International Journal of Obesity (2008) 32, 1137–1144
& 2008 Macmillan Publishers Limited All rights reserved 0307-0565/08 $30.00
www.nature.com/ijo
ORIGINAL ARTICLE
Validating stage of change measures for physical
activity and dietary behaviors for overweight women
AH Robinson1, GJ Norman2, JF Sallis3, KJ Calfas3, CL Rock2 and K Patrick2
1
Department of Psychiatry and Behavioral Sciences, Stanford University School of Medicine, Stanford, CA, USA;
Department of Family and Preventive Medicine, University of California, San Diego, CA, USA and 3Department of
Psychology, San Diego State University, San Diego, USA
2
Objective: To investigate the construct, concurrent and predictive validity of stage of change measures for physical activity (PA),
and intakes of fruit and vegetables (FVs), dietary fiber (FB) and dietary fat (DF) among a sample of overweight women.
Design: Subjects were 401 women (mean age ¼ 41, s.d. ¼ 8.7 years; mean body mass index ¼ 32.35, s.d. ¼ 4.6) recruited to
participate in a 12-month weight loss intervention trial. Concurrent validity tests included (1) self-report of current behavior, (2)
decisional balance (for example, pros and cons of behavior change), (3) self-efficacy, (4) the MTI Actigraph accelerometer
(for the PA staging measure), and (5) a food-frequency questionnaire (for all dietary staging measures). Predictive validity was
assessed through tests of the relationship between the baseline stage of change measures and their corresponding behavior
1-year later.
Results: Coefficient a-tests of internal consistency exceeded 0.70 on the majority of scales. Concurrent validity tests indicated
strong validity evidence for three staging measures and little validity for the DF staging measure (Z2 range, 0.02–0.18). All
staging algorithms demonstrated predictive validity (Z2 range, 0.04–0.126).
Conclusion: Staging measures can determine motivational readiness for overweight women, contribute to the standardization
of stage of change assessment and facilitate cross-study comparisons.
International Journal of Obesity (2008) 32, 1137–1144; doi:10.1038/ijo.2008.65; published online 20 May 2008
Keywords: stage of change; behavior change; Transtheoretical Model; physical activity; dietary behaviors; overweight women
Introduction
The Transtheoretical Model of health behavior change is
a multiple construct framework for understanding health
behavior and promoting behavior change.1 The model
postulates that people change behavior as they progress
through five stages: precontemplation, contemplation, preparation, action and maintenance. An individual is classified
into a stage based on their current behavior and readiness to
change that behavior.2
Stages of change have been used as a surveillance tool
for populations and population segments to determine
estimates and changes in motivational readiness.3 For
example, stage of change may be used to compare subjects’
readiness to change a specific behavior,4,5 their readiness
across multiple behaviors6 and/or their longitudinal shifts in
readiness to change.7–10
Correspondence: Dr AH Robinson, Department of Psychiatry and Behavioral
Sciences, Stanford University School of Medicine, 401 Quarry Road, Stanford,
CA 94305-5722, USA.
E-mail: [email protected]
Received 30 October 2007; revised 2 April 2008; accepted 5 April 2008;
published online 20 May 2008
The stage of change construct can facilitate tailoring
of interventions by matching intervention strategies to
individuals’ motivational readiness.11 For instance, a
stage-matched intervention for at-risk participants (either
overweight and sedentary or active smoker) was superior to a
nonstage-matched intervention in increasing the odds of
becoming more physically active, decreasing dietary fat (DF)
intake or quitting smoking.12 Stage of change for a single
behavior such as physical activity (PA) can be incorporated
across a number of studies as a common measure of
motivation and behavior,3,13 or several staging measures
for different behaviors can be used in multiple-risk behavior
and weight loss interventions.14–16
Few studies have tested the measurement validity of
PA and dietary behavior staging measures for use in
overweight adults, who are the focus of increasing
study.17 Validity evidence for stage of change measures is
often based solely upon comparisons to other self-report
measures of behavior.18–21 One study reported using
psychosocial constructs (for example, self-efficacy, decisional
balance) hypothesized to be related to the stages of change
to test concurrent validity for moderate PA in overweight
adults.22
Validating measures for overweight women
AH Robinson et al
1138
The present study investigated the validity of four staging
measures in a target population of overweight and obese
women enrolled in a weight loss intervention program. The
staging measures included four algorithms assessing readiness to change PA, and intakes of fruit and vegetables (FVs),
dietary fiber (FB) and DF. It was hypothesized that each
staging measure would be related to its corresponding
behavior (providing evidence for concurrent and predictive
validity) and to related psychosocial measures of self-efficacy
and the pros and cons of change (providing evidence for
construct validity).
Participants
Women were recruited through their primary care provider
to participate in a weight loss intervention trial. The trial
included women only because1 of the clinical importance of
investigating the health behaviors of overweight and obese
women seeking weight loss, and2 a second, companion trial
focused exclusively on men was conducted subsequently.
Eligibility criteria included having a body mass index (BMI
weight (kg)/height (m2)) between 25 and 40 kg m2, being
age 18–55, able to engage in moderate PA, able to speak and
read English and have access to the Internet. Thirty-seven
primary care providers at seven clinic sites sent letters to
women within the eligible age range informing them they
may be contacted to participate in a research study. Women
had the option to return a postcard indicating they did not
want to be contacted about the study. Over a 6-month
period, trained recruiters attempted to contact 1726 women
over telephone. Of the attempted contacts, 675 women were
not eligible to participate, 247 women were eligible but
declined to participate and eligibility was not determined for
392 women. A total of 401 women enrolled in the study and
287 women completed the 12-month measurement visit.
Loss to follow-up was not significantly related to group
or demographic characteristics. All study procedures were
approved by university and clinic institutional review
boards.
The mean age of the sample was 41 years (s.d. ¼ 8.7) and
the average BMI was 32.35 kg m2 (s.d. ¼ 4.6). Total 67% of
the sample reported being married or living with a partner,
47% were college graduates, 73% reported working full time.
Total 61% of women indicated their ethnicity was white
non-Hispanic, 20% indicated Hispanic, 7% indicated black
non-Hispanic, 5% indicated Asian or Pacific Islander and the
remaining 7% identified themselves as multiethnic or did
not state their ethnicity.
health recommendation for the behavior was met based on
behavior criteria cutoff scores, which determined the questions from the staging algorithm the participant completed
(see criteria descriptions and cutoff calculations below). If
the behavior criterion was met, then participants reported if
they have been meeting the recommendation for fewer or
more than 6 months and were classified into the action or
maintenance stages, respectively. If women were not meeting the health recommendation, then intention to change
their behavior was assessed. Based upon reported intention,
women were classified into either the precontemplation (no
intention to change within the next 6 months), contemplation (intention to change within the next 6 months) or
preparation (intention to change within the next 30 days)
stage. The 6-month criterion and general measurement
format were based on previously used staging measures and
recommendations.23,24
The criteria used to classify a participant into at least the
action stage of change for each behavior were based on
health guidelines in place at the time the study began.25
Table 1 presents the criteria used in each staging measure to
classify an individual into at least the action stage of change.
All questions were asked in reference to PA or dietary intake
over the previous 7 days. To be placed in at least the action
stage of change for PA, the subject had to respond
affirmatively to a single item inquiring whether she engaged
in moderate PA for at least 30 min on at least 5 days per week.
Fruit and vegetable stage was assessed with a single item
asking if the subject consumed at least five servings of FVs
per day, and an affirmative response placed the participant
into at least the action stage of change. Although PA and FV
intake were determined from single items, FB and DF intakes
were based on multiple items and an estimated number of
servings were computed. The FB criterion was determined
from seven questions about frequency of consuming highfiber foods (see Table 1). Five response options were weighted
as 0 ¼ ‘never’, 1 ¼ ‘rarely’, 2 ¼ ‘some of the time’, 15 ¼ ‘most
of the time’ and 16 ‘always’. A summated score of at least 75,
the equivalent of choosing 5 of the 7 items ‘most of the
time’, was the criteria for being in at least the action stage of
change. The DF criterion was determined from subject’s selfreported consumption of 26 high-fat foods and cooking
products (for example, margarine, butter, oil). The nine
weighted response options ranged from 0 ¼ ‘never’ to
14 ¼ ‘two or more per day.’ A score of no more than 35
servings per week or 5 servings per day placed participants
into at least the action stage of change. Calculation of the FB
and DF servings scores was conducted in real-time using
JavaScript during the computer administration of the
surveys.
Measures
Stage of change. Four separate staging algorithms classified
participants into one of five stages of change for PA, and
intake of FVs, FB and DF. First, it was determined if the
Validating measures. Minutes of moderate and vigorous PA
were measured with the MTI Actigraph accelerometer
(WAM 7164). This uniaxial accelerometer is small
(5.1 3.8 1.5 cm), lightweight (45 g) and worn on a belt
snugly around the waist. The Actigraph stored acceleration
Methods
International Journal of Obesity
Validating measures for overweight women
AH Robinson et al
1139
Table 1
Health behavior criteria used for stage of change algorithms
Behavior
Criterion to be classified in at least the action stage of change
Daily recommendation
Source of recommendation
Physical activity
Do at least 30 min of purposeful activitya on at least 5 days
per week
Eat at least five servings of fruit and vegetables (examples
of serving sizes provided)
Do at least five of the following high-fiber eating habits
regularly: Eat
1. breads and crackers made with whole grains,
2. whole, fresh or frozen fruits with skins,
3. starchy vegetables like potatoes, corn and peas,
4. fiber-rich cereals,
5. beans, peas or legumes as side dishes or instead of meat
as entrees,
6. whole, fresh or frozen vegetables with skins,
7. popcorn, dried or fresh fruit for a snack instead of chips
or pretzels.
No more than five servings of high-fat foods (based on 26
item food survey)
Do moderate physical activity for at least
30 min regularly, preferably daily
At least two daily servings of fruit
At least three daily servings of vegetables
At least six servings of grain products, with at
least three being whole grains
USDHHS, 2000 Aim 22–2
Fruit and vegetable
intake
Dietary fiber intake
Dietary fat intake
No more than 30% of energy intake
from total fat
USDHHS, 2000 Aim 19–5
and 19–6
USDHHS, 2000, Aim 19–7
USDHHS, 2000 Aim 19–9
Abbreviation: USDHHS, United States Department of Health and Human Services. aPurposeful physical activity was defined for study participants as ‘planned in
advance and involves continues movement in which your heart beats faster and your breathing is heavier’; study conducted in San Diego, CA, 2003–2004.
counts at 1-min intervals. The Actigraph has been shown to
be valid for quantifying activity levels in laboratory and field
settings.26,27 Data from the monitors were considered valid if
the monitor was worn for at least 3 of the 7 days and for at
least 10 h each day. A monitored hour was not considered
valid if the number of consecutive minutes of 0 counts
exceeded 30 min. Acceleration counts were translated into
minutes of moderate and vigorous PA using accepted cut
points of 1952 and 5725, respectively.28
The Fred Hutchinson Cancer Research Center Food
Frequency Questionnaire (FFQ) was used to estimate daily
FV servings, fiber grams per day and percent energy intake
from fat per day (kcal per day). This FFQ has yielded nutrient
estimates similar to those of 4-day food records and shortterm dietary recalls and has been useful in deriving estimates
of total fruit, vegetable, fiber and DF intake.29,30 Women
reported about foods eaten in the past month by indicating
the frequency and portion size of 122 foods and beverages.
Frequency responses ranged from ‘never or less than once
per month’ to ‘two or more times per day’ for foods and up
to ‘six or more times per day’ for beverages.
Decisional balance comprises two constructs called the
‘pros’ and ‘cons’ of behavior change that address cognitive
and motivational aspects of human decision-making.31 For
the current study, pros and cons of PA, FV consumption, FB
intake and eating high-fat foods were adapted from previously developed inventories.32,33 Scale items assessed what
women may perceive as positive and negative aspects of
changing consumption of fruits, vegetables, fiber (for
example, ‘fresh FVs are too expensive’ and ‘eating high-fiber
foods fills me up so I do not overeat’), high-fat foods (for
example, ‘I feel sluggish and heavy when I eat high-fat
foods’) and engaging in regular PA (for example, ‘I would
look better if I did regular PA’). Participants rated the
importance of each item on a 5-point Likert scale ranging
from 1 ‘not at all important’ to 5 ‘extremely important’.
Self-efficacy refers to one’s confidence about their ability
to perform a behavior.34 Self-efficacy to change dietary
intake was assessed by a six-item FV scale, eight-item fiber
scale and a five-item DF to assess confidence to increase daily
consumption of FVs and fiber (for example, ‘How confident
are you that you can choose foods that are high in fiber
when at a social event?’), and decrease daily consumption of
DF, respectively. A six-item scale assessed the women’s
confidence to engage in PA. All scales were adapted from
previous self-efficacy scales for eating and PA behaviors.32,33,35 Participants responded to each item on a
5-point Likert scale ranging from 1 ‘not at all confident’ to
5 ‘extremely confident’.
Procedure
The data analyzed in the present study were collected during
the baseline and 12-month measurement visits of the
intervention trial. All measures were completed as part of a
computer-based assessment battery. Participants completed
the measures in 30–40 min. The Actigraph accelerometer was
worn for 1 week after completing the assessments and
returned by mail. Valid accelerometer data were available for
309 women at baseline and 221 women at 12 months.
Analyses
Pros, cons and self-efficacy scale scores were computed by
averaging values for items pertaining to each scale. The
distributional properties of the summated scale scores were
examined and coefficient a-tests of internal consistency were
determined for each scale. Scales scores were transformed
and displayed as T-scores (mean ¼ 50, s.d. ¼ 10) in order to
place the scales on a common metric and ease comparisons
by stages of change.
International Journal of Obesity
Validating measures for overweight women
AH Robinson et al
1140
To investigate the validity of the staging measures, a series
of analyses of variance (ANOVAs) were specified with stage of
change as the independent variable and a criterion measure
as the dependent variable. Tukey’s honestly significant
difference post hoc tests determined significant differences
between stages following a statistically significant omnibus
ANOVA test. Z2-Values provided effect size estimates for each
test with values interpreted as small (0.01), medium (0.06)
and large (0.14) based on guidelines defined by Cohen.36
To test the predictive validity of the staging measures
ANOVA models were specified to include the independent
variables treatment group (treatment, control), baseline
stage of change and the treatment by baseline by stage of
change interaction. The dependent variable for reach model
was behavior measured at 12 months. Predictive validity
tests included only participants who completed assessments
at 12 months.
Analyses were conducted using SPSS version 15.0. All
statistical tests were two-tailed. P-values were adjusted for
conducting multiple statistical validity tests through a
Bonferroni correction. Specifically, PA, FV and DF stage of
change measures’ P-values were divided by five and FB was
divided by four, as five and four represent the total number
of estimates of the behaviors of interest and related
psychosocial constructs that were tested for PA, FV, DF and
FB, respectively. The correction yielded new a-levels of
Po0.01 for PA, FV and DF and Po0.0125 for FB. Predictive
validity tests were considered statistically significant at
Po0.05.
Results
Table 2 presents means, standard deviations and a-values for
each validation measure. a-Coefficients exceeded 0.70 on all
scales except the cons scales for PA, FV and FB. The restricted
response range on the cons scales, as indicated by scale
means below the midpoint of the full range of the scales,
resulted in reduced total variance that reduces the ability of
the items to covary and lower a-coefficients.
Physical activity estimates from the accelerometer were
21.2 (s.d. ¼ 15.7) minutes of moderate and 0.83 (s.d. ¼ 2.85)
minutes of vigorous PA per day. Dietary behavior estimates
from the FFQ indicated the women consumed 1.3 (s.d. ¼ 1.2)
servings of fruit per day, 2.2 (s.d. ¼ 1.5) servings of vegetables
per day, 17.0 (s.d. ¼ 8.9) gram per day of fiber and 35.4%
(s.d. ¼ 7.1) kcal per day from total fat.
Stages of change categories were collapsed when the
number of participants in at least one stage was below 20.
This resulted in combining the action and maintenance
stages for the PA (action, n ¼ 16; maintenance, n ¼ 33), FVs
(action, n ¼ 4; Maintenance, n ¼ 27) and FB(action, n ¼ 9;
maintenance, n ¼ 35) staging measures and combining the
precontemplation (n ¼ 10) and contemplation (n ¼ 74) stages
for the DF staging measure.
International Journal of Obesity
Table 2 Number of items, sample distribution characteristics and scale
internal consistency for the psychosocial scale measures
Number of items
Mean (s.d.)
Coefficient (a)
Physical activity
Pros
Cons
Self-efficacy
4
4
6
4.21 (0.66)
2.24 (0.78)
2.58 (0.80)
0.73
0.59
0.80
Fruit and vegetable intake
Pros
Cons
Self-efficacy
4
4
6
3.81 (0.80)
2.16 (0.73)
3.37 (0.87)
0.71
0.54
0.83
Dietary fiber intake
Pros
Cons
Self-efficacy
4
4
8
3.57 (0.95)
1.98 (0.78)
3.17 (0.86)
0.80
0.63
0.89
Dietary fat intake
Pros
Cons
Self-efficacy
4
4
5
2.50 (0.87)
3.76 (0.86)
2.72 (0.94)
0.72
0.72
0.89
Means are raw scale scores with a response range of 1–5; study conducted in
San Diego, CA, 2003–2004.
Table 3 presents the proportion of women in each stage of
change, the means and standard deviations of each validating variable by stage of change, and the summary statistics
from the ANOVA and post hoc tests. Nearly 88% of the
women were classified into the pre-action stages of change
(precontemplation, contemplation and preparation) for PA.
Levels of moderate and vigorous activity varied by stage with
women in the action-maintenance stage engaged in more
moderate and vigorous PA than those did in all other stages.
The pros and cons of PA demonstrated the expected pattern
across the stages of change with the pros generally increasing
and the cons decreasing from precontemplation to actionmaintenance. Self-efficacy generally increased with each
increase in stage of change. All five concurrent validity tests
were statistically significant (Po0.01 with the Bonferroni
correction) with effect sizes ranging from medium to large.
Almost all the women (92%) were in one of the first
three stages of change for FV consumption, with over 50%
of women classified into the preparation stage of change.
Women in the pre-action stages consumed about one daily
serving of fruit and about two daily servings of vegetables
compared to women in the action and maintenance
stage who consumed about three and four daily servings of
FVs respectively. The pros of change differed by one standard
deviation (39.2–49.4) between the precontemplation
and contemplation stages, whereas the cons of change score
was over half a standard deviation lower for the action and
maintenance stage compared to the pre-action stages.
Self-efficacy increased monotonically by stage of change.
All five concurrent validity tests for the FV staging measure
were statistically significant (Po0.01 with the Bonferroni
correction) with effect sizes ranging from small–medium
to large.
Validating measures for overweight women
AH Robinson et al
1141
Table 3
Means, standard deviations and ANOVA summary statistics for validating variables by corresponding stage of change measure
Stage of change
Physical activity stage of change, n (%)
P (Z2)
Tukey’s post hoc test
Pc, C, ProA+M
Pc, C, ProA+M
PcoA+M
Pc, CoPr
PcoC
Pc, C, Pr4A+M
Pc, C, ProA+M
Pc, CoPr
PcoC
Minutes of moderate activity per day , mean (s.d.)
Minutes of vigorous activity per daya, mean (s.d.)
Pros, mean (s.d.)
Pc
47 (11.7)
17.7 (13.4)
0.6 (1.7)
41.7 (11.1)
C
127 (31.7)
21.1 (15.5)
0.5 (1.3)
49.1 (9.9)
Pr
178 (44.4)
19.0 (12.7)
0.3 (0.8)
52.3 (8.7)
A+M
49 (12.2)
32.0 (22.6)
3.5 (6.9)
51.8 (9.5)
Po0.0001 (0.076)
Po0.0001 (0.135)
Po0.0001 (0.112)
Cons, mean (s.d.)
Self-efficacy, mean (s.d.)
53.8 (9.7)
41.4 (6.5)
50.8 (10.5)
48.4 (9.5)
49.7 (9.6)
51.2 (9.0)
45.3 (8.9)
58.1 (10.4)
Po0.0001 (0.046)
Po0.0001 (0.182)
Fruit servings per day, mean (s.d.)
Vegetable servings per day, mean (s.d.)
Pros, mean (s.d.)
Pc
23 (5.7)
1.0 (1.2)
1.8 (1.4)
39.2 (9.3)
C
131 (32.7)
1.1 (1.0)
1.8 (1.2)
49.4 (9.6)
Pr
216 (53.9)
1.3 (1.0)
2.1 (1.4)
51.4 (9.8)
A+M
31 (7.7)
2.8 (1.9)
4.2 (1.7)
50.9 (8.6)
Po0.0001 (0.143)
Po0.0001 (0.165)
Po0.0001 (0.078)
Cons, mean (s.d.)
Self-efficacy, mean (s.d.)
50.7 (11.5)
40.9 (11.5)
51.3 (9.7)
47.7 (9.3)
50.1 (9.9)
51.6 (9.5)
43.2 (8.3)
55.6 (8.9)
P ¼ 0.001 (0.042)
Po0.0001 (0.103)
Fiber (grams per day), mean (s.d.)
Pros, mean (s.d.)
Pc
24 (6.0)
15.0 (5.5)
40.5 (10.7)
C
140 (34.9)
16.3 (8.0)
48.8 (9.8)
Pr
193 (48.1)
16.7 (7.6)
51.7 (9.1)
A+M
44 (10.9)
21.6 (15.0)
51.5 (11.0)
Po0.0001 (0.070)
Po0.0001 (0.075)
Cons, mean (s.d.)
Self-efficacy, mean (s.d.)
49.7 (7.6)
41.8 (8.3)
51.9 (10.7)
47.9 (9.4)
49.4 (9.9)
51.6 (9.9)
46.6 (8.2)
54.0 (9.7)
P ¼ 0.012 (0.027)
Po0.0001 (0.085)
Pc+C
84 (21)
110.3 (41.9)
39.22 (5.5)
52.8 (8.9)
48.2 (11.0)
47.6 (9.1)
Pr
95 (23.7)
104.4 (48.7)
37.46 (5.6)
51.7 (11.2)
51.6 (9.7)
50.4 (10.3)
A
63 (15.7)
49.8 (15.2)
33.30 (7.9)
49.1 (9.5)
52.1 (9.7)
50.5 (10.4)
M
159 (39.7)
50.8 (18.1)
32.94 (7.2)
47.9 (9.5)
49.2 (9.5)
50.8 (10.0)
Po0.0001 (0.045)
Po0.0001 (0.140)
P ¼ 0.001 (0.042)
P ¼ 0.025 (0.023)
P ¼ 0.108 (0.015)
a
Fruit and vegetable intake stage of change, n (%)
Dietary fiber intake stage of change, n (%)
Dietary fat intake stage of change, n (%)
Fat (grams per day), mean (s.d.)
Percent energy from fat, mean (s.d.)
Pros, mean (s.d.)
Cons, mean (s.d.)
Self-efficacy, mean (s.d.)
Pc, C, ProA+M
Pc, C ,ProA+M
PcoA+M
PcoPr
PcoC
Pc, C, Pr4A+M
Pc, CoA+M
Pc, CoPr
PcoC
Pc, C, ProA+M
Pc, CoA+M
Pc, CoPr
PcoC
C4A+M
Pc, CoA+M
Pc, CoPr
Pc+C4A, M
Pc+C, Pr4A, M
Pc+C, Pr4M
NS
NS
Abbreviations: A, action; C, contemplation; M, maintenance; NS, not significant; Pc, precontemplation; Pr, preparation. aPhysical activity stage distribution for tests
with accelerometer derived variables was: Pc, 38 (12%); C, 92 (30%); Pr, 139 (45%), A and M, 40 (13%). Pros, cons and self-efficacy scales reported as
T-scores. Study conducted in San Diego, CA, 2003–2004.
Stage of change for FB had a similar pattern of results as FV
consumption, with 89% of the women in the first three
stages of change and 48% of women in the preparation stage.
Women in the action-maintenance stage consumed more
fiber than women in the pre-action stages. The pros
increased from precontemplation to contemplation whereas
the cons decreased between pre-action and the actionmaintenance stage. Self-efficacy increased monotonically
by stage of change. All five validity tests for the FB staging
measure were statistically significant (Po0.0125 with the
Bonferroni correction) with effect sizes ranging from small to
medium.
In contrast to the other stage of change measures, for the
DF staging measure, the precontemplation and contemplation stages were collapsed into one group comprising 21% of
the women. Nearly 56% of women were classified into the
action or maintenance stage of change for DF. Percent of
energy intake from fat was lower in the action and
maintenance stages compared to the pre-action stages and
this relationship had a large effect size. The pros of eating
high-fat foods were lower in maintenance stage compared to
the pre-action stages. Neither the cons of eating high-fat
foods nor self-efficacy differed by DF stage of change.
All four staging algorithms demonstrated predictive validity with baseline stage of change related to behavior
outcomes 1-year later (Table 4). Effect sizes for these
relationships ranged from 0.04 to 0.126. These ANOVA
models included treatment group and the treatment group
by stage of change interaction providing adjusted effects of
baseline stage of change predicting behavior estimates at 12
International Journal of Obesity
Validating measures for overweight women
AH Robinson et al
1142
Table 4
Predictive validity tests of stage of change at baseline and behavior outcomes at 12-months
Baseline stage of change
Behavior outcomes at 12-months
n
Physical activity
MVPA minutes per dayb
Precontemplation
Contemplation
Preparation
Action and maintenance
Fruit and vegetablec
Servings per day
Precontemplation
Contemplation
Preparation
Action and maintenance
Dietary fiber
(grams per day)
Precontemplation
Contemplation
Preparation
Action and maintenance
Dietary fatc
Percent energy from fat
Precontemplation
Contemplation
Preparation
Action and maintenance
29
63
101
28
15
96
154
21
18
99
137
32
57
66
47
116
Mean (s.d.)
17.80
23.65
23.52
29.72
3.02
4.42
4.77
7.80
15.82
17.31
16.17
22.27
36.36
33.23
30.37
32.46
F-testa
Z2
Tukey HSD
F(3,213) ¼ 3.35, P ¼ 0.020
Z ¼ 0.045
PcoA and M
F(3,278) ¼ 13.33, Po0.001
Z2 ¼ 0.126
Pc,C,Pr oA and M
F(3,278) ¼ 4.02, P ¼ 0.008
Z2 ¼ 0.042
Pc,C,Pr oA and M
F(3,278) ¼ 4.83, P ¼ 0.003
Z2 ¼ 0.050
Pc and C 4A and M
2
(11.59)
(16.89)
(17.91)
(16.19)
(2.01)
(2.68)
(3.04)
(3.30)
(8.01)
(8.75)
(8.34)
(10.19)
(8.63)
(7.53)
(8.12)
(9.48)
Abbreviations: A, action; C, contemplation; Pc, precontemplation; Pr, preparation; M, maintenance; MVPA, moderate to vigorous physical activity. Group by stage
of change interactions were included in each model but were not statistically significant. aF-test for baseline stage of change effect. bAnalysis conducted on square
root transformed MVPA minute per day but raw score means presented in table. cANOVA model includes statistically significant group (treatment, control) effect
(Po0.05). Study conducted in San Diego, CA, 2003–2004.
months. For the FV and DF models, treatment group had a
significant effect on behavior. However, there were no
significant interactions of treatment group and baseline
stage indicating no differential prediction of behavior at 12
months by treatment group.
Discussion
This study assessed the construct, concurrent and predictive
validity of four stages of change measures. The stage
measures determined motivational readiness to make behavior changes for PA, and intake of FVs, FB and DF. Strong
evidence for validity was found for three of the staging
measures (that is, PA, and intakes of FVs and FB), whereas
limited validity was found for the DF stage of change
measure. Validating variables included both estimates of
the target behavior and psychosocial constructs hypothesized to be related to stage of change, similar to a recent
construct validity study on PA stage of change measures.37
Such an approach not only allowed for tests of hypothesized
stage differences at several points among the five stages, but
also tests differences in behavior between pre-action stages
International Journal of Obesity
and the action and maintenance stages, which has often
served as the point of between-group comparisons in
previous studies.38–40
The distribution of individuals across stages of change for
a health behavior varies by the nature of the population.
Broad population-wide surveillance derived from proactive
(for example, random digit dialing), sampling methods can
provide estimates of the proportion who have reached a
health behavior recommendation and proportion who could
benefit by changing, and also their level of motivation to
change.41,42 A very different distribution of individuals
among stages would be found from a sample who signs-up
for an intervention program. These individuals will likely
report being in the preparation stage of change as they are
motivated to make changes on some aspect of their life.
However, in an area such as weight management in which
multiple diet and activity lifestyle changes are needed to
produce and sustain weight change, individuals may be
more motivated to change some behaviors than others.
Identifying stage of change on multiple behaviors can help
clinicians and others developing interventions to determine
which behaviors an individuals should target for change at
various points in the intervention.
Validating measures for overweight women
AH Robinson et al
1143
Limitations and strengths
Study limitations included the use of self-report measures to
assess the validity of the stage of change algorithms, all of
which were subject to potential participant inaccuracies and
biases. However, in addition to the self-report measures of
behavior, psychosocial scales that demonstrated adequate
internal consistency and an objective measure of PA were
used to assess validity. Future studies can incorporate
objective measures of these dietary behaviors to further
contribute to validating the FV and FB, and DF staging
measures. As previously noted, many women in this study
were classified into the preparation stage of change.
Although the present study was part of a larger weight loss
intervention trial, participants who joined the study were
expected to be ready to change their behaviors soon.
However, this pattern was notably different for DF intake
where the majority of the women were in the maintenance
stage of change. The reason for the large percent of women
in maintenance for DF may be attributable to subjects’
misperceptions and consequently, underestimation of their
DF intake. Another reason may be the discrepancy in the DF
staging measure between the staging criteria (that is, five
servings of high-fat foods) and the health recommendation
it was based on (that is, eat o30% of energy intake from total
fat). Previous research on DF stage measures has yielded
mixed results.43,44 One study noted discrepancy between DF
stage of change measure classification results when two
methods of classifying stage of change, self-reported intake
versus food-frequency questionnaire estimation, were compared.45 To improve validity outcomes, Horwath46 recommended DF stage measures inquire about clear dietary
behaviors (for example, similar to ‘five servings of FVs daily’)
rather than a general nutritional outcome (that is, intake no
more than 30% calories from DF). The authors acknowledge
that there is little evidence valid for the present study’s DF
stage of change measure and therefore do not recommend
it for use or further validation studies. Instead, interested
readers are referred to the DF staging measure validated by
Greene et al.43
There are several strengths of this study including a large
sample size, the use of validating variables that were
estimates of the behaviors of interest and related psychosocial constructs (for example, pros and cons), and the
assessment of various forms of validity including construct,
concurrent and predictive validity. The use of a longitudinal
design is a study strong point as it provided an opportunity
to examine predictive validity. This study also employed an
objective measure of PA. Another strength was that the
staging algorithms are based on the most recent health
guidelines issued by the United States Department of Health
and Human Services25 for health and weight management.
This study investigated the construct validity of four stages
of change measures (one PA and three dietary intake
measures) and found strong concurrent validity evidence
for three staging measures and predictive validity for all four
measures. These staging measures can facilitate the standar-
dization of PA, FV, FB and DF stage of change assessment in
descriptive and intervention studies, and may be helpful
in determining motivational readiness on a population
basis.41,42 Using common valid staging measures can facilitate cross-study comparisons, where assorted staging algorithms previously made such comparisons difficult and often
inappropriate.47
Acknowledgements
This work was partially supported by grants CA81495-01A2,
CA113828 and CA85873 from the National Cancer Institute.
Portions of this paper were presented at the Society of
Behavioral Medicine’s 25th Annual Sessions, Baltimore, MD,
March, 2004.
References
1 Prochaska JO, Velicer WF. The Transtheoretical Model of health
behavior change. Am J Health Promot 1997; 12: 38–48.
2 Reed GR, Velicer WF, Prochaska JO, Rossi JS, Marcus BH. What
makes a good staging algorithm: examples from regular exercise.
Am J Health Promot 1997; 12: 57–67.
3 Nigg C, Hellsten L, Norman G, Burbank P, Braun L, Breger R et al.
Physical activity staging distribution: establishing a heuristic
using multiple studies. Ann Behav Med 2005; 29 (suppl): 35–45.
4 Pearlman MY, Wernicke R, Thorndike F, Haaga DA. Stages of
change in smoking cessation: a comparison of expectancies
among precontemplators and contemplators. J Ration Emotive
Cogn Behav Ther 2004; 22: 131–147.
5 Prochaska JJ, Rossi JS, Redding CA, Rosen AB, Tsoh JY, Humfleet GL
et al. Depressed smokers and stage of change: implications for
treatment interventions. Drug and Alcohol Depend 2004; 76: 143–151.
6 Garrett NA, Alesci NL, Schultz MM, Foldes SS, Magnan SJ, Manley
MW. The relationship of stage of change for smoking cessation to
stage of change for fruit and vegetable consumption and physical
activity in a health plan population. Am J Health Promot 2004; 19:
118–127.
7 de Vet E, de Nooijer J, de Vries NK, Brug J. Determinants of
forward stage transition from precontemplation and ontemplation for fruit consumption. Am J Health Promot 2005; 19: 278–285.
8 Hughes JR, Keely JP, Fagerstrom KO, Callas PW. Intentions to quit
smoking change over short periods of time. Addict Behav 2005;
30: 653–662.
9 O’Hea EL, Boudreaux ED, Jeffries SK, Taylor CL, Scarinci IC,
Brantley PJ. Stage of change movement across three health
behaviors: the role of self-efficacy. Am J Health Promot 2004; 19:
94–102.
10 Schumann A, Meyer C, Rumpf H, Hannover W, Hapke U, John U.
Stage of change transitions and processes of change, decisional
balance, and self-efficacy in smokers: a Transtheoretical Model
validation using longitudinal data. Psychol Addict Behav 2005; 19:
3–9.
11 Jordan PJ, Nigg CR. Applying the Transtheoretical Model:
tailoring interventions to stages of change. In: Burbank PM,
Riebe D (eds). Promoting Exercise and Behavior Change in Older
Adults: Interventions with the Transtheoretical Model. Springer
Publishing Co.: New York, NY, US, 2002, pp 181–207.
12 Steptoe A, Kerry S, Rink E, Hilton S. The impact of behavioral
counseling on stage of change in the fat intake, physical activity,
and cigarette smoking in adults at increased risk of coronary
heart disease. Am J Public Health 2001; 91: 265–269.
International Journal of Obesity
Validating measures for overweight women
AH Robinson et al
1144
13 Hellsten L, Nigg C, Norman G, Burbank P, Braun L, Breger R et al.
Accumulation of behavioral validation evidence for physical
activity stage of change. Health Psychol (in press).
14 Dallow CB, Anderson J. Using self-efficacy and a Transtheoretical
Model to develop a physical activity intervention for obese
women. Am J Health Promot 2003; 17: 373–381.
15 Burkholder GJ, Evers KA. Application of the Transtheoretical
Model to several problem behaviors. In: Burbank PM, Riebe D
(eds). Promoting Exercise and Behavior Change in Older Adults:
Interventions with the Transtheoretical Model. Springer Publishing
Co.: New York, NY, US, 2002, pp 85–145.
16 Glanz K, Rimer BK, Lewis FM. Theory, research, and practice in
health behavior and health education. In: Glanz K, Rimer BK,
Lewis FM (eds). Health Behavior and Health Education. Jossey-Bass:
San Francisco, CA, 2002, pp 22–40.
17 Reibe D, Greene GW, Ruggiero L, Stillwell KM, Blissmer B,
Nigg CR et al. Evaluation of a healthy-lifestyle approach to weight
management. Prev Med 2003; 36: 45–54.
18 Clark PG, Rossi JS, Greaney ML, Riebe DA, Greene GW, Saunders
SD et al. Intervening on exercise and nutrition in older adults: the
Rhode Island SENIOR Project. J Aging Health 2005; 17: 753–778.
19 Cullen KW, Bartholomew LK, Parcel GS, Koehly L. Measuring
stage of change for fruit and vegetable consumption in 9–12-year
old girls. J Behav Med 1998; 21: 241–254.
20 de Oliveria M, Anderson J, Auld G, Kendall P. Validation of a tool
to measure processes of change for fruit and vegetable
consumption among male college students. J Nutr Educ Behav
2005; 37: 2–11.
21 Gulliver P, Horwath C. Women’s readiness to follow milk product
consumption recommendations: design and evaluation of a
‘stage of change’ algorithm. J Hum Nutr Diet 2001; 14: 277–286.
22 Sarkin JA, Johnson SS, Prochaska JO, Prochaska JM. Applying the
transtheoretical model to regular moderate exercise in an overweight population: validation of stages of change measure. Prev
Med 2001; 33: 462–469.
23 Richards-Reed G, Velicer WF, Prochaska JO, Rossi JS, Marcus BH.
What makes a good staging algorithm: examples from regular
exercise. Am J Health Promot 1997; 12: 57–66.
24 DiClemente C, Prochaska J, Fairhurst S, Velicer W, Velasquez M,
Rossi J. The process of smoking cessation: an analysis of
precontemplation, contemplation, and preparation stages of
change. J Consult Clin Psychol 1991; 59: 295–304.
25 US Department of Health and Human Services. Healthy People
2010: Understanding and Improving Health, 2nd edn. Government
Printing Office: Washington DC, 2000.
26 Nichols J, Morgan C, Chabot L, Sallis J, Calfas K. Assessment of
physical activity with the Computer Science and Applications,
Inc. accelerometer: laboratory versus field validation. Res Q Exerc
Sport 2000; 71: 436–443.
27 Welk GJ. Use of accelerometry-based activity monitors to assess
physical activity. In: Welk GJ (ed). Physical Activity Assessments for
Health-related Research. Human Kinetics: Champaign, IL, 2002,
pp 125–142.
28 Freedon PS, Melanson E, Sirard J. Calibration of the Computer
Science and Applications, Inc. accelerometer. Med Sci Sports Exerc
1998; 30: 777–781.
29 Kristal AR, Vizenor NC, Patterson RE, Neuhouser ML, Shattuck
AL, McLerran D. Precision and bias of food frequency-based
International Journal of Obesity
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
measure of fruit and vegetable intake. Cancer Epidemiol Biomarkers
Prev 2000; 9: 939–944.
Patterson RE, Kristal AR, Carter RA, Fels-Tinker L, Bolton MP,
Agurs-Collins T. Measurement characteristics of the Women’s
Health Initiative Food Frequency Questionnaire. Ann Epidemiol
1999; 9: 178–187.
Velicer WF, DiClemente CC, Prochaska JO, Brandenburg N.
A decisional balance measure for assessing and predicting
smoking status. J Pers Soc Psychol 1985; 48: 1279–1289.
Marcus BH, Rakowski W, Rossi JS. Assessing motivational
readiness and decision making for exercise. Health Psychol 1992;
11: 257–261.
Rossi SR, Greene GW, Rossi JS, Plummer BA, Benisovich SV, Keller
S et al. Validation of decisional balance and situational temptations measures for dietary fat reduction in a large school-based
population of adolescents. Eating Behav 2001; 2: 1–18.
Bandura A. Social Foundations of Thought and Action: A Social
Cognitive Theory. Prentice Hall: Englewood Cliffs, NJ, 1986.
Sallis JF, Pinski RB, Grossman RM, Patterson TL, Nader PR. The
development of self-efficacy scales for health-related diet and
exercise behaviors. Health Educ Res 1988; 3: 283–292.
Cohen J. Statistical Power Analysis for the Behavioral Sciences, 2nd
edn. Erlbaum: Hillsdale, NJ, 1988.
Schumann A, Estabrooks PA, Nigg CR, Hill J. Validation of the
stages of change with mild, moderate, and strenuous physical
activity behavior, intentions, and self-efficacy. Int J Sports Med
2003; 24: 363–365.
Johnson SS, Driskell MM, Johnson JL, Dyment SJ, Prochaska JO,
Prochaska JM et al. Transtheoretical model intervention
for adherence to lipid-lowering drugs. Dis Manag 2006; 9:
102–114.
Snelling AM, Adams TB, Korba C, Tucker L. Stages of change
algorithm for calcium intake by male college students. J Am Diet
Assoc 2006; 106: 904–907.
Walton J, Hoerr S, Heine L, Frost S, Roisen D, Berkimer M.
Physical activity and stages of change in fifth and sixth graders.
J Sch Health 1999; 69: 285–289.
Laforge RG, Velicer WF, Richmond RL, Owen N. Stage distributions for five health behaviors in the United States and Australia.
Prev Med 1999; 28: 61–74.
Velicer WF, Fava JF, Prochaska JO, Abrams DB, Emmons KM,
Pierce JP. Distribution of smokers by stage in three representative
samples. Prev Med 1995; 24: 401–411.
Greene GW, Rossi SR, Reed GR, Willey C, Prochaska JO. Stages of
change for reducing dietary fat to 30% of energy or less. J Am Diet
Assoc 1994; 94: 1105–1110.
Ni Mhurchu C, Margetts BM, Speller VM. Applying the
stages-of-change model to dietary change. Nutr Rev 1997; 55:
10–16.
Lechner L, Brug J, De Vries H, Van Assema P, Mudde A. Stages of
change for fruit, vegetable, and fat intake: consequences of
misconception. Health Educ Res 1998; 13: 1–11.
Horwath CC. Applying the transtheoretical model to eating
behaviour change: challenges and opportunities. Nutr Res Rev
1999; 12: 281–317.
Marshall SJ, Biddle SJ. The Transtheoretical Model of behavior
change: a meta-analysis of applications to physical activity and
exercise. Ann Behav Med 2001; 23: 229–246.