#2737—Social Psychology Quarterly—VOL. 69 NO.

1—69106-burke

Social Psychology Quarterly

2006, Vol. 69, No. 1, 81–96

Identity Change*
PETER J. BURKE
University of California, Riverside

This research examines two mechanisms by which persons’ identities change over time.
First, on the basis of identity control theory (ICT), I hypothesize that while identities
influence the way in which a role is played out, discrepancies between the meanings of
the identity standard and the meanings of the role performance will result in change.
That is, due to the hierarchical structure of identity systems, change will occur not only
to the role performance (to counteract the discrepancy), but also to the meanings of
identity standard over time (to bring them more into line with the disturbance). A sec-
ond mechanism of change in identities is hypothesized to result from persons holding
multiple identities that share meanings. Identities that share dimensions of meaning
influence each other’s standard to maintain the shared meaning at a common level.
Changing identity standards redefines who one is. Using data from the Marital Roles
Project, I examine the spousal and gender identities of newly married couples over the
first years of marriage as couples engage in routine activities. Hypotheses about change
in identities are supported; implications for extending identity control theory are dis-
cussed.

The question of how identities change that define who one is, and identity change as
has been a topic of theoretical interest for a change in these meanings. Then I point out
number of years (Burke and Cast 1997; that these meanings, even while providing the
Deaux 1993; Gecas and Mortimer 1987; standard against which self-relevant mean-
Kiecolt 1994, 2000; McNulty and Swann 1994; ings in the situation are judged, are them-
Serpe 1987). That identities change has not selves the product of higher levels in the
been in dispute, but the demonstration of the identity system; therefore they are dynamic
theoretical mechanisms involved in this and change, though at a much slower rate
change has not been resolved fully, in part than the meanings in the situation. Because
because such mechanisms must account for these self-meanings held in identity stan-
both the stability and the change of identities dards are dynamic, the conditions under
over time. Identity control theory (ICT), with which they may change can be seen readily in
its hierarchical view of identities as control the way the whole identity system operates;
systems, is able to address these issues. hypotheses about identity change may be
I begin by reviewing the central aspects derived from these principles.
of ICT that are relevant to our understanding
of both change and stability. First, to be clear IDENTITY CHANGE IN THE
about what identity change involves, I discuss CONTEXT OF ICT
the nature of identities as the self-meanings
Identities As Meanings

* An earlier version of this paper was presented at Within ICT, an identity is viewed as a set
the meetings of the Pacific Sociological Association of self-relevant meanings held as standards
held in Vancouver in April 2002. The research report- for the identity in question (Burke 1980,
ed here is based on data from a longitudinal study of
1991; Burke and Tully 1977). For the role
first-married couples, “Socialization Into Marital
Roles,” funded by NIMH Grant MH46828, under the identity of spouse, for example, the standard
direction of Irving Tallman, Peter J. Burke, and Viktor would include what it means to be a husband
Gecas. I would like to thank Jan Stets for her com- or a wife. Following the work of Osgood,
ments on an earlier version. Address all correspon-
Suci, and Tannenbaum (1957), meanings are
dence to Peter J. Burke, Department of Sociology,
University of California, Riverside, CA 92521–0419; understood to be one’s mediational respons-
peter.burke@ucr.edu. es to stimuli. These responses include not
81
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

82 SOCIAL PSYCHOLOGY QUARTERLY

only what are typically understood to be the els of “task-orientation” to perhaps consid-
denotative responses, but also the connota- ering levels of “dominance”). In the present
tive (including affective) responses. In this research I consider only changes in the
view, responses are assessed along the strength of the response, with the dimen-
dimensions that underlie the meanings. sions fixed.1
Osgood and his colleagues (1957) devised
the semantic differential as a method for the Managing Situational Meanings
quantitative measurement of meaning in
terms of the strength of a person’s response To see how identity meanings change,
along various underlying dimensions. Burke first we must understand that within ICT,
and Tully (1977) adapted this procedure to the identity standard, or set of meanings
allow measurement of self-meanings, or defining the identity, is part of a dynamic,
reflexive responses to the self as a stimulus, self-regulating control system that operates
and showed how to measure self-meanings when an identity is activated. To begin, I out-
line the basic identity system and its func-
along culturally shared and relevant dimen-
tioning; then, I discuss how the meanings
sions. Examples of such dimensions that
can change.
have been useful in understanding identities
In addition to the identity standard or
are masculinity/femininity for the gender
set of meanings defining who one is (as a
identity (Burke 1989; Burke and Tully 1977),
spouse), there is a set of perceptions of self-
“intellectualism” and “sociability” for the
relevant meanings in the social situation, as
student identity (Burke and Reitzes 1981),
illustrated in Figure 1. In identity 2, for
and “task-orientation” for a leader identity
example, depicted on the lower right-hand
(Burke 2003).
side of the figure, we see that when the iden-
For each identity there is a standard that
tity is activated, the individual perceives the
indicates the level of each dimension of
meanings implied by his or her ongoing
meaning (for example, of “task-orientation”
behavior in the situation (either observed
with respect to the leader identity), which
directly or received through reflected
defines the person’s identity: what it means
appraisals). As Burke and Reitzes (1981)
to be who one is. The meanings that define
showed, the link between identities and
an identity are the identity standards of any
behavior lies in the shared meanings of
group-, role-, or person-based identity such
each: people engage in behavior to create
as American, spouse, or “honest.” Change in
meanings that correspond to the meanings
identities thus refers to changes in the
of their identity standard.
meanings within the identity standard—
The perceptions of these self-relevant
changes, for example, in what it means to be
meanings are fed into the comparator, a
a spouse.
mechanism that compares one’s perceptions
Insofar as persons make the same
of self-relevant meanings with the self-
responses to stimuli (similar strength of
defining meanings of the identity standard.
response along the same dimensions), the
Differences between these two are output as
meaning is shared and the stimulus is a sig- an error or discrepancy signal. The discrep-
nificant symbol. Shared meaning allows ancy represents a lack of correspondence
communication as well as shared under- between the meanings in the identity stan-
standings and expectations. Measuring self- dard and the meanings in the situation.
meaning thus involves measuring the Another element is one’s social behav-
strength of a person’s responses to the self ior in the situation. As shown in Figure 1,
along the relevant dimensions (Burke and behavior is a function of the error or output
Tully 1977). Thus identities can change in of the comparator. The meanings implied by
two ways: by changes in the strength of the
response along a given dimension (e.g., how
1 Future research must deal with questions of shifts
“task-oriented” one is as a leader), and by
in the dimensions of meaning or of the emerging rele-
changes in which dimensions are relevant
vance of new dimensions: for example, changes in
for a particular identity (e.g., changing what what it means to be male or female that occur when
it means to be a leader from considering lev- one goes through puberty.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 83

Figure 1. Identity Model Showing Two Levels and Two Identities

the behavior change meanings in the situa- controlled (by modifying the meaningful
tion, thus altering reflected appraisals and behavior in the situation) to match mean-
perceptions, and reducing any discrepancy ings in the identity standard. Should some
(any error between perceptions of self-rele- disturbance in the situation (perhaps anoth-
vant meanings and the meanings held in the er’s behavior) cause the perceived self-
identity standard). This process of making meanings to deviate from the identity
self-perceptions match the identity standard standard, the person will behave so as to
is the process of identity verification. In this shift the perceived self-meanings back into
sense, an identity is a perceptual control sys- agreement with the standard.
tem (Powers 1973). When an identity is acti- The error or discrepancy between the
vated, 2 perceptions of meanings are perceptions and the identity standard not
only governs behavior, but also produces an
2 Activation is the process of bringing an identity
emotional response. We feel distress when
“on line” to control perceptions. Theoretically this the discrepancy is large or increasing; we
occurs when identity-relevant meanings are perceived feel good when the discrepancy is small or
in the situation. Perceptions that do not exist cannot decreasing (Burke and Harrod 2005; Cast
be controlled, and without relevant stimuli in the situ-
and Burke 2002). These emotional respons-
ation there can be no relevant perceptions (mean-
ings) to be controlled. As Oakes (1987) points out, es provide some motivation for reducing
however, this is not a matter of the attention-grabbing any discrepancies between perceptions and
properties of social stimuli but a combination of
accessibility and fit. Accessibility is the readiness of a
particular dimension of meaning to become activated erties of the situation are congruent with the dimen-
in a person; fit is the degree to which perceived prop- sion of meaning (Oakes 1987).
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

84 SOCIAL PSYCHOLOGY QUARTERLY

the standard, and in general for keeping Standards adjust at a much slower rate, how-
them small. ever; thus, in the normal course of events, we
may not notice that the standard has adjusted
Managing Self-Meanings at all. Yet, when our behavior, for some rea-
son, does not reduce the discrepancy, or when
Thus far the model assumes that the we are prevented from countering the distur-
identity standard is unchanging and that bance so as to change our perceptions, the
behaviors change situational meanings, standard will continue over time to change
thereby leading to changes in perceptions to toward the perceptions until the error signal
match the unchanging standard. This view, is reduced to zero — that is, until our percep-
however, is incomplete. The identity standard tions match the changed standard.
in ICT is itself the output of a higher-level
control system. This is illustrated in Figure 1, The Sources of Identity Change
where we see a lower-level control system
(for example, identity 2, with its identity stan- Identity change thus is ubiquitous in ICT,
dard, perceptions, comparator, and behav- but in normal circumstances the change is
ior), and a higher-level control system small and slow. Insofar as identity verifica-
(identity 0), in which the output of the system tion occurs and people successfully bring
(corresponding to the behavior in the lower- meanings in the situation into agreement
level system) is the identity standard of the with the meanings in their identity standards,
lower-level system. When the higher-level the standards are subject to little systematic
system is activated, it changes the meanings pressure to change in any particular direc-
tion. Slow fluctuations may occur around a
of the standard in the lower-level identity sys-
central value, but these should be small. The
tem as the mechanism by which the higher-
conditions for dramatic identity change are
level control system controls the higher-level
manifested only in unusual circumstances
perceptions.3
such as religious cults or the prisoner of war
This higher-level control system (for
camps in the Korean conflict (Schein 1958).4
example, an identity associated with a master
Smaller changes that accumulate over time
status or a personal identity that operates
are more common.
across role identities and situations) also pos-
Burke and Cast (1997), for example, doc-
sesses perceptions, standard, and compara- umented the changes in the meanings of the
tor, and is also a perceptual control system. If gender identities of newlywed husbands and
this higher-level identity system is to control wives that occur with the birth of their first
its perceptions, it must alter its outputs, which child. This birth represents a change of mean-
are the standards of the lower-level systems ings in the environment (a disturbance in the
(Burke 1997; Tsushima and Burke 1999). model) that persists and is not countered eas-
Because both higher- and lower-level control ily. Under these conditions, Burke and Cast
systems operate continuously when activat- found that husbands’ gender identity became
ed, the identity standards as well as percep- somewhat more masculine over the year fol-
tions and behaviors are changing in the lowing the child’s birth, while that of wives
situation. The higher- and lower-level control became somewhat more feminine. The pres-
systems operate at different rates, however. ence of a child changed the meanings in the
Behavior in the situation adjusts so as to interactional setting in a way that was not
alter perceptions of meaning at all levels and easily changed back or countered; thus a con-
thus to correct discrepancies as they occur. tinuing discrepancy was present between sit-
The standard also adjusts continuously so as uationally self-relevant meanings and the
to reduce discrepancies as they occur by
moving toward the current perceptions.
4 The prevention of any control over perceptions is

primary among the techniques used. Thus individuals

in these conditions are allowed to perceive only what
3 The higher-level perceptions are patterns or com-
the “captors” wish them to perceive. Over time, then,
binations of the perceptions from the lower-level sys- identity standards shift in the direction of the allowed
tem. perceptions.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 85
meanings contained in the couple’s identity 1982, 1994), and the degree to which each
standards. This prolonged discrepancy over identity is tied to other identities in the full
time allowed each of the identities to change set of identities held by this individual
slowly to match the perceptions of the new (Burke 2003; Smith-Lovin 2003; Thoits 1986).
self-relevant meanings in the situation; in this If commitment to one identity is stronger
way the discrepancies were reduced slowly than to another, the more highly committed
over time toward zero. This is the first way in identity may change less than that to which
which identities change. the commitment is smaller. That is, if one
Identity change also may occur when identity entails more ties to others, more oth-
people have multiple identities that are relat- ers will expect to see the meanings of that
ed to each other in the sense that they share identity expressed continually, and the costs
meanings and are activated at the same time of changing that identity will be greater.
(Burke 2003; Deaux 1992, 1993; Stets 1995). Similarly, if one identity is more salient than
Thus, as one controls perceptions of self-rele- another—that is, more likely to be activated
vant meanings to match the standard for one in a situation—more occasions exist to make
identity, they may become discrepant with demands for portrayal of particular mean-
the standard for another active identity if ings; thus it is more difficult to change that
they also are relevant for that identity.5 For identity standard.
example, one person’s gender identity as a Thus we identify two general sources of
woman may suggest that she must be strong systematic identity change in ICT: persistent
and independent, but her wife identity may problems with the verification of a particular
suggest that she must let her husband take identity, and multiple identities activated
the lead in family matters. Insofar as these together, whose verifications require oppos-
identities are activated at the same time and ing meanings to be manifested in the individ-
she cannot act on the basis of one without ual’s behavior. The difference between these
creating a discrepancy with respect to the sources of change lies in the source of the
other, the two identities are in conflict. She conflict of meanings. In the first case, the
cannot reduce both discrepancies at the same source is a disturbance to the meanings in the
time. external situation, causing them to be per-
To continue this example, as these dis- ceived as discrepant from the meanings of
crepancies persist, ICT suggests that the the identity standard. In the second case, it is
identity standards for both of the woman’s an internal conflict manifested when two
identities will shift slowly toward each other, identities, each controlling the same dimen-
becoming identical at some “compromise” sion of meaning, but to different levels, are
position so that meaningful behavior can ver- activated at the same time.6 In view of each of
ify both identities at the same time. She may these sources, the meanings in the identity
become less strong and independent in her standard(s) are likely to change in the service
gender identity; at the same time, she may of making identity verification possible. What
become less likely to let her husband always
it means to be who one is will change.
take the lead in family matters. In this case,
To make this explanation more concrete,
the meanings in both identity standards have
I consider a discrepancy7 or difference along
shifted. The extent to which each of the stan-
dards changes depends upon other factors
such as the degree of commitment to each of 6 This might happen when an identity developed in
the identities (Burke and Reitzes 1991; one context becomes activated in another context: for
Burke and Stets 1999; Stryker and Serpe example, trying to talk on the phone to one’s
1982), the degree of salience of each of the boyfriend while one’s husband is in the same room.
7 I use the term discrepancy here as a shorthand for
identities (Callero 1985; Stryker and Serpe
the difference along some dimension of meaning
between the meanings in an identity standard and the
meanings one perceives about who one is in a situa-
tion. This difference sometimes is called an error
5 Role conflict and status inconsistency are exam-
because the self-meanings in the situation are not the
ples of situations that may be interpreted as identity same as the reference of the self-meanings in the stan-
conflicts of this type. dard.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

86 SOCIAL PSYCHOLOGY QUARTERLY

the dimension of traditional femininity meanings of the identity standard on the
between the meanings in an identity standard same dimension.
and the identity-relevant meanings perceived
Again, with respect to the spousal identi-
in a situation. Let us say that a woman in her
ty, this hypothesis suggests a negative feed-
spouse identity sees herself in the situation as
back loop such that when self-meanings in
acting more traditionally feminine than she
the situation are more “traditionally femi-
defines herself in her identity standard: that
nine” than the self-meanings in the identity
is, she is coming across as too traditionally
standard, the person will increase the degree
feminine in her spouse role. This discrepancy
of femininity in her identity standard. As sug-
will have two effects. First, it will cause her to
gested above, however, this change is slow,
change her behavior so as to change the situ-
and the discrepancy therefore must persist
ational meanings (how she is coming across)
long enough to allow small changes to accu-
to be less traditionally feminine. Second, at
mulate. Such changes might be expected to
the same time, the discrepancy will act slowly
occur over time. This feedback also is nega-
to change the meanings held in her identity
tive because the effect of the discrepancy is to
standard so as to be more traditionally femi-
reduce the discrepancy over time.
nine, moving them closer to a match with the
The second source of identity change,
way she is coming across in the situation.
other identities, is the subject of the last
In additional, other identities that share
hypothesis.
dimensions of meaning (for example, a gen-
der identity) will cause a change in the mean- Hypothesis 3: A focal identity that shares a
ings in the spousal identity standard. For dimension of meaning with a second identity
example, if the person sees herself in her gen- will change in a positive direction the mean-
der identity as very feminine, this will influ- ings of the second identity on the same
ence the degree of femininity of her spousal dimension to be consistent with the focal
identity, and the degree of femininity of her identity (and vice versa).
spousal identity in turn will influence the Again, this feedback may be regarded as
degree of femininity of her gender identity. negative because the discrepancy between
These ideas are formalized in the following the degrees of meaning on some dimension
hypotheses: held by two identities is reduced by changes
Hypothesis 1: The discrepancy between per- in each of the identities. When two identities
ceptions and identity standard on a particular are trying to control the same dimension of
dimension of meaning will negatively influ- meaning to exist at different levels, they are
ence the meanings of the role performance in conflict. The conflict is resolved when the
on the same dimension. two identity standards come to agree on the
level at which the dimension of meaning
With respect to the spousal identity, for
should be set. These changes in identity stan-
example, this hypothesis suggests a negative
dards would be expected to occur slowly over
feedback loop such that when self-meanings
time. If the two identities do not share any
in the situation are more “traditionally femi-
dimensions of meaning, neither will influence
nine” than the self-meanings in the identity,
the other because they operate independent-
the person will decrease the degree of femi-
ly.
ninity in her role performance until the self-
meaning in the situation matches the
self-meaning in the identity standard and the METHOD
discrepancy is reduced to zero. This hypothe- To test these hypotheses, I investigate
sis pertains to changing meanings in the situ- two identities that share dimensions of mean-
ation, not in the identity; the latter is the topic ing. As suggested above, the focal identity is
of the next two hypotheses. the spouse identity; gender identity is the sec-
Hypothesis 2: The discrepancy between per- ond identity, which shares meanings with the
ceptions and standard on a particular dimen- first. The behavior that is involved pertains to
sion of meaning will positively influence the the spousal role.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 87
The data for this research come from a analysis are based on information from the
longitudinal study of marital roles that inves- interviews and on the daily diaries in all three
tigated marital dynamics in the first two years data collection points over the two-year span.
of marriage (Tallman, Burke, and Gecas
1998). The sample for this study was drawn Measures
from marriage registration records in 1991
and 1992 in two midsized communities in The spousal role identity (standard) was
Washington State. Of those couples recorded measured by asking respondents to rate each
in the marriage registry during this period, of eight spousal role activities by how much
about 45 percent (574 couples) met the crite- they felt that they should engage in that
ria for involvement: both spouses were over activity. It is not the role activities themselves
age 18, were involved in their first marriage, that are important, however, but what it
and had no children living with them. means to engage in those activities. Thus the
Of the couples meeting the criteria for sampling of role activities should capture
involvement in the longitudinal study, 286 important underlying dimensions of the
couples completed all the data collection in spousal identity of meaning. Response cate-
the first year. The couples do not differ signif- gories for all the included items ranged from
icantly from couples throughout the United “not doing that activity in the household”
States who marry for the first time. For exam- (coded 0) to “doing all of that activity in the
ple, their mean age is similar to the national household” (coded 4). These items were fac-
mean age of people marrying for the first tored and displayed a single underlying
time (about 25), and their mean educational dimension of meaning with an omega relia-
levels resemble the national level of persons bility of .90. Four items were reverse coded,
marrying for the first time (“some college”) as indicated in Table 1; the standardized
(Vital Statistics of the United States 1987). In items then were added to form a scale.8 We
the United States, first-married persons typi- did this for each of the three time points.9
cally are white (86%) (Vital Statistics of the High scores on the underlying dimension of
United States 1987). In the present sample, 89 meaning represent a more (traditionally)
percent are white, 3 percent are black (under- feminine definition of the spouse identity.
representing blacks nationally), and 9 per- The items are presented in Table 1.
cent are other minorities (overrepresenting The measure of the meanings of the per-
Asians and Hispanics nationwide). This sam- ceived spousal role performance was derived
ple reflects the racial distribution in from items in the daily diaries that each
Washington State (World Almanac and Book respondent kept. Respondents indicated the
of Facts 1992). extent (in time) to which each of eight activi-
Attrition was 15 percent from year 1 to ties was undertaken on each of the 28 days of
year 2, and 4 percent from year 2 to year 3. responses recorded in the daily diaries. These
These figures do not include the 13 couples activities included a variety of spousal activi-
who were separated or divorced after year 1 ties: cooking, cleaning, earning an income,
nor the 16 couples who were separated or talking and sharing with the family. Again,
divorced after year 2, who were no longer the actual items are less important than the
included in the sample. Couples who dropped
out of the study after the first or second year
were more likely to be young (p < .01), less 8 In this and the other scales, the meanings of the

highly educated (p < .01), and of a lower items emerge from the pattern of the participants’
responses rather than from any arbitrary assignment
socioeconomic status (p < .01).
of particular meanings to particular items.
Each data-collection period included a 9 We find some indication that in the third year, the

90-minute face-to-face interview, four one- items relating to cleaning, meal preparation, and
week daily diaries kept by respondents at 10- laundry became slightly more important in defining
week intervals, and a 15-minute videotaping the dimension of meaning measured here (indicating
a possible shift in the underlying dimension of mean-
of couples’ conversations as they worked to ing). To make the meanings we measured the same
solve areas of disagreement that they had over the three time periods, however, we used the
identified previously. The data for the current same (equal) weights in all the periods.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

88 SOCIAL PSYCHOLOGY QUARTERLY

Table 1. Items for the Spouse Identity Scale and the Spouse Role Performance Scale

A. Spouse Identity Scale:

To what extent should you be responsible for. . .?
—Providing for my family before children are born (R)
—Providing for my family after children are born (R)
—Preparing and serving meals
—Cleaning the house
—Shopping and groceries
—Washing and ironing
—Home repair (R)
—Yard work (R)
——Omega reliability (.90)

B. Spouse Role Performance Scale:

How much did you do this today?
—Providing for my family by working at a job or profession (R)
—Cooking and food preparation for my family
—Cleaning and housekeeping for my family
—Shopping for my family
—Recreational activities with my family
—Talking and sharing with my family
—Talking or visiting with my spouse’s family and relatives
—Talking or visiting with my own family and relatives
——Omega reliability (.83)

fact that they all tap into the same underlying that distinguished significantly between mas-
dimension of meaning for the behavior in culine and feminine meanings across the
question.10 We factored the items to show a three time points, using a discriminant func-
single dimension of underlying meaning; the tion. Applying this function to the self-ratings
item “providing for my family” was reverse yielded a gender identity scale on which high
coded as shown in Table 1. The standardized scores indicated more feminine meanings.12
items were summed to form a scale on which The omega reliability for the scale was .83.
a high score represented the more (tradition- The items are shown in Table 2.
ally) feminine activities for each of the three To be sure that the same dimension of
time points.11 The overall omega reliability meaning was captured by the measures for
for the scale was .83. The items also are listed the spousal role identity, the spousal role per-
formance, and the gender identity, we con-
in Table 1.
ducted a factor analysis of the nine scales
Following the procedures outlined by
(three measures by three time points). The
Burke and Cast (1997), we measured gender
results show a strong single factor with high
identity using the Burke and Tully (1977)
loadings on all measures (mean loading was
method on items taken from the Spence and
.75). We take this as confirmation of the com-
Helmreich (1978) Personal Attributes mon underlying dimension of meaning cap-
Questionnaire (PAQ). We selected 15 items tured by these measures.

10 Coltrane (2000) labels this dimension as repre- The Model

senting gender in its symbolic and performance
aspects. The structural equation model used to
11 As with the measure of the spousal identity, we
test the hypotheses is presented in Figure 2.
find some indication that over time the items relating Here we see the three concepts—the mean-
to cleaning and meal preparation became slightly
ings of the spouse identity standard, the
more important in defining the dimension of meaning
measured here (again, indicating a possible shift in
the underlying dimension of meaning). To make the
meanings we measured the same over the three peri- 12 We find no indication of any shift in the gender

ods, however, we used the same (equal) weights in all meanings measured over time for the present sample.
the periods. We used the same weights in all three time periods.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 89
Table 2. Gender Identity Items

01. Very aggressive—Not at all aggressive

02. Very independent—Not at all independent
03. Not at all excitable in a major crisis—Very excitable in a major crisis
04. Very active—Very passive
05. Very rough—Very gentle
06. Not at all helpful to others—Very helpful to others
07. Very competitive—Not at all competitive
08. Not at all kind—Very kind
09. Feelings not easily hurt—Feelings easily hurt
10. Not at all aware of the feelings of others—Very aware of the feelings of others
11. Never gives up easily—Gives up easily
12. Never cry—Cries very easily
13. Feel very superior—Feel very inferior
14. Very cold in relations with others—Very warm in relations with others
15. Very little need for security—Very strong need for security
00. Omega reliability .83

e1 e2
1 1
Gender Id 1 Gender Id 2 Gender Id 3

e3 e4
1 1
Spouse Id 1 Spouse Id 2 Spouse Id 3

Spouse Role 1 Spouse Role 2 Spouse Role 3

1 1 1

e5 e6 e7
Figure 2. Model of Spousal Identity Change

meanings of the spouse role performance, Central to this analysis is the representa-
and the meanings of the gender identity stan- tion of the discrepancy or difference between
dard—as they relate to each other over time. the perceived spousal identity meanings in
Meanings of the spousal role performance the situation (represented in the meanings of
are an immediate function of the meanings of the spousal role performance) and the mean-
the spousal identity. Meanings in the gender ings in the spousal identity standard. This dis-
crepancy is represented by the error term in
identity standard and the spousal identity
the structural equation model for the mea-
standard influence each other over time (in
sure of the role performance meanings, as
accordance with Hypotheses 2 and 3). No predicted by the meanings of the spouse
direct connection is present between mean- identity standard (e5, e6, and e7). Because
ings of the gender identity standard and the meanings in one’s own role performance
meanings of the spousal role performance, as reported in the diaries are perceived per-
though an indirect path exists through the formances, they represent the situational per-
meanings of the spousal identity standard. ceptions in the model. The difference
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

90 SOCIAL PSYCHOLOGY QUARTERLY

between perceptions and the identity stan- sion; the average correlation between these
dard is the error term representing the identities over the three time periods is .56.
extent to which the meanings of the spousal As the model in Figure 2 shows, each identi-
role performance are not predicted by the ty influences the other with a lag of one time
meanings of the spousal identity. Insofar as period to reflect the time such change
situational disturbances cause the perceived takes.13 According to the hypothesis, if one
meanings of the role performance to deviate identity is more traditionally feminine at one
from the performance predicted by the time period, it should increase the tradition-
meanings in the identity standard, an error ally feminine meanings of the other identity.
or discrepancy exists. The direction and mag- One way to think about all of these
nitude of this discrepancy along the mea- effects is to consider that we are modeling a
sured dimension of meaning have dynamic process in which each part is influ-
consequences, as predicted in Hypotheses 1 encing the others through time.
and 2. Disturbances cause the perceived self-rele-
Thus, for example, in Figure 2, error e5 vant meanings in the situation to become
represents that part of the perceived mean- discrepant from the identity standard. These
ings of the spousal role performance in year disturbances simultaneously begin to bring
1 which is not in line with (predicted by) the about changes in the role performance and,
spouse identity meanings. It may be a perfor- more slowly, in the meanings in the identity
mance that contains more or less traditional- standard. The rates of change in the two out-
ly feminine meanings than expected. comes are quite different. Identities change
According to Hypothesis 1, this discrepancy slowly; behavior in the situation is expected
(being more or less than expected) will neg- to change more rapidly.
atively influence the level of spousal role In addition, effects at one time period
performance in year 2. If the role perfor- theoretically should be identical to the same
mance is more traditionally feminine than effects at another time period (invariance
expected in year 1, Hypothesis 1 suggests over time). Thus the effect of gender identity
that it will become less so in year 2. This is on the spouse identity between years 1 and 2
the normal move toward counteracting the should be the same as the effect of gender
disturbance (or reducing the discrepancy) by identity on spouse identity between years 2
bringing the role performance more into line and 3. Also, the effect of gender identity at
with the performance predicted in the mean- year 1 on itself at year 2 should be the same
ings of the identity standard (Burke 1991; as the effect from year 2 to year 3. These
Burke and Reitzes 1981; Burke and Stets restrictions were added to the model for all
1999). effects. Insofar as this assumption is not true,
At the same time, according to the model will fail to fit the data.
Hypothesis 2, the discrepancy (error e5) will
act over time to change the meanings held in RESULTS
the identity standard itself (in year 2), so as
Before the model is estimated, Table 3
to bring it more into line with the perceived
presents the means, standard deviations, and
spousal role performance. Thus, if the per-
correlations among the variables without
ceived meanings of the spousal role perfor-
comment. I estimated the model using full-
mance are more traditionally feminine than
information maximum-likelihood proce-
expected, they will act over time to increase
dures; the results are presented in Table 4.
the traditionally feminine meanings of the The chi-square test for goodness of fit is ␹2(24)
spouse identity standard (Burke and Cast = 34.7 (p = .07). This indicates a good fit of
1997).
Hypothesis 3 relates to the impact of the
13 Alternative modeling with simultaneous influ-
meanings of one identity standard on the
meanings of another when that they share a ences between the two identities, or with combina-
tions of simultaneous and instantaneous influences,
dimension of meaning. In this case, the leaves the results almost unchanged, but these models
spouse identity and the gender identity share do not fit the overall data. These analyses are avail-
meanings on a masculine/feminine dimen- able on request.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 91
Table 3. Correlations, Means, and Standard Deviations Among the Variables

Variables
Variables .(1) .(2) .(3) .(4) .(5) .(6) .(7) .(8) .(9) Mean SD
Gender Id 1 (1) 1.00 .46 .31
Gender Id 2 (2) .79 1.00 .52 .32
Gender Id 3 (3) .78 .80 1.00 .55 .33
Spouse Id 1 (4) .57 .54 .59 1.00 .00 .69
Spouse Id 2 (5) .55 .53 .57 .87 1.00 .00 .69
Spouse Id 3 (6) .57 .55 .59 .84 .90 1.00 –.01 .72
Housework 1 (7) .30 .25 .30 .39 .38 .38 1.00 –.03 .49
Housework 2 (8) .31 .29 .32 .38 .41 .41 .65 1.00 –.01 .53
Housework 3 (9) .27 .25 .29 .34 .36 .40 .56 .71 1.00 –.01 .56

Table 4. Standardized Structural Equation Coefficients for Model Shown in Figure 3.

Gender Identity Spouse Identity Housework

Year 2 Year 3 Year 2 Year 3 Year 1 Year 2 Year 3
Gender Identity 1 .82* .— .07* .— .— .— .—
Gender Identity 2 .— .82* .— .07* .— .— .—
Gender Discrepancy 2 (e1) .— –.37* .— .— .— .— .—
Spouse Identity 1 .12* .— .83* .— .29* .— .—
Spouse Identity 2 .— .12* .— .92* .— .08* .—
Spouse Identity 3 .— .— .— .— .— .— .08*
Spouse Discrepancy 2 (e3) .— .— .— –.24* .— .— .—
Housework 1 .— .— .— .— .— .82* .—
Housework 2 .— .— .— .— .— .— .82*
Housework 3 .— .— .— .— .— .— .—
Housework Discrepancy 1 (e5) .— .— .06* .— 1.00 –.19* .—
Housework Discrepancy 2 (e6) .— .— .— .06* .— 1.00 –.19*
R2 .59 .71 .77 .83 .— .54 .52
Note: Chi-square goodness of fit: 26.1, df = 23, p = .30
*p < .01

the data to the model, including the assump- reduce the discrepancy. If the individual
tion of equal effects over time.14 engages in more than the amount of (tradi-
With respect to Hypothesis 1 pertaining tionally feminine) housework predicted by
to counteracting the disturbances, Table 4 the meanings of the spousal identity standard
shows that when a discrepancy exists in one year, he or she reduces the amount in
between the meaning of the spousal identity the next year. If the individual engages in less
standard and the meanings of the spousal than the predicted amount, he or she increas-
role performance, that discrepancy (e5 or e6 es the traditionally feminine spousal role
in Figure 2) has a negative effect (beta = –.19, behavior in the subsequent year.
p ≤ .01) on the meanings of the spousal role In regard to identity change, we see that
performance in the following year so as to identities in fact change slowly as the result
of disturbances in the situation. In the pre-
14 Although the model fit the data, a significantly
sent study the persistence of the spousal
better fit can be obtained by allowing the stability identity over the three time periods (two
coefficient for the spousal identity measure to years) is .81 (= .90 ⫻ .90) including direct and
increase over time from .83 to .92. With this change, indirect effects. This means that the weekly
none of the other parameters change, and the overall
persistence would be estimated (assuming
fit of the model is ␹2(23) = 26.1 (p = .30). Allowing such 1

a change might reflect early adjustments in the identi- 100 weeks) as .81100 or .998; when squared this
ty, which begins to stabilize over time. translates to 99.6 percent of the variance in
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

92 SOCIAL PSYCHOLOGY QUARTERLY

common for this identity from one week to sis. We see that the spousal identity is influ-
the next. Over the course of two years (100 enced positively (beta = .07, p ≤ .01) by the
weeks), however, only .812 = 66 percent of the gender identity in the prior year, and that the
variance is in common with the beginning gender identity is influenced positively (beta
point. Very small effects cumulate over time. = .12, p ≤ .01) by the spousal identity in the
In the present research, the gender iden- prior year as well. Because the prior level of
tity changes somewhat more; perhaps mar- each identity is controlled in the model, we
riage has caused larger changes to this are dealing, in effect, with changes over time.
identity. The persistence of this identity over The more traditionally feminine one’s gender
the two years is .672. The estimated weekly identity is in one year, the more one’s spousal
persistence is .996, hardly less stable than the identity becomes traditionally feminine in
spousal identity. Yet the long-term effects the next year, and the more traditionally fem-
show that the cumulative change is larger, inine or masculine one’s spouse identity in
with only 45 percent of the variance in com- one year, the more one’s gender identity
mon between year 1 and year 3. becomes feminine or masculine to match.
Hypothesis 2 concerns the first source of Finally, the model suggested that spousal
identity change and suggests that a discrep- role performance is a function of the spousal
ancy not only leads to behavior counteracting identity (this was the case: beta = .08, p ≤ .01).
a disturbance to meanings in the situation, as It also suggested, however, that spousal role
in Hypothesis 1. It also causes the meanings performance was not a direct effect of gender
in the standard to shift slowly over time in the identity, though an indirect effect was includ-
direction of the behavioral meanings in the ed in the model through the spousal role
situation, thus reducing the discrepancy by identity. Similarly, the implication of a dis-
identity change. As shown in Table 4, the dis- crepancy between the spousal identity stan-
crepancy between perceived spousal perfor- dard and role performance exerts direct
mance meanings and spousal identity effects only on the spousal identity (and on
meanings (e5 or e6 in Figure 2) exerts a posi- future role performance), but through its
tive effect (beta = .06, p ≤ .01) on subsequent effects on the spousal identity, it also exerts
levels of meaning in the spouse identity stan- an indirect effect on gender identity. Because
dard. This finding confirms Hypothesis 2, the overall model fits the data very well, no
concerning the first source of identity change. direct connection between gender identity
Persons who, because of situational con- and the spousal role performance is needed.
straints and disturbances, enact the spouse
role in a more traditionally feminine manner DISCUSSION
than is consistent with their spouse role iden-
tity will find that their spouse role identity The fact that identities act so as to resist
becomes somewhat more traditionally femi- change does not mean that they do not
nine over time. The identity changes over change over time. Identity change involves
time to become more consistent with the role changes in the meaning of the self: changes in
performance. This finding is consistent with what it means to be who one is as a member
Peterson and Gerson (1992), who show that of a group, who one is in a role, or who one is
situational constraints often determine the as a person. These meanings are held in the
amount of household labor in which hus- identity standard, the part of the identity that
bands and wives engage. I find here that serves as a reference for judging self-in-situa-
these constraints are translated into self- tion meanings. Identities’ resistance to
meanings as they persist over time. change gives them some stability; thus
Hypothesis 3 addresses a second mecha- change occurs only slowly in response to per-
nism of identity change, which suggests that sistent pressure.
when activated identities share dimensions of In the present paper I suggest two mech-
meaning, each will influence the other to anisms whereby identities change over time.
keep the levels on the shared dimension of The first is the slow change that occurs as the
meaning the same. The results displayed in meanings in the identity standard shift to be
Table 4 also are consistent with this hypothe- more like the self-relevant meanings that are
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 93
perceived in the situation. This is an adaptive slowly, gaining control where it can, and
response that allows individuals to fit into adapting where it must. This point is consis-
new situations and cultures where the mean- tent with the traditional symbolic interaction
ings are different. It can be viewed as a social- understanding of self as process. At the same
ization effect that might occur as individuals time, however, it acknowledges the structural
take on new roles and memberships. Because symbolic interaction view that identities exist
this process is slow, it is unlikely to result in within the structural framework of society
much change unless the perceptions are per- and are influenced by their position in that
sistently different from the standard. Also, framework. As Coltrane and Ishii-Kuntz
because the output behavior tends to change (1992) point out, men’s participation in
the situation more quickly so as to bring per- household labor is influenced by a number of
ceptions into alignment with the identity structural factors including income, occupa-
standard, such persistence in the discrepancy tion, and urban residence; these tend to shape
is unusual unless the person is in a new situa- ideology and, I would add, self-meanings.
tion. The present results, however, show that
this does happen. IMPLICATIONS
The second mechanism is also an adap-
tive response, in which two identities that A number of implications can be drawn
share some common dimension(s) of mean- from the present research results pertaining
ing in their standards become more like each to identity change. Consider first the effect of
other in their settings on that dimension other identities. In this paper I investigate the
when they are activated together. In the pre- relationship between two identities. I show
sent study, each of the identities (gender that when each identity tries to confirm a dif-
identity and spouse identity) shares a mascu- ferent level on some dimension of meaning
line/feminine dimension. If the standards for (for example, one identity wanting to be
the amount of masculinity/femininity differ more masculine than the other), each identi-
for the two identities, a discrepancy will exist ty standard adjusts slowly toward the level of
for one of the identities when the other the other: in effect, a compromise. If this is to
shows no discrepancy. Whatever the behavior happen, both identities must be activated in
in the situation, persistent discrepancies still the situation, each trying to verify itself from
will be present. Consequently we expect that a common pool of meanings in the situation.
the standards (with respect to For example, the identities of friend and son
masculinity/femininity) for the two identities might be enacted in a situation that involves
will shift so that they are the same. Being both parents and friends. The individual may
more feminine on one standard will bring be embarrassed by the way his parents treat
about more femininity on the other. him in front of his friends, and embarrassed
Conversely, being more masculine on the lat- by the way the friends act toward him in front
ter will bring about more masculinity on the of his parents. This is the traditional notion of
former. This process is also confirmed in the role conflict. One avoids such a conflict by
present data. Which identity changes more not entering situations in which parents and
will depend on other factors: for example, if friends are together (thus activating both
the individual is more strongly committed to identities simultaneously). On the other
the spouse identity or if that identity is more hand, if one encounters this situation fre-
salient, it should change less. quently, the meanings of both identities—
The confirmation of each of these son and friend—shift slowly toward com-
processes within identities begins to move monality, and the conflict again is avoided.
identity theory from the more static view of In this way, every new identity one takes
identities that characterizes much of the prior on, through role acquisition or membership
work to a more dynamic view of identities as in new groups, creates potential changes in
always changing (though slowly) in response other identities that may share dimensions of
to the exigencies of the situation. Insofar as meaning. If I take a job, join a club, or
an identity cannot change the situation (and become friends with a new person, each of
the meanings contained therein), it adapts these identities ultimately must “fit in” with
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

94 SOCIAL PSYCHOLOGY QUARTERLY

my other identities insofar as they are acti- made. As a result of the presence of such ele-
vated simultaneously in situations. As I have ments that have meanings discrepant with
shown, however, it is not merely that one our identities, some self-meanings change in
identity changes to “fit in” with existing iden- a way that reduces the dissonance or discrep-
tities; all identities must adjust to fit together ancy. We change the importance or promi-
so as to bring shared meanings to a common nence of certain elements, and we shift our
level. In this way, then, identities are always self-conceptions on dimensions of meaning
shifting, although generally in small amounts, so as to reduce the dissonance. The changes
because we resist situations that may require in identities resulting from most decisions
large changes. would be quite small; for very important or
To avoid situations that require large large decisions, however, fairly large changes
identity changes, we engage in strategies that can ensue. Nevertheless, as we have seen,
act to confirm our existing identities (Swann small changes occurring from everyday deci-
1990). These strategies include selective sions can cumulate.
interaction, in which we choose the persons A final source of identity change that
and situations with which we engage our- should be recognized is what I have called
selves so as to confirm our current identities disturbances in the situation: events that are
and avoid those which are likely to discon- beyond the control of individuals which
firm them. In addition, we display identity change the self-relevant meanings in the situ-
cues to let others know who we are, and ation. Burke and Cast (1997) discussed this
therefore how we should be treated. By look- the type of event; it is also the type that
ing the part, we convey those meanings which makes the news, such as a house fire, an air-
define us for others as we wish to be defined plane crash, or winning the lottery. Such
and understood. Finally, we use interpersonal events also include experiences such as bully-
prompts or other altercasting procedures ing on the playground or the deterioration of
(Weinstein and Deutschberger 1963) to a friendship, as far as these are relevant to
cause others to treat us in a manner consis- our identities. Finally, they include everyday
tent with our identities. activities, both our own and those of people
The present research has other implica- around us that change meanings in the situa-
tions for identity change as well. As Burke tion in unanticipated ways. Insofar as these
and Cast (1997) have shown, identities meanings have self-relevance and cannot be
change when self-relevant meanings in the fully countered or controlled, they will result
situation alter irrevocably. Such a change in some degree of change in our identities, as
would be manifested in the model depicted in shown in the present research on the normal
Figure 2 as an increase in the discrepancy daily performance of the spousal role.
represented (for example) by e5 or e6. In that
research the birth of a baby resulted in hus- REFERENCES
bands’ becoming more masculine and wives’
becoming more feminine. This change was Burke, Peter J. 1980. “The Self: Measurement
rather dramatic. From the cognitive disso- Implications From a Symbolic Interactionist
Perspective.” Social Psychology Quarterly
nance literature, however, we know that
43:18–29.
every decision leads to postdecision disso- ———. 1989. “Gender Identity, Sex, and School
nance because it creates self-relevant mean- Performance.” Social Psychology Quarterly
ings that are at odds with the identity we 52:159–69.
possess. This is the case because there are ———. 1991. “Identity Processes and Social
always some undesirable elements in the Stress.” American Sociological Review
alternative that was chosen, and some ele- 56:836–49.
ments in the alternative that was rejected ———. 1997. “An Identity Model for Network
Exchange.” American Sociological Review
which are nevertheless desirable.
62:134–50.
Although a situation may contain only a ———. 2003. “Relationships Among Multiple
few such discrepant elements (otherwise we Identities.” Pp. 195–214 in Advances in
probably would have chosen the other alter- Identity Theory and Research, edited by
native), they persist after the decision is Peter J. Burke, Timothy J. Owens, Richard T.
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

IDENTITY CHANGE 95
Serpe, and Peggy A. Thoits. New York: Relationships: The Self as Architect and
Kluwer Academic/Plenum. Consequence of Social Reality.” Journal of
Burke, Peter J. and Alicia D. Cast. 1997. “Stability Personality and Social Psychology
and Change in the Gender Identities of 67:1012–23.
Newly Married Couples.” Social Psychology Oakes, Penelope. 1987. “The Salience of Social
Quarterly 60:277–90. Categories.” Pp. 117–41 in Rediscovering the
Burke, Peter J. and Michael M. Harrod. 2005. “Too Social Group, edited by John C. Turner,
Much of a Good Thing?” Social Psychology Michael A. Hogg, Pamela J. Oakes, Stephen
Quarterly 68:359–74. D. Reicher, and Margaret S. Wetherell. New
Burke, Peter J. and Donald C. Reitzes. 1981. “The York: Blackwell.
Link Between Identity and Role Osgood, Charles E., George J. Suci, and Percy H.
Performance.” Social Psychology Quarterly Tannenbaum. 1957. The Measurement of
44:83–92. Meaning. Urbana: University of Illinois
———. 1991. “An Identity Theory Approach to Press.
Commitment.” Social Psychology Quarterly Peterson, Richard R. and Kathleen Gerson. 1992.
54:239–51. “Determinants of Responsibility for Child
Burke, Peter J. and Jan E. Stets. 1999. “Trust and Care Arrangements Among Dual-Earner
Commitment Through Self-Verification.” Couples.” Journal of Marriage and the Family
Social Psychology Quarterly 62:347–66. 54:527–36.
Burke, Peter J. and Judy C. Tully. 1977. “The Powers, William T. 1973. Behavior: The Control of
Measurement of Role Identity.” Social Perception. Chicago: Aldine.
Forces 55:881–97. Schein, Edgar H. 1958. “The Chinese
Callero, Peter L. 1985. “Role-Identity Salience.” Indoctrination Program for Prisoners of
Social Psychology Quarterly 48:203–14. War: “A Study of Attempted
Cast, Alicia D. and Peter J. Burke. 2002. “A Theory ‘Brainwashing’.” Pp. 311–34 in Readings in
of Self-Esteem.” Social Forces 80:1041–68. Social Psychology, edited by Eleanor E.
Coltrane, Scott. 2000. “Research on Household Maccoby, Theodore M. Newcomb, and
Labor: Modeling and Measuring the Social Eugene L. Hartley. New York: Henry Holt
Embeddedness of Routine Family Work.” and Co.
Journal of Marriage and the Family Serpe, Richard T. 1987. “Stability and Change in
62:1208–33. Self: A Structural Symbolic Interactionist
Coltrane, Scott and Masako Ishii-Kuntz. 1992. Explanation.” Social Psychology Quarterly
“Men’s Housework: A Life Course 50:44–55.
Perspective.” Journal of Marriage and the Smith-Lovin, Lynn. 2003. “Self, Identity, and
Family 54:43–57. Interaction in an Ecology of Identities.” Pp.
Deaux, Kay. 1992. “Personalizing Identity and 167–78 in Advances in Identity Theory and
Socializing Self.” Pp. 9–33 in Social Research, edited by Peter J. Burke, Timothy
Psychology of Identity and the Self-Concept, J. Owens, Richard T. Serpe, and Peggy A.
edited by Glynis M. Blackwell. London: Thoits. New York: Kluwer / Plenum.
Surrey University Press. Spence, Janet T. and Robert L. Helmreich. 1978.
———. 1993. “Reconstructing Social Identity.” Masculinity and Femininity: Their
Personality and Social Psychology Bulletin Psychological Dimensions, Correlates and
19:4–12. Antecedents. Austin: University of Texas
Gecas, Viktor and Jeylan T. Mortimer. 1987. Press.
“Stability and Change in the Self-Concept Stets, Jan E. 1995. “Role Identities and Person
From Adolescence to Adulthood.” Pp. Identities: Gender Identity, Mastery Identity,
265–86 in Self and Identity: Perspectives and Controlling One’s Partner.” Sociological
Across the Lifespan, edited by Terry Honess Perspectives 38:129–50.
and Krysia Yardley. Boston, MA: Routledge Stryker, Sheldon and Richard T. Serpe. 1982.
and Kegan Paul. “Commitment, Identity Salience, and Role
Kiecolt, K. Jill. 1994. “Stress and the Decision to Behavior: A Theory and Research
Change Oneself: A Theoretical Model.” Example.” Pp. 199–218 in Personality, Roles,
Social Psychology Quarterly 57:49–63. and Social Behavior, edited by William Ickes
———. 2000. “Self Change in Social and Eric S. Knowles. New York: Springer.
Movements.” Pp. 110–31 in Identity, Self, and ———. 1994. “Identity Salience and
Social Movements, edited by Sheldon Psychological Centrality: Equivalent,
Stryker, Timothy Owens, and Robert White. Overlapping, or Complementary
Minneapolis: University of Minnesota Press. Concepts?” Social Psychology Quarterly
McNulty, Shawn E. and William B. Swann. 1994. 57:16–35.
“Identity Negotiation in Roommate Swann, William B., Jr. 1990. “To Be Adored or to
 #2737—Social Psychology Quarterly—VOL. 69 NO. 1—69106-burke

96 SOCIAL PSYCHOLOGY QUARTERLY

Be Known?: The Interplay of Self- Examining Gender and Marital Status
Enhancement and Self-Verification.” Pp. Differences in Distress.” American
408–50 in Handbook of Motivation and Sociological Review 51:259–72.
Cognition, edited by E. Tory Higgins and Tsushima, Teresa and Peter J. Burke. 1999. “Levels,
Richard M. Sorrentino. New York: Guilford. Agency, and Control in the Parent Identity.”
Tallman, Irving, Peter J. Burke, and Viktor Gecas. Social Psychology Quarterly 62:173–89.
1998. “Socialization into Marital Roles: Vital Statistics of the United States. 1987. Marriage
Testing a Contextual, Developmental Model and Divorce. Vol. 3. Hyattsville, MD:
of Marital Functioning.” Pp. 312–42 in The National Center for Health Statistics.
Developmental Course of Marital Weinstein, Eugene A. and Paul Deutschberger.
Dysfunction, edited by Thomas N. Bradbury. 1963. “Some Dimensions of Altercasting.”
Cambridge, UK: Cambridge University Sociometry 26:454–66.
Press. World Almanac and Book of Facts. 1992. Microsoft
Thoits, Peggy A. 1986. “Multiple Identities: Bookshelf. Redmond, WA: Microsoft.

Peter J. Burke is professor of sociology at the University of California, Riverside, AAAS Fellow,
and the 2003 recipient of the ASA Section on Social Psychology’s Cooley-Mead Award for life-
time contributions to social psychology. Recent publications include “Extending Identity
Theory: Insights from Classifier Systems,” 2004, in Sociological Theory, “New Directions in
Identity Control Theory,” with Jan E. Stets, 2005, in Advances in Group Processes, and “Too
Much of a Good Thing,” 2005, with Michael Harrod, Social Psychology Quarterly.