]ournal of Substance Abuse, 2, 217-235 (1990)

Stages of Change Profiles in

Outpatient Alcoholism Treatment

Carlo C. DiClemente
Sheryl O. Hughes
University of Houston
Houston, Texas

While few believe in the uniformity myth about alcoholics, discovering rel-
evant dimensions that usefully divide the treatment population and guide in-
tervention has been a difficult task. This study evaluated a stages of change
assessment measure with a group of 224 adults entering outpatient alcoholism
treatment. Subjects' scores on the precontemplation, contemplation, action, and
maintenance stage subscales of the URICA were subjected to duster analysis
yielding five distinct and theoretically consistent profiles. Profiles were labeled
as follows: 1. Precontemplation (n = 63) 2. Ambivalent (n = 30) 3. Participation
(n = 51) 4. Uninvolved (n = 27), and 5. Contemplation (n = 53).
These five groups of subjects demonstrated no differences on demographic
characteristics but significantly differed on Alcohol Use Inventory subscales, a
temptation to drink and abstinence self-efficacy measure as well as several
outcome variables. Group differences support the validity of the cluster analytic
profiles, confirm the interpretation of profile groups, and provide interesting
contrasts consistent with the stages of change model. Classification of individuals
on the stages of change offers a useful perspective for alcoholism treatment
matching research.

The search for treatment-relevant typologies for alcoholism and substance

abuse problems has been extensive and frustrating. Exploration of a unitary
addictive personality has been unproductive (Sutker & Allain, 1988). Efforts
to classify and subtype individuals with substance abuse problems have focused
on severity of addiction (McLellan, Luborsky, Woody, & O'Brien, 1980),
typologies of use patterns (Morey, Skinner, & Blashfield, 1984; Wanberg &
Horn, 1983), multiple associated clinical features (Donovan, Kivlahan, &
Walker, 1986), and more theoretically based childhood and family charac-
teristics. Recently, the focus on treatment matching has turned from historical
and etiological factors towards the process of addictive behavior change
The authors wish to thank the personnel at the Texas Research Institute of Mental Sciences
who made this research possible. In particular we are grateful to Dr. Jack Gordon, Charles
Cortez, Glen Razak, Becky Re)'es, and case managers, Riva Okonkwo, Betty Hofker, Letitia
Miranda, Norman Weeden, Diane Newsome, and Gwen Scott. Special thanks is due to Robin
Morris, who consulted with us on the duster analytic procedures, and Mildred Dobson, who
typed the many revisions of the manuscript.
Correspondence and requests for reprints should be sent to Carlo C. DiClemente, Psychology
Department, University of ttouston, Houston, TX 77204-5341.

217
218 C. C. DiClemente and S. O. Hughes

(DiClemente & Prochaska, 1985; Marlatt, Baer, Donovan, & Kivlahan, 1988;
Miller & Hester, 1986; Prochaska & DiClemente, 1986a). Thus, fluid process
variables replace static personality characteristics or historical traits as the
key dimensions to be examined. Level of participation or stage of change in
this process of change would be central to this type of treatment matching.
It is axiomatic in alcoholism treatment that not everyone who arrives for
treatment actually shows up. Denial, resistance and differing levels of moti-
vation are considered significant problems in treatment participation as well
as outcome (Miller, 1985; Marlatt et al., 1988). Most often, readiness for
change is seen as a dichotomous phenomenon, a presence or absence of
motivation which is measured retrospectively. DiClemente and Prochaska
(1982, 1985) have conceptualized change as a stage phenomenon. From this
perspective intentional change is viewed as the movement from precontempla-
tion, (characterized by the individual being unaware of or unwilling to ac-
knowledge problems and engaging in little change process activity) to contem-
plation of the problem (marked by consciousness raising and decision-making
activity). T h e action stage follows with an increase in behavioral coping as
well as other change process activity. T h e final stage of successful change is
maintenance, which, in many ways, represents continued action to reinforce
and firmly establish the new behavior change into the individual's lifestyle
(Prochaska & DiCiemente, 1986a, 1986b).
Movement through these stages is seen as not merely linear, as described
above, but cyclical in nature. Individuals can begin to contemplate change,
then decide not to change and exit the cycle at that point. Action or
maintenance is often interrupted by relapse (i.e., a return to the problematic
behavior). Individuals, especially those with addictive behavior problems like
smoking, obesity, or alcoholism often make several revolutions through the
cycle either with or without formal intervention before achieving successful
change (Prochaska & DiClemente, 1984, 1986a; Schachter, 1982).
Identification and classification of stage of change status is a critical element
of the model and a focus of the current study. In previous research, stage
of change assessment has been accomplished by a series of questions regarding
attitudes and behaviors related to the problem behavior, specifically, cigarette
smoking (DiClemente & Prochaska, 1985). This assessment involves a dicho-
tomous branching procedure based on current smoking and cessation activity
as well as intentions to quit in the next month, 6 months, or year. While
this system provides for clean stage assignment, it does not take into account
more subtle attitudinal differences that could identify within or between stage
categories. T o address this problem, a questionnaire called the University of
Rhode Island Change Assessment Scale (URICA) has been developed that
provides a continuous measure of attitudes representing each of the stages
of change (McConnaughy, Prochaska, & Velicer, 1983). In its initial use with
several adult outpatient psychiatric samples, the URICA has demonstrated
solid psychometric properties for scale composition and theoretical consistency
(McConnaughy et al., 1983). Profiles produced using a cluster analytic tech-
nique to classify groups of subjects yielded five to eight distinct and relevant
Stages of Change Profiles 219

stage o f change profiles (McConnaughy, DiClemente, Prochaska, & Velicer,

1989; McConnaughy et al., 1983).
This study used the URICA with a sample o f adults with alcohol problems
who presented themselves to an outpatient alcoholism treatment program.
T h e questions addressed were (1) whether the scale would be useful with an
alcohol dependent population, and (2) whether we could identify distinct,
reliable, and relevant profiles for subgroups of these individuals using the
stages of change scale. Finally, we were interested in (3) whether these profiles
would differ in a predictable manner on other measures of alcohol use and
history, temptations to drink, and alcohol abstinence self-efficacy.

METHOD
Subjects
Subjects were 224 adults who applied for treatment to the Outpatient
Alcoholism T r e a t m e n t Program at the Texas Research Institute of Mental
Sciences over an 18-month period. Subjects all had serious drinking problems
but did not currently need medical detoxification nor long-term inpatient
treatment as assessed by clinical staff. These 224 subjects were predominantly
male (65%) with an average age of 33 years, ranging from 18 to 60. Most
were Caucasian (79%), wifh about 12% Blacks and 4% Mexican Americans.
In formal education, 23% had less than high school, 36% had completed
high school, 37% had some college, and 4% had some postgraduate education.
Approximately 40% o f these subjects were employed full or part time. In
terms o f current marital status, 32% never married, 25% were currently
married, and 42% were divorced or separated. As assessed by the Alcohol
Use Inventory (AUI; Wanberg, Horn, & Foster, 1977) which was normed
on an inpatient alcoholism treatment population, subjects represented a rather
serious level of alcohol problems. T h e y had been involved in problem drinking
for an average of 8 years; had a mean score o f 5 (range 1-10) on the general
alcoholism, alcoholic deterioration, and anxiety related to drinking subscales;
and scored at the inpatient mean on measures of withdrawal symptoms and
daily quantity o f alcohol.

Measures
Demographic Information Sheet
As part of the intake process, all subjects were asked a series of questions
which were coded on a standard intake form. These questions focused on
age, living situation, education, and employment status.

University o f Rhode Island Change Assessment Scale (URICA)

This scale was developed to measure the stages of change (McConnaughy
et al., 1983). It operationally defines four theoretical stages of change
(Precontemplation, Contemplation, Action, and Maintenance) identified by
220 C. C. DiClemenle and S. O. Hughes

DiClemente and Prochaska (1982, 1985). T h e scale consists of 32 items, with

eight items measuring each of the stage subscales. T h e items are written so
that they are relevant to change of a "problem" that is determined by the
subjects. Internal consistency for each scale was quite high (Coefficient Alpha
ranged from .88 to .89). Interscale correlations form a simplex pattern
consistent with the theory. Responses are given on a five-point Likert format
(1 = strong disagreement to 5 = strong agreement). Subscale scores are summed
and scores on each of the four stages are obtained for each subject.

Alcohol Use Inventory (AUI)

This 147 item, multiple-choice questionnaire is composed of 92 scales, 16
primary scales that measure the patterns of alcohol use, benefits of drinking,
and the effects o f drinking on personal and interpersonal adjustment (Wanberg
et al., 1977). In addition, higher order scales represent broader dimensions
o f alcohol use and one factor measures alcoholism in general. Responses are
given in a dichotomous presence-absence format or a 3-point response allowing
for some gradations. Scores are summed and plotted against a normative
sample of over 2000 hospitalized alcoholics.

Alcohol Abstinence Self Efficacy Scale (SE)

This scale consists of 49 items representing cues related to drinking.
Subjects are requested to respond how "tempted" they would be to drink
in each situation on a five-point scale (from not at all = 1 to extremely = 5).
Similarly, they are asked to rate how "confident" they are that they would
not drink in that situation on a similar 5-point Likert scale which represents
their self-efficacy (Bandura, 1982). Scores are summed separately for Temp-
tation and Self-efficacy. Similar scales developed for smoking and other
addictive behaviors have demonstrated relevance and solid psychometric prop-
erties (DiClemente, 1986). Initial reliability and validity estimated for this
scale demonstrated high internal consistency (Spearman & Brown = .95) for
each scale and a substantial negative correlation (-.58) between temptation
and self-efficacy (DiClemente, Gordon, & Gibertini, 1983).

Procedure
Clinical Care
Subjects were individuals who called the community outpatient alcoholism
treatment program. If callers needed inpatient treatment or emergency care,
they were given referrals to other programs. This program offered sliding
scale services so individuals without resources frequently came to the program
as a last resort and, if the physical symptoms of detoxification were under
control, were accepted into the program.
T h e treatment philosophy of this program relied on a case management
system and individualized treatment programming. Group therapy modules
included alcohol education, values clarification, alternatives to drinking, so-
cialization, and family components, as well as a more traditional insight-
Stages of Change Profiles 221

oriented group. Individual therapy sessions with the case manager and other
treatment staff were available on an as-needed basis. A medication evaluation
and follow-up chemotherapy sessions were used to assess and treat withdrawal
symptoms and other psychiatric problems. Involvement in AA and other
relevant self-help support groups was strongly advocated.but not made a
condition of treatment. T h e treatment program was housed in a compre-
hensive mental health treatment and research facility and referrals came from
other clinics as well as directly to the program.

Research
At intake, subjects individually completed the demographic information
questionnaire, the URICA, and the AUI. T h e Self-Efficacy Questionnaire
(SE) was added to the assessment battery several months after the initiation
o f the research program; hence, not all subjects received this questionnaire.
T h e percentage o f subjects who missed the SE scale was rather evenly
distributed across groups and ranged from 27% to 44%. In addition, the
possibility of selective attrition seems less likely since subjects usually completed
both URICA and the SE scale during the same intake session. Missing data
are due primarily to the absence of the SE Scale until several months after
the initiation o f the project.

RESULTS
A principal component analysis was used to reassess the factor structure
o f the U R I C A with this alcoholism treatment population. This analysis es-
sentially replicated the original four components. Several items, however,
appeared to have weak and inconsistent loadings on their original components.
Since we were going to use these scores in a subsequent analysis, we used a
conservative approach to measuring the constructs and eliminated four items,
the lowest loading item from each stage subscale of the U R I C A prior to the
subsequent data analyses. This yielded a seven-item subscale measuring each
o f the stages o f changeA (See Table 1, p. 222, for sample items for each
scale). Internal consistency (Cronbach's Alpha) was moderate to high for the
Precontemplation (.60), Contemplation (.75), Action (.82), and Maintenance
(.80) subscales. Mean scores for each subscale (possible range = 7 to 35) prior
to standardization were: Precontemplation: M = I 1.7 (SD = 3.8); Contempla-
tion: M = 31.3 (SD = 2.7); Action M = 27.0 (SD = 4.1); and Maintenance:
M = 26.3 (SD = 5.3).
In an attempt to find groups of subjects with similar stages of change
profiles, U R I C A scores for the first 167 subjects admitted to treatment were
subjected to a cluster analysis using a hierarchical agglomerative method
(minimum variance) with squared Euclidean distance as the distance measure.

' T h e full 8-item URICA Scale with initial scoring norms for both the psychotherapy sample
a n d the 7-item adaptation with norms for the alcohol treatment sample are available from the
first author.
 222 C.C. DiClemente and S. O. Hughes

Table I. Sample Items from the Stages of Change S c a l e

Stage Item
Precontemplatlon Stage
1. As far as I'm concerned, I d o n ' t have any problems that need changing.
2. I'm not the problem one. It doesn't make much sense for me to be here.

Contemplation Stage
I. I have a problem and I really think I should work on it.
2. I'm hoping this place will help me to better understand myself.

Action Stage
1. I am doing something about the problems that had been bothering me.
2. Anyone can talk about changing: I'm actually doing something about it.

Maintenance Stage
1. It worries me that I might slip back on a problem I have alre_ady changed so I am here to
seek help.
2. I thought once 1 had resolved the problem 1 would be free of it, but sometimes I still find
myself struggling with it.

Because cluster analyses may yield meaningless cluster results, these same
subjects were also subjected to another hierarchical agglomerative method
(complete linkage). These two methods resulted in a high degree o f consistency
and confirmed a clear cluster structure to the data set. Since the minimum
variance method has been shown to be the best method for clustering
psychological data (Punj & Stewart, 1983), the results from this method were
used for further analyses. Subsequently, six subjects were deleted from the
data set because their scores deviated from the main sample distribution.
T o determine the number of clusters, both the hierarchical tree and the
clustering coefficients were used. Five clusters appeared to adequately differ-
entiate groups of subjects. These five clusters were then subjected to a K-
means iterative partitioning clustering procedure (Punj & Stewart, 1983). This
method allows for relocation of any misassigned subjects to a more appropriate
cluster. In addition, it allows for the lowest within-cluster variance and the
highest between-cluster variance. As further support for the five-cluster so-
lution, less than 10% o f the subjects were actually placed in different clusters
using the iterative partitioning method. Therefore, cluster profiles remained
stable and the five-cluster solution was further confirmed.
To further ensure the internal validity of the five-cluster solution, data
alteration through the addition of subjects was used to examine the stability
of the cluster solutions (Morris, Blashfield, & Satz, 1981). Sixty-three subjects,
who were later admitted to treatment, were added to the original 161 subjects
and the K-means cluster analysis rerun. After the 63 subjects were added,
all of the original five clusters maintained their initial profile structures. T h e
added subjects distributed rather evenly over the five clusters with 22%
joining Cluster 1, 18% joining Cluster 2, 22% joining Cluster 3, 16% joining
Cluster 4, and 22% joining Cluster 5.
Stages of Change Profiles 223

T h e results o f the cluster analysis classified 224 subjects into five distinct
clusters which have unique profiles on the four stages of change subscales
of the URICA (see Figure la-e). These clusters are labeled and described
as follows:

1. Precontemplation Cluster: T h e 63 subjects in this cluster (Figure la) are

characterized by above-average scores on Precontemplation (M = 56.3),
particularly low scores on Contemplation (M = 38.9), and below-average
scores on Action (M = 47.6) and Maintenance (M = 47.6). These subjects
seem to be neither contemplating nor engaging in change. They seem,
instead, to be maintaining the status quo with respect to their alcoholism
problem and resisting the view that they have a problem.
2. Ambivalent Cluster: T h e 30 subjects in this cluster (Figure lb) are char-
acterized by above-average scores on all four stages (M = 52.5 for
Contemplation, M = 52.3 for Action, M = 56.5 for Maintenance) with
particularly high scores on Precontemplation (M = 64.5). This is an anom-
alous profile with a high level o f endorsement across the subscales. Subjects
with this profile seem somewhat reluctant or ambivalent about changing
their alcohol-problemed behavior and endorse conflicting statements.
3. Participation Cluster: T h e 51 subjects in this cluster (Figure lc) are well
below average on Precontemplation (M = 41.4) and above average on
Contemplation (M = 59.7), Action (M = 61.9), and Maintenance (M = 56.7).
These subjects seem to report a high level o f investment and involvement
in change.
4. Uninvolved or Discouraged Cluster: T h e 27 subjects in this cluster (Figure
ld) are characterized by below-average scores on all the stages (M = 46.1
for Precontemplation, M -- 45.8 for Contemplation), with particularly low
scores on Action (M = 37.7) and Maintenance (M = 32.2). These subjects
have a low level of endorsement overall and seem rather listless in
affirming their ability to take action. T h e y seem uninvolved in changing
their behaviors and may represent a group of subjects who have given
up on change.
5. Contemplation Cluster: T h e 53 subjects in this cluster (Figure le) are
characterized by low scores on Precontemplation (M = 44.5), a higher
level of endorsement on Contemplation (M = 54.6), and low to average
scores on Action (M = 46.4) and Maintenance (M = 51.7). These subjects
seem interested in changing, but have not yet begun to take action. T h e y
are best described as seriously thinking about change but not yet ready
for action.
 Figure 1 (a-e). Profiles of change in alcoholism treatment.
Precontemplation Cluster (n = 63)

T-Score

55

50

50 . . . . . . . . . . . . . . . . . . . . . . .

.45,

40.

35,

3O 3

! I I . I

Precontemp Contemplation Action Maintenance

Fig. l(a) Stages of Change

Ambivalent Cluster (n = 30)

T-Score

7~I
55

80

53

50 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~ . . . . . . .

45

4O

I
35

-ao

I I I !

Precontemp Contemplation Action Maintenance

FiB. l(b) Stages of Change
224
 Participation Cluster (n = 51)

T-Score

9 70

O5

~0 . . . . . . . . . . . . . . . .

45

40

35

30

! | 1 !

Precontemp Contemplation Action Maintenance

Fig. l(c) Stages of Change

Uninvolved Cluster (n = 27)

T-Score

7Q

B5

5o

55

513

45-

4Q 9

35,

~K3,

! ! |
i

Precontemp Contemplation Action Maintenance

Fig. l(d) Stages of Change
225
226 C. C. DiClemente and S. O. Hughes

Figurel (a-e). (Continued)

Contemplation Cluster (n = 53)

T-Scare

70-

85-

60-

45-

4O

35

3O

I I I I

Precontemp Contemplatian , Actian Maintenanco

Fig. l(e) Stages of Change

To examine the external (criterion related) validity of these 5 cluster

subtypes, we examined group differences on other measures obtained in the
study. Chi-Square analyses by group (n = 5) for gender, ethnicity, marital and
employment status, as well as levels of education and income were all non-
significant. Differences between the five cluster groups were also investigated
using multivariate analysis of variance (MANOVA) on the Alcohol Use In-
ventory primary and secondary subscales. For the MANOVAs, primary scales
were grouped according to theoretical concepts defined in the most recent
Guide to the Alcohol Use Inventory (Horn, Wanberg & Foster, 1987). These
include Benefits of Drinking (scales 1, 2, 7, and 15), Styles of Drinking (scales
3, 4, and 5), Consequences of Drinking (scales 9, 10, 11, 12, and 16), and
Concerns over Drinking (scales 6, 8, and 14). Differences among the cluster
groups were examined in a final MANOVA for the second-order scales and
in a single analysis of variance for the general alcoholism third-order scale.
The multivariate main effect for group was significant for all five MANOVAs
(Benefits of Drinking: F(16,660)=2.37, p=.002; Styles of Drinking:
F(12,574)= 1.90, p=.03; Consequences of Drinking: F(20,714)= 1.83, p=.015;
Concerns over Drinking: F(12,574)=3.00, p=.0004; Second-order scales:
F(20,710)=1.60, p=.046). The analysis of variance for the general alcoholism
scale was also significant: F(4,217)=4.72, p=.001. As shown in Tables 2 & 3,
Stages of Change Profiles 227

univariate F-tests were significant for 16 of 21 scales. Only gregarious drinking,

sustained drinking, drinking that follows marital problems, drinking that
provokes marital conflict, and obsessive-sustained drinking were not signifi-
cant. Post hoc tests were performed on each scale using the Tukey test to
control for familywise Type I error (Keppel, 1982). Results from these analyses
are presented in Tables 2 and 3 (pp. 228-229).
On the AUI primary subscales, the groups differed on scales related to
the benefits of drinking as well as concerns over drinking and consequences
of drinking (Table 2). On the Mental Benefit scale (Benefits of Drinking) the
Ambivalent cluster (#2) is significantly higher than the Precontemplation (#1),
Contemplation (#5), and Uninvolved (#4) groups. T h e Participation group
(#3) acknowledges that they have sought help for their alcohol problem more
often (Concerns over Drinking) and admit to higher levels of loss of control
and social role disruption (Consequences of Drinking) than the Precontem-
plation group (#1). T h e Contemplation group (#5) demonstrates the highest
level of post drinking worry (Concerns over Drinking) and differs significantly
on that scale from the Precontemplation group (#1). T h e clusters do not
differ on either gregarious or sustained drinking patterns. T h e differences
at the primary scale level seem related to motivational and behavioral di-
mensions which are logically and theoretically linked to the stage of change
most reflected by a particular profile.
Second- and third-order AUI scales represent more summary scores and
higher order constructs (Wanberg et al., 1977). At these levels, significant
group differences emerged primarily between the Precontemplation (#1) and
both the Contemplation (#5) and Participation (#3) profile groups. These
latter groups demonstrated higher levels of anxiety related to their drinking
behavior that is presumably a good prognostic treatment indicator (Wanberg
et al., 1977). In addition, both the Contemplation (#5) and Participation (#3)
groups acknowledge significantly higher levels of problems resulting from
their alcohol use, as well as general alcoholism, than their Precontemplation
counterparts (#1). It would be expected that subjects in earlier stages of
change would not see their behavior as problematic and could be expected
to be defensive or denying about the problem behavior. T h e differences on
the AUI scales support the validity of the clustering decision as well as the
cluster definitions.
Although data on the self-efficacy measure were not obtained for all
subjects, an A N O V A comparison of subjects in each cluster who had com-
pleted the self-efficacy scale was undertaken. Mean scores for subjects (n = 142)
on temptation was 160 (SD = 37) and on self-efficacy was 139 (SD = 41) for
this 49-item scale. Pearson first-order correlation of total scores on Temptation
and Self-efficacy for the entire sample was substantial and negative (r = -.64).
Table 4 (p. 231) presents the summary o f group differences on both
Temptation and Self-efficacy scales (range 49 to 245) as well as the mean
difference between temptation and self-efficacy for each group. Finally, Table
4 includes the Pearson first-order correlations between Temptation and Self-
efficacy scores for each group. These correlations were used to measure the
 E

c-
O

~%~ --~ 9 ,
c-

o
D E

8O .

0 =

r-
~ ~ ~ ~ ~ L ~

~b
<
b

O
T,

II
.<

..~
0 ~ U
,;V Q-~
V ~
,q

c~ ~ . _ ~.
~ 6"4 e-

-s '~ ~ "5
9~ ~ . - ~
_ ~ ~.~ 8 ,.
,,.~ . ~ ~ 0

~.~
el =o oo~ ~:~
9.--~ ~ ,~_~
H

228
 o
~,~ ~. ~- a.. J

o
e-,

o
o. c-

O4

o
E

e...
f_, ee~

o
L. tJ~
~

-_=
0 u

e~

.o

~EE
0
~ U
0

qg~'o
~

C ~ gE
~ ' - ~
~ ~

~'~
~ <<< 0"<
C

229
230 C. C. DiClemente and S. O. Hughes

relationship between these two variables (which is usually substantial with

correlations in the range o f - . 5 to -.7; DiClemente, 1986; DiCiemente,
Gordon, & Gibertini, 1983).
T h e Uninvolved group (#4) demonstrated the highest level o f Temptation
and the lowest level of Self-efficacy. T h e differences between these variables
was quite large, indicating a sense o f helplessness or hopelessness which
supports the impression that this cluster represents a group who seem dis-
couraged over the possibility of change. T h e Precontemplation group of
subjects (#1) reported the lowest level of temptation and were the only group
to have efficacy levels greater than temptation levels. Whether real or imag-
ined, this group tends to be quite confident in their ability to abstain from
drinking across a wide range of situations. This is consistent with their
tendency to report the lowest levels of deterioration, loss of control, and
general alcoholism on the AUI. Although it is difficult to determine completely
whether this group of subjects does not have a serious problem or are denying
the severity of the problem and inflating their sense o f confidence, lack of
significant differences on the daily quantity o f alcohol and withdrawal subscales
o f the AUI would support a denial or minimization interpretation.
T h e Ambivalent group (#9) demonstrated rather high levels o f temptation
but also had the highest level of self-efficacy of all the groups. Most notable
for this group is that the correlation between temptation and efficacy is small.
This relationship, which is so consistent among the other groups, is quite
disrupted in this group. This lack o f relationship in temptation and efficacy
assessment may explain their ambivalence, since they have difficulty evaluating
their ability to abstain with respect to their temptation to drink.
Most interesting, both the Participation (#3) and the Contemplation (#5)
groups had similar profiles on the self-efficacy measures. T h e y demonstrated
great consistency in their evaluations o f temptation and efficacy. T h e more
temptation they reported, the lower the efficacy they had. In addition, there
seems to be a rather moderate span between temptation and efficacy levels
which is not as large or overwhelming as in the Uninvolved group (#4) and
not as bold as in the Precontemplation group (#I).
It is impossible to determine causality from this cross-sectional data. There-
fore the results cannot confirm whether the readiness to change as demon-
strated by the stage profiles contributed to the efficacy assessment or whether
the type or accuracy of the self-efficacy assessment contributed to stage of
change profile. Nevertheless, group differences on the efficacy assessment
provide additional insight into the profiles established by the stages of change
scale.

CONCLUSION
Results indicate that the URICA can be used with individuals who come
to treatment for alcoholism problems and yields a series of five distinct profiles
o f subjects defined by a cluster analytic procedure. These five groups have
stage of change profiles clearly related to the precontemplation, contempla-
 e+
o

~ - '
~ 1
v

P: ,

t-
O

~ "
~ 1
r~

m
o

I..

o,~
V-~
oo.
~v 2.

. ~ rr-'

Ul

ii
8
++ ~'~ oog~
+

~Nu
231
232 C. C. DiClemente and S. O. Hughes

tion, and action stages of change and are quite consistent with theoretical
considerations. In addition, groups of subjects defined by the stages of change
profiles demonstrated significant differences on AUI subscales, temptations
to drink, and self-efficacy to abstain from drinking. Group differences on the
above variables support the external validity of the cluster analytic profiles,
confirm interpretation or labeling of the profiles, and provide interesting
contrasts consistent with the stage of change model.
This sample of subjects is relatively homogeneous when examined from
either the perspective of a common problem or basic demographic charac-
teristics. However, when seen through the perspective of the stages of change
scale, the subjects divide into interesting subgroups. Moreover, these groups
appear to be coherent, consistent with theory, and related to problem di-
mensions and change variables.
T h e profiles established in this study are similar to those found in previous
studies of psychiatric outpatients. McConnaughy et al. (1983) enumerated 9
profiles which include all of the five found in this study. Absent from this
study were profiles labeled Pre-participation, Maintenance, Noncontemplative
Action and Nonreflective Action. However, these seemed to be profiles which
represented not different shapes but level differences across variables. Thus,
the five profiles in this study are consistent with the profiles found in previous
studies.
T h e Precontemplation group (#1) affirmed strongly that they do not have
a problem. This appears inconsistent with their coming to an outpatient,
voluntary alcoholism treatment program. These 63 subjects might not have
a serious problem. They report lowest levels of worry or anxiety about
drinking, least loss of control and role maladaptation, and the lowest scores
on the general alcoholism and alcoholic deterioration scales of the AUI.
However, they do not report drinking significantly less than the other groups
and they do report withdrawal symptoms at levels equal to or greater than
other groups. Clearly, it appears that the precontemplators do have a problem,
but these subjects may be denying the alcohol problem or feel they have the
problem under self-control. Support for the latter interpretation comes from
efficacy evaluations. These subjects have the lowest temptation scores (M = 2.9)
of all the groups. More important, they have a level of abstinence self-efficacy
(M = 3.0) slightly higher than their temptations to drink. T h e i r denial seems
consistent across all dimensions.
T h e Ambivalent group (#2) is similar to the Precontemplation group in
their strong affirmation that they have little or no problem. However, they
seem more ready to be involved in the change process. On the AUI, these
30 subjects are similar to the majority of the others on almost all measures.
T h e notable exception involves self-enhancing and mental benefit drinking.
This group seems to rely on alcohol for some important psychological func-
tions. This could explain the roots of their ambivalence. In spite of the
problems with alcohol as reflected in their AUI scores and their rather high
mean temptation score (M=3.3), the Ambivalent group has the highest level
of abstinence efficacy of all the groups (M=3.1). T h e relationship between
Stages of Change Profiles 233

their efficacy and temptation scores is totally random, reflecting that some
process other than an accurate self-evaluation is occurring.
T h e Participation group (#3) appears most committed to the process of
change. These 51 subjects admit to high levels o f loss of control, deterioration,
maladjustment, and compulsive drinking which contribute to high general
alcoholism scales. T h e y have sought help from others in the past more than
the other groups and have one of the highest levels of anxiety related to
drinking. All o f these support the conceptualization of this group as ready
for active participation in the treatment process. However, this group has
serious alcohol-related problems and could still be expected to have difficulty
in achieving abstinence. Temptation and confidence levels seem appropriate
for treatment motivation.
T h e Uninvolved group (#4) appears lethargic with respect to the process
of change. T h e y do not appear to be unique in their endorsements on the
AUI subscales. As with the Ambivalent group, the key to understanding this
group of 27 subjects lies in their efficacy scores. These subjects seem unin-
volved or lethargic because they appear overwhelmed by .their dependence
on alcohol. T h e y have the highest level of temptation to drink of any group
(M = 3.6). In contrast, they have the lowest level of abstinence self-efficacy
(M = 2.3). This discrepancy could create a feeling of being rather helpless
or hopelessly addicted.
T h e Contemplation group (#5) is most like the Participation group both
in their investment in considering change and in their AUI profiles. However,
they have not sought prior help for their problem as much as their Partic-
ipation group counterparts. Thus, the 53 subjects may be contemplators
particularly because they have less prior experience with attempts to change
their alcohol behavior. This is consistent with research on smoking cessation
which shows clear differences among contemplators depending on whether
they have ever attempted to quit smoking (DiClemente, Prochaska, & Gross-
man, 1986).
Limitations of this study should be noted. Most of the measures used in
this study are self-report measures. Thus, data are limited to a single domain
and are subject to all the problems associated with self-report. T h e sample
of subjects tends to be from a public institutional treatment program, so
generalizability may be limited. Results may not be as applicable to a middle-
class or upper-class treatment population. Relationship of the groups to
treatment process and outcome is not established.
Even with these limitations, the results are promising in that they open
up a new area of research on readiness to change as measured by the stage
of change scale. Possibilities for treatment matching and additional research
are intriguing and more promising than previous classification or subtyping
attempts (Morey & Blashfield, 1981; Sutker & Allain, 1988). In future research,
replication of the cluster profiles with a different sample should be followed
by a classification procedure to identify single subject's profile grouping. Both
immediate and long-term outcome measures with both subjective and objective
234 C. C. DiClemente and S. O. Hughes

criteria should then be used to examine outcome relevance o f the profile

groups.

REFERENCES

Bandura, A. (1982). Self-efficacy mechanism in human agency. American Ps~'chologist, 37, 122-147.
DiClemente, C.C. (1986). Self-efficacy and the addictive behaviors. Journal of Social & Clinical
Psychology, 4, 302-315.
DiClemente, C.C., Gordon, J.R., & Gibertini, M. (1983, August). Self-efficacy and determinants of
relapse in alcoholism treatment. Paper presented at 91st Annual Convention of-the American
Psychological Association, Anaheim, CA.
DiClemente, C.C., & Prochaska, J.O. (1982). Self-change and therapy change of smoking behavior:
A comparison of processes of change in cessation and maintenance. Addicth,e Behaviors, 7,
133-142.
DiC[emente, C.C., & Prochaska, J.O. (1985). Processes and stages of self-change: Coping and
competence in smoking behavior change. In S. Shiffman& T.A. Wills "(Eds.), Coping behavior
and drug use. San Diego, CA: Academic.
DiClemente, C.C., Prochaska, J.O., & Grossman, A.A. (1986, August). Contemplation and action:
Treatment matching stages of smoking cessation. Presentation at the Annual Meeting of the
American Psychological Association, Washington, DC.
Donovan, D.M., Kivlahan, D.R., & Walker, R.P. (1986). Alcoholic subtypes based on multiple
assessment domains: Validation against treatment outcome. In M. Galanter (Ed.), Recent
developments in alrohol#m (Vol. 4, pp. 207-222). New York: Plenum.
Horn, J.L., Wanberg, K.W., & Foster, F.M. (1987). Guide to the alcohol use inventor3". Minneapolis,
MN: National Computer Systems.
Keppel, G. (1982). Design analysis: A researcher's handbook (2nd ed.). Englewood Cliffs~ NJ: Prentice
Hall.
Marlatt, G.A., Baer, J.S., Donovan, D.M., & Kivlahan, D.R. (1988). Addictive behaviors: Etiology
and treatment. Annual Review of Psycholog3., 39, 223-252.
McConnaughy, E.A., DiClemente, C.C., Prochaska,J.O., & Velicer, W'.F. (1989). Stages of change
in psychotherapy: A followup report. Psychotherapy, 26, 494-503.
McC0nnaughy, E.A., Prochaska, J.O., & Velicer, W.F. (1983). Stages of change in psychotherapy:
Measurement and sample profiles. Ps)'chotherapy: Theor); Research and Practice, 20, 368-375.
McLellan, A.T., Luborsky, L., Woody, G.E., & O'Brien, C.P. (1980). An improved diagnostic
evaluation instrument for substance abuse patients: The Addiction Severity Index.Journal
of Nen,ous and Mental Disease, 168, 26-33.
Miller, W.R. (1985). Motivation for treatment: A review with special emphasis on alcoholism.
Psrchological Bulletin, 98, 84-107.
Miller, W.R., & Hester, R.R, (1986). Matching problem drinkers with optimal treatments. In
W.R. Miller & N. Heather (Eds.), Treating addictive behaviors: Process of change (pp. 175-203).
New York: Plenum.
Morey, L.C., & Blashfield, R.K. (1981). Empirical classification of alcoholism: A review. Journal
of Studies on Alcohol, 42, 925-937.
Morey, L.C., Skinner, H.A., & Blashfield, R.K. (1984). A typology of alcohol abusers: Correlates
and implications. Journal of Abnormal Ps~'cholog~., 93, 408-417.
Morris, R., Blashfield, R., & Satz, P. (1981). Neuropsycbology and cluster analysis. Journal of
Clinical Neurops)cholog3", 3, 79-99.
Prochaska, J.O., & DiC[emente, C.C. (1984). The transtheoretical approach: Crossing traditional
boundaries of therap)'. Homewood, IL: Dow Jones Irwin.
Prochaska, J.O., & DiClemente, C.C. (1986a). Toward a comprehensive model of change. In
W.R. Miller & N. Heather (Eds.), Treating addicth'e behaviors (pp. 3-27). New York: Plenum.
Prochaska, J.O., & DiClemente, C.C. (1986b). The transtheoretical approach: Towards a sys-
tematic eclectic framework. In J.C. Norcross (Ed.), Handbook of eclectic psychotherap)" New
York: Brunner/Mazel.
Stages of Change Profiles 235

Punj, G., & Stewart, D. (1085). Cluster analysis in marketing research: Review & suggestions
for application. Journal of Z~far~elitlg Research, 20, 134-148.
Schachter, S. (1982). Recidivism and self-cure of smoking and obesity. Americm, Psychologist, 37,
436-444.
Sutker, P.B., & Allain, A.W. (1988). Issues in personality conceptualizations of addictive behaviors.
Journal of Consulting and Clinical Psychology, 56, 172-182.
Wanberg, K.W., & Horn, J.C. (1983). Assessment of alcohol use with multidimensional concepts
and measures. American Psychologist, 58, 1055-1060.
Wanberg, R.W., Horn,J.L., & Foster, F.M. (1977). A differential assessment model foi- alcoholism:
The scales of alcohol use inventory. Journal of Studies on Alcohol, 38, 512-543.