Patterns of Career Decision-Ma

Journal of Counseling Psychology
Patterns of Career Decision-Making Difficulties in 16 Countries: A Person-

Centered Investigation
Nimrod Levin, Yuliya Lipshits-Braziler, and Itamar Gati
Online First Publication, September 4, 2023. https://dx.doi.org/10.1037/cou0000704
CITATION
Levin, N., Lipshits-Braziler, Y., & Gati, I. (2023, September 4). Patterns of Career Decision-Making Difficulties in 16
Countries: A Person-Centered Investigation. Journal of Counseling Psychology. Advance online publication.
https://dx.doi.org/10.1037/cou0000704
Journal of Counseling Psychology
© 2023 American Psychological Association
ISSN: 0022-0167 https://doi.org/10.1037/cou0000704
Patterns of Career Decision-Making Difficulties in 16 Countries:

A Person-Centered Investigation
Nimrod Levin1, 2, Yuliya Lipshits-Braziler2, and Itamar Gati2, 3
1
Institute of Psychology, University of Lausanne
2
Seymour Fox School of Education, Hebrew University of Jerusalem
3
Department of Psychology, Hebrew University of Jerusalem
Classifying clients into career indecision types can facilitate tailoring interventions to each client’s specific
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
needs. The present research examined a typology of career indecision on 50 data sets (N = 31,527)
This document is copyrighted by the American Psychological Association or one of its allied publishers.
representing diverse populations from 16 countries on five continents. Latent profile analyses of participants’
responses on the Career Decision-Making Difficulties Questionnaire (CDDQ) revealed seven replicable career
indecision types across samples: unmotivated, unrealistic, generally uninformed, occupations-uninformed,
conflicted-uninformed, externally conflicted, and internally conflicted. Age emerged as a negligible predictor
of career indecision types, whereas gender predicted membership in the unmotivated type, with men twice
more likely to be unmotivated than women. The seven types were similarly predictive of career decision status,
decision certainty, and decision self-efficacy. These results largely support using the CDDQ to differentially
diagnose career indecision types based on 10 causes of career indecision in different countries, life stages,
and genders. Classifying individuals based on their patterns of career decision-making difficulties supports
tailoring individual career counseling or group interventions to clients’ needs.
Public Significance Statement

Analyzing 50 samples from 16 countries on five continents, we found that individuals can be classified
into seven career indecision types based on the causes of their career decision-making difficulties. By
identifying clients’ career indecision types, career counselors and psychologists can tailor the counseling
process to address clients’ specific needs, resulting in a more effective counseling experience for clients.
Keywords: career indecision, career decision-making difficulties, person-centered research, latent profile
analysis, career counseling
Supplemental materials: https://doi.org/10.1037/cou0000704.supp
Diagnosing clients’ problems is an essential step in providing 1996; Xu & Bhang, 2019). However, this approach has limitations,
effective counseling services tailored to client’s specific needs. In such as the lack of validated guidelines for prioritizing problems of
career counseling, the dominant approach for diagnosing problems comparable saliency and the complexity involved in integrating
involves applying multidimensional assessments to identify the multiple scores into a coherent diagnosis. An alternative approach,
causes of clients’ indecision (Brown & Rector, 2008; Gati et al., which resolves some of the challenges of translating multiple scores
Association. The findings have not been published or disseminated publicly

Nimrod Levin https://orcid.org/0000-0003-4643-4288 in any other way. Data are available at https://osf.io/fz2se/.
Yuliya Lipshits-Braziler https://orcid.org/0000-0001-6648-064X Nimrod Levin played a lead role in data curation, formal analysis,
Itamar Gati https://orcid.org/0000-0001-5201-1623 methodology, project administration, visualization, and writing–original
The authors are grateful for the generosity of their colleagues who shared draft and an equal role in conceptualization and investigation. Yuliya
their data, thereby making this research possible. The research reports the Lipshits-Braziler played a supporting role in methodology and project
analyses of previously collected secondary data by the authors (12 samples) administration and an equal role in conceptualization, investigation,
and other research teams (38 samples). Information on the source of each supervision, and writing–review and editing. Itamar Gati played a lead
sample can be found in Supplemental Material F in the online materials. This role in funding acquisition, a supporting role in methodology and project
research was supported by the Israel Science Foundation (Grant 636/16) administration, and an equal role in conceptualization, investigation,
and by the Samuel and Esther Melton Chair of Itamar Gati and was not supervision, and writing–review and editing.
preregistered. The authors have no known conflicts of interest to disclose. Correspondence concerning this article should be addressed to Nimrod
The authors thank Benny A. Benjamin, Michaël Parmentier, Moshe Levin, Institute of Psychology, University of Lausanne, Géopolis 5209,
Sluhovsky, and Tirza Willner for their comments on earlier versions of this article. CH-1015 Lausanne, Switzerland, or Itamar Gati, Department of
The main findings reported in this article were presented orally at the Psychology, Hebrew University of Jerusalem, Mount Scopus,
2022 Early Career Marathon of the International Association of Applied Jerusalem 9765418, Israel. Email: nimrod.levin@mail.huji.ac.il or
Psychology and at the 2023 Annual Conference of the Israeli Psychometric itamar.gati@huji.ac.il
1
2 LEVIN, LIPSHITS-BRAZILER, AND GATI
into diagnostic insights, involves differentially diagnosing clients indecision types described in 15 clustering and nonclustering
into career indecision types (e.g., Kelly & Pulver, 2003; Milot- studies, Gordon (1998) described three types of decided students
Lapointe et al., 2022; Rochlen et al., 2004). A recent study (very decided, somewhat decided, and unstable decided) and four
demonstrated the utility of the Career Decision-Making Difficulties types of undecided students (tentatively undecided, developmen-
Questionnaire (CDDQ; Gati et al., 1996), a multidimensional tally undecided, seriously undecided, and chronically indecisive).
measure of 10 specific causes of career indecision, for diagnosing In comparison, Kelly and Pulver’s (2003) review of eight clustering
replicable types of career indecision in two large U.S. samples studies provided a more parsimonious, threefold typology of career
(Levin et al., 2022). However, the applicability and generalizability indecision. Their first type included relatively decided individuals
of this CDDQ-based typology in other countries remain to be experiencing little career indecision and who do not seem to need
examined. career counseling (e.g., happy and work oriented in Lucas &
In the present research, we sought to investigate the applicability Epperson, 1988; confident-decided in Wanberg & Muchinsky,
and generalizability of a CDDQ-based differential diagnosis of 1992). Kelly and Pulver’s (2003) second type corresponded to
career problems among individuals from diverse countries. Using developmental indecision, which is considered to be the result of
latent profile analysis (LPA), we aimed to identify homogeneous lack of information (LI) or motivation (e.g., undecided and limited
groups of individuals with similar patterns of causes of career interests in Lucas & Epperson, 1988; indifferent-undecided in
indecision while confirming their presence and relevance in Wanberg & Muchinsky, 1992). Finally, Kelly and Pulver’s (2003)
different countries. Specifically, we analyzed data from 50 samples third type corresponded to chronic indecision, which is considered
and 16 countries to identify indecision types and then validated to be the result of a lack of information as well as emotional and
the derived typology by exploring the associations of types with personality factors such as anxiety, low self-esteem, or dependence
demographic predictors (age and gender) and career decision- on others (e.g., anxious and unclear on goals in Lucas & Epperson,
related outcomes (using three indicators of career decidedness). By 1988; anxious-undecided in Wanberg & Muchinsky, 1992). Table 1
doing so, we aimed to provide a more comprehensive understanding presents the types reported in these two literature reviews and eight
of the applicability and generalizability of a CDDQ-based typology representative clustering studies (see Supplemental Material A for
for classifying individuals in different countries into career indeci- an overview of 24 previous clustering studies).
sion types. The syntheses of Kelly and Pulver (2003) and Gordon (1998)
suggest classifying individuals into three or seven primary types
Career Indecision: A Framework for the based on the causes of their career indecision and their degree of
Diagnosis of Career Problems decidedness. However, previous clustering studies often yielded
inconsistent findings, namely in the number and characteristics of
The need for procedures to diagnose career problems was the identified types. Interestingly, inconsistencies emerged even in
recognized decades ago (see Jones & Lohmann, 1998; Osipow, studies clustering the same set of variables (Lucas & Epperson,
1999; Osipow & Winer, 1996). Campbell and Cellini (1981) 1988, 1990) as well as in studies clustering multiple scores derived
proposed four categories of career problems: (a) problems in decision from a single measure, such as the CDS (Argyropoulou et al.,
making, (b) problems in implementing plans, (c) problems in
2007; Rojewski, 1994; Savickas & Jarjoura, 1991) and the Career
organizational performance, and (d) problems in organizational
Factors Inventory (Akos et al., 2004; Chartrand et al., 1994). These
adaptation. In the vocational literature, problems associated with
inconsistencies, indicative of the relatively low reliability of the
career choices and their implementation—Campbell and Cellini’s
derived typologies, may explain why studies investigating the
(1981) first two categories—were often combined under the term
predictive validity of career indecision types failed to document
career indecision, defined as the inability “to make an appropriate and
meaningful differential effectiveness of interventions for different
required degree of commitment to an educational or vocational
career indecision types (Kelly & Pulver, 2003; Milot-Lapointe
direction” (Osipow, 1999, p. 234). This definition suggests that both
et al., 2022; Rochlen et al., 2004). This state of affairs underscores
undecided and decided individuals can experience indecision and
the need for a more reliable and valid method to classify clients into
that indecision could induce uncertainty or delay in implementing
career indecision types.
a decision.
Various explanations have been suggested for the inconsistency,
Various assessments were developed over the years to evaluate the
poor replicability, and low predictive validity of the reported
nature and severity of career indecision. For example, items included
typologies of career indecision. These include the clustering of
in the Career Decision Scale (CDS; Osipow et al., 1976), among
unreliable scores (e.g., in the case of the CDS, see Osipow &
the first career indecision assessments, were derived from clients’
Winer, 1996), the use of traditional cluster analytical techniques
problem statements relating, among other causes of indecision, to
self-knowledge, external conflicts, and vocational exploration. that are sensitive to sample characteristics and lack clear guidelines
Later assessments considered a wider variety of causes of career for selecting the optimal number of profiles (Hofmans et al., 2020;
indecision, including cognitive (e.g., needs for occupational or Spurk et al., 2020), and the clustering of large and nonoverlapping
self-knowledge; Chartrand et al., 1994), emotional (e.g., anxiety or sets of variables giving rise to different types depending on the
low self-esteem; Saka et al., 2008), and personality-related (e.g., variables used to derive the types (Brown & Rector, 2008; Kelly &
generalized indecisiveness; Jones & Lohmann, 1998) causes. Pulver, 2003). Krieshok (1998) concluded that
Parallel to the development of career indecision assessments, many have addressed this issue [regarding types of career indecision], but
considerable effort was devoted to designing procedures for a the lack of consistent findings makes it the clearest example of something
differential diagnosis of career problems by clustering clients into the field wishes would yield some certainty (so that it could inform
career indecision types. For example, reviewing the career practice), but that has not yet reached the certainty stage. (p. 216)
CAREER INDECISION TYPES IN 16 COUNTRIES 3
These two types differ

A recent effort to address some of the shortcomings of previous
Anxious and unclear on

Chronically indecisivec
clustering studies investigated the utility of a CDDQ-based typology
Choice anxious
of career indecision types (Levin et al., 2022). The CDDQ is
goals
typically used as a multidimensional assessment of 10 causes of
Chronically undecided
Chronically indecisive
Seriously undecided
Undecided-anxious
Chronic indecision
Chronic indecision
career indecision grouped into three major clusters: (a) lack of
Conflicted
readiness (comprising lack of motivation, general indecisiveness,
and dysfunctional beliefs); (b) lack of information (comprising
c
lacking information about the career decision-making process, the
quantitatively based on their anxiety levels (moderate vs. excessive). d A representative set of clustering studies conducted among both undecided and decided participants.
Derived from a synthesis of previous career indecision typology research (pp. 445–446). b These two types differ quantitatively based on their degree of decidedness.
self, occupations, and ways of obtaining additional information);
Caught in a
undecidedc
and (c) inconsistent information (comprising unreliable information,

Indecisive
Seriously
dilemma
internal conflict, and external conflict). Yet, using LPA, Levin et al.
(2022) demonstrated that the CDDQ can also be used for classifying
clients into five career indecision types—unmotivated, indecisive,
unrealistic, uninformed, and conflicted—in two U.S. samples. These

five types represent distinct patterns of co-occurring causes of career

Developmentally undecided
Undecided and limited
Transitional indecision
indecision. Specifically, the unmotivated, indecisive, and unrealistic

Developmentally
types were characterized by a single salient difficulty relating to a

Uninformed
undecidedb
interests
lack of motivation, general indecisiveness, or dysfunctional beliefs.

Developmental indecision
Undecided-indifferent
Then, whereas the uninformed and conflicted types included

individuals lacking information in multiple domains, individuals in
the conflicted type also reported salient external conflicts. These
Career indecision types
The Compatibility Among Career Indecision Types Reported in Previous Literature Reviews and Clustering Studies
recent findings support using the CDDQ as a differential diagnostic

tool in the United States.
Yet, the applicability and generalizability of Levin et al.’s
Ready to decide
Unmotivated
Crystallizing
preferences
Tentatively
undecidedb
Happy and
(2022) CDDQ-based typology in other countries remain to be

playful
tested. In this regard, 23 of the 24 previous clustering studies on

career indecision were conducted in either North America or
Europe, with none examining the replicability of types across
countries (see Supplemental Material A for an overview of the
studies). In fact, Gordon’s (1998) and Kelly and Pulver’s (2003)
Unstable decided
syntheses preceded the publication of any study conducted outside

Concerned-
Indecisive
decided
the United States. Nevertheless, Kelly and Pulver’s (2003) typology—

differentiating among decided, developmentally undecided, and
chronically indecisive individuals—appears compatible with
Career decided and confident
studies conducted later in Canada (Guay et al., 2006), Greece

Happy and work oriented
(Argyropoulou et al., 2007), and Portugal (Santos & Ferreira,

Somewhat decided
2012), thus supporting the likely generalizability and applicability

Very decided
Unrealistic
Decided
Decided
of indecision types across countries.

—
—
—
Moreover, although some have challenged the cross-cultural

equivalence of career indecision (Xu & Bhang, 2019), recent studies
provide ample evidence for the equivalence of the factor structure of
career indecision in different countries as measured by contempo-
rary assessments, such as the CDDQ (Levin et al., 2020, 2023) and
Very decided
the Career Indecision Profile (Xu & He, 2022). These findings
Confident-
decided
notwithstanding, some studies found that individuals in Asian

—
and African countries are more prone to career indecision than

individuals in the United States or Europe (Atitsogbe et al., 2018;
Levin et al., 2020; Willner et al., 2015). Thus, these findings suggest
that although the factor structure of career indecision rather
Wanberg and Muchinsky (1992)
replicates in different countries, career indecision may still

Lucas and Epperson (1988)
Santos and Ferreira (2012)
manifest differently in different contexts.

Kelly and Pulver (2003)a
Chartrand et al. (1994)

Source
Levin et al. (2022)
The Present Study

Guay et al. (2006)
Feldt et al. (2011)
Empirical findingsd
Rojewski (1994)
Literature reviews
Gordon (1998)
The main goal of the present study was to test the applicability
and generalizability of a CDDQ-based procedure for differentially
Table 1
diagnosing career problems among individuals from diverse

countries. This goal responds to calls for validating the relevance
of career assessments to different cultural contexts (Xu & Bhang,
a
2019; Xu & He, 2022). Attending to practical considerations, we Finally, we tested the utility of the identified types in predicting
sought to develop a typology of career indecision that could be indicators of career decidedness. Career decidedness is among the
implemented efficiently using a single assessment. We utilized most widely used indicators of career indecision types (see Gordon,
LPA to identify indecision types derived from the 10 causes of 1998; Kelly & Pulver, 2003), but it has been operationalized using
career indecision as measured by the CDDQ within and across various criteria, including whether individuals declared an academic
16 countries. We selected the CDDQ because (a) it is among the major (Larson et al., 1988), reported being decided (Wanberg &
most administered and psychometrically sound assessments of Muchinsky, 1992), or made concrete career plans (Guay et al., 2006).
career indecision (Xu & Bhang, 2019); (b) it offers a comprehen- However, Levin et al. (2022) argued that in diagnosing career
sive assessment of various cognitive (e.g., lack of information), problems, the causes of indecision should be distinguished from their
emotional (e.g., lack of motivation), and personality-related (e.g., consequences. Thus, to validate the identified types and test their
general indecisiveness) causes of career indecision; and (c) its utility in predicting relevant outcomes, we examined whether the
factor structure has been validated across countries, age groups, career indecision types differed in three career decidedness-related
and genders (Atitsogbe et al., 2018; Levin et al., 2020, 2023). outcomes: career decision status (range of considered alternatives
Despite the exploratory nature of clustering, we expected to detect [RCA]), decision certainty (the degree of certainty in choice), and
between three (Kelly & Pulver, 2003) and seven (Gordon, 1998) career decision self-efficacy (confidence in having the necessary skills
career indecision types across countries. As Table 1 depicts and for career decision making). Previous studies found that types
considering Levin et al.’s (2022) CDDQ-based typology, two characterized by a lack of information exhibit lower levels of
types of seemingly “decided” individuals are likely to be detected, decidedness (e.g., Levin et al., 2022; Multon et al., 2007), decision
with some individuals making a decision based on unrealistic certainty (e.g., Multon et al., 2007; Rojewski, 1994), and career
expectations (unrealistic) and others generally struggling with decision self-efficacy (e.g., Feldt et al., 2011; Guay et al., 2006)
decision making (indecisive). Then, two “developmental” indeci- than other types. Based on these findings, we hypothesized that
sion types were envisioned to emerge, with some individuals not types associated with the lack of information would be the least
sufficiently motivated to decide (unmotivated) while others needing decided, certain, and confident.
to engage in exploratory activities (uninformed). Finally, although
Levin et al. (2022) reported only one “chronic” indecision type
(conflicted), two chronic indecision types were reported in previous Method
studies, with some experiencing indecision due to internal conflicts Archival Data Request
while others being externally conflicted and seeking the approval of
others. To gather relevant data sets, we contacted researchers with
In addition to identifying career indecision types, we tested the access to CDDQ data through several channels: (a) tapping a list of
validity of the derived typology by examining whether age and researchers who had requested and signed the CDDQ usage
gender predict the likelihood of being classified into specific types. agreement, (b) using keyword searches in APA PsycNET and
Regarding age, theoretical accounts suggest that developmental Google Scholar, (c) inspecting reference lists of identified studies,
indecision-related types (i.e., associated with the lack of motivation or and (d) asking contacted researchers to distribute our request
information) should be more prevalent among younger individuals, among relevant colleagues. In what follows, we report how we
in contrast to chronic indecision-related types (i.e., associated with determined our sample size, all data exclusions (if any), all
emotional and personality difficulties) that should become more manipulations, and all measures in the study. As recommended
prevalent with age (Brown & Rector, 2008; Kelly & Pulver, 2003; (Nylund et al., 2007; Spurk et al., 2020), all available data sets
Osipow, 1999). These expectations, however, received only little resulting in a combined country sample of over 500 participants
support in previous research. For example, Di Fabio et al. (2015) were included in the analyses. Consequently, we pooled the analyzed
found that personality explains a larger percentage of the variance in data from 50 studies conducted by the authors (12 studies) or other
career indecision than age. Similarly, Levin et al. (2020) referred to research teams (38 studies). We verified that an institutional review
the relatively small effect sizes reported for age differences in the board had approved all studies (see Supplemental Material H for
literature and argued that specific stages and individual differences details on the sources and procedures of the included samples; data
likely influence career indecision more than age. In clustering are available at https://osf.io/fz2se/?view_only=98b8beac239c4aa
research, Levin et al. (2022) reported that age did not predict f8d24dc8f7e2e79f3, Levin et al., 2023; analysis codes are included
membership in career decision types. We thus expected that age in Supplemental Materials E1–E4). None of these studies applied
would not differentiate among indecision types. Regarding gender, clustering methods to identify career indecision types.
one of the most consistent gender differences reported in previous
studies is that men are more likely than women to lack motivation
Participants
for career decision making (Atitsogbe et al., 2018; Levin et al.,
2020). In clustering research, Meldahl and Muchinsky (1997) The 50 samples included data from 33,978 participants from
found that women, rather than men, were more likely to be classified 16 countries. In two countries, data were collected in two languages:
in types associated with decisiveness (Meldahl & Muchinsky, 1997), Canada (English and French) and Israel (Arabic and Hebrew). Thus,
but Levin et al. (2022) found that women were more likely than men participants were divided into 18 country-language samples. We
to be classified as indecisive than unmotivated. In light of these carried out a series of preliminary validity checks to ensure that only
findings, we hypothesized that gender would mainly emerge as a data of participants who completed the CDDQ with sufficient
predictor of the likelihood of being classified into types related to lack attention were analyzed. The data of 2,451 participants were
of motivation. excluded for several reasons: 745 (2.2%) due to incomplete CDDQ
data (one or more missing responses), 187 (0.6%) due to selecting further information about the psychometric properties of the
the same one or two responses, and 1,519 (4.5%) because their CDDQ, see Gati & Levin, 2014; Xu & Bhang, 2019).
responses to two validity items indicated questionable attention.
Across samples (N = 31,527), the age range of the 27,011 (85.7%)
participants with valid age data was 13–62 (M = 20.38 ± 6.18). Of
Career Decision Status
the 31,331 (99.1%) participants with gender data, 62.2% identified Nine studies (n = 7,888) included data on career decision status,
as women, 37.7% as men, and five participants indicated “mixed” measured with the RCA question (Saka et al., 2008). Participants
or “other” as their gender. Table 2 reports the characteristics of were asked to select one of six statements that best describes their
each of the 18 samples. current career decision status: (a) “I do not even have a general
direction,” (b) “I have only a general direction,” (c) “I am
deliberating among a small number of specific occupations,” (d) “I
Instruments am considering a specific occupation, but I would like to explore
The Career Decision-Making Difficulties Questionnaire other options before I make my decision,” (e) “I know which
occupation I am interested in, but I would like to feel sure of my

Participants’ causes of career indecision were assessed using nine choice,” and (f) “I am already sure of the occupation I want.” The
translations of the 34-item version of the CDDQ (Gati et al., 1996). RCA has been useful in measuring progress toward making a
The CDDQ assesses 10 causes of career indecision, grouped into career decision (Saka et al., 2008), measuring career decidedness
three major clusters: (a) lack of readiness (lack of motivation, general (Buzzetta et al., 2017), and assessing intervention effectiveness
indecisiveness, and dysfunctional beliefs); (b) lack of information (Buzzetta et al., 2017; Gati et al., 2013).
(lacking information about the career decision-making process, the
self, occupations, and ways of obtaining additional information);
(c) inconsistent information (unreliable information, internal con-
Decision Certainty
flicts, and external conflicts). Previous studies provided supporting Seventeen studies (n = 3,788) included data on participants’
evidence for the hypothesized factor structure, reliability, and decision certainty. Specifically, participants who had made an
measurement equivalence of the CDDQ, as measured by the nine educational or vocational choice were asked to indicate their
utilized translations across different countries and languages confidence in their choice on a 9-point Likert-type numerical
(Atitsogbe et al., 2018; Levin et al., 2020, 2023). The 34 items scale with word anchors only at the endpoints (very not confident
were presented on a 9-point Likert-type numerical scale with word to very confident). Hartung (1995) noted that measures of career
anchors only at the endpoints (does not describe me to describes indecision typically assess career certainty with one or two items.
me well). Scale scores are calculated as the means of items; higher Such an item is included in the CDDQ, but we located only one
scores indicate more severe decision-making difficulties (for published study that analyzed participants’ reported decision
Table 2
Sample Characteristics and LPA Solutions for the 18 Country-Language Samples
Sample characteristics LPA (within samples) Profile RFs (across samples)

Region Country Nparticipants Age Women | Mena K>5% Kfinal Entropy P1 P2 P3 P4 P5 P6 P7
Africa South Africa 1,472 21.46 ± 5.99 65% | 35% 6 6 0.83 9% 34% 19% 12% 10% 9% 7%
Togo 536 18.72 ± 2.63 43% | 57% 5 4 0.73 8% 34% 9% 22% 9% 15% 4%
America Canada (English) 4,045 25.44 ± 8.31 69% | 31% 6 6 0.80 5% 13% 35% 8% 17% 7% 15%
Canada (French) 877 17.40 ± 0.59 60% | 40% 9 6 0.87 7% 34% 17% 8% 10% 12% 11%
United States 794 20.03 ± 3.52 66% | 33% 5 5 0.85 7% 42% 12% 6% 13% 10% 10%
Asia China 964 23.31 ± 4.46 64% | 36% 8 7 0.79 12% 15% 27% 9% 18% 9% 9%
India 572 25.17 ± 7.62 31% | 37% 5 5 0.81 8% 23% 24% 7% 18% 17% 4%
Malaysia 945 18.92 ± 2.88 59% | 41% 7 6 0.81 8% 23% 26% 11% 14% 13% 6%
South Korea 1,319 19.20 ± 2.91 59% | 39% 4 4 0.77 4% 17% 25% 15% 25% 7% 7%
Australia Australia 4,600 18.12 ± 6.75 61% | 39% 7 7 0.78 8% 17% 33% 5% 16% 9% 12%
Europe Croatia 1,198 16.94 ± 1.47 66% | 33% 6 6 0.81 10% 29% 19% 19% 3% 8% 11%
France 1,805 18.78 ± 2.41 71% | 29% 7 7 0.77 14% 29% 23% 13% 5% 7% 9%
Greece 2,393 17.52 ± 3.67 64% | 36% 7 7 0.80 8% 30% 14% 28% 7% 8% 6%
Poland 639 18.47 ± 2.61 60% | 40% 7 7 0.82 16% 22% 17% 13% 11% 9% 11%
Switzerland 1,500 18.83 ± 4.27 56% | 44% 9 9 0.77 14% 24% 26% 12% 6% 6% 12%
Middle East Israel (Arabic) 1,203 17.04 ± 0.33 56% | 38% 5 5 0.86 3% 43% 18% 16% 10% 7% 3%
Israel (Hebrew) 3,678 23.47 ± 5.69 64% | 36% 9 7 0.81 2% 13% 43% 8% 13% 4% 16%
Turkey 2,987 15.37 ± 1.04 54% | 46% 5 5 0.82 4% 57% 6% 11% 7% 13% 2%
Note. LPA = latent profile analysis; K = number of profiles; RF = relative frequency; P1 = unmotivated; P2 = unrealistic; P3 = generally uninformed;
P4 = occupations-uninformed; P5 = conflicted-uninformed; P6 = externally conflicted; P7 = internally conflicted. Final LPA solutions with fewer profiles
than initially identified (based on minimal relative frequency) are presented in bold.
a
Percentages of women and men are only reported; four participants indicated “other” in Switzerland and one “mixed” in the United States; Detailed
information is provided in Supplemental Material H.
certainty. Specifically, among Greek students, the causes of in Supplemental Material E1). The models were estimated using
indecision explained 33% of the variance in participants’ decision 1,000 random start values to avoid model convergence on a local
certainty (Vaiopoulou et al., 2019). In addition, the test–retest maximum. We ensured that the maximal log-likelihood value was
reliability of a similar item was found to be r = .79 and was obtained at least 3 times.
associated with the congruence between vocational interests and We identified the optimal profile solutions by following a
choice (Tracey, 2010). multistep procedure (see details in Supplemental Material B). After
confirming there were no error messages, we located the largest k-
Career Decision Self-Efficacy profile solution with profiles of minimum relative frequency (RF)
of 5% to maximize the parsimony and meaningfulness of the
Eleven studies (n = 4,012) included data on participants’ career profiles. In doing so, we adhered to standard practices, rejecting
decision self-efficacy, measured by the Career Decision Self-Efficacy solutions that included profiles with fewer than 25 participants at
Scale–Short Form (CDSE-SF; Betz et al., 1996). The 25 items of the the level of each country-language sample (and above the
CDSE-SF measure participants’ confidence in five career decision- preferable rule of thumb of RF > 1%; Lubke & Neale, 2006;
making skills: accurate self-appraisal, gathering occupational informa- Spurk et al., 2020). As presented in the middle section of Table 2 in
tion, goal selection, making plans for the future, and problem solving. column K>5%, based on profile RFs, the initial solutions for each
The 25 items were presented on a 5-point Likert-type numerical scale country-language sample ranged between four and nine profiles
with word anchors only at the endpoints (no competence at all to each. Second, we inspected the Akaike’s information criterion
complete competence). We considered only the CDSE-SF total score in (AIC), Bayesian information criterion (BIC), and sample size–
the present study. Strong associations with career certainty, indecision, adjusted Bayesian information criterion (SABIC) values to ensure
and vocational identity supported the validity of the CDSE-SF (e.g., they decreased with the addition of profiles. BIC values increased
Betz et al., 1996; for a recent meta-analysis, see Udayar et al., 2020). in only the Togolese sample between the four- and five-profile
Betz et al. (1996) reported a reliability estimate of .94 for the total solutions, indicating the five-profile solution should be rejected in
CDSE-SF score. In the present study, the reliability of the total score this sample. Third, we examined the qualitative distinctiveness of
varied between .80 and .91 (Mdn = .88). profiles within the 18 country-language samples manually and by
clustering the centroids of the 115 retained profiles from the second
Results step across the 18 samples. These examinations revealed that four
samples included two or more profiles that differed quantitatively but
Analyses were conducted in five main stages. Following Levin
not qualitatively: Canada–French, China, Israel–Hebrew, and
et al. (2020), in the first stage, we computed 10 ipsative scores for
Malaysia; thus, in these samples, a more parsimonious solution
each participant (see details on the computation procedure in
with fewer profiles was preferred. Finally, we evaluated the
Supplemental Material C). Ipsative scores capture intraindividual
classification quality of the retained solutions. The chosen profile
differences and minimize response bias (Cheung & Chan, 2002).
The ipsative scores were then standardized across participants solutions—between four and nine profiles (see Table 2, column
and served as LPA indicators. In addition, each country-language Kfinal)—had entropy values within the acceptable range (Mdn = .81,
sample was randomly split into two subsamples to enable estimating range = .73–.87).1
the replicability of solutions. Across the 18 country-language samples, 10 different profiles
emerged. We labeled these 10 profiles according to their most salient
CDDQ cluster or scale scores: (a) lack of motivation, (b) dysfunctional
Multigroup Latent Profile Analysis Within beliefs, (c) lack of information, (d) lack of information about
Country-Language Samples occupations, (e) external conflicts-lack of information, (f) external
In the second analysis stage, we sought to identify the optimal conflicts, (g) general indecisiveness-unreliable information-internal
number of profiles in each of the 18 country-language samples. A conflicts, (h) general indecisiveness-lack of information about
challenge in LPA is determining the optimal profile solution, namely the self-internal conflicts, (i) general indecisiveness, and (j) lack
the number of profiles that provides the optimal classification of of information about the process and about the self. Columns K
individuals into groups. Three criteria often used for model selection and RF on the right in Table 3 present the number of samples in
are (a) the statistical adequacy, (b) the profiles’ meaning, and (c) the which these 10 profiles emerged and their RFs across samples,
solutions’ theoretical conformity (Hofmans et al., 2020; Spurk et al., respectively. Interestingly, as hypothesized, all country-language
2020). However, these criteria do not always indicate a clear
conclusion or a replicable solution. Thus, we integrated (d) 1
Compared to the analytical approach implemented in the main analyses,
replicability as an additional criterion for identifying the optimal the results of 18 series of one-sample LPA models (i.e., one for each country-
number of profiles by implementing multigroup LPA (Morin et al., language sample) were less informative for model selection. First, in these
alternative one-sample LPAs, AIC, BIC, and SABIC values consistently
2016) to directly estimate profile solutions while equating specific
decreased with the addition of profiles. Second, all 180 estimated models
parameters across the relevant samples. Specifically, to identify the provided significant bootstrapped likelihood ratio test values. Third, the
optimal profile solution in each country-language sample, we inspection of RFs supported choosing the same profile solution in 14 samples
conducted 18 series of multigroup LPA models, using Mplus 8.7, or selecting solutions with additional profiles in the Chinese, French, Greek,
specifying models with 2–10 profiles of distributional similarity and U.S. samples. Fourth, the estimated profile indicators were identical
across the solutions deemed optimal across the two analytical approaches.
between the two random subsamples. The distributional similarity is Thus, the estimation approach implemented in the main analyses either
estimated by constraining within-profile indicator means, variances, replicated the results of one-sample LPAs or supported choosing more
and profile frequencies (Morin et al., 2016; Mplus code is presented parsimonious solutions.
18
18
18
18
information; Ii = internal conflicts; Lp = lack of information about the decision-making process; Ls = lack of information about the self; RF = relative frequency; K = number of samples; LPA = latent
profile analysis. Values reflect column percentage (i.e., the relative frequency of individuals in each of the 10 profiles identified in individual LPA, separately for each of the seven profiles identified
Note. Rm = lack of motivation; Rd = dysfunctional beliefs; LI = lack of information; Lo = lack of information about occupations; Ie = external conflicts; Ri = general indecisiveness; Iu = unreliable
9
9
4
4
5
samples comprised four or more profiles, four of which were common
K
across all samples (Profiles 2, 3, 5, and 6).
4.09%
4.85%
8.95%
4.84%
2.82%
5.54%
2.49%
24.61%
29.65%
12.16%
RF
Multigroup Latent Profile Analysis Across
Country-Language Samples
In the third analysis stage, multigroup LPA was conducted
across the 18 country-language samples to identify an optimal and
conflicted
Internally
66.54%
59.55%
8.37%
2.47%
3.20%
2.68%
2.14%
2.16%
9.28%
24.77%
20.92%
replicable profile solution. Given expected variations in profile
sizes, 2–10 profile solutions of dispersion similarity were
estimated across the 18 samples by constraining only within-
profile indicator means and variances (Morin et al., 2016; Mplus
code is presented in Supplemental Material E2). Table 4 presents
the minimal profile RF, fit statistics, and classification quality

Externally
conflicted
81.53%
1.40%
0.89%
0.00%
0.13%
5.90%
1.38%
3.26%
0.97%
0.13%
8.51%
indices for these solutions. First, as Table 4 shows, solutions with
eight or more profiles included profiles with RFs of less than 5%,
thereby leading to their rejection. Second, we ensured that AIC,
BIC, and SABIC values consistently decreased with the addition
of profiles, which was confirmed across solutions. Third, we
Conflicted-uninformed
evaluated the qualitative distinctiveness of profiles, revealing that

the seven-profile solution encompassed qualitatively distinct
83.38%
0.23%
0.28%
1.33%
1.44%
4.32%
1.69%
1.20%
2.81%
11.80%
12.13%
profiles. Thus, after ensuring the adequacy of its classification

quality (entropy = .77), the seven-profile solution was determined
to be the best-fitting solution and retained for further analysis.
Figure 1 depicts the mean Z scores of the 10 CDDQ scales (i.e.,
across individuals and within variables) for each of the seven profiles.
Mean ipsative scores (i.e., within individuals and across variables) for
the seven-profile solution are reported in Supplemental Material D.
Across samples
Occupations-uninformed
The first and second profiles were labeled unmotivated (7.1%) and
Compatibility of Profiles Within (Rows) and Across (Columns) Country-Language Samples
unrealistic (25.8%) as they were high in lack of motivation and

76.85%
14.92%
11.73%
10.57%
2.56%
3.20%
3.52%
0.32%
7.20%
0.11%
7.78%
dysfunctional beliefs, respectively. The next two identified profiles

were characterized by lacking information. Specifically, individuals
in the third profile, labeled generally uninformed (25.2%), reported
lacking information in all four domains comprising lack of
information (the decision-making process, the self, occupations,
and ways of obtaining additional information). The fourth profile,
labeled occupations-uninformed (11.5%), was mostly character-
ized by lacking information about occupations. Last, the final three
uninformed
Generally
75.80%
6.90%
0.28%
3.29%
0.00%
7.79%
6.01%
12.56%
20.56%
38.52%
26.08%
identified profiles were characterized by difficulties related to

internal or external conflicts. The most pronounced difficulty of
individuals in the fifth (12.0%) and sixth (8.5%) profiles, labeled
across samples). Frequencies above 50% are presented in bold.
conflicted-uninformed and externally conflicted, respectively,

involved external conflicts. However, in addition to experiencing
Unrealistic
89.46%
external conflicts, individuals in the conflicted-uninformed profile

42.68%
26.45%
7.36%
2.00%
4.45%
1.20%
2.80%
3.41%
5.84%
9.18%
also reported lacking information in the four information domains,

relying on unreliable information, and enduring internal conflicts.
Finally, individuals in the seventh profile, labeled internally
conflicted (9.9%), reported high internal conflicts, unreliable
Unmotivated
information, and general indecisiveness. The right side of Table 2

73.18%
3.42%
2.75%
1.90%
0.57%
1.38%
9.36%
8.99%
6.98%
16.59%
16.71%
presents profile RFs separately for each of the 18 country-language

samples.
To further evaluate the classification quality of the seven-profile
solution, we calculated the RFs of individuals in each of the seven
profiles identified across samples, separately for each of the 10 profiles
identified within samples. Table 3 shows that when comparing the
Within samples
solutions within and across samples, most individuals were cross-

classified into the same groups (RFmean = 78.1%). For example, 75.8%
Table 3
Ri/Ls/Ii
Ri/Iu/Ii
Lp/Ls
of those classified as lack of information in the within-samples

Ie/LI
Rm
RF
Rd
Lo
Ri
LI
analyses were assigned to its equivalent generally uninformed profile

Ie
Table 4
Fit Indices and Classification Quality Indicators for the LPA Models Across Samples
Fit statistics Classification quality

K RFmin DF LL AIC BIC SABIC Entropy MPCPmin
2 44.49% 65 −519768.99 1039667.97 1040211.28 1040004.71 .72 .91

3 18.33% 93 −514623.62 1029433.25 1030210.60 1029915.04 .80 .90
4 9.76% 121 −512342.10 1024699.25 1025710.64 1025326.11 .82 .84
5 9.06% 149 −509814.56 1019927.11 1021172.54 1020699.02 .78 .78
6 7.66% 177 −508143.71 1016641.42 1018120.89 1017558.39 .79 .76
7 7.12% 205 −506881.79 1014173.58 1015887.09 1015235.60 .77 .73
8 4.95% 233 −505639.32 1011744.64 1013692.19 1012951.72 .77 .73
9 1.70% 261 −504521.24 1009564.49 1011746.08 1010916.63 .78 .73
10 2.69% 289 −503479.14 1007536.28 1009951.92 1009033.48 .77 .71
Note. LPA = latent profile analysis; K = number of profiles; RF = relative frequency; DF = degrees of freedom; LL = log likelihood; AIC = Akaike’s
information criterion; BIC = Bayesian information criterion; SABIC = sample size–adjusted Bayesian information criterion; MPCP = mean posterior
classification probability. The selected profile solution is presented in bold.
in the across-samples analyses. Fluctuations among profiles mostly Mplus code is presented in Supplemental Material E3). Table 5
occurred in profiles identified in only a few country-language samples shows the results of the multinomial logistic regressions for the
(see the right-most column in Table 3). associations with age and gender, on the one hand, and profile
classification, on the other hand. Although age emerged as a
statistically significant predictor, none of the 21 comparisons
Predictors of Career Indecision Types
indicated it as a meaningful predictor of profile membership
The fourth analysis stage examined age and gender as predictors regarding effect size (OR < 0.60 or OR > 1.68). Then, for gender,
of profile membership. We employed multinomial logistic regres- identifying as a man was associated with a greater likelihood of
sion to test age and gender as predictors of profile classification being classified as unmotivated than all other profiles (ORrange =
using the three-step approach for LPA with covariates and the 1.72–3.13). Indeed, 9.4% of men were classified as unmotivated
auxiliary R3STEP option in Mplus (Asparouhov & Muthén, 2014; compared with 5.4% of women. In addition, identifying as a
Figure 1
Standardized Means of the 10 CDDQ Scales for the Seven Profiles Across Samples
Note. Rm = lack of motivation; Ri = general indecisiveness; Rd = dysfunctional beliefs; Lp = lack of information about the decision-making
process; Ls = lack of information about the self; Lo = lack of information about occupations; La = lack of information about ways of obtaining
additional information; Iu = unreliable information; Ii = internal conflicts; Ie = external conflicts; CDDQ = Career Decision-Making Difficulties
Questionnaire.
Table 5
Results From the Categorical Latent Variable Multinomial Logistic Regression of the Associations of Age and Gender With Profiles
1 versus 2 1 versus 3 1 versus 4 1 versus 5 1 versus 6

Predictor Coef. OR Coef. OR Coef. OR Coef. OR Coef. OR
+ +
Age −0.04*** 0.96 0.02** 1.02 −0.02* 0.98 0.01 1.01 −0.01 0.99
Gender 0.58*** 1.80 0.77*** 2.16 0.97*** 2.65 0.86*** 2.36 0.54*** 1.72

Coef. OR Coef. OR Coef. OR Coef. OR Coef. OR
Age 0.04*** 1.04 0.06*** 1.08 0.02** 1.02 0.05*** 1.05 0.03*** 1.03
Gender 1.14*** 3.13 0.19*** 1.20 0.39*** 1.47 0.28*** 1.32 −0.04+ 0.96

Age 0.08*** 1.08 −0.04*** 0.96 −0.01** 0.99 −0.03*** 0.97 0.02*** 1.02
Gender 0.56*** 1.74 0.20** 1.22 0.09+ 1.09 −0.23*** 0.80 0.37*** 1.45

+
Age 0.03*** 1.03 0.01 1.01 0.06*** 1.06 −0.02** 0.98 0.03*** 1.03
Gender −0.11+ 0.89 −0.43*** 0.65 0.17+ 1.18 −0.32*** 0.73 0.28*** 1.32
6 versus 7
Coef. OR
Age 0.05*** 1.05

Gender 0.60*** 1.82
Note. ns = nonsignificant difference; Coef. = logit coefficient; OR = odds ratio. Gender was coded as a dichotomous variable (1 = man, 2 = woman).
Profile 1 = unmotivated; Profile 2 = unrealistic; Profile 3 = generally uninformed; Profile 4 = occupations-uninformed; Profile 5 = conflicted-uninformed;
Profile 6 = externally conflicted; Profile 7 = internally conflicted. ORs < 0.60 and ORs > 1.68 are presented in bold.
* p < .05. ** p < .01. *** p < .001. + ns.
man was associated with a greater likelihood of being classified as Similar differences across the profiles were also found in the
unrealistic than internally conflicted (OR = 1.74) and externally mean levels of decision certainty, χ2(df = 6) = 1,000.58, p < .001,
conflicted than internally conflicted (OR = 1.82). All other associations and career decision self-efficacy, χ2(df = 6) = 607.25, p < .001. As
with gender were negligible in terms of effect size (0.87 < ORs < 1.36). Figure 2b shows, decision certainty was significantly lower in
A series of multigroup confirmatory analyses provided support for the generally uninformed profile and significantly higher in the
the measurement invariance of the CDDQ across gender at the scalar unrealistic profile than in all other profiles. Among the remaining
level (see Supplemental Material F). five profiles, decision certainty was the lowest in the conflicted-
uninformed and internally conflicted profiles. Decision certainty in
the unmotivated and externally conflicted profiles was lower than
Outcomes of Career Indecision Types
in the occupations-uninformed profile. Finally, for career decision
The fifth and final analysis stage examined how the seven profiles self-efficacy, as Figure 2c shows, career decision self-efficacy was
differed in career decision status, decision certainty, and career significantly lower in the generally uninformed profile and significantly
decision self-efficacy. To this end, we tested an LPA model with higher in the unrealistic profile than in all other profiles. Career
distal outcomes using the Bolck–Croon–Hagenaars method (BCH; decision self-efficacy was lower in the conflicted-uninformed
Bakk & Vermunt, 2016; Mplus code is presented in Supplemental profile than in the remaining four profiles, which, in turn, did not
Material E4) via the BCH function in Mplus. As Figure 2a shows, significantly differ from one another. Taken together, these
career decision status differed among the seven types, χ2(df = 6) = findings support our hypotheses that profiles associated with a lack
1,583.92, p < .001. Specifically, paired comparisons revealed that of information will score lowest on the three decidedness outcomes
career decision status was significantly lower in the generally and that profiles associated with external or internal conflicts
uninformed profile and significantly higher in the unrealistic profile would also yield low levels in these outcomes.
than in all other profiles. Among the remaining five profiles, career
decision status was the lowest in the conflicted-uninformed profile.
Discussion
Then, the career decision status in the internally conflicted and
unmotivated profiles was lower than in the externally conflicted, The overarching aim of this study was to advance the
which, in turn, was lower than in the occupations-uninformed profile. development of a typology of career indecision that enables a
Figure 2
Mean Level Differences of Career Decision Status (N = 7,888), Certainty (N = 3,788), and Self-Efficacy (N = 4,012) by Profiles
Note. GU = generally uninformed; CU = conflicted-uninformed; IC = internally conflicted; UM = unmotivated; EC = externally conflicted;

OU = occupations-uninformed; UR = unrealistic; n.s. = nonsignificant difference.
differential diagnosis of career problems and is applicable and indecisive type (experiencing general indecisiveness) emerged in the
generalizable to diverse populations. As differential diagnosis present study in only four country-language samples. Individuals
entails classifying individuals into types based on a validated classified as indecisive in these four samples were assigned in the
procedure (Campbell & Cellini, 1981), we examined which types across-samples analyses to the unrealistic (43%), internally conflicted
could be discerned using the CDDQ (Gati et al., 1996) both (25%), or unmotivated (17%) types across samples (see Table 3,
within and across 18 country-language samples from countries row Ri), suggesting that indecisiveness is often linked to other self-
on five continents. Across samples, the best fit-to-data solution related causes of indecision. Then, we identified three types that
comprised seven types: unmotivated, unrealistic, generally unin- were not reported by Levin et al. (2022): the occupations-
formed, occupations-uninformed, conflicted-uninformed, externally uninformed, externally conflicted, and internally conflicted types.
conflicted, and internally conflicted. Age did not emerge as a Applying multigroup LPA in the present study likely enabled
meaningful predictor of types, but gender repeatedly predicted the identifying these additional indecision types.
likelihood of being classified as unmotivated compared to all other The findings of the present study also complement previous
types. Finally, a similar pattern of differences among the types studies clustering individuals to derive career indecision types based
emerged in three career decidedness outcomes: career decision status, on either single measures (e.g., Chartrand et al., 1994; Guay et al.,
decision certainty, and career decision self-efficacy. 2006; Savickas & Jarjoura, 1991) or multiple measures (e.g., Kelly
& Pulver, 2003; Lucas & Epperson, 1990; Santos & Ferreira, 2012).
Some studies advanced a threefold typology that included a decided
The Seven Career Indecision Types
type, a developmentally undecided type (typically characterized
Our findings demonstrated that seven career indecision types can by the lack of motivation or information), and a chronic indecisive
be identified using the 10 causes of career indecision measured by type (characterized by emotional or personality-related causes of
the CDDQ. Five types were each characterized by a single main indecision; Feldt et al., 2011; Guay et al., 2006; Santos & Ferreira,
cause of indecision: lack of motivation for unmotivated, dysfunc- 2012). This threefold typology is compatible with the results of our
tional beliefs for unrealistic, overall lack of information for within-sample analyses revealing that four types emerged across all
generally uninformed, lack of information about occupations for samples (see Table 3, column K): the unrealistic, generally uninformed,
occupations-uninformed, and external conflicts for externally conflicted-uninformed, and externally conflicted types. Further indicat-
conflicted. In addition, two types were characterized by multiple ing the lack of information and interpersonal conflicts as two major
causes of indecision: lack of information and external conflicts for differentiating factors among career indecision types (Xu & Bhang,
conflicted-uninformed and internal conflicts, unreliable information, 2019), these results also highlight dysfunctional beliefs as another
and general indecisiveness for internally conflicted. Descriptions of important differentiating factor to consider in future research and
these seven types and their similarities to types reported in previous practice (Xu, 2022).
literature are provided in Supplemental Material G.
Four of the seven career indecision types identified across countries
Predictors and Outcomes of Career Indecision Types
in the present study matched Levin et al.’s (2022) CDDQ-based five
types detected in two United States samples: the unmotivated, In the present study, we examined whether age and gender are
unrealistic, generally uninformed (uninformed in Levin et al., 2022), meaningful predictors of type membership. Considering the
and conflicted-uninformed (conflicted in Levin et al., 2022) types. substantial differences in sample characteristics (see Table 2),
Interestingly, in the present study, Levin et al.’s (2022) five types also we did not investigate whether one’s country is predictive of type
emerged in and replicated the five-profile solution for the United membership. However, our application of multigroup LPA
States sample (see Table 2). Nevertheless, Levin et al.’s (2022) ensured that the seven career indecision types identified in the
present study were applicable to individuals from all considered introducing confounds for which no information was available
countries. Notwithstanding recent research on career indecision and thus could not be controlled. Future research examining the
that has provided ample support for the cross-cultural equivalence prevalence of types among groups should ensure the equivalence of
of career indecision as measured by contemporary assessments the compared groups (e.g., applying the same study design to
(Levin et al., 2020, 2023; Xu & He, 2022), future research is compare groups of high school students from different countries).
needed to examine whether cultural contexts are predictive of the Second, most retrieved data sets were part of cross-sectional studies,
prevalence of career indecision types. thus reflecting the scarcity of longitudinal research that can better
Aligning with previous studies (e.g., Levin et al., 2022; Rojewski, capture causal relations. In our case, we validated the types using
1994), age did not emerge in the present study as a meaningful three concurrently measured outcomes, thus limiting the under-
predictor of types. This finding does not support the hypothesis of standing of the consequences of experiencing specific indecision
age being a significant predictor of developmental indecision- patterns over time. For example, individuals classified as unrealistic
related types (Brown & Rector, 2008; Osipow, 1999), namely those appear to be highly decided, certain of their choices, and confident in
characterized by a lack of motivation or information (Guay et al., their decision-making skills. However, they may be later at greater
2006; Kelly & Pulver, 2003; Santos & Ferreira, 2012). Rather, this risk of experiencing regret or a low degree of commitment
finding supports the claim that indecision is related more to specific (Xu, 2022). Future studies should employ a longitudinal design to
transitions than to age-related developmental patterns (Levin test the temporal stability of the types and their consequences over
et al., 2020). time. Finally, the underlying goal of identifying indecision types
For gender, previous studies have reported an absence of or is to better understand diverse career problems and how they can
merely negligible gender differences in career indecision (Atitsogbe be treated. However, because research on previous indecision
et al., 2018; Levin et al., 2020). Nevertheless, in studies that did find typologies failed to demonstrate the differential effectiveness of
such differences, among the most consistent finding was that men interventions across types (e.g., Kelly & Pulver, 2003; Milot-
are more likely to lack motivation than women (Levin et al., 2020, Lapointe et al., 2022), future studies are needed to evaluate the
2022; Meldahl & Muchinsky, 1997). Respectively, in the present effectiveness of interventions tailored to the unique needs of
study, the most robust gender difference was that men rather than specific indecision types.
women were almost twice as likely to be classified as unmotivated.
This finding complements previous studies indicating that women
Implications for Research and Practice
tend to invest more effort in the career decision-making process than
men (Gati et al., 2013; Levin et al., 2020). On a methodological level, one of the most challenging aspects to
To better differentiate the causes of career indecision from its reconcile in previous clustering research (on career indecision and in
consequences and to further validate the derived typology, we general) is that types identified in one study only partially overlap
also investigated how the seven emerged types differed in three with types identified in other studies. Indeed, in various studies,
decidedness outcomes: career decision status, decision certainty, different types emerged even when the same set of variables were
and career decision self-efficacy. For career decision status, two clustered (e.g., in the case of the CDS, see Argyropoulou et al., 2007;
lack of information-related types (generally uninformed, con- Guay et al., 2006; Rojewski, 1994; Savickas & Jarjoura, 1991). For
flicted-uninformed) included a larger proportion of undecided this reason, we argue that integrating replicability as a criterion when
individuals. In comparison, the internally conflicted and unmoti- identifying an optimal clustering solution is crucial. Accordingly, we
vated types were more decided but less decided than the externally utilized multigroup LPA to estimate replicable solutions, comparing
conflicted and occupations-uninformed types, and especially less profile solutions both within and across 18 country-language samples.
than the unrealistic type. Based on their career decision status, the In doing so, the percentage of individuals cross-classified in the
seven types were arranged into four statistically different groups same seven identified types ranged from 66% to 89% (M = 78%;
of types (see Figure 2a). However, in terms of decision certainty, see Table 3). In comparison, in a study on organizational commitment
the seven types were arranged into five groups (see Figure 2b), employing multilevel LPA (Kabins et al., 2016), the percentage of
indicating that career indecision types explained more variability individuals cross-classified in the same five types ranged from 14% to
of decision certainty than of career decision status. Still, the overall 66% (M = 46%), reflecting less stable classifications. This comparison
pattern of differences among the seven career indecision types illuminates the advantage of multigroup LPA over multilevel LPA for
in career decision status and decision certainty was highly similar deriving more stable and replicable classifications.
(r = .96). Finally, for career decision self-efficacy, only three In the career indecision literature, efforts to differentiate between
groups of types emerged (see Figure 2c), but, again, the pattern of transitory developmental indecision and chronic indecision continue
differences in self-efficacy was highly similar to that of career to occupy the counseling field (Gati & Levin, 2014; Kelly & Pulver,
decision status (r = .97). Thus, the seven types differed on the three 2003; Xu & Bhang, 2019). Our findings challenge this dichotomous
decidedness outcomes most similarly, with decision certainty differentiation in several ways. First, for developmental indecision,
offering the most nuanced measurement of decidedness. three different types corresponding to previous depictions of
developmental indecision (Guay et al., 2006; Kelly & Pulver,
2003; Santos & Ferreira, 2012) were identified in the present study
Limitations and Future Research
(unmotivated, generally uninformed, and occupation uninformed).
Before discussing the implications of our research, its limitations Similarly, three different types corresponding to previous depictions of
should be acknowledged. First, to examine age and gender as chronic indecision (Kelly & Pulver, 2003; Lucas & Epperson, 1990;
predictors of types, we relied on data compiled from 50 data sets. Santos & Ferreira, 2012) emerged in the present study (conflicted-
These data sets were collected using different study designs, likely uninformed, internally conflicted, and externally conflicted). Second,
age did not emerge as a significant predictor of types, thereby preferred direction), emotional (e.g., feeling confident and committed
challenging the developmental explanation underlying the to the choice), and behavioral (e.g., implementing the decision).
differentiation between developmental and chronic indecision.
Third, in the present study, one quarter of individuals were classified
as unrealistic, a type comprising many decided individuals with Conclusions
dysfunctional beliefs about the world of work or the career decision- Seven types capture the variations among individuals in the
making process. In this respect, the dichotomous differentiation patterns of the causes of their career indecision as measured by the
between developmental indecision and chronic indecision appears to CDDQ across countries, age, and genders. These career indecision
overlook both the potential needs of seemingly decided individuals types include both undecided and decided individuals. Diagnosing
and dysfunctional beliefs as a frequent cause of indecision. Taken each client’s career indecision type can promote intervention
together, these findings may suggest that the differentiation between planning regardless of whether clients have declared a preferred
developmental and chronic indecision would be better explained career alternative. However, further research is needed to determine
by the measured causes of career indecision (i.e., cognitive vs. which interventions suit each type. Providing counseling tailored
emotional; Feldt et al., 2011; Santos & Ferreira, 2012) than by age to clients’ career indecision type could facilitate helping them
differences. As such, these results advocate adopting a more advance in the career decision-making process, strengthening clients’
nuanced approach to career indecision that considers multiple confidence and commitment, and increasing clients’ satisfaction
types of indecision beyond the dichotomy of developmental with their career choices.
indecision versus chronic indecision.
Moreover, some studies argued that career indecision could be
experienced only by undecided individuals (Larson et al., 1988; References
Lucas & Epperson, 1990), whereas others viewed decidedness as
a differentiating factor among various types (Guay et al., 2006; Akos, P., Konold, T., & Niles, S. G. (2004). A career readiness typology and
typal membership in middle school. The Career Development Quarterly,
Multon et al., 2007; Santos & Ferreira, 2012; Wanberg & Muchinsky,
53(1), 53–66. https://doi.org/10.1002/j.2161-0045.2004.tb00655.x
1992). Our results systematically revealed that in terms of three
Argyropoulou, E. P., Sidiropoulou-Dimakakou, D., & Besevegis, E. G.
measured outcomes, all seven types included both undecided and (2007). Generalized self-efficacy, coping, career indecision, and vocational
decided individuals, thereby underscoring the utility of career choices of senior high school students in Greece: Implications for career
indecision as a construct to consider among ostensibly decided guidance practitioners. Journal of Career Development, 33(4), 316–337.
individuals. Moreover, our findings that the generally uninformed https://doi.org/10.1177/0894845307300412
and conflicted-uninformed types were consistently the least Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture
decided imply that exploration activities are among the first steps modeling: Three-step approaches using Mplus. Structural Equation
to becoming more decided, aligning with previous results (Levin Modeling, 21(3), 329–341. https://doi.org/10.1080/10705511.2014.915181
et al., 2022). The finding that individuals in the occupations- Atitsogbe, K. A., Moumoula, I. A., Rochat, S., Antonietti, J.-P., & Rossier, J.
uninformed type, lacking information mainly about occupations, were (2018). Vocational interests and career indecision in Switzerland and
Burkina Faso: Cross-cultural similarities and differences. Journal of
among the most decided suggests that dealing with self-related issues
Vocational Behavior, 107, 126–140. https://doi.org/10.1016/j.jvb.2018
(e.g., self-exploration and internal conflict) typically precedes the .04.002
exploration of the world of work in career decision making. Bakk, Z., & Vermunt, J. K. (2016). Robustness of stepwise latent class
Finally, our findings also support the utility of distinguishing modeling with continuous distal outcomes. Structural Equation Modeling,
between the causes of indecision and its consequences. From a 23(1), 20–31. https://doi.org/10.1080/10705511.2014.955104
theoretical standpoint, Gati et al. (1996) hypothesized that difficulties Betz, N. E., Klein, K. L., & Taylor, K. M. (1996). Evaluation of a short form
related to lack of readiness (e.g., lack of motivation, general of the Career Decision-Making Self-Efficacy Scale. Journal of Career
indecisiveness, and dysfunctional beliefs) while individuals are still Assessment, 4(1), 47–57. https://doi.org/10.1177/106907279600400103
relatively undecided are more likely to emerge before embarking on Brown, S. D., & Rector, C. C. (2008). Conceptualizing and diagnosing
the decision-making process. Our findings, however, showed that problems in career decision-making. In S. D. Brown & R. W. Lent (Eds.),
Handbook of counseling psychology (4th ed., pp. 392–407). Wiley.
indecision types characterized by a lack of readiness could present
Buzzetta, M. E., Lenz, J. G., & Kennelly, E. (2017). Comparing two groups
themselves at any stage of the decision-making process (e.g., the of student-athletes: Implications for academic and career advising.
unmotivated type) or among highly decided individuals (e.g., the NACADA Journal, 37(1), 26–36. https://doi.org/10.12930/NACADA-
unrealistic type). Gati et al. (1996) suggested that career indecision 15-041
represents a group of problems leading to “the same final outcome Campbell, R. E., & Cellini, J. V. (1981). A diagnostic taxonomy of adult
(i.e., the inability to make a career decision)” (p. 521), a view shared career problems. Journal of Vocational Behavior, 19(2), 175–190. https://
by the dual-process theory of career decision making (Xu, 2022). doi.org/10.1016/0001-8791(81)90057-9
However, our results align better with Campbell and Cellini’s (1981) Chartrand, J. M., Martin, W. F., Robbins, S. B., McAuliffe, G. J., Pickering,
diagnostic taxonomy of career problems, differentiating between J. W., & Calliotte, J. A. (1994). Testing a level versus an interactional
view of career indecision. Journal of Career Assessment, 2(1), 55–69.
career decision making and implementation problems. Similarly, the
https://doi.org/10.1177/106907279400200106
current findings align with Osipow’s (1999) postulation that
Cheung, M. W.-L., & Chan, W. (2002). Reducing uniform response bias
indecision could be experienced by decided but insufficiently with ipsative measurement in multiple-group confirmatory factor analysis.
committed individuals. Therefore, advancing our understanding of Structural Equation Modeling, 9(1), 55–77. https://doi.org/10.1207/S1532
career indecision and its subtypes would benefit from developing 8007SEM0901_4
more complex, multidimensional conceptualizations of decidedness, Di Fabio, A., Palazzeschi, L., Levin, N., & Gati, I. (2015). The role of
encompassing three key components—cognitive (e.g., identifying a personality in the career decision-making difficulties of Italian young
adults. Journal of Career Assessment, 23(2), 281–293. https://doi.org/10 Behavioral Research, 41(4), 499–532. https://doi.org/10.1207/s15327906
.1177/1069072714535031 mbr4104_4
Feldt, R. C., Ferry, A., Bullock, M., Camarotti-Carvalho, A., Collingwood, Lucas, M. S., & Epperson, D. (1988). Personality types in vocationally
M., Eilers, S., Meyer, L., & Nurre, E. (2011). Personality, career undecided students. Journal of College Student Development, 29(5),
indecision, and college adjustment in the first semester. Individual 460–466.
Differences Research, 9(2), 107–114. Lucas, M. S., & Epperson, D. L. (1990). Types of vocational undecidedness:
Gati, I., Krausz, M., & Osipow, S. H. (1996). A taxonomy of difficulties A replication and refinement. Journal of Counseling Psychology, 37(4),
in career decision making. Journal of Counseling Psychology, 43(4), 382–388. https://doi.org/10.1037/0022-0167.37.4.382
510–526. https://doi.org/10.1037/0022-0167.43.4.510 Meldahl, J. M., & Muchinsky, P. M. (1997). The neurotic dimension
Gati, I., & Levin, N. (2014). Counseling for career decision-making of vocational indecision: Gender comparability? Journal of Career
difficulties: Measures and methods. The Career Development Quarterly, Assessment, 5(3), 317–331. https://doi.org/10.1177/1069072797005
62(2), 98–113. https://doi.org/10.1002/j.2161-0045.2014.00073.x 00305
Gati, I., Ryzhik, T., & Vertsberger, D. (2013). Preparing young veterans for Milot-Lapointe, F., Le Corff, Y., & Savard, R. (2022). Processes and
civilian life: The effects of a workshop on career decision-making outcomes of individual career counselling for different types of clients:
difficulties and self-efficacy. Journal of Vocational Behavior, 83(3), 373– A latent profile analysis replication and extension. British Journal of
385. https://doi.org/10.1016/j.jvb.2013.06.001 Guidance & Counselling, 50(5), 677–686. https://doi.org/10.1080/03069

Gordon, V. N. (1998). Career decidedness types: A literature review. The 885.2021.1959518
Career Development Quarterly, 46(4), 386–403. https://doi.org/10.1002/ Morin, A. J. S., Meyer, J. P., Creusier, J., & Biétry, F. (2016). Multiple-group
j.2161-0045.1998.tb00715.x analysis of similarity in latent profile solutions. Organizational Research
Guay, F., Ratelle, C. F., Senécal, C., Larose, S., & Deschênes, A. (2006). Methods, 19(2), 231–254. https://doi.org/10.1177/1094428115621148
Distinguishing developmental from chronic career indecision: Self- Multon, K. D., Wood, R., Heppner, M. J., & Gysbers, N. C. (2007). A
efficacy, autonomy, and social support. Journal of Career Assessment, cluster-analytic investigation of subtypes of adult career counseling
14(2), 235–251. https://doi.org/10.1177/1069072705283975 clients: Toward a taxonomy of career problems. Journal of Career
Hartung, P. J. (1995). Assessing career certainty and choice status. ERIC Assessment, 15(1), 66–86. https://doi.org/10.1177/1069072706294508
Digest. https://eric.ed.gov/?id=ED391107 Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the
Hofmans, J., Wille, B., & Schreurs, B. (2020). Person-centered methods in number of classes in Latent Class Analysis and Growth Mixture Modeling:
vocational research. Journal of Vocational Behavior, 118, Article 103398. A Monte Carlo simulation study. Structural Equation Modeling, 14(4),
https://doi.org/10.1016/j.jvb.2020.103398 535–569. https://doi.org/10.1080/10705510701575396
Jones, L. K., & Lohmann, R. C. (1998). The Career Decision Profile: Using Osipow, S. H. (1999). Assessing career indecision. Journal of Vocational
a measure of career decision status in counseling. Journal of Career Behavior, 55(1), 147–154. https://doi.org/10.1006/jvbe.1999.1704
Assessment, 6(2), 209–230. https://doi.org/10.1177/106907279800600207 Osipow, S. H., Carney, C. G., & Barak, A. (1976). A scale of educational-
Kabins, A. H., Xu, X., Bergman, M. E., Berry, C. M., & Willson, V. L. vocational undecidedness: A typological approach. Journal of Vocational
(2016). A profile of profiles: A meta-analysis of the nomological net of Behavior, 9(2), 233–243. https://doi.org/10.1016/0001-8791(76)90081-6
commitment profiles. Journal of Applied Psychology, 101(6), 881–904. Osipow, S. H., & Winer, J. L. (1996). The use of the Career Decision Scale
https://doi.org/10.1037/apl0000091 in career assessment. Journal of Career Assessment, 4(2), 117–130. https://
Kelly, K. R., & Pulver, C. A. (2003). Refining measurement of career doi.org/10.1177/106907279600400201
indecision types: A validity study. Journal of Counseling and Development, Rochlen, A. B., Milburn, L., & Hill, C. E. (2004). Examining the process and
81(4), 445–453. https://doi.org/10.1002/j.1556-6678.2003.tb00271.x outcome of career counseling for different types of career counseling
Krieshok, T. S. (1998). An anti-introspectivist view of career decision clients. Journal of Career Development, 30(4), 263–275. https://doi.org/
making. The Career Development Quarterly, 46(3), 210–229. https:// 10.1177/089484530403000403
doi.org/10.1002/j.2161-0045.1998.tb00697.x Rojewski, J. W. (1994). Career indecision types for rural adolescents
Larson, L. M., Heppner, P. P., Ham, T., & Dugan, K. (1988). Investigating from disadvantaged and nondisadvantaged backgrounds. Journal
multiple subtypes of career indecision through cluster analysis. Journal of Counseling Psychology, 41(3), 356–363. https://doi.org/10.1037/
of Counseling Psychology, 35(4), 439–446. https://doi.org/10.1037/0022- 0022-0167.41.3.356
0167.35.4.439 Saka, N., Gati, I., & Kelly, K. R. (2008). Emotional and personality-related
Levin, N., Braunstein-Bercovitz, H., Lipshits-Braziler, Y., Gati, I., & aspects of career-decision-making difficulties. Journal of Career
Rossier, J. (2020). Testing the structure of the Career Decision-Making Assessment, 16(4), 403–424. https://doi.org/10.1177/1069072708318900
Difficulties Questionnaire across country, gender, age, and decision status. Santos, P. J., & Ferreira, J. A. (2012). Career decision statuses among
Journal of Vocational Behavior, 116(Pt. A), Article 103365. https:// Portuguese secondary school students: A cluster analytical approach.
doi.org/10.1016/j.jvb.2019.103365 Journal of Career Assessment, 20(2), 166–181. https://doi.org/10.1177/
Levin, N., Lipshits-Braziler, Y., & Gati, I. (2022). The identification 1069072711420853
and validation of five types of career indecision: A latent profile analysis Savickas, M. L., & Jarjoura, D. (1991). The Career Decision Scale as a
of career decision-making difficulties. Journal of Counseling Psychology, type indicator. Journal of Counseling Psychology, 38(1), 85–90. https://
69(4), 452–462. https://doi.org/10.1037/cou0000603 doi.org/10.1037/0022-0167.38.1.85
Levin, N., Lipshits-Braziler, Y., & Gati, I. (2023). Patterns of career Spurk, D., Hirschi, A., Wang, M., Valero, D., & Kauffeld, S. (2020). Latent
decision-making difficulties in 16 countries: A person-centered investiga- profile analysis: A review and “how to” guide of its application within
tion. https://doi.org/10.17605/OSF.IO/FZ2SE vocational behavior research. Journal of Vocational Behavior, 120,
Levin, N., Udayar, S., Lipshits-Braziler, Y., Gati, I., & Rossier, J. (2023). Article 103445. https://doi.org/10.1016/j.jvb.2020.103445
The structure of the Career Decision-Making Difficulties Questionnaire Tracey, T. J. G. (2010). Relation of interest and self-efficacy occupational
across 13 countries. Journal of Career Assessment, 31(1), 129–148. https:// congruence and career choice certainty. Journal of Vocational Behavior,
doi.org/10.1177/10690727221099226 76(3), 441–447. https://doi.org/10.1016/j.jvb.2009.10.013
Lubke, G., & Neale, M. C. (2006). Distinguishing between latent classes and Udayar, S., Levin, N., Lipshits-Braziler, Y., Rochat, S., Di Fabio, A., Gati, I.,
continuous factors: Resolution by maximum likelihood? Multivariate Sovet, L., & Rossier, J. (2020). Difficulties in career decision making
and self-evaluations: A meta-analysis. Journal of Career Assessment, Journal of Career Assessment, 30(2), 221–237. https://doi.org/10.1177/
28(4), 608–635. https://doi.org/10.1177/1069072720910089 10690727211036887
Vaiopoulou, J., Papavassiliou-Alexiou, I., & Stamovlasis, D. (2019). Career Xu, H., & Bhang, C. H. (2019). The structure and measurement of career
decision-making difficulties and decision statuses among Greek student indecision: A critical review. The Career Development Quarterly, 67(1),
teachers. Hellenic Journal of Psychology, 16(1), 74–94. 2–20. https://doi.org/10.1002/cdq.12159
Wanberg, C. R., & Muchinsky, P. M. (1992). A typology of career decision Xu, H., & He, R. (2022). Facilitating international research on career
status: Validity extension of the vocational decision status model. Journal indecision: Developing Career Indecision Profile-Short-5 in China and
of Counseling Psychology, 39(1), 71–80. https://doi.org/10.1037/0022- the U.S. Journal of Career Assessment, 30(4), 719–738. https://doi.org/10
0167.39.1.71 .1177/10690727221080449
Willner, T., Gati, I., & Guan, Y. (2015). Career decision-making profiles and
career decision-making difficulties: A cross-cultural comparison among
US, Israeli, and Chinese samples. Journal of Vocational Behavior, 88,
143–153. https://doi.org/10.1016/j.jvb.2015.03.007 Received January 6, 2023
Xu, H. (2022). Understanding the dysfunctionality of dysfunctional career Revision received June 21, 2023
decision-making beliefs: Ambiguity aversion as a general mechanism. Accepted June 28, 2023 ▪

Patterns of Career Decision-Ma

Uploaded by

Copyright:

Available Formats

Patterns of Career Decision-Ma

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Patterns of Career Decision-Ma

Uploaded by

Copyright:

Available Formats

Journal of Counseling Psychology

Patterns of Career Decision-Making Difficulties in 16 Countries: A Person-

Patterns of Career Decision-Making Difﬁculties in 16 Countries:

Public Signiﬁcance Statement

Supplemental materials: https://doi.org/10.1037/cou0000704.supp

Association. The ﬁndings have not been published or disseminated publicly

These two types differ

Anxious and unclear on

and (c) inconsistent information (comprising unreliable information,

unrealistic, uninformed, and conﬂicted—in two U.S. samples. These

ﬁve types represent distinct patterns of co-occurring causes of career

indecision. Speciﬁcally, the unmotivated, indecisive, and unrealistic

types were characterized by a single salient difﬁculty relating to a

lack of motivation, general indecisiveness, or dysfunctional beliefs.

Then, whereas the uninformed and conﬂicted types included

recent ﬁndings support using the CDDQ as a differential diagnostic

(2022) CDDQ-based typology in other countries remain to be

tested. In this regard, 23 of the 24 previous clustering studies on

syntheses preceded the publication of any study conducted outside

the United States. Nevertheless, Kelly and Pulver’s (2003) typology—

studies conducted later in Canada (Guay et al., 2006), Greece

(Argyropoulou et al., 2007), and Portugal (Santos & Ferreira,

2012), thus supporting the likely generalizability and applicability

of indecision types across countries.

Moreover, although some have challenged the cross-cultural

notwithstanding, some studies found that individuals in Asian

and African countries are more prone to career indecision than

replicates in different countries, career indecision may still

Santos and Ferreira (2012)

manifest differently in different contexts.

Chartrand et al. (1994)

Levin et al. (2022)

The Present Study

diagnosing career problems among individuals from diverse

occupation I am interested in, but I would like to feel sure of my

Sample characteristics LPA (within samples) Profile RFs (across samples)

the minimal profile RF, fit statistics, and classification quality

evaluated the qualitative distinctiveness of proﬁles, revealing that

proﬁles. Thus, after ensuring the adequacy of its classiﬁcation

unrealistic (25.8%) as they were high in lack of motivation and

dysfunctional beliefs, respectively. The next two identiﬁed proﬁles

identified profiles were characterized by difficulties related to

conﬂicted-uninformed and externally conﬂicted, respectively,

external conflicts, individuals in the conflicted-uninformed profile

also reported lacking information in the four information domains,

information, and general indecisiveness. The right side of Table 2

presents proﬁle RFs separately for each of the 18 country-language

solutions within and across samples, most individuals were cross-

of those classiﬁed as lack of information in the within-samples

analyses were assigned to its equivalent generally uninformed proﬁle

Fit statistics Classification quality

2 44.49% 65 −519768.99 1039667.97 1040211.28 1040004.71 .72 .91

1 versus 2 1 versus 3 1 versus 4 1 versus 5 1 versus 6

1 versus 7 2 versus 3 2 versus 4 2 versus 5 2 versus 6

2 versus 7 3 versus 4 3 versus 5 3 versus 6 3 versus 7

4 versus 5 4 versus 6 4 versus 7 5 versus 6 5 versus 7

Age 0.05*** 1.05

Note. GU = generally uninformed; CU = conflicted-uninformed; IC = internally conflicted; UM = unmotivated; EC = externally conflicted;

385. https://doi.org/10.1016/j.jvb.2013.06.001 Guidance & Counselling, 50(5), 677–686. https://doi.org/10.1080/03069