See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/236640870

Development and Validation of the Revised Identity Style Inventory (ISI-5):

Factor Structure, Reliability, and Validity

Article in Psychological Assessment · May 2013

DOI: 10.1037/a0032642 · Source: PubMed

CITATIONS READS
140 9,848

6 authors, including:

Michael Berzonsky Bart Soenens

SUNY Cortland Ghent University
110 PUBLICATIONS 7,271 CITATIONS 378 PUBLICATIONS 35,002 CITATIONS

SEE PROFILE SEE PROFILE

Koen Luyckx Dennis Papini

KU Leuven South Dakota State University
430 PUBLICATIONS 20,360 CITATIONS 42 PUBLICATIONS 1,826 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Bart Soenens on 26 May 2014.

The user has requested enhancement of the downloaded file.

 Psychological Assessment © 2013 American Psychological Association
2013, Vol. 25, No. 3, 893–904 1040-3590/13/$12.00 DOI: 10.1037/a0032642

Development and Validation of the Revised Identity Style Inventory

(ISI-5): Factor Structure, Reliability, and Validity

Michael D. Berzonsky Bart Soenens

State University of New York at Cortland Ghent University

Koen Luyckx and Ilse Smits Dennis R. Papini

Katholieke Universiteit Leuven Middle Tennessee State University
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Luc Goossens
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Katholieke Universiteit Leuven

Identity processing style refers to differences in how individuals process identity-relevant information as
they engage or manage to avoid the challenges of constructing, maintaining, and/or reconstructing a sense
of identity. The third version of the Identity Style Inventory (Berzonsky, 1992b) has been used to
operationally define identity styles in most empirical investigations. The objective of the present series
of studies was the development and validation of a new revised measure of identity processing style:
Identity Style Inventory—Version 5 (ISI-5). Initially a pool of 39 generic items was generated that
highlighted the processing of identity-relevant information on content-neutral issues such as personal
values, goals, problems, and the like. Three style scales were identified by Exploratory Factor Analysis:
A 9-item Informational-style scale; a 9-item Normative-style scale; and a 9-item Diffuse-avoidant style
scale. Confirmatory factor analysis on an independent sample indicated that this 3-factor solution
provided the best fit. Results from 5 studies provided evidence for the psychometric properties of the
scales. Scores on the 3 style scales demonstrated good test–retest reliability and internal consistency.
Theoretically predicted correlations between the ISI-5 scale scores and performance on measures of
identity status, content, and commitment, and measures of rational and automatic processing provided
evidence for their convergent and discriminant validity. It is concluded that the scales should be useful
for researchers interested in investigating individual differences in identity processing style. Limitations
and directions for future research are considered.

Keywords: identity style, commitment, rational processing, identity formation, psychometric

According to Erikson’s (1968) psychosocial theory, the process frame of reference for making decisions and interpreting experi-
of forming an individualized, well-integrated sense of identity ence and self-relevant information, which enables people to main-
plays a central role in personality development over the lifespan. A tain a meaningful sense of self-sameness and self-continuity de-
consolidated, well-integrated identity structure provides a personal spite the random events and inevitable changes they encounter
during their lives (Berzonsky, 2005a). Although Erikson’s theory
reflects his psychoanalytic training and is grounded in ego psy-
chology, he and others have highlighted the role that cognitive
This article was published Online First May 6, 2013. processes may play in identity formation (e.g., Erikson, 1964;
Michael D. Berzonsky, Department of Psychology, State University of
Inhelder & Piaget, 1958; Marcia, 1999). Findings from early
New York at Cortland; Bart Soenens, Department of Developmental,
Personality, and Social Psychology, Ghent University, Ghent, Belgium;
investigations of the role formal operational reasoning may play
Koen Luyckx and Ilse Smits, Faculty of Psychology and Educational in identity formation were inconsistent (see for example, Ber-
Sciences, School Psychology and Child and Adolescent Psychology, zonsky & Barclay, 1981; but compare Rowe & Marcia, 1980).
Katholieke Universiteit Leuven, Leuven, Belgium; Dennis R. Papini, Subsequent research, however, indicated there are reliable sty-
Department of Psychology, Middle Tennessee State University; Luc listic differences in the social-cognitive strategies individuals
Goossens, Faculty of Psychology and Educational Sciences, School use to construct, preserve, and/or reconstruct a sense of identity
Psychology and Child and Adolescent Psychology, Katholieke Univer- (Berzonsky, 2011). In particular, Berzonsky (1990, 2004) de-
siteit Leuven.
veloped a social-cognitive model of identity formation that
Dennis R. Papini is now at College of Arts and Sciences, South Dakota
postulates three different identity processing styles: informa-
State University.
Correspondence concerning this article should be addressed to Michael tional, normative, and diffuse-avoidant. These styles refer to
D. Berzonsky, Department of Psychology, Old Main Room 132, State reported preferences in the social-cognitive processes deployed
University of New York at Cortland, Cortland, NY 13045. E-mail: when individuals deal with or attempt to evade identity con-
[email protected] flicts and decisions (Berzonsky, 2011).

893
 894 BERZONSKY ET AL.

The purpose of the present investigation was to develop and with cognitive complexity, problem-focused coping, vigilant de-
validate a revised measure of these identity processing styles. First, cision making, open mindedness, personal effectiveness, and an
a brief theoretical description of the three processing styles and achieved or moratorium identity status (Berzonsky, 2011).
some relevant research is presented. The construction and devel-
opment of the revised Identity Style Inventory (ISI) is then de-
Normative Identity Processing Style
scribed. Finally, psychometric data relevant to the reliability and
convergent validity of the Identity Style Inventory—Version 5 Not everyone approaches potentially self-diagnostic information
(ISI-5) scale scores are presented. in a rational, open-minded fashion. Although people with a nor-
mative identity style are also conscientious, self-disciplined and
possess a strong sense of commitment and purpose, they tend to
Identity Processing Orientations
internalize and adhere to the goals, expectations, and standards of
Marcia (1966) developed the identity status paradigm to oper- significant others or referent groups in a relatively more automatic
ationally define Eriksonian identity formation. By crossing high manner. They have a foreclosed identity status, a limited tolerance
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and low levels of self-exploration (originally termed crisis) and for uncertainty and a strong need for structure and closure; their
This document is copyrighted by the American Psychological Association or one of its allied publishers.

commitment, Marcia identified four identity types or statuses: (a) primary goal is to defend and preserve their existing self-views and
identity achievement (highly committed following a period of identity structure (Berzonsky, 2004; Soenens, Duriez, & Goossens,
self-exploration), (b) identity moratorium (currently engaged in 2005).
self-exploration with limited commitment), (c) identity foreclosure
(highly committed with limited self-exploration), and (d) identity Diffuse-Avoidant Identity Processing Style
diffusion (limited commitment but not engaged in self-
exploration). A substantial body of literature has established reli- Individuals with a diffuse-avoidant style procrastinate and try to
able status differences along a number of social, personal, and avoid dealing with identity conflicts and decisions as long as
cognitive dimensions (see, e.g., Berzonsky & Adams, 1999; Mar- possible. When they have to act or make choices, their behavior is
cia, 1993, for reviews). Most identity research over the past four determined primarily by situational demands and consequences.
decades has been based on or inspired by the status paradigm. How they act depends to a large extent on where they are and who
The statuses are typically conceptualized as identity outcomes, they are with. Theoretically, people with high diffuse-avoidant
because exploration and commitment are confounded within each scores may possess commitments but their commitments are likely
status category. Consequently, a number of investigators have to be volatile and quickly accommodated in light of changing
attempted to focus more directly on the process by which identity situational demands, rewards, and circumstances (Berzonsky,
is formed (e.g., Berzonsky, 1990; Grotevant, 1987; Kerpelman, 2011; Berzonsky & Ferrari, 2009). This identity style is associated
Pittman, & Lamke, 1997; Whitbourne & Weinstock, 1979). In with an external locus of control, limited self-control, weak com-
particular, Berzonsky and Barclay (1981) hypothesized that Mar- mitments, self-handicapping attributions and behaviors, problem
cia’s (1966) four statuses reflected three different ways of dealing behaviors and a diffusion identity status (Berzonsky, 2011; Ber-
with or managing to avoid identity-relevant conflicts and issues: an zonsky & Ferrari, 2009).
informed, rational orientation; a more automatic, normative or
conforming orientation; and a procrastinating, diffuse-avoidant Theoretical Associations Between Identity Processing
orientation (see also Berzonsky, 1988, 2011).
Styles and Related Identity and Cognitive Constructs
Berzonsky (1990, 2011) proposed that these processing orien-
tations operate on different levels. Social-cognitive identity pro- The construct of identity style was developed in an effort to
cessing strategies comprise organized sets of the more elemental explain differences in how individuals categorized in Marcia’s
cognitive and behavioral responses individuals use to process and (1966) identity status categories approach or manage to avoid the
cope with identity-relevant information and conflicts. In contrast, tasks of forming, maintaining, and/or revising their identity struc-
identity processing style refers to relatively stable differences in ture (Berzonsky, 1988, 2011; Berzonsky & Barclay, 1981). Con-
the social-cognitive strategies that individuals typically prefer to sequently, performance on valid measures of identity style should
employ to negotiate identity conflicts. Although research suggests differentially predict scores on measures of identity status. Specif-
that late adolescents normally are capable of utilizing all three ically, an informational style should be linked to identity achieve-
social-cognitive strategies, there may be individual differences in ment; the normative style to identity foreclosure; and the diffuse-
how efficiently and consistently they are accessed and utilized avoidant style to identity diffusion.
(Berzonsky, 2011). Most research on these orientations has fo- Theoretically, the informational style should also be positively
cused on identity processing styles. associated with identity moratorium (Berzonsky, 1990), which is
characterized by ongoing self-exploration (Marcia, 1966, 1993).
However, research has demonstrated that commitment may sup-
Informational Identity Processing Style
press that relationship: Commitment is negatively correlated with
Individuals with an informational identity style are self- moratorium scores, but positively associated with informational
disciplined with a clear sense of commitment and direction. They scores (Berzonsky, 1990; Berzonsky & Neimeyer, 1994). Further,
are self-reflective, skeptical, and interested in learning new things style scores should account for unique variation in strength of
about themselves; they intentionally seek out, evaluate, and utilize identity commitment. That is to say, commitments may be strate-
self-relevant information, and they are willing to accommodate gically avoided (Berzonsky & Ferrari, 2009) or they may be
self-views in light of dissonant feedback. This style is associated formed in an intentional, reflective informative fashion or a more
 DEVELOPMENT AND VALIDATION OF REVISED IDENTITY STYLE 895

automatic normative manner (Berzonsky, 1990; Marcia, 1966, time thinking about my options” (informational) and “When I have
1993). to make a decision, I try to wait as long as possible to see what will
The type of self-attributes (see Cheek, 1989) people highlight in happen” (diffuse-avoidant). The revised ISI-2 Inventory comprised
the way they define themselves and their identity should vary with a 10-item informational style scale (coefficient ␣ ⫽ .62); a 10-item
identity processing styles (Berzonsky, 1994; Berzonsky, Macek, & diffuse-avoidant style scale (coefficient ␣ ⫽ .73); a nine-item
Nurmi, 2003). An informational style focuses on personal self- normative style scale (coefficient ␣ ⫽ .66); and the original
components including personal goals, values, ideas, and moral 10-item commitment scale (coefficient ␣ ⫽ .71).
standards. A normative style would primarily emphasize relatively Additional modifications to the normative and informational
stable collective self-attributes including religion, family, ethnic- scales were made in a third revision (Berzonsky, 1992b). The third
ity, and nationality. Given that a diffuse-avoidant style focuses on version (ISI-3) contained an 11-item informational style scale
situational demands and consequences (Berzonsky, 1990; Berzon- (coefficient ␣ ⫽ .70); a nine-item normative style scale (coeffi-
sky & Ferrari, 2009), it should be associated with more variable cient ␣ ⫽ .64); a 10-item diffuse-avoidant style scale (coefficient
social self-components such as reputation, popularity, and the ␣ ⫽ .76); and a 10-item commitment scale (coefficient ␣ ⫽ .71).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

impressions and reactions of others. Two-week, test–retest reliabilities of the ISI-3 scales were diffuse-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Identity processing styles include reasoning as well as identity avoidant r ⫽ .83; informational r ⫽ .87; normative r ⫽ .87; and
components (Berzonsky, 1990, 2004). An informational style pri- commitment r ⫽ .89 (Berzonsky, 1992b). Over the past two
marily involves deliberate, mentally effortful, reason-based prob- decades, the third version of the Inventory (ISI-3) or translations of
lem solving and decision making. However, continually seeking it have been used in numerous countries including Australia,
new information and reconsidering problem solutions and deci- Belgium, Canada, China, the Czech Republic, Denmark, Finland,
sions can become counterproductive. Therefore, an informational Germany Greece, India, Iran, Italy, the Netherlands, Pakistan,
style is postulated to reflect experience-based automatic as well as Poland, Slovakia, Spain, Switzerland, and Turkey (Berzonsky,
conscious rational reasoning (Berzonsky, 2008, 2011). A norma- 2005b, 2011). The ISI-3 scales have been found to have acceptable
tive style is postulated to be relatively more automatic in nature. psychometric properties (see Berzonsky, 1992b, 2004, 2011). In-
While automatic processing may ordinarily be adaptive and eco- ternal reliabilities of the ISI-3 style scores generally range from .60
nomical in terms of the expenditure of mental resources and effort, to .75, although estimates for translated versions, especially of the
it is prone to bias and distortion (Epstein, 1990). The diffuse- normative scale, have in some cases been lower (see Berzonsky,
avoidant style, being driven by situational demands and hedonistic 2011). Convergent validity has been established by theoretically
considerations, should be negatively associated with rational pro- consistent relationships between the style scores and performance
cessing (Berzonsky, 1990, 2008). on measures of identity status, identity emphases (i.e., personal,
social, or collective), cognitive reasoning, personal adjustment and
well-being, and personality dimensions (Berzonsky, 2011).
Measuring Identity Processing Styles
Despite these strengths, there are some potential limitations with
In an initial attempt to operationally define identity processing the ISI-3. In some cases, especially with translated versions of the
styles, Berzonsky (1989) created the original Identity Style Inven- normative scale, internal reliability of scores on the ISI-3 scales
tory (ISI-1). Statements that reflected the process dimension of the was found to be less than .60. To some extent, the moderate
identity statuses in the content domains measured in Marcia’s internal reliability may reflect the bandwidth of the theoretical
(1966) status interview were generated (six statements for each of construct, which comprises cognitive processing as well as identity
the three styles). Participants rated the extent to which each state- components, but some other considerations may be relevant. For
ment was “not like” or “like” them. A 10-item commitment scale instance, like a number of measures of identity (e.g., Adams, 1999;
was also included. The 5-week, test–retest reliability of scores on Balistreri, Busch-Rossnagel, & Geisinger, 1995), some of the item
the ISI-1 scales ranged from .78 to .80. Their internal reliabilities, statements in ISI-3 refer to different specific content domains such
however, were quite modest: .52 to .59. Despite the low internal as religious or political beliefs (see Smits et al., 2008; Vleioras,
reliability, scores on the scales generally differentiated between the 2007). Thus, it is not always clear whether some items are more
statuses in a theoretically predictable way: Positive correlations salient than others for a given participant. That is to say, the style
were found between the informational style and identity achieve- one endorses may depend on the personal relevance of the issue
ment scores, the normative style and identity foreclosure scores, being considered. For instance, some participants may endorse an
and performance on the diffuse-avoidant style and identity diffu- informational style with respect to personally relevant issues (e.g.,
sion scales (see also Berzonsky, 1990). One seemingly anomalous religious values) but a diffuse-avoidant style with respect to self-
finding was the lack of a positive correlation between the infor- irrelevant issues (e.g., political values), whereas others may en-
mational style and identity moratorium scores, which by definition dorse the same style across content domains. Both phenomena
involves ongoing effortful self-exploration. However, when com- represent substantive problems that may blur the interpretation of
mitment was statistically controlled, the hypothesized relationship the ISI-3 scores. In addition, these phenomena may suppress the
was obtained (see also Berzonsky, 1990). reliability of the style scores and may, more generally, represent a
In an effort to improve its psychometric properties, Berzonsky problem for the internal structure of the scales.
(1992a) revised the original Inventory (ISI-2). In this revision, On a related note, the salience of content domains may also vary
instead of relying exclusively on statements relevant to content across cultures or nations. Religious beliefs, for instance, may not
areas in Marcia’s (1966) Identity Status Interview, numerous “face be relevant to virtually all participants in some cultures or nations.
valid” statements about identity processing were generated: for On the other hand, most participants in other cultures or nations
instance, “When I have to make a decision, I like to spend a lot of may accept religious beliefs in an automatic, normative fashion
 896 BERZONSKY ET AL.

without informed deliberation. These issues may create equiva- (ISI-4), which has been used in a number of studies (e.g., Doumen
lence problems when the ISI-3 scales are being translated, which et al., 2012; Luyckx, Lens, Smits, & Goossens, 2010; Missotten,
may in part account for the relatively low internal-reliability esti- Luyckx, Vanhalst, Branje, & Goossens, 2011; Smits, 2009, Ch. 3;
mates obtained with some translated versions of the ISI-3 scales. Smits, Doumen, Luyckx, Duriez, & Goossens, 2011; Smits,
Likewise, content-domain salience may vary across age groups Soenens, Vansteenkiste, Luyckx, & Goossens, 2010). Although
(e.g., questions about education major will not be relevant to high this version has worked relatively well— e.g., the test–retest reli-
school adolescents). Wording style statements in a domain-neutral ability over a 1-week interval for the informational, normative, and
manner may, therefore, facilitate efforts to assess the cross- diffuse-avoidant scores were, respectively, .80, .85, and .87 (Smits,
national and cross-age generalizability of the identity style theory. 2009, Chapter 3)—it has some problems. The criteria for selecting
Another potential problem with the ISI-3 is that some of the the items were overly strict, which may have resulted in the
scale statements are worded in the present tense, whereas others removal of essential items. Perhaps as a consequence, the reliabil-
are worded in the past tense (see Smits et al., 2008). Due to this ity and validity of the scores on the normative scale in particular
mixture of retrospective items and items tapping into one’s current was questionable. Specifically, the internal reliability of scores on
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

endorsement of the identity styles, the ISI-3 scores may not be the ISI-4 normative scale tends to be less than .70, and it was found
This document is copyrighted by the American Psychological Association or one of its allied publishers.

interpreted unequivocally as representing one’s current identity to have low to nonsignificant correlations with the commitment
style. The blend of current and past processing of identity-relevant scale and to correlate highly with scores on the diffuse-avoidant
information reflected in ISI-3 scores may be particularly problem- scale (e.g., Doumen et al., 2012; Missotten et al., 2011; Smits et
atic in longitudinal research. To adequately examine changes in al., 2010, 2011). Given these problems, in the current set of studies
identity style scores across time, it is important to have an instru- we aimed to do further systematic psychometric and validation
ment that assesses individuals’ current identity processing styles analyses with the broader set of revised items so as to arrive at a
within each of the time points. next and improved version of the ISI, that is, the ISI-5.
The present investigation was an attempt to develop a revised
version of the Identity Style Inventory and to obtain evidence Phase 2: Evaluating the Factor Structure of the ISI-5
relevant to the test–retest and internal reliability and convergent
and discriminant validity of the scores of this revised measure.
Participants

Phase 1: Constructing the Revised Inventory (ISI-5) The participants for this phase of the investigation consisted of
two independent samples of undergraduate students enrolled at a
The revision process involved generating additional items for large southern university in Tennessee. Sample 1 consisted of 403
the three style scales and revising and adapting some of the participants (241 females and 162 males) with an age range from
statements contained in the third version of the Inventory (ISI-3). 17 to 26 years (M ⫽ 19.00, SD ⫽ 1.38). Sample 2 was composed
First, our goal was to create statements that referred to generic of 440 participants (314 females and 126 males) who varied from
identity categories (e.g., values, goals, beliefs, life decisions, per- 17 to 25 years in age (M ⫽ 19.25, SD ⫽ 2.04). The majority of the
sonal problems, and the like), rather than to ones that referred to participants were Caucasian. Although information about race/
specific identity domains (e.g., religion, political beliefs, college ethnicity was not obtained from the participants in Sample 1, 57%
major, occupation, and the like). Thus, interviewees would be able of the participants in Sample 2 self-identified as Caucasian, 16% as
to decide for themselves which personal problems, values, goals, African American, and 4% as Hispanic; 23% left the ethnicity
and so on they would focus on rather than having to respond to question blank or self-identified as other. All students, who vol-
ones relevant to a particular domain such as, for example, reli- unteered for extra-course credit, were administered the 39 identity
gious, political, or moral values. Second, statements were worded style and nine commitment items in large group settings. (Addi-
in the present tense. Finally, we attempted to focus on the pro- tional tests, described below, designed to provide validity data
cessing of identity-relevant information and not the outcome of were also administered to the participants in Sample 1.) All items
that process: for example, “When making important life decisions, were responded to on 1 (not at all like me) to 5 (very much like me)
I like to think about my options” (informational); “I automatically Likert-type scales.
adopt and follow the values I was brought up with” (normative);
and “Many times, by not concerning myself with problems, they
Exploratory Factor Analysis (Sample 1)
work themselves out” (diffuse-avoidant).
The first four authors who had all published research on identity Exploratory factor analyses (EFAs) were conducted to refine the
style in peer-reviewed journals, made judgments about the face item selection process. Sample 1 was used in this phase of the
validity of the pool of statements that had been generated. Thirty- research program. First, a principal factors extraction with varimax
nine style items were retained: there were 13 items for each style rotation was performed on the 39 style items. The scree test, which
scale. In addition, the authors made judgments about the 10 items plots the eigenvalues against the factors (Cattell, 1966), indicated
in the ISI-3 commitment scale. Some commitment items were that the items measured three factors. Twelve items with loadings
revised to refer to generic identity areas and all were stated in the less than .40 on their relevant factor and/or cross-loadings greater
present tense: for instance, “I know what I want to do with my than .35 were deleted. The final three-factor solution (see Table 1),
future” (commitment). One was deleted, which resulted in a nine- which accounted for 39% of the variance, consisted of a nine-item
item identity commitment scale. informational style scale (coefficient alpha ⫽ .77), a nine-item
On the basis of this broad pool of items, Smits et al. (2008) normative style scale (coefficient alpha ⫽ .75), and
developed a brief revised version of the Identity Style Inventory a nine-item diffuse-avoidant style scale (coefficient alpha ⫽ .79).
 DEVELOPMENT AND VALIDATION OF REVISED IDENTITY STYLE 897

Table 1
Factor Loadings of the Revised Identity Style Inventory Scales

EFA
(Sample 1)
CFA
Items 1 2 3 (Sample 2)

Diffuse-avoidant items
When personal problems arise, I try to delay acting as long as possible .58 ⫺.10 ⫺.01 .59
I’m not sure where I’m heading in my life; I guess things will work themselves out .57 ⫺.08 ⫺.09 .33
My life plans tend to change whenever I talk to different people .57 .00 .04 .54
Who I am changes from situation to situation .55 ⫺.03 .00 .46
I try not to think about or deal with problems as long as I can .53 ⫺.12 .20 .61
I try to avoid personal situations that require me to think a lot and deal with them on my own .51 ⫺.12 .26 .44
When I have to make a decision, I try to wait as long as possible in order to see what will happen .51 .12 .05 .50
⫺.14
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

It doesn’t pay to worry about values in advance; I decide things as they happen .46 .01 .34
⫺.22 ⫺.01
This document is copyrighted by the American Psychological Association or one of its allied publishers.

I am not really thinking about my future now, it is still a long way off .42 .36
Informational items
When making important decisions, I like to spend time thinking about my options ⫺.22 .57 ⫺.03 .55
When facing a life decision, I take into account different points of view before making a choice ⫺.03 .57 .02 .56
It is important for me to obtain and evaluate information from a variety of sources before I make
important life decisions .03 .57 .00 .54
When making important decisions, I like to have as much information as possible ⫺.15 .57 .04 .70
When facing a life decision, I try to analyze the situation in order to understand it ⫺.30 .55 ⫺.08 .63
Talking to others helps me explore my personal beliefs .13 .52 ⫺.13 .26
I handle problems in my life by actively reflecting on them ⫺.20 .50 ⫺.03 .50
I periodically think about and examine the logical consistency between my values and life goals ⫺.16 .46 .15 .43
I spend a lot of time reading or talking to others trying to develop a set of values that makes sense to me .22 .40 ⫺.08 .27
Normative items
I automatically adopt and follow the values I was brought up with ⫺.11 .06 .60 .60
I think it is better to adopt a firm set of beliefs than to be open-minded ⫺.01 ⫺.12 .58 .68
I think it’s better to hold on to fixed values rather than to consider alternative value systems ⫺.11 ⫺.11 .56 .73
When I make a decision about my future, I automatically follow what close friends or relatives expect
from me .33 .10 .56 .49
I prefer to deal with situations in which I can rely on social norms and standards .27 .09 .47 .40
I have always known what I believe and don’t believe; I never really have doubts about my beliefs ⫺.33 ⫺.04 .45 .44
I never question what I want to do with my life because I tend to follow what important people expect
me to do .23 .03 .45 .45
When others say something that challenges my personal values or beliefs, I automatically disregard what
they have to say .10 ⫺.26 .44 .43
I strive to achieve the goals that my family and friends hold for me .12 .31 .41 .41
Eigenvalue 4.49 3.20 2.86
Note. EFA ⫽ exploratory factor analysis; CFA ⫽ confirmatory factor analysis. Values in boldface type indicate the highest factor loading for each item.

A nine-item identity commitment scale was also retained: A sep- We first estimated a three-factor model, where each of the three
arate principal factor analysis indicated that all of the commitment identity styles was represented as a latent variable indicated by its
items had loadings greater than .40 on a single factor (coefficient corresponding items. To examine whether the three identity styles
alpha ⫽ .82). were distinct, this three-factor model was compared to three alter-
native two-factor models: that is, a model where the informational
Confirmatory Factor Analyses (Sample 2) and the normative items were combined into one construct (Alter-
native Model 1); a model where the informational and diffuse-
To evaluate the accuracy of the three-factor structure of the 27
avoidance items were combined (Alternative Model 2); and a
items obtained by the exploratory factor analysis, a CFA was
model where the normative and the diffuse-avoidance items were
performed with Sample 2. CFA was conducted using Lisrel 8.50
with Maximum Likelihood Estimation (Jöreskog & Sörbom, combined (Alternative Model 3).
1993). To evaluate model fit, we inspected the Satorra-Bentler The three-factor solution approached criteria for acceptable fit
Scaled chi-square (SBS-␹2, Satorra & Bentler, 1994) instead of the [SBS-␹2(321) ⫽ 866.39; SBS-␹2/df ⫽ 2.70; RMSEA ⫽ .06;
regular chi-square because the former corrects for data nonnor- SRMR ⫽ .08] and was clearly favored over each of the two-factor
mality. An SBS-␹2 to degree of freedom ratio (SBS-␹2/df) close to solutions, as indicated by significantly different chi-square statis-
3.0 indicates acceptable model fit (Kline, 2005). We also inspected tics [⌬SBS-␹2(2) ⫽ 2843.93; p ⬍ .001 for Alternative Model 1;
the root-mean-square error of approximation (RMSEA) and the ⌬SBS-␹2(2) ⫽ 2459.98; p ⬍ .001 for Alternative Model 2; ⌬SBS-
standardized root-mean-square residual (SRMR). According to Hu ␹2(2) ⫽ 529.49; p ⬍ .001 for Alternative Model 3]. In the final
and Bentler (1999), acceptable model fit is indicated by cutoff best fitting model, factor loadings ranged between .26 and .73 with
values of .08 or less for RMSEA and .09 or less for SRMR. In a mean of .47 (all ps ⬍ .001). In terms of correlations between the
order to compare models, SBS-␹2 difference tests were used. latent factors, the information-oriented style factor was unrelated
 898 BERZONSKY ET AL.

to the normative style factor (r ⫽ .04; p ⬎ .05) and negatively Procedures and Measures
related to the diffuse-avoidant style factor (r ⫽ –.23; p ⬍ .01). The
correlation between the normative style and the diffuse-avoidant To evaluate the convergent validity of the scores on the ISI-5,
style was not significant (r ⫽ .13; p ⬎ .05). the revised scales and a battery of measures designed to assess
To assess the generalizability of this factor structure, we exam- identity status, self-components, identity commitment, and cogni-
tive reasoning processes were administered to the participants in
ined whether it would replicate across gender. To this aim, we
Samples 1 and Sample 3. Cronbach alphas for Samples 1 and 3,
performed a multigroup CFA, comparing an unconstrained model
respectively, were: .77 and .74 (informational); .75 and .79 (nor-
(i.e., a model where factor loadings were allowed to vary between
mative); .79 and .83 (diffuse-avoidant); and .82 and .82 (commit-
male and female participants) to a constrained model (i.e., a model
ment). The means and standard deviations for the revised scales
where factor loadings were set equal across gender). The uncon-
appear in Table 2. Comparisons between female and male partic-
strained model did not have a significantly better fit compared to
ipants indicated that although females scored higher on commit-
the constrained model [⌬SBS-␹2(27) ⫽ 38.54; p ⬎ 05] in terms of
ment in both Sample 1, F(1, 400) ⫽ 3.89, p ⬍ .05 and Sample 3,
the chi-square statistic. This analysis suggests that the factor struc-
F(1, 171) ⫽ 3.92, p ⬍ .05 and lower on diffuse-avoidance in
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ture was comparable across gender. With Sample 2, the internal

Sample 3, F(1, 171) ⫽ 8.54, p ⬍ .01, the effect sizes were
This document is copyrighted by the American Psychological Association or one of its allied publishers.

reliabilities (coefficient alpha) for the scale scores were as follows:

extremely small (Table 2). The following criterion measures were
informational style .74; normative style .77; diffuse-avoidant style
also administered.
.71; and identity commitment .82.
Identity status. Identity status was operationalized with ide-
ological items from the Extended Objective Measure of Ego Iden-
Phase 3: Validity of Scores on the ISI-5 tity Status (EOM-EIS: Grotevant & Adams, 1984). The ideological
EOM-EIS comprises four eight-item identity status measures that
The previous analyses indicated that the factor structure and were responded to on a 1 (strongly disagree) to 5 (strongly agree)
internal consistency of the revised scale scores were sound. This Likert-type scale. Cronbach alphas for Sample 1 and Sample 3
phase of the research program focused on the reliability and were, respectively, .62 and .62 (achievement); .82 and .82 (fore-
convergent validity of the scores on the revised scales: Do scores closure); .71 and .72 (moratorium); and .61 and .60 (diffusion).
on the ISI-5 scales correlate with measures of other identity and Validity data for the scales are presented in Adams (1999).
cognitive processes in theoretically predictable ways? Cognitive reasoning processes. The Need for Cognition
scale (NFC: Cacioppo, Petty, & Kao, 1984), which measures the
extent to which individuals are motivated to engage in deliberate,
Participants effortful information processing, was used to assess intentional,
Four samples of undergraduate students enrolled at a large rational reasoning (18 items: e.g., “I would prefer complex prob-
southern university in Tennessee participated in this phase of the lems to simple ones”). Coefficient alpha was .90 in both Sample 1
investigation. Sample 1 consisted of the same 403 participants and 3. Reliability and validity data are provided in Cacioppo and
(241 females and 162 males) who participated in Phase 2: they Petty (1982) and Cacioppo et al. (1984). Automatic, intuitive
ranged in age from 17 to 26 years (M ⫽ 19.00, SD ⫽ 1.38). cognitive processing was assessed with the Faith in Intuition scale
Sample 3 comprised 174 participants (113 females and 61 males). from the Rational Versus Experiential Inventory (RVEI: Epstein,
Their ages ranged from 18 to 26 years (M ⫽ 18.97, SD ⫽ 1.29). Pacini, Denenes-Raj, & Heier, 1996). Epstein et al. (1996) found
The fourth sample consisted of 70 participants (40 females and 30 that experientially based, intuitive processing scores were associ-
males) who ranged in age from 18 to 30 years in age (M ⫽ 20.83, ated with emotion-based reasoning such as stereotypic thinking,
SD ⫽ 2.76). Sample five included 77 participants (47 females and superstition, and naive optimism: (12 items: e.g., “I believe in
30 males) who ranged in age from 18 to 31 years in age (M ⫽
20.43, SD ⫽ 2.50). Participants in Sample 5 filled out the ISI-5
twice, within a 2-week interval. Convergence of the ISI-5 scores Table 2
with those on measures of other identity processes and cognitive Means and Standard Deviations of Revised Scales
measures was evaluated using Samples 1 and 3. Sample 4 was used
to assess the convergence between scores on the revised ISI-5 Total Males Females
scales and those measured by the third version of the Identity Style Scale M SD M SD M SD Effect size
Inventory (ISI-3: Berzonsky, 1992b). The test–retest reliability of
the ISI-5 scores was evaluated with Sample 5. The majority of the Sample 1 (N ⫽ 403)
Informational 3.66 0.59 3.59 0.62 3.58 0.58 .00
participants were Caucasian. In Samples 4 and 5, respectively, Normative 2.76 0.63 2.69 0.65 2.81 0.61 .00
66% and 56% of the participants self-identified as Caucasian, 20% Diffuse-Avoidant 2.18 0.66 2.24 0.68 2.14 0.65 .00
and 29% as African American, 6% and 3% as Hispanic, 3% and Commitment 3.91 0.68 3.83 0.71 3.97 0.65 .01ⴱ
5% as Asian American, and the rest did not answer the ethnicity
Sample 3 (N ⫽ 174)
question or answered “other.” All of the participants volunteered Informational 3.70 0.56 3.66 0.60 3.72 0.54 .00
for extra-course credit and were administered the 27 identity style Normative 2.80 0.67 2.78 0.73 2.81 0.65 .00
and nine commitment items in large group settings. (Additional Diffuse-Avoidant 2.08 0.69 2.27 0.77 1.96 0.62 .05ⴱⴱ
tests, described below, designed to provide validity data were also Commitment 4.00 0.69 3.86 0.73 4.07 0.66 .02ⴱ
administered.) All items were responded to on 1 (not at all like me) Note. Means are adjusted for number of items in each scale.
ⴱ
to 5 (very much like me) Likert-type scales. p ⬍ .05. ⴱⴱ p ⬍ .01
 DEVELOPMENT AND VALIDATION OF REVISED IDENTITY STYLE 899

trusting my hunches”). Coefficient alpha was .83 in Sample 1 and increased from .10 (ns) to .35 (p ⬍ .01). Thus, it appears that
.82 in Sample 3. the predicted positive relationship between an informational
Self-components. The type of self-relevant information asso- processing style and identity moratorium was suppressed by
ciated with different identity processing styles was measured with commitment.
the Aspects of Identity Questionnaire (AIQ-IIIx) developed by One apparently anomalous finding involves the moderate
Cheek, Tropp, Chen, and Underwood (1994). Participants rated positive zero-order correlation between the diffuse-avoidant
how important various self-components or attributes are to their style and moratorium-status scores (Table 3). To a large extent
sense of self on a 1 (not important to my sense of who I am) to 5 this association may be due to the strong relationship between
(extremely important to my sense of who I am) Likert-type scale. Adams’ (1999) measures of the diffusion and moratorium iden-
The AIQ-IIIx contains three subscales: Personal self-components tity statuses, which load on a common factor. Controlling for
(10 items: e.g., personal goals, knowledge, values, and thoughts);
the effect of a diffusion status in the regression analyses atten-
Collective self-attributes (eight items: e.g., family, religion, race or
uated the relationship between diffuse-avoidant and moratorium
ethnicity, and country); and Social self-elements (seven items: e.g.,
scores.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

reputation, popularity, and impressions on others). Coefficient

The results in Table 3 provide evidence for the discriminant
This document is copyrighted by the American Psychological Association or one of its allied publishers.

alpha for Samples 1 and 3, respectively, were: .83 and .84 (per-
sonal); .76 and .78 (collective); and .78 and .77 (social). Validity as well as convergent validity of scores on the style scales.
data are presented by Cheek (1989) and Cheek et al. (1994). Correlations between the informational scale and the foreclo-
sure and diffusion status scales and the collective and social
identity scales were relatively low (r2 ⱕ .04). Likewise, limited
Correlational Results (Samples 1 and 3) associations were found between the normative scale and the
The correlations between the scores on each style variable and achievement, moratorium, and diffusion status scales and the
those on the validation-criterion measures appear in Table 3. In personal identity scale and between diffuse-avoidant scores and
both samples, the following theoretically predicted correlations
were found: (a) the informational style was positively associated
with identity achievement, commitment, an emphasis on personal 1
An anonymous reviewer suggested the possibility that similar items
self-components, and both rational and intuitive reasoning; (b) the about beliefs in the commitment and normative scales might have artifi-
normative style was positively correlated with identity foreclosure, cially inflated associations between scores on these two scales. Deleting the
commitment,1 an emphasis on collective self-components, and two belief items from the commitment scale and recomputing the correla-
intuitive reasoning processes; and (c) the diffuse-avoidant style tions and beta coefficients between the normative and commitment scales
in Tables 3 and 4. indicated that this was not the case. Correlations between
was associated positively with a diffuse identity status and an the normative and the nine-item and seven-item commitment scales were,
emphasis on social self-components and negatively with commit- respectively, .27 (p ⬍ .01) and .26 (p ⬍ .01) for Sample 1 and .31 (p ⬍ .01)
ment2 and rational reasoning. Correlations between the style vari- and .30 (p ⬍ .01) for Sample 3. Regression coefficients for the normative
ables for Sample 1 and 3, respectively, were informational and and the nine-item and seven-item commitment scales, with the effects of
normative rs ⫽ –.03 (ns) and .12 (ns); informational and diffuse- the other two style variables controlled were, respectively, .52 (p ⬍ .01)
and .49 (p ⬍ .01) for Sample 1 and .62 (p ⬍ .01) and .57 (p ⬍ .01) for
avoidant rs ⫽ –.19 (p ⬍ .01) and –.19 (p ⬍ .05); and normative Sample 3.
and diffuse-avoidant rs ⫽ .14 (p ⬍ .01) and .16 (p ⬍ .05). 2
An anonymous reviewer was concerned about possible overlap be-
Because of covariation between the ISI-5 variables and those tween items measuring diffuse-avoidance and items measuring commit-
within each block of validation measures (e.g., between the ment. We did a number of additional analyses to examine the degree of
overlap between the diffuse-avoidant style and commitment. We did these
status measures), a series of regression analyses was performed analyses on Sample 2 (i.e., the largest sample in the study). First, we did
to determine the unique relationship between the score on each a principal components analysis (PCA) on the items tapping into diffuse-
style variable and those on the validation measures. In each of avoidance and commitment. A two-component solution with varimax
these analyses, scores on a style variable were regressed on a rotation showed that only one item did not load as expected. The diffuse-
block of validation measures (e.g., the four status measures, the avoidance item reading “I am not sure where I’m heading in my life; I
guess things will work themselves out” did not load on the component
two cognitive measures, or the three identity components mea- defined by the diffuse-avoidance items (.10) and instead loaded negatively
sures) and the two style measures not being regressed. As (–.63) on the component defined by the commitment items. All other
shown in Table 4, the same theoretically consistent pattern of diffuse-avoidance items had their primary loading on the diffuse-avoidance
relationships found with the zero-order correlations was repli- component and did not have cross-loadings ⬎ .40 on the commitment
component. Conversely, all commitment items had their primary loading
cated. In addition, the predicted correlation between the infor- on the commitment component and did not have cross-loadings ⬎ .40 on
mational style and moratorium-status scores obtained in both the diffuse-avoidance component. Second, we did a confirmatory factor
samples when the other statuses and styles were controlled: analysis (CFA), thereby comparing two models: one in which the diffuse-
Sample 1 ␤ ⫽ .32, p ⬍ .01; Sample 3 ␤ ⫽ .33, p ⬍ .01. Because avoidance and commitment items loaded on separate factors and one in
commitment was positively correlated with the informational which these items loaded on a common factor. The fit of the two-factor
model was clearly superior to the fit of a one-factor model [⌬␹2(1) ⫽
style but negatively correlated with identity moratorium, ancil- 185.76; p ⬍ .001]. Modification indices again suggested that the fit of the
lary partial correlations were performed on both samples to model could be improved by allowing the item “I am not sure where I’m
evaluate whether commitment played a role in suppressing the heading in my life; I guess things will work themselves out” to load on the
relationship (see Berzonsky, 1990; Berzonsky & Neimeyer, commitment factor rather than on the diffuse-avoidance factor. Overall
then, it seems like the potential problem of item overlap is limited to only
1994). In Sample 1, when commitment was controlled the one item. We redid the main validation analyses of the diffuse-avoidance
correlation between the informational and moratorium variables scale without this item and found that this did not change the pattern of
increased from .09 (ns) to .29 (p ⬍ .01); in Sample 3 it results.
 900 BERZONSKY ET AL.

Table 3
Bivariate Correlations Between Identity Processing Styles and the Validation Criterion Measures

Informational Normative Diffuse-Avoidant

Validity measures Sample 1 Sample 3 Sample 1 Sample 3 Sample 1 Sample 3

Identity statuses
Achievement .31ⴱⴱ .35ⴱⴱ .12ⴱⴱ .19ⴱ ⫺.40ⴱⴱ ⫺.33ⴱⴱ
Foreclosure ⫺.15ⴱⴱ .01 .62ⴱⴱ .64ⴱⴱ .22ⴱⴱ .20ⴱⴱ
Moratorium .09 .10 ⫺.15ⴱⴱ ⫺.16ⴱ .47ⴱⴱ .48ⴱⴱ
Diffusion ⫺.14ⴱⴱ ⫺.08 ⫺.10 .01 .46ⴱⴱ .58ⴱⴱ

Cognitive reasoning processes

Rational .34ⴱⴱ .31ⴱⴱ ⫺.28ⴱⴱ ⫺.14 ⴚ.33ⴱⴱ ⴚ.37ⴱⴱ
Intuitive .19ⴱⴱ .29ⴱⴱ .20ⴱⴱ .30ⴱⴱ ⫺.03 ⫺.11
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Aspects of identity components

Personal .44ⴱⴱ .41ⴱⴱ .12ⴱ .11 ⫺.24ⴱⴱ ⫺.26ⴱⴱ
Collective .13ⴱ .18ⴱ
.44ⴱⴱ .53ⴱⴱ ⫺.08 ⫺.07
Social .17ⴱⴱ .20ⴱ .37ⴱⴱ .37ⴱⴱ .15ⴱⴱ .20ⴱⴱ

Commitment
Commitment .27ⴱⴱ .30ⴱⴱ .27ⴱⴱ .31ⴱⴱ ⴚ.56ⴱⴱ ⴚ.60ⴱⴱ
Note. Coefficients in bold were predicted to be significant.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01.

foreclosure, intuitive reasoning, and collective identity (r2 ⱕ Avoidant .87; and Commitment .85. Alpha coefficients for the
.04). scores on the ISI-3 scales were as follows: Informational .80;
Normative .77; Diffuse-Avoidant .87; and Commitment .85.
Relationships Between the ISI-5 and ISI-3 Scales The correlations between the scores on the ISI-5 and ISI-3
(Sample 4) scales were as follows: Informational r ⫽ .72; Normative r ⫽ .73;
Diffuse-Avoidant r ⫽ .79; and Commitment r ⫽ .75 (Table 5).
The convergent and discriminant validity of scores on the Although scores obtained with the ISI-3 and ISI-5 are not fully
ISI-5 style scales was also examined by administering the ISI-5 redundant, the strong convergence between both measures suggest
and ISI-3 (Berzonsky, 1992b) scales to the participants in that they essentially tap into the same underlying identity style
Sample 4. Cronbach alphas for the scores on the ISI-5 scales constructs. Evidence for the discriminant validity of the ISI-5 style
were as follows: Informational .86; Normative .82; Diffuse- scores was provided by the relatively low correlations between the

Table 4
Regression Coefficients: Identity Style Regressed on Validation Criterion Measures

Informational Normative Diffuse-Avoidant

Validity measures Sample 1 Sample 3 Sample 1 Sample 3 Sample 1 Sample 3

Identity statuses
Achievement .29ⴱⴱ .29ⴱⴱ .09ⴱ .13ⴱ ⫺.23ⴱⴱ ⫺.21ⴱⴱ
Foreclosure ⫺.17ⴱⴱ ⫺.11 .59ⴱⴱ .58ⴱⴱ .08 .01
Moratorium .32ⴱⴱ .34ⴱⴱ ⫺.19ⴱⴱ ⫺.24ⴱⴱ .32ⴱⴱ .27ⴱⴱ
Diffusion ⫺.16ⴱⴱ ⫺.08 ⫺.05 ⫺.03 .21ⴱⴱ .34ⴱⴱ

Cognitive reasoning processes

Rational .31ⴱⴱ .29ⴱⴱ ⫺.28ⴱⴱ ⫺.13 ⴚ.29ⴱⴱ ⴚ.31ⴱⴱ
Intuitive .17ⴱⴱ .26ⴱⴱ .22ⴱⴱ .29ⴱⴱ .01 ⫺.10

Aspects of identity components

Personal .45ⴱⴱ .38ⴱⴱ ⫺.11 ⫺.17ⴱ ⫺.32ⴱⴱ ⫺.28ⴱⴱ
Collective ⫺.07 ⫺.13 .40ⴱⴱ .55ⴱⴱ ⫺.14ⴱ ⫺.23ⴱ
Social .03 .12 .22ⴱⴱ .06 .34ⴱⴱ .41ⴱⴱ

Commitment
Commitment .31ⴱⴱ .27ⴱⴱ .52ⴱⴱ .62ⴱⴱ ⴚ.64ⴱⴱ ⴚ.71ⴱⴱ
Note. Beta coefficients in bold were predicted to be significant. Effects of the other two style variables and the other dependent variables within each block
are controlled.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01.
 DEVELOPMENT AND VALIDATION OF REVISED IDENTITY STYLE 901

Table 5 .82 to .89 (M ⫽ .85). Test–retest reliabilities of scores on the ISI-5

Correlations Between Scores on the ISI-5 and ISI-3 Identity scales ranged from .77 to .83.
Processing Style Scales Convergent and discriminant validity was evaluated by exam-
ining whether scores on the new scales correlated with measures of
Variable INFO ISI-3 NORM ISI-3 DIFF ISI-3 identity processes and cognitive reasoning in theoretically predict-
INFO ISI-5 .72ⴱⴱ .25ⴱ ⫺.35ⴱⴱ able ways. The validity data were consistent with predictions based
NORM ISI-5 .09 .73ⴱⴱ .10 on the identity style model (see Berzonsky, 1990, 2008, 2011).
DIFF ISI-5 ⫺.25ⴱ ⫺.15 .79ⴱⴱ Informational scores were positively correlated with identity
Note. ISI-5 ⫽ Identity Style Inventory—Version 5; ISI-3 ⫽ Identity achievement, a personal sense of identity, strength of identity
Style Inventory—Version 3 (Berzonsky, 1992b); INFO ⫽ Informational commitment and both rational and experientially based, automatic
Style; NORM ⫽ Normative Style; Diff ⫽ Diffuse-Avoidant Style. intuitive reasoning; they were not strongly correlated with those on
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. the foreclosure, diffusion, collective, and social identity scales.
Normative scores were positively linked with identity foreclosure,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

a collective sense of identity, strength of identity commitment and

scores on each ISI-5 scales and the other two ISI-3 style scales. For
This document is copyrighted by the American Psychological Association or one of its allied publishers.

automatic reasoning but negatively with rational reasoning. Rela-

example, scores on the ISI-5 normative scale did not correlate
tively low associations were obtained between normative scores
significantly with the scores on either the ISI-3 informational or
and those on the achievement, diffusion, moratorium, and personal
diffuse-avoidant scales (Table 5).
identity scales. Diffuse-avoidant scores were positively correlated
with identity diffusion and a socially based sense of identity that
Test–Retest Reliability (Sample 5) highlighted popularity and expectations of others and negatively
The test–retest reliability of the scores on the ISI-5 scales was associated with strength of identity commitment and rational rea-
evaluated by administering the scales twice to the 77 participants soning. Their discriminant validity was indicated by low relation-
in Sample 5 over a 2-week interval. At Time 1, coefficient alpha ships with those on intuitive reasoning and the foreclosure and
for the scores on the informational, normative, diffuse-avoidant, collective identity scales. Not only was the validation effort suc-
and commitment scales, respectively, were .82, .78, .89, and .89. cessful, the findings contribute to a burgeoning empirical literature
At Time 2, they were .80, .80, .89, and .89, respectively. The on identity processing styles (e.g., Berzonsky, 1990, 2004, 2011).
2-week test–retest reliabilities for the scores on the scales were as
follows: Informational r ⫽ .81; normative r ⫽ .78; diffuse- Directions for Future Research
avoidant r ⫽ .77; and commitment r ⫽ .83.
The data presented here indicate that scores on the ISI-5 are
reliable and valid indicators of the identity processing styles and
General Discussion
that they have a clear internal structure. Convergence of scores on
The purpose of the present investigation was to develop and the ISI-5 and ISI-3 scales demonstrates that scores on both the
validate a fifth revised measure of identity processing styles. In ISI-3 and ISI-5 are reliable and valid measures for assessing
contrast to previous style assessments, we sought to develop scales individuals’ identity processing styles. In our view, the most
consisting of items that dealt with content-neutral identity catego- important difference between the ISI-3 and the ISI-5 is that the
ries (e.g., life decisions, goals, beliefs, values, personal problems, ISI-5 is a relatively more generic measure of identity styles.
and so on) instead of statements that pertained to specific identity Contrary to the ISI-3, in which life domains are specified in some
domains (such as religious values, political beliefs, occupational of the items, the items of the ISI-5 refer to the processing of
aspirations, college major, and the like). Our goal was to generate identity-relevant information at a more wholesale, domain-neutral
items that would enable interviewees to decide for themselves level, which enables participants to frame an item (e.g., values)
which content domains were relevant to them instead of requiring within a domain of personal relevance (e.g., religious values,
them to focus on domains specified in the statement. Further, we political values, moral values, and so forth). We believe this type
generated statements worded in the present tense and ones that of assessment may be advantageous for two types of future re-
focused on the processing of identity relevant information rather search on identity styles, that is, longitudinal research and cross-
than the outcome of that processing. national research.
An initial list of 39 statements (13 for each style scale) was First, because of its domain-specific content, scores on the ISI-3
generated and refined via principal factor analyses (PFA). The may yield interpretation problems in longitudinal research. For
PFA yielded three style scales on which items loaded at least instance, if one were investigating changes in identity style scores
.40 with no cross-loadings greater than .35: A nine-item from early to late adolescence, observed changes in the ISI-3 style
Informational-style scale; a nine-item Normative-style scale; and a scores may occur for different reasons. One possibility is that
nine-item Diffuse-avoidant style scale. A confirmatory factor anal- adolescents may have changed their actual style of processing
ysis (CFA) performed on these 27 items indicated that a three- identity-relevant information over time (which would reflect real
factor solution provided the best fit. Further, the three-factor developmental change). However, another possibility is that the
solution held across gender. Across five samples the internal relevance to the adolescents of some of the life domains in
reliabilities (coefficient alpha) of the scores on the ISI-5 style the items may have changed. A third possible explanation is that
scales varied from .74 to .86 (Informational, M ⫽ .79); .75 to .82 the extent to which adolescents differentiate between life domains
(Normative, M ⫽ .79); and .71 to .89 (Diffuse-avoidant, M ⫽ .83). may have changed over time, which would influence the consis-
Coefficient alpha for scores on the Commitment scale varied from tency of their style of processing identity-relevant information
 902 BERZONSKY ET AL.

across content-domains. We hypothesize that this type of interpre- university in one U.S. state, it is necessary to ascertain whether the
tation problem may be better addressed with the ISI-5 because results generalize to similar aged students attending universities in
participants can interpret its items in terms of the life domains that other U.S. states as well as other countries.
are most relevant to them at the particular developmental stage
they are in. Future longitudinal research is needed to evaluate this Conclusion
hypothesis.
Further, we hypothesize that scores on the ISI-5 may provide a The generic nature of the ISI-5 processing scales enables inter-
less culturally biased assessment of identity styles than those on viewees to decide for themselves which identity content to focus
the ISI-3 scales and may, as such, be relatively more useful in upon. The factor structure, reliability, and validity of the scores on
cross-national research. One striking finding with the ISI-3 was the revised scales were found to be acceptable. Accordingly, the
that the reliability of its scores (and, in particular, the reliability of ISI-5 promises to be useful for researchers interested in investi-
the normative scores) was higher in U.S. samples compared to gating individual differences in identity processing style, including
non-U.S. samples (e.g., Berzonsky, Branje, & Meeus, 2007; Du- those researchers interested in examining identity styles from a
developmental and/or cross-cultural perspective.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

riez & Soenens, 2006). We speculate that this phenomenon may be

This document is copyrighted by the American Psychological Association or one of its allied publishers.

at least partially due to the domain-specific formulation of some of

the ISI-3 items. Some of the domains specified in the ISI-3 may be
3
comparatively less relevant in countries outside the United States An anonymous reviewer disputed our suggestion that the Identity Style
Inventory may be suitable for cross-cultural research arguing that the
and/or there may be between-country variability in the extent to model seemed to reflect an “overwhelmingly Western” self-directed em-
which adolescents’ style of processing differs between life do- phasis on informed decision making, which devalues a normative orienta-
mains. These between-country differences in the relevance of life tion to identity formation. However, numerous studies conducted in what
domains may attenuate the reliability of the ISI-3 scores and, more might be considered to be “Western countries” such as the United States,
Canada, Finland, and Belgium have found that a normative style is posi-
important, may hamper a fair and straightforward comparison of
tively associated with: being conscientious and self-controlled; having firm
the prevalence and correlates of identity style scores between commitments and a clear sense of purpose and direction; and having a
nations and samples with different cultural backgrounds.3 This positive sense of well-being such as high self-esteem and low depressive
issue needs to be addressed in future research. reactions (Berzonsky, 2011). Further, evidence suggests that neither the
informational nor normative styles are inherently adaptive or maladaptive.
For example, Smits et al. (2010) hypothesized that a normative (and
Limitations informational) could be adopted for autonomous (freely chosen) or con-
trolled (parental pressure and concerns about feelings of guilt and regret)
It should be noted that scores on the ISI-5 scales measure motives. They found that commitment and well-being were positively
perceived rather than actual processing of identity-relevant infor- correlated with autonomous motives for adopting a normative style but
negatively associated with controlled motives. This study raises the pos-
mation. As such, the processing scales are potentially subject to the
sibility that a normative style, where people respect the views and values
same threats as other self-report measures including deliberate of family or authorities, can indeed be adaptive when people endorse their
and/or inadvertent distortion and bias. In addition, individuals can decision to rely on family or authorities. Such a pattern of volitionally
only potentially report on processing operations to which they endorsed adoption of normative expectations is particularly likely to be
have access, which indicates that the scale scores may under- prevalent in non-Western countries. Clearly, more research is needed to
examine the cross-cultural meaning and correlates of the identity styles.
estimate the amount of processing, especially automatic pro-
cessing, that individuals are engaged in. The extent to which
References
self-reported processing scores correlate with actual perfor-
mance on information-processing tasks is a question that needs Adams, G. R. (1999). The objective measure of ego identity status: A
to be addressed. manual on theory and test construction. Guelph, Ontario, Canada: Uni-
Future research needs to consider whether scores on translated versity of Guelph.
versions of the scales will demonstrate the same psychometric Balistreri, E., Busch-Rossnagel, N., & Geisinger, K. F. (1995). Develop-
properties. As noted before, the internal reliability estimates of ment and preliminary validation of the Ego Identity Process Question-
naire. Journal of Adolescence, 18, 179 –192. doi:10.1006/jado.1995
scores on translated versions of the ISI-3 scales, especially the
.1012
normative scale, tend to be lower than those reported with English Berzonsky, M. D. (1988). Self-theorists, identity status, and social cogni-
versions of the scales. Related to this observation and related to our tion. In D. K. Lapsley & F. C. Power (Eds.), Self, ego, and identity:
claim that the ISI-5 is a promising instrument for cross-national Integrative approaches (pp. 243–262). New York, NY: Springer-Verlag.
research, an important direction for future research is to address doi:10.1007/978-1-4615-7834-5_12
the measurement equivalence of ISI-5 style scores across countries Berzonsky, M. D. (1989). Identity style: Conceptualization and measure-
and cultural groups. Once measurement equivalence has been ment. Journal of Adolescent Research, 4, 268 –282. doi:10.1177/
established, scholars may address cross-cultural similarities and 074355488943002
differences in the prevalence, developmental course, antecedents, Berzonsky, M. D. (1990). Self-construction over the life-span: A process
and outcomes of the scores on the ISI-5. Likewise, the participants perspective on identity formation. In G. J. Neimeyer & R. A. Neimeyer
(Eds.), Advances in personal construct psychology (Vol. 1, pp. 155–
in the present studies were predominantly Caucasian university
186). Greenwich, CT: JAI Press.
students who represented a relatively narrow range of age and Berzonsky, M. D. (1992a). Identity style and coping strategies. Journal of
education; it remains to be demonstrated whether scores on the Personality, 60, 771–788. doi:10.1111/j.1467-6494.1992.tb00273.x
ISI-5 scales are valid indicators of identity processing in partici- Berzonsky, M. D. (1992b). Identity Style Inventory (ISI-3): Revised ver-
pants of different ages, ethnic backgrounds, and/or levels of edu- sion. Unpublished measure, Department of Psychology, State University
cation. Also, because the participants were attending the same of New York, Cortland, NY.
 DEVELOPMENT AND VALIDATION OF REVISED IDENTITY STYLE 903

Berzonsky, M. D. (1994). Self-identity: The relationship between process Duriez, B., & Soenens, B. (2006). Personality, identity styles, and reli-
and content. Journal of Research in Personality, 28, 453– 460. doi: giosity: An integrative study among late and middle adolescents.
10.1006/jrpe.1994.1032 Journal of Adolescence, 29, 119 –135. doi:10.1016/j.adolescence.2004
Berzonsky, M. D. (2004). Identity processing style, self-construction, and .11.007
personal epistemic assumptions: A social-cognitive perspective. Euro- Epstein, S. (1990). Cognitive-experiential theory. In L. Pervin (Ed.), Hand-
pean Journal of Developmental Psychology, 1, 303–315. doi:10.1080/ book of personality theory and research (pp. 165–192). New York, NY:
17405620444000120 Guilford Press.
Berzonsky, M. D. (2005a). Ego-identity: A personal standpoint in a post- Epstein, S., Pacini, R., Denes-Raj, V., & Heier, H. (1996). Individual
modern world. Identity: An International Journal of Theory and Re- differences in intuitive-experiential and analytical-rational thinking.
search, 5, 125–136. doi:10.1207/s1532706xid0502_3 Journal of Personality and Social Psychology, 71, 390 – 405. doi:
Berzonsky, M. D. (2005b). Identity processing style and self-definition: 10.1037/0022-3514.71.2.390
Effects of a priming manipulation. Polish Psychological Bulletin, 36, Erikson, E. (1964). Insight and responsibility. New York, NY: Norton.
137–143. Erikson, E. (1968). Identity: Youth and crisis. New York, NY: Norton.
Berzonsky, M. D. (2008). Identity formation: The role of identity process- Grotevant, H. D. (1987). Toward a process model of identity. Journal of
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ing style and cognitive processes. Personality and Individual Differ- Adolescent Research, 2, 203–222. doi:10.1177/074355488723003
This document is copyrighted by the American Psychological Association or one of its allied publishers.

ences, 44, 645– 655. doi:10.1016/j.paid.2007.09.024 Grotevant, H. D., & Adams, G. R. (1984). Development of an objective
Berzonsky, M. D. (2011). A social-cognitive perspective on identity con- measure to assess ego identity in adolescence: Validation and replica-
struction. In S. J. Schwartz, K. Luyckx, & V. Vignoles (Eds.), Handbook tion. Journal of Youth and Adolescence, 13, 419 – 438. doi:10.1007/
of identity theory and research: Structures and processes (Vol. 1, pp. BF02088639
55–76). London, England: Springer. doi:10.1007/978-1-4419-7988-9_3 Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance
Berzonsky, M. D., & Adams, G. R. (1999). The identity status paradigm: structure analysis: Conventional criteria versus new alternatives. Struc-
Still useful after thirty-five years. Developmental Review, 19, 557–590. tural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118
doi:10.1006/drev.1999.0495 Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from
Berzonsky, M. D., & Barclay, C. R. (1981). Formal reasoning and identity childhood to adolescence. New York, NY: Basic Books. doi:10.1037/
formation: A reconceptualization. In J. A. Meacham & N. R. Santilli 10034-000
(Eds.), Contributions to human development (Vol. 5, pp. 61– 87). Basel,
Jöreskog, K. G., & Sörbom, D. (1993). LISREL VIII user’s reference guide.
Switzerland: Karger.
Mooresville, IN: Scientific Software.
Berzonsky, M. D., Branje, S. J. T., & Meeus, W. (2007). Identity process-
Kerpelman, J., Pittman, J. F., & Lamke, L. M. (1997). Toward a micro-
ing style, psychosocial resources, and adolescents’ perceptions of
process perspective on adolescent identity development: An identity
parent-adolescent relations. The Journal of Early Adolescence, 27, 324 –
control theory approach. Journal of Adolescent Research, 12, 325–346.
345. doi:10.1177/0272431607302006
doi:10.1177/0743554897123002
Berzonsky, M. D., & Ferrari, J. R. (2009). A diffuse-avoidant identity
Kline, R. B. (2005). Principles and practice of structural equation mod-
processing style: Strategic avoidance or self confusion? Identity: An
eling. New York, NY: Guilford Press.
International Journal of Theory and Research, 9, 145–158. doi:10.1080/
Luyckx, K., Lens, W., Smits, I., & Goossens, L. (2010). Time perspective
15283480802683607
and identity formation: Short-term longitudinal dynamics in college
Berzonsky, M. D., Macek, P., & Nurmi, J.-E. (2003). Interrelations among
students. International Journal of Behavioral Development, 34, 238 –
identity process, content, and structure: A cross-cultural investigation.
247. doi:10.1177/0165025409350957
Journal of Adolescent Research, 18, 112–130. doi:10.1177/
Marcia, J. E. (1966). Development and validation of ego identity status.
0743558402250344
Berzonsky, M. D., & Neimeyer, G. J. (1994). Ego identity status and Journal of Personality and Social Psychology, 3, 551–558. doi:10.1037/
identity processing orientation: The moderating role of commitment. h0023281
Journal of Research in Personality, 28, 425– 435. doi:10.1006/jrpe.1994 Marcia, J. E. (1993). The status of the statuses: Research review. In J. E.
.1030 Marcia, A. S. Waterman, D. R. Matteson, S. L. Archer, & J. L. Orlofsky
Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of (Eds.), Ego identity: A handbook for psychosocial research (pp.
Personality and Social Psychology, 42, 116 –131. doi:10.1037/0022- 22– 41). New York, NY: Springer-Verlag. doi:10.1007/978-1-4613-
3514.42.1.116 8330-7_2
Cacioppo, J. T., Petty, R. E., & Kao, C. (1984). The efficient assessment Marcia, J. E. (1999). Representational thought in ego identity, psychother-
of need for cognition. Journal of Personality Assessment, 48, 306 –307. apy, and psychosocial developmental theory. In I. E. Sigel (Ed.), De-
doi:10.1207/s15327752jpa4803_13 velopment of mental representation: Theories and application (pp. 391–
Cattell, R. B. (1966). The scree test for the number of factors. Multivariate 414). Mahwah, NJ: Erlbaum.
Behavioral Research, 1, 245–276. doi:10.1207/s15327906mbr0102_10 Missotten, L., Luyckx, K., Vanhalst, J., Branje, S., & Goossens, L. (2011).
Cheek, J. M. (1989). Identity orientations and self-interpretation. In D. M. Identity styles and conflict resolution styles in adolescence: Associations
Buss & N. Cantor (Eds.), Personality psychology: Recent trends and in mother– daughter dyads. Journal of Youth and Adolescence, 40,
emerging directions (pp. 275–285). New York, NY: Springer-Verlag. 972–982. doi:10.1007/s10964-010-9607-5
doi:10.1007/978-1-4684-0634-4_21 Rowe, I., & Marcia, J. E. (1980). Ego identity status, formal operations,
Cheek, J. M., Tropp, L. R., Chen, L. C., & Underwood, M. K. (1994, and moral development. Journal of Youth and Adolescence, 9, 87–99.
August). Identity orientations: Personal, social, and collective aspects doi:10.1007/BF02087928
of identity. Paper presented at the meeting of the American Psycholog- Satorra, A., & Bentler, P. M. (1994). Corrections to test statistics and
ical Association, Los Angeles, CA. standard errors in covariance structure analysis. In A. von Eye & C. C.
Doumen, S., Smits, I., Luyckx, K., Duriez, B., Vanhalst, J., Verschueren, Clogg (Eds.), Latent variable analysis: Applications in developmental
K., & Goossens, L. (2012). Identity and perceived peer relationship research (pp. 399 – 419). Thousand Oaks, CA: Sage.
quality: The mediating role of attachment-related emotions. Journal of Smits, I. (2009). Identity styles in adolescence: Measurement and its
Adolescence, 35, 1417–1425. associations with perceived parenting, personal well-being, and inter-
 904 BERZONSKY ET AL.

personal functioning (Unpublished doctoral dissertation). Catholic Uni- cents’ identity style. Journal of Youth and Adolescence, 39, 1343–1356.
versity of Leuven, Leuven, Belgium. doi:10.1007/s10964-009-9469-x
Smits, I., Berzonsky, M. D., Soenens, B., Luyckx, K., Goossens, L., Soenens, B., Duriez, B., & Goossens, L. (2005). Social-psychological
Kunnen, S., & Bosma, H. (2008). The Identity Style Inventory— 4: profiles of identity styles: Attitudinal and social-cognitive correlates in
Internal research report. Leuven, Belgium: Catholic University Leuven. late adolescence. Journal of Adolescence, 28, 107–125. doi:10.1016/j
Smits, I., Doumen, S., Luyckx, K., Duriez, B., & Goossens, L. (2011). .adolescence.2004.07.001
Identity styles and interpersonal behavior in emerging adulthood: The Vleioras, G. (2007). Not all identity style items refer to identity: Does it
intervening role of empathy. Social Development, 20, 664 – 684. doi: matter? Identity: An International Journal of Theory and Research, 7,
10.1111/j.1467-9507.2010.00595.x 255–262. doi:10.1080/15283480709336933
Smits, I., Soenens, B., Vansteenkiste, M., Luyckx, K., & Goossens, L. Whitbourne, S. K., & Weinstock, C. S. (1979). Adult development: The
(2010). Why do adolescents gather information or stick to parental differentiation of experience. New York, NY: Holt, Rinehart and
norms? Examining autonomous and controlled motives behind adoles- Winston.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Appendix
Revised Commitment Scale

1. I know basically what I believe and don’t believe. 6. I have clear and definite life goals.

2. I know what I want to do with my future. 7. I am not sure what I want out of life. (reversed)

3. I am not really sure what I believe. (reversed) 8. I have a definite set of values that I use to make personal
decisions.
4. I am not sure which values I really hold. (reversed)
9. I am emotionally involved and committed to specific
5. I am not sure what I want to do in the future. (reversed) values and ideals.
Received September 9, 2011
Revision received October 29, 2012
Accepted February 22, 2013 䡲

View publication stats