Open Journal of Social Sciences
2013. Vol.1, No.6, 12-17
Published Online November 2013 in SciRes (
Open Access
Evaluation of Decisional Balance in Change of Effective Stress
Management Behavior among Chinese University Participants
Using Item Response Theory
Ke Deng1, Akira Tsuda2, Satoshi Horiuchi3, Terumi Matsuda4
1Institute of Comparative Studies of International Cultures and Socities, Kurume University, Kurume, Japan
2Department of Psychology, Kurume University, Kurume, Japan
3Health Sciences University of Hokkaido, Japan Society for the Promotion of Science, Tokyo, Japan
4Graduate School of Psychology, Kurume University, Kurume, Japan
Received September 2013
The transtheoretical model defines behavior change as progression through five stages: precontemplation
(not ready), contemplation (getting ready), preparation (readiness), action, and maintenance. Decisional
balance (i.e., the relative weight of the pros and cons of making a change) is assumed to mediate stage
progression. As one progresses through the stages, the model predicts that the balance of pros increases
while that of cons decreases. Previous studies have confirmed this; these results may be attributed to dif-
fering response patterns to each item of the decisional balance measure across the stages. This study ex-
amines the relationship between decisional balance and the stages of change related to effective stress
management behavior (i.e., any healthy activity to manage stress) using a decisional balance measure
based on item response theory. The participants were 447 male and 602 female college students. A six-
item scale of decisional balance was developed. The balance of pros was significantly higher in later
stages such as action and maintenance stages relative to earlier stages such as precontemplation and con-
templation stages, while the opposite held for the cons. These results provide strong evidence that the
correspondence between decisional balance and the stage of change can be applied to stress management
Keywords: Effective Stress Management Behavior; Decisional Balance; Stage of Change; Item Response
Stress is an increasing problem in the youth (Darling,
McWey, Howard, & Olmsted, 2007). Participants enter univer-
sity in their late teens, an age at which many psychological
disorders develop, making them particularly vulnerable (Kess-
ler, 2007; Greenberg, 2010). This is supported by research sug-
gesting that participants experience increased levels of mental
illness and stress (Darling et al., 2007; Li, Bray, & Kehle, 2005;
Eisenberg, Gollust, Golberstein, & Hefner, 2007).
In China, 42% university participants do not practice effec-
tive stress management behavior measures (defined as any form
of activity practiced for at least 20 minutes and that reduces
perceived stress, such as regular relaxation, physical activity,
talking with others, and engaging in a social activity) (Deng,
Tsuda, Horiuchi, Kim, & Wu, 2012). Thus, the importance of
using a behavioral science model has been emphasized to un-
derstand how university participants manage daily stress (Ho-
riuchi, 2013). However, there remains a lack of population-
based stress management behavior research in China based on a
behavioral science model.
Transtheoretical model (TTM) based intervention studies
have been successful in guiding populations to initiate and
maintain stress management behavior (Evers, Prochaska, John-
son, Mauriello, Padula, & Prochaska, 2006; Prochaska, But-
terworth, Redding, Burden, Perrin, Leo, Flaherty-Robb, &
Prochaska, 2008). TTM proposes that individual stress man-
agement behaviors progress through five stages: precontempla-
tion, contemplation, preparation, action, and maintenance.
Transition through the stages is driven by changes in deci-
sional balance, which involves evaluating the pros and cons of
a target behavior. In addition, TTM posits that the balance of
pros and cons varies over time and is dependent on the stage of
change (Prochaska, DiClemente, & Norcross, 1992). According
to Hall and Rossi (2008), cons are more salient than pros in the
precontemplation stage. However, the relative weight of pros
and cons reverses between the contemplation and preparation
stages; that is, pros are more highly weighed in the contempla-
tion stage tha n in the precontemplation stage. In the preparation
stage, the cons are less impor tant than t hey were in ea rlier s tag es.
Four reliable and valid measures of decisional balance for
stress management behavior have been reported (Fava, Nor-
man, Levesque, Redding, Johnson, Evers, & Reich, 1998;
Evers et al., 2006; Mauriello et al., 2007; Horiuchi, 2012). In
these studies, reliability was confirmed using classical test
theory (CTT) and validity was confirmed by theoretically pre-
dicting the relationships between the stages of change. CTT,
however, presents some limitations in analyzing test items and
interpreting data; for instance, it generates a single reliability
estimate for an entire scale and sample dependent score inter-
pretation (Embretson & Reise, 2000). Therefore, this study uses
Open Access
item response theory (IRT) to test an alternative to CTT.
IRT is popular in the fields of educational measurement and
psychometrics. Its procedures provide distinct advantages over
those of CTT for item analysis, score interpretation, and relia-
bility estimates. Within the CTT model, item parameters are
sample dependent, meaning that an individual’s score may be
higher if the items are easy or may be lower if the items are
difficult, and the difficulty of the items may vary depending on
the abilities of individuals completing them. Furthermore, when
using CTT to measure abilities or attitudes of multiple groups,
it is difficult to make comparisons across groups unless the
same instrument is used, because scoring is relative to specific
tests and groups of respondents (Embretson & Reise, 2000).
For example, the score for cons is lower in the precontempla-
tion stage than in the action stage. However, in CTT, it is diffi-
cult to determine if there is a difference between the cons in the
precontemplation and action stages and to distinguish whether
an item among the cons matches the situation and abilities of
the subject.
Therefore, the results of previous studies could be attributa-
ble to varying response patterns in each item across the stages.
To rule out this possibility, it is necessary to develop a deci-
sional balance scale that examines the relationship between
decisional balance and each stage of change and more rigo-
rously apply TTM.
In this study, effective stress management behavior is de-
fined as a form of healthful activity practiced for at least 20
minutes such as exercising, meditating, relaxing, and seeking
social support. Unhealthful activities include consuming alco-
hol and drugs, overeating, or smoking. This definition is ap-
plied from the stage-based manual for adopting stress manage-
ment behavior (Pro-Change Behavior Systems, Inc., 2003). The
minimum length of time (20 minutes) was specified to provide
participants with a helpful time frame. One could argue that a
more constrained definition should be used here. The authors
believe, nevertheless, that it is suitable to use a flexible defini-
tion when focusing on daily self-care activities from a primary
prevention focus, since there is a wide variation in the types of
activities people use to manage stress (Horiuchi, Tsuda, Kim,
Hong, Park, & Kim, 2010). Furthermore, Horiuchi et al. (2010)
reported that college participants who effectively manage stress
for more than six months are less stressed than those who do
not, supporting the validity of this definition.
The potential effect of decisional balance on stress manage-
ment behavior makes it important to promote accurate estimates
of participants’ abilities. An appropriate first step in this proc-
ess is to develop a measure of decisional balance that provides
ample evidence for a reliable and valid interpretation of partic-
ipants’ scores. The current study is unique in that IRT was used
for scaling or establishing the relationship between participants’
item responses and their current stage of decisional balance.
This study uses IRT to develop a measure of decisional balance
and to examine the relationships between decisional balance
and the stages of change in effective stress management beha-
Participants included 1049 Chinese college freshmen, of
whom 52.3% were female. The mean age was 19.40, with a
standard deviation (SD) of 1.56 years. We recruited participants
during or after lectures. Only those who agreed to participate in
this study completed questionnaires.
Stage of Change
Stage of change was assessed using the Chinese language
version of Pro-Change staging algorithm (Deng et al., 2012).
First, the definition of effective stress management behavior
was provided to the participants. Then, participants were asked
if they experienced stress; only those who answered affirma-
tively were asked to respond to the change stage algorithm. The
participants were asked whether they practiced stress man-
agement behavior everyday and were requested to select one of
the following five items representing their stage of change: 1)
“No. I have no intention to begin in the next six months.” (pre-
contemplation); 2) “No. But I intend to begin in the next six
months.” (contemplation); 3) “No. But I intend to begin in the
next month.” (preparation); 4) “Yes. I have been practicing but
for less than six months.” (action); or 5) “Yes. I have been
practicing for at least six months.” (maintenance). 27 female
and 25 male participants reported they were not stressed and
were excluded from the following analyses. Individuals’ stages
of change could vary to some extent depending on situational
factors such as stressful daily events, so it was expected that t he
temporal stability of the stage of change would be moderate. A
two-week test-retest revealed moderate reliability (κ = .40).
Construct validity was confirmed by demonstrating that corre-
lations to decisional balance were consistent with the TTM
predictions (Horiuchi, Tsuda, Kobayashi, & Deng, 2009).
Decisional Balance
Decisional balance for stress management behavior was as-
sessed using the Pro-Change decisional balance measure (Evers
et al., 2006), comprising 12 items which are divided into two
subscales: pros and cons. Each subscale includes six items. An
example of an item from the pros is “If I used healthy strategies
to effectively manage my stress, I would be healthier.” An ex-
ample of item for the cons is “if I used healthy strategies to
effectively manage my stress, it would take too much time.”
Each item was rated according to the importance of each state-
ment on a five -point Likert scale (1 = not important to 5 = ex-
tremely important). The total scores of each of the six items
were calculat ed as scores for the pros and cons, respectively.
Statistical Analyses
Evaluation of Model Assumpt ions
An exploratory factor analysis was performed using the
maximum-likelihood method with Varimax rotation. Using the
results, we confirmed that each item loaded mainly on the ex-
pected factor. It was expected that odd-numbered items would
load more highly on pros, while the even-number items would
do so on the cons.
For the comparative fitness index (CFI), values above .90
generally indicate models with an acceptable fit. A root mean
square error of approximation (RMSEA) below .08 usually
indicates a reasonable fit, with a threshold of .05 providing a
strict criterion of quality of fit.
Item Response Theory and the Graded Response Model
The graded response model (GRM; Samejima, 1969) is an
Open Access
IRT model specifically designed for K-ordered polytomous
responses. In this study, there are five (K = 5) ordered genotype
categories for each item, which are the following: Y = 1 (not
important), Y = 2 (slightly important), Y = 3 (moderately im-
portant), Y = 4 (very important), and Y = 5 (extremely impor-
tant). We use θ, a continuous latent variable, to denote the gen-
eral level of an individual for decisional balance. Mathemati-
cally, GRM is specified by the following decisional balance
response function:
i ik
i ik
exp(a (θb)
P(Yk |θ)1exp(a (θb ))
In this study, θ represented the probability of giving a partic-
ular response given a specific level of pros or cons; k equaled
the observed participant response to a Likert-type item; ai
equaled the item’s ability to discriminate between those with
high or low levels of pros or cons; and bik equaled the difficulty
in moving from a response in a lower category (k 1) to the
next higher category (k) for item j.
In this formula, ai and bik, respective ly , are the discrimination
and category-specific parameters of decisional balance i, and P
Y k|θ
) denotes the cumulative probability of genotype
categories k or higher for the ith decisional balance. Because
GRM requires bi1 < bi2 < < bik, the probability of each geno-
type category is given by the following formula:
i ii
P(Yk |θ)P(Yk|θ) P(Yk+1|θ)==≥ −≥
Parameter Estimation
Parameter estimation was completed with EasyGRM (Ku-
matani, 2003). The participants’ response patterns to the pros or
cons items were analyzed using the maximum-likelihood esti-
mation procedures to approximate the four-step difficulty pa-
rameters (−4 < b < 4) and one discrimination parameter (a > .75)
for each item (Ironson, Smith, Brannick, Gibson, & Paul, 1989),
along with an information function, which provided evidence of
reliability. Following item parameter estimation, the expected
posteriori scoring procedure estimated one latent trait value
(pros or cons) for each participant.
Reliability Evidence
In IRT models, reliability is judged by the amount of infor-
mation provided by the individual items as well as by the entire
test or scale. In polytomous IRT models, the information func-
tions from each response category for an item are combined to
arrive at the information provided by the item (Dodd, de Ayala,
& Koch, 1995). Items that are more discriminating provide
more information about the latent trait than less-discriminating
items (Embretson & Reise, 2000).
A review of test information functions showed the amount of
information that each subscale collectively provided at all le-
vels of decisional balance. The information function was also
used to calculate the standard error of measurement for each
level of decisional balance. Smaller standard errors indicated
more precise measurements (Embretson & Reise, 2000).
External Validity
External validity was evaluated through the examination of
the relationship between decisional balance and the stage of
change related to stress management. Relationships between the
stage of change and decisional balance were compared with the
patterns observed across previous stress management behavior
studies (Riley & Fava, 2003; Evers et al., 2006; Mauriel lo et al.,
2007; Horiuchi et al., 2012).
To confirm that the decisional balance score differed across
the stages of change, we examined the differences in mean
values for the decisional balance for stress management beha-
vior across the stages using a two-way mul tiple analysis of va-
riance (MANOVA). According Mauriello e t al. (2007), women
reported significantly high rating for pros. Thus, we included
gender as well as the stage of change as one of the independent
Factor Structure Analyses
As expected, the two exploratory factor analyses showed two
statistically identifiable and interpretable factors: pros and cons
(Figure 1).
Confirmatory factor analysis showed that CFI and RMSEA
fulfilled the criteria (CFI = .931 and RMSEA = .063) (Figure
1). The inter-factor correlation between the pros and cons
was .13. Thus, these results confirm the two factors uncovered
by the study.
© 2004 Pro Cha nge Behavior Systems, Inc., all rights reserved
Figure 1.
Two-factor decisional balance measurement model.
Open Access
Parameter Estimation
Parameter estimates for all subscale items are presented in
Table 1. Slope parameters greater than .75 were considered
acceptable as recommended (Ironson et al., 1989). The partici-
pants’ response patterns to the decisional balance items were
used in the marginal maximum likelihood and expectation max-
imization estimation procedures to approximate the four diffi-
culty parameters and one discrimination parameter for each
item, along with an information function, which provided relia-
bility evidence. Following item parameter estimation, the ex-
pected posteriori scoring procedure estimated one latent deci-
sional balance value for each participant.
Reliability Evidence
Test information functions for the items showed that the
subscale for the pros provided the most precise information,
with low standard error when participants estimated pros’ levels
ranging from 2.5 to +.5 (Figure 2). On the other hand, the
subscale for the cons provided the most precise measurement
between cons levels of 1 and +2.5 (Figure 3). As previously
mentioned, this represents a significant difference between the
results of IRT and CTT scaling. While IRT procedures revealed
a difference in the precision of measurement between the vari-
ous levels of decisional balance, CTT procedures did not. Cron-
bach’s alpha, the typical CTT measure of reliability, was .78
and .70 for the pros and cons subscales, respectively.
External Validity (Decisio na l Balance by Stage of
The participants’ stage of change distribution was as follows:
precontemplation = 11%, contemplation = 15%, preparation =
15%, action = 38%, and maintenance = 20%.
ANOVAs indicated that within the stages, there were signif-
icant differences between pros [F(4, 866) = 5.83, p < .05, η2
= .04] and cons [F(4, 866) = 3.07, p < .05, η2 = .01]. The pros
were significantly higher in the preparation, action, and main-
tenance stages than in the precontemplation stage, and higher in
maintenance than in contemplation. Cons were significantly
higher in the precontemplation stage than in the contemplation,
action, and maintenance stages. The pros showed a maximum
average difference of .7 SD units and the cons showed a maxi-
mum average difference of .4 SD units between the precontem-
plation and maintenance stages (Table 2 and Figure 4).
To the best of our knowledge, this is the first study to apply
decisional balance to effective stress management behavior
among Chinese university participants. To do so, this research
has generated internally and externally valid measurements of
the pros and cons of practicing effective stress management
behaviors. These measures can serve as the foundation for the
future development of interventional measures. These results
demonstrated good construct validity for the TTM scale, mea-
suring decisional balance in a large sample of Chinese univer-
sity participants. In addition, decisional balance demonstrates
validity with stages of effective stress management behavior.
The selection of 12 items by using IRT resulted in decisional
balance scale as 6 items developed. The main reason for re-
moving the six items was that the low level of discrimination
that is not reflected much in answer to an item led to a differ-
ence in decisional balance of the subject. Items that exceed 4
(b) denote that most participants would answer not important
to deflection by floor effect. And shows a ceiling effect that
would answer “extremely important” for items that are more
than 4 (b) in reverse. Standard errors of the six items that have
been adopted for this study have a very low value; however,
there was no problem with the accuracy of the estimation. The
Table 1.
Item parameter estimates for pro and con items*.
No. Pros items b1 b2 b3 b4 Slope
5 If I used healthy strategies to effectively manage my stress, I would feel better about myself 2.95 2 .34 1.11 .28 1.3
7 If I used healthy strategies to effectively manage my stre s s, I would be more in control of my life 3.43 2.32 1.05 .18 1.46
9 If I used healthy strategies to effectively manage my stre s s, my relations hips with others would improve 2.93 2.08 0.69 .64 1.12
No. Cons items b1 b2 b3 b4 Slope
2 If I used healthy strategies to effectively manage my stress, I would take too much time 1.33 .00 1.64 3.23 .85
4 Efforts to manage my stress with healthy strategies would be disruptive to my daily life .90 .42 1.56 2.59 1.12
12 If I used healthy strategies to effectively manage my stress, it can be expensive .81 .39 1.62 2.60 1.36
Note: *© 2004 Pro Change Behav ior Syste ms, Inc., all rights reserved.
Table 2.
Summary of ra w scores on pros and cons.
Pros Cons
Stage N Mean SD Mean SD
Precontemplation 92 45.79 11.30 52.84 10.63
Contemplation 134 49.26 9.93 50.24 8.74
Preparat ion 142 50.41 9.79 49.08 9.36
Action 332 50.91 9.61 49.30 9.83
Maintenance 171 52.88 8.13 48.85 10.37
Open Access
Figure 2.
Information and standard err or for pro subscale.
Figure 3.
Information and standard error for con subscale.
Figure 4.
Pros and cons across the stages of change for effective stress manage-
ment. PC = Precontemplation, C = Contemplation, PR = P reparation, A
= Action, and M = Maintenance.
test information curve showed information at 2.5 to .5 le vel of
pros characteristic values and 1 to 2.5 of cons characteristic
values reached a peak for most people.
The results of IRT showed that the IRT model can be fitted
to decisional balance data and item information, test informa-
tion, and standard error can be helpful in determining item cha-
racteristics. We used an IRT model to analyze the decisional
balance data because its reliability can be compromised by the
various factors that might influence an individual’s response to
any single indicator measure. If all indicators are considered
together, the effect of this kind of variation for any single
measure is reduced, improving the reliability of our assessment
of decisional balance. IRT provides a solution to measuring
decisional balance across stage of change
Through confirmatory factor analyses, we developed a six-
item Chinese language version of the decisional balance with
two factors: the pros of effective stress management behaviors
and the cons of doing so. Two-factor decisional balance meas-
ures have been found across stress management behavior (Riley
& Fava, 2003; Evers et al., 2006; Mauriello, Rossi, Fava, Red-
ding, Robbins, Prochasa, & Meier, 2007; Horiuchi et al., 2012).
This study, like prior studies, showed the same factor structure.
The relative weights of pros and cons varied between partic-
ipants in the precontemplation stage and those in the action and
maintenance stages. We also observed a characteristic cross-
over in the relative importance of pros and cons between the
precontemplation stage and the action and maintenance stages.
As noted above, similar patterns have been observed in a wide
array of other behaviors. Moreover, the current study refines its
findings using IRT to develop a more accurate evaluation of the
decisional balance scale than seen in prior studies.
The specific progression through the stages of change in the
weighting of pros and cons related to stress management beha-
vior was consistent with findings for other behaviors. Previous
investigations of change across multiple behaviors predicted a
2:1 ratio of variation in pros to that in cons, with pros being an
average of 1.0 SD units higher in preparation and post-action
stages than in the precontemplation and contemplation stages,
and cons being an average of .5 SD units lower in the prepara-
tion and contemplation stages than in the preparation and post-
action stages. In this study, pros showed a smaller difference
between stages than expected (.7 SD unites), and cons dis-
played a greater difference between stages than expected (.4 SD
These findings lend themselves to a number of possible in-
terpretations. First, the results support the theme of cross-be-
havior consistency in the construct of decisional balance.
Second, assuming that the 7:4 ratio of variation in pros to
that in cons can be replicated across other cross-sectional and
longitudinal studies, this suggests that pros may be more im-
portant for the action and maintenance stages of stress man-
agement behavior than for the other samples and countries.
These findings are particularly interesting given that previous
research shows that a 7:4 ratio is different from that found in
decisional balance studies of routine medical behaviors.
Limitations and Future Direction s
This study has two limitations. First, the findings are based
on cross-sectional comparisons of individuals in each stage of
change. Although previous studies with other behaviors have
shown that cross-sectional patterns may be longitudinally rep-
licated, follow-up studies evaluating individual changes over
time would be important. The second limitation of this study is
that the participants only consisted of college participants.
Therefore, it is unknown whether the six-item Chinese version
of the PDSM would be appropriate for different populations. In
future studies, it will be necessary to examine whether the de-
veloped measure is suitable for different participants from dif-
ferent demographics.
Darling, C. A., McWey, L. M., Howard, S. N., & Olmstead, S. B.
(2007). College student stress: The influence of interpersonal rela-
tionships on sense of coherence. Stress and Health, 23, 215-549.
Deng, K., Tsuda, A., Horiuchi, S., Kim, E., & Wu, X. (2012). Devel-
opment of the Chinese version of the algorithm of stage of change
-4-3-2-10 1 2 3 4
Standard Error
-4-3-2-10 1 2 3 4
Standard Error
Cons Pros
Open Access
for stress management behavior confirmation of the reliability and
validity. Kureume University Psychological Research, 11, 1-7.
Dodd, B. G., de Ayala, R. J., & Koch, W. R. (1995). Computerized
adaptive testing with polytomous items. Applied Psychological Me-
thods, 19, 5-22.
Embreston, S.E., & Reise, S.P. (2000). Item response theory for psy-
chologists. Mahwah, NJ: Lawrence Erlbaum.
Evers, K. E., P rochask a, J. O., Jo hnson, J. L., Mauriel lo, L. M., Padula,
J. A., & Prochaska, J. M. (2006). A randomized clinical trial of a
population and transtheoretical model-based stress-management in-
tervention. Health Psychology, 25, 521-529.
Fava, J. L., Norman, G. J., Levesque, D. A., Redding, C. A., Johnson,
S., Evers, K., & Reich, T. (1998). Measuring decisional balance for
stress management. In Nineteenth annual scientific sessions of the
society of behavioral medicine, New Orleans, 25-28 March 1998.
Greenberg, J. S. (2010). Comprehensive stress management (12th ed.).
New York: McGraw-Hill Companies.
Hall, K. L., & Rossi, J. S. (2008). Meta-analytic examination of the
strong and weak principles across 48 health behaviors, Preventive
Medicine, 46, 266-274.
Horiuchi, S., Tsuda, A., Ki m, E., Hong, K.-S., Park, Y.-S., & Ki m, U.
(2010). Relationships between stage of change for stress manage-
ment behavior and perceived stress and coping. Japanese Psycho-
logical Research, 52, 291-297.
Horiuchi, S., Tsuda, A., Kim, E., Morita, T., & Kobayashi, H. (2013).
Effectiveness of the expert system for modifying effective stress
management behavior bas ed on the transtheoretical mod el in college
students. Stress Management Research, 9, 9-24.
Horiuchi, S., Tsuda, A., Kobayashi, H., & Prochaska, J. M. (2012).
Reliability and validity of the Japanese language version of Pro-
Change’s decisional balance measure for effective stress manage-
ment (PDSM). Japanese Psychological Research, 54, 128-136.
Horiuchi, S., Tsuda, A ., Tanaka, Y., Oka mura, H., Yaji ma, J., & Tsuda,
S. (2009). Stage of change distribution in stress management beha-
vior in Japanese university students. Japanese Journal of Health
Promotion, 11, 1-8.
Ironson, G. G., Smith, P. C., Brannick, M. T. , Gibson, W. M., & Paul,
K. B. (1989). Construction of a job in general scale: A comparison of
global, composite, and specific measures. Journal of Applied Psy-
chology, 74, 193-200.
Kessler, R. C., Gollust, S. E ., Golberstein, E., & Walters, E. E. (200 5).
Lifetime prevalence and age-of-onset distributions of DSM-IV dis-
orders in the National Comorbidity Survey replication. Archives of
General Psychiatry, 62, 593-602.
Mauriello, L. M ., Rossi, J. S., Fav a, J. L ., Reddin g, C. A., Ro bbins , M.,
Prochasa, J. O., & Meier, K. S. (2007). Assessment of the pros and
cons of stress management among ado lescents: Development and va-
lidation of a decisi o nal b alan ce measure. American Jou rn al o f Hea lth
Promotion, 22, 140-143.
Pro-Change Behavior Systems, Inc. (2003). Road to healthy living: A
guide for effective stress management. Kingston, RI: Pro-Change Be-
havior Systems, Inc.
Prochaska, J. O., Butterworth, S., Redding, C. A., Burden, V., Perrin,
N., Leo, M., Flahertty-Robb, M., & Prochasak, J. M. (2008). Initial
efficacy of MI, TTM tailoring an d HRI’s with multiple behaviors for
employee health promotion. Preventive Medicine, 46, 226-231.
Prochaska, J . O., DiCl emente, C . C., & Norc ross J. C. (1 992). In search
of how people change: Applications to addictive behaviors. Ameri-
can Psychologist, 47, 1102-1114.
Riley, T. A., & Fava, J. L. (2003). Stress and transtheoretical model
indicators of stress management behaviors in HIV-positive women.
Journal of Psychosomatic Research, 54, 245-252.
Samejima, F . (1969). Estimation of latent ability using a response pat-
tern of graded scores. Psychometrika Monograph No. 17, Richmond:
Psychometric Society.