Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Journal of Applied Psychology

1991, Vol.76, No. 5. 644-657

Copyright 199! by the American Psychological Association, Inc


0021-9010/91/J3,00

Time Urgency: The Construct and Its Measurement


Frank J. Landy and Haleh Rastegary
Center for Applied Behavioral Sciences
Pennsylvania State University

Julian Thayer and Caran Colvin


Pennsylvania State University
Time urgency may be an important construct in industrial and organizational psychology. Preliminary analyses have indicated that time urgency may be multidimensional, and available self-report
measures have been criticized on psychometric grounds. The present research addressed the dimensionality of time urgency. Behaviorally anchored rating scales (BARS), in which behavioral
statements are substituted for qualitative anchors, were used to construct time urgency measures.
The BARS technique produced multidimensional measures of time urgency that possessed adequate reliability and construct validity The scales were tested on a wide variety of subjects. It is
concluded that time urgency is a multidimensional construct. Relationships between the time
urgency measures, job satisfaction, and work stress are discussed in light of previous research
findings.

Time urgency and time orientation have been topics of interest in industrial and organizational psychology for some time.
For example, one of the first industrial psychologists, Hugo
Munsterberg (1913), conducted research on the manner in
which people keep track of and utilize time in applied settings.
More recent treatments of time orientation in organizations
have been provided by McGrath and Rotchford (1983),
Schriber and Gutek (1987), and Edwards, Baglioni, and Cooper
{1990). Several forces have driven this interest. One is the potential relationship between the approach to time as a valuable
resource and the experience of strain at work. The notion is that
some people are more concerned about time and its passage
than others, and, as a result, may be more prone to suffer the
physical and psychological symptoms associated with strain
when time demands are high.
In fact, a great deal of research in clinical and organizational
psychology has been devoted to illustrating links between behavior patterns in work and leisure, on the one hand, and coronary heart disease, on the other. A pervasive construct in this
research has been the Type A behavior pattern (TABP), a pattern
defined partially by a heightened sense of time urgency.
In addition to the concern for stress and resulting strain,
there is reason to be interested in the extent to which organiza-

tions and individuals match with respect to time orientations.


As an example, one might suspect that if a time-urgent occupation (e.g. emergency room nursing) is filled by an individual not
particularly sensitive to the passage of time, problems in motivation, satisfaction, and performance will result. In fart, several
studies have demonstrated that time availability is implicated
in dissatisfaction and burnout in the nursing profession (Dewe,
1988; Maslach & Jackson, 1982,1984; Moses & Rothe, 1979).
With respect to the position of time urgency in the industrial
stress paradigm, the approach that has received the most attention over the years has been that of M. Friedman and Rosenman (1974). On the basis of observations of cardiac patients in
their private practice during the 195O's, M. Friedman and Rosenman identified a behavioral pattern that appeared related to
cardiovascular disease. They labeled this syndrome the Type A
behavior pattern and defined it as
a characteristic action-emotion complex which is exhibited by
those individuals who are engaged in a relatively chronic struggle
to obtain an unlimited number of poorly defined things from their
environment in the shortest period of time and, if necessary,
against the opposing effects of other things or persons in this same
environment. (M. Friedman & Rosenman, 1974, p. 67)
In contrast, the Type B behavior pattern was characterized as a
much more relaxed and noncompetitive approach to life, a polar opposite of the Type A pattern. Thus, time use and time
urgency were proposed as basic elements of this problematic
life style (Rosenman, 1978).
Since the early work of M. Friedman and Rosenman, many
researchers have sought to distill and define the element(s) in
the Type A pattern that are the "active ingredients" leading to
heart disease. This research took two directions. The first was
the traditional route of the differential psychologist. Self-report
scales were developed that would permit the collection of data
from a large number of subjects in short periods of time. Examples of these scales are the Jenkins Activity Survey (JAS; Jen-

We would like to thank Bruce Friedman for suggestions regarding


study design, data collection, and analysis and Stacey Kohler for identifying useful stress literature, in addition, the following individuals
helped in the development of research materials and data collection:
LeVonne LeFever, Rebecca Jones, Michelle Ludwig, and Steve Popkin,
We would also like to thank five anonymous reviewers for the comments on earlier versions of this article,
Caran Colvin is now at San Francisco State University
Correspondence concerning this article should be addressed to
Frank J. Landy. Center for Applied Behavioral Sciences, Pennsylvania
State University, 207 Research Building D, University Park, Pennsylvania 16802.
644

645

TIME URGENCY
kins. Zyzanski, & Rosenman, 1971) and the Framingham scales
(Haynes, Levine, Scotch, Feinleib, & Kannel, 1978). The JAS
has been the most widely used of these devices (Wright, 1988).
The second measurement method was more clinical in nature.
It became known as the Type A Structured Interview or SI
(Rosenman, 1978) and assessed behavioral characteristics apparent to an interviewer, characteristics such as speech patterns
and postural or gestural cues. Because of the time-consumitig
nature of the SI, researchers have tended to depend on self-report measures of time urgency
Jenkins Activity Survey
The JAS is a multi-item self-administered questionnaire. The
instrument was designed for adult working men and yields a
total score as well as scores on the subscalesof Speed and impatience (S), Hard Driving (H), and Job Involvement (J) (Jenkins
et al., 1971; Jenkins, Zyzanski, & Rosenman, 1979). Scoring of
the JAS is based on a complicated weighting scheme developed
from discriminant function equations based on a prospective
adult sample used to establish the predictive validity of the SI. A
large body of work on TABP has been done with college students, and the JAS was adapted for use with a college population by Glass (1977). The adaptation involved removing items
that dealt with work-related activities. Thus, the student version
of the JAS yields a total score and two subscale scores for Speed
and Impatience (S) and Hard Driving (H). The student and
adult forms of the JAS are identical, with the exception of the
missing job-involvement items and the elimination of certain
work-related words in other items.
In recent years, substantial concern has been raised about the
value of available self-report measures of both time urgency
and the more general TABP. In addition, there has been increasing disaffection with the construct of time urgency and greater
interest in the construct of hostility in the consideration of
behavioral contributions to coronary heart disease (CHD) and
other hypothesized symptoms of straiti. The SI technique and
the anger and hostility scores have seemed to be more promising predictors of CHD (Booth-Kewley & Friedman, 1987; H. S.
Friedman & Booth-Kewley, 1988; Matthews, 1982, 1988;
Wright, 1988). Thus, in large measure, this disaffection has resulted from the disappointment experienced by epidemiological researchers who have not been able to achieve high levels of
prediction of CHD with time urgency measures. The wisdom
of this shift from time urgency to hostility, of course, depends
on the assumption that the time urgency measures that have
been abandoned were psycho metrically sound. In the present
research, we revisited the issue of time urgency from a more
traditional measurement perspective and examined the extent
to which time urgency may have been poorly assessed in earlier
studies. If that is the case, tbe abandonment of time urgency for
hostility may be precipitous.
Burnam, Pennebaker, and Glass (1975) defined time urgency
as an accelerated pace. It is the tendency on the part of an
individual to consider time as a scarce resource and to plan its
use carefully. Time urgency is clearly a perspective on the part
of the individual and, as such, represetits an individual-difference variable. There is some question as to whether it is governed by constitutional mechanisms or is exclusively a learned

response. Although studies of heritability (e.g., Rahe, Hervig, &


Rosenman, 1978) among monozygotic twins have yielded little
evidence for a genetically transmitted aspect to TABP, there is
some evidence of chronically different metabolic rates and
other physiological differences between Type A and Type B
individuals (Wright, 1988). There have been no studies of tbe
possible constitutional basis of the specific time urgency factor
in the TABP In all likelihood, there is either an interaction
between physiology and environment with respect to the development of a sense of time urgency or a stress-vulnerability
mechanism in which certain individuals (e.g., those with an
abundance of a particular neurotransmitter or hormone) are
vulnerable to the development of a sense of time urgency in
particular environments (e.g., an occupation that places a great
emphasis on the judicious use of time).
Although there is a reasonable literature base on time estimation and time perception in experimental settings (e.g.. Fraisse,
1963; Frankenhauser, 1959), the treatment of time orientation
in organizational settings has not been as detailed. McGrath
and his colleagues (McGrath & Rotchford, 1983) suggested the
relationship between reward systems and the allocation of time
at the workplace, but few empirical data related to actual time
usage exist.
Schriber and Gutek (1987) pointed out that "time is a basic
dimetision of organizations. How time is partitioned, scheduled, and used has both dramatic and subtle influences on organizations and people in them" (p. 642). Within organizations, it
is easy to find examples of various points on a time urgency
continuum. Engineering development departments are often
accused of a lack of awareness of time constraints, of ignoring
deadlines, of missing projected dates for prototype development. In contrast, marketing and sales departments in these
same organizations are often accused of creating arbitrary or
self-imposed deadlines and pretenditig that they are externally
imposed. In this example, the engineering department would
be characterized as low on time urgency and the marketing and
sales departments as high on that dimension. These variations
in time urgency (at both the individual and the group level) can
cause difficulties in an organizational setting. Schriber and Gutek suggested that the extent of the difficulty will be proportional to the discrepancy between the time awareness of an
individual and the norms or expectations of the organization.
The research we conducted is related to these concepts of
time urgency, time awareness, and time use. For convenience,
we use the term lime urgency to refer to the construct of interest. But, in all likelihood, a constellation of variables is involved. As others have clearly demonstrated. Type A individuals are also more acutely aware of time, tend to schedule many
events in shorter and shorter periods of time, and so forth.
Thus, the time urgency component of TABP ought to include
facets of awareness, scheduling, and so on.
Dimensionality of Time Urgency
As indicated above, the work of Schriber and Gutek (1987) as
well as others suggests that time urgency is a rich construct with
many potential facets. It is equally clear that the current measures of time urgency do not reflect that richness. A recent
study by Fdwards et al. (1990) illustrates the problem rather

646

LANDY, RASTEGARY. THAYER, AND COLVIN

clearly. They examined three current measures of TABPthe


Framingham questionnaire, the Bortner questionnaire and the
JASin a sample of 352 business executives. After careful psychometric analysis, Edwards et al. concluded that none of these
devices is suitable as a measure of TABP They came to this
conclusion for three reasons; (a) The measures seem to tap different underlying constructs, throwing the notion of a universal
TABP syndrome into doubt; (b) each measure contains substantial measurement error; and (c) the measures mask a fundamental multidimensionality in the constructs they purport to
measure.

oped total to well over 100 items, many of the items are direct duplications. These duplications were omitted, yielding 65 unique items. The
resulting scale was completed by 190 undergraduate psychology students, and the responses were subjected to a principal components
analysis with both orthogonal and oblique rotations.

Results
The results of that analysis were provocative. As Edwards et
al. (1990) suggested, we were able to identify distinct components in the new hybrid questions. There was a clear competitive, hard-driving factor. But the most interesting result was
that there were several distinguishable time urgency components, a result also reported by Edwards et al. (1990).

Edwards et al.'s (1990) analysis is particularly illuminating.


Using the three measures that were the focus of the study, they
identified seven dimensions of time urgency, each with a different focus. These dimensions were general speed, doing many
things at once, eating fast, putting words in the mouth of others,
Study 1: Developtnent of a Likert-Type
impatience, punctuality, and time pressure Edwards et al. came
Time Urgency Scale
to three major conclusions. First, they suggested that none of
Method
the scales is adequate as a stand-alone measure of TABP In fact,
they suggested that global measures of TABP be abandoned.
On the basis of the results of our first analysis of the Likert-format
Next, they concluded that, if one were to combine scales to
hybrid questionnaire, we reanalyzed only those items that dealt with
achieve greater coherence and psychometric integrity, it would
time urgency or speed and formed a new reduced set of items for a
be appropriate to recognize the multidimensionality of several
principal factors analysis. In addition, we included those items that
of the underlying constructs and, in particular, the construct of
defined the competitive, hard-driving component extracted in the eartime urgency Thus, Edwards et al. recommended that relier analysis. These latter items were included to illustrate the discrimisearchers concentrate on facets of TABP rather than on an overnant validity of any time urgency measures that emerged from the
analysis of the new, reduced intercorrelation matrix. From a measurearching construct. Finally, they suggested that the multidimenment and scale construction perspective, we believed it would be usesionality present in the scales should not be seen as the definiful to demonstrate that the competitive, hard-driving items would detive view of the related constructs because their solutions
fine
a unique factor, thus providing at least partial evidence for the
represented "the best that could be obtained from the available
construct validity of the various time urgency dimensions.
items" (p. 452). They suggested that a complementary strategy
The combination of the time urgency and competitive, hard-driving
would be to "develop items that convincingly represent the doitems yielded a new scale of 33 items. The reduced 33-item intercorremain specified by these definitions, combine these items to
lation matrix was subjected to a principal factors analysis with an
form unidimensional measures, and test these measures for
oblique rotation (promax). Communality estimates were the squared
internal consistency, external consistency, and construct validmultiple correlations of each variable with all other variables. Rotaity" (p. 452). In the research reported in this article, we accomtions were completed on differing numbers of factors extracted to deplished those goals by concentrating on the development of
termine the most psychologically meaningful solution. A scree test
instruments for measuring time urgency In addition to the
(Cattell, 1966) indicated a point of inflection in extracted variance
Bortner scale, the Framingham scale, and the JAS. we also conbetween the fourth and fifth factors. As a result, three, four,five,and
six factors were rotated and examined for meaningful ness. The fivesidered the Thurstone Activity Scale (Thurstone, 1949), another
factor solution was most interpretable.
instrument often used to measure time urgency

Results
Pilot Study
Method
Edwards et al. (1990) maintained the formats of the measures examined (the Bortner scale, the Framingham scale, and the JAS) with the
exception of a minor modification of the Bortner response scale. However, the formats of the three measures differ substantially For example, the Bortner scale has 5 response alternatives (11 as modified by
Edwards et al.). the Framingham scale has either 2 or 4altertiatives, and
the Thurstone scale has only 3 alternatives. The JAS (Form C) has
response alternatives varying from 2 to 8. and the anchors are not
uniform but vary from item to item.
In an attempt to impose some uniformity on the response scales
associated with each of these instruments, we developed a Likert-type
format to accompany each of the questions from the four scales (Thtjrstone. JAS, Framingham, and Bortner). Thus, a new scale was developed. Although the four scales from which the current scale was devel-

The results of the promax rotated factor analysis are presented in Table I. In this table, items are given brief descriptors
to identify content and are coded to indicate from which of the
four scales (Thurstone, JAS, Framingham, or Bortner) they
came. These results are based on an analysis of a 33 X 33 intercorrelation matrix. Items are arranged under the factor titles
with the appropriate loadings for the item on the factor. As can
be seen, several clear factors emerged. The first factor represents competitiveness. The second factor relates to eating behavior. The third factor seems to relate to a general style of time
urgency or concern for time. The fourth factor seems more
specific and has a stronger task orientation to it than the third
factor. The fifth factor is less clear than the others but seems to
pick up some aspects of speech patterns or interpersonal dominance through communication activities. The internal consistency reliabilities (coefficient alpha) for the scales were as fol-

647

TIME URGENCY
Table 1
Principal Factor Analysis (Promax Rotation)
Factor
Item

Source

Slow doing things


Pressed for time
Eat too fast
Work is slow and deliberate
Go "all out"
Need to excel
Much less responsible
Talk rapidly
Eat rapidly
Bossy or dominating
"Put words in mouth"
Pressed for time
Restless and fidgety
Never in a rush
Eat more slowly
Hard driving
Htirrying
Speak louder
Work slowly
Set deadlines
Linger over a meal
Hard driving and competitive
Do things in a hurry
Satisfying myself
Ambitious
Others rate me as easy going
Work fast
Eat too quickly
Slow, deliberate talker
Consider myself as easy going
Try to persuade others
In a hurry
Work quickly and energetically

B
F
J
T
B
F
J
T
T
F
J

F
T
B
J
B
J

T
T
J
T
F
J
B
B
J
T

F
B
J
T
T

Comp)etitiveness

Eating
Behavior

General
Hurry

Task-Related
Hurry

Speech
Pattern

.26
.06
--.05
.00

-.02

-.30
.81
.00

,5/
-.14
-.03

-.11

.03
.82

.02
-.11

-,02

.67

-.04
-.02

J6

-.01

-,10

-.04
-.02
.85

,01
.10

.47
.06
.12
,19
-48
,05

715
-.03

.15

-.06

.16

.52
.03

-.03
-04
,03
.06
.79

.00

-.10

.13
.09

.79

-.06
-.08

-.01

.21
.06
.06
.56

.26
.35
.38
.00
.71

-.03
,14
.19
.06

-.12
.02
.65
,15
.05

-.08
.10
.89
.15

-.02
-.05
-.03
.21

-.08
.26

.02
.17
.8}
.14
.54
,04
,03
,21
.02
.23
.23

-.26
-.06
-.04

-.01
-.11

.35

.02
.29
.68
.01
.28
.34
.02
.27
.54

.47
.06

,20
,17
.27

-.14
-.10

-.19

-.10

,06
,03

-.26
-,17

-,03

,05
,55

-.04

.32
A9
.10
.31

.36

.16

-,10
-,01

.08

-,04

-,15

.05
.14
.46

.12

-.29
-.20

.31
.13
.08

.01
.07
.39

-.08

.06
.07

.12
.18
.36
.24

-.20
.08
.20
.01

-.10
.05

.37
.23
.04

Note. Items theoretically belonging to each factor are underlined. Items that actually loaded on each
factor are in italics. Items indicating slow responses were reversed prior to analysis. B = Bortner scale; F =
Framingham scale; T = Thurstone Activity Scale; and J = Jenkins Activity Survey.

lows: Competitiveness = .81, Eating Behavior = .89, General


Hurry = .81, Task-Related Hurry = .72, and Speech Pattern =
.69. The alphas were computed from the observed correlations
among the items underlined in Table 1.
There was substantial similarity between the dimensions
identified in the present analysis and those identified by Edwards et al. (1990), which were eating fast, general speed, doing
tnany things at once, time pressure, and putting words in the
mouths of others. This should come as no surprise because
Edwards et al.'s analysis depended on three of the four scales
analyzed in this study {although Edwards et al. did not standardize the response format).
The factor intercorrelations are reported in Table 2. As can
be seen, there was a low to moderate level of factor intercorrelation, indicating that, although distinct, these facets are not completely independent of one another.
Since our original student data were gathered, additional
data have been gathered from both student and nonstudent
samples, who responded to the 33-item Likert-type scale. The
nonstudent samples included clerical workers (n = 213), manu-

facturing supervisors {n= 91), and professional surveyors ( =


178). The student samples included introductory psychology
students {n = 642) and students enrolled in computer science
courses {n = 44). These additional data permit a broader statement of internal consistency and scale intercorrelations. The
internal consistency values for this larger sample were as follows: Competitiveness = .76, Eating Behavior = .89, General
Hurry = .78, Task-Related Hurry = .75, and Speech Pattern .60. The intercorrelations for the students appear above the

Table 2
Interfactor Correlation Among Likert-Type Suhscales
Subscale

Competitiveness
Eating Behavior
General Huny
Task-Related Hurry
5. Speech Pattern

.17
.19
.39
.19

1.
2.
3.
4.

.25
.30
.27

.34
.27

.30

648

LANDY, RASTEGARY THAYER, AND CQLVIN

diagonal in Table 3 and the intercorrelations for the nonstudents, below the diagonal.
Finally, because it was possible to administer the Likert-type
scales to a subsample of the clerical workers (n = 132) on two
separate occasions separated by 4 months, we used a LISREL
analysis to estimate test-retest reliability for those subscales.
This was accomplished by specifying the items that compose
each factor, determining the loading of those items on the factor
for Titne 1 and Time 2, and then computing the correlations of
the same latent factor across the two occasions. To account for
retest effects, the correlation between the measurement errors
of the same items across the two occasions was set free to be
estimated. If time urgency is to be considered a stable attribute
of an individual, test-retest data are valuable. These test-retest
reliabilities appear in the diagonal of Table 3. All but one were
quite high. We used LISREL to estimate test-retest reliabilities
because the traditional test-retest model assumes uncorrelated
measurement errors across occasions, an assumption that may
be unrealistic. LISREL allows for the estimation of correlated
measurement errors. However, LISREL does result in somewhat higher reliability estimates than might be obtained with
traditional test-retest correlational procedures because it effectively corrects the coefficients for unreliability of measurement
at each time of measurement.
The combination of the reliability data and the scale intercorrelation data suggest that, although there was a modest positive relation among the subscales, there also was a great deal of
reliable and unique variance associated with each.

Discussion
As was found in the earlier study by Edwards et al. (1990),
time urgency was not a unidimensional construct but a multidimensional one with several distinct facets. Although in a few
instances the reliability (internal consistency) of these subscales
was lower than one might desire, for the most part the values
reached acceptable levels, indicating the psychometric integrity
of the facets. The test-retest reliabilities were reasonable as
well. In addition, the modest positive intercorrelations among
the subscales suggests that, although there may be a generalized
tendency toward time urgeticy in many different behaviors (i.e.,
there is a constellation of variables that constitute and support
time urgent behavior), the facets are certainly not redundant.
Thus, the combination of the factor analysis results, the sub-

1.
2.
3.
4.
5.

Competitiveness
Eating Behavior
General Hurry
Task-Related Hurry
Speech Pattern

(.93)

.16

.11
.42
.48
.42

(.90)

.28
.32

.36
.42
.44

(.95)

.35
.29
.36
,38

.21

(.70)

.15
.14
.25

(.90)
.32
.34

The suggestion of a multidimensional nature to time urgency


is an important finding. It implies that relationships between
various facets of time urgency and other variables of interest
{e.g., CHD, perceived stress, job satisfaction, etc.) may differ. In
other words, there may be little or no relationship between
CHD and task-related time urgency, but the relationships between eating or speech-related time urgency and outcomes such
as CHD or anxiety may be substantial. Such differential relationships would have been impossible to detect in earlier studies because all facets of time urgency were combined to yield a
single score {as in the JAS-S factor). Although we could not use
our current data to identify what these differential relationships
might be, we can conclude that earlier research efforts to identify relationships between a unidimensional time urgency concept as measured by self-report questionnaires and health-related variables {e.g., CHD) were likely to provide equivocal and
disappointing results because all facets of time urgency had
been aggregated. Edwards et al. (1990) urged the development
of a measuring device that permits high-quality assessment of a
multidimensional construct. In Study 2, we undertook the development of such a device.

Study 2: Development of Behaviorally Anchored


Rating Scales

Table 3
lntercorrelation Among Likert-Type Subscales
Subscale

scale results (reliability), and the earlier results of Edwards et al.


suggests that there was good reason to pursue the development
of a multidimensional time urgency measure. Furthermore, it
appears that a self-report measure of time urgency might have
sufficient psychometric integrity if it were carefully developed.
One final aspect of the analysis of the Likert-type items
added support to our conclusion that the time urgency factors
were not artifacts or due to method variance: The competitive,
hard-driving factor emerged from the analysis intact and distinct from the time urgency subscales. This result provides additional evidence of discriminant validity as specified by
Campbell and Fiske (1959). In other words, with these scale
items, the construct of time urgency can be separated from the
construct of competitiveness. An important operation in construct validation is to demonstrate that a measure of the construct of interest is independent of other measures with different meaning. In the present case, there were two purposes for
such a demonstration. The first purpose was to illustrate that
method variance had been minimized. This is an analytic concern. The second purpose was to demonstrate that the construct of interest has some independent meaning, that it contributes new information and not simply old information with a
new name. We return to the issue of construct validity later.

Note. Correlations below the diagonal are based on the data from the
nonstudent samples (N - 482). Correlations above the diagonal are
based on the student samples (N = 876). Test-retest reliability coefficients (diagonal) are based on data from clerical workers (A" = 132).

When we examined instruments previously used to assess


time urgency (e.g., the Bortner scale, the JAS, the Framingham
scale, and the Thurstone Activity Scale), it was clear that different scales included different time urgent items. Because the
scales developed in the Likert-fortnat study were modified versions of items appearing in previously published questionnaires, there was no assurance that even these scales fairly represented the various facets of time urgency. As a result, a second
study was initiated to determine the dimensionality of time
urgency by using an independent methodology This study gave
us the opportunity to further explore the dimensionality of

649

TIME URGENCY
time urgency without relying on the conceptions of earlier researchers.
The behaviorally anchored rating scale (BARS) technique
was developed in 1963 by Smith and Kendall as a way of gathering performance ratings resistant to traditional forms of judgment error. In addition to performance measurement, the
BARS technique has been used for the measurement of motivational constructs through self-ratings and the ratings of observers (Landy & Guion. 1970).

Table 4
Behaviorally Anchored Rating Scale Dimensions
and Their Definitions

Since its introduction, the BARS technique has replaced traditional graphic rating as the method of choice for gathering
performance ratings that are good estimates of true performance. The nature of scale development seems to provide structural protection against many of the common forms of rating
error. The literature on this scaling technique is substantial and
has been reviewed in several recent sources (Landy. 1989;
Landy&Farr, 1980,1983).
The BARS technique offered a number of advantages for the
measurement of time urgency. First, early stagesof BARS development involve the identification of individual dimensions of a
multidimensional construct. In addition, because BARS have
frequently demonstrated acceptable psychometric integrity (reliability and resistance to common forms of rating error), such
measures of time urgency ought to provide higher quality information than that provided by currently available instruments.
Therefore, we decided to use the BARS technique to develop
graphic rating scales for facets of time urgency

Eating Behavior

Dimension

Definition

Awareness of Time

The extent to which an individual is aware of


the exact time of day, regardless of the
environment or circumstances. The extent
to which a person is aware of important
dates, such as birth dates, test dates, etc.
The extent to which time plays a role in the
manner in which individuals plan or eat
various meals.
The extent to which a person can be
characterized as being in constant motion,
even while "resting."
The extent to which a person creates or
maintains a list of things to do during the
day or during the week.
The extent to which an individual schedules
activities and keeps to that schedule. The
schedule might include leisure, personal,
or work activities. This also includes the
extent to which an individual apportions
time for particular activities.
The extent to which an individual exhibits
rushed speech patterns. This would include
talking fast, interrupting others, and
finishing the sentences of others.
The extent to which an individual creates or
appears to be controlled by external
deadlines.
The extent to which a person engages in
actions directed toward saving time
through more efficient planning or action.
The extent to which a person can tolerate
tardiness in others.

Nervous Energy
List Making
Scheduling

Speech Patterns

Deadline Control
Time Saving

Method
Dimension identification. Six undergraduate research assistants
who had been working on the earlier phasesof the time urgency project
met with Frank J. Landy for 2 hr to generate possible dimensions of
time urgency. They were told to brainstorm and suggest potential
names of dimensions without worrying about the definitions of these
dimensions. In the course of this brainstorming. approximately 30
labels were generated and listed on a blackboard. When no more labels
could be generated, the group examined the entire list and eliminated
those labels that were synonymsof others on the list. In addition, labels
were combined to yield more comprehensive categories, and new labels
were generated to cover the new category. A total of 12 dimensions were
produced in this phase. The next step was to generate conceptual definitions for the dimension labels. In the process of developing these
definitions, it became clear that three of the labels were covered by
other dimensions on the list. Thus, when the definitions had been
completed, we were left with 9 dimensions of time urgency. These 9
dimensions and their definitions appear in Table 4.
Behavioral anchor development. The six judges were asked to develop three high, three moderate, and three low anchors for each of the
nine dimensions. In theory, this could produce approximately 500 potential anchors. In fact, it was difficult for some judges to develop the
full nine items for each scale. Furthermore, many of the examples were
similar or identical across judges. As a result, a total ofl81 potential
anchors or examples were identified and edited to form the pool of
behavior examples for the translation phase.
Relranstation. The 181 items were presented to an independent
group of 16 undergraduate students for retranslation. The task of these
judges was to place the items back into the categories for which they
were originally prepared.
Items were retained for scaling if 75% of the judges assigned the item
to the dimension for which it was originally written. On the basis of this
criterion, a number of items were eliminated from further consider-

Tolerance of
Tardiness

ation. Thenumberofitemsthat were initially written for a dimension,


as well as the number of items that survived the retranslation process,
are presented in Table 5. As can be seen from that table, some items
were deleted and one dimension (time saving) was left with too few
items to form a reasonable scale. Thus, behavioral examples for eight
scales were considered in the next phase, the scaling phase.
Scaling. In the scaling phase, 46 undergraduate student judges assigned scale values to the behavioral examples. Judges considered each
behavioral example and assigned a scale value ranging from 1 to 7,
based on their decisions about how much of the time urgency aspect
that particular example represented.
A substantial number of items survived the scaling process. In this
phase, we were looking for items that had relatively low standard deviations (approximately 1.00 or below) and mean values that were representative of different points on the scales. In general, it is desirable to have
four or more anchors at different points on a scale. Reviews of the rating
literature show that the most robust range of scale points is between 4
and 8 (Landy & Farr, 1980). In fact, there were sufficient numbers of
anchors in many of the scales to form parallel forms of the rating
scales. The only dimension for which there were not enough anchors to
form parallel forms was deadline control. The intolerance of tardiness
dimension was dropped from further consideration because too few
anchors survived the retranslation process to form even one scale.
The scaling data were gathered in two separate sessions. As a check
on the overall agreement among judges, we correlated mean item values for each dimension from each of the two sessions (see Table 5). The
correlations between sets of mean values for each dimension were

650

LANDY, RASTEGARY, THAYER. AND COLVIN

Table 5

Characteristics of Behaviorally Anchored Rating Scales (BARS) in Development


Dimension
Development stage
Item survival
No, of initial items
No, of survival items
Correlations between sets of
mean values for each dimension
of the BARS"
Before item deletion
After item deletion
Parallel form reliability'
Time I
Time 2
Combined
Test-retest reliabilities for
the BARS'
Set A
SetB
Combined
' Dimension deleted,
samples (tt = 168).

Awareness Speech
Nervous
List
Eating
of Time Patterns Scheduling Energy Making Behavior
21
16

20
18

21
20

13
12

18
13

31
22

Deadline Intolerance
Time
Control of Tardiness Sharing'
25
10

19
13

16
3

.96
.97

.97
.97

.97
.91

.98
.98

.98
.99

.98
.94

.63
.79
.78

.60
.69
.70

.64
.64
.68

.82
,82
.86

,69
.72
.75

.67
.83
.84

_
_
_

.78
.63
.80

.72
.72
.79

.66
.82
.81

.73
.83
.84

.72
.78
.83

.72
.75
,82

,66

_
_

.97
.97

' These correlations are based on the responses of 46 judges. ' These reliabilities are estimated from one of the student

computed before any items were eliminated (r = .94) and again after
items were eliminated because their standard deviations were greater
than one (r = .97). Agreement between the two separate groups of
judges regarding the scale values of the various items was very high,
Intraclasscorreiation coefficients computed on the scale valuesof individual items ranged from .91 to ,99.
Thus, the BARS developmental process produced two parallel rating scales with multiple anchors for six dimensions and a single rating
scale for the seventh dimension. An example of one of the scales appears in Figure 1. These scales were then used for estimating reliability
Reliabitily estimates. The resulting 13 scales were completed by 168
undergraduate students enrolled in an introductory psychology class.
Four weeks later, the students completed the two sets of scales again.
From these two administrations, we estimated the parallel-form and
test-retest reliabilities of the scales.

Results
The parallel-form reliabilities for the scales are presented iti
Table 5. Because the scales were administered on two separate
occasions, it was possible to estimate the parallel-form reliability for each occasion independently as well as to collapse across
occasions for estimation purposes. As can be seen from the
table, most values were acceptable. It seems reasonable to conclude that, for six ofthe facets identified, parallel forms of the
rating scales can be used. Remember that there were insufficient anchors to form a second scale for the dimension of deadline control.
The test-retest reliabilities for the BARS also are presented
in Table 5. Again, because parallel scales were administered on
the two occasions (with the exception of deadline control), it
was possible to estimate the test-retest reliability of each set
and of the combined sets. These three sets of values are presented in Table 5. As can be seen from the table, most test-retest reliabilities were acceptable, although the few values in the

.60s are only minimally adequate. The term acceptable implies


that there should be some stability to time awareness and urgency. Nevertheless, because this is exploratory research, the
term acceptable is defined by reference to earlier research on
other trait ratings. Because the nature of BARS rating produces
single-item scores, internal consistency reliabilities cannot be
estimated.
Intercorrelations among scales. The dimensionality of the
resulting scales can be addressed by considering the intercorrelations among those scales. Although the scale development
process should have produced relatively independent scales, it
is useful to examine the extent of this independence empirically. Because the scales were initially field tested, we had data
from several different populations, including both students and
employed adults. Respondents varied in age. gender, occupation, and educational level. In total, we had data from 1,358
subjects on the primary set of rating scales. The intercorrelations of the dimensions for student and nonstudent samples
appear in Table 6. As can be seen in that table, the intercorrelations were low Because the reliability ofthe scales (both parallel-forms and test-retest) is acceptable, this lack of intercorrelation may be interpreted as resulting from differences in true
scores rather than from large error variance in each rating.
These data suggest that the subjects were capable of using these
scales in both a reliable and a discriminating manner. For the
interested reader, the intercorrelations between the time urgency BARS and the Likert-type measures (described in Study
1) also are presented in Table 6. Because only two dimensions of
the different approaches were conceptually similar (eating and
speech), inferences must be limited. Nevertheless, it is encouraging to see the modest correlations for the speech dimension and
the more substantial correlationsforthe eating dimension. This
is additional evidence, albeit marginal, of the wisdom of the
multifaceted approach to time urgency.

TIME URGENCY
1. Awareness of Time
The extent to which an individual is aware of the exact time of day,
regardless of the environment or circumstances. The extent to which
a person Is aware ot important dates such as binhdays, tests, etc.

High
I glance at my watch frequently during the
day
Even when I can't look at a watch or clock, I
know approximately what time it is.

Average

I often lose track ot time when I am engaged


in an activity.
Sometimes I remember the birthday of a
close friend or relative a day or two after
it has passed,
I must often be reminded of important dates,
I seldom look at my watch or a clock.

651

the construct being tapped is unique, rather than simply a new


name for an old construct, one should be able to demonstrate
that the new construct is uncorrelated with measures of these
distinct constructs. To provide this type of evidence, we administered the BARS and other measures to a sample of clerical
workers attending a conference at Pennsylvania State University The additional measures included the State-Trait Anxiety
Scale (Spieiberger et al.. 1979) and the Job Descriptive Index
(JDI; Smith. Kendall, & Hulin, 1969), which provides a measure
of job satisfaction. The logic was that time urgency ratings
might simply represent generalized or specific anxiety or possibly job dissatisfaction. The intercorrelations between these
measures and the time urgency measures appear in Table 7.
Although there were several significant intercorrelations, for
the most part these intercorrelations were low and show no
pattern of relationship with the time urgency scales that would
suggest that the time urgency facets are simply surrogates for
other better known constructs, such as job satisfaction or anxiety More will be said about the correlations between the JDI
scores and the time urgency scores in the General Discussion
section.

Finally, Spence, Helmreich, and Pred {1987; Spence, Pred, &


Helmreich, 1989) published a new version of the JAS that is
intended to eliminate many of the earlier problems. In this
form, two dimensions are isolatedimpatience/irritability and
achievement striving. The impatience/irritability scale was also
administered to our clerical sample because it was conceptually
linked to our notion of time urgency The correlations between
Occasionally, 1 forget what day of the week
Low
it is.
the impatience/irritability score and the time urgency scores
were significant (see Table 7). Once again, this is good news
because it demonstrates that the BARS do correlate with other
measures intended to assess the same general construct. The
Figure I. An example of a behaviorally anchored scale.
fact that the correlations are not higher is probably because
Spence et al.'s JAS revision, like its predecessor, mixes facets of
time orientation and urgency to produce a single subscore. In
addition. Spence et al.'s revision includes general irritability in
Time urgency and demographic variable.^. Both tnen and
the subscore, further obscuring the time urgency facets. Neverwomen were represented in our various samples. In addition,
theless, even with this "noise," the intercorrelations were signifiacross all samples there was substantial variation in the age of
cant and in the expected direction.
respondents. In exploring possible covariates of the time urgency construct, we felt that these two demographic variables
would be of interest. The correlations appear in Table 7. AlStudy 3: Construct Validity of the Measures
though several correlations were statistically significant, the
only substantial correlation with respect to gender was for the
From a construct validity perspective, the results of the cleriBARS list-making dimension. Women had a greater tendency
cal worker study are supportive. The time urgency scales correto make lists as a way of arranging activities. The only substanlated with constructs that seem related and did not correlate
tial correlation for age was with the BARS time awareness dihighly with constructs that seem unrelated or at least are not
mension for the collapsed student and nonstudent sample.
identical to the time urgency construct. Nevertheless, to be sure
Younger respondents tended to be more aware of time than
that something other than response bias or method variance is
older respondents. The extreme homogeneity of the student
represented by the data, a more formal treatment of construct
sample with respect to age and the mean differences between
validity was warranted. In essence, the question to be addressed
the student and nonstudent samples with respect to age may
was whether the multiple dimensions of time urgency, and in
account for the significant overall correlation and the absence
fact the overarching construct itself, can be independently conof within-group correlations. Again, although there were other
firmed. Traditionally, such independent confirmation has
significant correlations in addition to the awareness-of-time retaken two forms. The first form is to demonstrate that two
lationship, the values were low.
different methods of measuring the same attribute reach the
Time urgency and other measures. As indicated earlier, one same conclusion. This is manifest in multitrait-multimethod
analyses. In a modest way, this was illustrated in the correlaoperation in the examination of the construct validity of an
tions between the Likert and BARS measures of eating and
instrument is demonstrating its independence from other inspeaking
behavior. A second method for demonstrating construments (representing distinct constructs). The logic is that if

652

LANDY, RASTEGARY, THAYER, AND COLVIN

Table 6

Interrelationship Among Behaviorally Anchored Rating Scales (BARS) and Between BARS
and Likert Scales for Student and Nonstudent Samples
BARS
BARS

Likert-type scales
5

Competitiveness

Eating
Behavior

General
Hurry

Task-related
Hurry

Speech
Pattern

-.03

.06

.01
.39*
-.02
.25*

Student sample (M ^ 876)


1. Awareness of Time
2. Speech Pattern
3. Scheduling
4. Nervous Energy
5. List Making
6. Eating Behavior
7. Deadline Contro!

.05

.33*

.05

.10*

.04

.14*
.10*

.06

.21*
.10*

.08
.11*
.18*
.12*
-.04

.21*
.06

.33"
.08

.30*
.05

.19*
.12*
.28*
.14*
.14*
.07

.12*
-.04
.50*

.30*

.02

.03
.16*
.05
.25'
.16*
-.07
.08*

.03
.01

.09*
-.08*
.25*
.08*

.07

.04

.09*

.lr

.01

Nonstudent sample ( -482)


1. Awareness of Time
2. Speech Pattern
3. Scheduling
4. Nervous Energy
5. List Making
6. Eating Behavior
7. Deadline Control

.16*

.10*

.04

.13*
.18*
.10*

.16*
.12*
.10*
.15*

.11*
.12*

.20*

.08

.02

.02

.09

.06

.14*
.21*
.36*
.22*

.16*
.14*
.22*
.22*
.12*
.22*

.18*
.02
.17*
.10*
.57*
.26*

.20*
-.01
.20*
.12*
.13*

.09

.03

.12*

.25*
.20*

-.01

.06

.05

.36*

.13*
.17*
.12*

.06

.23*

.14*
.18*
.14*

*/K.O5.

struct validity is the muititrait-multirater approach. Here, we


set out to demonstrate the integrity of the concept by demonstrating that independent judges could agree on the presence or
absence of an attribute. In the present case, we compared the
self-reports of our subjects with the reports of others who had an
opportunity to observe them.

Method
The BARS were administered to a hetergeneous employed sample
consisting of office and professional workers from two separate organizations. Twenty-eight engineers from a medical equipment manufacturing company and 25 service representatives from a teachers' professional organization made up an employed sample of 53. Respondents
were asked to have their spouses also complete the scales but describing the respondents rather than themselves. The completed scales were
returned in a sealed envelope for analysis. Thus, the traits were the
various dimensions of the BARS, and the raters wereself and spouse.

Results
The results were encouraging. Using the correlations from
Table 8, we conducted an analysis of variance (ANOVA) to assess the multitrait-multi rater data. This procedure was articulated by Stanley (1961), who demonstrated that it was possible
to obtain traditional mean square estimates for the model from
the variance and covariance among the measures. This technique has been further illustrated in applications to the trait
ratings often gathered in industrial and organizational settings
(King, Hunter. & Schmidt, 1980; Schmitt & Stults, 1986). The
results of the ANOVA analysis are presented in Table 9. In this
analysis, there were four sources of variance: (a) the ratee component, which represents overall agreement on ratees across raters

and traits and corresponds to convergent validity; (b) the


Ratee x Trait component, which represents the degree of rater
discrimination among traits, or discriminant validity; (c) the
Ratee X Rater component, which represents covariation that
might be considered halo; and (d) error. The critical issue in
assessing the convergent and discriminant validity of the scales
is the amount of variance due to each source. As a result, variance components were also estimated. These components allow inferences to be made about effects relative to unexplained
variance (error) while controlling the sample size. These variance components revealed adequate agreement among raters
on traits (convergent validity) and good differentiation among
traits (discriminant validity). Although error variance appeared
to be high, this is neither unusual (Boruch, Larkin. Wolins, &
MacKinney, 1970; Dickinson & Tice, 1973; Kavanagh, MacKinney; & Wolins, 1971) nor unexpected (Schmitt & Stults,
1986). The high error variance is due, in part, to the fact that the
ANOVA design does not recognize correlated errors among
traits (Schmitt & Stults, 1986).
The intercorrelation matrix (Table 8) seems to point to scheduling and, to a lesser extent, deadline control as problematic
dimensions. Therefore, we recalculated the ANOVA, eliminating scheduling. The variance component (see Table 9) associated with validity coefficients increased (convergent validity
from . 19 to .21, discriminant validity from .29 to .36), and the
error variance component decreased (from .52 to .43). Next,
deadline control was dropped from the analysis, but this deletion did not further increase overall validity estimates or reduce
error estimates.
To assess the characteristics of the present scales, we found it
i nformative to compare the variance estimates in this study and
those typically found in muititrait-multirater analyses of rating

653

TIME URGENCY
Table 7
Interrelationship Between Behaviorally Anchored Rating Scales (BARS) and Other Variables or Scales
BARS
Variable or scale
Age'
Student sample (n == 876)
Nonstudeni sample (n = 482)
Overall sample
Gender"
Student sample (n = 876)
Nonstudent sample (n = 482)
Overall sample
State-Trait Anxiety Scale
(Spielberger et al.. 1979)"*
State Anxiety
Trait Anxiety
Job Descriptive Index
(Smith, Kendall, & Hulin. 1969)=
Work

Awareness
of Time

Speech
Patterns

Scheduling

.08
-.09
-.27*

.01
-.10*
-.10*

-.06
-.04

.09*
-.17*
-.03

.11*
-.02
.05

-.01

.16*
.20*

-.15*
-.20*

.17-

-.02

Pay

Promotion
Supervisor
Coworkers
General
Impatience/Irritability scale of new JAS
(Spence, Helmreich. & Pred, 1987)^

.04

.16*
.08

.23*

.07

Nervous
Enei^y

List
Making

Eating
Behavior

-.03
-.04
-.11*

-.07
-.03
-.11*

-.01

.01

.01

-.02

-.07
-.07

-.08
-.17*
-.12*

.11*
-.09*
.02

.04

.06
-.03
.02

.37*

-.14
-.16*

.14

.25*

-.09
-.09
-.07
-.03
-.18*
-.07

.10
.08
.07
.07
.07
.12

.03
-.06
-.03
-.02
-.16*
-.03

.14
.02

.30*

.12

.36*

.14

.02

.09

.07

.23*

E>eadline
Control

.03
.10

.19*
.05

.18*

.17*
.10

.14
.07
.06
.11

.02
.03
.09
.06
.04

.23*

.16*

.23*

.07

.17*
.02
.05
.09

^vte. JAS = Jenkins Activity Survey.


' For the overall sample. M = 29.76. SD = 13.46; for the student sample. M = 19.53, SD - 2.92; and for the nonstudent sample, M = 42.72, SD 9.96. '' Male coded as I. female coded as 2. In the nonstudent sample, there were 258 men and 305 women; in the student sample, there were 221
men and 242 women. " Based on a sample of clerical workers (A'= 231). ** Higher values are associated with lower anxiety. ' Positive correlations mean that respondents displaying impatience on the new JAS also displayed time urgency on the BARS.
p < .05.

scales. To make such a comparisoti, we used Kitig et al.'s (1980)


estimates of convergent variance, discrimitiatit variance, halo,
and error for 11 empirical studies. A comparison ofKinget al.'s
estimates with those obtained in the current study showed that

the current scales demonstrated higher trait variance and lower


halo variance than is common for such scales. In addition, the
ratee and error variance components were similar to values
found in other studies.

Table 8
Multitrait-Multirater Matrix for Behaviorally Anchored Rating Scales of Time Urgency
Scale
Self-ratings
1. Awareness of Time
2. Speech Pattern
3. Scheduling
4. Nervous Energy
5. List Making
6. Eating Behavior
7. Deadline Control
Spouse ratings
8. Awareness of Time
9. Spwech Pattern
10. Scheduling
11. Nervous Energy
12. List Making
13. Eating Behavior
14. Deadline Control

.15
.15

.31*

.30*

.11

.03

tl

12

10

.09

.17
.02

.21

.34*

-.06

.47*

.31*

.09

.11
.17

-.08
.49*

13

14

-.04

.28

-.19

-.25
-.01

.11

.20

-.11
.19

.35*
-.18

.56*

.26

.15
.14

.36*
.32*

-.10
.12
.14
.17

.12
.17
.17
.26

.33*
.35*
-.10
.54*
.02

-.12

-.04

.01
.06

-.04
.40*

.51*
.23

.25

.07
.12
.18
.05

.22

.14
.24

Note. The sample included 28 male and 25 female respondents.


*p<.Q5.

.09

.56*
-.04
.36*

.08
.24
.29
.11
.18

.55*
.31*

.01

-.24

.27

.09
.10
.15
.03

.43*

.30*

.44*

.21

.24
.23

.57*

.15

654

LANDY, RASTEGARY, THAYER. AND COLVIN


Table 9

Summary of the Analysis of Variancefor Self- and Spouse Ratings of Time Urgencv

Source

df

MS

Variance
component

Variance
component
without
scheduling

Ratees
Ratees x Traits
Ratees X Raters
Error

52
312
52
312

4,04
0.32
1.53
0.52

7,48*
0.59
2.83

,19
.29
.05
.52

.21
.36
.02
.43

Typical
estimates'
2$
.08
.31
.38

' These values appeared in an analysis by King. Hunter, and Schmidt (1980) of typical multitrait-multirater matrices.
p < .05.

The purpose of this analysis was to show that the construct of


time urgency is more than simply method variance or response
bias. To do that, it should be demonstrated that the attribute
can be anchored in some reality outside of self-reports. We were
able to demonstrate this independent reality by showing the
convergence and discriminance of self-reports and spouse reports. Although the sample was small, construct validity was
demonstrated in the multitrait-multirater analysis. When this
validity is coupled with the earlier data on reliability and internal consistency, support for the construct seems quite reasonable and responsive to Edwards et al.'s (1990) call for unidimensional measures with internal consistency, external consistency,
and construct validity.

General Discussion
Measurement of Time Urgency
A number of conclusions can be drawn from the current
data. The first and most obvious is that the time urgency BARS
investigated represent reasonable instruments for gathering
self-report data on how individuals perceive and administer
time. The scales that have been available previously (e.g, the
JAS. the Framingham scale, the Bortner scale, and the Thurstone Activity Scale) have been disappointing in contrast with
the SI {Booth-Kewley & Friedman. 1987; Matthews, 1988). This
may have been either the result of inadequate initial development or a failure to recognize the multidimensional nature of
the time urgency construct (Fdwards et al,. 1990). The current
scales demonstrate acceptable levels of reliability and sufficient
evidence of discriminant and convergent (i.e.. construct) validity to warrant careful consideration.
Because the BARS were developed with student samples, it is
reasonable to speculate whether the same results might have
been found if nonstudent subjects had been involved. It seems
likely that the current results are not sample specific. We draw
this conclusion for several reasons. First, the patterns of reliabilities for the BARS were highly similar for the student and nonstudent samples. In addition, the multitrait-multirater analysis
confirmed the construct validity of the measures and facets in
an employed sample. Finally, it is worth noting that, in the
BARS, only one anchor contains a word (test) that relates to
student populations. This single word is deleted when the scales

are used with employed samples. In terms of both empirical


and logical considerations, the dimensions and scales do not
appear to be sample specific.
The identification of a reasonable method for gathering time
urgency data is particularly timely In the last few years, a certain disenchantment with the construct and the measurement
of time urgency has appeared. In large part, this has been due
to the inability of researches to show links between measures
of time urgency and consequent variables of interest. Many
researchers are suggesting that hostility should be considered
the culprit (i.e.. active ingredient) in various disease entities
linked to TABP For example, Williams and colleagues (Williams, 1984; Williams, Haney, et al., 1980; Williams, Lane, et al.,
1982) have argued forcefully for hostility as the active ingredient
in TABP. Recently, Matthews (1988) added support to that suggestion. However, Ohman, Nordy, and Svebak (1985) and
Wright (1988) have argued for the role of time urgency in cardiovascular reactivity In addition, these researchers suggest that
time urgency may be a causal factor in emergent hostility and
that the link to CHD involves both constructs in an interactive
manner Wright (1988) indicated the need for a better time urgency measure. Although there seems to be agreement that the
SI provides the best index of TABP. the SI is labor intensive and
of little use for widespread epidemiological and field research.
The current scales for time urgency may represent an alternative method for producing at least one component score of that
interviewa time urgency score.
In light of the current results, abandoning time urgency for
hostility in the context of understanding work strain and CHD
may not be warranted. At the very least, it is precipitous to
abandon the time urgency construct before it has been given a
careful examination with appropriate methods. It may well be
that hostility is a more central or interactive construct, as
Wright suggests, but it seems clear that researchers will not be
able to evaluate that possibility until a reasonable time urgency
measure is available.

Dimensionality of Time Urgency


The second conclusion that seems appropriate is that time
use and time urgency are multidimensional constructs. Edwards et al. (1990) demonstrated this with an analysis of the
JAS. the Framingham questionnaire, and the Bortner question-

TIME URGENCY
naire. To those measures, we added the Thurstone Activity
Scale and found similar dimensionality The data clearly indicate the independence of some traditional item types, for example, eating behavior and speech patterns. Because these dimensions are measured by demonstrably reliable methods, it is safe
to conclude that these differences are real and not artifacts of an
unreliable data set. In fact, the matrix of intercorrelations
among the BARS shows uncharacteristically low levels of intercorreiation compared with what is commonly found in different facets of omnibus self-report measures. The data support
the notion that there is a constellation of facets that define or
constitute time urgency. This constellation includes such dimensions as time awareness, scheduling, deadline setting, and
so forth. Perhaps the term time use or time perception or time-re-

lated behavior should be substituted for time urgency Nevertheless, given the wide use and understanding of the term time
urgency, and because it is ultimately this urgency that may be
the culprit in certain health-related disorders, it is advisable to
continue using the term time urgency, at least for the present.
Nevertheless, the BARS technique clearly identified different
facets of time urgency If one examines the measures that have
been used to provide time urgency scores in the past, our data
confirm the conclusion of Edwards et al. (1990), Matthews
(1988), and others that apples, oranges, and peaches have been
thrown together in the same basket. At leastfiveofthe present
BARS appear ready for operational use, but scheduling and
deadline control may need additional refinement. The fact that
they correlate substantially with each other suggests that they
could be combined to form a single scale.

Time Urgency in Organizational Settings


Earlier studies have pointed to the dimensionality of time.
Dapkus (1985) used a clinical interview to explore time awareness in individuals and found several distinct dimensions in the
temporal schemes of participatits. These dimensions included a
changing awareness of time, a recognition of time as a scarce
resource, and the tempo of time or the perception ofthe speed
with which time passes.
The role of time in studies of stress in the workplace is almost
axiomatic. As an example, Kirmeyer (1988) used time in her
definition of role overload. She cited the earlier work of Kahn,
Wolfe, Quinn. Snoek, and Rosenthal (1964), who portrayed
overload as too much to do in the time available. Earlier studies
(e.g., Beehr, Walsh, & Taber, 1976; Caplan & Jones. 1975) demonstrated that perceptions of overload on the part of employees
are correlated with greater levels of effort, fatigue, anxiety, and
job dissatisfaction. In the data we gathered from the clerical
sample, the BARS speech dimension correlated negatively with
the nervous energy dimension and the JDI satisfaction with
co-workers scale. This suggests that the time urgent individual
may see co-workers as obstacles, moving more slowly than the
worker would like. Glass, Snyder, and Hollis (1974) demonstrated that Type A individuals reacted with impatience when
the completion of a task was delayed by another person. Kirmeyer concluded that overload is clearly implicated in stress
perception and that TABP is also involved, but called for more
work to investigate the dynamics by which these two variables

655

are related. We suggest that they are tied together through feelings of time urgency.
The work of Schriber and Gutek (1987) is quite interesting in
light ofthe current results. They were able to identify 12 dimensions of an organizational time perspective, including schedules
and deadlines, punctuality, awareness of time, and allocation of
time. Schriber and Gutek concluded that time is part of the
culture of organizations and, as such, represents a body of normative expectations. Some organizations expect people to be
on time for work, but others are not so concerned about punctuality Some organizations expect people to set and meet deadlines, and others are less structured. In some organizations, an
awareness of time is central to acceptance, whereas in others it
is less relevant. Schriber and Gutek suggested that individual
satisfaction and happiness may be related to the extent that
workers' patterns of time awareness match organizational expectations. The positive correlation between the BARS awareness of time dimension and the JDI general satisfaction score is
particularly interesting. It appears that people who are generally aware of time are more satisfied, overall, than those who are
less aware of time. From Schriber and Gutek's perspective, this
might suggest an organizational environment in which time is
an important parameter and in which those who are aware of
time adapt more effectively In fact, the satisfaction data came
from clerical workers in secondary schools. On the surface, this
environment seems to be one in which careful time monitoring
and awareness would be adaptive. This would account for the
positive correlations observed between the BARS time awareness scale and the JDI work and general scales. The post hoc
nature of these analyses make the results less definitive and
more speculative. Nevertheless, our results do lend some support to mechanisms proposed by earlier research. Schriber and
Gutek (1987) suggested a framework for understanding dissatisfaction, withdrawal, and productivity variations in organizations. They suggested that these are all consequences of the
degree of match between organizational expectations with respect to time and its use and individual time orientations.
Schriber and Gutek developed an attractive set of variables for
assessing the time orientation of organizations. Our scales may
provide a means of assessing the individual-differences side of
Schriber and Gutek's equation.

Future Research in Time Urgency


Edwards et al. (1990) suggested that the discovery of multidimensionality in TABP and its facets (i.e., time urgency) should
be followed by a reconstruction and reconsideration of existing
instruments and the independent development of new instruments. The research we have reported follows that suggested
line. The BARS represent an independent consideration ofthe
construct of time urgency without the constraints of existing
instruments or views.
Now that there are some reasonable measures of time urgency and use, a good deal of parametric work remains to be
done. More must be learned about the extent to which these
measures are related to basic measures of such constructs as
state and trait anxiety, anger, locus of control, and other variables that have been implicated in stress reactions. How do
these scales relate to various demographic characteristics, such

656

LANDY, RASTEGARY THAYER, AND COLVIN

as age, gender, race, occupational status, and educational status? The data we have presented in this article are only a beginning with respect to the articulation of the full construct validity of the measures.
Assuming that the measures continue to demonstrate adequate reliability and validity, there are a number of issues that
should be investigated. The ftrst is the extent to which feelings
of time urgency can be affected by environmental variables.
Hurrell (1985) demonstrated the impact of machine-paced
work on time urgency as measured by a modified version of the
Thurstone Activity Scale (Thurstone, 1949). This result should
be replicated with more precise measures of time urgency, such
as those described in this article. Johansson and Aronsson
(1984) suggested that computer environments induce a time
urgent type of behavior. This interaction between environments and individual-difference variables in the context of
time urgency was also suggested by Schriber and Gutek (1987),
as described earlier. With more refined instruments to measure
time urgency, this hypothesis can now be more completely examined.

sciousness, achievement striving, and the Type A coronary prone


behavior pattern. Journal of Abnormal Psychology 84. lb~19.
Campbell, D. X, & Fiske. D. W (1959). Convergent and discriminant
validation by the multitrait-multimethod matrix. Psychological Bulletin. 56. ^\-\QS.
Caplan, R. D, & Jones, K. W (1975). Effects of workload, role ambiguity and Type A personality on anxiety, depression and heart rate.
Journal of Applied Psychology. 60. 713-719.
Cattell, R. B. (1966). The scree test of the number of significant factors.
Multivariate Behavioral Research. I. 245-276.
Dapkus, M. (1985). A thematic analysis of the experience of time. Journal of Personality and Social Psychology. 49, 408-419.
Dewe, P J. (1988). Investigating the frequency of nursing stressors: A
comparison across wards. Social Science and Medicine. 26, 375-380.
Dickinson, T. L., & Tice, T E. (1973). A multitrait-multimethod analysis of scales developed by retranslation. Organizational Behavior and
Human Performance. 9. 421-438.
Edwards, J. R., Baglioni, A. J., Jr., & Cooper, C. L. (1990). Examining
relationships among self-report measures of Type A behavior pattern: The effects of dimensionality, measurement error, and differences in underlying constructs. Journal of Applied Psychology. 75.
440-454.
We are currently conducting a study in which the time urFraisse, P (1963). The psychology of time New York: Harper & Row.
gency scales will be validated with behavioral measures. High
Frankenhauser, M. (1959). The estimation of time: An experimental
and low time urgent subjects are exposed to various delays or
study Stockholm: Almquist & Wiksell.
lags in using computer or word-processing equipment. It is anFriedman, H. S., & Bootfi-Kewley, S. (1988). Validity of the Type A
ticipated that the time urgent subjects will show greater disconstruct: A reprise. Psychological Bulletin, 91. 293-323.
comfort and physiological arousal than will the non-time-urFriedman, M., & Rosenman, R. (1974). Type A behavior and your heart.
gent subjects.' Boucsein (1987) presented preliminary data sugNew York: Knopf.
gesting that there should be a significant interaction between an
Glass,
D. C. (1977). Behavior patterns, stress, and coronary diseaseorganismic variable, such as time urgency, and an environmenHillsdale,
NJ: Eribaum.
tal variable, represented by computer lag time. We are using
Glass, D C, Snyder. M. L., & Hollis, J. F (1974). Time urgency and the
Boucsein's experimental paradigm to test this proposition.
Type A coronary-prone befiavior pattern. Journal of Applied Social
Finally, we have begun a program of epidemiological rePsychology. 4,125-\4O.
search directed toward examining the relationship between the
Haynes, S. G., Levine. S., Scotch, N., Feinleib, M., & Kannel, W B.
(1978). Tfie relationship of psychosocial factors to coronary fieart
various measures of time urgency and use and other life-style
disease in the Framingham study: I. Methods and risk factors. Amerivariables and events, such as leisure patterns and cardiovascucan Journal of Epidemiology. 107. 362-383.
lar incidents.
Hurrell, J. J. (1985). Machine paced work and Type A behaviour pattern. Journal of Occupational P.sychotogy 58.15-26.
' We would like to thank Marianne Frankenhauser for suggesting
Jenkins, C D., Zyzanski. S. J., & Rosenman, R. H. (1971). Progress
severalfield-basedstudies for the validation of the time urgency scales,
toward validation of a computer-scored test for the Type A coroas well as Wolfram Boucsein for sharing data and designs with us. We
nary-prone behavior pattern. Psychosomatic Medicine. 33,193-202.
would also like to thank Walter Freytag and Gerry Brandon for their
Jenkins, C D, Zyzanski, S. J., & Rosenman, R. H. (1979). Jenkins
help in obtaining muititrait-multirater data.
Activity Survey manual (Form C). New York: Psychological Corp.
Johansson, G., & Aronsson, G. (1984). Stress reactions in computerized administrative work. Journal of Occupational Behaviour. 5.159Referetices
181.
Beehr, T A.. Walsh, J. T. & Taber, T. D. (1976). Relationshipof stress to Kahn, R. L., Wolfe, D M., Quinn, R. P., Snoek, J. S., & Rosenthal. R. R.
(1964). Organizational stress: Studies in role conflict and ambiguity
individually and organizationally valued states: Higher order needs
New York: Wiley
asa moderator. Journal of Applied Psychology. 61. 41-47.
Kavanagh. M. J., MacKinney, A. C, & Wolins, L. (1971). Issues in
Booth-Kewley, S., & Friedman, H. S. (1987). Psychological predictors
managerial performance: Multitrait-multimethod analyses of ratof heart disease: A quantitative review. Psychological Review. 101.
ings. Psychological Bulletin. 75. 34-49.
343-362.
Boruch, R. F, Larkin, J. D., Wolins. L., & MacKinney, A. C. (1970).
King, L. M., Hunter, J. F., & Schmidt. F. L. (1980). Halo in a multidiAlternative methods of analysis: Multitrait-muhi method data. Edumensional forced-choice performance evaluation scale. Journal of
cational and Psychological Measurement. 30, 833-853.
Applied Psychology. 65. 507-516.
Boucsein, W (1987). Psychophysiological investigation of stress inKirmeyer, S. (1988). Coping with competing demands: Interruption
duced by temporal factors in human-computer interaction. In M.
and Type A Pattern. Journal of Applied Psychology 73. 621 -629.
Frese, E. Ulich, & W Dzida (Eds.), Psychological issues of human- Landy, F J. (1989). The psychology of work behavior. Monterey, CA:
computer interaction in the workplace (pp. 163-165). Amsterdam:
Brooks-Cole.
North-Holland.
Landy, F. J., & Farr. J. L. (1980). Performance rating. Psychological
Burnam. M. A., Pennebaker, J. W, & Glass, D. C. (1975). Time conBulletin, 87. 72-\01.

657

TIME URGENCY
Landy, F. J,, & Farr, J. L, (1983). The measurement ofwork performance:
Methods, theory and application. San Diego, CA: Academic Press.
Landy, F J., & Guion, R. M. (1970). Development of scales for the
mtatsuremtmof work motivation. Organizalional Behavior and Human Performance. 5, 93-103.
Maslach, C, & Jackson, S, E. (1982), Burnout in health professions: A
social psychological analysis. In G, S. Sanders & J. Suls (Eds.), Social
psychology of health and illness (pp. 227-251). Hillsdale, NJ: Erlbaum.
Masiach, C. & Jackson, S, E. (1984), Burnout in organizational settings. Applied Social Psychology Annual, 5. 133-153.
Matthews, K.. A, (1982). Psychological perspectives on the Type A behavior pattern. Psychological Bulletin. 91. 293-323.
Matthews, K, A, (1988), Coronary heart disease and Type A behaviors:
Update on and alternative to the Booth-Kewley and Friedman (1987)
quantitative review. Psychological Bulletin. 104. 373-380.
McGrath, J, E., & Rotchford. N. L. (1983), Time and behavior in organizations. In B, M. Staw & L, L. Cummings (Eds,), Research in organizational behavior (Vol. 5, pp. 57-101). Greenwich, CT: JAi Press,
Moses, B., & Rothe, A. (1979). Nursepower, American Journal of Nursing. 81. \145-ll 56.
Munsterberg, H, (1913). Psychology and industrial efficiency. Boston:
Houghton-Mifflin.
Ohman. A., Nordy, H., & Svebak, S. (1985). Task-induced cardiac activation and components of the Type A behavior pattern. Psychophysiotogy 22, 606,
Rahe, R. H,. Hervig, L,, & Rosenman, R, H. (1978). Heritability of
Type A behavior. Psychosomatic Medicine, 40. 478-486.
Rosenman, R, H. (1978). The interview method of assessment of the
coronary-prone behavior pattern. In T. M. Dembroski, S, M, Weiss,
J. L. Shields, S, G, Haynes, & M. Feinleib (Eds,), Coronary-prone
behavior {pp. 55-70). New York: Springer-Verlag,
Schmitt, N.. & Stults, D, M. (1986), Methodology review: Analysis of
multitrait-multimethod matrices. Applied Psychological Measurement. 10, 1-22.
Schriber, J. B,, & Gutek, B. A, (1987). Some time dimensions of work:
Measurement of an underlying aspect of organization culture. Journal of Applied Psychology 72, 642-650.

Smith, P. C. & Kendall, L. M. (1963), Retranslation of expectation: An


approach to the construction of unambiguous anchors for rating
scales. Journal of Applied Psychology, 47. 149-155.
Smith. P. C, Kendall, L. M., & Hulin, C, L, (1969). The measurement of
satisfaction in vi'ork and retirement: .4 strategy for the study of atti-

tudes. Chicago: Rand-McNally.


Spence, J. T. Helmreich, R. L.. & Pred, R. S. (1987), Impatience versus
achievement strivings in the Type A pattern: Differential effects on
students' health and academic achievement. Journal of Applied Psychology. 72. 522-528,
Spence, J, T, Pred, R, S,. & Helmreich, R. L, (1989). Achievement
strivings, scholastic aptitude, and academic performance: A followup to "Impatience Versus Achievement Strivings in the Type A Pattern," 7(9wma/o/'/lpp//ct//'5>'(.'/io/o^>: 74. 176-178,
Spieiberger, C. D, Jacobs, G, A., Crane. R, G., Russell, S, F, Westberry,
L., Barker, L,, Johnson, E, H., Knight. J., & Marks, E. (1979), Preliminary manual for the State-Trait Personality Inventory (STPI). Tampa:
University of South Florida Human Resources Institute.
Stanley, J. C. (1961), Analysis of unreplicated three-way classifications,
with applications to rater bias and trait independence. Psychometrika, 26,205-219.
Thurstone, L. L. (1949). Thurstone Temperament Schedule. Chicago:
Science Research Associates.
Williams, R, B, (1984), An untrusting heart. The Sciences. 24. 31-36,
Williams, R. B.. Haney, T. L.. Lee, K. L., Kong, W, Blumenthal. J. A., &
Whalen. R. (1980). Type A behavior, hostility, and coronary atherosclerosis. Psychosomatic Medicine. 42. 539-549,
Williams, R. B., Lane, J, D. Kuhn, C M., Melosh. W, White, A. D, &
Schanberg, S. M. (1982), Type A behavior and elevated physiological
and neuroendocrine responses to cognitive tasks. Science. 2IS. 483485,
Wright, L, (1988). The Type A behavior pattern and coronary artery
disease. American Psychologist. 43. 2-14.
Received October 1,1990
Revision received April 29,1991
Accepted May 1,1991

Correction to Edwards, Baglioni, and Cooper (1990)


In the article "Exatnitiing the Relationships Among Self-Report Measures of the Type A Behavior Pattern: The Effects of Dimensionality, Measurement Error, and Differences in Underlying
Constructs," by Jeffrey R. Edwards, A. J. Bagliotii, Jr., and Cary L. Cooper (Journal of Applied
Psychology 1990, Vol. 75, No. 4, pp. 440-454). the loading for Item 8 of the JAS-AB in Table 4
(404) is misplaced. This loading should be utider the HC column, not the SI column.

You might also like