You are on page 1of 18

Journal of Applied Psychology

1999, Vol. 84, No. 4, 496-513

Copyright 1999 by the American Psychological Association, Inc.


0021-9010/99/S3.00

Flexible and Compressed Workweek Schedules:


A Meta-Analysis of Their Effects on Work-Related Criteria
Boris B. Baltes

Thomas E. Briggs, Joseph W. Huff, Julie A. Wright,


and George A. Neuman
Northern Illinois University

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Wayne State University

Meta-analytic techniques were used to estimate the effects of flexible and compressed
workweek schedules on several work-related criteria (productivity/performance, job satisfaction, absenteeism, and satisfaction with work schedule). In general, the effects of both
schedules were positive. However, the effects of both flextime and compressed workweek
schedules were different across the outcome criteria (e.g., compressed workweek schedules
did not significantly affect absenteeism). Thus, the level of positive impact associated with
either schedule is dependent on the outcome criterion under consideration. Further, several
variables were found to be moderators of flexible work schedules. For example, highly
flexible flextime programs were less effective in comparison to less flexible programs, and
the positive benefits of flextime schedules were found to diminish over time.

Alternative work schedules, such as flextime and compressed workweeks, have been adopted by an increasing
number of organizations over the past several decades
(Pierce & Dunham, 1992). A recent report that surveyed 1,035 organizations found that 66% offered flexible
work schedules (up 6% from the year before) and 21%
offered compressed work schedules (Hewitt Associates
LLC, 1995).
Much of the increased use of alternative work schedules
is due to societal changes, such as increasing numbers of
women in the workforce, dual-career households, and
work-leisure time expectations (Hochschild, 1997; Pierce,
Newstrom, Dunham, & Barber, 1989; Ronen, 1984). These
changes have increased employee demands for flexibility in
their work schedules so that they can better adjust to and
master life outside the workplace. The positive benefits of
these alternative work schedules for employees' quality of
life outside of work are well documented (Lee, 1983; Meij-

man, 1992; Ronen & Primps, 1981; Stevens & Elsworth,


1979; Thierry & Meijman, 1994). However, research results
regarding benefits to the employing organizations that have
implemented these alternative work schedules are far more
ambiguous (Pierce et al., 1989). This question is the primary
focus of the present study.
Organizational gains that are presumed to result from
alternative work schedules are many and diverse, but they
generally include increased employee job satisfaction, reduction of overtime, decreased absenteeism, and increased
productivity (deCarufel & Schaan, 1990; Pierce et al.,
1989). However, although originally assumed to have primarily positive effects on both the employee and the organization, alternative work schedules can have unintended
negative effects. These negative consequences include increased need for managerial planning, the inability of the
supervisor to be present at all times when employees are on
the job, and extra implementation costs (Coltrin &
Barendse, 1981). In addition, Nollen (1981) has proposed
that alternative work schedules may create problems with
interface and coverage with suppliers and customers, as all
units are not working on the same schedule.
Because organizational use of these alternative work
schedules is growing, it seems imperative that researchers
provide organizational leaders with the information needed
to determine whether an alternative work schedule would be
beneficial or detrimental. In this vein, the next section
addresses several problems with the current literature that
leave open the questions of if, when, and how these workschedule interventions are effective for organizations (Dun-

Boris B. Baltes, Psychology Department, Wayne State University; Thomas E. Briggs, Joseph W. Huff, Julie A. Wright, and
George A. Neuman, Department of Psychology, Northern Illinois
University.
A preliminary version of this article was presented at the 104th
Annual Convention of the American Psychological Association,
Toronto, Ontario Canada, August, 1996. We thank Rob Altmann
and Ken McGraw for their comments on a draft of this article.
Correspondence concerning this article should be addressed to
Boris B. Baltes, Psychology Department, 71 West Warren, Wayne
State University, Detroit, Michigan 48202. Electronic mail may be
sent to bbaltes@sun.science.wayne.edu.
496

497

WORK SCHEDULES

ham, Pierce, & Castaneda, 1987; Golembiewski & Proehl,


1978, 1980; Pierce et al., 1989).

mance, overall job satisfaction, absenteeism, and satisfaction with work schedule.

Overview of Alternative Work Schedules

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Problems With the Current Literature


There are several deficiencies within the available evidence that can be identified. First, with few exceptions, the
research has not been based on theoretical models (Pierce et
al., 1989; Thierry & Meijman, 1994). Second, researchers
have not always agreed in their interpretation of the evidence on the effects of alternative work schedules. Although
earlier literature reviews have revealed overall positive effects of certain alternative work schedules (Golembiewski
& Proehl, 1978, 1980; Ronen & Primps, 1981), a more
recent literature review (Dunham et al., 1987) revealed that
the results of studies on work-schedule interventions were
highly mixed. For example, the effect of flextime schedules
on work-related criteria (e.g., productivity) are highly variable and range from zero or little change to substantial
positive change (Dunham et al., 1987; Pierce et al., 1989).
These mixed results may point to the existence of moderators (e.g., employee type) that need to be identified to better
understand the relationship between these alternative work
schedules and various outcome measures. Third, much of
the literature is nonexperimental in nature and is "strongly
characterized by (1) anecdotal reports of flexible workinghour systems, (2) the use of nonstandardized research
scales, (3) failure to include statistical treatment of the
reported data, and (4) the absence of other systematic datacollection strategies" (Pierce & Newstrom, 1983, p. 247).
Consequently, the internal validity of the individual studies
included in previous literature reviews is questionable,
which could, in turn, affect the validity of each of the earlier
reviews. Finally, very few researchers have looked at alternative work schedules as multidimensional (Pierce et al.,
1989). For example, the particular design features of a
flextime schedule, such as amount of core hours, have rarely
been considered as moderators of the effectiveness of a
flextime intervention.
In the current study, we have attempted to address these
problems by introducing two substantial improvements over
previous research. First, we used two theoretical models and
prior research to formulate hypotheses about the effects of
these alternative work schedule interventions on various
outcomes. Second, we used quantitative meta-analytic techniques to more accurately assess the effects of flexible and
compressed workweek schedules on multiple criteria. Finally, we were able to assess how specific moderators
influenced the effects of both alternative work schedules. To
accomplish these goals, we collected data from experimental studies that examined the effects of flextime and/or
compressed workweek schedules on at least one of the
following four work-related criteria: productivity/perfor-

Alternative work schedules are schedules that do not fit


the fixed 8-hr day, 40-hr week definition. Examples of
common alternative work schedules include flexible working hours (flextime), compressed workweek, part time, and
telecommuting. This study considers two of the most common alternative schedules: flextime and compressed workweek. It may be helpful to the reader at this juncture to point
out that shiftwork arrangements, although alternative, were
excluded from this study because they do not always match
this definition of an alternative work schedule.1
Flextime
Under a flextime schedule, employees exercise a decision
regarding the time of day they will arrive at and leave from
work. The employer creates a band of core time where each
employee must be present (normally 9 or 10 a.m. to 2 or
3 p.m.). For example, a flexible work schedule where all
employees have to be present from 10 a.m. to 3 p.m. would
have 5 core hours. Employees are free to arrive before the
core start time and leave after the core finish time, but
typically there is a limit as to how early the employees can
arrive and how late they can leave (e.g., cannot start before
7 a.m. and cannot stay past 9 p.m.). Another important
flextime characteristic is the degree of carryover that is
permitted. Some organizations do not permit any carryover
of hours (i.e., the employee must work 8 hr per day),
whereas others permit carryover on a weekly basis (i.e., no
requirement for 8 hr per day but must work 40 hr per week),
and a few organizations even allow monthly carryover.
Our review of the literature revealed that flextime schedules are used almost exclusively in nonmanufacturing organizations. This may be because of the fact that a flextime
schedule is more difficult to implement in continuous process operations, such as assembly lines (Ronen, 1981). That
is, allowing employees to attend work at different times
does not mesh well with the interdependence required
among workers in a manufacturing setting.
Compressed Workweek
Under a compressed workweek schedule, the workweek
is compressed into fewer than 5 days by increasing the
1

Shiftwork schedules are often still 8-hr day 40-hr workweeks


with abnormal starting and ending times. Thus, many of these
schedules are not alternative according to our definition. Furthermore, these schedules introduce factors not applicable to other
alternative work schedules, such as constantly trying to adjust from
day to night shifts as the shift schedule dictates.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

498

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

number of hours an employee is required to work per day.


The most common form of compressed workweek in the
United States is the 4-day, 40-hr workweek (4/40), in which
employees typically work four, 10-hr days (Latack & Foster, 1985; Pierce et al., 1989). Commonly, employees will
have either Friday or Monday off, extending their weekend
to three days. In more recent years, 3/36, 3/38, and 3/40
schedules have been adopted by some organizations.
Our review of the compressed workweek literature revealed that these schedules are most commonly used in
manufacturing settings, which may be because of two reasons. First, because compressed workweek schedules still
require workers to attend work at the same time, they meet
the interdependence requirement of assembly line (i.e, manufacturing) settings. Second, manufacturing organizations
typically do not offer services that require employees to be
present at more regular time intervals (e.g., MondaySaturday) to serve customers.
Hypothesized Benefits of Alternative
Work Schedules
Early narrative reviews attempting to summarize the benefits of alternative work schedules failed to use a theoretical
model to formulate hypotheses. However, a few attempts to
construct theoretical models to explain the benefits of alternative work schedules have been made. Pierce and Newstrom (1980) used the work adjustment model (Dawis,
England, & Lofquist, 1968) to explain how flextime schedules influence employees' attitudes and behaviors. The
work adjustment model leads to the prediction that high
correspondence between an employee's abilities and the
ability requirements of the job should lead to high role
performance. Further, high correspondence between an employee's needs and the reinforcement system of the work
environment should lead to more positive job attitudes.
Work adjustment is high when individuals fulfill their work/
role requirements and the organization simultaneously fulfills the needs of the individual. More recent research has
modified the model so that moderating relationships also
exist between job attitudes and job performance (Dawis &
Lofquist, 1984).
Another theoretical model, which we believe can be used
to explain the effects of alternative work schedules, is
Hackman and Oldham's (1976) job characteristics theory.
The basis of this model is that core characteristics of the job
(e.g., autonomy, task identity, etc.) induce psychological
states that in turn lead to outcomes such as job performance
and job satisfaction. The introduction of alternative work
schedules can affect the core characteristics of a job and
thus work outcomes. For example, a flextime schedule
should positively affect employees' sense of autonomy,
which in turn increases job satisfaction.
To develop hypotheses regarding the effects of both

alternative work schedules on our four work-related criteria,


we drew on the aforementioned theoretical models, the
results of prior alternative-work-schedule research, and research findings from other areas (e.g., person-job fit, job
satisfaction, job performance, etc.). These are presented in
the following sections.
Flextime
Productivity/Performance
Using the framework of the work adjustment model,
Pierce and Newstrom (1980) suggested that flexible working schedules affect employees' performance in the following ways: They may allow individuals to make more efficient use of their own circadian rhythms (the normal 24-hr
physiological cycle) and may decrease the amount of stress
(e.g., work arrival related stress) experienced by employees.
Employees making more efficient use of their circadian
rhythms should result in a higher correspondence between
their abilities and the ability requirements of the job.
Person-job fit research supports the idea of congruence
between the individual and the job environment leading to
higher performance (Caldwell & O'Reilly, 1990; Chatman,
1988). Although research results on the relationship between job stress and job performance are mixed, it seems
safe to assert that there is a negative relationship between
negative reactions to stressful job conditions and job performance (Jamal, 1984, 1985; Parker & Kulik, 1995; Sullivan & Bhagat, -1992). Thus, if reduced job stress leads to
a reduction in negative reactions, then one should expect to
see an increase in job performance.
The implementation of a flextime schedule also gives
employees more job autonomy. Hackman and Oldham's
(1976) job characteristics theory would predict that increased job autonomy should lead to increased job performance. Indeed prior research has linked increased job autonomy to higher job performance (Dodd & Ganster, 1996;
Roberts & Foti, 1998).
Finally, prior research has indicated no decrements in
performance with the introduction of a flextime schedule
and some increases (Pierce et al., 1989). Thus, with respect
to productivity/performance, we expect that the introduction
of a flexible work schedule will have positive effects.
Absenteeism
Organizational attendance (lower absenteeism) should increase as the amount of discretionary time increases (Pierce
et al., 1989). Employees under a flextime schedule can more
easily respond to work-nonwork conflicts, which can reduce employee stress. Prior research has linked decreased
employee stress to decreased absenteeism (Parker & Kulik,
1995). Also, motivation to attend may be enhanced by
increased organizational loyalty and job satisfaction result-

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES

ing from the implementation of a flexible work schedule


(Pierce et al., 1989). Both organizational commitment and
job satisfaction have been positively linked to increased
organizational attendance (Gellatly, 1995; Somers, 1995;
Song, Daly, Rudy, Douglas, & Dyer, 1997) especially with
respect to voluntary absences (Sagie, 1998). Finally, misuses of sick leave may no longer be necessary because the
employee can adjust his or her time of attendance (Ronen,
1981). Prior literature has supported the hypothesis that
attendance is positively affected with the advent of a flextime schedule, with frequent dramatic drops in absenteeism
being reported by organizations (Pierce et al., 1989; Ronen,
1981). Thus, with respect to absenteeism we expect that the
introduction of a flexible work schedule will have positive
results (i.e., lower absenteeism).
Job Satisfaction and Satisfaction With Schedule
Both Pierce and his colleagues (1989), using the work
adjustment model, and Ronen (1981), using Herzberg's
taxonomy of needs, concluded that the introduction of a
flextime schedule should lead to more positive job attitudes
(i.e., job satisfaction and satisfaction with schedule). Several reasons exist for this conclusion. First, employees'
needs for autonomy/independence can be met by the introduction of a flextime work schedule that can help the
employee fulfill self-actualization needs (Ronen, 1981).
Ronen's theory coincides with Hackman and Oldham's
(1976) theory of job characteristics, which predicts that
increased autonomy leads to increased job satisfaction. Indeed, previous research has found that increased job autonomy is positively linked to job satisfaction (Fried, 1991;
Fried & Ferris, 1987; Roberts & Foti, 1998). More specifically, Macan's (1994) research indicated a positive relationship between employees' perceived control of time and
job satisfaction. Furthermore, prior altemative-workschedule research has, for the most part, supported the idea
that job attitudes are favorably affected by the introduction
of a flexible work schedule (Pierce et al., 1989; Ronen,
1984). Therefore, we expect that the introduction of a flextime work schedule will lead to increased job satisfaction
and satisfaction with schedule.
Compressed Workweek
Productivity/Performance
Using the circadian rhythm approach, Pierce and his
colleagues (1989) suggested that there are only a few hours
a day where employees enjoy their peak period and perform
at optimal levels. Thus, having employees work longer
hours (as is required in a compressed workweek work
schedule) should increase the amount of time they are
working at suboptimal levels. Within the framework of the
work adjustment model, this decrease in performance is

499

linked to lower congruence between the employee's abilities and the ability requirements of the job. Research on
person-job fit supports the notion that a decrease in personjob fit (i.e., decrease in congruence between the employee's
abilities and the ability requirements of the job) would lead
to decreased job performance (Caldwell & O'Reilly, 1990).
Furthermore, prior research has shown that fatigue increases
with the advent of a compressed workweek schedule
(Ronen, 1984), which also could negatively affect performance. In addition, if increased fatigue is associated with
increased employee stress, then one would expect to see a
decrease in productivity/performance.
The results of prior compressed-workweek-schedule research have been mixed (Pierce et al., 1989), with productivity either improving or staying the same after the implementation of a compressed workweek work schedule
(Ronen, 1984). Thus, although theoretically we would expect the implementation of a compressed workweek schedule to lead to lower productivity, prior research does not
support this claim. Because of these apparent contradictions, we felt it wiser to make no hypotheses regarding
the impact of a compressed workweek schedule on
productivity/performance.
Absenteeism
As with flextime, the advent of a compressed workweek
schedule should lead to more discretionary time, which in
turn should lead to increased organizational attendance.
Employees enjoying 3-day weekends should be better able
to balance work and nonwork demands. Being able to more
easily respond to work-nonwork conflicts should reduce
stress, and as stated earlier, decreased employee stress has
been linked to decreased absenteeism (Parker & Kulik,
1995). Furthermore, prior research strongly suggests that
employee absenteeism may decrease following the implementation of a compressed workweek study (Pierce et al.,
1989). Thus, we expect that the introduction of a compressed work schedule will have positive effects on
absenteeism.
Job Satisfaction and Satisfaction With Schedule
According to Ronen (1984), compressed workweek
schedules can affect job attitudes by enhancing or facilitating production. Specifically, "increases in responsibility,
autonomy, and job knowledge resulting from implementing
the schedule may be associated with more positive attitudes
toward the job itself (Ronen, 1984, p. 57). As mentioned
earlier, Hackman and Oldman's (1976) model would predict that positive changes in these types of job characteristics lead to higher job satisfaction. Prior research has shown
mixed results, but in general positive changes in job attitudes can be expected with the implementation of a com-

500

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

pressed workweek work schedule (Pierce et al., 1989).


Thus, we expect that job satisfaction and satisfaction with
schedule will be positively affected by the advent of a
compressed workweek work schedule.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Moderators
Considering the evidence provided by primary-data studies, it seems likely that a number of variables may moderate
the relationships between alternative work schedules and
the previously mentioned criteria. Moderators that have
been suggested by earlier researchers include flexibility of
the flextime schedule (Pierce & Newstrom, 1983) and time
since schedule implementation (Ivancevich & Lyon, 1977).
Furthermore, we hypothesized that employee type and
methodological rigor could function as moderators.2 We
have used our conceptual framework and prior research to
formulate some hypotheses regarding the effects of our
moderators. However, these hypotheses are exploratory in
nature.
Employee Type
The effects of alternative work schedules may vary as a
function of employee type. Specifically, managers and professionals may be less affected by schedule interventions
than general employees (i.e., blue-collar, administrative
support, service employees, etc.), particularly if they already possess a large amount of autonomy regarding their
work schedules before the introduction of the alternative
schedule. The theoretical underpinning for this effect is that
because the managerial employees may already possess
freedom in their schedule, the official implementation of a
flexible or compressed workweek work schedule would not
increase the correspondence of the work environment to
their needs. That is, managers' working conditions already
satisfy their need for autonomy, and thus the introduction of
a formal alternative work schedule may not lead to higher
levels of need satisfaction. Therefore, it is hypothesized that
managers and professionals will be less affected by alternative schedules than general employees.
Flexibility of the Flextime Schedules
With respect to flextime interventions, high amounts of
flexibility (e.g., fewer daily core hours), coupled with employees having the option to change the pattern of hours
worked without management approval, may produce more
positive effects than less flexible flextime schedules (Pierce
& Newstrom, 1983).3 That is, increased flexibility may lead
to a higher correspondence between employee needs (e.g,
need for autonomy) and the work environment and thus
increase the positive effects on various outcome criteria
(e.g., job satisfaction, job performance, absenteeism, etc.).
Increased flexibility should lead to lower levels of employee

stress also enhancing any positive outcomes. Therefore, it is


hypothesized that more flexible flextime interventions will
lead to larger positive effects than less flexible flextime
interventions.
Time Since Schedule Intervention
The amount of time that the program has been in place at
the time its effects are ascertained is likely to be an important moderator. Ivancevich and Lyon (1977) found that the
long-term impact of a 4/40 workweek was not as positive as
the short-term impact. Prior research has demonstrated that
extrinsic rewards may have a temporary effect on employees (Ronen, 1981). That is, the effects of a work schedule
intervention may wane over time as employees eventually
become accustomed to the new amount of freedom. The
employees may adjust their perceptions and desire even
more discretionary time (i.e., an increase in needs), thereby
decreasing the level of correspondence between employee's
needs and the reinforcement system of the work environment that was initially attained. Thus, the benefits of alternative work schedules that are perceived as extrinsic (e.g.,
improved job conditions) may have temporary effects.
Therefore, it is hypothesized that the effects of alternative
schedules should decrease over time.
Methodological Rigor
The magnitude of the observed effects of alternative work
schedules on work-related criteria may vary as a function of
methodological rigor of the studies evaluating these effects.
Some researchers have suggested that low methodological
rigor may attenuate the size of effects observed in studies
(e.g., Bullock & Svyantek, 1983), whereas others have
suggested that low experimental rigor may inflate the size of
effects observed in studies (e.g., Terpstra, 1981). Given
these mixed results, no specific hypothesis is proposed.
However, we felt that an investigation of methodological
rigor might add to the extant literature.
2

Type of intervention (flextime vs. compressed workweek) was


also considered a moderator but was dropped from our analyses
because type of organization was confounded with intervention
type. The flextime studies included in these analyses were done
almost exclusively in nonmanufacturing environments, whereas
the compressed workweek studies were done predominantly in
manufacturing environments (confounding type of organization
with intervention type).
3
Other dimensions of a flexible work schedule, such as carryover, variability of employees schedule, and supervisor's role, may
also be important in determining the flexibility of a flexible work
schedule. Unfortunately, the studies included in this meta-analysis
only allowed us to consider core hours as a moderator.

501

WORK SCHEDULES

Method

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Sample of Studies
Computer-based literature searches were conducted on Psychological Abstracts (PsycLIT: 1974 to February, 1997), ABI/INFORM (1977 to February, 1997), Business Periodicals Index
(1977 to January, 1997), and Dissertation Abstracts (1891 to
March, 1997). These searches were conducted using the following
key terms: alternative work schedules; all forms of compressed
workweek (e.g., workweek, work week, 3/36, 3/38, 4/40); and all
forms of flextime (e.g., flexitime, flextime, flex time). A manual
search of all articles uncovered by the broad term work schedules
was also conducted. Reference lists of numerous review articles,
books, and chapters of books were searched, as well as the reference lists of all located studies. Finally, 20 larger U.S. corporations
who were reported to use alternative work schedules were contacted for possible data.

Inclusion Criteria for Studies


Studies were initially selected if they (a) evaluated a flextime
and/or compressed workweek schedule and (b) included a prepost, control-experimental, or normative-experimental comparison. This was done to ensure that all effect sizes included in this
meta-analysis reflected the difference between an experimental
group working under an alternative work schedule and a control
group not working under an alternative work schedule. Thus, the
effect sizes calculated would indicate the effect that each intervention had on work-related criteria. Unfortunately, because many of
these studies are qualitative in nature and often based on anecdotal
evidence, these criteria resulted in the loss of a large amount of the
original studies. Our problem with finding experimental alternative
work schedule studies is not unique. For example, Ralston, Anthony, and Gustafson (1985) evaluated over 100 flextime studies
and found that only a few met their experimental criteria. Furthermore, from the responses received from the corporations we contacted, we have deduced that many (if not most) businesses have
not conducted formal evaluations of their alternative work schedules. Finally, only studies that had the necessary statistics to be
included in a meta-analysis could be used.
Our final criteria for inclusion in the sample resulted in a total
of 29 published sources and 2 unpublished sources remaining for
analysis, several of which contained more than one study. Several
sources in the sample had examined the effects of alternative work
schedules across different employee groups (e.g., professionals,
managers, general employees) and across different units or divisions of an organization. In these cases, separate effect sizes were
calculated and entered into the analyses. Also, some sources had
examined the effects of alternative work schedules across time
(e.g., 3 months, 6 months, and 12 months). As time since intervention was a key moderator of interest in the meta-analysis, a
dichotomy of 6 or less months from intervention to data collection
(i.e., short) and more than 6 months from intervention to data
collection (i.e., long) was used to calculated effect sizes. Six
months was chosen as the cut point because it provided for a nearly
even split between the number of studies that fell into both the
short and long time since intervention groups. If multiple effect
sizes from a study fell into either time frame, they were averaged

into a single effect size; otherwise, separate effect sizes were


entered into the analyses (it should be pointed out that only one
study had such longitudinal data). Finally, in one study (Dunham
et al., 1987), the authors had evaluated the effects of both a
flextime intervention and a compressed workweek intervention
(using different samples), and therefore, separate effects were
calculated for these samples. As a result, the original 31 sources
were coded into 39 separate substudies. Because these 39 substudies measured multiple work-related criteria, we were provided
with 69 effect sizes. The key characteristics of the 39 substudies
are presented in Table 1. Although our inclusion criteria did result
in a large number of studies being excluded from our metaanalysis, we believe that the organizations in the studies we included make up a diverse and representative sample. For example,
within both schedule interventions, we had governmental versus
nongovernmental organizations, and the employee type varied
from managers to blue-collar workers. Furthermore, within flextime we had organizations representing the finance, insurance, and
governmental sectors. Also, from our own knowledge of the studies both included and excluded from the study, no discerning factor
was observed that would distinguish organizations that did conduct
formal evaluations from those that did not.

Variables Coded From Each Study


The following information was coded from each report: (a)
outcome criteria, (b) sample size, (c) alternative work schedule
introduced (flextime, compressed), (d) time since schedule implementation at the time of evaluation (6 months or less, more than 6
months), (e) flexibility of the flextime schedule (less than 5 core
hours';' 5 or more core hours), (f) sample employee type (general
employees, professional or management, mixed), and (g) methodological rigor (high, low).4
The outcome criteria from the studies were coded into the
following dimensions: productivity, supervisor-rated performance, self-rated performance, absenteeism, job satisfaction,
and satisfaction with schedule. Productivity, supervisor, and
self-performance ratings were coded as separate outcome criteria because of the low correlations found between these
different types of productivity/performance measures in previous research (Conway & Huffcutt, 1997; Hoffman, Barry, &
Holden, 1991).5 Criteria in the productivity dimension
included only objective criteria (e.g., amount of claims processed, etc.) measured at either the employee or unit level.6 The
supervisor and self-rated performance dimensions were made
4

Although many other variables were initially coded from each


study (e.g., location of sample, gender ratio of sample), they were
not included in the analyses because of the small number of studies
that had provided this information.
5
We thank an anonymous reviewer for pointing out that type of
performance criteria would be a logical moderator.
6
Individual and unit level productivity data were combined
because of the small number of effect sizes associated with productivity. The effect sizes associated with productivity in both
work schedules were homogenous (see Tables 3 and 4). This
indicated that no statistical differences existed between productivity effect sizes measured at the individual and unit levels.

502

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

Table 1
Summary of Characteristics of Alternative Work Schedule Substudies

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Characteristic
Type of organization
Manufacturing
Other
Employee type
Employees
Man ./prof.
Mixed
Unknown
Criteria
Productivity
Performance supervisor-rated
Performance self-rated
Job satisfaction
Absenteeism
Satisfaction with schedule
Degree of flexibility

Low
High

Unknown

All substudies
(n = 39)

Flextime
(n = 27)

Compressed
(n = 12)

9
30

1
26

8
4

20
3
15
1

15
3

5
0
7

9
4
5
25
13
13

8
1

5
17
8

9
7

16
4

0
4
4

8
5
4

Experimental rigor

Low
High
Time since intervention
Short (6 months or less)
Long (more than 6 months)
Unknown

15
24

18

16
19
4

13
11
3

6
6
3

8
1

Note. One study (Dunham, Pierce, & Casteneda, 1987) is represented twice, once in flextime and once in
compressed workweek. Man./prof. = manager/professional.

up of various subjective scales. All absenteeism criteria were


objective and consisted of company attendance records. Finally,
the job satisfaction and satisfaction with schedule dimensions
included various self-rating scales.
Methodological rigor was determined by the design of the study.
Specifically, each study was coded as one of the following 14
designs: (1) one sample t (i.e., normative-experimental comparison); (2) unmatched, experimental-control; (3) unmatched, longitudinal, experimental-control; (4) matched, experimental-control;
(5) matched, longitudinal, experimental-control; (6) pre-post with
different samples; (7) pre-post with the same samples; (8) prepost, longitudinal; (9) pre-post, experimental-control with a focus
on the pre-post comparison; (10) interrupted time series; (11)
pre-post, experimental-control with a focus on the interaction
effect; (12) experimental-control comparison of pre-post difference scores; (13) experimental-control comparison of pre-post
difference scores, longitudinal; and (14) pre-post, experimentalcontrol with pre scores serving as a covariate.
We generated a measure of experimental rigor that was based on
the experimental design used and the method of sample selection.
First, the above designs were reduced to five categories of increasing rigor. Specifically, Designs 1-3 were given a rigor score of 1;
Designs 4-6 were given a rigor score of 2; Designs 7-8 were
given a rigor score of 3; Designs 9-10 were given a rigor score of
4; and finally, Designs 11-14 were given a rigor score of 5.
Second, a rigor score of 1 was added to the existing rigor score if
the sample selection was random or representative random. Given
the low number of studies included in the sample, these scores

were further reduced to a low versus high rigor dichotomy, with


the low-rigor category consisting of studies assigned a rigor score
of 3 or less, and the high-rigor category consisting of studies
assigned a rigor score of 4 or more.
The studies were coded by three of the authors, and intraclass
correlation coefficients (Shrout & Fleiss, 1979) were calculated for
all the moderators used in the analysis. These correlation coefficients were as follows: amount of flexibility (1.00), time since
intervention (.88), methodological rigor (1.00), and employee type
(.91). All disagreements were resolved through discussion.

Meta-Analytic Procedures
Computation of Effect Sizes and Outlier Analysis
The first step in the analysis involved converting the results of the
various studies to a common statistic. The results were converted
to 69 Pearson correlations (r), reflecting the degree and valence
(positive vs. negative) of the relationship between the type of schedule
(i.e., standard vs. alternative) and the work-related criteria. These
conversions were done with Johnson's (1993) DSTAT computer
program. These Pearson correlations were then converted into a total
of 69 d statistics using the aforementioned program. A list of the
Pearson correlations calculated for each of the 39 substudies in each
of the six criteria and information about the attributes used in our
moderator analysis are provided in Table 2.
The computation of r was based on (a) Fisher's F ratio or t tests
for 50% of the effects; (b) means and standard deviations or error

503

WORK SCHEDULES
Table 2
Study Characteristics and Effect Sizes for Alternative Work Schedules
Effect size (r)

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Study
Bohen & Viveros-Long (1981)
Calvasina & Boxx (1975)A
Calvasina & Boxx (1975)B
Coston(1973)
Dalton & Mesch (1990)
Dalton & Todor (1984)
Dunham, Pierce, & Casteneda (1987)A
Dunham, Pierce, & Casteneda (1987)B
Evans (1975)
Golembiewski & Hilles (1977)
Golembiewski, Yeager, & Hilles (1975)A
Golembiewski, Yeager, & Hilles (1975)B
Goodale & Aagaard (1975)
Harvey & Luthens (1979)
Hausser (1980)
Hicks & Klimoski (1981)
Hodge & Tellier (1975)
Ivancevich (1974)
Ivancevich & Lyon (1977)A
Ivancevich & Lyon (1977)B
Kim & Campagna (1981)
Krausz & Freibach (1983)
Maklan (1977)
McGuire & Lira (1986)
McGuire & Liro (1987)
Millard, Lockwood, & Luthans (1980)
Morgan (1977)
Narayanan (1982)
Narayanan & Nath (1984)A
Narayanan & Nath (1984)B
Narayanan & Nath (1984)c
Orpen (1981)
Ralston (1989)
Ralston & Flanagan (1985)1
Ralston & Flanagan (1985)2
Schein, Maurer, & Novak (1977)A
Schein, Maurer, & Novak (1977)B
Venne (1993)
Welsch & Gordon (1980)

Prod.

Sup.
perf.

Self
perf.

Job
sat.

Abst.

Sch.
sat.

.000

FX

.086
.081
.111
.075

CW
CW
CW
FX
FX
CW
FX
FX

.078
.001
.134

.343
.423
.054

.231
.392
.714

FX

.020
.094
.277

.278
.380

.000
.503
.018

.077
.463
.569
.056
.497

.310
.006
.332

.129
.000
.076

.019
-.012
-.082
.060
-.197
.348
.261

.180
.330
.049
.134
-.078
.061

.205
.000

Sch.3
int.

-.073
-.142

.040
.010
.019
.041
.356
.328

.218
.177

.870

...

.292
.005
.000

FX

FX
CW

FX

high
low

FX
CW
FX
FX
CW
FX
FX
FX

FX
FX

FX
FX
FX

FX
.075

.099

high
high
high
high
high
high
high
high
low
low
low
high
high

CW

FX

.213

high
low
high
low
high

high
low

FX

.276
.460

high

FX
FX
FX
CW
CW
CW
CW

FX

.840
.859

Core
flex."

Timec
long
long
short
long
long
short
short
short
long
long
long
long
short
long
long
long
long
long
long
short
long

long
short
short
short
short
short
short
short
long
short
long
short
short
long

Org
type"
O
M
M

M
0
O
O
O
0
O

O
O

M
O

O
O
O
M
M
M
O

0
M

O
O

0
M

O
O
O

0
O

O
O

0
O

O
O
O

Job

type"

Rigor5

EMP
BMP
EMP
EMP
EMP

low
low
low
low
high
high
high
high
low
high
high
high
low
high
high
low
low
high
high
high
high
low
low
low
high
high
low
high
high
high
high
high
high
high
high
low
low
high
low

EMP
MIX
MIX
EMP
MIX
EMP
M/P
MIX

EMP
MIX
EMP
MIX
MIX

MIX
MIX
MIX

EMP
MIX
MIX
MIX
MIX
EMP
EMP
M/P
M/P

EMP
MIX
EMP
EMP
EMP
EMP
EMP
EMP

Note. Positive effect sizes refer to positive effects of intervention (i.e., absenteeism effect sizes have been reversed). Superscript A, B, C refer to different
samples of participants (e.g., different units, different divisions). Superscript 1 and 2 refer to different posttest times (e.g., 6 months, 12 months). Dashes
indicate cells in which data were applicable but not obtained. Prod. = production; sup. perf. = supervisor performance; sat. = satisfaction; abst. =
absenteeism; sch. sat. = schedule satisfaction.
a
Schedule intervention: FX = flextime; CW = compressed workweek. b Flexibility: high = flextime core hours of 5 or less; low = flextime core hours
of more than 5. c Time since intervention: short = 6 months or less since intervention; long = more than 6 months since intervention. d Organization
type: M = manufacturing, O = other. e Job type: EMP = employee; M/P = manager/professional; MIX = mixed group. f Experimental rigor: on the
basis of experimental design and participant selection method.

terms for 18% of the effects; (c) means and estimated error terms
for 16% of the effects; (d) proportions of standard and alternativework-schedule participants using extreme category responses on
measures with two or more response categories for 10% of the
effects (in cases where frequencies were reported on a strongly
satisfied to strongly dissatisfied scale, proportions were calculated
by comparing the extreme category of strongly satisfied to all other
categories combined); and (e) chi square for 2% of the effects.7 In
three cases, the authors stated that there were no significant differences without reporting any statistic or providing information
that would allow for the calculation of an effect size. In these

cases, a Pearson correlation (r) of zero was used. It should be


noted that in all of these cases, the sample size exceeded 100;
therefore, it can be assumed that the value of zero was a relatively
accurate estimate of the actual effect size obtained. There were

If several comparisons were conducted using the same criterion, more accurate estimates of the Pearson correlations were
obtained by estimating the standard deviation from significance
levels and means and then using the lowest estimate of the standard deviations to estimate the effect sizes.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

504

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

also two cases where the authors stated that the results were
statistically significant without reporting any statistic that would
allow for the calculation of an effect size. In cases where a
significance level was not provided, a significance level of .05 was
assumed, and this significance level and the sample size were used
to estimate the effect size (Glass, McGaw, & Smith, 1981). In
cases where the actual significance level was provided (e.g., .01,
.001), these values and the sample sizes were used to estimate the
Pearson correlation (r).
To estimate the relative stability of unbiased effect-size magnitudes, separate schematic plot analyses were conducted (Light,
Singer, & Willett, 1994) for each criterion variable in the flextime and
compressed workweek samples, as recommended by Hedges and
Olkin (1985). No outliers or extreme values were found in the compressed workweek sample. However, three outliers were found in the
flextime sample: one outlier in productivity, one in job satisfaction,
and one in schedule with satisfaction. Because these outliers made up
such a small percentage of our original sample they were eliminated
from the remaining analyses to ensure that any moderator effects that
were found were not of a spurious nature.

effects' of moderators were conducted first. Then, additional


weighted multiple-regression analyses were conducted using the
moderators, as well as different work-related criteria as predictors,
to ensure that the categorical differences associated with each
moderator were unique. Furthermore, by placing all the moderators into a regression model, we were able to test whether controlling for other moderators would reveal the true relationship between any particular moderator and our study effect sizes.
The homogeneity of within-class effect sizes, as well as the
significance of the between-class effects were assessed using
Hedges and Olkin's (1985) statistical procedures, which are incorporated into Johnson's (1993) DSTAT computer program. The
between-class effect was estimated by QB, which has a chi-square
distributionp 1 degrees of freedom (p is the number of classes).
The homogeneity of effect sizes within each class was estimated
by Qw, which also has a chi-square distribution with k 1 degrees
of freedom (k is the number of effect sizes within the class).

Statistical Methods

Flextime

The overall progression of analyses was based on Hedges and


Olkin's (1985) approach to meta-analysis. Categorical model analyses of the effects within schedule intervention type as well as the

All of the results with respect to the flextime categorical


analysis that are mentioned in the following sections are
presented in Table 3.

Results
Categorical Analyses

Table 3
Effects of Flextime Work Schedules on Positive Work Outcomes With Between- and WithinHomogeneity Tests Across Study Characteristics

Moderator

Mean
weighted
effect
size (d)

Overall
Type of work-related
criteria
Productivity
Performance self-rated
Absenteeism
Job sat.
Sat. with schedule
Employee type
Employee
Manager/prof.
Degree of flexibility0

41

4,492

0.30

95% CI forrf

0.26

0.35

Mean
weighted
(r)

(a,r

.15

(sjb

1004.55**
193.57**

4
5
8
16
8

316
563
554

0.45
0.04
0.93
0.15
0.32

0.26
-0.06
0.83
0.09
0.20

0.64
0.14
1.03
0.21
0.44

.22
.02
.42
.07
.16

3,936
556

0.41
0.01

0.36
-0.09

0.46
0.11

.20
.01

Low

11

1,298

High

27

2,617

0.49
0.28

0.40
0.23

0.57
0.34

.24
.14

Time since schedule


intervention"1
Short
Long
Methodological rigorc

Homogeneity tests

1,034
2,025

1.57
7.08
728.95**
48.90**
24.50*
47.99**

31

913.49**
7.37
16.21**
503.17**
462.63**

0.66
22
15

2,037
1,753

0.35
0.30

0.29
0.23

0.41
0.38

639.25**

.17
.15

343.00**
28.39**

Low

10

High

31

1,356
2,136

0.11
0.37

0.03
0.32

0.19
0.42

.05
.19

38.28**
937.89**

Note. Significant effect sizes are indicated by confidence intervals that do not include 0. Positive effect sizes
refer to positive effects of intervention (i.e., effects for absenteeism have been reversed), k = number of effect
sizes; CI = confidence interval; sat. = satisfaction; prof. = professional.
a
Significance indicates effects differ as a function of study characteristics. b Significance indicates rejection of the
hypothesis of homogeneity. c High = less than 5 core hours; low = 5 or more core hours. d Short = 6 or less
months since intervention; long = more than 6 months since intervention. e Low = low rigor; high = high rigor.
* p < .01. **p < .001.

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES

Work-related criteria. Most hypotheses regarding the


effect of flextime work schedules on work-related criteria
were supported as indicated by positive effect sizes whose
95% confidence intervals did not include zero. Flexible
work schedules favorably influenced productivity, job satisfaction, absenteeism, and satisfaction with work schedule.
However, flexible work schedules did not seem to have an
effect on self-rated performance. Furthermore, the test of
the categorical model for type of work-related criteria was
significant, x*(4, N = 41) = 193.57, p < .001, indicating
that the various work-related outcomes are affected quite
differently by the introduction of a flextime work schedule.
Absenteeism was the most effected by the introduction of a
flextime schedule (d = .93), whereas self-rated performance was the least affected (d = .04).
Employee type. As predicted, the three flextime studies
that included managers and professionals revealed an effect
that was not significantly different from zero (d = .01).
Furthermore, the <2W statistic for the flextime work schedule
indicates that this result is homogenous across studies, and
thus, a reliable estimate of the effect size, ^(7, N =
8) = 7.37, ns. Thus, it would appear that managers and
professionals were not appreciably affected by the implementation of flextime work schedules. Employees, on the
other hand, were affected by the advent of a flextime work
schedule (d = .41).
Flextime flexibility. Flextime interventions can vary
as a function of the number of core hours of mandatory
attendance (see footnote of Table 2). The number of core
hours provides a good operational definition of the degree of flexibility of the flextime schedule, and one of the
research questions of interest to us was whether increased
flexibility enhances organizational and employee outcomes. Contrary to our hypothesis, less flexible schedules (5 or more core hours) resulted in larger effect sizes
than more flexible schedules (less than 5 core hours)
across all positive work outcomes, ^(1, N = 38) = 16.21,
p < .001.8
Time since implementation. Most human resource practitioners should be concerned that the effects of an intervention may be short-lived. An intervention that initially
creates positive results can return to baseline levels over
time. Contrary to our hypotheses, however, the short interval flextime intervention effects were not larger than the
long interval flextime intervention effects across all positive
work outcomes, x*(l, N = 37) = 0.66, ns.
Methodological rigor. High-rigor studies showed larger
effect sizes than low-rigor studies across all positive work
outcomes, ^(1, N = 41) = 28.39, p < .001. Thus, it would
appear that high-rigor studies allowed researchers to find
stronger effects across all the outcome criteria included in
the flextime studies.

505

Compressed Workweek
All of the results with respect to the compressed workweek categorical analysis that are mentioned in the following sections are presented in Table 4.
Work-related criteria. Support for our hypotheses regarding the effects of compressed workweek schedules on
work-related criteria were mixed. Compressed workweek
schedules positively affected supervisor performance ratings, job satisfaction, and satisfaction with work schedule
but did not affect productivity. Contrary to our hypotheses,
however, absenteeism was not significantly affected. The
test of the categorical model for type of work-related criteria
was significant, ^(4, N = 25) = 74.85, p < .001. This
indicates that the introduction of a compressed workweek
schedule influences the various work-related criteria
differently.
Time since implementation. Contrary to our predictions,
a reduced effect of length of intervention was not found in
the compressed workweek interventions, ^(1, N =
23) = 0.04, ns.
Methodological rigor. There was not significant difference between high- and low-rigor compressed workweek
studies, ^(1, N = 25) = 1.12, ns. However, because of the
small number of low-rigor studies, this result should be
interpreted cautiously.
Weighted Regression Analyses
To determine which study characteristics were uniquely
related to our effect sizes, we conducted weighted
multiple-regression analyses. These regressions are
weighted because the variances of each individual effectsize estimate are inversely proportional to the sample size
of the study (Hedges & Olkin, 1985). The study characteristics are entered as predictors, effect size (d) as the
criterion, and w as the weighting factor. Thus, this analysis gives more weight to effect sizes that are estimated
more reliably. Because the standard errors for the regression coefficients were incorrect, by a factor of the square
root of the residual mean square (see Hedges & Olkin,
1985, p. 174), they needed to be corrected using Johnson's (1993) DSTAT program. The unstandardized partial regression coefficients from the multiple-regression
analyses indicate the association of each study characteristic with the ds, while statistically controlling for the
other variables in the regression analyses.
The study characteristics were dummy coded in both
regression analyses with four orthogonal dummy vectors
8
Positive work outcomes refers to an analysis of effects across
all criteria (productivity, job satisfaction, absenteeism, and
satisfaction with schedule) after reversing the effect sizes for
absenteeism.

506

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

Table 4
Effects of Compressed Work Schedules on Positive Work Outcomes With Betweenarid Within-Homogeneity Tests Across Study Characteristics

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Study characteristics
Overall
Type of work-related
criteria
Productivity
Performance sup.
rated
Absenteeism
Job satisfaction
Sat. with schedule
Time since schedule
intervention
Short"
Long
Methodological rigor
Lowb
High

Total
n

Mean
weighted
effect
size (d)

25

2,921

0.29

95% CI ford

0.23

0.34

Mean
weighted
W

Homogeneity tests

(QbY

.14

(2Jd
210.58**

74.85**
4
4

770
312

0.04
0.42

-0.07
0.27

0.15
0.57

.02
.21

2.51
16.09*

5
8
4

507
855
477

0.01
0.59
0.40

-0.13
0.48
0.25

0.14
0.69
0.55

.00

87.71**
21.17**
8.26*

.28
.19
.04

8
15

490
1,883

0.29
0.27

0.16
0.21

0.42
0.34

.14
.14

8
17

1,089
1,832

0.25
0.31

0.16
0.24

0.34
0.38

.13
.15

28.21**
173.16**
1.12

64.65**
144.82**

Note. Significant effect sizes are indicated by confidence intervals that do not include 0. Positive effect sizes
refer to positive effects of intervention (i.e., effects for absenteeism have been reversed), k = number of effect
sizes; CI = confidence interval; sup. = supervision; sat. = satisfaction.
"Time since intervention: short = 6 or less months since intervention, long = more than 6 months since
intervention. b Rigor: low = low rigor, high = high rigor. c Significance indicates effects differ as a function
of study characteristics. d Significance indicates rejection of the hypothesis of homogeneity.
*p<.0l. **/?<.001.

representing the different work-related criteria. This


dummy coding allowed us to make more specific comparisons between the various work-related criteria than
was allowed in the categorical analyses. The behavioralattitudinal vector (vector 1) compared the two behavioral
criteria (productivity and absenteeism) against the two
attitudinal criteria (job satisfaction and satisfaction with
schedule). The attitudinal vector (vector 2) compared the
two attitudinal criteria (job satisfaction and satisfaction
with schedule), and the behavioral vector (vector 3) compared the two behavioral criteria (productivity and absenteeism). Finally, the productivity/performance vector
(vector 4) compared the productivity criterion to the
performance criterion. This comparison was possible because in each regression only one performance criterion
(self ratings vs. supervisor ratings) was present.
Finally, the assumption of independence was violated in
both regressions (i.e., multiple effect sizes from the same
study), which can lead to increased Type-1 error rate. To
address this issue, we calculated an adjusted alpha level as
recommended by Stevens (1996). For both regression analyses, an alpha level of .01 was required. It should be
mentioned that this method was very conservative because
only half (53%) of our 39 substudies had more than one
effect size, only a few (20%) had more than two effect sizes,
and none had four effect sizes.

Flextirhe
In the fiextime regression analysis (see Table 5) all but
one of the study characteristics had significant regression
coefficients. Managers and professionals were less affected
by fiextime schedules; less flexible schedules resulted in
larger effect sizes than more flexible schedules; high-rigor
studies showed larger effect sizes than low-rigor studies,
and the various work-related criteria were affected quite
differently by the introduction of a fiextime work schedule.
Specifically, behavioral outcomes were more greatly effected than attitudinal outcomes, productivity/performance
effect sizes were larger than effect sizes associated with
absenteeism, and effect sizes associated with productivity
were greater than effect sizes measured through self-rated
performance scales. Thus, the regression results supported
our categorical analyses with one important exception. The
regression analyses found that the time since schedule intervention produced a significant negative unstandardized
regression coefficient (B = -.60, p < .001). The negative
value indicates that lower effect sizes are found as the time
of criterion measurement, after the intervention is introduced, increases. It seems that controlling for other study
characteristics has allowed time since schedule intervention
to explain a significant amount of variance in our effect
sizes. Follow-up partial correlation analyses indicated that
when one controls for degree of flexibility a significant

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES
effect of time since schedule intervention on the flextime
effect sizes occurs.9 This result, not found in the categorical
analysis, demonstrates the importance of conducting
weighted multiple-regression analyses. Categorical analyses
do not allow researchers to discover such interactions
among moderators.
The QR statistic indicated that a substantial and significant proportion of variance (40%) in the effect-size estimates is explained by the study characteristics. However,
the QE statistic was also significant, indicating that a significant proportion of the variance in the effect sizes was not
explained by the study characteristics. This result indicates
that other moderators may exist.

Table 6
Partial Multiple Regression Coefficients for Study
Characteristics Predicting Positive Work Outcomes in
Compressed Workweek Studies (k = 21, N = 2,373)
Positive work outcomes
Predictor

ft

Time since schedule intervention


Methodological rigor
Vector 1
Vector 2
Vector 3
Vector 4
Intercept

-.01
-.32
-.22**
.15**
.10
-.37**

-.01
-.35
-.52
.22
.17
-.55

Overall R2
Compressed Workweek

Qr

In the compressed-workweek weighted-regression analysis (see Table 6), all but three of the study characteristics
had significant regression coefficients. Thus, the regression
analyses supported the results of our compressed-workweek
categorical analysis. That is, short interval intervention effects were similar to the long interval intervention effects;
high-rigor and low-rigor studies are similar; and various

Table 5
Partial Multiple Regression Coefficients for Study
Characteristics Predicting Positive Work Outcomes in
Flextime Studies (k = 36, N = 3,790)
Positive work outcomes
Predictor

ft

Employee type
Degree of flexibility
Time since schedule intervention
Methodological rigor
Vector 1
Vector 2
Vector 3
Vector 4
Intercept

-.24**
-.94**
-.60**
.96**
.22**
-.03
-.33**
49**
1.40

-.13
-.57
-.39
.47
.29
-.03
-.20
-.32

Overall R2
Qr

.40
387.23**
591.40**

Note. In the dummy vector design for degree of flexibility, low flexibility = 0 and high flexibility = 1. For time since schedule intervention,
short = 0 and long = 1. For methodological rigor, low = 0 and high = 1.
In the four dummy vectors for type of positive work outcome, productivity
and absenteeism = 1 and job satisfaction and satisfaction with schedule =
1 in Vector 1 (comparing behavioral vs. attitudinal outcomes). In Vector 2, productivity and absenteeism = 0 and job satisfaction = 1 and
satisfaction with schedule = 1 (comparing attitudinal outcomes). In
Vector 3, productivity = 1; absenteeism = -1; job satisfaction = 0;
satisfaction with schedule = 0 (comparing behavioral outcomes). In Vector 4, productivity = 1; self-performance ratings = 1; absenteeism = 0;
job satisfaction = 0; satisfaction with schedule = 0. k = number of effect
sizes; B = unstandardized partial multiple regression coefficient; )3 =
standardized partial multiple regression coefficient.
**p < .001.

507

Qe

.88
.52
101.33**
93.84**

Note, k = number of effect sizes. In the dummy vector design for time
since schedule intervention, short = 0 and long = 1. For methodological
rigor, low = 0 and high = 1. In the four dummy vectors for type of positive
work outcome, productivity and absenteeism = 1 and job satisfaction and
satisfaction with schedule = -1 in Vector 1 (comparing behavioral, vs.
attitudinal outcomes). In Vector 2, productivity and absenteeism = 0 and
job satisfaction = 1 and satisfaction with schedule = 1 (comparing
attitudinal outcomes). In Vector 3, productivity = 1; absenteeism = -1;
job satisfaction = 0; satisfaction with schedule = 0 (comparing behavioral
outcomes). In Vector 4, productivity = 1; supervisor performance ratings = 1; absenteeism = 0; job satisfaction = 0; satisfaction with
schedule = 0. B = unstandardized partial multiple regression coefficient;
/3 = standardized partial multiple regression coefficient. ** p < .001.

work-related criteria were affected quite differently by the


introduction of a compressed workweek work schedule.
Specifically, attitudinal outcomes were more greatly affected than behavioral outcomes, effect sizes associated
with job satisfaction were larger than effect sizes associated
with satisfaction with schedule, and effect sizes associated
with productivity were much smaller than effect sizes measured through supervisor-rated performance scales.
Although the QR statistic indicated that a substantial and
significant proportion of variance in the effect-size estimates (52%) were explained by the study characteristics, the
QE statistic was also significant indicating that a significant
proportion of the variance in the effect sizes was not explained. As mentioned earlier, this result may indicate the
presence of additional moderators.
9
An anonymous reviewer suggested that the time since schedule intervention could only be appropriately tested with studies
that included pretest scores. Thus, to confirm our flextime and
compressed workweek results, we conducted both regressions with
only pre-post test studies (about two thirds of the original sample).
Time since schedule intervention was still found to be a significant
negative predictor in the flextime regression (B = .64, p <
.001), and it was not a significant predictor in the compressed
workweek regression. Both of these findings supported our earlier
results.

508

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Discussion
The primary rationale for this study was to provide stateof-the-art evidence regarding the effects of flextime and
compressed workweek schedules on organization-relevant
outcomes. Specifically, this study attempted to address several problems in prior alternative-work-schedule research
by using meta-analytic techniques to predict and assess the
impact of flexible and compressed workweek schedules on
several work-related criteria. This study also tested a number of hypotheses derived in a cumulative sense from past
empirical findings and previously introduced theories.
In discussing our results, we attempt to integrate the
findings of this study with previous research and discuss the
implications of our findings for organizational decision
making. We begin with a review of the specific effects and
moderators of both intervention types on work-related
criteria.
Specific

Effects

In general, the effects of both flextime and compressed


workweek schedules were positive and consistent with our
predictions with a few important exceptions. Furthermore,
differences in effect sizes were found within intervention
type across the outcome criteria.

schedule. Contrary to predictions, however, absenteeism did


not decrease. Perhaps most surprising was that although
productivity was not positively affected, the supervisorrated performance criteria did show a positive increase. This
result may seem surprising because the more objective
measure of productivity indicated no increase. However,
prior research has found very low correlations between
objective and subjective measures (Alexander & Wilkins,
1982).
The nonsignificant effect size reported for absenteeism
suggests that compressed workweek schedules do not enhance the motivation of employees to attend. However,
because this effect size was only calculated from five studies, it should be interpreted cautiously.
As with flextime, the size of effects across the criteria
was significantly different for compressed workweek. However, contrary to the flextime findings, the compressedworkweek effect sizes associated with behavioral workrelated criteria (productivity and absenteeism) were smaller
than those for attitudinal work-related criteria (job satisfaction and satisfaction with schedule). Because our hypothe:
ses did predict that the compressed workweek would not
affect behavioral work-related criteria as positively (i.e.,
only absenteeism should be positively affected) as attitudinal work-related criteria, this result is not that surprising.
Moderators

Flextime
The predictions made for flextime were upheld in all but
one case. Flexible work schedules had positive effects on
employee productivity, job satisfaction, satisfaction with
work schedule, and employee absenteeism. However, the
sizes of these effects were significantly different. For example, the effect size associated with absenteeism was
significantly larger than that for productivity. This result is
consistent with the conjectures made by Pierce et al. (1989),
that an alternative work schedule would be more likely to
impact attendance and/or retention than directly impact
worker effectiveness. Contrary to expectations, self-rated
performance was not positively affected by the introduction
of a flextime schedule. The fact that self-rated performance
was not affected is surprising given that productivity increased. However, research on the psychometric properties
of self-rated performance scales has found that the selfreport of performance tends to be more lenient than ratings
made by others (Ford & Noe, 1987). It could be that a
ceiling effect came into play with respect to self-rated
performance; that is, there was no room for self-rated performance to improve.
Compressed Workweek
As predicted, compressed workweek schedules did positively affect job satisfaction and satisfaction with work

Overall, four significant moderator effects were found for


flextime work schedules, whereas compressed-workweek
effect sizes exhibited no significant differences across
moderators.
Employee Type
In general, flextime work schedules demonstrated positive effects on work-related outcome criteria for general
employees, whereas they had no effect for professionals and
managers. This finding is consistent with our hypothesis
that alternative work schedules are unlikely to benefit those
who already have a high degree of work autonomy. However, because of the small number of studies that included
managers, these results need further replications.
Flextime Flexibility
A significant yet counterintuitive finding of the present
analysis was the diminished effectiveness of highly flexible
flextime programs in comparison to less flexible programs.
This finding is consistent, however, with earlier research
where it had been found that the gains incurred from a
highly flexible schedule may be offset by the extra control
required to monitor the number of hours worked by the
employee (Coltrin & Barendse, 1981). Furthermore, it is
possible that the increased flexibility may have become

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES

more of an inconvenience for employees than a benefit. For


instance, employees may experience negative consequences
from flextime when they cannot communicate and/or cooperate with other employees because they are not at work
during the same time period (Nollen, 1981). In this vein,
Ronen stated that "since the total work force is available
only during core time, problems with scheduling are inherent in flextime and can affect communication, supervision,
and task performance, especially if tasks are highly interdependent" (Ronen, 1981, p. 65). Thus, too much flexibility
may cause problems for employees who rely on one another
for task completion. Furthermore, as pointed out by an
anonymous reviewer, when flexibility is very high, individual managers may have more control over how the flextime
schedule is actually implemented. This fact could cause
variations between organizations and thus also be responsible for our finding.
It should be pointed out that there are other measures of
flexibility (e.g., carryover, supervisor's role, etc.). Because
most studies did not provide this information, we could not
take them into account as possible moderators. It is plausible that oneor even allof these other important factors
are confounding our findings with regard to degree of
flexibility.
Time Since Intervention
In the flextime regression analysis, when other variables
were controlled for, a significant difference for time since
schedule intervention was found. This result indicates that
flexible work schedules may have waning effects over time.
The decrease in positive outcomes seen with a flextime
intervention is important for human resource practitioners to
consider. As argued earlier, this may be a direct result of
employees becoming accustomed to the new schedules and
eventually accepting them as the norm (Ronen, 1981).
Methodological Rigor
The results of these analyses were somewhat inconsistent. That is, methodological rigor only affected flextime
effect sizes. High-rigor flextime studies exhibited larger
effect sizes than their low-rigor counterparts. The fact that
high-rigor studies led to larger effect sizes for flextime
studies is encouraging in that it helps promote the idea that
well-developed studies with stringent methodological standards can lead to more precise and, in this case, stronger
results.
Integration With Previous Research
and Practical Implications
The most recent major integrative review of the evidence
for both flextime and compressed workweek schedules is
contained in Pierce et al. (1989).10 By and large, our flex-

509

time, findings are consistent with this Pierce et al. review,


which reported flexible work schedules to have generally
positive effects across all of the criteria we considered (i.e.,
productivity, absenteeism, job satisfaction, and satisfaction
with schedule).
However, beyond the Pierce et al. (1989) review, we did
find that these effects varied significantly across criteria and
that positive effects were not found for managers/professionals. Furthermore, we found that these effects differed as
a function of time since schedule implementation and degree of flexibility. Specifically, the findings that the effects
of flexible interventions seem to decline with time and that
too much flexibility may actually decrease the positive
effects of this intervention on work-related criteria provide
new evidence that we deem to be important from both
practical and theoretical standpoints.
For example, our results concerning the degree of flextime flexibility suggest that human resource practitioners
need to carefully examine the work that is being done by
individuals in their organization to determine the degree of
interdependence between jobs. Too much flexibility (in
terms of core hours) for employees with highly interdependent jobs may lead to lesser gains for the company than a
low flexibility schedule. Also, our results indicate that human resource practitioners may see a reduction in the initial
positive gains as time goes by after the introduction of a
flextime intervention. To gain a better insight into this
probtem, it seems imperative to link work-related findings
with trajectories of job and life satisfaction, including those
associated with nonwork contexts, such as family life. It is
also important to point out that because of the relatively
small number of studies available the effects of these moderators (e.g., time since schedule intervention) need to be
interpreted cautiously. Further research investigating these
effects is needed.
The compressed-workweek literature is not nearly as
extensive as the literature concerning flexible work schedules. Regarding compressed workweek schedules, our results also generally support earlier findings (Pierce et al.,
1989). However, there are a few important exceptions. Job
satisfaction was positively affected, and contrary to earlier
studies (Goodale & Aagaard, 1975; Latack & Foster, 1985),
we found that this type of schedule did not affect absenteeism rates.
As mentioned earlier, one of the most significant findings
of the present meta-analysis is the differential effects of the
two schedule types across the four criterion measures. These
differences demonstrate the need to examine the effects of
interventions across a range of potential organizational out10
This meta-analysis included 15 studies not cited in the original Pierce et al. (1989) review, including two that were completed
after 1989 (Dalton & Mesch, 1990; Venne, 1993).

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

510

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

come criteria. Past meta-analyses of organizational developmental interventions (e.g., Guzzo, Jette, & Katzell, 1985)
have focused on a single class of outcome measure (e.g.,
performance or attitudinal) and on broader classes of intervention types (e.g., technostructural interventions). The
present findings demonstrate the need to examine specific
interventions across a range of outcomes and the development of detailed intervention profiles. For example, on the
basis of our findings, an organization implementing a compressed workweek schedule may consider it a failure if they
are attempting to lower absenteeism. On the other hand, the
same organization may be very satisfied with the introduction of their compressed workweek if their outcome criterion is job satisfaction. Thus, implementation of an intervention may result in organizational gains depending on the
outcome criterion being considered.
In summary, it is important to note that both flextime and
compressed workweek schedules had primarily positive and
no negative effects on work-related criteria. These positive
benefits are consistent with historical changes toward more
alternative work schedules, and as such they should ease
employers' worries over the outcomes they will experience
with the implementation of a flexible or compressed workweek work schedule. Similarly, the findings should be reassuring to organizations that may view the demand for
alternative work schedules as originating from outside of
their own work-related contexts, such as societal changes in
dual-career households and work-leisure time expectations.
However, the results presented in this study also make it
clear that employers and employees are well advised to
work together to ensure that alternative work schedules
provide the most positive benefits to individuals and
organizations.
Limitations and Future Research
Meta-analytic studies can also be used to identify weaknesses in research and subsequently avenues for further
research. From such a point of view, several limitations of
prior research, as well as the current study, deserve attention. With respect to prior research, the relatively small
number of alternative work schedule programs that have
been formally evaluated distresses us. Because organizations are increasingly using these new work schedules, it
seems desirable that more formal evaluations be done so
that future researchers can more accurately assess the benefits and/or losses associated with alternative work schedules. Finally, these formal evaluations should use a multidimensional conceptualization (e.g., carryover or supervisor
role) of the alternative work schedule being investigated.
With respect to the current study, we wish to point out the
following. First, on a methodological level, although both
regression models explained a significant amount of variance in their respective effect sizes, they also left a signif-

icant amount of variance unexplained. It is probable that


unknown moderators may be related to the effect sizes in
both the flextime and compressed workweek samples. For
example, the result that a low degree of flexibility is better
than a high degree of flexibility may change with the
introduction of other moderator variables, such as carryover. Second, because our setting of a 6-month cutoff in
the time since schedule intervention dichotomy was arbitrarily set to ensure us somewhat equal groups, future research should attempt to replicate these findings with a
more suitable design and, perhaps more importantly, a theoretically inspired time scale. Third, because of the small
number of effect sizes associated with any specific criterion
measure, we could not assess whether our moderator variables, such as flextime flexibility, have the same effects
across all the outcome criteria. Thus, future research should
attempt to assess moderators at the individual criterion
level. Fourth, although we attempted to use theoretical
models to generate hypotheses, the studies included in the
meta-analysis did not allow us to make a direct test of these
theoretical assumptions. For instance, we hypothesized that
because flextime meets employees' needs for autonomy, job
satisfaction will increase. However, we were not able to
directly test this proposition because available studies did
not measure employees' need for autonomy. Therefore, we
encourage future research to use measures that directly tap
the mediating variables that are part of the major theoretical
frameworks in the field. In general, altemative-workschedule research would benefit from the testing of models
that include individual level variables (e.g., need for autonomy). These variables may help shed further light on outcome differences found in the alternative-work-schedule
literature. For example, it may be that the reason flextime
interventions have varying effects on job satisfaction is that
different types of employees have different levels of need
for autonomy. Furthermore, the integration of these individual level variables into alternative-work-schedule research
would be necessary for any meaningful cross-cultural research attempting to discern differences, if any, that the
implementation of alternative work schedules may have in
different cultures.
There are also issues of theory in other areas of industrial
psychology, such as training, where an increased concern
for the role of multiple contexts of life and historical time
has been observed (Warr, 1994). In this vein, as one evaluates the effects of alternative work schedules, it is important to recognize that the effect patterns obtained are context
dependent. Two such context dependencies seem of particular significance. One is the link between work contexts and
other contexts of life such as family functioning and leisure
activity. It would seem important in future work on alternative work schedules to link these contexts more explicitly
than past research has done. The second relevant context is
historical time. There are major historical changes in aspects

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES

of social communication, technology, and work environments, and it has been found that incentives that motivate
and satisfy older workers are irrelevant to younger workers
(Forteza & Prieto, 1994). Thus, it may be that baby
boomers, who grew up with parents who worked in a more
structured work environment, are more affected by the
advent of an alternative work schedule than a later generation of employees who have come to expect such considerations on the part of their employer. It seems desirable,
therefore, in future studies and meta-analyses, to include
such comparative and historical dimensions. Moreover, because almost all of the formal evaluations were done quite
some time ago, it is important to conduct more up-to-date
evaluations to determine whether societal changes may have
changed the impact that these interventions may have on a
more modern workforce.
In summary, we believe that this meta-analysis provides
both researchers and practitioners with the most accurate
assessment of these two alternative work schedules that has
been presented to date. Furthermore, the results of this
meta-analysis and their theoretical implications will hopefully serve as a framework and/or inspiration for future
flextime and compressed workweek research.

References
References marked with an asterisk indicate studies used
in the meta-analysis.
Alexander, E., & Wilkins, R. D. (1982). Performance rating validity: The relationship of objective and subjective measures of
performance. Group and Organization Studies, 7, 485-496.
*Bohen, H. H., & Viveros-Long, A. (1981). Balancing jobs and
family life: Do flexible work schedules help? Philadelphia: Temple University Press.
Bullock, R. I, & Svyantek, D. J. (1983). Positive-findings bias in
positive findings research. Academy of Management Proceedings, 221-224.
Caldwell, D. F., & O'Reilly, C. A., III. (1990). Measuring personjob fit with a profile-comparison process. Journal of Applied
Psychology, 75, 648-657.
*Calvasina, E. J., & Boxx, W. R. (1975). Efficiency of workers on
the four-day workweek. Academy of Management Journal, 18,
604-610.
Chatman, J. A. (1988). Matching people and organizations: Selection and socialization in public accounting firms. Unpublished doctoral dissertation, University of California, Berkeley.
Coltrin, S. A., & Barendse, B. D. (1981). Is your organization a
good candidate for flextime? Personnel Journal, 60, 712-715.
Conway, J. M., & Huffcutt, A. I. (1997). Psychometric properties
of multisource performance ratings: A meta-analysis of subordinate, supervisor, peer, and self-ratings. Human Performance, 10, 331-360.
*Coston, R. D. (1973). An investigation of the economic effects of
the four-day workweek on women apparel workers in Georgia.
Unpublished doctoral dissertation, University of Arkansas, Little Rock.

511

*Dalton, D. R., & Mesch, D. J. (1990). The impact of flexible


scheduling on employee attendance and turnover. Administrative Science Quarterly, 35, 370-387.
*Dalton, D. R., & Todor, W. D. (1984). Unanticipated consequences of union-management cooperation: An interrupted time
series analysis. Journal of Applied Behavioral Science, 20,
253-264.
Dawis, R. V., England, G. W., & Lofquist, L. H. (1968). A theory
of work adjustment (Rev. ed.). University of Minnesota Studies
in Vocational Rehabilitation XXIII, Bulletin 47.
Dawis, R. V., & Lofquist, L. H. (1984). A psychological theory of
work adjustment: An individual differences model and its applications. Minneapolis, MN: University of Minnesota Press.
deCarufel, A., & Schaan, J. L. (1990). The impact of compressed
workweeks on police job involvement. Canadian Police College Journal, 14, 81-97.
Dodd, N. G., & Ganster, D. C. (1996). The interactive effects of
variety, autonomy, and feedback on attitudes and performance.
Journal of Organizational Behavior, 17, 329-347.
*Dunham, R. B., Pierce, J. L., & Castaneda, M. B. (1987). Alternative work schedules: Two field quasi-experiments. Personnel
Psychology, 40, 215-243.
*Evans, M. G. (1975). A longitudinal analysis of the impact of
flexible working hours. Studies in Personnel Psychology, 6,
1-10.
Ford, K. J., & Noe, R. A. (1987). Self-assessed training needs: The
effects of attitudes toward training, managerial level, and function. Personnel Psychology, 40, 39-53.
Forteza, J. A., & Prieto, J. M. (1994). Aging and work behavior. In
HL-C. Triandis, M. D. Dunnette, & L. M. Hough (Eds.), Handbook of industrial & organizational psychology (pp. 447-484).
Palo Alto, CA: Consulting Psychologists Press.
Fried, Y. (1991). Meta-analytic comparison of the Job Diagnostic
Survey and Job Characteristics Inventory as correlates of work
satisfaction and performance. Journal of Applied Psychology, 76, 690-697.
Fried, Y., & Ferris, G. R. (1987). The validity of the job characteristics model: A review and meta-analysis. Personnel Psychology, 40, 287-322.
Gellatly, I. R. (1995). Individual and group determinants of employee absenteeism: Test of a casual model. Journal of Organizational Behavior, 16, 469-485.
Glass, G. V., McGaw, B., & Smith, M. L. (1981). Meta-analysis in
social research. Beverly Hills, CA: Sage.
*Golembiewski, R. T., & Hilles, R. (1977). Drug company workers like new schedules. Monthly Labor Review, 2, 65-69.
Golembiewski, R. T., & Proehl, C. (1978). A survey of empirical
literature on flexible workhours: Character and consequences of
a major innovation. The Academy of Management Review, 3,
837-853.
Golembiewski, R. T., & Proehl, C. (1980). Public sector applications of flexible workhours: A review of available experience.
Public Administration Review, 40, 72-85.
*Golembiewski, R. T., Yeager, S., & Hilles, R. (1975). Factor
analysis of some flextime effects: Attitudinal and behavioral
consequences of a structural intervention. Academy of Management Journal, 18, 500-509.
*Goodale, J. G., & Aagaard, A. K. (1975). Factors relating to

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

512

BALTES, BRIGGS, HUFF, WRIGHT, AND NEUMAN

varying reactions to the 4-day workweek. Journal of Applied


Psychology, 60, 33-38.
Guzzo, R. A., Jette, R. D., & Katzell, R. A. (1985). The effects of
psychologically based intervention programs on worker productivity: A meta-analysis. Personnel Psychology, 38, 275-291.
Hackman, J. R., & Oldham, G. R. (1976). Motivation through the
design of work: Test of a theory. Organizational Behavior and
Human Performance, 16, 250-279.
*Harvey, B. H., & Luthens, F. (1979). Flextime: An empirical
analysis of its real meaning and impact. MSU Business Topics, 27, 31-36.
*Hausser, D. (1980). The impact of flexitime on organizational
functions: A field study. Washington, DC: U.S. Government
Printing Office.
Hedges, L. V., & OIkin, I. (1985). Statistical methods for metaanalysis. Orlando, FL: Academic Press.
Hewitt Associates LLC. (1995). Work and family benefits provided
by major U.S. employers in 1994. New York: Author.
*Hicks, W. D., & Klimoski, R. J. (1981). The impact of flexitime
on employee attitudes. Academy of Management Journal, 24,
333-341.
Hochschild, A. R. (1997). The time bind: When work becomes
home and home becomes work. New York: Metropolitan
Books.
*Hodge, B. J., & Tellier, R. D. (1975). Employee reactions to the
four-day week. California Management Review, 18, 25-30.
Hoffmann, C. C., Barry, N. R., & Holden, C. (1991). A comparison of validation criteria: Objective versus subjective performance measures and self versus supervisor ratings. Personnel
Psychology, 44, 601-619.
*Ivancevich, J. M. (1974). Effects of the shorter workweek on
selected satisfaction and performance measures. Journal of Applied Psychology, 59, 111-111.
*Ivancevich, J. M., & Lyon, H. L. (1977). The shortened workweek: A field experiment. Journal of Applied Psychology, 62,
34-37.
Jamal, M. (1984). Job stress and job performance controversy: An
empirical assessment. Organizational Behavior and Human
Performance, 33, 1-21.
Jamal, M. (1985). Relationship of job stress to job performance: A
study of managers and blue-collar workers. Human Relations, 5,
409-424.
Johnson, B. T. (1993). DSTAT 1.10: Software for the meta-analytic
review of research literatures. Hillsdale, NJ: Erlbaum.
*Kim, J. S., & Campagna, A. F. (1981). Effects of flexitime on
employee attendance and performance: A field experiment.
Academy of Management Journal, 24, 729-741.
*Krausz, M., & Freibach, N. (1983). Effects of flexible working
time for employed women upon satisfaction and absenteeism.
Journal of Occupational Psychology, 56, 155-159.
Latack, J. C., & Foster, L. W. (1985). Implementation of compressed work schedules: Participation and job redesign as critical factors for employee acceptance. Personnel Psychology, 38,
75-92.
Lee, R. A. (1983). Flexitime and conjugal roles. Journal of Occupational Behavior, 4, 297-315.
Light, R. J., Singer, J. D., & Willett, J. B. (1994). The visual
presentation and interpretation of meta-analysis. In H. Cooper &

L. V. Hedges (Eds.), Handbook of research synthesis. New


York: Sage.
Macan, T. H. (1994). Time management: Test of a process model.
Journal of Applied Psychology, 79, 381-391.
*Maklan, D. M. (1977). The four-day workweek: Blue-collar adjustment to a nonconventional arrangement of work and leisure
time. New York: Praeger.
*McGuire, J. B., & Liro, J. R. (1986). Flexible work schedules,
work attitudes, and perceptions of productivity. Public Personnel Management, 15, 65-73.
*McGuire, J. B., & Liro, J. R. (1987). Absenteeism and flexible
work schedules. Public Personnel Management, 16, 47-59.
Meijman, T. F. (1992). Verkoten van de werkweek en verlangen
van de wekdag [Shortening the workweek and lengthening the
workday]. Amsterdam: Studiecentrum Arbeid en Gezondheid.
*Millard, C. W., Lockwood, D. L., & Luthans, F. (1980). The
impact of a 4-day workweek on employees. MSU Business
Topics, 28, 31-37.
*Morgan, F. T. (1977). Your (flex)time may come. Personnel
Journal, 56, 82-85.
*Narayanan, V. K. (1982). Hierarchical level and the impact of
flexitime. Industrial Relations, 21, 216-230.
*Narayanan, V. K., & Nath, R. (1984). The influence of group
cohesiveness on some changes induced by flexitime: A quasiexperiment. Journal of Applied Behavioral Science, 20, 265276.
Nollen, S. D. (1981). The compressed workweek: Is it worth the
effort? Industrial Engineer, 13, 58-64.
*Orpen, C. (1981). Effect of flexible working hours on employee
satisfaction and performance: A field experiment. Journal of
Applied Psychology, 66, 113-115.
Parker, P. A., & Kulik, J. A. (1995). Burnout, self- and supervisorrelated performance, and absenteeism among nurses. Journal of
Behavioral Medicine, 18, 581-599.
Pierce, J. L., & Dunham, R. B. (1992). The 12-hour work day: A
48-hour, eight-day week. Academy of Management Journal, 35,
1086-1098.
Pierce, J. L., & Newstxom, J. W. (1980). Toward a conceptual
clarification of employee responses to flexible working hours: A
work adjustment approach. Journal of Management, 6, 117134.
Pierce, J. L., & Newstrom, J. W. (1983). The design of flexible
work schedules and employee responses: Relationships and
process. Journal of Occupational Behavior, 4, 247-262.
Pierce, J. L., Newstrom, J. W., Dunham, R. B., & Barber, A. E.
(1989). Alternative work schedules. Needham Heights, MA:
Allyn and Bacon.
*Ralston, D. A. (1989). The benefits of flextime: Real or imaging Journal of Organizational Behavior, 10, 369-373.
Ralston, D. A., Anthony, W. P., & Gustafson, D. J. (1985).
Employees may love flextime, but what does it do to the
organization's productivity? Journal of Applied Psychology, 70,
272-279.
*Ralston, D. A., & Flanagan, M. F. (1985). The effect of flextime
on absenteeism and turnover for male and female employees.
Journal of Vocational Behavior, 26, 206-217.
Roberts, H. E., & Foti, R. J. (1998). Evaluating the interaction

This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

WORK SCHEDULES

between self-leadership and work structure in predicting job


satisfaction. Journal of Business & Psychology, 12, 257-267.
Ronen, S. (1981). Flexible working hours. New York: McGrawHill.
Ronen, S. (1984). Alternative work schedules: Selecting, implementing, and evaluating. Homewood, IL: Dow Jones-Irwin.
Ronen, S., & Primps, S. B. (1981). The compressed workweek as
organizational change: Behaviors and attitudinal outcomes.
Academy of Management Review, 6, 61-74.
Sagie, A. (1998). Employee absenteeism, organizational commitment, and job satisfaction: Another look. Journal of Vocational
Behavior, 52, 156-171.
*Schein, V. E., Maurer, E. H., & Novak, J. F. (1977). Impact of
flexible working hours on flexibility. Journal of Applied Psychology, 62, 463-465.
Shrout, P. E., & Fleiss, J. L. (1979). Intraclass correlations: Uses in
assessing rater reliability. Psychological Bulletin, 86, 420-428.
Somers, M. J. (1995). Organizational commitment, turnover and
absenteeism: An examination of direct and interaction effects.
Journal of Organizational Behavior, 16, 49-58.
Song, R., Daly, B. J., Rudy, E. B., Douglas, S., & Dyer, M. A.
(1997). Nurses' job satisfaction, absenteeism, and turnover after
implementing a special care unit practice model. Research in
Nursing and Health, 20, 443-452.
Stevens, E. D., & Elsworth, R. (1979). Flexitime in the Australian
public service: Its effect on nonwork activities. Public Personnel Management, 8, 196-205.

513

Stevens, J. (1996). Applied multivariate statistics for the social


sciences. Mahwah, NJ: Erlbaum.
Sullivan, S. E., & Bhagat, R. S. (1992). Organizational stress, job
satisfaction, and job performance: Where do we go from here?
Journal of Management, 18, 353-374.
Terpstra, D. E. (1981). Relationship between methodological rigor
and reported outcomes in organizational development evaluation research. Journal of Applied Psychology, 66, 541-543.
Thierry, H., & Meijman, T. (1994). Time and behavior at work. In
H. C. Triandis, M. D. Dunnette, & L. M. Hough (Eds.), Handbook of industrial & organizational psychology (pp. 341-414).
Palo Alto, CA: Consulting Psychologists Press.
*Venne, R. A. (1993). Alternative worktime arrangements: The
compressed workweek. Unpublished doctoral dissertation, University of Toronto, Toronto, Ontario, Canada.
Warr, P. (1994). Age and employment. In H. C. Triandis, M. D.
Dunnette, & L. M. Hough (Eds.), Handbook of industrial &
organizational psychology (pp. 485-550). Palo Alto, CA: Consulting Psychologists Press.
*Welch, J. R., & Gordon, D. (1980). Assessing the impact of
flexitime on productivity. MSU Business Horizons, 23, 61-65.

Received March 13, 1998


Revision received September 23, 1998
Accepted September 28, 1998

You might also like