule revealed that a statistically significant
difference existed based on work schedule, F(4, 437) = 2.65, p < .05. Post hoc tests identified a
difference between third shift and extra board dispatchers with third shift dispatchers having a
lower overall health self-rating (see Table B-1 in Appendix B).
Figure 5 contains the frequency distribution for dispatcher sick days. While the majority of
dispatchers rated their health as good or excellent, this group averaged 5.6 workdays lost due to
illness annually. In comparison, U.S. employed adults with paid sick time averaged 3.6 d of sick
time (U.S. Department of Health and Human Services, 2006, p. 49). One possible explanation
for the higher rate of sick days among dispatchers is the stressful nature of the job that may lead
to stress-related medical conditions. Popkin, Gertler, and Reinach (2001) found a higher rate of
stress-related medical conditions among dispatchers than U.S. population norms. Another

23

70
65.5

60

50

Percent

40

30

19

20
13.3
10
2
0
Poor

Fair

Good

Excellent

Rating

Figure 4. Self-assessment of overall health
25

20.9
20.1
20

Percent

15

11.9

10

9.2

8.7

7.1
5.9

5.5

5

4.3
3.4

3

0
0

1

2

3

4

5

6

7

8

Days

Figure 5. Workdays lost due to illness in past year

24

9

10+

possible explanation is that some dispatchers, particularly those working an irregular schedule,
may mark off sick to catch up on sleep.
3.3.6 Incidence of Sleep Disorders
The Wisconsin Sleep Cohort Study, a longitudinal study of cardiopulmonary sleep disorders
among middle-aged working adults, estimated that 2 percent of women and 4 percent of men
have sleep apnea (Young, et al., 1993). (The definition of sleep apnea for this study was an
apnea-hypopnea score of 5 or higher and daytime hypersomnolence.) The National Sleep
Foundation (NSF) and the National Institutes of Health report the numbers from the Wisconsin
study as an estimate of the prevalence of sleep apnea among U.S. adults. Some sleep researchers
hypothesize that the prevalence of sleep apnea may in fact be higher because many remain to be
diagnosed. According to the Wisconsin study, 9 percent of women and 24 percent of men have
undiagnosed sleep-disordered breathing, a condition that in some people results in excessive
daytime sleepiness.
Of the 443 participants in this study, 41 or 9.3 percent reported having a diagnosed sleep
disorder. Of those with a diagnosed disorder, 63.4 percent reported receiving treatment. A
separate question inquired about sleep apnea. Of the total group, 7.4 percent reported having
sleep apnea. (This implies that 1.9 percent have a sleep disorder other than sleep apnea.) Twothirds of those with sleep apnea are receiving treatment. Two possible reasons exist as to why
the dispatchers’ incidence of sleep apnea is higher than the reported norm for U.S. middle-aged
working adults. Railroad and labor educational programs, as well as media publicity, in recent
years may have made this group of railroad employees more cognizant of the symptoms of sleep
apnea and its consequences and caused them to seek medical evaluation. In addition, dispatching
is a sedentary job that requires sitting in one place for an 8-h period. Sedentary jobs may lead to
weight gain, which exacerbates sleep apnea.
3.3.7 Consumption of Caffeinated Beverages
NSF reports that 250 mg of caffeine a day, the equivalent of a soda and a couple of coffees,
generally poses no harm. Almost all participants reported consuming caffeinated beverages on a
daily basis (90.7 percent), and those who did averaged 3.5 beverages a day. Based on this level
of caffeine consumption, dispatchers are within normal healthy limits, and their sleep, in general,
is not likely disrupted due to caffeine unless caffeine consumption occurs close to bedtime (NSF,
2002a).
3.3.8 Summary of Dispatcher Demographic Characteristics
At the time of the survey, 87 percent of the dispatchers held trick dispatcher jobs, and the
remainder worked as either an assistant chief or other non-Hours of Service position. The
average dispatcher is 47 yr old and has worked as a dispatcher for 14 yr. Nearly half of all
dispatchers are 50 yr or older, and 14 percent are female. Female dispatchers have, on average,
4 yr less experience and are 3.5 yr younger. Seventy-six percent are married, but very few have
children under the age of 2 yr. Nearly 85 percent rate their health as good or excellent, but
dispatchers’ use of company-paid sick days exceeds that of U.S. working adults with paid sick
days. Over 7 percent of dispatchers report having sleep apnea, and another 2 percent have sleep

25

disorders other than sleep apnea. A third of those dispatchers with a sleep disorder remain
untreated.
3.4

Job Characteristics

This section explores several aspects of the dispatcher’s job, including work schedule, hours
worked, workday breaks, commute time, and sources of workplace stress.
3.4.1 Work Schedule
Work schedules fell into five basic categories: permanent first, second, and third shifts; relief;
and extra board. Researchers defined shifts based on the starting time of the workday as follows:
first shift–workdays starting between 4:30 a.m. and 10 a.m., second shift–starting after 10 a.m.
and extending less than 4 h after 12 a.m., and third shift–any shift running at least 4 h after 12
a.m. Relief schedules are those that forward rotate through shifts but work the same days each
week. Finally, extra board jobs are those positions requiring filling in for regular and relief
dispatchers, and hence the schedule is irregular.
Nearly 70 percent of dispatchers worked the same shift each day. The remainder was split
between relief (19 percent) and extra board (12 percent). Table 8 compares the work schedules
of trick dispatchers (i.e., those covered by Hours of Service) and assistant chiefs (i.e., those not
covered by Hours of Service). Both trick dispatchers and assistant chiefs worked across each
schedule type. One notable difference between the schedules of the two groups is that 14 percent
of trick dispatchers worked the extra board, whereas only one assistant chief reported working
the extra board. Railroads usually have trick dispatchers who are qualified to work their assistant
chief positions. When they have a need to fill in for an absent assistant chief, a trick dispatcher
fills the vacancy rather than having an extra board for assistant chief jobs. This is why it is rare
to find someone holding an assistant chief position with an extra board schedule.
Table 8. Work schedule by type of job and overall (percent)
Job Type
Work Schedule*

Trick Dispatcher
(n = 385)

Assistant Chief
(n = 58)

Total
(n = 443)

1

26.2

19.0

25.3

2

19.5

31.0

21.0

3

22.3

25.9

22.8

Relief

18.2

22.4

18.7

Extra Board

13.8

1.7

12.2

*1 = first or day shift, 2 = second or evening shift, 3 = third or night shift

Table 9 presents age and experience data by work schedule. Dispatchers working first shift are
older than those working other schedules, F(4, 438) = 8.87, p < .05. (See post hoc test for age in
Appendix B, Table B-2.) Likewise, first shift dispatchers have more years of experience than

26

those working other schedules, and extra board dispatchers have the least experience, F(4, 433) =
33.80, p < .05. (See post hoc test for experience in Appendix B, Table B-3.)
Table 9. Work schedule by age and experience (yr)
Age

Experience

Work Schedule

Mean

Median

Mean

Median

1

50.5

52.0

21.5

21.9

2

46.6

50.0

13.7

11.8

3

45.7

47.0

13.1

10.3

Relief

46.0

49.0

12.5

11.8

Extra Board

42.7

43.5

5.6

3.1

A variety of shift rotation combinations for relief jobs exists, but by far the most common is two
first shifts, followed by two second shifts, and a third shift. The Hours of Service Law precludes
backward shift rotation for dispatchers. Figure 6 provides a breakdown of the shift patterns of
relief dispatchers.
50
45.8
45

40

35

Percent

30

25

20
15.8
15
9.6

10

7.2

6

4.8

5

3.6

3.6

3.6

11222

11333

22223

0
11223

12233

11112

23333

22233

Shift Pattern

Figure 6. Relief schedules

27

Other

Because relief jobs rotate through various shifts during the workweek, and because extra board
jobs work irregular shift patterns, researchers were interested in the actual shifts worked as a
function of schedule type. Table 10 presents the actual shifts worked by work schedule. Results
indicate that those working permanent shifts rarely work outside their typical schedules, but due
to staff shortages, vacations, sickness, and other absences, this can occur. Relief jobs worked
primarily second shift (40 percent), followed by first shift (33 percent), and third shift (27
percent).
Table 10. Actual shifts worked by job schedule (percent)
Work Schedule

Shift Worked

1
(n = 112)

2
(n = 93)

3
(n = 101)

Relief
(n = 83)

Extra
Board
(n = 54)

1

96.5

1.6

1.9

32.9

42.1

2

3.0

97.9

3.4

39.7

34.7

3

0.5

0.6

94.7

27.0

22.9

It was surprising to find such a large percentage of extra board dispatchers working first shift.
The expectation was that extra board dispatchers would be split evenly among the three shifts
with perhaps a slightly smaller proportion on third shift. (Passenger railroads and some freight
railroads combine desks at night and on weekends so fewer opportunities exist to work third
shift.) A possible explanation for the survey results is that these dispatchers were filling in for
experienced dispatchers who take vacation during periods of good weather. (The majority of
survey respondents completed the survey from the end of April through the end of June.) In
addition, since regular first shift dispatchers are the most experienced and therefore accrue the
largest number of vacation and sick days, this is likely the shift that most frequently uses an extra
board dispatcher, regardless of the time of year.
Shift variability can lead to fatigue if it disrupts a worker’s normal sleep pattern. Investigation of
shift variation provided a means to estimate work schedule variability. The study defined a
variation in shift as a change in shift (e.g., from first to second) from the previous day. If the
previous day was a non-workday, then no shift variation occurred. The start of the workday
determined the shift categorization, as previously stated. Table 11 presents the number of shift
variations for relief and extra board dispatchers throughout the 2-week study period.
Another surprise was the large percentage of extra board dispatchers who had no shift variation
through the study period. As stated above, however, this may be explained by extra board
dispatchers who fill in for more experienced permanent shift dispatchers when they take vacation
during periods of good weather. If the dispatcher did not have 2 consecutive rest d, then the
definition of shift variability may also result in underestimation of extra board dispatcher shift
variability.
Even though one would expect all relief dispatchers to have some shift variations due to the
nature of this work schedule, two dispatchers in the survey reported working the same shift
throughout the study period, therefore having no shift variation.
28

Table 11. Shift variability for relief and extra board work schedules (percent)
Work Schedule

Number of Shift Variations
(in 2-week period)

Relief

Extra Board

0

2.4

35.2

1

9.6

22.2

2

32.5

27.8

3

19.3

13.0

4

33.7

1.9

5+

2.4

.0

3.4.2 Hours Worked
The study collected data on a nominal workweek, a typical workweek, and actual hours worked.
In the background survey instrument, both trick dispatchers and assistant chiefs reported a
nominal workweek of 40 h. During the study period, trick dispatchers worked only slightly more
than this. In contrast, assistant chiefs averaged approximately 45 h of work per week. In fact,
one-quarter of assistant chiefs worked 48 h or more per week, which is 8 h more than their
nominal schedules require. The 75th percentile for trick dispatchers’ workweek was 43 h, which
is only 3 h more than their nominal schedules.
Actual work was similar to reported typical values for both groups. This suggests that the time
period for this study was a typical one with respect to dispatcher work schedules. In addition,
assistant chiefs, who are not covered by the Hours of Service Law, do indeed work substantially
more than their nominal 40-h workweek. Table 12 presents typical and actual work for trick
dispatchers and assistant chiefs.
Table 12. Typical and actual work for 2-week period by job type (h:min)
Trick Dispatcher

Assistant Chief

Mean

Median

Std.
Dev.

25th %

75th %

Mean

Median

Std.
Dev.

25th %

75th %

Typical
Work

82:22

80:00

6:06

80:00

80:00

90:01

80:00

12:54

80:00

96:00

Actual
Work

81:09

80:25

9:49

77:52

86:06

89:19

85:33

13:34

80:11

96:56

Number
of Workdays

10.1

10

1.1

10

11

10.6

10

1.6

10

12

3.4.3 Breaks
Unlike railroad crafts such as signalmen and MOW workers, who have contractual provisions for
a meal break after 4 h on the job, railroad dispatchers have no such contractual provisions. As
29

such, a trick dispatcher could potentially sit at his/her desk monitoring communications and
controlling track movements without a break for up to 9 h, or even 13 h in the case of an
emergency. Assistant chiefs, because they are not covered by the daily limitations of the Hours
of Service Law, can potentially work 12 to 16 consecutive h without a break. This is somewhat
concerning knowing that mental fatigue can accumulate as a result of time on task (Popkin,
Reinach, & Gertler, 2001).
Dispatchers occasionally leave their desk to use the bathroom or to bring food back to the desk,
but this is not always possible, however, due to the nature of the job and the workload that a
dispatcher handles. Leaving the desk means potentially missing important communication from
train crews, MOW, or others that may require immediate response.
Each day of the study period, participants recorded the number of breaks they took and the
duration of their longest break of the day. Trick dispatchers averaged 2 breaks per day, with the
longest break averaging 8 min. Assistant chiefs averaged 1½ break per day, with the longest
break averaging 10 min (see Table 13).
Table 13. Number and length of longest break per day by job type
Number of Breaks
Job

Length of Longest Break
(min)

Mean

Median

Mean

Median

Trick Dispatcher

2.06

2

8

7

Assistant Chief

1.53

1

10

10

Figure 7 presents the number of breaks per day by job type. Almost one-third of the time, both
trick dispatchers and assistant chiefs took no breaks during their shift. Researchers treated blank
fields in the diary as missing data rather than zero; therefore, it is likely that the percentage of
shifts when dispatchers did not have a break was even greater than the survey results indicate.
Although median values for break length on third shift were higher than those on first and second
shift, researchers found no differences in the average number of breaks by shift (see Table 14).

30

35
31.2
30

Trick Dispatchers
Assistant Chiefs

29.1
26

25
21.5
20.4

20.4
19.2

Percent

20
17
15

13.1
12

10

5

0
0

1

2

3

4 or more

Number of Breaks

Figure 7. Number of breaks per day by job type

Table 14. Breaks by shift worked
Number of Breaks

Length of Longest
Break (min)

Shift

Mean

Median

Mean

Median

1

2.05

2

8

7

2

1.95

2

8

6

3

1.99

2

8

10

Assistant chiefs, despite averaging fewer breaks than trick dispatchers, took longer breaks.
Approximately one-quarter of breaks taken by assistant chiefs were 10 min or more in length,
compared to only 13 percent of breaks taken by trick dispatchers. Figure 8 presents a more
detailed breakdown of longest break length by job type.

31

50
46.9

46.6
Trick Dispatchers
Assistant Chiefs

45
39.8

40

35
31

Percent

30

25

22.4

20

15

13.3

10

5

0
1–5

6–10

>10

Length of Longest Break (min)

Figure 8. Length of day’s longest break by job type
3.4.4 Commute Time
For the purpose of this study, commute time refers to travel between the dispatcher’s home and
his/her work site. Figure 9 depicts the average dispatcher workday, including commute times,
for trick dispatchers and assistant chiefs. Both groups of dispatchers travel over 35 min to work.
This commute is longer than the U.S. Census Bureau (2004) estimate of 25.5 min for U.S.
workers, excluding those who work at home. The commute home tends to be longer because,
according to notes in the log books, many dispatchers do errands or other activities on the way
home. The primary difference in the length of the workday for the two types of dispatcher jobs
is in the length of the work period. On average, assistant chiefs work 40 min longer each day
than trick dispatchers. Both trick dispatchers and assistant chiefs arrive 10 to 15 min before the
start of their shift to begin the transition from the dispatcher working the position on the previous
shift.
Figure 10 shows the distribution of commute start times for dispatchers. The tri-modal
distribution reflects commutes for first, second, and third shifts. Although most first shift
positions began at 6:30 a.m., some began as early as 4:30 a.m., and hence some of the commutes
began as early as 4 a.m.

32

Commute to Work

Time Before Start Work

Work

Commute Home

11.5

Trick Dispatchers

35.7

484

50

12.7

Assistant Chiefs

38.7

524

0

100

200

48

300

400

500

600

700

Minutes

Figure 9. Commute time and workday by job type
25
21.7

21.1

20
16.8
14.3

Percent

15

12.3
11
10

5

Figure 10. Start of commute to work

.

.
01
–1
2

a.
m

p.
m
10
:

01
–1
0
8:

6:

01
–8

p.
m

p.
m

.

.

.
4:

01
–4
2:

01
–6

p.
m

.
p.
m
1–
2

12
:0

a.
m
–1
2
1

10
:0

Time

33

1

0.5

0.3

p.
m
.

a.
m
.
01
–1

0

a.
m
8:

1–
8
6:
0

1–
6
4:
0

.

a.
m
.

.
a.
m
1–
4
2:
0

12
:0

1–
2

a.
m
.

0

0.5

0.4

0

3.4.5 Sources of Stress
In the background survey, dispatchers rated job-related sources of stress. They rated stress using
a Likert scale with values from 1 to 4 with 1–no stress, 2–a little stress, 3–stressful, and 4–very
stressful. The top four sources of stress for both trick dispatchers and assistant chiefs were
management policies, surges in workload, ambiguous rules or procedures, and responding to
emergencies (see Figure 11). While many of the comments in the trick dispatchers’ daily logs
bemoaned the lack of regular breaks, lack of break time was in the middle of the list of stressors
for trick dispatchers. As shown in Table 15, in 4 of the 14 sources of stress, statistically different
ratings existed between the 2 groups of dispatchers. Trick dispatchers assigned significantly
higher ratings of stress than assistant chiefs to the following categories: loss of sleep, inadequate
time off, ambiguous rules or procedures, and lack of break time. The difference in the ratings for
loss of sleep is likely due to a difference in job schedule. A substantial percentage of trick
dispatchers work the extra board and must constantly adjust their sleep patterns to obtain proper
rest. A higher stress rating for lack of break time by trick dispatchers supports the finding that
assistant chiefs are getting longer breaks. With regard to inadequate time off, trick dispatchers
typically have less seniority than assistant chiefs and therefore do not have as much vacation
time. Finally, the difference in stress rating with respect to ambiguous rules may result from the
fact that assistant chiefs do not necessarily apply or exercise the rules as frequently as trick
dispatchers.
3.5
Trick Dispatchers
Assistant Chiefs
3

2.5

Rating

2

1.5

1

0.5

M
an

ag
e

m
en
tp
Su
ol
rg
ic
es
ie
in
s
w
or
A
kl
m
oa
bi
gu
d
ou
s
C
ru
om
le
Em
s
R
m
er
es
un
g
po
en
ic
ns
at
ci
io
ib
es
n
ili
p
ty
ro
fo
bl
ro
em
th
s
er
s'
La
sa
ck
fe
of
ty
br
ea
k
In
tim
ad
eq
e
ua
te
st
Pr
Lo
af
es
f
ss
su
o
re
fs
C
oo
to
le
ep
rd
fin
in
i
sh
In
at
ad
io
a
eq
n
ta
w
sk
ua
ith
te
ot
tim
he
e
rd
of
ep
f
ar
tm
La
en
ck
ts
of
co
nt
ro
Jo
l
b
se
cu
rit
y

0

Source of Stress

Figure 11. Sources and levels of stress by job type
34

Dispatchers had different sources of stress based on their work schedule. Appendix B, Table B-4
provides ANOVA results for sources of stress by work. The statistically significant differences
were the following:
•

Permanent first shift dispatchers were less troubled by job security than permanent third
shift and extra board dispatchers.

•

Permanent third shift, relief, and extra board dispatchers rated inadequate time off as a
greater source of stress than did permanent first shift dispatchers.

•

Dispatchers working permanent third shift rated sleep loss a greater stressor than those
working permanent first or second shift.

•

Extra board dispatchers were more concerned about lack of control over their work
schedule than all other schedule groups.

•

Extra board dispatchers were less concerned about coordination with other departments
than dispatchers working permanent first, second, or third shifts.

•

Extra board dispatchers felt less pressure to finish a task than those working permanent
first and second shifts.

•

Ambiguous operating rules or procedures were less a source of stress for extra board
dispatchers than for permanent first and second shift dispatchers.

•

Extra board dispatchers rated communication problems as less of a source of stress than
all other schedule groups except permanent third shift.

3.4.6 Job Characteristics Summary
The around-the-clock nature of railroading requires railroads to staff dispatching centers with
three 8-h shifts. Dispatchers working permanent first, second, or third shifts rarely work outside
their typical schedules. The most common relief rotation is working two first shifts, followed by
two second shifts, and a third shift.
Surprisingly, a large percentage of extra board dispatchers worked first shift. This is most likely
because first shift dispatchers are the oldest and most experienced, therefore likely to have and
use the most vacation and sick time. This may also explain why more than one-third of extra
board dispatchers had no shift variation through the study period.
Both trick dispatchers and assistant chiefs reported a nominal workweek of 40 h. Trick
dispatchers worked just slightly over their nominal amount, while assistant chiefs worked
approximately 45 h per week. One-quarter of assistant chiefs worked 48 h or more per week, 8 h
more than their nominal schedules require. Because the Hours of Service Law does not apply to
assistant chiefs, the additional work hours could be either an extra day of work or more than 8 h
in a single day. Actual work was similar to reported typical values for both groups, suggesting
that the time period for this study was a typical one with respect to dispatcher work schedules.
Railroad dispatchers have no contractual provisions for breaks. Almost one-third of the time,
both trick dispatchers and assistant chiefs took no breaks during their shift. Assistant chiefs take
fewer but longer breaks than trick dispatchers.

35

Table 15. Stress ratings by job type
Trick
Dispatcher

Assistant
Chief

Management policies

2.98

2.95

t (440) = .29, p = .776

Surges in workload

2.86

2.71

t (439) = 1.25, p = .211

Ambiguous rules or
procedures

2.85

2.57

t (440) = 2.13, p < .05*

Responding to emergencies

2.74

2.81

t (440) = -.56, p = .573

Communication problems

2.69

2.43

t (440) = 1.93, p = .054

Responsibility for safety of
others

2.68

2.48

t (440) = 1.44, p = .152

Lack of break time

2.66

2.14

t (439) = 3.57, p < .05*

Inadequate staffing

2.52

2.36

t (440) = 1.13, p = .261

Loss of sleep

2.51

2.22

t (440) = 2.34, p < .05*

Pressure to finish a task

2.47

2.57

t (440) = -.83, p = .407

Inadequate time off

2.39

2.07

t (440) = 2.14, p < .05*

Coordination with other
departments

2.31

2.43

t (439) = -.96, p = .337

Lack of control over work
schedule

1.97

1.79

t (438) = 1.27, p = .207

Job security

1.96

1.88

t (440) = .57, p = .570

Source of Stress

Significance Test

Note: 1 = no stress, 2 = a little stress, 3 = stressful, 4 = very stressful
*
statistically significant at α = .05

Trick dispatchers and assistant chiefs had similar sources of stress. In general, trick dispatchers
had higher stress ratings than assistant chiefs, with ratings on sleep loss, lack of time off,
ambiguous rules, and lack of breaks being significantly higher. Some differences existed in
stress ratings by work schedule.
3.5

Sleep Characteristics

This study examined primary sleep, as well as naps. For first and second shift dispatchers,
primary sleep refers to nighttime sleep. For third shift dispatchers, this study defined primary
sleep as the first sleep period following the end of the individual dispatcher’s work period. An
alternative way to define primary sleep for third shift dispatchers would have been to select the
longest sleep period of the day, but this would have complicated the design of the log book.
Designing a log book that dispatchers working all shifts could use presented a challenge in terms
of where to record the primary sleep period and the related sleep quality ratings. To facilitate
36

design and use of the log book for all study participants, the daily log book contained only one
sleep period for each day that had sleep quality ratings. The log book sleep section was
chronological with the first entry being the nighttime or primary sleep period. First and second
shift dispatchers recorded their nighttime sleep in this section upon arising in the morning.
When a dispatcher worked third shift, he/she recorded his/her morning sleep period in this
section. The next page of the log’s sleep section provided space to record sleep periods or naps
that occurred later in the day.
Some adjustment of the nap data was necessary to accurately identify naps versus split primary
sleep periods. The diary instructions asked dispatchers to record split nighttime sleep as a
primary sleep period plus a nap. (This scheme permitted a more accurate computation of
nighttime sleep.) For nap entries on non-workdays, if the nap began between 12 a.m. and 7 a.m.,
then the researchers added nap duration to primary sleep duration. For first shift dispatchers’
workday entries, if the nap began after the person went to sleep, but before he/she began the
commute to work, then the nap was considered split primary or nighttime sleep and was added to
the primary sleep duration. For second and third shift dispatchers, if the nap began less than 1 h
after the recorded wake-up time or before 6 a.m., then the nap time was added to primary sleep
time and was not a part of the nap analysis.
The ending time of the sleep period determined its assignment to a calendar day. The
categorization of each calendar day as a workday or a non-workday depended upon whether or
not a work period began on that day. In other words, if a sleep period ended on a day on which
the individual began a work shift, then the sleep period was in the category workday sleep.
Understanding the method for categorizing sleep as workday versus non-workday is important
when reviewing the sleep results.
The following sections discuss primary and total daily sleep, sleep quality, sleep latency, and
naps or supplementary sleep periods.
3.5.1 Primary and Total Daily Sleep
The study considered both total daily sleep and the primary sleep period. Table 16 contains the
sleep data for total daily sleep by type of day and work schedule. These data indicate that
dispatchers working first shift jobs get considerably less sleep on workdays than those with other
work schedules, but first shift dispatchers appear to make up for their loss of sleep on nonworkdays. When interpreting these data, it is important to keep in mind that on the last day of
the work cycle, a dispatcher working third shift may choose to take only a short morning nap
after work or to not sleep at all until that night. This means that his/her non-workday sleep for
the first non-workday could be 0 h or merely a nap for a few hours. Similarly, the night before
the dispatcher’s first third shift, he/she usually gets a full night’s sleep. For this reason, the
workday sleep for the third, relief, and extra board jobs is higher than what might be expected for
dispatchers who work third shift, and the non-workday sleep appears understated. For example,
the workday sleep for permanent third shift dispatchers is 6:52 (h:min), and their non-workday
sleep is 6:16.
Results from a meta-analytic review of shiftworker sleep provide data that can be compared with
the results from the present study. Pilcher, Lambert, and Huffcutt (2000) examined 168 primary
studies of shiftworkers and selected 36 for their meta analysis. These 36 studies met the
following criteria: the study had to report on actual shiftwork, not laboratory-based shiftwork;
37

the shift length had to be 8 h in duration and had to be either permanent evening, permanent
night, or rotating; and the study had to include data on self-report of sleep length. Sleep length
for their control group, non-shiftwork day workers, came from a 1998 NSF report. The article
does not specify the data collection instrument (log book versus questionnaire) for these studies.
Further, it is not clear from the article if sleep refers to the primary sleep period or total daily
sleep.
Table 16. Total sleep by type of day and work schedule (h:min)
Type of
Day

Workday

NonWorkday

Standard
25th
75th
Deviation Percentile Percentile

Work Schedule

Mean

Median

1

6:23

6:26

0:46

5:50

6:55

2

7:05

7:05

0:49

6:25

7:42

3

6:52

6:58

1:11

5:58

7:40

Relief

7:00

7:08

0:56

6:22

7:39

Extra Board

6:57

6:49

1:05

6:13

7:54

1

7:53

7:58

1:13

7:04

8:47

2

7:22

7:14

0:55

6:45

8:02

3

6:05

6:00

1:35

5:18

7:03

Relief

6:16

6:23

1:34

5:29

7:14

Extra Board

7:21

7:15

1:48

5:57

8:29

Table 17 compares the total workday sleep of dispatchers, by type of work schedule, with the
data from the Pilcher, Lambert, and Huffcutt (2000) study. In comparison with the shiftwork
norms, first and second shift railroad dispatchers get less sleep than other shiftwork populations,
and first shift dispatchers get the least sleep. The reason that the third shift and relief dispatchers
appear to get more sleep is most likely because of how this study assigned sleep to workdays
versus non-workdays. The researchers also surmise that the other shiftwork studies used a
questionnaire that inquired about usual sleep duration on workdays. This type of questionnaire
would lead to a shorter sleep duration for third shift than a daily log book study, such as the one
used for this study, that includes the nighttime sleep before the first workday in the average
workday sleep.

38

Table 17. Total sleep by work schedule versus shiftwork norms for workdays (h:min)
25th
Percentile

75th
Percentile

Shiftwork
Norm*

Work Schedule

Mean

Median

Standard
Deviation

1

6:23

6:26

0:46

5:50

6:55

7:00

2

7:05

7:05

0:49

6:25

7:42

7:34

3

6:52

6:58

1:11

5:58

7:40

6:36

Relief

7:00

7:08

0:56

6:22

7:39

6:39

Extra Board

6:57

6:49

1:05

6:13

7:54

n/a

*Source: Pilcher et al., 2000.
n/a = not available

Since relief and extra board dispatchers work different shifts over the course of their workweek,
it is appropriate to examine the dispatcher sleep patterns by shift actually worked. Table 18
presents total and primary sleep by shift actually worked. The difference between the primary
and total sleep for dispatchers working third shift reflects that the majority of dispatchers
working this shift employ a split sleep strategy on workdays. Among shiftworkers, second shift
workers typically get the most sleep because they do not have to get up early to go to work. First
shift workers get less sleep than second, and third shift people get the least sleep. This is the
case with the dispatcher results for primary sleep. Once again, the primary sleep for third shift
includes the nighttime sleep that occurred the night before the first third shift workday. For both
total and primary sleep, all shifts are statistically different from each other, F(2, 4398) = 137.49,
p < .05 and F(2, 4398) = 243.13, p < .05.
Table 18. Total and primary sleep on workdays by shift worked versus shiftwork norms
(h:min)
Dispatchers
Shift Worked

Total Sleep

Primary Sleep

Shiftwork
Norm*

1

6:19

6:08

7:00

2

7:14

7:06

7:34

3

6:59

5:54

6:36

*Source: Pilcher et al., 2000.

Effect size is a measure of the degree to which a relationship exists between two factors. By
computing the effect size for the relationship between total sleep and shift worked, and primary
sleep and shift worked, it is possible to compare the relative effect of shift worked on the two
sleep metrics. The effect size for total sleep is .24, and for primary sleep it is .32. According to
Cohen (1988), an effect size of .25 is a medium effect size. Using Cohen’s criteria, the effect
size for total sleep is small, and that for primary sleep is medium. This means that shift worked
has more of an effect on primary sleep than total daily sleep.
39

Table 19 provides additional descriptive statistics for primary sleep by shift worked. Both the
total and primary sleep results for first shift appear low in comparison with the shiftwork norms.
More than 75 percent of those working a first shift get less than the shiftwork norm of 7 h of
nighttime sleep. The reason may be that 91 percent of first shift dispatchers begin work by
7 a.m., and 43 percent start by 6:30 a.m. As Figure 10 illustrates, over 20 percent of the
dispatcher commute trips to work begin before 6 a.m. This early start of the commute to work
requires arising between 5 and 5:30 a.m. Getting 7 h of sleep requires the first shift dispatcher to
go to sleep by 10 p.m., a bedtime which most find difficult to accommodate.
Table 19. Primary sleep by shift worked (h:min)
Median

Standard
Deviation

25th
Percentile

75th
Percentile

Shiftwork
Norm*

6:08

6:15

1:07

5:29

6:55

7:00

2

7:06

7:10

1:30

6:10

8:05

7:34

3

5:54

5:55

2:00

4:30

7:15

6:36

Shift
Worked

Mean

1

*Source: Pilcher et al., 2000.

Analysis of primary sleep on workdays by age group did not reveal any differences,
F(4, 438) = .14. This analysis used the 10-yr age groupings of Figure 3 (e.g., 20-29, 30-39).
Comparison of dispatcher sleep with results from surveys of other adult groups is difficult
because the dispatchers are a shiftwork population. The NSF 2002 Sleep in America Poll
provides data on nighttime sleep on workdays for U.S. adults. Because third shift dispatchers do
not have nighttime sleep, their sleep data is not comparable to the NSF data. Permanent first and
second shift dispatchers, however, sleep at night so their sleep data is comparable to the NSF
data. Figure 12 presents a frequency distribution of workday sleep for this group of dispatchers
in comparison with the results from the NSF survey. Approximately 59 percent of U.S.
permanent first and second shift dispatchers are getting less than 7 h sleep on work nights in
contrast with 39 percent of U.S. adults. The proportion getting less than 6 h sleep (25.4 percent)
is considerably larger than that for U.S. adults.
A total of 11 dispatchers averaged less than 5 h of sleep on workdays. For nine of these, nonwork reasons may have contributed to the lack of sleep. Seven reported multiple personal and
family factors or an illness, and two people reported a change in sleeping habits. For the
remaining two, no information existed in either the background survey or the daily log to explain
the lack of sleep.
The National Health Interview Survey (NHIS) provides another set of normative sleep data for
comparison with dispatchers. The Centers for Disease Control, U.S. Department of Health and
Human Services, conducts this annual survey to obtain information on a number of health-related
issues. It includes the question, “On average, how much sleep do you get in a 24-hour period?”
This data is slightly different than that from the NSF 2000 Sleep in America Poll. NSF
differentiates between work nights and weekends while the NHIS data do not. In addition, NHIS
employs a rounding scheme to convert all responses to a whole number of hours. The interview
protocol rounded responses of half-hours or more to the next whole hour so, for example, 6½ h
40

50

45

43.4

40

35

29

Percent

30
25.4
25

30
28.3
U.S. Adults

24

Dispatchers

20

15

15

10

5

2.9

0
<6

6–6.9

7–7.9

≥8

Hours

Source: U.S. adult data from NSF, 2002b.

Figure 12. Nighttime sleep on workdays for permanent first and
second shift dispatchers versus U.S. adults
was entered as 7 h. Applying the same rounding scheme to the dispatcher data made comparison
with the NHIS results possible (see Figure 13). The categories in this comparison differ from
those in Figure 12 due to the NHIS rounding. In addition, the dispatcher data in Figure 13 use
average total daily sleep across both workdays and non-workdays for the dispatcher population.
This comparison, similar to the comparison in Figure 12, shows that dispatchers are getting less
sleep than U.S. adults, but the difference is not as great as the comparison with NSF data.
Adults getting less than an average of 7 to 8 h of sleep per night are chronically sleep-deprived.
The primary manifestation of sleep loss is excessive daytime sleepiness, but depressed mood and
poor memory or concentration may also result. Chronic sleep loss has serious consequences for
job performance and health. Performance effects of sleep loss can affect the dispatcher’s ability
to work safety and efficiently. Belenky et al. (2003) have shown that performance declines
initially with mild to moderate sleep restriction of 7 and 5 h, and after a few days it stabilizes at a
less than fully rested level. The relevant performance effects include the following (Institute of
Medicine, 2006):
•

Involuntary microsleeps occur.

•

Attention to intensive performance is unstable, with increased errors of omission and
commission.

•

Cognitive slowing occurs in subject-paced tasks, while time pressure increases cognitive
errors.

•

Response time slows.

41

45

38.8

40

35
31.9
30

29.9

29.2

Percent

25.8
25
U.S. Adults
Dispatchers
20

18.6
16.9

15

8.9

10

5

0
≤6

7

8

≥9

Hours

Source: U.S. adult data from National Health Interview Survey, 2005.

Figure 13. Sleep in 24 h for dispatchers versus U.S. adults
•

Performance declines in short-term recall of working memory.

•

Performance requiring divergent thinking deteriorates.

The job performance of those dispatchers getting less than 7 h of sleep on workdays is likely
compromised. Recent research has revealed that sleep loss has neurobehavioral effects, such as
those listed above, that often go unrecognized by the affected individuals (Van Dongen,
Mullington, & Dinges, 2003). These sleep-deprived dispatchers are probably unaware of the
extent of their performance degradation and the increased risk of error.
3.5.2 Sleep Ratings
Dispatchers recorded subjective ratings for primary sleep on both workdays and non-workdays.
They rated their ease of falling asleep, ease of arising, length of sleep, quality of sleep, and
alertness upon arising. The ratings shown in Table 20 used a Likert scale ranging from 1 to 5,
with 1 being the lowest or worst rating, while 5 indicated the highest or best.
Overall, dispatchers gave higher ratings to their non-workday sleep than their workday sleep
(see lower right-hand section of Table 20). While all of the work schedule groups rated their
non-workday sleep of higher quality than their workday sleep, only the first shift dispatchers’
ratings were statistically different for workdays and non-workdays.

42

Table 20. Ratings of primary sleep (mean of means) by work schedule and type of day
First Shift

Ease of
Falling
Asleep
Ease of
Arising
Length of
Sleep
Quality of
Sleep
Alertness
Upon
Arising

Second Shift

Workday

NonWorkday

Significance Test

Workday

NonWorkday

3.67

4.09

t(215) = -4.60, p < .05

3.98

4.01

t(179) = -0.20, p = .85

3.18

3.61

t(215) = -4.12, p < .05

3.18

3.40

t(179) = -1.81, p = .07

3.06

3.75

t(215) = -7.36, p < .05

3.30

3.45

t(179) = -1.36, p = .18

3.31

3.80

t(215) = -5.51, p < .05

3.42

3.58

t(179) = -1.24, p = .22

3.25

3.71

t(215) = -4.64, p < .05

3.31

3.46

t(179) = -1.23, p = .22

Third Shift

Ease of
Falling
Asleep
Ease of
Arising
Length of
Sleep
Quality of
Sleep
Alertness
Upon
Arising

Relief

Workday

NonWorkday

Significance Test

Workday

NonWorkday

4.04

4.14

t(197) = -0.99, p = .32

3.86

4.14

t(161) = -2.51, p < .05

3.16

3.30

t(196) = -1.25, p = .21

3.19

3.08

t(161) = 0.83, p = .41

3.07

3.20

t(197) = -1.18, p = .24

3.23

3.05

t(161) = 1.52, p = .13

3.35

3.63

t(197) = -2.49, p < .05

3.46

3.42

t(161) = 0.30, p = .76

3.28

3.45

t(197) = -1.53, p = .13

3.27

3.24

t(161) = 0.26, p = .79

Extra Board

Ease of
Falling
Asleep
Ease of
Arising
Length of
Sleep
Quality of
Sleep
Alertness
Upon
Arising

Significance Test

Significance Test

All

Workday

NonWorkday

Significance Test

Workday

NonWorkday

3.62

3.80

t(105) = -1.18, p = .24

3.85

4.06

t(865) = -4.15, p < .05

2.97

3.11

t(105) = -0.90, p = .37

3.15

3.33

t(864) = -3.17, p < .05

3.14

3.41

t(105) = -1.83, p = .07

3.15

3.38

t(865) = -4.42, p < .05

3.32

3.30

t(105) = 0.15, p = .88

3.37

3.58

t(865) = -4.02, p < .05

3.21

3.38

t(105) = -1.33, p = .19

3.27

3.47

t(865) = -3.76, p < .05

43

Significance Test

Dispatchers’ qualitative ratings of their sleep differed based on the shift actually worked.
Table 21 contains these ratings. Second shift ratings were statistically different for all categories
except ease of arising. The results for length and quality of sleep and alertness upon arising are
consistent with the fact that second shift dispatchers have the longest primary sleep period. With
the exception of ease of falling asleep, ratings from first and third shift dispatchers are similar.
Correlations between length of primary sleep period and the four sleep period ratings were all
significant. The largest correlation (.53) was between the length of the primary sleep period and
the rating for adequacy of the length of the sleep period.
Table 21. Workday primary sleep period ratings by shift worked
Shift Worked
Sleep Characteristic

Correlation
with Primary
Sleep (r)

1

2

3

Ease of Falling Asleep

3.67

3.90

4.01

.05*

Ease of Arising

3.14

3.20

3.10

.25**

Length of Sleep

3.05

3.37

3.04

.53**

Quality of Sleep

3.32

3.46

3.34

.28**

Primary Sleep (h:min)

6:08

7:06

5:54

--

*Significant at p < .05, **Significant at p < .01

3.5.3 Sleep Quality and Sleep Disorders
A total of 41 dispatchers (9.3 percent) reported having a diagnosed sleep disorder. Twenty-six of
those dispatchers (63.4 percent) reported receiving treatment for their disorder. The remainder
with a diagnosed sleep disorder reported that their problem was untreated. (Two survey
participants did not answer the questions regarding sleep disorders.)
Analysis of the sleep ratings included a comparison across three groups: (1) the untreated sleep
disorder group (n = 15), (2) the treated sleep disorder group (n = 26), and (3) those with no
diagnosed sleep disorder or the normal group (n = 400). Table 22 contains the average sleep
ratings for the three groups. While the average ratings indicate that those in the untreated sleep
disorder group had lower ratings and a shorter primary sleep period, no statistically significant
differences by sleep disorder status existed (see Appendix B, Table B-5).
An analysis of sleep ratings for workdays and non-workdays combined produced similar results
with no statistically significant differences by sleep disorder status (see Appendix B, Table B-6).
The small number of dispatchers in the two sleep disorder categories makes achieving statistical
significance difficult.

44

Table 22. Sleep ratings and duration by sleep disorder status (workdays only)
Untreated Sleep
Disorder (n = 15)

Treated Sleep
Disorder (n = 26)

Normal (n = 400)

Ease of Falling Asleep

3.81

3.84

3.85

Ease of Arising

2.80

3.21

3.17

Length of Sleep

2.78

3.21

3.17

Quality of Sleep

2.99

3.39

3.39

Primary Sleep (h:min)

6:12

6:27

6:24

Sleep Characteristic

3.5.4 Sleep Latency
Sleep latency is the time from lights out until the beginning of sleep. Sleep experts consider the
adult norm to be 20 min or less. Table 23 displays the sleep latency results based on the
dispatchers’ sleep log entries. These results refer to the primary sleep period, not supplementary
sleep periods or naps.
Table 24 contains similar sleep latency data by shift actually worked.
Table 23. Sleep latency by type of day and job type (min)
Workdays

Non-Workdays

Job Type

Mean

Median

> 20 min

Mean

Median

> 20 min

Trick Dispatcher

23

15

35%

20

15

29%

Assistant Chief

23

15

34%

19

15

26%

Table 24. Sleep latency by shift worked (min)
Shift

Mean

Median

> 20 min

1

26

15

40%

2

24

15

35%

3

19

15

26%

Sleep latency does not differ based on job type. On workdays and non-workdays, mean and
median sleep latency for both groups is nearly the same. In addition, the majority of trick
dispatchers and assistant chiefs are falling asleep within the 20-min adult norm. On nonworkdays, both groups fall asleep slightly more quickly than on workdays. This is likely due to
the lack of work-related stress on non-workdays and the cumulative fatigue of the workweek.

45

Examination of the sleep latency data by shift actually worked reveals that those working third
shift fall asleep most rapidly. This is a statistically significant difference, F(2, 4385) = 27.99,
p < .05. Individuals are typically not accurate judges of how long it takes them to fall asleep, so
drawing conclusions from the dispatchers’ self-report data requires caution. Some dispatchers
who read in bed or watched TV may not have reported their true lights out time, so their sleep
latency would be overstated if this were the case.
3.5.5 Naps
Data from participants’ daily logs indicate that the frequency of napping is a function of the
individual’s work schedule (see Table 25). Those working permanent third shift napped most
frequently. This group averaged 4.9 naps in the study’s 2-week period. Second shift dispatchers
were the least likely to nap and averaged 1.6 naps during this period. Over half of second shift
dispatchers reported taking no naps during the study period while nearly a third of the third shift
dispatchers reported 8 or more naps in 2 weeks. Many third shift dispatchers employ a split
sleep strategy on workdays, which accounts for the high number of naps for this group.
Similarly, relief and extra board dispatchers will use this sleep strategy when they work third
shift. During the study, as reported in Table 10, relief dispatchers worked third shift on
27 percent of the reported relief work periods, and extra board dispatchers worked third shift
23 percent of the time.
Table 25. Number of naps in 2-week period by work schedule
Number of Naps
Work Schedule

Mean

Median

1

2.4

1

2

1.6

0

3

4.9

4

Relief

3.7

3

Extra Board

2.5

2

Dispatcher nap length varies by work schedule and type of day (see Table 26). First and second
shift dispatchers take longer naps on non-workdays than on workdays. They are most likely
compensating for loss of sleep during the workweek. Those working third, relief, and extra
board schedules most likely take longer naps on workdays because they employ a split sleep
strategy when working third shift. The differences in nap length by work schedule on nonworkdays were not statistically significant, F(4,135) = .68, p = .604. For workdays, however, the
differences were significant, F(4,283) = 16.39, p < .05. (Appendix B, Table B-7 contains the
Tukey honestly significant difference (HSD) post hoc test results.)

46

Table 26. Mean nap length by work schedule and type of day (h:min)
Type of Day
Work Schedule

Workday

Non-Workday

1

1:02

1:35

2

1:12

1:26

3

2:12

1:51

Relief

1:57

1:41

Extra Board

1:46

1:48

Nap length differed by shift worked, F(2, 1048) = 143.90, p < .05. Dispatchers working third
shift tended to nap twice as long as those working the other two shifts (see Table 27). These
differences are statistically significant (see post hoc test results in Appendix B, Table B-8).
Again, this is likely because of a split sleep strategy when working third shift.
Table 27. Nap length on workdays by shift worked (h:min)
Shift Worked

Mean

Median

25th
Percentile

75th
Percentile

1

1:05

0:58

0:30

1:20

2

1:10

1:00

0:35

1:30

3

2:13

2:00

1:20

3:00

Start time of naps differs depending upon the type of day. As shown in Figure 14, nearly half
(49 percent) of all first naps on workdays began between 4 and 8 p.m. First shift dispatchers
returning from work and third shift dispatchers taking a pre-work nap account for these naps.
The second most popular nap start time was 2 to 4 p.m., a time when first shift dispatchers were
likely to nap after returning home from work. Individuals napping during their commute to or
from work account for the limited number of early morning naps. (Some dispatchers reported
sleeping during their train ride or carpool trip.)
Naps on non-workdays occur at different times of the day than those taken on workdays. Naps
on non-workdays tend to start between 12 and 4 p.m. (see Figure 15). This is a time that
coincides with the afternoon nadir in the circadian cycle.
3.5.6 Sleep Characteristics Summary
Dispatchers working permanent first shift jobs get considerably less sleep on workdays than
those with other work schedules, but first shift dispatchers appear to make up for their loss of
sleep on non-workdays. In comparison with sleep norms for other shiftwork populations,
dispatchers averaged less primary daily sleep than the shiftwork norms for comparable groups.

47

30.0

25.3

24.3

25.0

20.0

Percent

17.7

15.0

9.9

9.4

10.0
6.0
5.0

3.5

2.7
0.8

.–
12

a.
m
.

p.
m
.
10
:0
1

p.
m

8:
01
–1
0

p.
m
6:

4:

01
–8

01
–6

01
–4

p.
m

.

.

.
p.
m

p.
m
.
2:

.–
12

12
:0
1–
2

p.
m
.

.
a.
m

10
:0
1

6:

4:

01
–8

01
–6

01
–4
2:

a.
m

a.
m

a.
m

.

.

.

.
a.
m
12
:0
1–
1

0.3

8:
01
–1
0

0.0

a.
m

0.1
0.0

Start time

Figure 14. Nap start times for workdays
35.0

30.0

30.0

25.1

Percent

25.0

20.0

15.0

13.4
10.5

10.0

8.5
5.3

4.9

5.0
1.6
0.0

12

1
1–
:0

0.0

0.0

.
m
a.
2

4
1–
:0

m
a.

0.8

0.0

.
–6
01
4:

m
a.

.
–8
01
6:

a.

.
m
8:

0
–1
01

.
m
a.

10

1
:0

.
m
a.

2
–1

p.

.
m
2
1–
:0
12

p.

m

.
–4
01
2:

p.

m

.
–6
01
4:

p.

m

.
–8
01
6:

Start time

Figure 15. Nap start times for non-workdays

48

p.

.
m
0
–1
01
8:

p.

m

.

1
:0
10

p.

m

12
.–

m
a.

.

The nighttime sleep of permanent first and second shift dispatchers is less than that of U.S.
adults. One quarter of these dispatchers get less than 6 h of nighttime sleep in contrast with
15 percent of U.S. adults. The majority of first shift dispatchers begin work by 7 a.m., making it
difficult to get adequate nighttime sleep.
Nap frequency is a function of the individual’s work schedule. Dispatchers working permanent
third shift averaged 4.9 naps during the 2-week period of the survey, while more than half of the
second shift dispatchers took no naps. Nap length also varied by shift worked, with third shift
dispatchers taking naps that were on average 2:13, nearly twice that of those who worked first or
second shift.
Sleep quality ratings were higher for non-workdays versus workdays. These differences were
statistically significant for only the permanent first shift dispatchers. Sleep quality ratings were
lower for those with untreated sleep disorders, but no statistically significant differences existed,
most likely due to the small number in the untreated sleep disorder category.
3.6

Alertness

This section presents the survey results with respect to self-assessments of alertness. The study
collected self-assessments of alertness in two ways. The background survey had questions about
overall alertness, and dispatchers recorded their self-assessments of alertness at different points
during the workday in their daily log. One section discusses alertness relative to job type,
another relative to work schedule, and the third relative to sleep disorders.
3.6.1 Job Type
Through questions on the background survey, dispatchers rated their overall alertness at work
and after work. Assistant chiefs generally reported being alert at work more often than trick
dispatchers and were less likely to be mentally drained after work (see Tables 28 and 29). These,
however, are not statistically significant differences, χ2(2, n = 443) = 1.89, p = .39 and χ2(2, n =
443) = 3.31, p = .19.
Table 28. Alertness at work by job type (percent)
Trick
Dispatcher

Assistant
Chief

7.8

6.9

Frequently

47.3

56.9

Occasionally

41.3

34.5

3.6

1.7

Alert at Work?
Always

Never

49

Table 29. Mentally drained after work by job type (percent)
Drained After
Work?

Trick
Dispatcher

Assistant
Chief

Always

17.1

8.6

Frequently

46.8

46.6

Occasionally

34.8

41.4

1.3

3.4

Never

3.6.2 Work Schedule
Examination of these responses by job schedule did, however, result in statistically significant
differences overall, χ2(8, n = 443) = 16.26, p < .05. Table 30 contains the responses to the
background survey questions regarding alertness at work by work schedule. Pairwise
comparison of the work schedule groups revealed that first and second shifts were statistically
different from third, and second differed from relief (see Appendix B, Table B-9, for the
statistical test results). Those dispatchers working a permanent first or second shift reported that
they were more frequently alert at work than those working third shift. Similarly, second shift
dispatchers reported that they were more frequently alert at work than those working a relief
schedule. Although the data in Table 30 indicate that the dispatchers working the extra board
report being alert at work less frequently than those working other work schedules, the small size
of this group relative to the other work schedule groups was most likely responsible for the lack
of statistical significance in comparisons with this group.
Table 30. Alertness at work by work schedule (percent)
Work Schedule
Alert at Work?

1

2

3

Relief

Extra
Board

Always

8.9

9.7

5.9

8.4

3.7

Frequently

55.4

58.1

39.6

41.0

46.3

Occasionally

33.0

30.1

47.5

47.0

50.0

2.4

2.2

6.9

3.6

0.0

Never

Table 31 summarizes the responses to the question, “How often do you feel mentally drained at
the end of your work period?” The responses from the five work schedule groups were not
statistically different, χ2(2, n = 442) = 9.42, p = .31.

50

Table 31. Mentally drained after work by work schedule (percent)
Work Schedule
Drained After
Work?

1

2

3

Relief

Extra
Board

7.1

9.7

5.9

3.6

1.9

Frequently

38.4

30.1

40.6

33.7

38.9

Occasionally

47.3

43.0

42.6

48.2

53.7

7.1

17.2

10.9

14.5

5.6

Always

Never

Data from dispatchers’ daily logs revealed some differences in alertness levels based on shift
worked. For all shifts, alertness peaked after the commute to work but then declined throughout
the day, as shown in Table 32. The decline was greatest for third shift and least for first shift.
These results are consistent with the time of day when the dispatcher from each shift arrives
home and how long the dispatcher has been awake. The first shift dispatcher arrives home midafternoon while the second shift dispatcher arrives home around midnight, a time when most
adults are ready for nighttime sleep. The third shift dispatcher arrives home after being awake
through the night and trying to counteract the body’s need for sleep.
Table 32. Alertness throughout the day by shift worked
Shift Worked
Time of Rating

1

2

3

Upon Awakening

3.20

3.36

3.24

After Commute to Work

3.59

3.80

3.65

Mid-Shift

3.51

3.43

2.97

After Arriving Home

3.05

2.69

2.26

Table 33 contains the results of a linear regression analysis to explain alertness as a function of
time on shift using the data in Table 32 . The β coefficients indicate the more rapid decline of
alertness over the work period on second and third shifts relative to first shift. An alternative
formulation for this model that incorporated shift worked as an independent category variable
did not have better predictive power than the three separate models.
Table 33. Linear regression results by shift for alertness and time on shift
Shift

β

Constant

R2

1

-.14

3.79

.86

2

-.28

4.14

.96

3

-.35

4.00

.99

51

3.6.3 Sleep Disorders
Dispatchers with untreated sleep disorders had lower alertness ratings throughout the workday
than those with treated or no sleep disorders (see Table 34). These results do not reveal a
statistically significant difference due to the small number with untreated sleep disorders (see
Appendix B, Table B-10). The sleep literature contains numerous studies documenting the
performance risks of untreated sleep disorders, in particular sleep apnea. A fatigue education
program for dispatchers should caution all dispatchers about these risks.
Table 34. Alertness and sleep disorders (workdays only)
Untreated Sleep
Disorder (n = 15)

Treated Sleep
Disorder (n = 26)

Normal
(n = 400)

Upon Awakening

3.03

3.25

3.28

After Commute to Work

3.48

3.51

3.70

Mid-Shift

2.95

3.31

3.33

After Arriving Home

2.35

2.64

2.70

Time of Rating

3.7

Textual Data

The daily log included two separate spaces for participants to record any comments regarding
their sleep and work periods each day. This section presents an overview of participants’
comments on their sleep and work experiences throughout the 2 weeks of the study.
Commenting on sleep or work experiences in the daily log book was not a requirement of
participation. Rather, the diaries provided an opportunity for dispatchers to qualify part of their
day. As such, some participants chose not to comment, while others commented frequently. For
this reason, a statistical analysis of these comments was not possible. Researchers, however,
scanned a number of participant log books to determine common themes presented in the
comments and performed a simple tabulation of the frequency of topics mentioned. The
following themes emerged from this review:
1. Sleep
2. Workload
3. Alertness/Fatigue
4. Naps
5. Work Schedule
6. Stress
7. Personal Issues (family, leisure activities)
8. Weather
9. Territory
10. Communication
52

11. Safety
12. Management
13. Responsibility
Researchers used ATLAS.ti® software V5.2.8 to autocode comments based on keyword
searches and tally the number of comments made under each topic area. Table 35 lists the
keywords that were the basis for each topic area search.
These categories were not mutually exclusive. As a result, many of the comments fell into more
than one category. For example, a comment on work schedule may also be counted in the stress
category, or a sleep comment may be in both the alertness/fatigue and naps categories. The most
frequently mentioned topics in the section for sleep comments were sleep, naps, personal issues,
and work schedule. Comments on workload, alertness/fatigue, stress, and sleep dominated the
work experience section. The comments complement the quantitative survey results by
providing personal examples of the effect of work or sleep patterns present in the survey data. In
this way, the comments provide a more complete picture of dispatchers’ fatigue-related concerns.
Table 36 presents more detailed results of keyword searches.
The selected comments that appear following Table 36 illustrate the consequences of the work
and sleep patterns in the survey data. For example:
•

Many of the comments related to Sleep touch upon the difficulty of sleeping during the
day for those working third shift, the challenge in adjusting or maintaining one’s sleep
schedule when working a variable schedule, and the general deficiency or poor quality of
sleep due to work schedules.

•

Comments on Workload related to heavy traffic volume and the large variety of
responsibilities a dispatcher must handle. Many of these comments linked workload to
the lack of breaks.

•

A number of participants commented on the difficulty of staying Alert on third shift.
Others (assistant chiefs) mention Fatigue due to long workdays. Many dispatchers
commented that being busy at work made it easier to stay alert.

•

The comments with regard to Naps illustrate how some survey participants, primarily
those working third shift, have split sleep patterns. Those individuals typically sleep in
the morning after work and then nap in the evening in preparation for their next night
shift.

•

Comments on Work Schedule concern primarily the difficulty in working relief or extra
board jobs and how those jobs affect one’s personal schedule, including sleep, as
mentioned above.

•

Dispatchers mentioned Stress in relation to a variety of issues, such as traffic volume,
workload, communication difficulty, weather issues, train delays, and dealing with train
and engine crews and track crews.

•

Some survey participants described Personal Issues that affected their sleep.

53

Table 35. Keywords used for each topic area search
Topic

Keywords

Alertness (mental)/Fatigue (physical)

Alert*, aware*, awake, attentive*, watchful,
vigilant, prepared, fatigue*, tired, sleepy,
exhaust*, weary, energy, weak*

Communication

Communicat*, dispatcher, track gang, track
crew, track department, signal department,
yardmaster, radio, communication
equipment

Management

Manage*, boss, supervis*, company, policy,
organization, administration, control, chief,
rule*, staff, inadequate

Naps

Nap*

Personal Issues (family, etc)

Personal, private, family, domestic, son,
daughter, wife, kid*, baby, father, mother,
grand*, relative*, child*, husband

Responsibility

Responsib*, duty, blame, reliab*,
accountab*

Safety

Safe*, accident, incident, injury, casualty,
error, protect*

Sleep

Sleep*, rest, bed

Stress

Stress*, pressure, strain, anx*, worr*, tense

Territory

Desk, territory, coverage, area, region

Weather

Weather, heat, hot, degrees, temperature,
cold, freezing, wet, rain, snow, sleet, light*,
dark, sun, ice, climate, condition*

Work Schedule

Schedule, overtime, unscheduled, shift, 3rd,
third, extra*, night shift, relief, break, hours,
lunch, double

Workload

Busy, work load, workload

Note: ATLAS.ti search logic uses the symbol “*” as a wildcard. For example searching for “stress*,” would result in all
words starting with s-t-r-e-s-s and would include any ending (such as stressful, stressor, etc.).

54

Table 36. Frequency of comments by topic area and source
Source of Comments
Topic

Work Log

Total

1877

298

2175

17

678

695

Alertness/Fatigue

109

393

502

Naps

432

36

468

Work Schedule

169

178

347

33

308

341

258

47

305

Territory

2

55

57

Communication

3

38

41

Safety

0

28

28

Weather

0

28

28

Management

9

8

17

Responsibility

2

9

11

2911

2104

5015

Sleep Log

Sleep
Workload

Stress
Personal Issues

Total

The limited comments relating to Management and Responsibility were not meaningful and are
therefore not presented below.
Selected comments by topic follow:
Sleep
•

“Sleeping during the day is not the same as sleeping at night. I’ve been on third shift for
8 years and I have never gotten used to it.”

•

“Being on a relief job you never have enough time to catch up on your rest because you
only have one and a half days off each week, and how much of that is really a day off?”

•

“Did not get enough sleep because worked second shift on Friday and first shift on
Saturday.” (assistant chief)

•

“Was not good sleep due to job called me at 0600 to be at work at 0630. I was scheduled
to work at 1430, so my sleep was planned on that schedule.” (extra board dispatcher)

•

“Woke up from 1200-1215. Even on off days I try to stay on same sleep schedule. If I
keep trying to change my schedule around, I will be more tired, grumpy, and have less
energy. Feel groggy all day.”

55

•

“My change of assignments also meant going from third shift to afternoons so my sleep is
now much better! It took about 4 months to get 6 straight hours.”

•

“It’s a day off and I slept more. The extra nap helped a lot. I went to bed at 2200 hours.
I use my days off to catch up on lost sleep so I don’t get much else done on my days off
except SLEEP.”

Workload
•

“Sunday was a little lighter workload, but trains, duties, responsibilities, rules increase
regularly. Eating, taking a break is next to impossible because the workload increases
continuously! We are also responsible for numerous weather alerts, warnings, and have
to constantly devote excess time to notifying every train and document data–feel like a
weather reporter.”

•

“Extremely busy. Multi-tasking non-stop for full shift. Very stressful. Had to rest in
parking lot before feeling fit to drive home.”

•

“Very busy day today–barely had a chance to eat or go to the bathroom. By 6 p.m. I was
really tired, physically and mentally. Each phone call began to be overwhelming halfway
through the shift. Days like this would be so much easier if you would get a 30 minute
break away from the desk.”

•

“Sunday night–Did not sleep very soundly. I think the anticipation of the coming day’s
events (workload) makes sleeping difficult.”

•

“Very busy. Break long enough to go to the restroom and get coffee. No time for lunch.
Signal trouble late in shift.”

•

“Because of enormous increase in traffic volumes, every day is extremely busy and
intense. Opportunities for breaks are few, if any, due to the constant demand for my
attention. The long, intense mental workout leaves me mentally drained and tired by end
of day. This usually leaves me needing a nap in the afternoon, which disturbs a normal
sleep at night and unrested in the morning.”

•

“Regarding ‘breaks’ during the work day: we do not have ‘break’ time built into the
work schedule. The only break I ever take is to use the restroom or get my lunch from
the refrigerator and return to my work site. I eat lunch while continuing my work. Time
passes really fast during the work day. Most days I’m mentally exhausted at the end of
the day.”

Alertness/Fatigue
•

“Typical day for the night shift. It’s like a pattern of not enough sleep and always tired.”

•

“I work days, sleep nights, get good rest. Things were different when I worked nights. I
was tired all the time.”

•

“Work was slow. When it’s that way, it’s hard to stay awake sometimes. Busier is
better.”

•

“Very good work day. A steady day where time went by quickly and felt rested and alert
all day.”

56

•

“Another very long night… Fighting to stay awake, drank 7 cups of coffee and took 3
‘NoDoz’ pills to keep going. So tired I started to invert numbers and forgot what I was
going to say to people.”

•

“Did not sleep. Marked off work for third shift starting tonight at 2230. Due to being
dangerously exhausted.”

•

“It seems that I just reached the point of total exhaustion. I need to try to go to bed at the
same time every night, but being on the extra board makes it hard to. The last 2 weeks
have been the exception as I had somewhat of a normal shift. Not always true!!!”

•

“I feel very alert when leaving work, but after 10 minutes in the car ‘winding down’ the
fatigue envelopes me.”

•

“I did not feel very alert all day, nodding off frequently at my desk. No matter how many
breaks I took, I couldn’t seem to fully concentrate on my work because I was so tired.
I’m looking forward to tomorrow, a rest day, to catch up on some much needed sleep. At
this point of the week, I’m burned out.”

•

“3rd day to work 12 hour shifts and very tired!”

Naps
•

“Generally when my shift changes from evening shift to night shift, I split my sleep. So
I will mostly take a nap for abut 3 or 4 more hours of sleep.”

•

“Took a short nap. Didn’t want to waste any of my day and a half off.”

•

“Had to take nap before work, may have to double.”

•

“Short nap to relieve stress–felt better after nap.”

•

“Took a nap–trying to stay rested in case called for third shift. Did not get the call for
third until 11:40 p.m. for an 11 p.m. shift. Had just gone to bed and dozed off.”

•

“Decent night’s sleep but will spend the remainder of the day dreading third shift tonight.
Short nap prior to work I find to be very helpful, makes the night at least tolerable.”

•

“Slept extra today–coming off rest day and going to work third–extra sleep so I don’t
have to take a nap.”

Work Schedule
•

“Had to work a double shift last night. By the time I got home had only one thing on my
mind. Go to bed. Was worried about getting into a car accident on way home I was so
tired.”

•

“The hardest part of the job–working this swing shift–is staying alert on third shift. The
workload is OK, but I’m never able to get enough rest to work effectively.”

•

“Was told I was working third shift. Went to bed and was called to take a rest day and
come first shift on the 5th. Yo-yo sleeping messes me up. ‘Circadian rhythm out of
whack.’”

•

“Making the transition to third shift one night a week is very difficult. The next day, ‘my
so called day off,’ I am pretty much exhausted because if I sleep too long when I get
57

home, I won’t sleep that night. Have to get back on a regular schedule. Overall work
day OK but I feel like I gradually go downhill.”
•

“Always seem to have trouble falling asleep after day off. Originally supposed to work
second trick today, but when we got home yesterday evening found a message saying I’m
working first trick, which required me to rush and get all my stuff ready and my lunch, so
instead of winding down before sleep time, I go to sleep all keyed up. I really believe
when they call you and change your schedule on your rest day, extra compensation
should be given.”

•

“On call–don’t know if I’ll work. Sleeping in case they call me for third shift. They did
not call.”

•

“Very tired all night. Tough to go from day schedule to night schedule. Also slow night
for trains, doesn’t help at all.”

•

“Going from second to third shift is tough. I don’t always get a chance to take a nap so I
am usually very tired by the end of my shift.”

•

“People on third shift should be allowed short naps if needed. I get very tired at night for
no reason.”

Stress
•

“Busy day. Student present today, adding to stress. Lots of impatient T&E and MOW
employees making comments about their delays for lack of being able to get in contact
with the dispatcher. This in turn causes more stress on the dispatch. Stayed over 45
minutes to help smooth the operation for the shift transfer.”

•

“Many problems at work due to weather problems. Being tired and having a busy work
shift makes for a very long night. Fatigue intensifies on rough work shift. Stress makes
things worse.”

•

“More stress today than usual. Trains close to Hours of Service Law. Plans made by
trainmasters, and dispatcher not made aware of plans.”

•

“Lately work has been really stressful. You have people constantly giving you a hard
time. They call and call on the radio despite the fact that they can hear that you are
talking to someone else. THEY mess up on something and you somehow get dragged
into it...It is like babysitting 60 people all at once who should know how to act like adults
but don’t.”

•

“Total chaos. Train traffic heavy and late, MOW people anxious to get time on track.
Too much to be done by one dispatcher safely and efficiently by the rules and
instructions. I chose the safe route.”

•

“Medium traffic. Steady work. Many communication and computer issues. Crew in
emergency did not notify me, had trains going by–unprotected. I was stressed…”

Personal Issues
•

“Wanted to sleep more but have to work this afternoon and I’ll take time away from
family if I do.”

58

•

“On every other Saturday and Sunday I have only a 2-hour (plus or minus) window to
sleep. I’m divorced and have split custody of my two children. The kids are brought
here around 10:00 and I bring them to their mother’s at 20:20.”

•

“Difficult falling asleep–daughter has strep and worried about her–wife with difficulties
at her work.”

•

“Mother’s Day–spent time with grandchildren and cut back on sleep. I’m used to being
tired.”

•

“Due to having worked third shift the night before, I only sleep a few hours so I will be
able to fall asleep at a normal time tonight and spend time with my children during the
day.”

Territory
•

“I work a very busy desk that keeps me alert throughout the shift.”

•

“Another stressful day. Rank 4 out of 5. Too much territory to control safely by one
dispatcher.”

•

“I was shifted off my regular desk. I worked a much busier desk. By 12:30 p.m. I was
bushed. By 1:30 p.m. I was completely tired and ready for a nap. 1½ hours still left.”

Communication
•

“Trains in emergency. Ongoing radio and CADS problems. Crossing failures–one radio
tower has been down for weeks. It makes communicating with certain areas nearly
impossible. Intermittent radio outages randomly occur on all other towers and constant
repeating. The transmissions do not seem to be going out, as well as the fact the radio
quality [is poor].”

•

“The radio problems are ridiculous. In my opinion, they are unsafe. It’s not just the
radios. It’s the noise, speakers, air box, etc… When I try to talk to another dispatcher on
the intercom system there is a lot of feedback, same with my open lines to yards. The
computer system has so many glitches that it is causing more work than it saves.”

Safety
•

“0825–rear end collision between two loaded coal trains. Total chaos. Crewmembers
injured. Millions of dollars damage. Main lines blocked. The kind of shift no dispatcher
ever wants, but must be ready for.”

•

“Full throttle. Many trains and the trauma of a crossing incident. I am so fatigued
mentally and physically.”

Weather
•

“Another hard day. Lots of weather-related problems and account maintenance work lots
of trains went dead due to hours of service and our responsibilities, including lining up
transportation as well. Life of a dispatcher is constantly changing.”

•

“Stress with weather conditions to protect. Wind in excess of 50 mph and flash floods
until 0200.”

59

•

“Blizzard conditions–stressful, switches, etc. Won’t lock up. Vans delayed. Trains dead
on hours of service, etc.”

•

“Tornado warnings. Heavy rain. Signal problems. Very stressful.”

60

4. Findings and Recommendations
Analysis of the data from this study provides some insights into the demographics of the U.S.
dispatcher population, as well as how their work schedules and sleep patterns affect their
alertness on the job. The data came from a random sample of U.S. dispatchers who are members
of ATDA. Nevertheless, the results are representative of the Nation’s dispatcher population at
the time of the survey for two reasons. First, approximately two-thirds of the Nation’s
dispatchers are ATDA members. In addition, it is a reasonable assumption that the work
schedules and sleep patterns of dispatchers in non-union dispatching jobs and at railroads with
different labor representation are similar to those of the ATDA dispatchers because of the work
hour limitations of the Hours of Service Law.
This section presents the key findings of the study, as well as some recommendations for
methodological changes for future field studies of this nature. The section concludes with some
suggestions for additional research using the survey data from this study.
4.1

Key Study Findings

The following subsections highlight the study’s key findings with respect to dispatcher
demographics, work periods, and sleep patterns and alertness.
4.1.1 Dispatcher Health
Dispatchers average 5.6 workdays lost annually due to illness. In contrast, U.S. employed adults
with paid sick time average 3.6 d of sick time. Several possible explanations exist for the more
frequent usage of sick days among dispatchers. First, this is a shiftwork population, and
shiftworkers tend to have a higher rate of medical conditions, such as gastrointestinal problems
and headaches. The stressful nature of the job may also lead to medical problems. Finally, some
dispatchers, especially those working third shift or irregular schedules, may find it necessary to
mark off sick to catch up on sleep.
Due to the around-the-clock nature of railroading, dispatchers will always be shiftworkers. If
dispatchers understand how the body responds to irregular work schedules and working at night,
and if they have strategies for coping with shiftwork, they will be better able to manage this work
arrangement and to minimize its health consequences. An initial dispatcher training program
should include education on shiftwork, as well as the potential health consequences of
inadequate sleep. For dispatchers who have had no shiftwork or fatigue training, annual rules
training might provide the opportunity to cover this information.
4.1.2 Work Periods
Both trick dispatchers and assistant chief dispatchers have a nominal workweek of 40 h, but
assistant chiefs, on average, work more hours per week. Since the Hours of Service Law does
not apply to assistant chiefs, on occasion they will work 12 or even 16 h in 1 d. Based on the
survey data, the primary reason that the assistant chiefs worked more hours than the trick
dispatchers was that they worked an extra workday. A quarter of the assistant chiefs averaged

61

1 d or more of overtime per week during the study. A regular work pattern like this with only 1
rest day does not allow adequate time for rest and recovery from the workweek.
The 8-h workday of a dispatcher plus commute time should provide adequate time for sleep and
personal activities, but many aspects of their work schedules may prevent this. First shift for
over 90 percent of the dispatchers begins by 7 a.m., necessitating an early wake-up time that
compromises nighttime sleep. Third shift, relief, and extra board dispatchers must contend with
varying and unpredictable work schedules, as well as sleeping at times when their bodies say
they should be awake. For assistant chiefs, the situation has the added complication that no
limitation exists on the number of hours that they may work in a day; thus two shifts in 1 d and
backward shift rotation may occur.
Dispatchers working relief and extra board jobs experience variability in their shift start times.
Although the survey indicated that relief dispatchers have more shift variability than extra board
dispatchers, relief dispatchers work the same schedule each week so at least the variability is
predictable. In contrast, an extra board dispatcher who is not filling in for a regular dispatcher on
vacation or other extended leave has no predictability in his/her schedule from week to week.
Because of the way that this analysis computed shift variability, the extra board dispatcher’s
variability may be underestimated.
Since dispatchers do not have any planned or guaranteed breaks during their 8-h shift, the
researchers were surprised to find that lack of breaks was not a major source of stress to either
trick dispatchers or assistant chiefs. Perhaps dispatchers have become accustomed to their few
short breaks and the need to eat meals at their desks.
4.1.3 Sleep Patterns and Alertness
Overall, dispatchers are a sleep-deprived group. They get less sleep than other shiftwork
populations and less sleep than the norm for U.S. adults. Only dispatchers working first shift
appear to be making up for their poor workday sleep on rest days. When working a third shift,
many dispatchers employ a split sleep strategy, which is common in other shiftwork populations
as well. The dispatchers’ overall lack of sleep is a concern since research has shown that
performance decrements occur with less than 7 h sleep, particularly if it is consistently at this
level. Even more disconcerting is that nearly 40 percent of dispatchers averaged 6 h or less sleep
overall (for both workdays and non-workdays), and these individuals, who perform safety critical
jobs, are probably unaware of the extent of their performance degradation and the potential
health consequences of lack of sleep.
One of the dispatcher daily log books included the comment that participating in the survey
made this dispatcher aware of the extent of the dispatcher’s lack of sleep. The same entry
included the comment that as a result of the survey, this dispatcher planned to make an effort to
get more sleep. This one experience illustrates that fatigue education would likely make
dispatchers aware of their inadequate sleep, hopefully leading to some changes in sleep patterns.
The incidence of sleep apnea among U.S. dispatchers exceeds the U.S. adult norm. This may
reflect an awareness of the symptoms of sleep apnea and its consequences. A third of those with
sleep apnea have gone without treatment. Railroads and unions should continue their education
programs pointing out the possible performance and health consequences of untreated sleep
apnea and encouraging those with sleep apnea to accept treatment. Initial dispatcher training and

62

periodic rules training provide the opportunity for fatigue education that includes discussion of
sleep disorders.
Dispatchers’ self-ratings of alertness throughout the workday increased after the commute to
work and then declined from mid-shift. This pattern was the same regardless of shift worked,
but the greatest decline occurred with those working third shift.
4.2

Recommendations for Improvements in Study Procedures

This was the third FRA-sponsored study of work schedules and sleep patterns of railroad
workers. The experiences of each successive study have resulted in improvements in the
subsequent one. Based on the experience of this study, several methodological improvements
should be a part of any future studies to collect work schedule and sleep pattern data. Because
this was the first survey of a railroad shiftwork population, the first two of the four
recommendations below focus on issues unique to a shiftwork population. The recommended
changes are the following:
•

Design and test instructions for third shift workers to use in recording their primary sleep
period. Despite having instructions specifically for third shift dispatchers, some third
shift dispatchers did not record their sleep on the correct day. Several things can
potentially solve this problem. First, the instructions require re-evaluation. In addition,
the pilot test of the survey instruments for another shiftwork population could include a
hypothetical third-shift day to record. Examination of the pilot test participants’
responses to this hypothetical example is a means to evaluate whether or not the revised
instructions are effective. Finally, if at least half of the participants in a future pilot test
are third shift workers or workers on a relief schedule that includes third shift, then
adequate evaluation of these instructions will occur. (OMB regulations limit the pilot test
to a total of nine participants.)

•

Separate the question on job type (e.g., trick dispatcher, assistant chief, other) from the
one covering work schedule. Question 5 on the background survey asked “What type of
dispatcher job do you currently work?” Possible responses were the following: regular,
relief, extra board, assistant chief, other. These categories are not mutually exclusive.
For example, a dispatcher could work a regular schedule as an assistant chief. This
question should have had the following three responses: trick dispatcher, assistant chief,
other. A separate question, under “Your Work Schedule” should have asked, “What type
of work schedule do you work?” with the following possible responses: regular first
shift, regular second shift, regular third shift, relief, extra board. In spite of this poorly
worded question, the researchers were able to effectively categorize each survey
respondent’s job type and work schedule using the information in the background survey.

•

Include a definition of break and guidance on recording break information. The survey
instructions did not define a break and did not provide any guidance on recording breaks.
Based on log book comments, some dispatchers did not record any breaks because they
do not get official planned breaks. The intent was to capture any opportunity for the
dispatcher to leave his/her desk. The instructions should also specify to record 0 if the
dispatcher took no breaks during the shift.

63

•

4.3

Include a question on the background survey that asks, “How many times have you been
married?” Both shiftwork and railroad careers can lead to marital and family difficulties.
Railroaders frequently report instances of divorce due to the life style of a railroader. For
this reason, an additional question on the number of marriages would be helpful.
Recommendations for Additional Research

The researchers have developed an effective procedure and methodology for characterizing the
work schedules and sleep patterns of railroad workers. Each of the three populations surveyed to
date have had some differences, but the researchers used the same basic approach with the
signalmen, MOW workers, and dispatchers. Additional studies of other railroad populations
using this methodology would provide a more complete picture of work schedule and fatigue
issues in the industry. Yardmasters, a group not subject to the limitations of the Hours of Service
Law, have job responsibilities similar to a dispatcher and would be a candidate group for such a
survey. Pollard (1996) collected data on sleep and work patterns of locomotive engineers using a
sample of convenience rather than a random sample. Another survey of engineers would provide
a more systematic assessment of these groups. Conductors, who have work schedules similar to
those of locomotive engineers, might be surveyed concurrently.
A number of biomathematical models exist for predicting human fatigue and alertness. FRA has
sponsored a number of studies to improve and apply the Sleep, Activity, Fatigue, and Task
Effectiveness model (SAFTE) to locomotive engineers (Hursh et al., 2004). SAFTE model
developers could use the data from the dispatcher survey, as well as the data from the two earlier
studies, to further refine this model and predict how the typical railroad worker schedule may be
affecting on-the-job alertness. SAFTE also provide a means to examine the potential of
alternative work schedules to reduce fatigue and risk of an error.
Further analysis of the dispatcher survey data, as well as that from the signalmen and MOW
workers, is possible. Additional analyses might examine the relative likelihood (i.e., odds ratio)
of getting less than a minimum amount of sleep, such as 6 h, as a function of shift worked or
work period start time and length of the work period. A similar analysis could estimate the
likelihood of a workday nap as a function of shift worked. Comparisons across the different
craft groups are possible to determine if differences exist between groups covered by the Hours
of Service Law and those that are not.

64

5. References
Barnes-Farrell, J., & Piotrowski, M. (1989). Workers’ perceptions of discrepancies between
chronological age and personal age: you’re only as old as you feel. Psychology and Aging,
4, 376-377.
Barnes-Farrell, J., & Piotrowski, M. (1991). Discrepancies between chronological age and
personal age as a reflection of unrelieved worker stress. Work & Stress, 5, 177-187.
Belenky, G., Wesensten, N.J., Thorne, D.R., Thomas, M.L., Sing, H.C., Redmond, D.P., Russo,
M.B., & Balkin, T.J. (2003). Patterns of performance degradation and restoration during
sleep restriction and subsequent recovery: a sleep dose-response study. Journal of Sleep
Research, 12, 1-12.
Cohen, J. (1988). Statistical Power Analysis for the Behavior Sciences. Hillsdale, NJ:
Lawrence Erlbaum Associates, Inc.
Gertler, J., & Viale, A. (2006). Work Schedules and Sleep Patterns of Railroad Signalmen.
(DOT/FRA/ORD-06/16). Washington, DC: Federal Railroad Administration.
Gertler, J., & Viale, A. (2006). Work Schedules and Sleep Patterns of Railroad Maintenance of
Way Workers. (DOT/FRA/ORD-06/25). Washington, DC: Federal Railroad
Administration.
Hursh, S.F., Redmond, D.P., Johnson, M.L., Thorns, D.R., Belenky, G., Balkin,
T. J., Storm, W.F., Miller, J.C., & Eddy, D.R. (2004). Fatigue models for applied research
in warfighting. Aviation Space and Environmental Medicine, 75, 3, Supplement, A44-53.
Institute of Medicine. (2006). Sleep Disorders and Sleep Deprivation. Washington, DC:
National Academies Press.
Levy, P., & Lemeshow, S. (1999). Sampling of Populations: Methods and Applications. New
York: John Wiley & Sons, Inc.
National Sleep Foundation. (2002). Sleep Problems? Watch The Caffeine, Says The National
Sleep Foundation. Retrieved December 30, 2004, from
http://www.sleepfoundation.org/PressArchives/caffcalc.cfm.
National Sleep Foundation. (2002). 2002 “Sleep in America” Poll. Washington, DC: author.
Pilcher, J., Lambert, B.J., & Huffcutt, A.I. (2000). Differential effects of permanent and rotating
shifts on self-report sleep length: a meta-analytic review. Sleep, 23(2), 155-163.
Pollard, J. (1996). Locomotive Engineer’s Activity Diary. (DOT/FRA/RRP-96/02).
Washington, DC: Federal Railroad Administration.
Popkin, S., Gertler, J., & Reinach, S. (2001). A Preliminary Examination of Railroad
Dispatcher Workload, Stress, and Fatigue. (DOT/FRA/ORD-01/08). Washington, DC:
Federal Railroad Administration.
U.S. Census Bureau. (2005). Marital Status of People 15 Years and Over, by Age, Sex,
Personal Earnings, Race, and Hispanic Origin. Retrieved October 31, 2006, from
http://www.census.gov/population/socdemo/hh-fam/cps2005/tabA1-all.pdf.
65

U.S. Census Bureau. (2004). Journey to Work: 2000. Retrieved August 25, 2006, from
http://www.census.gov/prod/2004pubs/c2kbr-33.pdf.
U.S. Department of Health and Human Services. (2006). Summary Health Statistics for U.S.
Adults: National Health Interview Survey, 2004. Retrieved December 28, 2006, from
http://www.cdc.gov/nchs/data/series/sr_10/sr10_228.pdf.
Van Cauter, E., Leproult, R., & Plat, L. (2000). Age-related changes in slow wave sleep and
REM sleep and relationship with growth hormone and cortisol levels in healthy men. JAMA,
284(7), 861-867.
Van Dongen, H., Maislin, G., Mullington, J., & Dinges, D. (2003). The cumulative costs of
additional wakefulness: dose-response effects on neurobehavioral functions and sleep
physiology from chronic sleep restriction and total sleep deprivation. SLEEP, 26, 117-126.
Webb, W.B. (1992). Sleep, the Gentle Tyrant. Bolton, MA: Anker Publishing.
Young, T., Palta, M., Dempsey, J., Skatrud, J., Weber, S., & Bare, S. (1993). The occurrence of
sleep-disordered breathing among middle-aged adults. New England Journal of Medicine,
328, 1230-5.

66

Appendix A.
Survey Materials
This appendix contains copies of the following survey materials:
•

Cover letter to ADTA members from ADTA President

•

Instructions to participants on making entries in the Daily Log

•

Dispatcher Background Survey

•

Dispatcher Daily Log (1 day)

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

Appendix B.
Supporting Data
This appendix contains detailed data summaries that support the technical analysis in the main
sections of this report.
Table B-1. Tukey HSD post hoc test results for health status and work schedule
Work
Schedule

2

3

Relief

Extra
Board

1

p = 1.0

p = .268

p = .988

p = .533

2

--

p = .324

p = .994

p = .555

3

--

--

p = .645

p < .05

Relief

--

--

--

p = .337

Table B-2. Tukey HSD post hoc test results for age and work schedule
Work
Schedule

Relief

Extra
Board

2

3

1

p < .05

p < .05

p < .05

p < .05

2

--

p = .957

p = .989

p = .084

3

--

--

p = 1.000

p = .275

Relief

--

--

--

p = .234

Table B-3. Tukey HSD post hoc test results for years as dispatcher and work schedule
Work
Schedule

3

1

p < .05

p < .05

p < .05

p < .05

2

--

p = .985

p = .889

p < .05

3

--

--

p = .993

p < .05

Relief

--

--

--

p < .05

85

Relief

Extra
Board

2

Table B-4. Significant Tukey HSD ANOVA and post hoc test results for
sources of stress and work schedule
Work Schedule

3

Relief

Extra Board

Lack of control over work schedule
F(4, 435) = 19.07, p < .05
p = .0

1

p = .0

2

p = .0

3

p = .0

Relief

p = .0
Loss of sleep
F(4, 437) = 4.54, p < .05

1

p = .042

2

p = .027

p = .039

Coordination with other departments
F(4, 436) = 6.08, p < .05
p = .016

1

p = .0

2

p = .018

3

p = .013
Pressure to finish a task
F(4, 437) = 4.88, p < .05
p = .041

1

p = .001
p = .014

2

Ambiguous operating rules or procedures
F(4, 437) = 2.94, p < .05
1

p = .016

2

p = .033
Job security
F(4, 437) = 3.40, p < .05

1

p = .017

p = .030

Communication problems
F(4, 437) = 4.04, p < .05
1

p = .004

2

p = .011

Relief

p = .047

86

Work Schedule

3

Relief

Extra Board

Inadequate time off
F(4, 437) = 5.14, p < .05
p = .014

1

p = .017

p = .002

Table B-5. Significance tests for sleep ratings by sleep disorder status (workdays)
Sleep Characteristic

ANOVA

Ease of Falling Asleep

F(2, 438) = .03, p = .967

Ease of Arising

F(2, 438) = 1.71, p = .181

Length of Sleep

F(2, 438) = 2.38, p = .094

Quality of Sleep

F(2, 438) = 2.23, p = .109

Table B-6. Significance tests for sleep ratings by sleep disorder status (all days)
Sleep Characteristic

ANOVA

Ease of Falling Asleep

F(2, 438) = .03, p = .968

Ease of Arising

F(2, 438) = .62, p = .538

Length of Sleep

F(2, 438) = .88, p = .415

Quality of Sleep

F(2, 438) = 1.24, p = .290

Table B-7. Tukey HSD post hoc test results for nap length on workdays by work schedule
Work
Schedule

2

3

Relief

Extra
Board

1

p = .931

p < .05

p < .05

p < .05

2

--

p < .05

p < .05

p = .113

3

--

--

p = .489

p = .132

Relief

--

--

--

p = .869

87

Table B-8. Tukey HSD post hoc test results for nap length by shift worked
Shift
Worked

2

3

1

p = .770

p < .05

2

--

p < .05

Table B-9. Significance tests for alertness ratings by work schedule
Work
Schedule

2

3

Relief

Extra Board

1

χ2(2, n = 205) = .27,
p = .87

χ2(2, n = 213) = 7.57,
p < .05

χ2(2, n = 195) = 4.53,
p = .10

χ2(2, n = 166) = 3.79,
p = .15

2

--

χ2(2, n = 194) = 9.72,
p < .05

χ2(2, n = 176) = 6.25,
p < .05

χ2(2, n = 147) = 5.28,
p = .07

3

--

--

χ2(2, n = 184) = .55,
p = .76

χ2(2, n = 155) = .85,
p = .65

Relief

--

--

--

χ2(2, n = 137) = 1.33,
p = .51

Table B-10. Significance tests for alertness ratings by sleep disorder status (workdays only)
Time of Rating

ANOVA

Upon Awakening

F(2, 438) = .82, p = .440

After Commute to Work

F(2, 438) = 1.42, p = .243

After Lunch

F(2, 438) = 1.79, p = .169

After Arriving Home

F(2, 438) = 1.38, p = .253

88

Appendix C.
Adjustments to the Data
Population Means versus Mean of Individual Means
Some analyses of the daily log data calculated a mean for each survey participant and then the
analysis used the individual means. Analysis of the following measures used this method: actual
hours worked (for 2 weeks), primary and total sleep by type of day and job schedule, sleep
latency by job type and by type of day, and all sleep disorder comparisons. All other analysis of
the data from the daily logs pooled the data from all participants without first computing a mean
for each individual. For example, the analysis of commute time used the mean of the data for all
workdays in the survey data. The researchers used this latter approach where it was desirable to
characterize a typical day rather than the individual dispatcher’s experience.
Relief Jobs
The schedule for a relief job may require the dispatcher to work different days and/or shifts.
Two survey participants held relief jobs that worked the same shift each day but did not work the
same desk each day. Since the focus of this study was on work schedules and sleep patterns, the
analysis of the survey data treated these two relief jobs as straight shift jobs.
Type of Work Schedule
A few dispatchers held jobs that consisted of some days on a specific shift and the remainder on
the extra board. For these jobs, the category for the majority of days determined the type of
work schedule for that dispatcher. For example, if the dispatcher worked 3 d on second shift and
2 d on the extra board, then the job was a second shift job for purposes of analysis by work
schedule.
Break Time
Since this study sought to determine the extent to which dispatchers are able to take breaks away
from their desk, the analysis of break time did not include breaks that occurred while a
dispatcher was in a training class, rode a train for territory familiarization, or was in a meeting.
Comments in the log books made it possible to identify these situations.

89

Abbreviations and Acronyms
ACRE

Association of Commuter Rail Employees

ANOVA

analysis of variance

ATDA

American Train Dispatchers Association

d

day

FRA

Federal Railroad Administration

h

hour

HSD

honestly significant difference

in

inch

min

minute

mo

month

MOW

maintenance of way

n/a

not available

NARAP

North American Rail Alertness Partnership

NHIS

National Health Interview Survey

NSF

National Sleep Foundation

OMB

Office of Management and Budget

SAFTE

Sleep, Activity, Fatigue, and Task Effectiveness model

TCU

Transportation Communications International Union

yr

year

91