Content uploaded by Shawn Arent
Author content
All content in this area was uploaded by Shawn Arent on Sep 14, 2015
Content may be subject to copyright.
Arent, S., Landers, M., et al
The Effects of Exercise on Mood in
Older Adults: A Meta-Analytic
Review.
Readers are reminded that copyright subsists in this extract and the work from which it was taken.
Except as provided for by the terms of a rightsholder’s licence or copyright law, no further copying,
storage or distribution is permitted without the consent of the copyright holder.
The author (or authors) of the Literary Work or Works contained within the Licensed Material is or are
the author(s) and may have moral rights in the work. The Licencee shall not cause or permit the
distortion, mutilation or other modification of, or other derogatory treatment of, the work which would
be prejudicial to the honour or reputation of the author.
Arent, S., Landers, M., et al (2000), 'The Effects of Exercise on Mood in Older Adults: A Meta-
Analytic Review.', In The Journal of Aging and Physical Activity, Vol.8, , , , pp.407-430. ISSN:
1063-8652
This is a digital version of copyright material made under licence from the publisher, and its accuracy
cannot be guaranteed. Please refer to the original published edition.
Licenced for use at University of Derby for the Current Issues in Business Psychology course(s)
running during the period 01-SEP-07 to 31-AUG-08
Permission reference: H1063-8652_8(407-430)70151
Journal of Ageing and Physical Activity, 2000, 8, 407-430
© 2000 Human Kinetics Publishers, Inc.
The Effects of Exercise on Mood in Older Adults:
A Meta-Analytic Review
Shawn M. Arent, Daniel M. Landers, and Jennifer L. Etnier
This meta-analysis examined the exercise-mood relationship in older adults.
158 effect sizes (ESs) from 32 studies were grouped into experimental-versus-
control, gains, and correlational ESs. Each study was coded for moderator
variables related to descriptive, design, participant, exercise, and mood-
assessment characteristics. Experimental-versus-control ESs for negative (NA)
and positive affect (PA) were 0.35 (p < .05) and 0.33 (p> .05), respectively, with
an overall ES of 0.34, p < .05. The gains ESs for NA and PA in an exercise group
were 0.39 (p < .05) and 0.35 (p < .05), respectively, with an overall ES of 0.38,
p < .05. All effects were significantly greater than those for the control groups.
Correlational ESs of 0.47 and 0.42 were found for NA and PA, respectively. It
was concluded that chronic exercise is associated with improved mood in the
elderly. Moderating variables and implications for exercise prescription to
improve mood in the elderly are discussed.
Key Words: elderly, affect, physical activity
Senior citizens have become the fastest growing segment of our population. In 1880,
less than 3% of the total population was older than 65. By 1980, older adults
constituted almost 12% of the population. There are now more than 36 million
elderly people in the United States, and this number is expected to double by the year
2030. According to current estimates, this would mean that the elderly would
represent 22% of the total population (Hoeger & Hoeger, 1996). The 65-and-over
group is growing twice as fast as the rest of the population. One reason for this trend
is increasing life expectancies. Unfortunately, increased life expectancy does not
necessarily equate to increased quality of life.
One of the central issues raised by increased longevity is that of net gain in
active functional years versus total years of disability and dysfunction. Muscular
strength in most individuals is well maintained until about 45 years of age but then
deteriorates by 5-10% per decade thereafter (Rogers & Evans, 1993). Furthermore,
loss in muscle mass appears to be the major cause of strength decrease in the elderly
(Rogers & Evans). Muscle weakness can compromise everyday activities, leading
to dependence on others to help perform day-to-day tasks and increasing the risk of
injury resulting from falls. One way to potentially offset or rectify many of the health
The authors are with the Department of Exercise Science and Physical Education at
Arizona State University, Tempe, AZ 85287-0404.
407
408 • Arent, Landers, and Etnier
problems and physical deterioration incurred with aging is through exercise (Cress
et al., 1991; Fiatarone et al., 1990; Frontera, Meredith, O'Reilly, Knuttgen, &
Evans, 1988; Hagberg, Graves, & Limacher, 1989; Kohrt, Malley, & Coggan, 1991;
Makrides, Heigenhauser, & Jones, 1990). Of all age groups, the elderly have the
most to gain by being active because of the effects of exercise on improved body
composition, fitness, longevity, ability to perform personal care activities, and the
management of arthritis or other debilitating conditions. Furthermore, there is
convincing evidence that greater aerobic power, muscle strength, and flexibility
allow the elderly to perform at levels equivalent to people 10-20 years younger,
which often equates to a dramatically improved quality of life.
In addition to declines in physical performance with the passage of time, as
individuals progress beyond 60 years of age there also tends to be an increased
prevalence of mood disturbance (i.e., increased negative affect and decreased
positive affect; Fillingim & Blumenthal, 1993; Pfeiffer, 1977) and decrements in
some cognitive abilities (Botwinick, 1977; Jacewicz & Hartley, 1987; Schaie,
1994). Age-related changes (i.e., physical, mental, and social) often present
challenges to emotional control and stability that exceed those experienced by
younger individuals (Spirduso & Mackie, 1995). Physical deterioration, functional
losses, loss of friends and family, and even retirement provide unique and poten-
tially overwhelming challenges to the older adult. The fact that almost twice as
many individuals over 65 commit suicide each year (Crandall, 1991) might provide
some insight into the ramifications of these "challenges" associated with aging. A
recent study at the University of Washington (Unutzer et al., 1997) found that
elderly individuals with strong depressive symptoms incurred about $5,000 more
in medical costs over 4 years than did elderly adults without depression. Unutzer et
al. attribute increased costs to the possibility that depression in the elderly exacer-
bates other medical problems.
A meta-analysis on the effects of exercise on depression (North, McCullagh,
& Tran, 1990) concluded that exercise can help decrease depression—even in those
who are not initially depressed. North et al. indicated that this effect appears to be
even more pronounced in the elderly, but this population was unfortunately not
included in the final analysis. Similarly, Landers and Petruzzello (1994) conducted
a meta-analysis exploring the effects of physical activity on anxiety. Although
exercise was found to be associated with a reduction in anxiety, the effect of age as
a moderator variable was not specified. Thus, although it is generally agreed that
elderly populations physically respond to exercise in a manner similar to that of
younger populations, it is not as clear whether they respond in a similar fashion
psychologically, especially with regard to changes in mood.
Anxiety and depression are linked in that they are concepts typically included
under the more general rubric of affect, or mood. Other concepts commonly
explored and associated with research on mood include anger, vigor, fatigue,
confusion, pleasantness, and euphoria (Tuson & Sinyor, 1993). Lazarus (1991)
contends that, of these constructs, only anxiety, anger, and euphoria represent true
affect. Lazarus also maintains that the terms mood and affect do not represent the
same thing but, rather, different time periods on the emotional continuum. Mood,
according to Lazarus, represents a transient state, whereas affect represents something
Effects of Exercise on Mood in Older Adults • 409
more enduring. In an attempt to remain consistent with the current literature,
however, this study accepts as a precept that the terms mood and affect can be (and
usually are) used interchangeably and that all of these concepts (depression,
anxiety, anger, vigor, fatigue, confusion, pleasantness, and euphoria) can be
included under the rubric of mood. In fact, Gauvin and Brawley (1993) argue for this
approach when examining exercise and mood. They suggest that this might be better
suited to the understanding of the relationship of affect and exercise because the
models that can be derived from it are intended to be broad and encompassing
conceptualizations of affective experience. A model of affect that has a wider focus
is more likely to capture the nature of exercise-induced affect. Gauvin and Brawley
consistently use the words mood and affect interchangeably when referring to the
effects of exercise along a positive-negative dimension.
The terms mood, affect (positive and negative), psychological benefits, and
well-being are often used interchangeably when assessing the effects of exercise on
the elderly (or most other populations, for that matter). Nonetheless, studies differ
greatly on the components and concepts they include under these labels. For
example, in the exercise literature, the phrase psychological benefits often includes
improvements in cognitive functioning or decreases in stress reactivity (Tuson &
Sinyor, 1993). Neither of these concepts is necessarily indicative of improvements
in mood or affect, however. Well-being also tends to be a rather broad term that can
include physical well-being, cognitive functioning, and life satisfaction, in addition
to psychological states commonly associated with affect and mood (Tuson &
Sinyor). Although affect does appear to be interchangeable with mood in most of the
exercise literature, this might not always be the case when referring to psychological
benefits or well-being. Therefore, if it is concluded that an exercise treatment did not
produce psychological benefits or changes in well-being, this might or might not
indicate failure to produce a change in mood state, depending on the constructs
included under psychological benefits or well-being.
There have been inconsistencies in the literature reviews regarding the
efficacy of exercise for improving mood in the elderly. Although there have been
some narrative reviews that have espoused the beneficial effects of exercise on
mood in the elderly (Crase & Rosato, 1979; Johnson-Pawlson & Koshes, 1985;
Shephard, 1984, 1990), there have been others that have presented a less conclusive
view of this effect (Brown, 1992; Fillingim & Blumenthal, 1993; Netz & Jacob,
1994). Part of the problem might stem from the aforementioned inconsistencies in
regard to which constructs the reviewers thought comprised the term mood.
Differences in the conclusions reached by the reviewers might also be
attributed to inherent shortcomings of narrative reviews. Narrative reviews are
relatively subjective assessments of the current literature and rely on the "vote-
count" method to determine the presence or absence of an effect. They typically do
not include all available studies from a given area of research but, rather, those
studies the author believes fit well within the intended discussion. Methods are
typically subjective and unreported, leaving a problem of replicability. Narrative
reviews are limited by both the studies chosen and the limitations inherent to those
studies. These reviews typically depend on significant or nonsignificant findings,
rather than on the magnitude of a potential effect. Furthermore, there are numerous
410 • Arent, Landers, and Etnier
differences among studies specifically examining the effects of exercise on mood
in the elderly that would limit a traditional review's ability to assess these variables'
moderating effects on the findings. Some of these differences include (a) differ-
ences in measures of mood employed, (b) differences in exercise protocols
(including type, duration, intensity, and frequency), (c) differences in participants'
preexercise fitness levels, (d) differences in study design (i.e., random assignment,
presence of control groups, etc.), (e) differences in the components of mood
assessed, and (f) differences between acute exercise and chronic exercise. Fortu-
nately, there is a method whereby the reviewer can objectively assess how design
characteristics influence effect size through classification of findings by moderator
variables.
A meta-analysis (Glass, 1977) can provide what the traditional narrative
review is typically incapable of—a statistically objective, replicable, comprehen-
sive analysis of the effect that an independent variable has on a dependent variable.
In order to do this, the results of all available studies (published and unpublished)
are quantified to a standard metric, the effect size (ES), that then allows the
researcher to use statistical techniques as a means of analysis. Also, by combining
participants across studies, one is able to greatly increase the statistical power of the
analysis. The lack of statistical power is common to behavioral science research
because of the relatively low numbers of participants combined with large numbers
of variables (Cohen, 1988). The increased power available through meta-analytic
techniques often allows the reviewer to detect small, but significant, trends not
detected in single studies or narrative reviews. Essentially, a meta-analysis allows
determination of the magnitude of an effect.
As previously stated, related meta-analyses (Landers & Petruzzello, 1994;
North et al., 1990) have failed to examine the influence of age (particularly "old
age") as a moderating variable in the exercise-mood relationship. Considering that
the elderly often experience challenges to emotional control above and beyond
those typically experienced by younger adults (Spirduso & Mackie, 1995), and that
North et al. have suggested that the effects of exercise on depression appear to be
even more pronounced in the elderly, it is beneficial to devote a meta-analysis to
examining the impact of exercise on mood in this previously overlooked population.
It is logical, particularly when taking into account the physical and psychological
decrements that often accompany aging, to assume that exercise might be even more
important for mood improvement in the elderly than it is in younger populations.
Relying on the current reviews, however, does not allow for assessing the accuracy
of this statement. This study will improve and extend the information provided by
existing reviews. The primary purpose of the study was to use meta-analytic
techniques to test several hypotheses.
Hypothesis 1
Exercise (cardiovascular, resistance training, or a combination of the two) improves
mood (both positive and negative) in the elderly relative to a control condition
(Crase & Rosato, 1979; Johnson-Pawlson & Koshes, 1985; Shephard, 1984, 1990).
A number of hypotheses based on study characteristics that could potentially
moderate the effects of exercise on mood are also examined.
Effects of Exercise on Mood in Older Adults • 411
Hypothesis 2
Studies with longer training protocols produce the greatest improvements in mood.
North et al. (1990) found the greatest effect on depression in studies employing a
protocol longer than 15 weeks. In addition, although the elderly respond to training
in a fashion similar to that of a younger population, the magnitude of the response
is often lower, and more time is required to achieve greater fitness improvements
(Shephard, 1984, 1990).
Hypothesis 3
Studies using participants with the poorest initial health status produce the largest
effect sizes. It is assumed that these individuals have the most to gain from the
training and, as such, have the most room for improvement (law of initial values).
Additionally, Hatfield, Goldfaib, Sforzo, and Flynn (1987) suggested that absence
of improvements in mood might be the result of including elderly participants with
initially high fitness/health levels or positive mood scores.
Hypothesis 4
Studies reporting significant fitness gains are associated with the largest effect sizes.
This is based on the cardiovascular-fitness hypothesis, which contends that physi-
ological improvements provide the underlying mechanism for psychological changes
with exercise. Therefore, a greater physiological improvement is expected to
coincide with a greater psychological improvement.
Method
SELECTION AND INCLUSION OF STUDIES
Studies were included in the analysis if they investigated the effects of physical
activity or exercise on some construct of mood in older adults. Based on the
information provided in each study, the activity or exercise was classified as either
cardiovascular exercise, resistance training, or a combination of these. This classi-
fication process, particularly for studies examining "physical activity," was based
on activities (and activity groupings) described by the American College of Sports
Medicine (1995) in their exercise prescription guidelines. Although classifying
activities such as gardening and housework was not an issue for the experimental
studies (none examined these types of activity), a fourth classification level termed
"unspecified physical activity" was added for the correlational studies, some of
which examined a broader or less defined classification of physical activity. For the
purposes of this analysis, at least one of the following three conditions must also
have been met to satisfy the definition of older adults: (a) The mean age of the study
sample was >65 years; (b) there must have been at least one exercising group with
a mean age >65 years; or (c) if mean ages were not provided, the age range of
exercise participants must have had a lower bound of at least 60 years. These
412 • Arent, Landers, and Etnier
parameters were established based on the societal definition of elderly (>65 years
of age) and from age ranges or mean ages included in studies that used samples from
senior centers or nursing homes.
All English-language studies meeting these criteria and available before
March 1998 that cou ld be lo cated were included. Most of the studies were identified
through computer searches of PsycLit, ERIC, SPORTDiscus, Dissertation Ab-
stracts, HealthStar, and MedLine. Key terms used in the searches focused on
exercise (e.g., strength training, aerobic, physical activity), participant classifica-
tion (e.g., elderly, older adults), and mood adjectives (e.g., mood, depression,
happiness). A complete listing of key terms is available from the primary author.
Cross-checking of references in the related literature and hand searches of Psycho-
logical Abstracts and Social Science Citation Index, as well as of relevant journals
in the areas of gerontology, psychology, and exercise science, were also used to
ensure an exhaustive literature search.
The literature search initially identified over 100 potentially useful studies.
On review, 36 of these studies met the necessary criteria for inclusion. Of these, four
were excluded from the analysis because of insufficient data to calculate ESs. The
remaining 32 studies were included in the analysis. Three databases were con-
structed using these 32 studies, and they were analyzed independently. This was
done in order to include correlational studies, studies that did not use a control group
for comparison, and studies that used groups that were not equivalent at pretest on
the dependent variable of interest. This was also done in order to include all possible
comparisons while attempting to minimize violations of the assumption of indepen-
dence. Although this procedure might not have completely ruled out this violation,
it was considered to be superior to the other alternatives—picking one representa-
tive ES for each study or combining all ESs for a study into one average ES. In either
of the latter cases, potentially useful information could have been lost. Instead, a
total of 168 ESs for the 32 studies were calculated. These were then divided into ESs
from exp erimental-versus-control-group comparisons (n = 61), gains ESs (pre-post
comparisons, n = 83), and correlational ESs (n = 24).
CALCULATION O F ESs
In this meta-analysis, Hedges's g was the ES measure of choice. Hedges's (1981)
formula for computing this ES is
In the case of gains ESs, ME- MC becomes Mpost - Mpre. If the necessary means
and standard deviations were not available, ES was calculated using F, t, r, or p
values if available, as outlined by Rosenthal (1994). ESs were corrected for positive
bias resulting from small sample sizes (Hedges & Olkin, 1985) and were weighted
by the inverse of the variance (Hedges & Olkin).
Effects of Exercise on Mood in Older Adults • 413
A negative ES for studies assessing negative affect would indicate a relative
improvement (i.e., lower scores would be better). In order to evaluate the effect of
exercise on global mood, improvements in mood were reflected by positive ESs. To
keep ES direction consistent across negative and positive affect, negative-affect ESs
were multiplied by - 1 . As a result, any improvement in mood was reflected by
positive ES values. Average ESs were calculated as outlined by Hedges and Olkin
(1985).
CODING O F ST UDIES
The included studies were coded for a number of characteristics based on a priori
decisions regarding potential moderator variables for the exercise-mood relation-
ship in the elderly. These characteristics were classified as design and descriptive
characteristics, participant characteristics, exercise characteristics, and mood-
assessment characteristics. Separate forms were used for each of the three databases
(experimental-vs.-control ES, gains ES, or correlational ES). Moderator variables
were identified through the previous related meta-analyses (Landers & Petruzzello,
1994; North et al., 1990) and suggestions made by authors in the gerontology
literature.
Design and Descriptive Characteristics. Glass, McGaw, and Smith (1981)
have argued that studies should not be excluded from re view based on methodologi-
cal rigor. Instead, they argue for examining study design issues in an a posteriori
fashion. Studies were coded for information regarding publication status (published
vs. unpublished) and comparison-group activity (for experimental-vs.-control ES
only). Studies were also coded for number of threats to internal validity based on the
description of threats by Campbell & Stanley (1963).
Participant Characteristics. Studies were coded for health status and fitness
level of the participants before treatment. This information was based on the
assessment of the author(s) of each study and/or on demographic and fitness-
assessment data. These data were then compared with standards established by the
American College of Sports Medicine (1995). In addition, experimental-versus-
control-ES studies were coded for equivalency of treatment, and control groups, on
the dependent variable of interest (mood) at pretest. This latter characteristic was
determined through t tests of the pretest means if equivalency had not been
evaluated by the author of each study.
Exercise Characteristics. Because exercise was the independent variable of
interest, a number of potential moderating characteristics were identified from the
available literature and previous meta-analyses. Studies were coded for exercise
type (or treatment type in general for gains-ES analysis), exercise paradigm (acute
vs. chronic), frequency of exercise, time per session, weeks of participation, and
intensity of exercise. In order of preference, intensity categories were determined
from (a) American College of Sports Medicine (1995) definitions, (b) National
Strength and Conditioning Association (Baechle, 1994) definitions, or (c) classifi-
cation by individual study authors. In addition, significance of cardiovascular-
fitness improvement was coded based on information provided in individual studies
regarding significantly increased VO2max (or predicted VO2max). In the rare cases that
this information was not provided, categorization was based on significant de-
creases in resting heart rate or significant improvements in resting blood pressure.
414 • Arent, Landers, and Etnier
Mood-Assessment Characteristics. Each study was coded for assessment of
positive and/or negative affect. Improvement in positive affect was defined as an
increase in score for a particular scale on which a higher score would be considered
"better." Likewise, improvement in negative affect was defined as a decrease in
score for a particular scale on which a lower score would be considered better. These
dimensions of affect were identified as described by Watson and Tellegen (1985)
in their two-factor structure of affect.
CODER RELIABILITY
The primary author coded all studies. Orwin (1994) has suggested that, when this
is the case, potential coder drift should be assessed. This process involved selecting
10 of the coded studies at random and recoding them. A per-case agreement rate
(number of variables coded the same divided by the total number of variables coded)
was calculated for each study. A mean agreement rate of .90 was required to be
considered acceptable.
Analyses
Separate analyses were conducted for each of the three categories of ES. After
calculating the ESs for each study, overall ESs were calculated for experimental-
versus-control-ES, gains-ES, and correlational-ES data. Average ESs were also
calculated for each level of the moderator variables. Tests for homogeneity were
conducted. This procedure is similar to an ANOVA in that the total variance (QT)
can be partitioned into between (QB) and within (QW) variance (Hedges & Olkin,
1985). These analyses allow the researcher to determine whether or not all the ESs
are derived from a homogeneous sample. If the sample is heterogeneous, one can
then use the test of homogeneity to determine which moderators are potentially
accounting for differences in ES.
QT is tested against a critical value of a X2 distribution (df= number of ES -
1) to determine whether all ESs are homogeneous. If QT is not significant, the ESs
are assumed to be homogeneous, and, except for testing a priori hypotheses,
analysis stops. If Q
T is significant, however, this indicates that the ESs are
heterogeneous. If this is the case, moderator-variable effects are examined by
comparing the appropriate QB to a corresponding critical value of a X2 distribution
(df = number of levels of the moderator variable - 1). In the event that Q
B is
significant, the particular moderator variable contributes to differences among ESs.
In this case, confidence intervals are computed to test whether or not average ESs
at each level are significantly different from zero. In addition, follow-up tests are
conducted to determine differences between the levels of each significant modera-
tor variable. These contrast procedures for fixed-effects models are outlined by
Hedges and Olkin (1985). Within each moderator category, only moderator levels
with at least five effect sizes were included in the analyses in order to improve
interpretability and stability of results. If Q H is significant, QW can be tested (using
df = number of ESs - number of levels of the corresponding moderator) to determine
the homogeneity within the levels of the moderating variable. A significant Q
W
simply implies that there are different groups of ESs within the levels of the
Effects of Exercise on Mood in Older Adults • 415
moderator variable of interest. In this case, interactions among variables can be
tested.
Because of the lack of information needed to calculate ES variances within the
correlational studies, one-way ANOVAs were used to test moderator variables if the
ESs demonstrated a normal distribution (Wolf, 1986). Significant omnibus F values
were further examined using Scheffé post hoc tests, as has been done in previous
meta-analyses (Etnier et al., 1997; Petruzzello, Landers, Hatfield, Kubitz, &
Salazar, 1991). Confidence intervals were used to test whether ESs were different
from zero. As with the other two data sets, only moderator levels with at least five
effect sizes were included in the analysis.
Results
The assessment of coder drift using per-case agreement rate (Orwin, 1994)
indicated that reliability was acceptable. The mean agreement rate for the 10
randomly selected studies was .98. Agreement rates ranged from .94 to 1.00. Results
for each grouping of ESs are presented separately.
EXPERIMENTAL-VERSUS-CONTROL-GROUP ES
The overall mean ES for mood, based on a total of 61 ESs, was 0.24 (SD = 0.50),
which was significantly greater than zero. The test of homogeneity was also
significant, QT(60) = 114.10, p < .001, warranting further examination of potential
moderator variables. Before continuing, however, the effect of nonequivalence on
the dependent variable (mood) at pretest was assessed. The QB value was signifi-
cant, QB(2) = 29.38, p < .001. Studies using a comparison group that had initially
better mood at pretest were associated with an average ES of -0.34 (SD = 0.25, p <
.05, n = 10). Because one of the basic assumptions of comparing groups in
experimental research at posttest is that they are equivalent on the dependent
variable of interest at pretest, these ESs (n = 10) were remove d from further analy sis.
It was determined that a negative effect of this magnitude could significantly
confound moderator-variable conclusions and was not representative of the overall
effect. This removal, however, did not affect inclusion in gains-ES analysis, as
nonequivalence was effectively controlled for because of the pre/post nature of the
comparisons. The overall mean ES for mood was recalculated based on the
remaining 51 ESs and was found to be 0.34 (SD = 0.45, p < .05). This indicates that
global mood improved 0.34 of a standard deviation with exercise, thus supporting
Hypothesis 1. QT was recalculated and was still found to be significant, QT(50) =
80.76, p < .005. A total of 7 out of 10 moderator variables had a QB value significant
at the p < .05 level. All moderator variables, their Q B values, significance lev els, and
corresponding ESs are summarized in Table 1. Significant moderator variables are
examined next in greater detail.
Design and Descriptive Characteristics. Of the design and descriptive charac-
teristics coded, only comparison-group activity had a significant QB value.
Use of a no-treatment comparison group, motivational comparison group, or
a flexibility/yoga comparison group produced average ESs significantly different
from zero. Use of a motivational control group was associated with the largest
Moderator variable
Significance
df level ES SD n p
Design and Descriptive Characteristics
Publication status
published
unpublished
Comparison-group activity
no treatmenta
motivationalb
flexibility/yogac
exercise
c
3.57
19.68
1
3
n.s.
p<.001
0.25
0.45
046
0.72
0.09
0.10
0.30
0.56
0.35
0.43
0.26
0.60
29
22
23
6
12
10
*
*
*
*
*
Exercise Characteristics
Type of exercise
cardiovascular
resistance training
mixed
Frequency
<3 day/weeka
>3 day/weekb
Time per session
variable
(self-selected)a
35-45 minb
>45 minc
Weeks of participation
1-6 weeksa
7-12 weeksa
>12 weeksb
Intensity of exercise
high
a
medium
a
lowb
Exercise-group fitness increase
not reporteda
yes
b
no
3.89
9.17
12.80
7.40
7.38
6.94
2
1
2
2
2
2
n.s.
p < .005
p < .005
p < .025
p<.05
p<.05
0.55
0.26
0.38
0.49
0.69
0.28
0.86
0.05
0.36
0.48
0.45
0.19
0.29
0.26
0.58
0.56
0.26
0.30
0.46
0.43
0.43
0.45
0.42
0.50
0.49
0.40
0.59
0.39
0.27
0.46
0.40
0.48
0.61
0.41
7
30
10
11
8
43
7
9
33
14
10
23
9
30
12
9
35
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
416 • Arent, Landers, and Etnier
Table 1 Homogeneity Tests (QB) and Post Hoc Analyses for Moderator
Variables: Experimental-Versus-Control ES
(continued)
Effects of Exercise on Mood in Older Adults • 417
Table 1 (continued)
Note: Variables that do not share at least one common superscript differ significantly at
p < .05. n.s. = nonsignificant.
*Differ from zero at p < .05
effect, but use of a no-treatment group was still associated with a larger effect than
were comparison groups doing flexibility/yoga training or exercise differing in
intensity and/or type from the experimental exercise group. The latter two were not
different from each other.
Exercise Characteristics. The Q B value was significant for 6 of the 7 modera -
tor variables describing exercise characteristics: frequency of exercise, time per
session, weeks of participation, intensity of exercise, exercise-group fitness in-
crease, and comparison-group fitness increase.
The largest average ES for frequency of exercise was related to the studies in
which participants exercised <3 days per week. This effect was significantly
different from zero. It was also significantly different from the average ES
associated with exercising >3 clays per week (which was also significantly different
from zero).
Surprisingly, the largest average effect for time per session was associated
with exercise bouts that were self-selected and thus variable in duration. Exercise
bouts in which time was controlled and of a sufficient duration to potentially
produce cardiovascular benefit (i.e., 35-45 min) were not significantly different
from zero. The exception was exercise that lasted >45 min, which was significant.
All levels of weeks of participation were associated with ESs significantly
greater than ze ro. The l argest average ESs were associated with 1 - to 6- and 7- to 12-
week protocols, which were not different from each other but were both signifi-
cantly greater than protocols lasting longer than 12 weeks. This result does not
provide support for Hypothesis 2 and might indicate one way in which elderly
participants respond differently than do younger participants to exercise.
Moderator variable
Significance
QBdf
level
ES SD n p
no
a
Exercise Characteristics (continued)
Comparison-group fitness increase
not reported
a
yes
b
no
c
10.84 2
p<0.1
0.53
-0.02
0.37
0.53
0.53
0.34
11
7
33
*
*
Mood-Assessment Characteristics
Mood assessed
positive affect
negative affect
0.002
1
n.s.
0.33
0.35
0.71
0.41
6
45
*
418 • Arent, Landers, and Etnier
Intensity of exercise was also found to be a significant moderator variable. All
levels were found to be significantly different from zero, but the average effect
associated with low-intensity exercise was significantly greater than that associated
with either medium or high intensities. The latter two did not differ from each other.
Although a significant Q W value, Q W (48)= 73.38, p < .025, was observed, a detailed
analysis of an Intensity x Exercise-Type interaction was not feasible because of the
small number of ESs constituting a number of the cells of the interaction.
The presence or absence of an exercise-group fitness increase was a signifi-
cant moderating variable, and all levels were significantly different from zero.
Surprisingly, studies that reported no fitness increase were associated with a larger
ES than were studies reporting a fitness increase. This would appear to contradict
Hypothesis 4. However, there was also a large effect associated with studies that did
not report changes in fitness altogether. This might confound the conclusions
somewhat, because it is impossible to determine under which classification these
nine ESs might actually fit.
The presence or absence of a comparison-group fitness increase was also a
significant moderator. Significant effects were associated with studies in which a
fitness increase was not reported or it was found that there was no fitness increase
in the comparison group. A fitness increase in the comparison group produced a
negative average effect but was not significant. This finding provides some indirect
support for Hypothesis 4, because the smallest difference in mood between the
treatment and comparison groups occurred when there was an increase in fitness in
the comparison group. In studies in which there was not an increase in fitness in the
comparison group, the treatment group had a relative improvement in mood.
Mood-Assessment Characteristics. Type of mood assessed (PA or NA) was
not associated with a significant QB value. This would indicate that both PA and NA
were equally improved with exercise, because both levels are significantly different
from zero. This provides further support for Hypothesis 1.
GAINS ES (PRETE ST-POST TEST COMPARISONS)
Although the homogeneity value was relatively large, it failed to reach significance,
QT(82) = 98.77, p > .05. Therefore, only moderator variables related to a priori
hypotheses were examined. These moderators are summarized in Table 2. Average
global-mood ESs were first determined for all exercise groups versus all control
conditions. The average ES for global mood in exercisers was 0.38. The average ES
for global mood in control groups was 0.06. The average ES for global mood was
significantly greater in exercisers than in control groups, p < .001; Hypothesis 1 was
thus supported. To further analyze these predicted differences between exercisers
and controls, differences in positive and negative affect were assessed. Type of
mood (PA vs. NA) was broken down by group (exercise vs. control). Exercise was
associated with an average ES of 0.35 (SD = 0.43, n = 12, p < .05) for positive affect
and an average ES of 0.39 (SD = 0.41, n = 38, p < .05) for negative affect. These
values were not different from each other (p < .05), thus providing further support
for Hypothesis 1. The corresponding ESs for PA and NA in controls were 0.16 (SD
= 0.25, n = 5,p< .05) and 0.04 (SD = 0.24, n = 28, p > .05), respectively. Using gains
scores, exercise was associated with an average improvement in mood (both
Effects of Exercise on Mood in Older Adults • 419
Table 2 Homogeneity Tests (QB) and Moderator Variable Summaries:
Gains ES
Note: Because of a nonsignificant Q
T statistic, comparisons between group levels were
conducted only for those variables related to a priori hypotheses. Variables that do not
share at least one common superscript differ significantly at p < .05. n.s. = nonsignificant.
*Differ from zero at p < .05.
A priori
moderator variable
Activity grouping
controlb
exercise
a
Type of activity
no activitya
motivational controla
yoga/flexibilitya
cardiovascular exerciseb
resistance trainingc
mixed (cardiovascular
+ resistance)b
Intensity
no HR increasea
highb
medium
b
lowb
Weeks of participation
0 weeks (acute)a
1-6 weeksa
7-12 weeksc
>12 weeksa
Initial health status
not reported
healthy + active
healthy + sedentary
mixed (diseased
+ healthy/sedentary)
Fitness increase
not reporteda
yesb
noa
Q
B
16.10
31.27
12.99
15.84
5.78
14.20
df
1
5
3
3
3
2
Significance
level
p < .001
p < .001
p< .01
p<.005
n.s.
p < .001
ES
0.06
0.38
-0.01
0.12
0.12
0.26
0.80
0.37
0.04
0.29
0.38
0.34
0.06
0.46
0.26
0.10
0.35
0.27
0.19
0.44
0.16
0.48
0.16
SD
0.25
0.41
0.24
0.28
0.18
0.45
0.24
0.24
0.26
0.45
0.45
0.31
0.36
0.44
0.31
0.29
0.40
0.38
0.33
0.46
0.34
0.45
0.29
n
33
50
15
9
9
30
8
12
24
12
25
22
12
21
29
21
12
6
47
14
25
26
32
p
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
420 • Arent, Landers, and Etnier
positive and negative affect) in the elderly equal to almost two fifths of a standard
deviation. This would be considered at least a moderate effect.
Type of activity was examined in order to determine whether type of exercise
was associated with different ESs. Type of activity was significant and was further
differentiated for comparisons of individual levels. Cardiovascular exercise,
resistance-training exercise, and mixed (cardiovascular + resistance training) exer-
cise were all associated with effects significantly different from zero. Resistance
training produced significantly better effects than all other types of activity. Yoga/
flexibility, motivational control groups, and no-treatment controls had average ESs
that were not different from each other or from zero. Taken as a whole, Hypothesis
1 appears to be supported, because the exercise groups produced the largest ESs
when contrasted with the activities associated with the comparison groups.
The effects of exercise intensity on mood in the elderly were also evaluated.
High, medium, and low intensities were all significantly different from zero but not
from each other. Only non-heart-rate-increasing activities failed to reach signifi-
cance. It appears that the significance of intensity is a result of the difference
between nonexercise activities and exercise. There was also a significant effect for
weeks of participation. Acute studies were not associated with effects different from
zero. Although studies lasting longer than 12 weeks were not different from the
acute studies, they were associated with an effect different from zero. Training
protocols lasting 1-6 or 7-12 weeks were different from zero and from each other
(as well as from the acute and >12-week studies), with protocols of 1-6 weeks being
associated with the largest effects. Based on these results, Hypothesis 2 does not
appear to be supported. Although significant improvements in mood are seen for
some of the more lengthy participation periods, there are also obvious benefits
derived from shorter studies. A nonsignificant QW value prevented further analysis
of interactions.
Initial health status was examined in order to test Hypothesis 3, but this was
not a significant moderator variable. All levels were associated with effects
significantly different from zero. Most of the ESs came from studies using healthy/
sedentary or mixed populations. The mixed samples were composed of healthy/
sedentary individuals and diseased individuals. Exercise was associated with
significantly improved mood across all levels of initial health status.
Hypothesis 4 received considerable support based on the significance of a
cardiovascular-fitness variable. Although all levels were significantly different
from zero, studies in whic h a cardiovascular-fitness increase was reported were
associated with a significantly larger average ES than were studies with no fitness
increase or studies not reporting fitness increases. This would suggest that mood
improvement in the elderly might be moderated by physiological improvement.
CORRELATIONAL ES
The overall mean ES for correlational studies was 0.46 (SD = 0.27, n = 24, p< .05),
indicating that mood is better in elderly individuals who participate in physical
activity/exercise than in those who do not. This result is consistent with the
magnitude of the effects seen for experimental-versus-control ESs and gains ESs.
The large effect was seen for both positive (ES = 0.42, SD = 0.38, n=6, p< .05) and
negative (ES = 0.47, SD = 0.24, n = 18, p < .05) affect across all forms of physical
Effects of Exercise on Mood in Older Adults • 421
activity. Further analysis of types of activity was not feasible because of the small
number of ESs contributing to a number of the levels of this moderator variable. A
plot of the means revealed that these ESs were distributed normally, allowing
further analysis of moderator variables by way of the omnibus F test and contrasts.
The only moderator variable that had sufficient numbers of ESs in all levels to
warrant testing, however, was publication status, which was not a significant
moderator, F (1,22) = .239, p > .05. Both published (ES = 0.48, SD = 0.19, n = 13)
and unpublished (ES = 0.43, SD = 0.35, n = 11) studies were significantly different
from zero but not different from each other. Although the small number of
contributing correlational studies does not lend itself to in-depth analysis, the
overall results appear to provide support for the conclusions reached with the
experimental-versus-control ESs and gains ESs.
Discussion
The results of this meta-analysis indicate that, overall, exercise is associated with
improved mood in the elderly, thus supporting the main hypothesis. In studies
comparing an exercise group with some form of a control group, exercise is related
to enhanced mood. This improvement is almost two fifths of a standard deviation,
ES=0.34, and appears to be the result of equivalent improvements in both negative
and positive affect, ES = 0.35 and 0.33, respectively. In studies comparing pre- to
posttest changes in mood, exercise is also associated with improved mood in the
elderly. Again, this improvement is almost two fifths of a standard deviation, ES =
0.38, and the effect appears to be relatively equivalent for both positive and negative
affect, ES = 0.35 and 0.39, respectively. Correlational studies also support these
results. Physically active elderly individuals appear to have an enhanced global
mood in comparison with physically inactive elderly individuals. This difference is
almost one half of a standard deviation, ES = 0.48. Taken as a whole, the results are
remarkably consistent across ES groupings and provide considerable support for the
beneficial effects of exercise on mood in the elderly.
These results are consistent with the conclusions of a narrative and two meta-
analytic reviews examining the effects of exercise as an alternative treatment for
anxiety and depression among younger and older adults (Landers & Petruzzello,
1994; Moore & Blumenthal, 1998; North et al., 1990). The magnitude of the overall
effects found in the present study is in the range of findings in other meta-analyses
that have examined the effects of exercise on depression (North et al.) and anxiety
(Landers & Petruzzello) in younger adults. Although Moore and Blumenthal's
narrative review with older adults focused on a specific construct included under
mood (i.e., depression) and on primarily aerobic exercise, the overall conclusions
support the role of exercise in reducing negative affect. As with any literature
review, however, the conclusions reached by Moore and Blumenthal and by this
meta-analysis are somewhat limited by the methodological problems inherent to the
included studies.
In particular, this meta-analysis has identified a number of concerns associ-
ated with this area of research. The first concern is the number of studies that
included groups that were not equivalent on mood at pretest. Comparisons with
control groups that were initially better at pretest produced deceptively large
422 • Arent, Landers, and Etnier
negative ESs that did not appear representative of the effect in general. By using
gains scores as a measure of magnitude rather than experimental-control compari-
sons, nonequivalence is effectively removed. In the experimental-versus-control
data set, the average ES obtained once comparison groups that were better at pretest
were removed from further analysis (ES = 0.34) was almost identical to that
obtained for exercise groups using gains scores (ES=0.38). This effect was actually
reversed (ES = -0.34) in studies using comparison groups that had better mood
scores at pretest. The effect of nonequivalence of mood scores at pretest alone might
help explain why there are so many inconsistencies in narrative reviews and why the
use of meta-analytic techniques can help clarify and identify these effects.
Another area of concern is the lack of studies examining the effects of exercise
on positive affect in the elderly. Although the effect of exercise on PA was
equivalent to that seen for NA, the number of ESs on which these conclusions are
based were quite discrepant. Only six ESs were derived from studies examining PA
for experimental-versus-control-ES analysis, compared with the 45 ESs for NA.
Gains-ES analysis for PA was based on only 17 ESs, compared with the 66 for NA.
The research focus appears to be on reducing bad mood rather than increasing good
mood. Although it might seem to be a semantic issue, exercise might be much more
appealing to the elderly if they were told it could make them feel "good" rather than
simply "less poorly."
In addition to using moderator variables to identify sources of concern in the
literature, examining certain moderator variables can also provide some insight into
potential mechanisms driving the exercise-mood relationship. Exercise type is one
such variable. In the two previous related meta-analyses, resistance training was
sometimes associated with improvements in mood (North et al., 1990), and at other
times it was associated with no improvements in mood (Landers & Petruzzello,
1994). In the present meta-analysis with older adults, the experimental-versus-
control and gains-score analyses revealed larger ESs for resistance training than for
cardiovascular exercise. The average gains ES for resistance training was particu-
larly large (0.80) and was considered more representative than the experimental-
versus-control ES. The latter ES was influenced by the type of control group
employed, which often consisted of a treatment that would also improve, thereby
minimizing the difference between groups. With the gains ES the comparisons were
pre-post comparisons within the same group of participants and thus better
reflected the effect of the treatment. The average effects associated with mixed
(cardiovascular + resistance training) exercise also support the influence of resis-
tance training over that of cardiovascular training alone. These results are not
surprising when one considers how physically and functionally beneficial strength
training is for an elderly population.
A decline in mood might well be expected to accompany the age-related loss
of functional abilities caused by deterioration of muscle mass and strength. Strength
training might provide the necessary stimulus to improve daily functioning. As seen
in previous studies (Cress et al, 1991; Fiatarone et al., 1990; Frontera et al., 1988),
the elderly obviously respond well physically to resistance training. Psychological
improvements might coincide with these physical improvements. The other possi-
bility is that weight training increases self-efficacy and self-confidence through a
series of graded mastery experiences. This graded mastery, in turn, might allow for
more effective coping, which could lead to improved mood.
Effects of Exercise on Mood in Older Adults • 423
Social-cognitive theory (Bandura, 1986) contends that perceptions of en-
hanced capabilities lead to increases in positive affect, and mastery has been
identified as one possible mechanism that could explain the exercise-mental health
relationship. There is considerable evidence to suggest that self-efficacy is related
to affective responses associated with exercise (McAuley, 1991; McAuley, Bane,
& Mihalko, 1995; McAuley & Coumeya, 1992). Furthermore, the use of physical
activity as a mastery experience that leads to increased self-efficacy has been
demonstrated in older adults (McAuley, Courneya, & Lettunich, 1991; McAuley,
Lox, & Duncan, 1993). Ewart, Stewart, Gillilan, and Kelemen (1986) examined the
usefulness of adding weight training to a walking/jogging exercise program for
elderly coronary artery disease patients. Results indicated that arm- and leg-strength
efficacy was enhanced in the weight-training group, a finding that might be
particularly important in light of the functional relevance of these abilities for the
elderly. It is important to acknowledge the potential benefits of strength training (in
addition to cardiovascular training) for the elderly and to increase the pool of studies
employing this form of exercise. It is obviously a viable form of exercise for this
population, but this area is severely lacking in well-developed psychological studies
including it in their protocols.
In addition to the psychologically oriented social-cognitive theory, a physi-
ologically oriented hypothesis has also been suggested as an explanation for the
exercise-mood relationship. This hypothesis, known as the monoamine hypothesis
(Morgan & 0' Connor, 1988), posits that neurotransmitters in the brain (which have
been linked to depression, anxiety, and other mood constructs) are changed by both
acute and chronic exercise. Most of the research examining this hypothesis,
however, has focused on acute bouts of aerobic exercise and younger participants.
Considering the lack of effect for a single acute bout of exercise (and the larger
effects seen for resistance training) in this study, the monoamine hypothesis might
not be the most plausible explanation for the exercise-mood relationship seen in the
elderly. Further research into this potential mechanism is needed.
It also does not appear that the distraction hypothesis is supported. The
distraction hypothesis posits that exercise acts as a "time-out" from daily stressful
events. A caveat to this is that exercise should not produce greater effects than do
control conditions that conceivably also produce such a time-out. As found in this
study, however, improvements in mood for motivational control groups and yoga/
flexibility were associated with mean ESs that were significantly lower (and not
different from zero) than those for exercise (cardiovascular, resistance training, or
mixed).
In addition to providing a means of assessing proposed mechanisms, modera-
tor variables can also be used to examine contentions made by other authors.
Brown's (1992) contention that the lack of significant findings for mood effects
with exercise in the elderly is a result of insufficient training intensity was not
supported by the current findings. In fact, the greatest improvements in mood were
associated with the lowest intensity of exercise when analyzing the results for the
experimental-versus-control ESs. Even when looking at pre-post comparisons
(gains ES), all levels of intensity were at least equivalent. It might be that some of
these samples of elderly individuals were in such an unfit condition that even low-
intensity exercise produced perceived fitness gains. The effect of intensity on mood
might also be influenced by the type of exercise. The influence of exercise intensity
424 • Arent, Landers, and Etnier
on mood in the elderly obviously deserves further exploration, especially with
studies that control for potentially confounding variables (e.g., duration, initial
fitness levels, exercise type).
The findings for the effects of study duration (i.e., number of weeks) were not
consistent with those found by North et al. (1990). In that meta-analysis, which did
not include elderly participants, reductions in depression were larger for studies
using training protocols longer than 15 weeks. The present analysis found that
improved mood was not necessarily related to increased number of weeks of
participation. In general, the effects appeared to be rather equally distributed across
levels of weeks of participation for chronic exercise. This would appear to provide
little support for Hypothesis 2. Nonetheless, future research should examine
whether these findings, as well as those for intensity, interact with initial fitness
levels (McAuley, Mihalko, & Bane, 1996). One potential confound that is often
encountered when trying to determine the effects of chronic exercise (particularly
on mood states) is the impact of the most recent bout of acute activity. Although it
is possible that improvements in mood were the result of an acute bout of exercise
before posttest rather than the result of changes that took place with chronic activity,
this possibility is made less plausible in light of the fact that the effect size for acute
exercise was small (ES=0.06) and nonsignificant. Although the potential influence
of an acute bout cannot be ignored, the results do not point to this as the most likely
explanation for the changes in affect.
Dunn and Blair (1997) suggest a number of criteria for evaluating scientific
literature for public policy decisions. These criteria include consistency, whether
the study is sequenced appropriately, whether the results are plausible and coherent,
the strength of relation in terms of relative risk, and whether there is a biological
gradient. In terms of this meta-analysis, the findings that are relevant to these criteria
are as follows: (a) 82% of the effect sizes were in the predicted direction, demon-
strating that exercise is related to improvements in mood; (b) 78% of the effect sizes
from the experimental studies were in the positive direction, supporting the
predicted sequencing that exercise preceded the improvements in mood; (c)
findings suggest that plausible mechanisms might include graded mastery experi-
ences and/or improvements in self-efficacy, as well as fitness improvements being
implicated as having at least some role; and (d) considering the magnitude of the
effects found, risk of exercise appears relatively small—only one study reported any
exercise dropouts because of injury. Taken as a whole, this meta-analysis demon-
strates at least the consistent findings, the appropriate cause-and-effect sequencing,
the plausible and coherent results, and the minimal risk needed to justify making
conclusions concerning public policy.
Evidence for a biological gradient, though, is less conclusive because of the
nature of the findings for the role of fitness improvements. That is, there was not
absolute agreement across the ES groupings concerning the importance of a fitness
improvement for mood improvements. However, this may beg an important
question: Are we examining the most important aspects of fitness improvements in
this particular population? The most accepted means of determining a fitness
improvement has historically been to assess VO2max changes. This practice unfor-
tunately ignores a relatively important (but often overlooked) component of fitness
that might be of greater functional importance to the elderly—strength. Future
Effects of Exercise on Mood in Older Adults • 425
studies should take this into consideration when determining which measures might
be the most useful and meaningful indicators of a fitness improvement in an aged
population. Furthermore, future research must examine these effects while making
a concerted effort to better assess the impact of exercise dose-response issues, such
as exercise intensity and duration, on mood improvements in the elderly.
Summary
Based on these findings, it is apparent that exercise is associated with significant
pre- to posttreatment improvements in mood in the elderly, especially when
compared with a no-treatment control, motivational control, or yoga/flexibility
condition. Based on the significant contribution of variables such as fitness im-
provements, low initial fitness levels, and frequency of exercise, it would appear
that these mood improvements are the result (at least somewhat) of physiological
improvements resulting from exercise. The lack of significance (at least in the
predicted "physiologically better" direction) of variables such as intensity of ex-
ercise and weeks of training, however, indicates that these effects might be caused
by much more than this. It might very well be that it is the combination of the graded
mastery experience provided by exercise and the objective physiological improve-
ments that causes mood enhancement in the elderly.
"Mood-improving" effects are seen for all types of exercise—particularly
resistance training—if examining pre- to posttest changes in mood. The most
consistent improvements in mood also appear to be associated with exercise done
fewer than 3 days per week, exercise done for more than 45 min or based on
participant "needs," and low- to medium-intensity exercise. Future research should
focus on further examining the effects of exercise on PA and on the effects of
strength training for the elderly. There is much to be learned regarding optimal
intensity, frequency, and duration of resistance-training programs in this popula-
tion, particularly in diseased segments of the population. Furthermore, a concerted
effort should be made to examine the effects of a combined cardiovascular and
strength-training protocol. This is potentially the most physiologically beneficial
exercise protocol for this age group, yet studies examining it are severely lacking.
As a final note, research in this area must make more of an attempt to use groups that
are equivalent at baseline on the mood measure of interest. Small, but significant,
improvements in mood with exercise can be washed out if this is not controlled for
in the course of study design.
References
References marked with an asterisk indicate studies included in the meta-analysis.
American Co llege o f Sports Medicine. (1995). ACSM's guidelines for exercise testing and
prescription (5th ed.). Baltimore: Williams & Wilkins.
Baechle, T.R. (1994). Essentials of strength training and conditioning. Champaign, IL:
Human Kinetics.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory.
Englewood Cliffs, NJ: Prentice-Hall.
426 • Arent, Landers, and Etnier
*Beal, R.K. (1985). The effect of a dance/movement activity program on the successful
adjustment to aging in the active/independent older adult. Unpublished doctoral
dissertation, University of Cincinnati.
*Bennett, J., Carmack, M.A., & Gardner, V.J. (1982). The effect of a program of physical
exercise on depression in older adults. Physical Educator, 39, 21-24.
*Blumenthal, J.A., Emery, C.F., Madden, D.J., George, L.K., Coleman, R.E., Riddle, M.W.,
McKee, D.C., Reasoner, J., & Williams, R.S. (1989). Cardiovascular and behavioral
effects of aerobic exercise training in healthy older men and women. Journal of
Gerontology, 44, M147-M157.
*Blumenthal, J.A., Emery, C.F., Madden, D.J., Schniebolk, S., Walsh-Riddle, M., George,
L.K., McKee, D.C., Higginbotham, M.B., Cobb, F.R., & Coleman, R.E. (1991). Long-
term effects of exercise on psychological functioning in older men and women.
Journal of Gerontology, 46, P352-P361.
*Blumenthal, J.A., Schocken, D.D., Needles, T.L., & Hindle, P. (1982). Psychological and
physiological effects of physical conditioning on the elderly. Journal of Psychoso-
matic Research, 26, 505-510.
Botwinick, J. (1977). Intellectual abilities. In J.E. Birren & K.W. Schaie (Eds.), Handbook
of the psychology of aging (pp. 580-605). New York: Van Nostrand Reinhold.
*Brassington, G.S. (1993). Relationship between aerobic exercise and sleep quality in
elderly individuals. Unpublished master's thesis, San Jose State University.
Brown, D.R. (1992). Physical activity, ageing, and psychological well-being: An overview
of the research. Canadian Journal of Sport Science, 17(3), 185-193.
*Buccola, V.A., & Stone, W.J. (1975). Effects of jogging and cycling programs on physio-
logical and personality variables in aged men. Research Quarterly, 46,134-139.
Campbell, D.T., & Stanley, J.C. (1963). Experimental and quasi-experimental designs for
research. Chicago: Rand McNally.
*Carpenter, J.R. (1982). The effect of exercise on several physiological and psychological
variables among three elderly female subjects. Unpublished doctoral dissertation,
West Virginia University, Morgantown.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale,
NJ: Erlbaum.
Crandall, R.C. (1991). Gerontology: A behavioral science approach. New York: McGraw-
Hill.
Crase, D., & Rosato, F.D. (1979). Exercise and aging: New perspectives and educational
approaches. Educational Gerontology, 4, 367-376.
Cress, M.E., Thomas, D.P., Johnson, J., Kasch, F.W., Cassens, R.G., Smith, E.L., & Agre,
J.C. (1991). Effect of training on VO
2max
, thigh strength, and muscle morphology in
septuagenarian women. Medicine and Science in Sports and Exercise, 23, 752-758.
*Delpape, B.L. (1986). The effects of a short-term exercise program on three levels of
function in an elderly population. Unpublished doctoral dissertation, The University
of Southern Mississippi.
Dunn, A.L., & Blair, S.N. (1997). Exercise prescription. In W.P. Morgan (Ed.), Physical
activity and mental health (pp. 49-62). Washington, DC: Taylor & Francis.
*Emery, C.F.,&Gatz,M. (1990). Psychological and cognitive effects of an exercise program
for community-residing older adults. The. Gerontologist, 30, 184-188.
Etnier, J.L., Salazar, W., Landers, D.M., Petruzzello, S.J., Han, M., & Nowell, P. (1997). The
influence of physical fitness and exercise upon cognitive functioning: A meta-
analysis. Journal of Sport & Exercise Psychology, 19, 249-277.
Effects of Exercise on Mood in Older Adults • 427
Ewart, C.K., Stewart, K.J., Gillilan, R.E., & Kelemen, M.H. (1986). Self-efficacy mediates
strength gains during circuit weight training in men with coronary artery disease.
Medicine and Science in Sports and Exercise, 18, 531-540.
*Favilla, G. (1992). The role of self-efficacy as a mediator in the relationship between muscle
strength training and mood in the elderly. Unpublished doctoral dissertation, The
California School of Professional Psychology, San Diego.
*Ferretti, A.J. (1987). Effects of exercise on young and old adults applying different program
strategies. Unpublished doctoral dissertation, The University of Southern Missis-
sippi.
Fiatarone, M.A., Marks, E.C., Ryan, N.D., Meredith, C.N., Lipsitz, L.A., & Evans, W.J.
(1990). High intensity strength training in nonagenarians, effects on skeletal muscle.
Journal of the American Medical Association, 263, 3029-3034.
Fillingim, R.B., & Blumenthal, J.A. (1993). Psychological effects of exercise among the
elderly. In P. Seraganian (Ed.), Exercise psychology: The influence of physical
exercise on psychological processes (pp. 237-254). New York: Wiley.
*French, S.L. (1995). The self-management of osteoarthritis in older women. Unpublished
doctoral dissertation. Queen's University, Kingston, Ontario, Canada.
Frontera, W.R., Meredith, C.N., O'Reilly, K.P., Knuttgen, W.G., & Evans, W.J. (1988).
Strength conditioning in older men: Skeletal muscle hypertrophy and improved
function. Journal of Applied Physiology, 64, 1038-1044.
Gauvin, L., & Brawley, L.R. (1993). Alternative psychological models and methodologies
for the study of exercise and affect. In P. Seraganian (Ed.), Exercise psychology: The
influence of physical exercise on psychological processes (pp. 146-171). New York:
Wiley.
Glass, G.V. (1977). Integrating findings: The meta-analysis of research. Review of Research
in Education, 5, 351-379.
Glass, G.V., McGaw, B., & Smith, M.L. (1981). Meta-analysis in social research. Beverly
Hills, CA: Sage.
Hagberg, J.M., Graves, J.E., & Limacher, M. (1989). Cardiovascular responses of 70- to 79-
year-old men and women to exercise training. Journal of Applied Physiology, 66,
2589-2594.
*Hatfield, B.D., Goldfarb, A.H., Sforzo, G. A., & Flynn, M.G. (1987). Serum beta-endorphin
and affective responses to graded exercise in young and elderly men. Journal of
Gerontology, 42, 429-431.
*Hawkins, W.E., Duncan, D.F., & McDermott, R.J. (1988). A health assessment of older
Americans: Some multidimensional measures. Preventive Medicine, 17, 344-356.
Hedges, L.V. (1981). Distribution theory for Glass' estimator of effect size and related
estimators. Journal of Educational Statistics, 6, 107-128.
Hedges, L.V., & Olkin, I. (1985). Statistical methods for meta-analysis. New York:
Academic Press.
Hoeger, W.W.K., & Hoeger, S.A. (1996). Fitness & wellness (3rd ed.). Englewood, CO:
Morton.
*Holder, J.D. (1982). The effects of low level aerobic activity upon systolic blood pressure,
heart rate, and depression levels among the elderly. Unpublished doctoral disserta-
tion, Boston University School of Education.
Jacewicz, M.M., & Hartley, A.A. (1987). Age differences in the speed of cognitive
operations: Resolution of inconsistent findings. Journal of Gerontology, 42, 86-
428 • Arent, Landers, and Etnier
Johnson-Pawlson, J., & Koshes, R. (1985). Exercise is for everyone. Geriatric Nursing, 6,
322-325.
*Knittle, V.R. (1988). The effect of twelve weeks of exercise on depression, self-esteem, and
locus-of-control in a wellness program sample of women aged 65 and older.
Unpublished master's thesis, The University of Arizona.
Kohrt, W.M., Malley, M.T., & Coggan, A.R. (1991). Effects of gender, age, and fitness level
on response of VO2max to training in 60 -71-year-olds. Journal of Applied Physiology,
71, 2004-2011.
Landers, D.M., & Petruzzello, S. (1994). Physical activity, fitness, and anxiety. In C.
Bouchard, R.J. Shephard, & T. Stevens (Eds.), Physical activity, fitness, and health
(pp. 868-882). Champaign, IL: Human Kinetics.
Lazarus, R.S. (1991). Progress on a cognitive-motivational-relational theory of emotion.
American Psychologist, 46, 819-834.
Makrides, L., Heigenhauser, G.J.F., & Jones, N.L. (1990). High intensity endurance training
in 20-to 30- and 60- to 70-year-old healthy men. Journal of Applied Physiology, 69,
1792-1798.
McAuley, E. (1991). Efficacy, attributional, and affective responses to exercise participa-
tion. Journal of Sport & Exercise Psychology, 13, 382-393.
McAuley, E., Bane, S.M., & Mihalko, S.L. (1995). Exercise in middle-aged adults: Self-
efficacy and self-presentational outcomes. Preventive Medicine, 24, 319-328.
McAuley, E., & Courneya, K.S. (1992). Self-efficacy relationships with affective and
exertion responses to exercise. Journal of Applied Sport Psychology, 22, 312-326.
McAuley, E., Courneya, K.S., & Lettunich, J. (1991). Effects of acute and long-term exercise
on self-efficacy responses in sedentary, middle-aged males and females. The Geron-
tologist, 31, 534-542.
McAuley, E., Lox, D.L., & Duncan, T. (1993). Long-term maintenance of exercise, self-
efficacy, and physiological change in older adults. Journal of Gerontology, 48, P218-
P223.
McAuley, E., Mihalko, S.L., & Bane, S.M. (1996). Acute exercise and anxiety reduction:
Does the environment matter? Medicine and Science in Sports and Exercise, 11, 143-
182.
*McAuley, E., Shaffer, S.M., & Rudolph, D. (1995). Affective responses to acute exercise
in elderly impaired males: The moderating effects of self-efficacy and age. Interna-
tional Journal of Aging and Human Development, 41, 13-27.
*McNeil, J.K., LeBlanc, E.M., & Joyner, M. (1991). The effect of exercise on depressive
symptoms in the moderately depressed elderly. Psychology and Aging, 6, 487-488.
*Molloy, D.W., Beerschoten, D.A., Borrie, M.J., Crilly, R.G., & Cape, R.D.T. (1988). Acute
effects of exercise on neuropsychological function in elderly subjects. Journal of the
American Geriatric Society, 36, 29-33.
Moore, K.A., & Blumenthal, J.A. (1998). Exercise training as an alternative treatment for
depression among older adults. Alternative Therapies in Health and Medicine, 4, 48-
56.
Morgan, W.P., & O'Connor, P.J. (1988). Exercise and mental health. In R.K. Dishman (Ed.),
Exercise adherence: Its impact on public health (pp. 91-121). Champaign, IL: Human
Kinetics.
*Morton, A.G. (1992). Effects of an exercise program on elderly residents of a retirement
complex. Unpublished master's thesis, San Jose State University.
Effects of Exercise on Mood in Older Adults • 429
National Center for Health Statistics. (1993). Health, United States, 1992. Hyattsville, MD:
Public Health Service.
Netz, Y., & Jacob, T. (1994). Exercise and the psychological state of institutionalized elderly:
A review. Perceptual and Motor Skills, 79, 1107-1118.
*Nieman, D.C., Warren, B.J., Dotson, R.G., Butterworth, D.E., & Henson, D.A. (1993).
Physical activity, psychological well-being, and mood state in elderly women.
Journal of Aging and Physical Activity, 1, 22-33.
North, T.C., McCullagh, P., & Tran, Z.V. (1990). Effect of exercise on depression. Exercise
and Sport Science Reviews, 18, 379-415.
Orwin, R.G. (1994). Evaluating coding decisions. In H. Cooper & L.V. Hedges (Eds.), The
handbook of research synthesis (pp. 139-162). New York: Russell Sage Foundation.
*Palinkas, L.A., Wingard, D.L., & Barrett-Connor, E. (1990). Chronic illness and depressive
symptoms in the elderly: A population-based study. Journal of Clinical Epidemiol-
ogy, 43, 1131-1141.
*Parks, C. J. (1979). The effects of a physical fitness program on body composition, flexibility,
heart rate, blood pressure, and anxiety levels of senior citizens. Unpublished doctoral
dissertation, The University of Alabama, Tuscaloosa.
*Perri, S., II, & Templer, D.I. (1985). The effects of an aerobic exercise program on
psychological variables in older adults. International Journal of Aging and Human
Development, 20, 167-172.
Petruzzello, S.J., Landers, D.M., Hatfield, B.D., Kubitz, K.A., & Salazar, W. (1991). A meta-
analysis on the anxiety-reducing effects of acute and chronic exercise: Outcomes and
mechanisms. Sports Medicine, 11, 143-182.
Pfeiffer, E. (1977). Psychopathology and social pathology. In J.E. Birren & K.W. Schaie
(Eds.), Handbook of the psychology of aging (pp. 650-671). New York: Van Nostrand
Reinhold.
*Pierce, E.F., & Pate, D.W. (1994). Mood alterations in older adults following acute exercise.
Perceptual and Motor Skills, 79, 191-194.
*Reiter, M.A. (1981). Effects of a physical exercise program on selected mood states in a
group of women overage 65. Unpublished doctoral dissertation, Columbia University
Teachers College, Missouri.
Rogers, M.A., & Evans, W.J. (1993). Changes in skeletal muscle with aging: Effects of
exercise training. Exercise and Sport Sciences Reviews, 21, 65-102.
Rosenthal, R. (1994). Parametric measures of effect si ze. In H. Cooper & L.V. Hedges (Eds.),
The handbook of research synthesis (pp. 231-244). New York: Russell Sage Founda-
tion.
*Ruuskanen, J.M., & Ruoppila, I. (1995). Physical activity and psychological well-being
among people aged 65 to 84 years. Age and Ageing, 24, 292-296.
Schaie, K.W. (1994). The course of adult intellectual development. American Psychologist,
49, 304-313.
*Segebartt, K.S., Nieman, D.C., Pover, N.K., Arabatzis, K., & Johnson, M. (1988).
Psychological well-being in physically active and inactive healthy young old to very
old women. Annals of Sports Medicine, 4, 130-136.
Shephard, R.J. (1984). Management of exercise in the elderly. Canadian Journal of Applied
Sport Sciences, 9, 109-120.
Shephard, R.J. (1990). The scientific basis of exercise prescribing for the very old. Journal
of the American Geriatrics Society, 38, 62-70.
430 • Arent, Landers, and Etnier
*Silberman, J. (1985). The effects of type of physical fitness training, actual fitness &
perceived fitness on depression & self-concept in the elderly. Unpublished doctoral
dissertation, Hofstra University, New York.
Spirduso, W.W., & Mackie, K.S . (1995). Health, exercise, and emotional function. In W.W.
Spirduso (Ed.), Physical dimensions of aging (pp. 279-301). Champaign, IL: Human
Kinetics.
*Tack, B.B. (1991). Dimensions and correlates of fatigue in older adults with rheumatoid
arthritis. Unpublished doctoral dissertation, University of California, San Francisco.
*Tsutsumi,T. (1997). The effects of strength training on mood, self-efficacy, cardiovascular
reactivity, and quality of life in older adults. Unpublished doctoral dissertation,
Boston University.
Tuson, K.M., & Sinyor, D. (1993). On the affective bene fits of acu te aerobic exercise: Taking
stock after twenty years of research. In P. Seraganian (Ed.), Exercise psychology: The
influence of physical exercise on psychological processes (pp. 80-122). New York:
Wiley.
Unutzer, J., Patrick, D.L., Simon, G., Grembowski, D., Walker, E., Rutter, C, & Katon, W.
(1997). Depressive symptoms and the cost of health services in HMO patients aged
65 years and older: A 4-year prospective study. Journal of the American Medical
Association, 277, 1618-1623.
Watson, D., & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological
Bulletin, 98,219-235.
Wolf, F.M. (1986). Meta-analysis: Quantitative methods for research synthesis. Beverly
Hills, CA: Sage.
