Journal of Counseling Psychology Copyright 1996 by the American Psychological Association, Inc.

1996, Vol. 43, No. 4, 510-526 0022-0167/96/$3.00

A Taxonomy of Difficulties in Career Decision Making

I t a m a r Gaff and M i n a K r a u s z S a m u e l H. O s i p o w
Hebrew University of Jerusalem Ohio State University

A general theoretical taxonomy of career decision-making difficulties, based on decision

theory, has been developed. To examine the proposed taxonomy empirically, a questionnaire
was constructed in which the various possible difficulties in the theoretical model were
represented by respective statements. The questionnaire was administered to a sample of 259
young Israeli adults who were at the beginning of their career decision-making process and
to an American sample of 304 university students. The observed relations among the 10
scales, which represent the 10 theoretical categories of difficulties, and those among the items
within 2 selected categories, were similar in the 2 samples and compatible with the expected
relations derived from the theoretical model. The implications for career counseling and
research are discussed.

The growing rate of change in the world of work in-
creases the number of career transitions individuals make
during their lifetime. The quality of the career decisions
made during these transitions is significant for both the
individual and society. Making a career decision is, how-
ever, a complex process. Although some people make such
decisions easily, at least apparently, others face difficulties
in making their career decisions, and many seek profes-
sional help. One of the central aims of career counseling is
to facilitate the career decision-making process of counsel-
ees and, in particular, to help them overcome the difficulties
they encounter during this process. Therefore, identifying
the unique difficulties that prevent individuals from reach-
ing a decision is an essential step in providing them with the
help they need.
The construct of career indecision has been used to refer
to the problems individuals may have in making their career
decision (for a review, see Slaney, 1988). Because of its
recognized significance, career indecision is one of the
central research issues of career psychology (Betz, 1992;
Itamar Gati and Mina Krausz, Department of Psychology, He-
brew University of Jerusalem, Jerusalem, Israel; Samuel H. Osi-
pow, Department of Psychology, Ohio State University.
This research was supported by Grant 94-139 from the United
States-Israel Binational Science Foundation and carded out at the
National Council of Jewish Women Research Institute for Inno-
vation in Education of the Hebrew University. We thank Naomi
Fassa for valuable discussions; Nitza Shitz, Leah Hass, Tzachi
Ashkenazi, Hadas Benari, Mary Hill, and Hadassa Keren for their
help in data collection and analyses; and Daniela Ben-Dor, Naomi
Goldblum, Liat Kibari, Aluma Reiss, and Orit Trumper for their
comments on an earlier version of this article.
Correspondence concerning this article should be addressed to
Itamar Gati, Department of Psychology, Hebrew University,
Jerusalem, Israel 91905, or to Samuel H. Osipow, Department of
Psychology, Ohio State University, 1885 Neil Avenue Mall, Co-
lumbus, Ohio 43210. Electronic mail may be sent via Interuet to
Itamar Gati at msgati@pluto.mscc.huji.ac.il or to Samuel H. Osi-
pow at sosipow@magnus.acs.ohio-state.edu.
Fouad, 1994; Meier, 1991; Tinsley, 1992). Indeed, research
on career development and choice has devoted much atten-
tion to the categorization of various problems related to
indecision. However, some of these discussions have been
purely theoretical, without any empirical testing (e.g.,
Campbell & Cellini, 1981; Miller, 1971), whereas others
have had mainly an empirical focus in attempting to develop
various measures of career indecision. Furthermore, it ap-
pears that these two lines of research, the theoretical and the
empirical, have been conducted independently of one an-
other and by different groups of investigators.
In his introduction to the special issue on indecision in the
Journal of Vocational Behavior, Tinsley (1992, p. 211)
suggested that "efforts to relate research on career indeci-
sion to the decision making process and to develop a theo-
retical context for the construct of indecision will lead to
important advances in our research on these constructs."
The present article attempts to meet this challenge. Specif-
ically, the goal of this article is to present a new theoretical
model of difficulties in career decision making based on an
adaptation of decision theory to the context of career deci-
sions and to report an empirical test of this model in two
different samples.
Career Indecision
Various theoretical approaches have been used to deal
with career indecision, each emphasizing a different aspect.
For example, the psychodynamic approach (e.g., Bordin &
Kopplin, 1973) tried to classify the individual's problems
according to their internal sources instead of their observed
symptoms. The developmental approach (e.g., Osipow &
Fitzgerald, 1996; Super, 1953) suggested that the typical
problems faced during the decision process correspond to
the normal stages of career development, highlighting in
particular the concept of vocational maturity (Crites, 1978).
The vocational interests approach (e.g., Holland, 1985; Roe,
1956; Salomone, 1982) proposed that insufficient crystalli-

510
 DIFFICULTIES IN CAREER DECISION MAKING
zation of interests is among the major factors preventing the
individual from reaching a decision. Because each of these
theoretical approaches (as well as some others) has focused
on a single major facet of the process of career decision
making, their categorizations are limited to the particular
facet adopted and hence are neither comprehensive nor
inclusive (Rounds & Tinsley, 1984). Furthermore, in many
cases, the empirical tests of these theoretical approaches
have supported them only partially.
The empirical research on career indecision has focused
on developing various measures for examining individual
differences in career indecision. These measures include the
Career Decision Scale (CDS; Osipow, Carney, & Barak,
1976; Osipow, Carney, Winer, Yanico, & Koschier, 1987;
Osipow &Winer, 1996); the My Vocational Situation Scale
(MVS; Holland, Daiger, & Power, 1980); the Vocational
Decision Scale (VDS; Jones & Chenery, 1980); the Career
Decision Profile (CDP, a revision of the VDS; Jones, 1989);
the Behavioral Indecision Scale (BIS; Fuqua & Hartman,
1983); the Career Decision Making Self-Efficacy Scale
(CDMSE; Taylor & Betz, 1983); the Career Decision Di-
agnostic Assessment (CDDA; Bansberg & Sklare, 1986;
Larson, Busby, Wilson, Medora, & Allgood, 1994); the
Career Factors Inventory (CFI; Chartrand, Robbins, Morrill,
& Boggs, 1990); the Career Barriers Inventory (Swanson &
Tokar, 1991); and the Career Belief Inventory (CBI; Krum-
boltz, 1991, 1994). Most of the research involving these
measures has been carried out independently of theoretical
conceptualizations (Tinsley, 1992).
The state of affairs reflected in this concise review of the
theoretical and empirical research has led us to the conclu-
sion that a new framework, which combines a theoretical
analysis and empirical tests, may provide a unique and
significant contribution to research on career indecision.
Decision Theory and Career Decision M a k i n g
Decision theory, on which the proposed taxonomy of
difficulties relies, has been playing an increasingly impor-
tant role in understanding the processes involved in career
decision making (e.g., Brown, 1990; Gati, 1986; Jepsen &
Dilley, 1974; Katz, 1966; Mitchell & Krumboltz, 1984;
Neimeyer, 1988; Osipow, 1987; Osipow & Fitzgerald,
1996; Phillips, 1994; Pitz & Harren, 1980; Walsh & Osi-
pow, 1988). According to the normative theory of decision
making, the best decision is the one that best helps to
achieve the decision maker's goals. These goals are repre-
sented by the individual's preferences with respect to the
various attributes of the alternatives under consideration. A
rational career decision maker should choose the alternative
with the highest utility, where the utility of each alternative
is a function of the perceived gap between the individual's
preferences and the alternative's characteristics in each of
these attributes. Utility theory is a normative model that
may be regarded as a prescription for the best method for
making decisions (Baron, 1988; Brown, 1990).
Career decisions have the following features: There is an
individual who has to make a decision, there are a number
511
of alternatives to select from, and there are many attributes
or aspects that are considered in the comparison and eval-
uation of the various alternatives. In addition to these char-
acteristics, which are common to many decisions, career
decisions also have certain unique features. First, the num-
ber of potential alternatives is often fairly large (e.g., the
number of occupations, colleges, majors, or potential em-
ployers). Second, there is an extensive amount of informa-
tion available on each alternative. Third, a large number of
aspects (e.g., length of training, degree of independence,
type of relationship with people) is required to adequately
characterize the occupations and the individual's prefer-
ences in a detailed and meaningful way (e.g., Gaff, Garty, &
Fassa, 1996; Gati, Osipow, & Givon, 1995; Katz, 1993;
Lofquist & Dawis, 1978). Fourth, uncertainty plays a major
role with respect to both the individual's characteristics
(e.g., present and future preferences) and the nature of
future career alternatives (Gati, 1990; Gelatt, 1989).
Proposed T a x o n o m y
Assumptions Underlying the Proposed Taxonomy
Relying on decision theory, we first developed a model of
an "ideal career decision maker." This term refers to a
person who is aware of the need to make a career decision,
is willing to make it, and is capable of making the "right"
decision (i.e., a decision using an appropriate process and
most compatible with the individual's goals). The complex-
ity of the process of career decision making, as noted
earlier, makes it difficult for most people to be ideal career
decision makers. We define any deviation from the ideal
career decision maker as a potential problem that may affect
the individual's decision process in one of two possible
ways: (a) by preventing the individual from making a de-
cision or (b) by leading to a less than optimal decision.
The process of career decision making can be separated
into distinct components (e.g., Brown, 1990; Gati, Fassa, &
Houminer, 1995; Katz, 1966; Pitz & Harren, 1980), each of
which presumably involves different kinds of difficulties.
Therefore, the various possible difficulties the individual
may face during the process of career decision making can
be classified into distinct categories such that difficulties
with common features are included in the same category
(Campbell & Cellini, 1981). Specifically, the classification
of the difficulties into categories was based on the following
criteria: (a) belonging to the same stage or component of the
process of career decision making; (b) having the same
assumed source; (c) having similarity in the hypothesized
possible impact of the difficulty (i.e., halting the process or
leading to a less than optimal decision); and (d) having
similarity in the type of intervention needed to overcome it.
Furthermore, we assume that the individual may have a
single difficulty or a combination of them (Campbell &
Heffernan, 1983) and that these difficulties may be located
in one or several categories. However, we expect that the
difficulties within each category will co-occur more often
than those from different categories. On the other hand,
512 GATI, KRAUSZ, AND OSIPOW
problems from different categories can also be associated in
certain cases because, as Campbell and Heffernan (1983)
pointed out, problem categories cannot and should not be
completely independent of each other.
Features of the Proposed Taxonomy
The proposed taxonomy was developed through interac-
tion and sequential interplay between theoretical consider-
ations and empirical testing. Specifically, to ensure the
comprehensiveness of the proposed model, as well as its
relevance for real-life contexts, we elicited descriptions of
career decision difficulties from 200 career counselees and
10 expert career counseling psychologists. We compared
these lists with the theoretical model to ensure that it in-
cluded all important and relevant difficulties and introduced
a few changes in the set of difficulties included in the
categories.
The proposed taxonomy itself is hierarchic (Fleishman &
Quaintance, 1984), in which broad categories of difficulties
are separated into categories and then subcategories based
on finer distinctions. Accordingly, during the diagnostic
process each difficulty of an individual can be classified as
to its major category and then into finer categories and
subcategories (Campbell & Heffernan, 1983). Specifically,
in the proposed taxonomy the difficulties are divided into
three major categories. The first, Lack of Readiness, in-
cludes four categories of difficulties that precede the en-
gagement in making a specific career decision. Each of the
other two major categories, Lack of Information and Incon-
sistent Information, includes three categories of difficulties
that arise during the actual process of career decision mak-
ing. The three major categories of difficulties and the 10
CAREER DECISION HAKIN6 DIFFICULTIES
1
PRIOR TO BEGINNING THE PROCESS
I I
LACK OF
READZNESS
DUE TO
LACK OF INDECI- DYSFUNCTIONAL LACK OF SELF
MOTIVATION SIVENESS MYTHS KNOWLEDGE
ABOUT PROCESS
(STEPS)
Figure 1. The initial theoretical taxonomy of career decision-making difficulties.
specific categories of the proposed taxonomy are summa-
rized in Figure 1.
The 10 Difficulty Categories
Within the major category of Lack of Readiness, a dis-
tinction was made between two pairs of categories. One pair
includes difficulties that are related to lack of motivation to
engage in the career decision process and general indeci-
siveness concerning all types of decision making. The other
pair includes difficulties related to dysfunctional myths
(e.g., irrational expectations) about the process of career
decision making and lack of knowledge about the steps
involved in this process.
Within the major category of Lack of Information, a
distinction was made among three categories: lack of infor-
mation about self, lack of information about occupations,
and lack of information about ways of obtaining additional
information. We assume that, because the last two catego-
ries refer to external, objective information, they are more
closely related to each other than to the first one (which
involves the individual).
Within the major category of Inconsistent Information, a
distinction was made among three categories: unreliable
information, which includes difficulties related to unreliable
or fuzzy information; internal conflicts, which include con-
flicts within the individual; and external conflicts, which
include conflicts involving the influence of significant oth-
ers. Here too, we assume that the last two categories of
conflict are more closely related to each other than to the
first.
DURING THE PROCESS
I
J I
LACK OF INCONSISTENT
INFORMATION INFORMATION
ABOUT DUE TO
r I
0CCUPATZONS WAYS OF UNRELZABLE INTERNAL EXTERNAL
OBTAINING INFORMATZON CONFLICTS CONFLICTS
INFOR/4ATZON
 DIFFICULTIES IN CAREER DECISION MAKING
Specific Difficulties Within Categories
The elaborated theoretical model, which includes 44 spe-
cific difficulties representing the various categories (and
subcategories), is presented in the Appendix. Each of these
44 specific difficulties represents a distinct type of problem,
where these distinctions are based on both theoretical con-
siderations and apparent practical significance. It is impor-
tant to note that the degree of differentiation is not equal
across the 10 categories. Rather, the six categories associ-
ated with the difficulties that arise during the engagement in
the process were refined by a systematic division into sub-
categories, whereas the four categories included in the ma-
jor category of Lack of Readiness were not (for lack of
theoretical rationale or practical need).
Specifically, within the category of lack of information
about self, a distinction was made between lack of infor-
mation regarding preferences ("What do I want?") and lack
of information regarding the individual's perceived capabil-
ities ("What can I do?"). Each of these two subcategories
was further divided into difficulties associated with lack of
information at present and difficulties associated with un-
certainty about the future. Similarly, within the category of
lack of information about occupations, a distinction was
made between lack of information about the very existence
of occupational alternatives and lack of information about
the characteristics of the alternatives. Each of these two
subcategories was further divided into lack of information at
present and lack of information about the future. Within the
category of lack of information about ways of obtaining
additional information, a distinction was made between
ways of obtaining information about the self (e.g., career
counseling services) and ways of obtaining information
about the occupations (e.g., career information centers).
Within the category of unreliable information, a distinc-
tion was made among three subcategories, according to the
different kinds of unreliable information. Such unreliability
may be related to (a) the individual's preferences, (b) his or
her perceived capabilities, or (c) the occupational alterna-
fives regarded as relevant. Within the category of internal
conflicts, a distinction was made between conflicts associ-
ated with incompatible preferences and those arising from
the gap between preferences and capabilities. The first sub-
category was further divided into conflicts among preferred
aspects, conflicts among occupations, and conflicts between
an occupation and a desirable aspect (e.g., between the
desire to be a nursery school teacher and to have a very high
income). The second subcategory was further divided into
insufficient abilities and abilities higher than required. Fi-
nally, within the category of external conflicts, a distinction
was made between conflicts among external sources (e.g.,
between two significant others) and conflicts between the
individual and an external source. Each of these subcatego-
ries of conflicts was further divided into conflicts expressed
in terms of relevant aspects (i.e., what criteria or factors to
take into consideration in the decision) or in terms of
alternatives (i.e., what careers the individual should
choose).
Note that important distinctions that underlie decision
5 13
theory are represented explicitly in the proposed taxonomy.
These distinctions are (a) between alternatives (i.e., careers)
and the characteristics (i.e., aspects) that can be used to
compare and evaluate them; (b) between specific missing
information and the possible ways of obtaining that infor-
mation; (c) between preferences and capabilities; and (d)
between lack of information about the present and lack of
information about the future (explicitly representing the
uncertainty involved).
Implementation Issues
We did not include in the proposed taxonomy problems
related to the implementation of the decision made because
we believe that such problems typically have antecedents in
the process of career decision making. Thus, lack of imple-
mentation is assumed to stem from the fact that the process
by which the decision was reached was problematic, and
certain difficulties were not dealt with adequately. For ex-
ample, lack of means to realize a plan may stem from the
fact that the issue of financial resources to implement the
decision was not dealt with adequately; lack of information
about how to realize a preferred alternative may stem from
lack of information about that alternative or additional
sources of information; delaying the implementation of a
preferred alternative may stem from lack of confidence in
the process of career decision making, which, in turn, may
be associated with lack of knowledge about the steps and
characteristics of this process.
Goals o f This Research
In this research we tested the hypothesis that the empirical
structure of career decision difficulties conforms to the
theoretical model. However, because of the complexity of
the proposed model and its multilevel nature, we decided to
separate the test into two parts. First, we examined whether
the empirical relations among the 10 categories conform to
their theoretical structure. That is, we examined whether the
10 categories of difficulties are divided into the hypothe-
sized three major categories (Lack of Readiness, Lack of
Information, and Inconsistent Information) and whether the
structure within each major category conforms to the ex-
pected pattern of relations among them. Second, we exam-
ined whether the structures of difficulties within two of the
categories--lack of information about the self and incon-
sistent information--conform to their hypothesized internal
structure. These two categories were selected for three rea-
sons: (a) We believe that dealing with difficulties in these
two categories is significant for the career counseling pro-
cess; (b) their internal structure is systematic and well
defined; (c) the number of difficulties included in each of
these categories (8 and 6, respectively) makes such a test
possible (when only 3 difficulties are included in a category,
its internal structure cannot vary much).
To conduct these tests, we developed a questionnaire that
includes 44 statements corresponding to the 44 difficulties
in career decision making included in the theoretical model
514 GATI, KRAUSZ, AND OSIPOW
and defined 10 scales corresponding to the 10 categories.
This questionnaire was administered in two different cross-
cultural contexts: an Israeli sample o f y o u n g adults w h o
w e r e at the b e g i n n i n g o f their process of career decision
m a k i n g and an A m e r i c a n sample o f university students. The
use o f two culturally distinct samples enabled us to e x a m i n e
the generality and stability o f the p r o p o s e d t a x o n o m y . A f t e r
a discussion o f the results o f these two samples, the impli-
cations for theory, research, and the career counseling prac-
tice are explored.
Method
The questionnaire used in this study was developed especially
for the purpose of testing the proposed taxonomy. Hence, before
using it to test the theoretical taxonomy, we had to carry out a
number of steps to ensure that it adequately represents the model
and that its psychometric properties are satisfactory. We begin by
briefly describing the development of the questionnaire, then turn
to our main goal of testing the model itself. The initial construc-
tions of the questionnaire as well as its pretests were carried out in
Israel (in Hebrew), and only the last version was administered in
both Israel and the United States (after translation into English).
D e v e l o p m e n t o f the Career Decision Difficulties
Questionnaire ( C D D Q )
Initial construction of the questionnaire. On the basis of the
theoretical taxonomy, we constructed a set of items, each of which
corresponds to one of the 44 difficulties in the theoretical model.
We decided to use only one representative item for each difficulty
in order to prevent cognitive overload and motivational decrease
(due to an overly long questionnaire). Then, to test the correspon-
dence between the items in the questionnaire and the difficulties in
the theoretical model, we asked five experienced career counseling
psychologists (3 MA and 2 PhD) to match each of the items in the
questionnaire with one particular difficulty from those included in
the theoretical model. Mismatched items were revised, reworded,
or replaced. Next, on the basis of the proposed 10 categories of
difficulties, we defined 10 scales, each of which is based on the
particular set of items representing the set of difficulties included
in that category.
Revisions of the questionnaire on the basis of empirical data.
We asked 200 young adults, ages 20 to 22 years, who were at the
end of their obligatory military service and thus about to make
their first major career decision, to complete the questionnaire.
Their responses were analyzed to ensure that the psychometric
properties of the questionnaire were satisfactory. Specifically, we
first computed the mean and standard deviation of the responses
for each item to ensure differentiation. Second, we computed the
score of each scale, where this score was defined as the mean of
the responses to the items included in that scale. Then we com-
puted the item-scale correlation (excluding the item itself) and the
item-other scales correlations (i.e., the correlations between the
item and the scores of the scales to which the item does not
belong). Problematic items were located using the criteria pro-
posed by Meir and Gati (1981). Third, we carried out a cluster
analysis of the intercorrelation matrix among the responses to the
44 items. Fourth, we assessed the scale reliabilities (Cronbach's
alpha), as well as the reliability of the complete questionnaire.
Finally, using retest data from the same group of participants, we
examined the test-retest reliability of the items, the scales, and the
whole questionnaire. Some of the items were revised on the basis
of these analyses.
Using the revised questionnaire, we collected additional data
from another group of approximately 200 young adults whose
characteristics were similar to those of the first group. We then
analyzed their responses to the revised questionnaire in the same
manner and made the necessary revisions in the questionnaire.
This process was repeated four times (with different versions of
the questionnaire and different groups of about 200 young adults)
until the psychometric properties of the questionnaire were found
satisfactory and its internal structure had been consolidated.
Last R o u n d o f Data Collection
Participants. Two different cross-cultural samples participated
in the research: an Israeli sample and an American sample. The
Israeli sample consisted of 147 men and 112 women, ages 19 to 23
years (Mdn and mode = 21, M = 20.68, SD = 0.74), who were
nearing the end of their military service. Ninety-eight percent of
the participants had 12 years of schooling. The questionnaires of
two older participants (25 and 28 years) were not included in the
analyses. The American sample included 304 students (186 men
and 105 women; 13 did not indicate gender) at a large midwestern
university in the United States. The age range was 17 to 23 years
(Mdn and mode = 18, M = 18.70, SD = 1.10). Most participants
(74%) were freshmen, 18% were sophomores, and 6% were jun-
iors. Within the population from which the sample was drawn, the
majority are Caucasian (71%); other ethnic groups include Asian
Americans (15%), African Americans (6%), Hispanic Americans
(1%), Native Americans (1%), and others (those of mixed race and
non-Americans; 5%).
CDDQ. The first page of the questionnaire included general
background information: age, sex, years of education, present job,
the degree to which the participant is considering a career decision
(very, little, not at all), whether the participant has some career
alternative in mind, and his or her degree of confidence in that
alternative (on a 9-point scale). The following pages included 44
items, each one corresponding to a particular difficulty. The par-
ticipants were asked to rate (on a 9-point scale) the degree to which
the difficulty represented by each item (e.g., "It is usually difficult
for me to make a decision") described them (from 1 = does not
describe me to 9 = describes me well). Finally, the participants
were asked to rate the overall severity of their difficulties in
making a career decision (from 1 = not severe at all to 9 = very
severe) and to list additional difficulties preventing them from
making a career decision.
Because the research questionnaire was developed in Israel (in
Hebrew), it had to be translated into English for the American
sample. The questionnaire was first translated in Israel by a native
Hebrew speaker and then examined by three other native Hebrew
speakers; all four are bilingual. Next, the translation was examined
by a native English speaker who lives in Israel and after 20 years
may be regarded as bilingual as well. Finally, the English version
of the questionnaire was edited in the United States by a native
English speaker.
Procedure. The Israeli participants responded to the question-
naire at the beginning of a 3-day workshop called "Orientation to
Civilian Life," organized by the Israeli Veterans Administration.
In responding to the questionnaire, participants gave their consent
to use their responses in the research. Fewer than 5% did not return
the questionnaire. To allow us to assess test-retest reliability, we
asked the participants to complete the questionnaire again on the
third day of the workshop. Because at the end of the workshop the
soldiers would return to their original unit and could not be
 DIFFICULTIES IN CAREER DECISION MAKING 515

reached, we had only this short interval for the test-retest esti-
mates. After responding the second time (n = 164), the partici-
pants were given an orientation lecture geared toward making
career decisions and were given an explanation for being asked to
fill in the questionnaire twice. The American participants re-
sponded to the questionnaire in groups of 15 to 25 students, as part
of the course requirement in introductory psychology. All partic-
ipants returned their questionnaires.
Analyses. First, the psychometric properties of the question-
naires were examined, focusing in particular on the scales' means,
standard deviations, scale intercorrelations, Cronbach alpha, and
test-retest reliability. The cross-cultural comparison made it pos-
sible to test the psychometric equivalence of the questionnaires
(Fouad, 1993). Second, to derive the empirical structure of the 10
scales and the empirical internal structure of the two selected
categories, we used cluster analysis. Cluster analysis is used to
partition the stimuli of a given data set into relatively distinct
groups. One of its prevalent uses is to test a particular psycholog-
ical classification on the basis of prior knowledge or theory (Bor-
gen & Barnett, 1987).
In this study we used a clustering algorithm called ADDTREE
(Sattath & Tversky, 1977). ADDTREE is especially attractive
because it represents the proximity matrix in the form of an
additive or "path length" tree, in which the variables are divided
into clusters and subclusters according to the proximity between
them (as reflected in the degree of correlation between them).
Thus, the clustering analysis by ADDTREE of the intercorrelations
among the 10 scales and the intercorrelations among the difficul-
ties within the two selected categories allowed us to directly
compare the empirical structures with the hypothesized theoretical
structures. Another advantage of ADDTREE is that, unlike many
other clustering programs available as part of standard statistical
packages (e.g., SAS, SAS Institute, 1982; SPSS x, SPSS, Inc.,
1986), no decisions about the method or technique of clustering are
needed (i.e., single linkage, complete linkage, average linkage, or
Ward's, 1963, minimum variance technique). For additional de-
tails on ADDTREE, see Sattath and Tversky (1977); for the results
of its applications to a variety of datasets, see Tversky and
Hutchinson (1986).
Results
Because there were no systematic or meaningful differ-
ences between men and women in the scale scores, and the
pattern of correlations among the scales was highly similar
for men and women (Spearman's r = .89 and .90 for the
Israeli and the American samples), we report only the ag-
gregated results.
Psychometric Properties of the Questionnaire
Israeli sample. The means, standard deviations, and re-
liabilities of the scale scores in the Israeli sample are pre-
sented in the left-hand side of Table 1. The table also
presents the number of items per scale. The intercorrelations
among the scale scores in the Israeli sample are presented in
Table 2, above the diagonal. During the initial phases of
these analyses (as well as in those of the previous versions
of the questionnaire) we found that, unlike the structure in
Figure 1, the category of lack of knowledge about the
process was much closer to the three categories in the major
category of Lack of Information than to the three categories
in the major category of Lack of Readiness. Hence, in the
following analyses this category was regarded as belonging
to the difficulties of the major category of Lack of Infor-
marion. As can be seen in Table 1, the standard deviations
of the scale scores show an acceptable variability, and the
correlations among the scale scores (Mdn = .37 and range
= .06-.78) indicate an adequate differentiation among
them.

Table 1
Means, Standard Deviations, and Reliabilities of the Scales in the Israeli (n = 259)
and American Samples (n = 304)
American
Israeli sample sample
Number
Scale of items M SD a Retest M SD a
Lack of Readiness
Lack of Motivation 3 2.80 1.76 .66 .62 2.72 1.52 .53
Indecisiveness 4 3.80 1.85 .76 .72 4.52 1.73 .69
Dysfunctional Myths 3 4.52 1.51 .29 .51 5.11 1.57 .40
Lack of Information
Lack of Knowledge About the Process 3 4.04 2.28 .86 .64 3.63 2.08 .86
Lack of Information About Self 8 3.44 1.79 .87 .68 3.51 1.88 .91
Lack of Information About Occupations 4 4.62 2.37 .90 .62 3.96 2.15 .88
Lack of Information About Ways of
Obtaining Additional Information 2 3.52 2.19 .67 .50 3.22 1.87 .66
Inconsistent Information
Unreliable Information 6 3.12 1.68 .78 .67 3.34 1.72 .79
Internal Conflicts 7 3.42 1.69 .79 .66 3.57 1.59 .75
External Conflicts 4 2.28 1.75 .88 .67 2.64 1.90 .88
Lack of Readiness 10 3.71 1.23 .70 .67 4.16 1.02 .63
Lack of Information 17 3.84 1.78 .93 .74 3.60 1.78 .95
Inconsistent Information 17 3.04 1.53 .91 .72 3.27 1.45 .89
Total 44 3.49 1.36 .95 .80 3.60 1.32 .95
516 GATI, KRAUSZ, AND OSIPOW

Table 2
lntercorrelations Among the CDDQ Scales in the Israeli (n = 259; Above DiagOnal) and American Samples
(n = 304; Below Diagonal) '
Inconsistent
Lack of Readiness Lack of Information Information Clusters
Scale 1 2 3 4 5 6 7 8 9 10 11 12 13 14
1. Rm - - .34 .11 .27 .42 .23 .22 .38 .42 .37 .68 .37 .44 .52
2. Ri .29 -- .22 .46 .51 .28 .37 .35 .35 .28 .83 .49 .38 .58
3. Rd - . 0 5 .23 -- .16 .18 .06 .18 .14 .15 .14 .55 .17 .16 .28
4. Lp .46 .48 .05 -- .69 .63 .60 .55 .56 .36 .46 .84 .58 .77
5. Ls .45 .50 .09 .76 -- .63 .57 .69 .65 .48 .56 .91 .70 .88
6. Lo .41 .41 .02 .70 .73 -- .69 .55 .55 .32 .29 .86 .55 .74
7. La .28 .39 .01 .66 .66 .74 -- .53 .54 .37 .38 .77 .56 .72
8. Iu .40 .52 .08 .64 .76 .64 .65 -- .78 .62 .42 .70 .91 .84
9. Ii .46 .47 .10 .63 .68 .64 .59 .73 -- .67 .45 .68 .94 .86
10. Ie .32 .34 .08 .33 .38 .38 .35 .50 .44 -- .38 .46 .82 .67

11. R .57 .84 .55 .51 .54 .43 .36 .52 .53 .38 - -.52 .47 .67
12. L .47 .51 .06 .86 .95 .89 .80 .77 .72 .41 .54 -- .72 .92
13. I .47 .53 .11 .65 .74 .67 .65 .90 .89 .72 .57 .78 -- .90
14. Total .54 .64 .18 .81 .90 .82 .75 .87 .84 .58 .70 .94 .92 --
Note. CDDQ = Career Decision Difficulties Questionnaire; Rm = Lack of Motivation scale; Ri = Indecisiveness scale; Rd =
Dysfunctional Myths scale; Lp = Lack of Knowledge About the Process scale; Ls = Lack of Information About Self scale; Lo = Lack
of Information About Occupations scale; La = Lack of Information About Ways of Obtaining Additional Information scale; Iu =
Unreliable Information scale; Ii = Internal Conflicts scale; Ie =Extemal Conflicts scale; R = Lack of Readiness cluster; L = Lack of
Information cluster; I = Inconsistent Information cluster.

The reliability (Cronbach alpha) of the scales varies sig-
nificantly, as can be seen in Table 1. The reliability of the
scale of Dysfunctional Myths (Rd) is very low, revealing
that a positive endorsement of a particular dysfunctional
myth is not associated with the endorsement of other myths.
This finding was consistent across the various versions of
the questionnaire and replicates those of Gati (1996b). Other
scales had moderate-to-high reliabilities, ranging from .66
for the scale of Lack of Motivation to .90 for Lack of
Information About Occupations (the median scale reliability
was .78). The low reliability of Dysfunctional Myths con-
tributed to the relatively low reliability of the major cate-
gory of Lack of Readiness (.70); the reliabilities of the two
other major categories were much higher: Lack of Informa-
tion = .93, and Inconsistent Information = .91. The reli-
ability of the whole questionnaire was .95. The pattern of
test-retest reliabilities was generally similar to that of Cron-
bach alpha, although the reliabilities were lower and the
differences among the scales smaller (see Table 1).
American sample. The means, standard deviations, and
reliabilities of the scale scores in the American sample are
presented on the right-hand side of Table 1. The intercor-
relations among the scale scores in the American sample are
presented in Table 2 below the diagonal. As in the Israeli
sample, the standard deviations of the scale scores of the
American sample reflect an acceptable variability, and the
correlations among the scales (Mdn = .46, range = - . 0 5 -
.76) indicate an adequate differentiation among them.
The reliability of the scales varied significantly and in the
same pattern as that of the Israeli sample: from .40 for the
scale of Dysfunctional Myths to .91 for the scale of Lack of
Information About Self (the median scale reliability was
.77). As with the Israeli sample, the reliability of the major
category of Lack of Readiness was the lowest, .63 (which,
again, can be partially attributed to the unreliability of the
Dysfunctional Myths scale), the reliability of the major
category of L a c k of Information was .95, and that of the
Inconsistent Information was .89. The reliability of the
whole questionnaire was .95.
The similarity between the Cronbach alpha reliabilities
obtained in the two samples is reflected in a Spearman
rank-order correlation of .94 for the 10 scales. This simi-
larity indicates that the variance in reliabilities is inherent to
the scales and not especially associated with the different
procedures and settings in which the two data sets were
collected. Furthermore, the rank order of the correlations
between the 10 scales in the two studies was .90 (which is
reflected in the similarity of Figures 2 and 3), indicating a
stable pattern of relations among them across cultures. Fi-
nally, the rank-order correlation between the scale means
was .85, reflecting the similarity in the relative severity of
the difficulties in the two samples.
Empirical Structure of the 10 Scales
Israeli sample. Figure 2 presents the empirical structure
of the 10 scales obtained by an A D D T R E E analysis of the
intercorrelation matrix shown in Table 2. The clustering
structure in Figure 2 adequately summarized the empirical
relations among the scales: The linearly accounted-for vari-
ance by the distances in the clustering structures was 94%.
The distance between any pair of scales in this clustering
structure is represented by the sum of the horizontal seg-
ments on the shortest path connecting them. Thus, scales
within the same cluster are generally more closely related
than those belonging to different clusters.
 DIFFICULTIES IN CAREER DECISION MAKING 517

........................................ Motivation
R
.................................. Indecisiveness
..... i
........................................................... Dysfunctional myths
.......... Lack o f information about the s e l f

....
Lol ............... Lack o f knowledge about the process

..................... Lack o f information about occupations

........ i
. . . . . . . . . . . . . . . . . Lack o f information about ways o f o b t a i n i n g additional
information
.......... Unreliable information
II
................. Internal conflicts
--I
........................ External c o n f l i c t s

Figure 2. The empirical structure of the 10 difficulty categories--Israeli sample (n = 259). R =

Readiness; LoI = Lack of Information; II = Inconsistent Information.

As can be seen in Figure 2, the 10 scales are grouped into
three clusters corresponding to the three major categories:
Lack of Readiness, Lack of Information, and Inconsistent
Information. Furthermore, there is a distinction between the
first category, which includes difficulties arising prior to
the beginning of the process of career decision making, and
the last two categories, which include difficulties arising
during the decision-making process. The cluster of Lack of
Readiness includes, as expected, the scales of Lack of
Motivation, Indecisiveness, and Dysfunctional Myths, al-
though in a different internal arrangement. However, the
scale of Lack of Knowledge About the Process joined the
cluster of Lack of Information instead of the cluster of Lack
of Readiness. As stated, this finding also emerged in the
previous versions of the questionnaire. The cluster of Lack
of Information includes, in addition to this scale, the scales
of Lack of Information About Self, Lack of Information
About Occupations, and Lack of Information About Addi-
tional Sources of Information, as suggested by the theoret-
ical model. The third cluster, Inconsistent Information, in-
cludes, as hypothesized, the scales of Unreliable
Information, Internal Conflicts, and External Conflicts.
American sample. Figure 3 presents the empirical struc-
ture of the 10 scales, obtained by the ADDTREE analysis of
the intercorrelation matrix shown in Table 2. The clustering
structure in Figure 3 adequately summarizes the empirical
relations among the scales: The linearly accounted-for vari-
ance of the data by the distances in the clustering structure
is 98%.
As can be seen from Figure 3, the empirical structure of
the 10 scales in this sample is similar to that in the Israeli
sample. First, as for the Israeli sample, there is a distinction
between the difficulties that arise prior to beginning the
process of career decision making and those arising during
this process. Second, the cluster of Lack of Readiness
includes the scales of Lack of Motivation, Indecisiveness,
and Dysfunctional Myths, and, again, the last two are paired
together. Third, the cluster of Lack of Information includes
the same four scales as in the Israeli sample, with a similar
internal structure. Specifically, Lack of Information About
Occupations and Lack of Information About Ways of Ob-
taining Additional Information are paired with each other
and separated from the other two scales--Lack of Knowl-
edge About the Process and Lack of Information About
Self. It is important to note that Lack of Knowledge About
the Process is located in the cluster of Lack of Information
rather than in the Lack of Readiness cluster, just as in the
Israeli sample; in this both samples differ from the theoret-

............................... Motivation

_R__ - ............. Indecisiveness

....................................................... Dysfunctional myths

........................... External conflicts

..... Lack of information about the self

___L°I_ - ........ Lack of knowledge about the process

m m . . . . . Lack of information about occupations

/
. . . . . . . . . . Lack of information about ways of obtaining additional information
....... Unreliable information
--I
II ......... Internal conflicts

Figure 3. The empirical structure of the 10 difficulty categories--American sample (n = 304).

R = Readiness; LoI = Lack of Information; II = Inconsistent Information.
518 GATI, KRAUSZ, AND OSIPOW

ical model. Finally, the scales of Unreliable Information and
Internal Conflicts are grouped in the major cluster of Incon-
sistent Information. However, in contrast to what we ex-
pected, the scale of External Conflicts is located not in this
major cluster, as in the Israeli sample, but rather in the Lack
of Readiness cluster.
Empirical Structure of Difficulties Within Categories
Israeli sample. Figure 4a presents the empirical struc-
ture of the eight difficulties within the category of lack of
information about self, obtained by an ADDTREE analysis
of the intercorrelations between the responses to the eight
items representing the eight difficulties related to this cate-
gory in the Israeli sample. As expected, the difficulties in
this category were grouped into two subclusters correspond-
ing to the two subcategories: lack of information about
preferences ("What do I want?") and lack of information
about capabilities ("What can I do?"). In each of the two
clusters, the two difficulties representing lack of informa-
tion at present are found together (Difficulties 16 and 17,
and 14 and 15, respectively; see Appendix), as proposed in
the theoretical model. However, the respective two difficul-
ties concerning the future constitute a well-defined pair only
in the cluster of preferences (Difficulties 20 and 21).
Figure 4b presents the empirical structure of the six
difficulties related to the category of unreliable information
in the Israeli sample. Compatible with the theoretical model,
the difficulties in this category were grouped into three
pairs: (a) inconsistency about preferences (about desirable
aspects, or preferred careers), (b) inconsistent information
about capabilities (abilities and personality traits), and (c)
inconsistent information about the occupational alternatives
(what careers there are and what they are like).
American sample. Figure 5a presents the empirical
structure of the eight difficulties within the category of lack
of information about self in the American sample. The
empirical structure supports the distinction between lack of
information about preferences ("What do I want?") and lack
of information about capabilities ("What can I do?"). Fur-
thermore, in each of these clusters there is a pair of diffi-
culties representing lack of information at present (Difficul-
ties 16 and 17, and 14 and 15, respectively; see the
Appendix) and a pair of difficulties related to lack of infor-
mation concerning the future (Difficulties 20 and 21, and 18
and 19, respectively). Thus, the internal structures of lack of
information about self in both studies are highly similar, but
only that of the American sample is fully compatible with
the theoretically expected structure.
Figure 5b presents the empirical structure of the six
difficulties associated with unreliable information in the
American sample. The structure reveals three pairs of dif-
ficulties: (a) unreliable preferences, Co) unreliable informa-
tion about capabilities, and (c) unreliable information about
career alternatives. This structure is nearly identical to that

a. Within the category o f lack o f information about the s e l f

Present .................................. Abilities

............. i
~-I .......... Future ................................................. Pe~ity
.................................. Abilities
. Future Personality t r a i t s

Present ................................... Occupations

Pr i ............... ! ............................. Aspects

I Future ........... Occupations

........................ i
......... Aspects

b. Within the category o f qnr~liable information

Cap ............................................................... Abilities

......... i
................................. Personali t y t r a i t s

Pr ........... Occupations
...................................... i
.............. Aspects

Occ .................. Existence

......................................................
.................. Characteristics

Figure 4. The structure of difficulties within categories--Israeli sample. Cap = Capabilities;

Pr = Preferences; Occ = Occupations.
 DIFFICULTIES IN CAREER DECISION MAKING 519

a. Within the category o f lack o f informatiQo ~bo~t the s e l f

Present ............................................. Abilities

. . . . . . . . . . . . . . . . . . . . . . . i
___C)p______ - ................................... Personality traits

Future ................................ Abilities

........ i
............................................ Personality traits

Present .................................. Occupations

............. i
___Pr. . . . . . . -................................ Aspects

Future ............ Occupations

.................... i
....... Aspects

b. Within the category o f unreliable information

Cap ................................. Abilities

............................ i
............................. Personality t r a i t s

Pr --- Occupations
..................................... i
...... Aspects

Occ .................. Existence

.............. i
...... Characteristics

Figure 5. The structure of difficulties within categories--American sample. Cap = Capabilities;

Pr = Preferences; Occ = Occupations.

obtained in the Israeli sample data and compatible with the divided into more refined subcategories; these subcategories
expected theoretical structure. and the 4 other categories were broken down into 44 spe-
cific difficulties, each representing a distinct type of prob-
Discussion lem.
To test the proposed taxonomy, we constructed the
The central aim of this research was to develop and CDDQ, in which each of the 44 difficulties was represented
empirically test a taxonomy of difficulties in career decision by one respective item. On the basis of the theoretical
making. In the initially proposed taxonomy, outlined in model, 10 scales were defined for the questionnaire, each
Figure 1 and elaborated in the Appendix, the 44 difficulties including the set of items corresponding to the specific
were expressed in terms of concepts adopted from decision difficulties constituting the respective category. This ques-
theory and then adapted to the area of career decisions. The tionnaire was then administered to two samples from dif-
basic distinction in the proposed taxonomy was between ferent social-cultural contexts: an Israeli sample of young
difficulties arising prior to the beginning of the process of adults at the beginning of their process of career decision
career decision making and those arising during this pro- making and a sample of American university students. Two
cess. The major category of difficulties that arise prior to the levels of the proposed taxonomy were tested: the relations
beginning of the process, which was defined as Lack of among the 10 categories and the relations among specific
Readiness, included four categories: lack of motivation, difficulties that compose two selected categories.
indecisiveness, dysfunctional myths, and lack of knowledge The empirical pattern of relationships among the 10
about the process. The difficulties arising during the process scales, which represent the 10 difficulty categories in the
were divided into two major categories: Lack of Informa- theoretical model, was generally similar to the hypothesized
tion, which included three categories (lack of information pattern in both the Israeli and the American samples, with
about self, lack of information about occupations, and lack one clear deviation involving the scale of Lack of Knowl-
of information about ways of obtaining additional informa- edge About the Process. In the proposed taxonomy this
tion) and Inconsistent Information, which also included category was located in the major category of Lack of
three categories (unreliable information, external conflicts, Readiness because we assumed that this type of difficulty
and internal conflicts). Six of the 10 categories were further arises before the individual needs to make a particular
520 GATI, KRAUSZ, AND OSIPOW
career decision and can be prevented by appropriate prep-
aration. Empirically, however, this category was consis-
tently included in the cluster of Lack of Information in both
samples. Two reasons may account for this deviation from
the theoretical model. First, although knowledge of the steps
involved in career decision making may be acquired prior to
beginning the process of career decision making, one need
not apply this knowledge until the process has actually
begun. Accordingly, the participants probably felt a need for
knowledge of the steps involved in the process only when
they actually started the process of making a particular
decision and not before. Second, this knowledge has a
cognitive basis (such as lack of information about the self or
occupations), whereas the other categories in the cluster of
Lack of Readiness are more emotionally based. In light of
this consistent and reasonable pattern of relations, we intro-
duced the relevant changes in the theoretical taxonomy; the
revised taxonomy, which we now believe to be more ade-
quate, is presented in Figure 6.
Differences Among the Difficulty Categories
As can be seen from the list of difficulties in the Appen-
dix, certain parts of the taxonomy are refined more than
others. Two factors contributed to this variation in the
degree of refinement. First, our theoretical knowledge var-
ies with respect to the different possible difficulties. For
example, although we have enough knowledge about lack
of information about the self to make two additional levels
of within-category distinctions, for indecisiveness, as of yet
at least, we do not. Second, the practical significance of
CAREER DECISION MAKING DIFFICULTIES
PRZOR TO BEGZNNZNG THE PROCESS
I
LACK OF
READZNESS TNFORNATZON
DUE TO
LACK OF INDECZ- DYSFUNCTZONAL THE CAREER S E L F
~OTZVATZON SZVENESS MYTHS DECZSZON
MAKZNG PROCESS
(STEPS)
Figure 6. The revised theoretical taxonomy of career decision-making difficulties.
such refinements seems to vary among the categories. Spe-
cifically, within some categories different difficulties re-
quire distinct courses of action (e.g., lack of information
about preferences may require the help of a counselor,
whereas lack of information about abilities requires relevant
tests). Within other categories, however, the same type of
intervention is required for all difficulties (e.g., the various
dysfunctional myths may require face-to-face counseling).
Indeed, the results of this study indicate that some elab-
orations are justified. The empirical internal structure of the
two selected categories, lack of information about self and
unreliable information, was almost identical to the proposed
taxonomy. The initial organization of the difficulties within
each of these categories was very similar in the two sam-
ples, although the Israeli sample showed a slight deviation
from the theoretical model within one of the two subcate-
gories of lack of information about self. The compatibility
of the within-category structure of the difficulties with the
theoretical model in these two categories, and the consis-
tency of the within-category structures in the two samples,
increase the likelihood of locating not only difficulty cate-
gories but also more refined subcategories. Future research
may contribute to the further elaboration and refinement of
the difficulties within other categories as well.
Another difference among the 10 categories is the degree
of co-occurrence of the difficulties belonging to them. In
general, as can be seen from the medium-to-high reliabili-
ties of the 10 scales, the specific difficulties that compose
each category tend to correlate across individuals. The only
exception to this rule was the scale of Dysfunctional Myths.
The unique feature of this category is the very low corre-
DURZNG THE PROCESS
I
I I
LACK OF INCONSZSTENT
INFORNAT][ON
ABOUT DUE TO
0CCUPATZONS ~AYS OF UNRELZABLE INTERNAL EXTERNAL
0BTAZNZNG INFORNATZON CONFLZCTS CONFLZCTS
INFORHATZON
 DIFFICULTIES IN CAREER DECISION MAKING
lation between the items representing the various dysfunc-
tional myths (range = .02-. 19 and .03-.26 in the Israeli and
the American samples, respectively), which results in the
low scale reliability (.29 and .40, respectively). This pattern
shows that the endorsement of one myth was not related to
the endorsement of others. As mentioned, a similar pattern
of low correlations between myths was also found by Gati
(1996b). Nevertheless, in spite of the low consistency of the
Dysfunctional Myths scale, we decided not to delete or
change this category because of its prevalence and potential
impact on the process of career decision making (Nevo,
1987). We strongly recommend, however, that the unique
nature of difficulties related to this category be further
investigated.
Comparison Between the Two Samples
The participants in the American sample differ from those
in the Israeli sample in many respects, including age (and
thus, presumably, maturity), language, and social-cultural-
economic background. Nevertheless, the findings in these
two samples almost replicated each other. Specifically, the
patterns of reliabilities of the 10 scales in the two samples
are very similar; the empirical structures of the 10 scales are
similar in both the three main clusters and their internal
structure; the structures of difficulties within the category of
lack of information about self are very similar, and those
within the category of unreliable information are practically
identical.
However, the findings of the two samples differed in the
location of the scale of External Conflicts. In light of the
great overall similarity, this difference is puzzling. To find
out what caused it, we reexamined the specific items rep-
resenting the difficulties included in that scale and discov-
ered a small but apparently significant inaccuracy in the
translation. In the Hebrew version the items mentioned lack
of agreement between a significant other and the individual
regarding some aspect of the process, whereas in the En-
glish translation the word used was approval. Unlike agree-
ment, which implies symmetry between the involved parties
(i.e., the self and the significant other), approval implies
asymmetry. Such asymmetry may imply that the individu-
al's position is inferior relative to that of the involved
significant other. Thus, in the American sample these dif-
ficulties joined the cluster of Lack of Readiness, which is
more closely related to issues of maturity, rather than In-
consistent Information cluster, which is more closely related
to incompatible preferences and conflicting opinions.
Another possible explanation for this difference involves
the difference in the characteristics of the participants in the
two studies: The Israeli sample was composed of slightly
older individuals (age 21), who were nearing the end of their
2 or 3 years of compulsory military service, a period that
typically has significant effects on the young people, and
during which they usually become more independent and
responsible. The majority of the American sample (74%),
on the other hand, was composed of college freshmen (ages
18-19), who filled out the questionnaire during their first
521
quarter of college. It may be assumed that in this stage of
life they were just at the beginning of the process of sepa-
ration from home and therefore might have been more
dependent on their families than the Israelis. This may have
contributed to the observed pattern that External Conflicts
were associated with the Lack of Readiness cluster in the
American sample, but not in the Israeli sample.
Implications
Theory. One of the questions raised by the empirical
research on the various indecision measures, and on the
CDS (Osipow et al., 1987) in particular, concerns the di-
mensionality of career indecision. Two approaches have
been proposed for this issue. One considers indecision to be
multidimensional, whereas the other considers it a unidi-
mensional construct. The multidimensional approach sug-
gests that indecision has a number of well-defined dimen-
sions, which, in the context of the CDS, have typically been
summarized by four factors: Diffusion, Support, Approach,
and External Barriers (e.g., Martin, Sabourin, Laplante, &
Coallier, 1991; Shimizu, Vondracek, & Schulenberg, 1994;
Shimizu, Vondracek, Schulenberg, & Hostetler, 1988; Von-
dracek, Hostetler, Schulenberg, & Shimizu, 1990). In con-
trast, the unidimensional approach suggests that, due to the
observed instability in the number and content of the factors
underlying the various measures, there is not enough sup-
port for multidimensionality (e.g., Osipow, 1987; Savickas,
Carden, Toman, & Jarjoura, 1992). This lack of consistency
and stability with regard to the dimensions underlying the
various indecision measures can be attributed to measure-
ment errors and the lack of a theoretical framework (Tins-
ley, 1992). Indeed, when the question of dimensionality was
examined for the rationally developed Career Factors In-
ventory (Chartrand et al., 1990), the results supported the
hypothesized four-factor structure (Career Choices Anxiety,
Generalized Indecisiveness, Need for Career Information,
and Need for Self-Knowledge). Note, however, that these
factors are quite different from those identified in the CDS.
The theoretical framework underlying the present re-
search, as well as its empirical findings, suggest a new
perspective concerning the dimensionality of career indeci-
sion. Instead of trying to discover dimensions in the con-
struct of career indecision, we have suggested a taxonomy
for the various difficulties constituting this construct. The
assumption underlying this endeavor is that career indeci-
sion is not a single type of problem with different symptoms
but rather a group of problems that typically lead to the
same final outcome (i.e., the inability to make a career
decision). Accordingly, to better understand this construct,
it is necessary to investigate its components as well as the
relationships among them (i.e., its structure). The first step
in such an investigation was carded out in this research, and
we regard the reported findings as promising. Specifically,
we found that the various difficulties associated with inde-
cision can be classified into 10 meaningful categories and
that some of these categories can be further classified into
meaningful subcategories. Moreover, these categories and
522 GATI, KRAUSZ, AND OSIPOW
subcategories were found to have a well-defined structure
that is generally compatible with the theoretical model. In
light of this, we believe that future work on indecision
should focus less on the question of dimensionality and
more on the development of a comprehensive, theoretically
sound, and practically significant taxonomy.
The proposed taxonomy of difficulties in career decision
making is compatible with previous research on indecision.
For example, according to Shimizu et al. (1988), diffusion
(i.e., lack of confidence and structure), one of the four
factors in the CDS, is related to feelings of confusion,
discouragement, and lack of experience or information re-
garding career decision making. In the proposed model,
these problems were included in the major category of Lack
of Readiness and were defined as lack of motivation, inde-
cisiveness, and dysfunctional myths. These difficulties ap-
pear to have an affective source and to be associated with
low career maturity (Crites, 1978). Approach-approach
conflict, another factor in the CDS, is represented by three
difficulties in our category of internal conflicts: conflicts
among incompatible desirable aspects, conflicts among in-
compatible desirable occupations, and conflicts between a
desirable occupation and an incompatible desirable aspect.
Lack of knowledge about self and lack of information about
occupations, two of the factors in the CFI (Chartrand et al.,
1990), are represented here by two categories and are fur-
ther divided, systematically, into subcategories. Problems
related to disapproval by significant others, one of the
variables measured by the Career Barriers Inventory (Swan-
son & Tokar, 1991), or to authority orientation, a variable
measured by one of the subscales of the CDDA (Bansberg
& Sklare, 1986), are represented here by the category of
external conflicts. As can be seen from these examples, the
proposed taxonomy reorganizes previously encountered dif-
ficulties into theoretically based categories, elaborates and
refines these categories, and includes additional difficulties
that were not explicitly included in previous classifications.
Moreover, although this taxonomy is cognitively oriented, it
does encompass some affectively based difficulties (e.g.,
those included in the categories of lack of motivation,
indecisiveness, and internal conflicts).
Future research. The proposed taxonomy was tested in
this study by analyzing the responses of deliberating young
adults from two different cross-cultural samples. Although
the two samples provide important evidence for the gener-
ality of the taxonomy, it is important to replicate these tests
in other samples with different characteristics, for example,
a sample of adult career decision makers (ages 40-60) and
samples of various racial and ethnic minorities. Another
way to examine the proposed taxonomy is to elicit career
counselors' judgments about the difficulties of their coun-
selees. Career counselors can be asked to rate the degree to
which each difficulty or each of the 10 categories of diffi-
culties is characteristic of a particular counselee. Then the
structure of difficulties can be derived from these ratings by
procedures similar to those used in the present research and
can also be compared with the theoretical model. This may
provide another perspective on the structure of difficulties
and serve to validate the questionnaire.
The relationship among the various measures of career
indecision has been the focus of several empirical investi-
gations (e.g., Fuqua & Newman, 1989; Tinsley, Bowman, &
York, 1989). Following this line of research, simultaneous
collection of data on the CDDQ and other career indecision
instruments, for example the CDS or the CFI, may make it
possible to identify the major aspects of indecision tapped
by the latter. For instance, data based on the last group of
the American sample of participants (n = 19) reveal a high
correlation (.71) between the total career indecision score on
the CDS and the total CDDQ score. Further research should
investigate this in a larger sample and also investigate the
empirical relations between the CDDQ scales and the four
suggested factors of the CDS. This may shed light, albeit
indirectly, on the dimensionality of the CDS, an issue that
has been the focus of many studies, with inconclusive
results. Similarly, a comparison between the CDMSE (Tay-
lor & Betz, 1983) and the CDDQ may reveal what catego-
ries of difficulties are more closely associated with self-
efficacy in career decision making in general and the five
dimensions underlying the CDMSE in particular. In the
same time, such investigations would contribute to the val-
idation of the CDDQ and the taxonomy underlying it.
A different program of research on indecision has focused
on the classification of subtypes of individuals according to
their problems in career decision making (e.g., Callanan &
Greenhaus, 1992; Larson, Heppner, Ham, & Dugan, 1988;
Lucas & Epperson, 1990; Rojewski, 1994; Savickas &
Jarjoura, 1991). This line of research assumes that there is a
systematic variance in the sets of problem combinations
across individuals; however, classifying individuals into
subtypes is based on the particular theoretical framework
that underlies the taxonomy of the problems or difficulties.
Future research should explore the contribution of the
present taxonomy of difficulties to the classification of
subtypes of individuals.
The categorization of the difficulties in the proposed
taxonomy can be linked to the different consequences (and
hence also interventions) that each difficulty may lead to.
Specifically, we assume that, in general, difficulties involv-
ing lack of readiness may prevent the individual from be-
ginning the decision-making process, difficulties involving
lack of information may stop the process once it has been
started, and difficulties involving inconsistent information
(i.e., unreliable information that is not checked out or un-
resolved conflicts) may stop the process or lead to a deci-
sion that is less than optimal. The relative impact and
severity of each of these major categories, as well as that of
each of the 10 categories, needs to be further studied with
respect to both the process of career decision making and its
outcomes.
Counseling. A proper diagnosis of counselees' prob-
lems in career decision making is necessary (although, of
course, not sufficient) to provide deliberating individuals
with the help they need. As mentioned earlier, the purpose
of this research was to develop and test a theoretical frame-
work for difficulties in career decision making. Neverthe-
less, it seems that the questionnaire developed to test the
proposed taxonomy has the potential to become, after the
 DIFFICULTIES IN CAREER DECISION MAKING
necessary changes, an additional important tool for individ-
ual career counseling. Specifically, the comprehensiveness
and the theoretical basis of the CDDQ permits the assess-
ment of the individual's difficulties at three different levels
of specificity: (a) three major categories (i.e., Lack of
Readiness, Lack of Information, and Inconsistent Informa-
tion); (b) 10 categories (represented by the 10 scales); and
(c) specific difficulties (at the level of specific items). De-
pending on the counselor's judgment, all three levels can be
assessed, or the assessment can focus on only the one or two
levels that seem most relevant.
Thus, the questionnaire may be used, first, for an initial
screening of clients who seek career counseling, by direct-
ing the clients to the various intervention options available
according to the CDDQ results. For example, those who
have difficulties in dealing with conflicts could be directed
to personal counseling; those who lack information about
their abilities, to further assessment; and those who lack
information on educational or career options, to the occu-
pational library. Second, the information from the CDDQ
administered before the first counseling session could pro-
vide relevant data about the counselee that may help the
counselor plan and guide the counseling process. Third, the
questionnaire may be used as a "needs assessment" instru-
ment to collect information about the kinds of difficulties in
career decision making that frequently occur in particular
groups (e.g., 12th-grade students, college freshmen). This
may facilitate the design of interventions that best suit the
needs of each group. Finally, it may serve, like other in-
struments that assess career indecision, to evaluate the ef-
fectiveness of career interventions, by comparing the coun-
selees' responses before and after the intervention.
However, all these important and useful applications should
be implemented only after the CDDQ is transformed from a
research questionnaire aimed at testing a theoretical model
into a validated clinical instrument.
Computer-assisted career guidance systems. The diag-
nosis of difficulties is also very important for computer-
assisted career guidance systems. One of the problems as-
sociated with the use of these systems is that often the users
do not use them effectively because they contain a great
deal of information and many components, so that the user
"gets lost" (Gati, 1996a). Discovering the individuals' dif-
ficulties at the beginning of their dialogue with the comput-
erized system should make it possible to direct them to
those components in the system that may provide the infor-
mation or guidance they need or to recommend that they
approach a career counselor in those cases in which the
computerized system cannot help (e.g., solving interper-
sonal conflicts).
Conclusion
The significance of this research is linked to the changes
in the world of work, a setting that is becoming increasingly
dynamic. These changes imply an increase in the number of
career transitions individuals make during their lifetime and
therefore in the frequency with which they must make
523
career decisions. Locating and understanding the difficulties
faced by individuals during their process of career decision
making thus have significant practical implications. The
results of this research may contribute to the facilitation of
these career decisions and hence indirectly help to increase
the quality of life in one of its more significant aspects,
namely, work.
References
Bansberg, B., & Sklare, J. (1986). The career decision diagnostic
assessment. Monterey, CA: CTB/McGraw-Hill.
Baron, J. (1988). Thinking and deciding. New York: Cambridge
University Press.
Betz, N. E. (1992). Career assessment: A review of critical issues.
In S. D. Brown & R. W. Lent (Eds.), Handbook of counseling
psychology (pp. 453-484). New York: Wiley.
Bordin, E. S., & Kopplin, D. A. (1973). Motivational conflict and
vocational development. Journal of Counseling Psychology, 20,
154-161.
Borgen, F. H., & Barnett, D. C. (1987). Applying cluster analysis
in counseling psychology research. Journal of Counseling Psy-
chology, 34, 456-468.
Brown, D. (1990). Models of career decision making. In D. Brown
& L. Brooks (Eds.), Career choice and development (2nd ed.,
pp. 395-421). San Francisco: Jossey-Bass.
Callanan, G. A., & Greenhaus, J. H. (1992). The career indecision
of managers and professionals: An examination of multiple
subtypes. Journal of Vocational Behavior, 41, 212-231.
Campbell, R. E., & Cellini, J. V. (1981). A diagnostic taxonomy of
adult career problems. Journal of Vocational Behavior, 19,
175-190.
Campbell, R.E., & Heffernan, J.M. (1983). Adult vocational
behavior. In W. B. Walsh & S. H. Osipow (Eds.), Handbook of
vocational psychology (Vol. 1, pp. 223-260). Hillsdale, NJ:
Erlbaum.
Chartrand, J.M., Robbins, S.B., Morrill, W.H., & Boggs, K.
(1990). The development and validation of the career factors
inventory. Journal of Counseling Psychology, 37, 490-501.
Crites, J. O. (1978). Theory and research handbook for the Career
Maturity Inventory. Monterey, CA: CTB/McGraw-Hill.
Fleishman, E.A., & Quaintance, M.K. (1984). Taxonomies of
human performance. Orlando, FL: Academic Press.
Fouad, N. A. (1993). Cross-cultural vocational assessment. Career
Development Quarterly, 42, 4-13.
Fouad, N.A. (1994). Annual review 1991-1993: Vocational
choice, decision-making, assessment, and intervention. Journal
of Vocational Behavior, 45, 125-176.
Fuqua, D.R., & Hartman, B.W. (1983). A behavioral index of
career indecision for college students. Journal of College Stu-
dent Personnel, 24, 507-512.
Fuqua, D. R., & Newman, J.L. (1989). An examination of the
relations among career subscales. Journal of Counseling Psy-
chology, 36, 487-491.
Gati, I. (1986). Making career decisions: A sequential elimination
approach. Journal of Counseling Psychology, 33, 408-417.
Gati, I. (1990). Why, when, and how to take into account the
uncertainty involved in career decisions. Journal of Counseling
Psychology, 37, 277-280.
Gati, I. (1996a). Computer-assisted career counseling: Challenges
and prospects. In M. L. Savickas & B. W. Walsh (Eds.), Hand-
book of career counseling theory and practice (pp. 169-190).
Palo Alto, CA: Davies-Black Publishing.
Gati, I. (1996b). Towards career decision making. Unpublished
524 GATI, KRAUSZ, AND OSIPOW
manuscript, Department of Psychology, Hebrew University of
Jerusalem (in Hebrew).
Gati, I., Fassa, N., & Houminer, D. (1995). Applying decision
theory to career counseling practice: The sequential elimination
approach. Career Development Quarterly, 43, 211-220.
Gaff, I., Garty, Y., & Fassa, N. (1996). Using career-related aspects
to assess person-environment fit. Journal of Counseling Psy-
chology, 43, 196-206.
Gati, I., Osipow, S. H., & Givon, M. (1995). Gender differences in
career decision making: The content and structure of prefer-
ences. Journal of Counseling Psychology, 42, 204-216.
Gelatt, H. B. (1989). Positive uncertainty: A new decision-making
framework for counseling. Journal of Counseling Psychology,
36, 252-256.
Holland, J.L. (1985). Making vocational choices: A theory of
vocational personalities and work environments (2nd. ed.).
Englewood Cliffs, NJ: Prentice-Hall.
Holland, J. L., Daiger, D.C., & Power, P.G. (1980). My Voca-
tional Situation. Palo Alto, CA: Consulting Psychologists Press.
Jepsen, D. A., & Dilley, J. S. (1974). Vocational decision making
models: A review and comparative analysis. Review of Educa-
tional Research, 44, 331-349.
Jones, L. K. (1989). Measuring a three-dimensional construct of
career indecision among college students: A revision of the
Vocational Decision Scale--the Career Decision Profile. Jour-
nal of Counseling Psychology, 36, 477-486.
Jones, L. K., & Chenery, M. F. (1980). Multiple subtypes among
vocationally undecided college students: A model and assess-
ment instrument. Journal of Counseling Psychology, 27, 469-
477.
Katz, M.R. (1966). A model for guidance for career decision
making. Vocational Guidance Quarterly, 15, 2-10.
Katz, M.R. (1993). Computer-assisted career decision making:
The guide in the machine. Hillsdale, NJ: Eflbaum.
Krumboltz, J. D. (1991). Manual for the Career Beliefs Inventory.
Palo Alto, CA: Consulting Psychologists Press.
Krumboltz, J. D. (1994). The Career Beliefs Inventory. Journal of
Counseling & Development, 72, 424-428.
Larson, J. H., Busby, D. M., Wilson, S., Medora, N., & Allgood, S.
(1994). The multidimensional assessment of career decision
problems: The career decision diagnostic assessment. Journal of
Counseling & Development, 72, 323-328.
Larson, L.M., Heppner, P.P., Ham, T., & Dugan, K. (1988).
Investigating multiple subtypes of career indecision through
cluster analysis. Journal of Vocational Behavior, 35, 439-446.
Lofquist, L. H., & Dawis, R.V. (1978). Values as secondary to
needs in the Theory of Work Adjustment. Journal of Vocational
Behavior, 12, 12-19.
Lucas, M.S., & Epperson, D.L. (1990). Types of vocational
undecidedness: A replication and refinement. Journal of Coun-
seling Psychology, 37, 382-388.
Martin, F., Sabourin, S., Laplante, B., & Coallier, J.C. (1991).
Diffusion, support, approach, and external barriers as distinct
theoretical dimensions of the Career Decision Scale: Discon-
firming evidence? Journal of Vocational Behavior, 38, 187-197.
Meier, S. T. (1991). Vocational behavior, 1988-1990: Vocational
choice, decision-making, career development interventions, and
assessment. Journal of Vocational Behavior, 39, 131-181.
Meir, E.I., & Gati, I. (1981). Guidelines for item selection in
inventories yielding score profiles. Educational and Psycholog-
ical Measurement, 41, 1011-1016.
Miller, R. G. (1971). Development of a taxonomy of human per-
formance: Design of a systems task vocabulary (Tech. Rep.
A1R-72 6/2035-3/71-TRJI). Silver Spring, MD: American Insti-
tutes for Research.
Mitchell, L. K., & Krumboltz, J. D. (1984). Research on human
decision making: Implications for career decision making and
counseling. In S. D. Brown & R. W. Lent (Eds.), Handbook of
counseling psychology (pp. 238-280). New York: Wiley.
Neimeyer, G. J. (1988). Cognitive integration and differentiation
in vocational behavior. The Counseling Psychologist, 16, 440-
475.
Nevo, O. (1987). Irrational expectations in career counseling and
their confronting arguments. Career Development Quarterly, 35,
239-250.
Osipow, S.H. (1987). Manual for the Career Decision Scale.
Odessa, FL: Psychological Assessment Resources.
Osipow, S.H., Carney, C.G., & Barak, A. (1976). A scale of
educational-vocational undecidedness: A typological approach.
Journal of Vocational Behavior, 9, 233-243.
Osipow, S. H., Carney, C., Winer, J., Yanico, B., & Koschier, M.
(1987). The Career Decision Scale (rev. ed.). Odessa, FL: Psy-
chological Assessment Resources.
Osipow, S.H., & Fitzgerald, L.F. (1996). Theories of career
development (4th ed.). Boston, MA: Allyn & Bacon.
Osipow, S.H., & Winer, J. (1996). Career assessment and the
Career Decision Scale. Journal of Career Assessment, 4, 117-
130.
Phillips, S. D. (1994). Choice and change: Convergence from the
decision-making perspective. In M.L. Savikas & R.W. Lent
(Eds.), Convergence in career development theories (pp. 155-
163). Palo Alto, CA: Consulting Psychologists Press.
Pitz, G., & Harren, V. A. (1980). An analysis of career decision
making from the point of view of information processing and
decision theory. Journal of Vocational Behavior, 16, 320-346.
Roe, A. (1956). The psychology of occupations. New York: Wiley.
Rojewski, J. W. (1994). Career indecision types for rural adoles-
cents from disadvantaged and nondisadvantaged backgrounds.
Journal of Counseling Psychology, 41, 356-363.
Rounds, J. B., & Tinsley, H. E. A. (1984). Diagnosis and treatment
of vocational problems. In S. D. Brown & R. W. Lent (Eds.),
Handbook of counseling psychology (pp. 137-177). New York:
Wiley.
Salomone, P.R. (1982). Difficult cases in career counseling: II.
The indecisive client. Personnel and Guidance Journal, 60,
496-500.
SAS Institute. (1982). SAS user's guide: Statistics. New York:
Author.
Sattath, S., & Tversky, A. (1977). Additive similarity trees. Psy-
chometrika, 42, 319-345.
Savickas, M. L., Carden, A. D., Toman, S., & Jarjoura, D. (1992).
Dimensions of career decidedness. Measurement and Evaluation
in Counseling and Development, 25, 102-112.
Savickas, M. L., & Jarjoura, D. (1991). The Career Decision Scale
as a type indicator. Journal of Counseling Psychology, 38,
85-90.
Shimizu, K., Vondracek, F. W., & Schulenberg, J. E. (1994). Uni-
dimensionality versus multidimensionality of the Career Deci-
sion Scale. Journal of Career Assessment, 2, 1-14.
Shimizu, K., Vondracek, F. W., Schulenberg, J. E., & Hostetler,
M. (1988). The factor structure of the Career Decision Scale:
Similarities across selected studies. Journal of Vocational Be-
havior, 32, 213-225.
Slaney, R. B. (1988). The assessment of career decision making. In
W. B. Walsh & S. H. Osipow (Eds.), Career decision making
(pp. 33-76). Hillsdale, NJ: Erlbaum.
 DIFFICULTIES IN CAREER DECISION MAKING 525

SPSS, Inc. (1986). SPSSx user's guide, Edition 2. New York:
McGraw-Hill.
Super, D. E. (1953). A theory of vocational development. Ameri-
can Psychologist, 8, 185-190.
Swanson, J. L., & Tokar, D. M. (1991). Development and initial
validation of the career barriers inventory. Journal of Vocational
Behavior, 39, 344-361.
Taylor, K. M., & Betz, N. E. (1983). Applications of self-efficacy
theory to the understanding and treatment of career indecision.
Journal of Vocational Behavior, 22, 63-81.
Tinsley, H. E. A. (1992). Career decision making and career inde-
cision. Journal of Vocational Behavior, 41, 209-211.
Tinsley, H. E. A., Bowman, S.L., & York, D.C. (1989). CDS,
MVS, Vocational Rating Scale, and Decision Rating Scale: Do
they measure the same thing? Journal of Counseling Psychol-
ogy, 36, 115-120.
Tversky, A., & Hutchinson, W. J. (1986). Nearest neighbor anal-
ysis of psychological spaces. Psychological Review, 93, 3-22.
Vondracek, F. W., Hostetler, M., Schulenberg, J. E., & Shimizu,
K. (1990). Dimensions of career indecision. Journal of Coun-
seling Psychology, 37, 98-106.
Walsh, W.B., & Osipow, S. H. (Eds.). (1988). Career decision
making. Hillsdale, NJ: Erlbaum.
Ward, J. H. (1963). Hierarchical grouping to optimize an objective
function. Journal of the American Statistical Association, 58,
236-244.

Appendix

The 44 Difficulties Included in the Elaborated Theoretical Taxonomy

Prior to the beginning of the process
Lack of Readiness
Lack of Motivation (Rm)
1. Unwillingness to make a career decision
2. Work is not perceived as the most important thing in life
3. Feeling that time will lead to the "right" career choice
Indecisiveness (Ri)
4. A general difficulty in making decisions
5. A general need for conf'mnation and support for decisions
6. A general tendency to avoid commitment
7. A general fear of failure
Dysfunctional Myths (Rd)
8. The belief that entering a career will solve personal problems
9. The belief that there is an ideal career which can fulfill all aspirations
10. The belief that a career choice is a one-time thing and a life-long obligation
Lack of Knowledge About the Process of Career Decision Making (Lp)
11. About the steps involved in making a career decision
12. About the factors to take into consideration
13. About how to combine information concerning the self and career alternatives
Duringthe process
Lack of Information
Lack of Information About Self (Ls)
14. Lack of information about abilities
15. Lack of information about personality traits
16. Lack of information about preferred career alternatives
17. Lack of information about career-related preferences
18. Lack of information about abilities in the future
19. Lack of information about personality traits in the future
20. Lack of information about career alternatives that will be preferred in the future
21. Lack of information about career-related preferences in the future
Lack of Information About Occupations (Lo)
22. Lack of information about the variety of career or training alternatives
23. Lack of information about the characteristics of the career or training alternatives that interest the individual
24. Lack of information about the variety of future career or training alternatives
25. Lack of information about the future characteristics of the career or training alternatives
Lack of Information About Ways of Obtaining Additional Information (La)
26. Lack of information about ways of obtaining additional information about the self
27. Lack of information about ways of obtaining additional information about career and training alternatives

(Appendix continues on next page)

526 GATI, KRAUSZ, AND OSIPOW

Appendix (continued)
Inconsistent Information
Unreliable Information (Iu)
28. Unreliable information about abilities
29. Unreliable information about personality traits
30. Unreliable information about preferred career alternatives
31. Unreliable information about career-related preferences
32. Unreliable information about the existence of a particular career or training alternative
33. Unreliable information about the characteristics of career or training alternative(s)
Internal Conflicts (Ii)
34. Unwillingness to compromise
35. Several equally attractive career alternatives
36. Dislike of accessible career alternatives
37. Something in a preferred career alternative is undesirable
38. Preferences that cannot be combined in one career alternative
39. Abilities are insufficient for the requirements of the preferred career alternative
40. Abilities exceed those required in the preferred career alternative
External Conflicts (Ie)
41. Disagreement between a significant other and the individual concerning the desirable career alternative
42. Disagreement between a significant other and the individual concerning the desirable career-related characteristics
43. Disagreement between different significant others concerning the recommended career alternative
44. Disagreement between different significant others concerning the recommended career-related characteristics

Received March 20, 1996

Revision received May 28, 1996
Accepted May 30, 1996