Abstract
This research extends prior work by integrating self-esteem into the identity theory model. I focus on understanding the processes by which self-esteem acts as a moderator of feedback from others about the self. Differences in the moderating effect of self-esteem due to the relationship of the individual to the person who is the source of feedback about the self are also considered. The criminal identity and the worker identity are studied using data collected from an incarcerated population. Results confirm that the relationship between feedback from others about the self and self-views is moderated by self-esteem. Analyses reveal that participants with high self-esteem report self-views that correspond more with feedback from others than participants with low self-esteem. The moderating properties of self-esteem operate differently for different sources of feedback about the self and for different identities, and this finding has implications for identity theory.
The symbolic interactionist perspective holds that self-views form reflexively based on feedback from others during social interaction, but this position neglects attention to the role of the individual in forming self-views. The concept of the “reflexive self” also lacks sufficient explanation for differences among individuals who share the same kinds of identities (mother, father, student, teacher, etc.), although we know such differences do exist (and can be verified empirically). Sheldon Stryker (1968:559) suggests that research centered on the reflexive self “has suffered from a failure to go beyond Mead’s formulation of the self as that which is an object to itself” and calls for extension of theory and research investigating the role of individual agency in developing the reflexive self-view.
By conceptualizing the self as strictly defined by early symbolic interactionists, researchers tend to neglect extending early works to allow this perspective to adequately incorporate the role of individual agency for self-views (Gecas and Schwalbe 1983). Since individuals do not have the same self-views even for the same kinds of identities, it is important to extend our understanding about possible sources of variation in self-views among individuals within a symbolic interactionist framework in order to strengthen the applicability of the theory.
Previous work examines the buffering role of self-esteem in protecting self-views against inconsistent feedback from others about the self (Baumeister 1998; Campbell 1990; Cast and Burke 2002). These studies indicate that one’s level of self-esteem influences perceptions of feedback from others about the self; thus, self-esteem may be a source of individual differences in self-views. The current study considers the broader implication of self-esteem for self-views as well as the process by which the two relate. I also consider the possibility that self-esteem may not have the same effects on the relationship between feedback from others and self-views in all situations.
Emphasis on the buffering quality of self-esteem for self-views in previous work also neglects the fact that self-esteem may act as a buffer in some situations but not in others. This study is the first to postulate self-esteem as a moderator (rather than a buffer) of self-views by specifically placing it into the perceptions component of the identity model. I suggest that self-esteem is a moderator in that it is a relevant part of the process by which feedback from others is interpreted by the self on a regular basis. Thus, self-esteem plays a consistent role in forming self-views rather than only playing a role as a buffer when inconsistent feedback is detected. This work is also the first to address the main research objective of uncovering the process through which self-esteem asserts a relationship with self-views. The current study extends research on the relationship between self-esteem and self-views in three important ways. First, drawing upon identity control theory and research on self-esteem, self-verification, and self-enhancement, I specify the role of self-esteem in forming self-views. Next, I examine the process through which self-esteem moderates feedback from others about the self by testing two competing approaches to explaining how self-esteem asserts influence on perceptions of feedback from others about the self. Lastly, I examine variations in the intensity of the moderation dependent upon different interaction partners. To test my predictions, I use survey data collected from incarcerated criminal offenders, focusing on their criminal and worker identities. The unique context of incarcerated individuals will be discussed in detail in the theory section.
Principles of Self-Verification and Self-Enhancement
Theories of self suggest identity processes are motivated by self-verification or by self-enhancement. In self-verification, the goal is to bring perceptions about the self when performing a given identity (self-in-situation) in line with the meanings or expectations for the identity held in the identity standard (self-in-meaning) (Stets and Burke 2005; Swann, Rentfrow, and Guinn 2003). For self-enhancement, the goal is for perceptions about the self-in-situation to exceed the identity standard (Sedikides 1993; Sedikides, Gaertner, and Toguchi 2003; Swann 1990; Swann et al. 1987). The meanings in the identity standard are regarded as the cognitive component of the self (Burke 1980; Ervin and Stryker 2001). Identity standards consist of a combination of shared cultural definitions of a particular identity (e.g., daughter, father, professor) and definitions derived from personal experiences. The identity standard thus serves as a behavioral guide, or roadmap, for how an individual claiming a particular identity should act when engaging that identity.
For example, when people claim the student identity they engage in certain behaviors associated with being a student. These may include studying, doing homework, and attending classes because these behaviors are prescribed by the identity standard for the student identity. In self-verification, people claiming the student identity should be satisfied if perceptions of the self-in-situation for the student identity suggest they are performing these behaviors in a manner consistent with the identity standard for the student identity. In self-enhancement, only perceptions of the self-in-situation that exceed the identity standard bring the individual satisfaction.
For either perspective, behavior gives off meanings during interaction that are directed by the identity standard for the particular identity activated. During interaction, feedback about how others see the individual performing an identity is communicated through reflected appraisals. Reflected appraisals represent the way individuals think others see them with respect to performance of a particular identity. Reflected appraisals influence self-views, which represent one’s perceptions about the self-in-situation for the identity at issue (Stets and Burke 2005). Perceptions are assessed for congruence with the identity standard through a mechanism called the comparator.
If the comparator detects a match between perceptions of the self-in-situation and the identity standard, identity verification occurs. If the comparator detects a discrepancy between perceptions of the self-in-situation and the identity standard, identity nonverification occurs. For the self-verification approach, identity nonverification (either in the positive or the negative direction) has negative consequences and requires an adjustment in order to align these two components more closely (Burke 1991). For the self-enhancement approach, only nonverification in the negative direction has negative consequences and requires an adjustment to achieve perceptions about the self-in-situation that exceed expectations of the self-in-meaning (Swann 1990).
Potential adjustment strategies for identity nonverification include making behavioral adjustments in an attempt to alter feedback from others, selective interaction only with others who either verify or enhance one’s identity (or selectively processing feedback as verifying or enhancing), or changing the identity standard so that it matches feedback from others (Burke 2006; Kiecolt 1994). In cases where an identity is repeatedly nonverified (despite attempts at resolving discrepancy), an individual may choose to abandon the identity and/or the identity may become less relevant than other identities for the individual (Burke and Stets 2009). The outcome of any of these strategies is a change in one’s perception about the self-in-situation.
This model of identity processes suggests that individuals have different self-views reflecting different perceptions of feedback from others. However, the model does not specify a mechanism for explaining how people develop different perceptions of feedback from others. I suggest that self-esteem is one such mechanism. Conceptualizing self-esteem as part of the perceptions component of the identity model can account for individualized interpretations of the same kind of feedback from others for two different people.
Self-Esteem and Self-Views
Self-esteem as an independent component of self has been the subject of many studies (Baumeister 1993; Rosenberg 1979; Tesser 1988). Previous studies focus primarily on causes of variations in the amount of self-esteem among individuals (Harter 1993), the pursuit of self-esteem as a motive for behavior (Swann 1990), or self-esteem as a buffer from negative information about the self (Baumeister 1998; Burke 1991; Campbell 1990; Cast and Burke 2002). To date, there has been little research on self-esteem as an integral part of the process of identity (for exceptions, see Cast and Burke 2002; Ervin and Stryker 2001) despite the fact that it is an ever-present component of self.
Self-esteem has been identified as “the evaluative component of the self-concept” (Owens and Serpe 2003). It is generally considered as comprised of self-worth (degree to which an individual feels to be a person of value) and self-efficacy (degree to which an individual feels to be a capable or efficacious person) dimensions (Cast and Burke 2002; Gecas and Schwalbe 1983). Within an identity theory framework, self-esteem levels are affected by identity verification. If an identity is verified, increased feelings of worth and efficacy are generated because people feel they are of value and are capable when self-in-situation matches the identity standard. However, if a particular identity is not verified, decreased feelings of worth and efficacy are generated instead (Cast and Burke 2002). This suggests a direct relationship between self-esteem and self-views to the extent that identity verification is a source of self-esteem. 1 There are two distinct approaches to explain the relevance of self-esteem for self-views.
First, Baumeister (1998) suggests individuals with high self-esteem are constantly seeking to support their egotistical self-views. 2 As a result, they are more likely to interpret reflected appraisals as consistent with self-views even if they are not. Individuals with greater self-esteem use different techniques to support positive self-views. For example, they may make internal attributions for success and external attributions for failure, point out defects in information used by others to paint a negative picture of self, interact only with others who provide enhancing feedback, or selectively process only feedback about the self that is favorable (Baumeister 1998). These techniques are similar to the selective interaction and selective processing of feedback from others described as a strategy for dealing with identity nonverification.
Based on Baumeister’s (1998) description of the selective processing tendencies of people with high self-esteem, I expect that reflected appraisals will be more consistent with self-views for high self-esteem individuals than for low self-esteem individuals. The following hypothesis is used to test Baumeister’s selective processing approach:
Hypothesis 1: Reflected appraisals are more consistent with self-views for high self-esteem individuals than for low self-esteem individuals.
Second, Campbell (1990) offers a different approach for explaining the relevance of self-esteem for self-views. She suggests that the difference between high self-esteem and low self-esteem individuals is in certainty and clarity of self-views. She argues we should not look at high self-esteem people as unconditionally positive and low self-esteem people as unconditionally negative. Instead, we should focus on effects of self-esteem on stability of self-views. Campbell suggests individuals with more self-esteem have more clear and consistent self-views, which are less susceptible to influence from views of others, than those of individuals with lower self-esteem. Campbell posits that low self-esteem people possess plasticity of self-views (Campbell 1990). Plasticity refers to the idea that self-views of those with low self-esteem are more bendable (like plastic), so that they are more easily influenced by external information (e.g., reflected appraisals) than self-views of high self-esteem individuals (Brockner 1984; Tesser 2001).
In contrast to Baumeister’s ideas, Campbell’s work suggests that the techniques employed to process information about the self from others are similar to the strategy of changing one’s identity standard to match the views of others about the self-in-situation described with respect to identity nonverification. Specifically, Campbell’s work suggests high self-esteem individuals are more likely to hold self-views independent of situational conditions (e.g., others’ views), while low self-esteem individuals are more “plastic” and likely to bend or tailor self-views to match situational conditions. The following hypothesis tests this relationship in order to compare Campbell’s stability of self approach with Baumeister’s selective processing approach:
Hypothesis 2: Reflected appraisals are less consistent with self-views for high self-esteem individuals than for low self-esteem individuals.
The buffering role of self-esteem for self-views has been previously examined from an identity theory perspective. Cast and Burke (2002) use data from newly married couples to assess the role of self-esteem as a buffer for the self when appraisals from the spouse are inconsistent with self-views. Self-esteem is conceptualized as a resource that can be summoned to protect the self from conflicting appraisals. The results suggest that self-esteem does moderate discrepant appraisals by diluting negative emotional outcomes associated with identity nonverification. The current study extends this research by examining the process by which self-esteem acts as a moderator (as opposed to strictly a buffer) for self-views.
Reflected Appraisals
Previous empirical work suggests that the context of interaction is relevant to understand the role of reflected appraisals in forming perceptions about the self-in-situation. This includes one’s relationship to others providing feedback for reflected appraisals and differences in the way perceptions of self-in-situation are formed for different identities. For example, Burke (1991) posits that appraisals from others who are significant or familiar to participants have a stronger relationship with self-views than appraisals from strangers. He suggests that relationships with significant and/or familiar others require a greater investment and involve an established pattern of interaction that supports identity verification for both parties. Feedback from significant and familiar others is more meaningful for influencing perceptions of self-in-situation than feedback from strangers because individuals come to rely on this consistent behavior pattern to verify a given identity.
Status and legitimate authority also matter in determining the relationship between reflected appraisals and self-views. Cast, Stets, and Burke (1999) studied the effects of appraisals from others on self-views in the context of newly married couples’ relationships. They found that views of both the self and the spouse held by the spouse with lower status conform to the appraisals of the spouse with higher status. In addition, Stets and Asencio (2008) examine data from a laboratory study to assess the influence of legitimate status on the relationship between appraisals and self-views. In that study, respondents report more intense emotional reactions to feedback considered inconsistent with their own self-views when it came from someone who they thought was a supervisor, rather than when it came from someone who they did not think was a supervisor.
Similarly, Asencio (2011) tests for differences in the influence of reflected appraisals from significant others, familiar others, and legitimate others on self-views for the criminal identity using a sample of incarcerated offenders. Findings indicate that reflected appraisals from significant others and familiar others have a significant influence on criminal self-views while reflected appraisals from legitimate others have no influence on criminal self-views in the context of incarceration. 3
The findings discussed previously indicate the relevance of the source of feedback for examining the relationship between reflected appraisals and self-views. The current study tests this by comparing the implications of self-esteem and reflected appraisals from three different sources for self-views. 4 Self-esteem may moderate reflected appraisals differently depending upon who the source of feedback is in relation to the individual participating in the study.
Sources of Reflected Appraisals in the Current Study
The context of incarceration is unique for identity processes in that it limits the number of identities an individual can play out, the number of interaction partners an individual may have, and the type of interaction within which an individual may participate. These restrictions make the identities and interaction partners that are available highly salient for incarcerated offenders (Asencio 2011). The participants in the current study are allowed to have visitation with significant others only on an infrequent basis (once a month). Despite infrequent contact, interaction partners who are categorized by participants as significant others represent primary agents of socialization and are likely to be a major source of reflected appraisals about the self.
The inmate participants of this study are much more likely to interact on a daily basis with other incarcerated inmates. The inmates are housed together, which means they have 24-hour contact with at least one cellmate in addition to frequent daily contact with other inmates in the same unit. They perform certain activities such as housekeeping chores, meals, and attending classes offered at the facility together. During the period of incarceration, other inmates are the peers of the offenders included in the study. These peers are the people with whom the most frequent contact occurs. The frequency of interaction suggests that reflected appraisals from this group are highly relevant for perceptions about the self-in-situation. This may be particularly true for the criminal identity, which is the basis of interaction between inmates.
Correctional officers (guards) represent a third group of individuals who are highly relevant as a source of reflected appraisals for this study. The relationship between offenders and guards is unique in several ways. First, the guards are responsible for controlling possible behaviors and interaction partners of offenders, giving them a semblance of legitimate status. This legitimate status typically comes as authorization from superiors rather than endorsement from offenders. Authorization is typically one-sided in that guards believe they are in control but inmates do not respect guards or feel guards have the right to control them. As a result, authorization is likely to have less influence than endorsement in this context (Zelditch 2006).
While guards as a class come into contact with incarcerated participants on a daily basis, the actual correctional personnel acting as guards are rotated daily. This is a relatively standard procedure at many correctional facilities designed to avoid personal relationships (and potential corruption) to develop between inmates and guards. This rotation makes the nature of reflected appraisals from guards (legitimate others) substantively different from that of significant others and frequent others (peers) in that guards are a generalized group of others while significant others and peers consist of specific individuals with whom participants interact regularly.
Asencio and Burke (2011) use longitudinal data from incarcerated criminal offenders to examine reciprocity between reflected appraisals and perceptions of self-in-situation for three separate identities (criminal, drug user, and worker). Their findings support the idea that reflected appraisals operate differently for different types of identities. Reflected appraisals and self-views play roughly equal roles for perception of self-in-situation for the criminal identity. However, the perception of self-in-situation for the drug user identity is almost entirely constructed of reflected appraisals. Interestingly, the perception of self-in-situation for the worker identity is almost entirely constructed of self-views.
Asencio and Burke (2011) suggest that the criminal identity has a tangible component in that one has been formally tried and convicted prior to incarceration, which may impact the perceptions of self-in-situation independent of reflected appraisals. However, interpretation by the criminal justice system as part of the adjudication process allows for others’ views to play a potentially equal role in forming perceptions about self-in-situation for the criminal identity. On the other hand, the worker identity can be composed almost entirely of tangible output in the form of task completion (which is not subject to interpretation). This tangibility may make individuals less reliant on reflected appraisals for perceptions of the self-in-situation with respect to the worker identity than they are for the criminal identity (or any other identity with a lesser amount of tangible output). The current study compares the moderating influence of self-esteem for reflected appraisals on both the criminal identity and the worker identity.
Following previous findings that reflected appraisals from peers and significant others are more relevant for self-views than reflected appraisals from guards for incarcerated subjects, I anticipate that self-esteem will moderate these effects in a similar pattern. Consistent with identity theory, I do not expect any differences in the moderation process for the criminal and worker identities. Theoretically, the process should be the same though the experience of the identities may differ. The following hypothesis is tested to assess the role of self-esteem in determining variation in the relationship between reflected appraisals from different sources and self-views for the current study:
Hypothesis 3: Self-esteem will moderate the relationship of reflected appraisals from significant others and peers with self-views to a greater extent than it will moderate the relationship of reflected appraisals of guards with self-views.
Methods
Sample
Interviewed in 2005, the participants in this study were 171 male and female incarcerated felony offenders housed at a medium security correctional facility. The sample was obtained specifically for the purpose of examining the criminal identity in the incarceration context. Data were collected during a series of site visits over a six-month period to the correctional facility using a survey. Participants included in the study were attending general educational classes while incarcerated. The investigator was permitted to administer the survey during regularly scheduled classroom hours so as not to disrupt daily routines of participants or present security issues. The participants were asked to voluntarily complete a survey about how they see themselves. Of all the participants asked to participate, only one refused.
The sample consisted of 97 males and 74 females (N = 171) with a mean age of 29.7, made up of primarily whites (84) and Hispanics (60), with a small portion of African Americans (15). The racial/ethnic breakdown reflects the population in the Southern California region where the incarceration facility is located. A majority of the participants were married (138). In general, these characteristics are in line with those reported by the Bureau of Justice Statistics for the jail population at the time (2006) in the United States.
Measures
Criminal self-view
The measure of criminal self-view is created from an item asking respondents to report “How you see yourself right now” on a semantic differential scale anchored with “law-abiding” (1) and “unlawful” (5). This single item method of measuring self-views has been established in prior studies (Asencio 2011; Asencio and Burke 2011; Stets and Harrod 2004). The anchors for this item (law-abiding and unlawful) were selected using a combination of theoretical relevance and information gained from informal discussions with inmates similar in characteristics to the sample.
Worker self-view
The measure of worker self-view is created in the same manner as that of criminal self-view. Respondents were asked to report “How you see yourself right now” on a semantic differential scale anchored by “non-worker” (1) and “worker” (5). Scores are coded such that a higher score indicates the subject has a higher worker self-view.
Reflected appraisals
To measure reflected appraisals, respondents were presented with the same indicators used in the self-view measure and asked to report for each characteristic in three separate sections how they think significant others see them, other inmates see them, and guards at the jail see them. This method of measuring reflected appraisals has been established in prior studies (Asencio 2011; Asencio and Burke 2011).
Self-esteem
Self-esteem was assessed through the Rosenberg self-esteem scale (Rosenberg 1979). The 10-item scale was included as part of the survey participants were asked to complete about how they see themselves (α = 85). The Rosenberg self-esteem scale consists of a 10-item Likert scale in which respondents rate themselves on a four-point scale ranging from “strongly agree” to “strongly disagree.” Sample items from the scale include “On the whole, I am satisfied with myself,” “At times, I think I am no good at all,” and “I take a positive attitude toward myself.” Negatively worded items from the Rosenberg scale are reverse coded such that a higher score indicates positive self-esteem. The scores for all 10 items on the Rosenberg scale are added together to create a single self-esteem indicator with scores ranging from 10 to 40. Though recent social psychological literature defines self-esteem as a combination of self-worth and self-efficacy, the current study only employs a measure of self-worth, which is consistent with Baumeister and Campbell’s definition of self-esteem as an overall evaluation of the self (Baumeister et al. 2003).
Analyses
Ordered logistic regression models are used to assess the moderating effect of self-esteem on the relationship between reflected appraisals and self-view. Ordered logistic regression is an appropriate technique for these data as the distance between each response category of the dependent variable is not known (Long 1997; Long and Freese 2006; McKelvey and Zavoina 1975). The reflected appraisals and self-esteem variables are standardized with a mean of zero to allow for an equal comparison among relationships. To assess moderation effects of self-esteem on reflected appraisals, interaction variables are created and then standardized as well. The interaction variables are included in the ordered logistic models presented in the next section.
Results
Table 1 presents the means and standard deviations of variables in the study. The table reveals a mean criminal self-view of 2.28. The means for reflected appraisals from significant others is 2.71, reflected appraisals from peers 2.54, and reflected appraisals from guards is 3.36. The mean worker self-view is 1.62. The means for reflected appraisals from significant others is 1.93, reflected appraisals from peers is 2.10, and reflected appraisals from guards is 2.69. T-tests confirm that the means for self-views are significantly lower than the means for reflected appraisals from all sources for both identities.
Means and Standard Deviations for Variables (N = 171)
Note: RA = reflected appraisals.
Table 2 presents the distribution of scores for the criminal and worker self-views along with the distribution of scores for the various reflected appraisals. The results show that respondents are more likely to view themselves at the midpoint or lower for the criminal identity. Respondents are also more likely to report that significant others and peers rate them lower for the criminal identity than guards at the jail would. The distribution is even more negatively skewed for the worker self-view and reflected appraisals. For the worker self-view, respondents report that reflected appraisals from all three groups of others (significant others, peers, and guards) are more likely to be low for the worker identity.
Distribution of Key Variables (N = 171)
Note: RA = reflected appraisals.
Table 3 presents correlations of the variables in the study. The correlations show the variables of interest (self-view, reflected appraisals, and self-esteem) are significantly correlated with one another. However, the demographic variables such as age, gender, and race are not significantly correlated with any of the substantive variables of interest. Self-esteem is negatively correlated with the self-view for both identities under study.
Correlations among Variables
Note: RA = reflected appraisals.
p < .05.
Table 4 shows the ordered logistic model for the effects of reflected appraisals from significant others and self-esteem on the criminal self-view, as well as the percentage change in predicted odds (OR) for the criminal self-view due to a standard deviation increase in each predictor. 5 The results establish strong and significant direct relationships between reflected appraisals from significant others (OR = 1.99, 99.4 percent change in odds), reflected appraisals from peers (OR = 2.13, 109.8 percent change in odds), self-esteem (OR = .58, –42.1 percent change in odds) and the criminal self-view. The data also show that self-esteem does act as a moderator of the relationship between reflected appraisals and self-views. As self-esteem increases, the relationship between reflected appraisals of significant others and the criminal self-view becomes significantly stronger (OR = 1.45, 43.2 percent change in odds), as does the relationship between reflected appraisals of peers and criminal self-view (OR = 1.30, 28.2 percent change in odds). This means that for every standard deviation increase in self-esteem, the impact of reflected appraisals from significant others on the predicted odds of self-views increases by 1.30. Similarly, for every standard deviation in self-esteem, the impact of reflected appraisals from peers on the predicted odds of self-view increases by 1.30.
Ordered Logit Model for Criminal Self-View by Reflected Appraisals and Self-Esteem (N = 171)
Note: SE = standard errors.
p < .05.
Hypothesis 1 tests Baumeister’s selective processing approach, which suggests self-esteem moderates the relationship between reflected appraisals from significant others and self-views for the criminal identity such that they will be more congruent; this hypothesis is supported. Respondents with high self-esteem are more likely to hold self-views that are similar to others’ views of the self. Support for Campbell’s stability of self hypothesis would require that respondents with high self-esteem demonstrate self-views that were less congruent with the views of others as this would reflect the independent nature of self-views for high self-esteem people suggested by Campbell’s work.
Figures 1 through 3 provide a visual representation of the moderation effect of self-esteem for the relationship between reflected appraisals and self-view. Categories of low, middle, and high (coded 1, 2, and 3, respectively) were created for the self-esteem and reflected appraisal variables. For self-esteem, the middle category includes those scoring within one standard deviation of the mean for self-esteem (24.69–36.13), the low category includes those with scores below 24.69, and the high category includes those with scores above 26.13. For reflected appraisals, middle category for reflected appraisals includes respondents scoring within one half of a standard deviation to allow for a more equal distribution between groups since there is a smaller range of possible scores for reflected appraisals than for self-esteem (significant others = 2.06–3.36; peers = 1.89–3.19; guards 2.59–4.13). The low category for reflected appraisals from each group consists of respondents scoring in the range below the lower limit of the middle category, and the high category for each group consists of respondents scoring above the upper limit of the middle category for reflected appraisals. The graphs were created using the line graph functionality in SPSS version 19.0.
Reflected Appraisals from Significant Others by Self-Esteem
Reflected Appraisals from Peers by Self-Esteem
Reflected Appraisals from Guards by Self-Esteem
Figure 1 shows that the higher the self-esteem level, the more sharply the criminal self-view increases when reflected appraisals from significant others for the criminal identity also increase. A similar pattern is illustrated in Figure 2 for reflected appraisals from peers. The pattern for the moderation of self-esteem on the influence of reflected appraisals from guards differs from that of significant others and peers in that self-esteem interacts with reflected appraisals primarily for those only in the highest category of self-esteem. This reflects the nonsignificant results of the ordered logit analysis for the moderation of self-esteem on the relationship between reflected appraisals and guards for the criminal self-view.
The ordered logistic model for the worker self-view by reflected appraisals and self-esteem is presented in Table 5 along with the percentage change in odds per standard deviation increase in the predictors. The effects here demonstrate a similar pattern as the results for the criminal self-view. However, there is a slightly different pattern depending upon the source of reflected appraisals. The worker identity differs from the criminal identity in that reflected appraisals from all three sources have strong and significant relationships with the self-view (significant others OR = 2.25, 125.6 percent change in odds; peers OR = 1.71, 70.4 percent change in odds; guards OR = 1.65, 64.4 percent change in odds). The direct effect of self-esteem is also significant for the worker identity (β = .48, –51.3 percent change in odds).
Ordered Logit Model for Worker Self-View by Reflected Appraisals and Self-Esteem (N = 171)
Note: SE = standard errors.
p < .05.
For participants with higher self-esteem, the results demonstrate the relationship between reflected appraisals of peers (OR = 1.30, 27.5 percent change in odds) and reflected appraisals of guards (OR = 1.37, 36.3 percent change in odds) with worker self-views is significantly stronger than it is for those with lower self-esteem. This means that for every standard deviation increase in self-esteem, the impact of reflected appraisals from peers on the predicted odds for self-views increases by 1.30. Similarly, for every standard deviation increase in self-esteem, the impact of reflected appraisals from guards on the predicted odds for self-view increases by 1.30. The results support Baumeister’s selective processing approach reflected in Hypothesis 1. Table 4 also provides support for Hypothesis 3, in that the moderation effect of self-esteem is more intense for reflected appraisals of peers than for reflected appraisals of guards in the context of worker self-view.
Figures 4 through 6 contain the graphs of the interactions presented in Table 5 for clarification. Self-esteem is represented by the low, middle, and high categories with the same cut points as in the graphs for the criminal identity. The same technique of using a half of a standard deviation around the mean to create the categories for reflected appraisals for the criminal identity was used to create the categories for reflected appraisals for the worker identity. The range for the middle category for reflected appraisals from significant others for the worker identity is 1.30 to 2.56, reflected appraisals from peers 1.45 to 2.75, and reflected appraisals from guards 1.98 to 2.84. The low category for reflected appraisals from each group consists of respondents scoring in the range below the lower limit of the middle category, and the high category for each group consists of respondents scoring above the upper limit of the middle category for reflected appraisals.
Reflected Appraisals from Significant Others by Self-Esteem
Reflected Appraisals from Peers by Self-Esteem
Reflected Appraisals from Guards by Self-Esteem
Figure 4 shows that lower levels of self-esteem are associated with a closer relationship between reflected appraisals from significant others and the worker self-view than higher levels of self-esteem. This reflects the nonsignificant interaction term in the ordered logit analyses. Figures 5 and 6 show a different pattern for the interaction effect from that of Figure 4. Figure 5 shows that as self-esteem increases, the reflected appraisals from peers more closely match the worker self-view. Figure 6 confirms a similar pattern for reflected appraisals from guards.
Discussion
As suggested by Baumeister’s selective processing theory, these results show a tendency for participants with high self-esteem to pay more attention to reflected appraisals that confirm self-views. This means that people with high self-esteem are more likely to accomplish identity verification (and resolve discrepancies) by employing strategies such as interpreting feedback from others as verifying even if it is not, ignoring reflected appraisals that are nonverifying, and/or interacting only with others who provide verifying feedback. The results suggest that self-esteem is a mechanism that influences one’s perceptions of feedback from others about the self.
The findings with respect to different identities suggest reflected appraisals from different groups of others may be more or less relevant for different categories of identities. It is possible the less tangible output of the criminal identity means participants rely more on reflected appraisals from those who “know” them better (significant others and peers) than on reflected appraisals from guards who are virtually strangers. It is also possible the affective component of self (self-esteem) is not activated when evaluating reflected appraisals from guards, with whom respondents have little investment in establishing and maintaining consistent interaction patterns (as opposed to significant others and peers).
The worker identity has a more tangible output than the criminal identity (e.g., task completion). In the context of incarceration, output consists of completing jobs assigned to inmates in the barracks (e.g., cleaning, cooking, laundry, etc.). These are tasks that significant others are not witness to as they are not in the correctional facility. This helps explain why self-esteem may be activated with respect to reflected appraisals of peers and guards but not significant others for the worker identity. More work is needed in order to determine the reason for the difference in findings.
Collecting data on whether identity verification is accomplished in separate interaction with significant others, peers, and guards may help explain why there is a negative direct relationship between self-esteem and self-views for both identities under study. Identity theory predicts that there is a direct and positive relationship between self-esteem and identity verification. However, the data limitations for the current study did not allow for any hypothesis testing with respect to identity verification. It is possible that the negative relationship between self-esteem and self-views (indicating that participants with low self-esteem were less likely to claim the criminal and/or worker identities) is the result of identity nonverification. According to identity theory, identity verification is a source of increasing self-esteem. This means that identity nonverification for the criminal and/or worker identity results in lower self-esteem levels. The tendency of individuals to disengage identities that are repeatedly nonverified (Burke and Stets 2009) may result in a low self-view for the nonverified identity that corresponds with a low self-esteem level. This relationship could be tested by analyzing the relationship between self-esteem, self-views, and verification of particular identities.
In addition, the findings regarding self-esteem must be interpreted in light of the restrictions placed on interactions of participants as regulated by the correctional facility. Participants in this study are not in a position to seek out alternative interaction partners freely if their current partners are not supporting self-views through verification. The findings may thus reflect the absence of choice of interaction partners. Participants may interpret reflected appraisals as more closely matching self-view because they have no alternative source for supportive reflected appraisals, rather than indicating that high self-esteem leads people to interpret reflected appraisals as consistent with self-views.
It is also possible that the restricted environment of the correctional facility works in the opposite direction and facilitates selective processing. The incarcerated participants are in a relatively easy position to cut themselves off from others on the “outside” who provide nonverifying feedback about the self. For example, an incarcerated person may simply choose not to accept visits from those who provide conflicting views of self. While this is also a potential option outside of the incarceration facility, it is much easier to deny interaction with another person (especially a family member) if there is literally no physical access possible. The limited nature of this sample means that while the results are useful for empirically observing the relationships under study, they are not generalizable to the population at large. Future research should test the selective processing and stability of self approaches in mainstream populations to see if the effects are different for people who do not have restrictions placed on social interaction.
Identity meaning may play a role in the relationship between reflected appraisals, self-esteem, and self-views. In some cases, the criminal identity is viewed as a negative identity that one would not want to possess. In other cases, the criminal identity may be worn with a badge of pride. It may signify physical superiority or a stylish outlaw kind of status that others in the same social circle strive to achieve. Although identity theory does not hypothesize differences in identity processes based upon identity meaning, it does specify that more self-esteem is derived from verification of identities that are more important to an individual. Considering whether an identity is positive or negative to an individual (rather than general societal categorization as such) and how salient it is relative to other identities held may be important determinants for which identities are sources of self-esteem for different people.
Treatment and Rehabilitation of Criminal Offenders
Treatment programs aimed at social-cognitive behavior modification for criminal offenders provide instruction, counseling, and other interventions aimed at changing participants’ self-views. These programs use the assumption of labeling theory that individuals will behave in accordance with some categorization applied to them by an external source (a “self-fulfilling prophecy”) (Becker 1963; Cullen and Gendreau 2000; Lemert 1951; Matza 1964; Tannenbaum 1938). The curriculum for such programs presume that offenders hold negative self-views as criminals and aim to instill changes in self-views by providing reflected appraisals (labels) that are positive. The idea is that if negative self-views are counteracted with enough positive appraisals, positive self-views will emerge. This type of intervention is on the rise particularly for criminal offenders who are found to be drug addicted, mentally ill, and/or both.
Typically, these programs are run by employees of the criminal justice system (in many cases they actually take place in incarceration facilities). In light of the results of this study, involvement of significant others and peers may yield greater success. The results indicate that those who participate in treatment programs may be more likely to adjust self-views according to labels applied by people they actually care about (as opposed to administrators of a program employed by the criminal justice system).
From an identity theory perspective, individuals seek to have identities verified through interaction. The implication is that offenders who successfully complete a sociocognitive intervention program (success defined as changing self-views from negative to positive) will be more likely to seek interaction with others who verify their new, positive self-views. Most criminal offenders return to the same social environment (and engage in the same interaction patterns with the same interaction partners) as prior to incarceration. Identity theory indicates that proximal social structure (the small group of family, friends, coworkers, etc. that an individual interacts with on a regular basis) is highly relevant for identity processes as it dictates (and restricts) who interaction partners are (Merolla et al. 2012; Serpe 1987; Stryker, Serpe, and Hunt 2005). In most cases, social structural position is beyond the control of an individual as one may lack the means necessary to move out of a given neighborhood. This makes it imperative to extend education and counseling programs aimed at changing an individual’s self-views to include interaction partners in the proximate social structure.
Both theoretical and empirical evidence suggest that the culture of a given neighborhood is often a risk factor for criminal behavior. In many cases interaction partners in the proximate social environment are also criminals and therefore provide positive feedback and social support for engaging in criminal behavior as well as opportunities to learn techniques of crime (Akers 1985; Sampson and Laub 1993; Sutherland 1947). In other cases, the proximate social environment may dictate social rules and values that are risk factors for crime such as maintaining a “tough” reputation in the neighborhood through demonstrations of physical strength and violence (Anderson 1999). If sociocognitive treatment and rehabilitation programs are to be successful at changing self-views of criminals, then interaction patterns and partners must also change. Future research on the outcomes of sociocognitive programs aimed at changing criminal self-views should include examination of participants’ social environment both before and after the program. This will allow for an assessment of whether changes in the proximate social structure correspond to changes in self-views.
In summary, the study presented here extends prior work on identity and self-esteem and addresses the call for symbolic interactionists to provide theoretical and empirical explanations of where individual agency exists in identity processes. The results demonstrate that self-esteem is a unique characteristic contributing to individualistic perceptions of reflected appraisals about the self. Since each person has a different level of self-esteem, no two people can interpret appraisals from others about the self in the same way, thus will not have the same self-views as one another for the same identity. Therefore, the criticism that symbolic interactionism does not attend to sources of individual differences in self-views as suggested by Stryker (1968) and Gecas and Schwalbe (1983) is addressed by acknowledging (and empirically verifying) individual levels of self-esteem as a source of differences in self-views.
Footnotes
Acknowledgements
The author would like to acknowledge Jan E. Stets, the anonymous reviewers, and the SPQ editor for comments on earlier drafts of this manuscript.
1
Due to limitations of the data on identity verification in the current research, the direct relationship of self-esteem and self-views are analyzed and presented. However, no specific predictions about the direction of the direct relationship between self-esteem and self-views are formulated for this study.
2
Individuals with high self-esteem are typically considered more confident than those with low self-esteem. In fact, they are sometimes considered arrogant or conceited and even egotistical (Baumeister 1993; Baumeister, Smart, and Bodel 1996). Baumeister and colleagues suggest that threats to egos of high self-esteem individuals may cause violent behavior (
). Despite this somewhat negative picture of high self-esteem individuals, it is typically regarded as a positive trait because it is associated with positive self-evaluations.
3
Asencio’s (2011) findings with respect to legitimate status contradict the findings presented earlier by . Explanations for this contradiction include the natural setting of the correctional context in the Asencio (2011) study as opposed to the laboratory setting in the Stets and Asencio (2008) study. It is possible that the concept of legitimaticy may be substantively different for participants in the lab than for the incarcerated participants.
4
Previous studies regarding the influence of feedback from others on self-views within an identity theory framework use actual appraisals as a measure of feedback from others rather than reflected appraisals (for exceptions, see Asencio 2011; Asencio and Burke 2011). This means that prior studies report self-views for the identities under study that are compared to actual reports from others (spouses, legitimate others) of how they would rate the individual on the identities under study. In contrast, the current study (as well as Asencio 2011 and Asencio and Burke 2011) measures reflected appraisals by asking respondents to report how they think others would rate them on a particular identity. This measurement approach follows the theoretical postulate of identity theory that actual appraisals from others are first interpreted into reflected appraisals before asserting an influence on self-views ().
5
A Brant test conducted in Stata version 11.2 confirmed that parallel regression assumptions were not violated for this model confirming that ordered logit is an appropriate method of analyses (). The percentage change in odds of the dependent variable for every standard deviation change in the significantly related independent variables was obtained using the listcoeffiecient command in STATA 11.2.