This document discusses the procedures and methodology for conducting an experiment comparing the effectiveness of cognitive behavioral therapy (CBT) and yoga for treating stress. It outlines how to determine the population and recruit a representative sample, randomly assign participants to treatment groups, standardize the administration of CBT and yoga as the independent variables, blind assessors to condition assignments, and address potential confounds and ethical concerns. The goal is to design a controlled experiment that validly compares the two stress management therapies and allows results to be generalized to the target population.

1. SAMPLING
For what population do you want to test the new therapy?
Students should be able to give a concrete population to which
they would like to generalize their results. For example, perhaps
we are interested in seeing if our treatment is effective for all
veterans with PTSD. Again, further specification can make
implementation easier, but of course less specification might
enable broader generalization, as long as we sample well.
Describe your sample. Name 5 specific characteristics.
Students should be able to state who is in their sample, from
where they are obtaining their sample, and list at least 5
characteristics that are relevant to the example. Examples may
be age, gender, ethnic diversity, SES, or things more specific
for your psychological disorder. They will need to have these
characteristics when they consider the representativeness of the
sample to the population. These answers should be clear and
specific.
What does it mean to generalize? Why is generalization
important?
Use the definition from the list of concepts: “Generalization
means the results of the study can be applied to people other
than the actual participants of the study and to circumstances
outside the actual study.”
Mention externalvalidity as it relates to generalization &
representativeness. Use the vocabulary, or use the “language of
experimental method.” This is why defining our population and
sample clearly is important.
What specific methods will you use to recruit your participants?
Again, this will depend on your example so make sure it is
appropriate for your population/sample. For example, if you are
using veterans, it may make sense to recruit from the VA. Be
careful not to introduce samplingbias, which leads to an

2. unrepresentative sample, e.g. only people who use social media.
How could your methods of recruitment affect the
characteristics of your sample? Why is that important?
Consider vocabulary terms like selection, how
representativeness is affected, samplingbias, etc. within the
context of the example. The idea is to minimize a systematic
sampling bias and increase representativeness.
What incentives could you provide to increase participation?
How could these incentives affect or confound the results?
Reference information about the potential effects of incentives,
including samplingbias and consequent unrepresentativeness,
demandcharacteristics, subject bias, ethics risks, and other
potential confounds. Consider the possibility of
misinformation/faulty data in order to continue participation
and receive the incentive.
PROCEDURE (THERAPISTS)
Describe the basic procedure of your experiment.
Students can take some liberty here. They may need some
assistance designing how the specific treatments can be
implemented.
What specific part of the procedure will be different for the
different groups?
Identify the independent variable (the therapy) only.
How will you determine who the therapists are for the different
groups?
Consider competence, standardization, how therapists
operationalize the independent variable, and control for
experimenter effects with multiple therapists and randomization
of the therapists.
How will you make sure the therapy is provided correctly?
Students should come up with strategies to operationalize
treatment. Strategies could include training for competency,
monitoring of therapists, and using manualized treatments.
Will the therapist know to which therapy condition specific
subjects are assigned or can they be blind to the treatment

3. condition? If they can be blind, how will you achieve this?
A therapist can not be blind to the therapy they are providing.
What characteristics of the therapist could affect the results? If
so, how could you control for these potential experimenter
effects?
Examples of characteristics could include: demographics,
appearance, competence, personality (e.g. empathy,
interpersonal style, etc.).
These variables should be controlled with randomization across
these variables (except for competence), balanced conditions,
and if not we could evaluate the data (e.g. by gender of
therapist) after collection of the data for some of these effects.
Give an example of how the therapist’s expectations or bias
could affect or confound the results?
These confounds may result in a loss of experimental control if
they introduce any other variables, communicate their
expectations, or even inadvertently bias the subjects (see
examples from Ppt). Give special consideration to procedure
and the possibility of the therapist communicating expectations.
How could you control for these possibilities?
Similar to the response above: operationalization of
therapy/instruction & training for therapists, so no confounds
are introduced from experimenterbias. Also, randomization of
therapist assignment can help.
PROCEDURE (OUTSIDE THERAPY)
Describe the basic procedure of your experiment.
This will be specific to your example. Students may need
assistance with this but encourage them to brainstorm how they
would conduct the experiment.
What specific part of the procedure will be different for the
different groups?
Anything outside of the different therapies is NOT part of the
independent variable, so nothing should be different outside the
different types if therapy. Otherwise, this will introduce

4. confounds.
What instructions will you give to the participants?
Students should brain storm ideas but they need to make sure to
standardize the instructions to have experimental control.
Who will measure or assess the dependent variable?
Again, this is dependent on the example. Students may advocate
for an independent third party, who can be blinded to the
treatment conditions.
Will the instructor or the assessor know to which therapy
condition different subjects are assigned or can they be blind to
the treatment condition? If they can be blind, how will you
achieve this? Answer for each one:
The person giving instructions (instructor)
The assessor measuring variables
Students can blind the instructors/assessors by ensuring that the
instructions/assessment procedure are the same and the
instructor is never informed as to which treatment subjects
receive, even inadvertently by the subjects themselves. For this
reason, the instructions should be clearly-structured with no
idle conversation.
What characteristics of the instructor or the assessor could
affect the results? If so, how could you control for these
potential experimenter effects?
Very similar to the previous section. Examples of
characteristics could include: demographics, appearance,
competence, personality (e.g. empathy, interpersonal style,
etc.).
These variables should be controlled with randomization,
balanced conditions, and if not they should evaluate the data
after collection of the data for some of these effects.
Give an example of how the instructor/assessor’s expectations
or bias could affect or confound the results?
Example may be loss of experimental control if they introduce
any other variables, communicate their expectations, or even

5. inadvertently bias the subjects (see examples from Ppt).
How could you control for these possibilities?
Examples of strategies could include: standardized (controlled)
instruction & assessment procedure, so no confounds introduced
from experimenter bias and study is well-controlled. Also
randomization of instructor & assessor assignment (if using
multiple) can help.
SUBJECTS/PARTICIPANTS
How will you determine who is allocated to which group?
Randomization: Participants should be randomly assigned so
that each participant has an equal chance of being in either
treatment group. This is done to reduce the likelihood of
confounding variables between groups.
What would be an example of an uncontrolled group difference
that could be a confounding variable? Why is that important?
Give any number of examples (gender, age, diagnosis, etc.).
Mention how uncontrolled group differences are a threat to
internal validity.
How can you control for group differences?
Randomization.
What expectations might subjects themselves have about the
experiment’s results and how could these expectations influence
the results? How could you control for this?
Address subject bias within the context of the example.Subject
bias: Participants’ behavior is influenced by what they think the
experimenter’s expectations are
Give an example of the “good subject” and the “negativistic (or
bad) subject.”
Give an example of each that works with the experiment
example.
“Good participants” behave according to what they think the
experimenter’s expectations are, to help prove the hypothesis.

6. “Negativistic participants” behave to disprove what they think
the experimenter’s hypothesis is.
If subjects are concerned about how they are being evaluated,
how could this evaluation apprehension affect or confound the
results? How could you control for this?
Define evaluation apprehension and come up with an example of
how this can effect results. Examples could be enhanced
performance, increased anxiety, falsifying data. A potential
control could be reassuring participants that there is “no correct
answer,” and no judgment of desirability.
What other subject motives and goals could affect the results?
How could you control for these?
Students could discuss potential effect of incentives, also
possible experimenter effects on subjects. Subjects’ bias might
be specific to certain experimenters.
ETHICS
How would you protect the subjects’ privacy?
Anonymity should be ensured when possible, and there needs to
be established confidentiality with therapist.
What possible psychological harm could occur to subjects as a
result of this research? How could this risk be minimized?
This will depend on your example. A few general risks could be
creating anxiety, if subjects don’t improve then therapy could
be “wasting time”, treatment could make things worse.
How will you achieve informed consent?
Provide a clear general description about how they will inform
participants of the study, the ability to stop anytime, what the
benefits and risks are, and that they will sign statement that
they understand.
Would you deceive the subjects in any way?
We should not need to deceive subjects, but if we do, we need
to provide a good rationale as to why. Possible reasons could be
to limit inaccuracies of reporting and to possibly omit specifics
to keep participants blinded to conditions (relevant below).
What information about the study would you provide to the

7. subjects before they participate?
As mentioned above, this should include the procedure, general
purpose (treatment outcome), risk and protection from harm,
ability to stop, instructions of general expectations. This should
be the same for all subjects.
What information would you withhold from subjects as you
describe the study before they participate? What is the purpose
of withholding this information?
Be specific as to what information you omit and why. This
could mean keeping participants blinded to the treatment
specifics to minimize subject bias.
What would you address in your debriefing of subjects after the
study?
Participants should be told about the assigned conditions, the
specific purpose of the study (to test therapy differences), and
possibly the results of study at a later time.
How would you address any negative effects that might occur?
Be specific.
Students should come up with a strategy of how they would
attempt to correct any potential problems. One strategy could be
providing a therapy that is known to be effective.
What other ethical concerns can you identify for your study?
How will you address these?
Possible concerns could be using incentives, withholding
effective treatment, using deception, breach of confidentiality,
and possible emotional distress depending on the treatment. For
each ethical concern, students should attempt to find a solution.
Control and Validity Vocabulary Terms
Internal validity - the measurements accurately indicate the
variables they are designed to measure AND the results
accurately indicate the conclusion that the independent variable
(not something else) affects the dependent variable in this
particular way.
In other words, the measurements, results, and conclusions that

8. are part of (internal to) the experiment itself are all valid.
Generalizability or External Validity- the results of the study
can be applied to circumstances outside the actual study and to
people other than the actual subjects of the study. In other
words, the results are valid for circumstances and people
external to the specific study itself. So, we can generalize our
results outside the specific study.
Population- All of the people in a group
Sample- A portion of the population who are directly involved
in the study
Representativeness –the extent to which a given sample
accurately reflects or represents the particular population as a
whole.
Sampling bias – The sample is not reflective or representative
of the target population; perhaps demographic characteristics
are systematically different from the larger population, or a
particular variable has influenced the sample that is selected to
participate.
Allocation- Deciding which participants are in which research
condition or group
Randomization- Assigning participants randomly to the
different research conditions
Experimental control- Minimizing confounding variables in a
study by standardizing or holding them constant
Confounds: Anything that varies with the independent and/or
dependent variables in a study, and could thus interfere with
internal validity.
Demand Characteristics- Subtle clues that may make a
participant believe that a specific outcome or behavior is
expected
Subject bias- Participants’ behavior is influenced by what they
think the experimenter’s expectations are
Experimenter bias- The experimenter gives clues to participants
regarding their expectations.
Experimenter intentionally or unintentionally alters results
through:

9. · Choosing participants/samples that will lead to the
expected/desired outcome
· Choosing a procedure that influences outcomes: e.g., forcing
students to participate in a study for class credit
· Choosing an inappropriate measurement tool
· Providing demand characteristics
Experimenter Effects- Participants’ behavior is influenced by
characteristics or individualized behavior of the experimenter or
other people who conduct the study (instead of the independent
variable)
Incentives for Participation- Rewards for participating in the
study, perhaps in a certain way
Subject Bias-
“Good participants” behave according to what they think the
experimenter’s expectations are, to help prove the hypothesis.
“Negativistic participants” behave to disprove what they think
the experimenter’s hypothesis is.
Placebos – a fake treatment that causes an effect simply because
participant thinks it will
Blinding - making participants unaware of their condition
Evaluation Apprehension- Participant’s behavior is a response
to the presence of an evaluator
· Performance increases or is consistent with experimenter
expectations in order to get a favorable evaluation
· Performance decreases or is inconsistent due to worry about
evaluation
Ethics- Ensuring that participants are treated safely, fairly, and
are exposed to no unnecessary risks.
Risk- The probability of harm or injury, physical or emotional,
as a result of participation in a study
Minimal risk- Probability and magnitude of harm anticipated is
not greater than those ordinarily encountered in daily life or
routine examination
Informed Consent- Participants are informed of study
procedures and any and all potential risks prior to agreeing to
participate in a study.

10. · This includes being informed that they may discontinue their
participation in the study at any time
Deception- Participants are misled or wrongly informed about
the study’s aims or procedure
Debriefing- Participant is given a general idea of what is being
investigated and why
Privacy - individual’s control over the extent, timing, and
circumstances of sharing him/herself with others)
Confidentiality- Refers to data; an agreement made regarding
the disclosure of information
Anonymity- No identifiers (e.g., name, address, telephone
number) are collected that link information/data to the
individual from whom they were obtained.
Institutional Review Board- A committee that reviews and
approves research involving human subjects and ensures that all
human subject research is conducted in accordance with federal,
institutional, and ethical guidelines
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalC
ode=sbeh20
Download by: [Eastern Michigan University] Date: 25 June
2016, At: 17:25
Cognitive Behaviour Therapy
ISSN: 1650-6073 (Print) 1651-2316 (Online) Journal homepage:
http://www.tandfonline.com/loi/sbeh20
Stress Management: A Randomized Study of

11. Cognitive Behavioural Therapy and Yoga
Jens Granath , Sara Ingvarsson , Ulrica von Thiele & Ulf
Lundberg
To cite this article: Jens Granath , Sara Ingvarsson , Ulrica von
Thiele & Ulf Lundberg (2006)
Stress Management: A Randomized Study of Cognitive
Behavioural Therapy and Yoga,
Cognitive Behaviour Therapy, 35:1, 3-10, DOI:
10.1080/16506070500401292
To link to this article:
http://dx.doi.org/10.1080/16506070500401292
Published online: 03 Feb 2007.
Submit your article to this journal
Article views: 3268
View related articles
Citing articles: 79 View citing articles
http://www.tandfonline.com/action/journalInformation?journalC
ode=sbeh20
http://www.tandfonline.com/loi/sbeh20
http://www.tandfonline.com/action/showCitFormats?doi=10.108
0/16506070500401292
http://dx.doi.org/10.1080/16506070500401292
http://www.tandfonline.com/action/authorSubmission?journalCo
de=sbeh20&page=instructions
http://www.tandfonline.com/action/authorSubmission?journalCo
de=sbeh20&page=instructions
http://www.tandfonline.com/doi/mlt/10.1080/165060705004012

12. 92
http://www.tandfonline.com/doi/mlt/10.1080/165060705004012
92
http://www.tandfonline.com/doi/citedby/10.1080/165060705004
01292#tabModule
http://www.tandfonline.com/doi/citedby/10.1080/165060705004
01292#tabModule
Stress Management: A Randomized Study of Cognitive
Behavioural Therapy and Yoga
Jens Granath, Sara Ingvarsson, Ulrica von Thiele and Ulf
Lundberg
Department of Psychology and Centre for Health Equity Studies
(CHESS), Stockholm
University, Stockholm, Sweden
Abstract. In this study, a stress management program based on
cognitive behavioural therapy
principles was compared with a Kundaliniyoga program. A
study sample of 26 women and 7 men
from a large Swedish company were divided randomly into 2
groups for each of the different forms
of intervention; a total of 4 groups. The groups were instructed
by trained group leaders and 10
sessions were held with each of groups, over a period of 4
months. Psychological (self-rated stress
and stress behaviour, anger, exhaustion, quality of life) and
physiological (blood pressure, heart rate,
urinary catecholamines, salivary cortisol) measurements
obtained before and after treatment showed
significant improvements on most of the variables in both
groups as well as medium-to-high effect
sizes. However, no significant difference was found between the

13. 2 programs. The results indicate that
both cognitive behaviour therapy and yoga are promising stress
management techniques. Key
words: stress management; intervention; cognitive behavioural
therapy; yoga; catecholamines; cortisol
Received January 30, 2003; Accepted October 10, 2005
Correspondence address: Ulf Lundberg, Department of
Psychology and CHESS, Stockholm
University, SE-106 91 Stockholm, Sweden. Tel: +468163874.
Fax: +468167847.
E-mail: [email protected]
Stress-related health problems, such as
chronic fatigue, muscular pain and burnout,
have increased dramatically in modern socie-
ties in recent years (European Commission,
2000). A variety of stress-management tech-
niques is used to address this problem. In a
meta-analysis of occupational stress-reducing
interventions, van der Klink and colleagues
(2001) distinguished 4 intervention types:
cognitive-behavioural interventions, relaxa-
tion techniques, multimodal programs, and
organization-focused interventions. They con-
cluded that all of the intervention types were
effective, but cognitive-behavioural interven-
tions were more effective than the other types.
In a review of work-site stress management
interventions, including muscle relaxation,
meditation, biofeedback, cognitive-behavioural
skills and combinations of these techniques,
Murphy (1996) concluded that the effective-
ness of the interventions varied accord-
ing to the health-outcome measure used.

14. Cognitive-behavioural skills were more effec-
tive for psychological outcomes, whereas
muscle relaxation techniques were more
effective for physiological outcomes. Using
a combination of techniques (e.g. muscle
relaxation plus cognitive-behavioural skills)
seemed to be more effective across outcome
measures than using a single technique. A
similar recommendation is given by Jones
and Johnston (2000), who concludes that a
stress management intervention should
include psycho-education and analyses of
individual reaction in combination with
management techniques, such as relaxation,
assertiveness training, time management and
cognitive restructuring.
Yoga is an ancient Indian practice focusing
on breathing and physical exercises, thereby
combining muscle relaxation, meditation and
physical workout. There is a plethora of yoga
schools, of which one of the most common is
Kundaliniyoga. Kundaliniyoga is characterized
# 2006 Taylor & Francis ISSN 1650-6073
DOI 10.1080/16506070500401292
Cognitive Behaviour Therapy Vol 35, No 1, pp. 3–10, 2006
D
ow
nl
oa
de

15. d
by
[
E
as
te
rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a
t
17
:2
5
25

16. J
un
e
20
16
by exercises (kriyas) that stimulate the blood
flow and energy supply to the brain, the
nervous system and the glands in the end-
ocrine nervous system (Singh-Khalsa, 1998).
Although there is a lack of controlled studies,
yoga is regarded as a promising method for the
treatment of stress-related problems (Fersling,
1997). Several studies have shown yoga to be
promising for both physiological (Murugesan,
Govindarajulu, & Bera, 2000) and psychologi-
cal outcome measures (Malathi, Damodaran,
Shah, Patil, & Maratha, 2000).
In many randomized controlled studies a
comparison is made between an active treat-
ment and a wait list control group, which
enables control for history, spontaneous
recovery and other confounders (Sackett,
Haynes, Tugwell, & Guyatt, 1991). Although
this a common approach with many advan-
tages, it has also been criticized. From
medicine, a common criticism is that it is
unethical to withhold a treatment from a
patient who needs it (Feldman, Wang, Willan,

17. & Szalai, 2003). Others emphasize that the
possibility of being in a control group might
be regarded as unacceptable to potential
participants, hence increasing the risk of
non-compliance and drop-out (Schafer,
1982). This is particularly true for interven-
tions in the workplace, where organizations
rarely agree to participate in true experiments
and randomized controlled trails with non-
treatment or wait list control groups.
In the present randomized study, 2 active
methods for stress management, cognitive
behaviour therapy and Kundaliniyoga, are
compared. It is hypothesized that both methods
have positive effects on perceived stress, stress
behaviour, vital exhaustion, anger, quality of
life, blood pressure, heart rate, catecholamine
and cortisol levels, but that cognitive behaviour
therapy and yoga have different impacts on
various outcome measures.
Method
Design
Following recruitment and informed consent
and before assessment, the participants were
assigned randomly to 1 of 2 conditions: yoga
and cognitive behaviour therapy, in either an
all-female group or a mixed group (making 4
groups in total). Data were collected pre- and
post-treatment. The study was approved by
the local ethics committee.
Participants

18. Participants were recruited from a large
Swedish company in the financial sector. All
personnel received information about the
design of the project and the 2 methods, and
were invited to participate through the com-
pany’s personnel department. Participation in
the study was free of charge, but the company
provided financial means for analyses of the
physiological data. It was emphasized that
there was no opportunity to choose between
the 2 methods, and the participants were
requested not to share information about the
specific content of the methods with each
other during treatment. Twenty-seven women
and 10 men with self-reported stress-related
problems agreed to participate in the study.
Four individuals dropped out. In all, 33
participants completed the treatments: 17 on
the cognitive behaviour therapy program, and
16 on the yoga program. Two participants
missed the physiological measurements after
treatment. Accordingly, physiological data are
available for only 31 of the 33 individuals,
despite all participants having completed the
questionnaire. The main characteristics of the
participants are presented in Table 1.
Table 1. Characteristics of study participants.
Cognitive
behaviour
therapy Yoga
n519 n518
Gender

19. Male 5 (26%) 5 (28%)
Female 14 (74%) 13 (72%)
Level of education
Swedish high school 6 (44%) 7 (35%)
University 11 (56%) 9 (65%)
Marital status
Married/cohabiting 11 (65%) 14 (88%)
Single 3 (18%) 1 (6%)
Divorced/widowed 3 (18%) 1 (6%)
Number of children
0 9 8
1 2 5
>2 6 3
4 Granath, Ingvarsson, von Thiele and Lundberg COGNITIVE
BEHAVIOUR THERAPY
D
ow
nl
oa
de
d
by
[
E
as
te

20. rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a
t
17
:2
5
25
J
un
e
20
16

21. Procedure
The 10 sessions in each program ranged over a
4-month period during the winter and spring of
2000. There was a difference in preferences
with regard to the scheduling of the sessions.
Yoga was held weekly (with 1 week’s interrup-
tion between sessions 6 and 7 due to the school
vacation), whereas cognitive behaviour ther-
apy was initially held weekly (first 4 weeks),
followed by 3 sessions held once every other
week, and finally by 3 sessions once every 3
weeks. In order to end the 2 training programs
at approximately the same time, the yoga
groups started a few weeks later than the
cognitive behaviour therapy groups. The all-
female cognitive behaviour therapy group had
2 female instructors (SI, UvT), whereas the
mixed group had just 1 male instructor (JG). A
trained yoga instructor led both yoga groups.
The cognitive behaviour therapy sessions were
performed at the company premises and the
yoga sessions at locations nearby.
Measurements for all participants were
taken 2 weeks prior to the first cognitive
behaviour therapy group session and 2 weeks
after the final group sessions. Measurements
took place in a room at the company’s
premises. Upon arrival for the measurements,
the participants returned a questionnaire that
they had completed at home. Following a
short explanation of procedures, they were
instructed to sit down and relax for 5 minutes

22. before measurements of blood pressure and
heart rate were taken. The measurements were
repeated after 5 and 10 minutes. Finally, urine
samples for catecholamines and saliva sam-
ples for cortisol determinations were obtained.
During the measurements, relaxing classical
music (The Four Seasons by Vivaldi and Cello
Suites by Bach) was played on a compact disk
player. The conditions were the same for pre-
and post-measurements.
Following post-measurements, all partici-
pants were offered individual feedback on the
outcomes of their tests.
Intervention programs
The cognitive behaviour therapy program
consists of a modified version of a treatment
used for coronary heart disease (Burell, 1996).
Each session in the cognitive behaviour
therapy program was divided into 5 sections:
relaxation, discussion on home assignments,
psycho-education, management techniques,
and introduction of new home assignments.
For relaxation, the principles of ‘‘applied
relaxation’’ (Öst, 1987) were used. The home-
work assignments followed the same basic
structure, including 4 parts: registration tasks
(e.g. stress behaviour, anger and irritation,
problems, etc.), daily drills (training new
management techniques, such as listening
without interrupting, eating slowly, etc.), case
studies, and relaxation training. The psycho-
education section consisted in the presenta-
tion of stress-related topics, such as the

23. psychophysiology of stress, theoretical stress
models, time urgency, irritation and anger.
The management methods employed con-
cerned problem-solving, assertiveness train-
ing, goal setting, time management, cognitive
and behavioural restructuring and relapse
prevention. To ensure a high degree of control
over the content of the sessions, they followed
a pre-set manual (including a written manu-
script). Each session was followed by a meet-
ing in which group leaders evaluated the
content of the session.
The main focus of the yoga program was
on physical exercise. Sessions 1–3 had their
origin in a yoga program designed for back
treatment. In sessions 4–6 a program invol-
ving basic movements, normally used as
an introduction to Kundaliniyoga, was
implemented. Sessions 7–9 were aimed at
balancing body, energy and mind. The final
session involved exercises that can be used
daily to alleviate tension in the shoulders,
neck and head, or when some extra energy is
needed. At each session, participants were
given a compendium that contained a theore-
tical account of a specific theme, and advice,
suggestions and yoga exercises relevant to
that theme. Each session included a 15-
minute discussion of different topics, such
as life behaviours, restoration, reflection, self-
respect, physical exercise and food/eating
habits. The yoga participants also received
home assignments, which involved physical
exercises and reading through the compen-
dium in question. Various themes were

24. presented, including personal goals, breath-
ing, body postures, meditation and mantra
knowledge and intuition.
To ensure that the 2 treatments differed,
both were manual-based. Group leaders were
informed of the necessity of closely following
the manuals.
VOL 35, NO 1, 2006 Stress management through CBT and yoga
5
D
ow
nl
oa
de
d
by
[
E
as
te
rn
M
ic
hi
ga

25. n
U
ni
ve
rs
it
y]
a
t
17
:2
5
25
J
un
e
20
16
Questionnaires
Five scales covering different aspects of stress
were used for the outcome measures. The first
dependent measure, for general stress level,
consisted in a Swedish translation of the

26. ‘‘Perceived Stress Scale’’ (PSS) (Cohen,
Kamarck, & Mermelstein, 1983; Eskin &
Parr, 1996). The PSS is a 14-item scale
designed to measure the degree to which life
situations are appraised as stressful. The
internal consistency reliability for the
Swedish version is 0.82 (Cronbach’s alpha).
The second dependent measure, concerning
the extent of Type-A behaviour, was the 20-
item scale ‘‘Daily Stressors’’ (Burell, 1996).
The third dependent measure was a Swedish
version of the ‘‘Maastricht Questionnaire’’
(Appels, Höppener, & Mulder, 1987), which
measures exhaustion. The scale, consisting of
19 items, was designed to measure the degree
of vital exhaustion. As a fourth dependent
measure, a Swedish version of the MMPI-2
Anger Sub-scale was used (Graham, 1990).
Since 1 question was misinterpreted by a large
number of participants, the original 15-item
scale was transformed into a 14-item scale,
with 1 item removed for statistical analysis.
The fifth dependent measure was quality of
life, which was measured using the Quality of
Life Inventory (QOLI) (Frisch, Cornell,
Villanueva, & Retzlaff, 1992). The QOLI
consists of questions of importance and
satisfaction regarding 16 areas in life (in total
32 items). Unfortunately, no data on relia-
bility are available for the present versions of
‘‘Daily stressors’’, ‘‘exhaustion’’, ‘‘MMPI-2
Anger sub-scale’’ and QOLI. In order to
make the scales more homogeneous, all items
except those on the ‘‘Quality of Life
Inventory’’ were presented with 5 response
categories, regardless of the original number.

27. Physiological measures
Urinary catecholamine. Subjects were instructed
not to smoke, use snuff or drink coffee or
alcohol, and to avoid any activities that might
expose them to stress during the 3 hours
preceding the time of measurement.
They were instructed to empty their blad-
ders, and note the exact time of voiding, about
2 hours before the time of measurement.
Urine was voided again at the time of
measurement, and the exact period in time
from previous voiding was calculated.
Following measurement of sample volume,
the pH of the sample was adjusted to 3.0 with
6 M HCl. A 20 ml volume was frozen
(218 ˚C) prior to analysis for adrenaline and
noradrenaline by high-pressure liquid chro-
matography (HPLC) (Riggin & Kissinger,
1977). After determining the concentration
of catecholamines in the sample, values were
multiplied by volume and divided by time
(pmol/min).
Cortisol in saliva. Cortisol in saliva was
measured through use of a standard centri-
fugation tube (Salivette; Sarstedt Inc.,
Rommelsdorf, Germany), which contains a
small cotton roll that is ‘‘chewed’’ upon for a
couple of minutes to obtain a sufficient
amount of saliva. Tubes with saliva samples
were frozen (218 ˚C) until centrifuged and
analysed for cortisol by radioimmunoassay
(RIA).
Blood pressure and heart rate. Systolic and

28. diastolic blood pressure and heart rate were
measured by an automatic digital blood
pressure device (DS-140, A&D Company,
Japan). The median of 3 measurements was
used for the statistical analyses.
Statistical analysis
SPSS, version 9.0, was used for the statistical
analysis. Mean scores on the outcome mea-
sures at pre- and post-treatment for both
groups were analysed by means of a 2-way
repeated measures ANOVA. Group differ-
ences at pre- and post-treatment were tested
using an independent-samples t-test. Two-
tailed tests were used. Effect sizes were
calculated according to Cohen’s (1988) d
statistic. For each scale the magnitude of
change from pre- to post-treatment was
defined as (Mpre2Mpost)/SDpooled, where,
SDpooled5![(SDpre2+SDpost2)/2]. Positive
effect sizes represent improvements in stress
and other symptoms (e.g. reductions in
problems)) with the exception of QOLI, where
negative effect sizes represent improvements.
Results
Questionnaire data
The changes in scores on the various variables
following the intervention programs are sum-
marized in Table 2.
6 Granath, Ingvarsson, von Thiele and Lundberg COGNITIVE
BEHAVIOUR THERAPY
D

29. ow
nl
oa
de
d
by
[
E
as
te
rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a
t

30. 17
:2
5
25
J
un
e
20
16
Perceived stress (PSS) decreased signifi-
cantly among participants on both programs
(cognitive behaviour therapy: t(16)56.14,
pv0.001; yoga: t(15)52.89, pv0.01). Rat-
ings on the stress behaviour scale (cognitive
behaviour therapy: t(16)54.80, pv0.001;
yoga: t(15)52.66, pv0.05) and of exhaustion
(cognitive behaviour therapy: t(16)53.69,
pv0.01; yoga: t(15)52.91, pv0.01) also
decreased significantly for both programs.
Ratings of anger decreased significantly
for cognitive behaviour therapy but not
for yoga (cognitive behaviour therapy:
t(16)53.61, p50.0002; yoga: t(15)51.97,
p50.07). The increase in QOLI was non-
significant in both cognitive behaviour ther-
apy and yoga (cognitive behaviour therapy:
t(16)521.80, p50.09; yoga t(15)521.25,

31. p50.23).
Effect sizes varied from d520.44 (QOLI)
to d51.42 (PSS) for cognitive behaviour
therapy and from d520.19 (QOLI) to
d50.87 (Exhaustion) for yoga.
Physiological measures
The results from the measures of catechola-
mines in urine (adrenaline and noradrenaline),
salivary cortisol, systolic blood pressure,
diastolic blood pressure and heart rate are
presented in Table 3.
In the yoga group, the noradrenaline levels
decreased significantly between pre- and post-
measurements, t(14)53.15, p50.007, but not
in the cognitive behaviour therapy group
(t(15)51.12, p50.20). The decrease in adrena-
line was not significant for any of the 2
treatments but approach significance in the
cognitive behaviour therapy group (cognitive
behaviour therapy: t(15)52.07, p50.06; yoga
t(14)50.44, p50.66). The difference for nor-
adrenaline between the groups approached
significance (F(1,29)53.18, p50.085), but
this was not the case for adrenaline
(F(1,29)50.48, p50.495). For cortisol there
was no significant change in either group
(cognitive behaviour therapy: t(15)51.27,
p50.22; yoga: t(14)51.59, p50.13).
Table 2. Means and standard deviations of questionnaire data
for the cognitive behaviour therapy and yoga
treatment at pre- and post-treatment and effect sizes (Cohen’s d)
from pre- to post-treatment.

32. Cognitive behaviour therapy Yoga
Pre n517 Post n517 Pre n516 Post n516
M SD M SD d M SD M SD d
PSS 2.19 0.40 1.54 0.51 1.42 2.06 0.48 1.67 0.47 0.82
Stress behaviour 2.25 0.49 1.72 0.41 1.17 2.26 0.52 1.92 0.45
0.70
Exhaustion 1.85 0.62 1.28 0.68 0.88 1.82 0.74 1.25 0.56 0.87
Anger 1.15 0.46 0.81 0.45 0.75 1.23 0.51 0.96 0.44 0.57
QOLI 1.95 0.92 2.46 1.34 20.44 2.21 1.32 2.46 1.37 20.19
PSS5Perceived Stress Scale; QOLI5Quality of Life Inventory.
Table 3. Means and standard deviations of physiological data
for the cognitive behaviour therapy and yoga
treatment at pre- and post-treatment and effect sizes (Cohen’s d)
from pre- to post-treatment.
Cognitive behaviour therapy Yoga
Pre n516 Post n516 Pre n515 Post n515
M SD M SD d M SD M SD d
HR 64.4 11.89 60.9 8.30 0.34 66.2 9.57 61.4 7.47 0.56
SBP 117.6 13.71 112.9 8.78 0.41 115.5 11.88 113.7 12.23 0.15
DBP 79.4 9.22 76.6 7.74 0.33 76.6 8.28 78.4 9.94 20.20
Adrenaline 40.9 22.0 31.2 13.7 0.53 48.2 18.6 44.7 25.0 0.16
Noradrenaline 238.3 81.57 216.8 66.3 0.29 273.6 87.4 197.4
46.6 1.09
Cortisol 8.77 5.41 10.76 7.23 0.31 8.73 3.89 12.8 9.48 0.56
HR5heart rate, SBP5systolic blood pressure; DBP5diastolic

33. blood pressure.
VOL 35, NO 1, 2006 Stress management through CBT and yoga
7
D
ow
nl
oa
de
d
by
[
E
as
te
rn
M
ic
hi
ga
n
U
ni
ve
rs

34. it
y]
a
t
17
:2
5
25
J
un
e
20
16
Heart rate was not significantly lower after
either of the 2 programs, but approaching
significance in the yoga group (yoga
t(14)51.99, p50.07; cognitive behaviour ther-
apy t(15)51.51, p50.15). The reverse was
true for the decrease in systolic blood
pressure (SBP) (cognitive behaviour therapy
t(15)51.99, p50.07; yoga t(14)50.69,
p50.50). Mean diastolic blood pressure
(DBP) after cognitive behaviour therapy had
decreased from 79.4 to 76.6 mmHg, but this
change was not significant (t(15)51.48,

35. p50.16). Mean diastolic blood pressure after
yoga had increased from 76.6 mmHg to 78.4,
but again the change was not significant
(t(14)520.88, p50.39). The difference
between the 2 programs did not reach
significance (F(1,29)52.72, p50.11).
Effect sizes varied from d50.29 (noradrena-
line) to d50.53 (adrenaline) for cognitive
behaviour therapy and from d50.15 (SBP)
to d51.09 (noradrenaline) for yoga. Cortisol
had a negative effect size, implicating an
increased value, of d520.31 for cognitive
behaviour therapy and d520.56 for yoga.
Discussion
In this study, both cognitive behaviour
therapy and yoga programs resulted in a
statistically significant reduction in scores on
almost all stress-related subjective and phy-
siological variables. Overall, the effect size for
the questionnaire data was generally medium
to large (Cohen, 1992), whereas the effect size
for the physiological measures was lower,
ranging from small to medium (with the
exception of yoga and noradrenaline, where
the effect size was large). Thus, the hypothesis
was confirmed. No statistically significant
difference between the 2 intervention pro-
grams was found for any of the variables, but
they differed in effect size: cognitive behaviour
therapy had larger effect size on all ques-
tionnaire data whereas there was a more
complex picture regarding the physiological
data. The failure of showing a statistical

36. significant difference between the 2 treatments
could be due to the differences in effect size,
which can be caused by the small group sizes,
and thereby low statistical power.
Unlike the other variables, cortisol levels
increased, although non-significantly. Cortisol
levels are known to vary considerably between
individuals and over time (Cummins &
Gevirtz, 1993), and due to seasonal fluctua-
tions (Maes et al. 1997). It has also been shown
that individuals exposed to chronic stress
(Yehuda, Teicher, Trestman, Levengood, &
Siever, 1996) or report a stressful work
situation (Kurina, Schneider, & Waite, 2004)
may have markedly reduced cortisol levels,
thereby suggesting that an increase in cortisol
levels could be considered an improvement.
The contradictory findings make it difficult to
interpret the results. Further studies are
needed.
Given the difference in approaches to stress
management in the 2 interventions – one
mainly physical, the other mainly mental and
behavioural – and drawing on the conclusions
of Murphy (1996), the 2 treatments might
have different effects on different outcome
measures. Although our results do not pro-
vide enough evidence for any firm conclusions
to be drawn, the larger effect of cognitive
behaviour therapy on adrenaline and that of
yoga on noradrenaline along, suggests that
the specific contribution of cognitive beha-
viour therapy is mental relaxation, whereas

37. the specific contribution of yoga is physical
relaxation (Lundberg, 2000).
Even though arguments can be presented
for the benefits of comparing 2 active treat-
ments – most importantly the added value of
comparing a new treatment with an existing
treatment rather than with no treatment at all
– the choice of design does have disadvan-
tages. The most important lies in the lack of
untreated controls. The outcome improve-
ments found might be due to changes caused
by factors other than the intervention pro-
grams, such as seasonal changes, general
changes in workload and stress, or there
may simply be a regression towards the mean
(since the participants who volunteered for the
study had high levels of perceived stress).
However, a general decrease in workload and
stress does not seem likely in the present case.
During the period of the intervention pro-
grams, time pressure and workload in Sweden
increased considerably, as did work absentee-
ism due to stress-related disorders (Wennberg,
2001). In addition, during the study a
reorganization was performed at the com-
pany, details of which were announced only
a few weeks before the post-treatment mea-
surements were taken. This reorganization
8 Granath, Ingvarsson, von Thiele and Lundberg COGNITIVE
BEHAVIOUR THERAPY
D
ow

38. nl
oa
de
d
by
[
E
as
te
rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a
t
17

39. :2
5
25
J
un
e
20
16
involved changes in work tasks, colleagues,
leaders and work environment for the study
participants, factors that are known to con-
tribute to stress (European Agency for Safety
and Health at Work, 2000).
Another disadvantage of considering 2
active treatments in general, and the treat-
ments in this study in particular, lies in the
risk of inter-treatment contamination.
Although efforts were made to avoid con-
tamination, one explanation for the lack of
difference in outcome between the treatments
is that they did not differ sufficiently in
content. In line with the Jones and Johnston
recommendation (2000) for most effective
treatment, the cognitive behaviour therapy
groups included a reference to physical inter-
vention, e.g. ‘‘applied relaxation’’, whereas
the yoga groups, although having a main

40. focus on physical exercise, also talked about
stress-related issues.
One risk of assigning participants at ran-
dom to 2 different treatment groups is always
that some might be disappointed with their
assigned group. Although no participant
refused to participate in their group, and all
participants completed their treatment, a few
stated on their post-treatment evaluation form
that they would have preferred the other
treatment. These objections were divided
equally between the 2 groups That is, there
were both yoga participants who would rather
been trained in cognitive behaviour therapy
and cognitive behaviour therapy participants
who would have preferred training in yoga.
Despite their comments, there was still a
significant reduction in stress-related pro-
blems among them, which may provide
evidence that the treatments would be even
more effective if participants were able to
choose between them.
The largest shortcoming of the current
study is probably the lack of follow-up data
to assess long-term change. With only 1 post-
treatment assessment, no conclusions could be
drawn about the process of change.
Despite the methodological shortcomings
of the study, the consistent pattern in the
outcome variables and the medium-to-large
effect sizes indicates that both cognitive
behaviour therapy and yoga are promising
stress management techniques. Future studies

41. are needed to replicate the results, and
methodological issues need to be resolved.
Since both methods are multimodal, and in
light of the recent development in cognitive
behaviour therapy, there is a need to generate
further data on the effects of the different
models.
Acknowledgements
The authors are indebted to Maj-Lis
Bäckström and Laura Ikäheimo Lundberg at
SEB for providing facilities and encouraging
employees to participate in the study, to
Marie Öfvergård for running the yoga train-
ing program, to Dr Gunilla Burell for
instructions regarding the cognitive beha-
vioural therapy program, and to Ann-
Christine Sjöbeck for performing the hor-
mone assays. Financial support was obtained
from the Bank of Sweden Tercentenary
Foundation (UL). All authors share equal
responsibility for this manuscript. JG, SI
and UvT also served as instructors for
the cognitive behaviour therapy intervention
program.
References
Appels, A., Höppener, P., & Mulder, P. (1987). A
questionnaire to assess premonitory symptoms
of myocardial infarction. International Journal
of Cardiology, 17, 15–24.
Burell, G. (1996). Group psychotherapy project

42. new life: treatment of coronary-prone beha-
viours for patients who have had coronary
artery bypass graft surgery. I. R. Allan &
S. Schidt (Eds), Heart and Mind. The Practice
of Cardiac Psychology. Washington DC:
American Psychological Association.
Cohen, J. (1988). Statistical Power Analysis for the
Behavioural Science (2nd edn). Hillsdale, NJ:
Erlbaum.
Cohen, J. (1992). A power primer. Psychological
Bulletin, 112, 151–159.
Cohen, S., Kamarck, T., & Mermelstein, R. (1983).
A global measure of perceived stress. Journal of
Health and Social Behaviour, 24, 385–396.
Cummins, S. E., & Gevirtz, R. N. (1993). The
relationship between daily stress and urinary
cortisol in a normal population: an emphasis on
individual differences. Behavioral Medicine, 19,
129–134.
Eskin, M., & Parr, D. (1996). Introducing a Swedish
Version of an Instrument Measuring Mental
Stress. Stockholm: Stockholms Universitet.
Reports from the Department of Psychology.
European Agency for Safety and Health at Work
(2000). Research on Work-Related Stress.
Luxembourg: Office for Official Publications
of the European Communities.
VOL 35, NO 1, 2006 Stress management through CBT and yoga
9

43. D
ow
nl
oa
de
d
by
[
E
as
te
rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a

44. t
17
:2
5
25
J
un
e
20
16
European Commission (2000). Guidance on Work-
Related Stress. Spice of Life or Kiss of Death?.
Luxembourg: Office for Official Publications of
the European Communities. Employment and
Social Affairs, Health and Safety at Work.
Feldman, B., Wang, E., Willan, A., & Szalai, J. P.
(2003). The randomized placebo-phase design
for clinical trials. Physical Therapy in Sport, 4,
129–136.
Fersling, P. (1997). Naturligt övernaturligt.
[Natural Supernatural]. Köpenhamn:
Politikens Forlag.
Frisch, M. B., Cornell, J., Villanueva, M., &

45. Retzlaff, P. J. (1992). Clinical validation of
the Quality of Life Inventory: a measure of life
satisfaction for use in treatment planning and
outcome assessment. Psychological Assessment,
4, 92–101.
Graham, J. R. (1990). MMPI-2: Assessing
Personality and Psychopathology. New York:
Oxford University Press.
Jones, M. C., & Johnston, D. W. (2000). Reducing
distress in first level and student nurses: a
review of the applied stress management
literature. Journal of Advanced Nursing, 32,
66–74.
Kurina, L., Schneider, B., & Waite, B. (2004).
Stress, symptoms of depression and anxiety,
and cortisol patterns in working parents. Stress
and Health, 20, 53–64.
Lundberg, U. (2000). Catecholamines. In G. Fink
(Ed.), Encyclopedia of Stress. San Diego:
Academic Press.
Maes, M., Mommen, K., Hendrickx, D., Peeters,
D., D’Hondt, P., & Ranjan, R., et al. (1997).
Components of biological variation, including
seasonality, in blood concentrations of TSH,
TT3, FT4, PRL, cortisol and testosterone in
health volunteers. Clinical Endocrinology, 46,
587–598.
Malathi, A., Damodaran, A., Shah, N., Patil, N., &
Maratha, S. (2000). Effect of yogic practices on

46. subjective well being. Indian Journal of
Physiological Pharmacology, 44, 202–206.
Murphy, L. R. (1996). Stress management in work
settings: a critical review of the health effects.
American Journal of Health Promotion, 11,
112–35.
Murugesan, R., Govindarajulu, N., & Bera, T. K.
(2000). Effect of selected yogic practices on the
management of hypertension. Indian Journal of
Physiological Pharmacology, 44, 207–210.
Öst, L. -G. (1987). Applied relaxation: Description
of a coping technique and review of controlled
studies. Behaviour Research and Therapy, 25,
397–409.
Riggin, R. M., & Kissinger, P. T. (1977).
Determination of catecholamines in urine by
reverse-phase liquid chromatography with elec-
trochemical detection. Analytical Chemistry, 49,
2109–2111.
Sackett, D., Haynes, R., Tugwell, P., & Guyatt, G.
(1991). Clinical Epidemiology: A Basic Science
for Clinical Medicine. Philadelphia, PA:
Lippincott, Williams & Wilkins.
Schafer, A. (1982). The ethics of the randomized
clinical trial. New England Journal of Medicine.
307, 719–724.
Singh-Khalsa, D. (1998). Hjärnans långa liv [The
Long Life of the Brain.]. Stockholm: Svenska
Förlaget.

47. van der Klink, J. J., Blonk, R. W., Schene, A. H., &
van Dijk, F. J. (2001). The benefits of interven-
tions for work-related stress. American Journal
of Public Health, 91, 270–276.
Wennberg, A. (Ed.) (2001). Work Life 2000.
Quality in Work. Scientific Reports from the
Workshops. Stockholm: National Institute for
Working Life.
Yehuda, R., Teicher, M. H., Trestman, R. L.,
Levengood, R. A., & Siever, L. J. (1996).
Cortisol regulation in posttraumatic stress
disorder and major depression: a chronobiolo-
gical analysis. Biological Psychiatry, 40, 79–88.
10 Granath, Ingvarsson, von Thiele and Lundberg COGNITIVE
BEHAVIOUR THERAPY
D
ow
nl
oa
de
d
by
[
E
as
te

48. rn
M
ic
hi
ga
n
U
ni
ve
rs
it
y]
a
t
17
:2
5
25
J
un
e
20
16