Reconsidering Labels and Primary Deviance

The Reflected Appraisals Model

Matsueda’s (1992) reflected appraisals model specifies a symbolic interactionist explanation of law violation (Heimer and Matsueda 1994). Central to the interactionist perspective is the role-taking process: When individuals encounter a problematic situation, they take the role of significant others, view themselves and situations from the perspectives of these others, and fit their actions into a social interaction (Mead 1934). Thus, the self is formed through role-taking and is composed of reflected appraisals or views of oneself from the perspective of others (Felson 1985). In applying the interactionist perspective to delinquency, Matsueda (1992) specifies reflected appraisals as a rule violator as a primary link between appraisals of others and law-violating behaviors. The reflected appraisals model maintains that adolescents who begin to view themselves as rule violators from the perspective of others act upon their rule-violating identity and engage in delinquency. This is consistent with early labeling perspectives focused on the self-fulfilling prophecy or the idea that individuals see themselves in the way others see them and eventually behave in ways that align with how they are viewed by others (Becker 1963; Lemert 1951, 1972; Tannenbaum 1938).

Embedding the reflected appraisals process within its broader social context, Matsueda (1992) offers that social status characteristics—social class, race, urban residence, and residence in a female-headed household—inform appraisals of adolescents as rule violators and reflected appraisals regardless of actual rule-violating behaviors (Heimer and Matsueda 1994; Matsueda and Heimer 1997; Rios 2011). Thus, the reflected appraisals model effectively subsumes the status characteristics and secondary deviance hypotheses identified by Paternoster and Iovanni (1989) as core components of labeling theory (see also Farrington and Murray 2014).

The status characteristics hypothesis proposes that disadvantaged social positions inform law-violating appraisals, even among those who have not engaged in behaviors warranting the label. For instance, some adolescents are labeled as troublemakers due to socioeconomic disadvantage, minority status, or residence in a home with a nontraditional family structure. Consistent with this, Rios (2011) reports that impoverished Black and Latino parents sometimes seek parenting advice from conventional institutions, such as schools, that often criminalize the demeanor and behaviors of Black and Latino adolescents. This leads parents ultimately to respond to their children as delinquents in their efforts to parent them effectively. Similarly, an impoverished, single, Black mother in S. Elliott, Powell, and Brenton’s (2015:360) study relays that in order to protect her children from racist encounters, she strictly monitors their behaviors and “strips their beds [and] their pillow cases, just like a jail cell.” That is, she treats her children like criminals in her efforts to help them avoid being treated as such by others. Perhaps unsurprisingly, empirical research shows that African American adolescents are more likely than other adolescents to view themselves as troublemakers from the perspective of others, including their parents, even if their behavioral repertoires are similar (Heimer and Matsueda 1994; Matsueda 1992). Delinquent self-identities, therefore, hold the potential to explain the onset of delinquency as the result of delinquent identities being shaped by factors other than prior misbehavior. Unfortunately, researchers have not focused explicit, empirical attention on whether the reflected appraisals process applies in cases where adolescents have not engaged in prior delinquency but have been falsely appraised as rule violators by significant others (cf. Triplett and Jarjoura 1994). This is somewhat surprising, given the implications for broadening labeling perspectives to address not only secondary deviance but also primary deviance.

The reflected appraisal model focuses centrally on the secondary deviance hypothesis, which emphasizes the role of labeling in the continuation of law violation. Among Lemert’s (1951) leading contributions to the labeling perspective is his differentiation between primary and secondary deviance. Primary deviance is the original deviant act and is attributed by labeling perspectives to structural, cultural, and individual factors emphasized in etiological theories of delinquency. Secondary deviance is the outcome of the labeling/reflected appraisal process and includes the behaviors adolescents engage in to live up to other people’s expectations of them as rule violators or delinquents (Lemert 1951; Matsueda 2001). A lingering critique of the labeling perspective is its inability to explain primary deviance. If the reflected appraisals process is stimulated by false appraisals and produces law violation among delinquency abstainers, the labeling perspective has more to offer by way of understanding primary deviance than conceded by proponents of the perspective.

The Falsely Appraised

Becker (1963) introduced labeling theory to the falsely accused category in his classification of people based on their relationship to rule violation and labeling. The falsely accused is a person who has not broken any rules but is labeled a rule violator, the pure deviant has broken rules and has been labeled as such, the conformist has not violated rules and has not been labeled a rule violator, and the secret deviant has violated rules but has not been so labeled (Becker 1963; Matsueda 2006; Rains 1975). The two categories of most theoretical importance to researchers interested in how societal labels influence self-identities and subsequent behaviors are the true deviants and the falsely accused, as these are the groups that have received the imputed rule-violating stigma.

Attention to the falsely accused in criminology has focused primarily on identifying the characteristics of those who are falsely appraised as rule violators (Bowers 2008; Chevigny 1969; Klemke and Tiedeman 1990; Menard and Pollock 2014). Research demonstrates that minorities and economically marginalized individuals are disproportionately the targets of false formal accusations and false informal appraisals (Bowers 2008; Chevigny 1969; Cray 1967; Heimer and Matsueda 1994; Klemke and Tiedeman 1990; Matsueda 1992; Menard and Pollock 2014; Pollock and Menard 2014). Insufficient attention has been afforded to understanding if or how false accusations and appraisals trigger delinquency onset.

The reflected appraisals model makes the implicit assumption that identity processes operate similarly for true delinquents and falsely appraised adolescents. That is, falsely appraised individuals—like their true delinquent counterparts—are expected to accept their tarnished identity and live up to it by engaging in delinquency. It is likely, however, that falsely appraised adolescents engage in more efforts than their delinquent counterparts to resist or tune out delinquent appraisals and thereby are less prone to accept rule-violating appraisals into their self-identities. Indeed, research shows that both informal appraisals and prior law violation are strong, independent predictors of delinquency (Heimer and Matsueda 1994; Koita and Triplett 1998; Matsueda 1992). It remains to be seen if law-violating behaviors and informal appraisals interact in the production of delinquency such that law-violating informal appraisals are more potent predictors of delinquent identities among those who have engaged in delinquency than among those who have abstained from delinquency.

Even if false appraisals prove relatively impotent for identity processes among delinquency abstainers, such appraisals likely remain relevant for understanding delinquency onset. Koita and Triplett’s (1998) assessment of the reflected appraisals model across race and sex shows that parental appraisals influence reflected appraisals of self and delinquency as expected among White adolescents. Among Black males, however, parental appraisals influence delinquency even though they are not incorporated into self-identities. This suggests that informal appraisals influence delinquency through social processes unrelated to self-identity (Paternoster and Iovanni 1989; Sampson and Laub 1997). For instance, prior research suggests that social exclusion from conventional institutions and conventional relationships that help people avoid delinquent subcultures and crime are an important link between law-violating stigmas and future offending (Bernburg 2009; Bernburg and Krohn 2003; Heimer and Matsueda 1994; Paternoster and Iovanni 1989; Sampson and Laub 1997; Wiley, Slocum, and Esbensen 2013). Although identity processes are likely to vary across true delinquents and falsely appraised adolescents, there is little reason to expect that the social exclusion resulting from labels operates differently across these groups. Thus, continuity in law violation would appear to be more strongly influenced by labels than the onset of delinquency among falsely appraised adolescents when taking into consideration only the identity and social exclusion processes highlighted by labeling and interactionist perspectives.

Falsely appraised adolescents, however, may be pushed into delinquency by processes specific to their unwarranted labels. General strain and defiance theories offer that the injustice or illegitimacy associated with being falsely appraised as a rule violator can trigger emotions that lead to delinquency as corrective action or angry defiance (Agnew 1992, 2006; Sherman 1993). General strain theory maintains that stresses lead to negative emotions that trigger delinquency as a means of corrective action. The disjuncture between individual ideals of justice and the reality of unjust treatment is a stress Agnew (2001) identifies as particularly likely to produce law violation. False appraisals as a rule violator can be conceptualized as a form of unjust treatment that leads to negative emotions and delinquency as corrective action. Thus, false appraisals can lead to delinquency because it is stressful to be misidentified in a negative light, not because the experience of being falsely appraised leads adolescents to incorporate delinquent views into their self-identities, as proposed in the reflected appraisals model.

Defiance theory similarly emphasizes negative emotions as a link between perceived punitive mistreatment and delinquency. In identifying the conditions under which punishments produce varying deterrence or defiance effects, the theory maintains that criminal defiance is likely when punishments are perceived as unfair, the target of punishment feels alienated from the source of the sanction, and the target of punishment feels the sanction is aimed at his or her person rather than at a particular behavior (Sherman 1993). Criminal defiance results under these circumstances because the target of punishment feels a sense of angry pride that leads to offending. The falsely appraised are likely to view negative appraisals as unfairly punitive, feel alienated from those who falsely appraise them, and experience the appraisals as a condemnation of self and not (nonexistent) behavior. Thus, primary deviance among the falsely appraised can be conceptualized as a defiant response to the imposition of unwarranted appraisals.

Overall, we conclude that the reflected appraisals model may prove suitable principally for understanding continuity in delinquency among adolescents who have engaged in law violation prior to receiving their delinquent appraisal. This is because falsely appraised adolescents may be more prone than true delinquents to resist troublemaking labels, thereby effectively interrupting the process through which informal labels are incorporated into self-identities but not necessarily mitigating the influence of informal labels on behavior. Adolescents who have engaged in delinquency may be more prone to accept delinquent labels into their self-identities, as evidenced by the fact that both delinquent appraisals and prior involvement in delinquency are strong predictors of reflected appraisals as a rule violator (Bartusch and Matsueda 1996; Koita and Triplett 1998; Matsueda 1992). False appraisals are likely to lead to law violation among adolescents who did not earn their labels by engaging in rule violation through additional social processes not articulated in the reflected appraisals model. In particular, the falsely appraised may be vulnerable to experiencing negative emotions associated with the injustice or illegitimacy of being falsely appraised as troublemakers by significant others.

Hypotheses

Using Matsueda’s (1992) reflected appraisals model, we offer specific hypotheses about how the influence of appraisals by significant others is likely to vary across individuals who have engaged in delinquency prior to being appraised as troublemakers and those for whom rule-violating appraisals are unwarranted based on their prior behaviors. In doing so, our emphasis is on the extent to which reflected appraisals are shaped by parental appraisals and on the potency of reflected appraisals in mediating the influence of parental appraisals on continuity in delinquency among true delinquents or delinquency onset among falsely appraised adolescents.

Our hypotheses for those who have engaged in delinquency prior to being appraised as troublemakers derive directly from Matsueda’s (1992) reflected appraisals model. We propose alternative hypotheses for individuals who did not earn their labels behaviorally, offering that false appraisals are likely to lead directly to delinquency, even when accounting for their influence on reflected appraisals of self, because such appraisals are likely to be experienced as unjust, illegitimate, and/or stressful.

Data

We assess our hypothesized differences in how the reflected appraisals model applies to understanding continuity in delinquency among true delinquents and onset of delinquent among falsely appraised adolescents using the National Youth Survey (NYS) data originally used by Matsueda (1992) to assess the reflected appraisals model (see also Bartusch and Matsueda 1996; De Coster and Heimer 2001; Heimer and Matsueda 1994; Koita and Triplett 1998). The NYS is a national probability sample of 11- to 17-year-olds in the United States from 1976. Seventy-three percent of the final sample of adolescents (1,725) agreed to participate in the survey. They were interviewed in their homes for the first time in 1977 and then reinterviewed annually.

One parent of each youth was also interviewed during the first wave. Following Matsueda’s (1992) original analysis, our analysis uses variables from the first three waves. The attrition rate over the three-year span was low, 4 percent in 1978 and 6 percent in 1979, and comparisons of respondents across waves indicate that loss by demographic variables and law violation did not substantially influence the underlying distributions on these variables (D. Elliott, Huizinga, and Menard 1989). We use the data from 1,086 youths who reported having engaged in delinquency during the second wave and 586 youths who reported they did not engage in delinquency during this time.

Although the NYS was collected four decades ago, it remains the only longitudinal, national data set that contains information on status characteristics, parental appraisals, reflected appraisals, and delinquency. Given that recent decades have witnessed changes in race and gender relations, the NYS would not be appropriate for documenting contemporary race and sex gaps in labeling, delinquent identities, and delinquency. The data, however, remain valuable for testing generalized theoretical linkages between status characteristics, labeling, identities, and delinquency across groups of true delinquents and falsely appraised youth (see Collins’s [1989] distinction between the recency of data required for descriptive research vs. theory-oriented research).

Specification of Models

The constructs specified in our theoretical discussion are measured indirectly by responses to interview questions, which can contain errors in measurement. We address measurement error in parental appraisals and reflected appraisals by specifying a measurement model as well as a substantive model. The measurement model allows us to estimate and control for bias resulting from response error in these measures. The substantive and measurement models are identical to those used in Matsueda’s (1992) original study. The substantive model is diagrammed in Figure 1; the measurement model is omitted from the diagram for simplicity. The specific survey items used to measure theoretical constructs, which align with those used in Matsueda’s (1992) study, are listed in Appendix Table A1. Because we hypothesize differences in effects among adolescents who have been delinquent prior to the identity processes specified herein as compared to youths who have not been delinquent recently, we specify and estimate the models separately for delinquents and nondelinquents and then test for the equality of coefficients across groups by conducting chi-square difference tests on models with and without equality constraints across groups.

OPEN IN VIEWER

Figure 1.

Structural model of the reflected appraisals process.

Delinquents and nondelinquents are differentiated using a 24-item delinquency scale from time 2 that comprises property, violent, and drug offenses (D. Elliott, Huizinga, and Ageton 1985). ¹ The scale is dichotomized to create two categories: 0 for those who had not engaged in any delinquent acts in the previous year and 1 for those who had engaged in delinquency during this time. Our strategy for differentiating delinquents and nondelinquents is consistent with Paternoster’s (1989) strategy for assessing differences in decisions to offend versus decisions to desist.

Before describing the models, it is worth reiterating that the only difference between our models and those presented by Matsueda (1992) is that we estimate models across groups of delinquents and nondelinquents, whereas he controls for prior delinquency in a pooled model of delinquents and nondelinquents. This allows us to assess our hypothesized interactions between delinquency status and identity processes. Using the same data and modeling strategy as Matsueda allows for a stringent test of our challenge to the implicit assumption that the reflected appraisals process applies similarly to true delinquents and falsely appraised adolescents.

Consistent with Matsueda’s (1992) modeling, Figure 1 shows that our substantive model consists of the following blocks of variable: (1) a vector of exogenous variables capturing social structural positions measured at time 1, (2) parental appraisals at time 1, (3) reflected appraisals of self at time 2, and (4) self-reported delinquency at time 3. The ordering of the variables in the model follows the theoretical specification of the reflected appraisals process: Parental appraisals influence reflected appraisals, which ultimately inform delinquency. The time ordering of variables follows the theoretical ordering to assist in making causal inferences.

The vector of exogenous variables captures social status characteristics measured at wave 1, including family income, Black racial status, residence in a single-parent household, and urban residence. Each of these variables is a single-indicator dummy variable, coded 0 for advantaged and 1 for disadvantaged position. Youth’s age and sex are included also because these variables are strong correlates of delinquency. Inclusions of variables capturing structural disadvantage allows for consideration of the status characteristics hypothesis that informal appraisals are shaped not only by behavior but also by social positions.

The vector of parental appraisal variables includes four constructs capturing parental appraisals of children as likely to succeed, sociable, distressed, and rule violators. Although we are particularly interested in parent appraisals of adolescents as rule violators, as indicated in the hypotheses, we include all the parent appraisals variables incorporated in Matsueda’s (1992) model for comparability purposes. Parental appraisals of youth as likely to succeed is a single-indicator item, measured by parent reports of their child as likely to be successful. The remaining parent appraisal variables are latent variables. Parental appraisals as sociable is measured by two observable indicators asking parents to report how well-liked their child is and how well their child gets along with others. Parental appraisals as distressed is captured by parent reports that their child is often upset and that their child has problems. The final parent appraisal variable—parental appraisal as a rule violator—is a two-indicator variable capturing reports from parents that their child is a troublemaker and that their child breaks rules.

The reflected appraisals variables include four latent variables: reflected appraisals as likely to succeed, sociable, distressed, and a rule violator. Reflected appraisals as likely to succeed is measured by three items asking youth to report whether their parents view them as likely to succeed, friends view them as likely to succeed, and teachers view them as likely to succeed. Reflected appraisals as sociable are captured by six variables asking separately whether parents, friends, and teachers view them as likable and as getting along with others. Reflected appraisals as distressed are captured similarly by six questions asking the extent to which parents, friends, and teachers view them as often upset and as having problems. Finally, reflected appraisals as a rule violator are captured by six items asking adolescents to report whether parents, friends, and teachers view them as troublemakers and as rule violators. Following Matsueda (1992), our measurement model includes correlations for identical measures that differed only with respect to the referent significant other (e.g., breaks rules from the standpoint of parents, teachers, and friends) and among measures of a given latent construct that referred to a similar significant other (e.g., breaks rules from the standpoint of teacher and troublemaker from the standpoint of teacher).

Our outcome variable, general delinquency, is measured at wave 3 and asks adolescents to self-report involvement in violent, property, and drug offenses during the year between the wave 2 and wave 3 interviews. The general delinquency scale comprises the same 24 items as the scale used to differentiate delinquents and nondelinquents for the splitting of the sample. We use the natural log of the mean delinquency scale to correct for skewness. ²

Estimation of the Models and Results

The substantive and measurement models are estimated simultaneously using the maximum-likelihood procedures in LISREL version 8. The model fits the data well for delinquents (L² = 840.07, df = 434) and nondelinquents (L² = 731.62, df = 434). Indeed, the root-mean-square error of approximation is .028 for delinquents and .030 for nondelinquents, which indicates a close fit between the model and the data for both groups (MacCallum, Browne, and Sugawara 1996). Additionally, the model produces a comparable fit index score of .98 for both groups, which indicates an acceptable fit for the model (Jöreskog and Sorbom 1996).

Correlations among latent variables are presented in Appendix Table A2 for delinquents (above the diagonal) and nondelinquents (below the diagonal); descriptive statistics and estimates for the measurement models are presented in Appendix Table A3. The correlations evidence strong, positive relationships between parental rule-violating appraisals, reflected appraisals of self as a rule violator, and subsequent delinquency among both previously delinquent and nondelinquent adolescents. The means for delinquents and nondelinquents in Appendix Table A3 show that those who had been delinquent prior to being labeled engaged in more subsequent delinquency than their previously nondelinquent counterparts. In addition, delinquents are more likely to have been appraised as delinquent and distressed and are more likely to see themselves from the perspective of others as delinquent and distressed. They are less likely to view themselves positively as sociable and successful from the perspective of others.

The estimates for the measurement model show that the observed indicators are reasonably valid and reliable measures of unobserved theoretical constructs. The metric slopes column shows that all the observed variables significantly load on their respective latent factors, and the validity coefficients (standardized loadings) for indicators of the parent appraisals and reflected appraisals are relatively high, ranging from .54 to .92 for delinquents and .61 to .86 for nondelinquents. The results across delinquents and nondelinquents verify Matsueda’s (1992) substantive claim that youth reflected appraisals from parents, peers, and teachers coalesce into a noncompartmentalized self. And, the indicators of the parental appraisal and adolescent reflected appraisal variables appear to be satisfactory measures of the theoretical constructs. Table 1 presents the total effects of the truly exogenous variables and parental appraisals on delinquency for those who engaged in delinquency prior to being appraised as troublemakers (delinquents) and for those who did not earn their labels behaviorally (nondelinquents); and the unstandardized (b) and standardized (β) parameter estimates of direct effects for the substantive model are presented in Tables 2 (delinquents) and 3 (nondelinquents).

OPEN IN VIEWER

Table 1.

Total Effects of Exogenous Variables and Parental Appraisals on Delinquency.

Independent Variables	Delinquents (1)	Nondelinquents (2)
(1) Agea	.009*** (.002)	.003*** (.001)
(2) Black race	−.032** (.010)	.010 (.004)
(3) Urbana	.022** (.008)	.007 (.004)
(4) Female-headed household	.025** (.009)	.016** (.005)
(5) Family income	−.004* (.002)	−001 (.001)
(6) Female sexa	−.045*** (.007)	−.013*** (.003)
(7) Parent appraisal—sociable	−.004 (.012)	.005 (.008)
(8) Parent appraisal—successful	.003 (.006)	.003 (.004)
(9) Parent appraisal—distressed	−.011 (.008)	−.003 (.005)
(10) Parent appraisal—rule violatora	.054*** (.010)	.026** (.007)

Note: Numbers in parentheses are standard errors.

^aSignificant group difference p < .01.

*p < .05. **p < .01 (two-tailed test). ***p < .001.

OPEN IN VIEWER

Table 2.

Unstandardized and Standardized Direct Effects for the Structural Model (Delinquents).

Dependent Variables	(1) Age	(2) Race	(3) Urban	(4) Single- parent Family	(5) Family Income	(6) Sex	Parent Appraisals				Reflected Appraisals
							(7) Sociable	(8) Success	(9) Distressed	(10) Rule Violator	(11) Sociable	(12) Success	(13) Distressed	(14) Rule Violator	R 2
Parent appraisals
1. Sociable	.001 (−.009) .003	−.021 (−.05) −.017	−.005 (−.039) −.005	−.021 (−.045) −.02	.014 (−.008) .076	−.02 (−.034) −.022	—	—	—	—	—	—	—	—	.01
2. Success	−.012 (−.01) −.035	.058 (−.06) .033	−.026 (−.047) −.018	−.055 (−.054) −.037	.0278* (−.01) .100	.029 (−.04) .022	—	—	—	—	—	—	—	—	.02
3. Distressed	.029** (−.011) .09	.027 (−.062) .016	.039 (−.048) .028	.108 (−.057) .074	−.030** (−.01) −.115	.032 (−.042) .026	—	—	—	—	—	—	—	—	.04
4. Rule violator	−.026* (−.013) −.072	.259*** (−.074) .137	.101 (−.053) .065	.036 (−.067) .022	−.021 (−.012) −.073	−.080* (−.04) −.057	—	—	—	—	—	—	—	—	.06
Reflected appraisals
5. Sociable	.009 (−.006) .049	.007 (−.037) .008	.015 (−.023) .02	−.007 (−.033) −.009	.003 (−.006) .019	.043* (−.024) .071	.195*** (−.05) .263	.023 (−.024) .044	−.001 (−.033) −.001	.006 (−.032) .012	—	—	—	—	.09
6. Success	.001 (−.011) −.002	−.076 (−.062) −.046	.061 (−.043) −.046	.03 (−.056) −.021	.0268* (−.01) −.104	.012 (−.041) −.01	.128 (−.08) −.096	.0808* (−.04) −.085	.017 (−.055) −.017	−.107 (−.055) −-.123	—	—	—	—	.08
7. Distressed	−.007 (−.011) −.021	.111 (−.062) .068	−.098* (−.047) −.073	−.012 (−.055) −.009	−.034** (−.01) −.137	.172*** (−.042) .142	−.054 (−.078) −.04	−.025 (−.04) −.027	.103 (−.0551) .107	.1518* (−.055) .175	—	—	—	—	.14
8. Rule violator	.015 (−.012) .043	−.207** (−.068) −.11	−.091 (−.052) −.06	.116 (−.06) .072	−.009 (−.011) −.031	−.215*** (−.045) −.155	−.063 (−.086) −.041	.008 (−.044) .008	−.039 (−.061) −.081	.434*** (−.064) .439	—	—	—	—	.2
Delinquency
9. Delinquency	.009*** (−.002) .153	−.026** (−.01) −.088	.021** (−.007) .086	.016* (−.008) .061	−.003* (−.002) −.074	−.024*** (−.007) −.119	−.007 (−.012) −.108	.001 (−.006) .008	.002 (−.008) .011	.018* (−.008) .116	.062*** (−.015) .19	.002 (−.007) .013	−.022** (−.008) −.119	.074*** (−.008) .461	.29

Note: Standard errors are in parentheses; standardized coefficients are boldfaced. N = 1,069, λ² = 840.07, df = 434, root-mean-square error of approximation = .028.

*p < .05. **p < .01. ***p < .001 (two-tailed test).

OPEN IN VIEWER

Table 3.

Unstandardized and Standardized Direct Effects for the Structural Model (Nondelinquents).

Dependent Variables	(1) Age	(2) Race	(3) Urban	(4) Single-parent Family	(5) Family Income	(6) Sex	Parent Appraisals				Reflected Appraisals
							(7) Sociable	(8) Success	(9) Distressed	(10) Rule Violator	(11) Sociable	(12) Success	(13) Distressed	(14) Rule Violator	R 2
Parent appraisals
Sociable	.024* (.010) .111	−.144* (.069) −.116	.019 (.051) .019	−.039 (.069) −.032	−.010 (.011) −.054	−.015 (.043) −.017	—	—	—	—	—	—	—	—	.03
Success	.022a (.012) .074	−.130 (.081) −.077	.046 (.059) .033	−.068 (.080) −.041	−.018b (.012) −.072	−.054 (.051) −.044	—	—	—	—	—	—	—	—	.02
Distressed	−.005a (.015) −.016	.194 (.102) .109	−.067 (.074) −.046	.052 (.101) .030	.007a (.015) .027	−.028 (.064) −.022	—	—	—	—	—	—	—	—	.01
Rule violator	−.028* (.013) −.110	.304*** (.086) .204	.050 (.062) .041	.121 (.085) .083	−.024 (.013) −.101	−.159** (.054) −.147	—	—	—	—	—	—	—	—	.14
Reflected appraisals
Sociable	−.007 (.008) −.041	.003 (.054) .003	−.001 (.039) −.002	.008 (.053) .008	.005 (.008) .032	.103** (.036) .144	.191* (.082) .241	−.014 (.042) −.023	.020 (.053) .037	−.024 (.068) −.036	—	—	—	—	.07
Success	.008 (.012) .028	−.246** (.082) −.156	.69 (.059) .053	.099 (.079) .064	−.002 (.013) −.007	.085 (.052) .074	−.102 (.120) −.081	.042 (.062) .045	−.102 (.080) −.115	−.232* (.104) −.219	—	—	—	—	.13
Distressed	.002 (.011) .008	.124 (.076) .086	.000 (.054) .000	.060 (.073) .042	−.043** (.012) −.198	−.130*a (.046) .023	−.104 (.112) −.090	−.046 (.057) −.054	.270** (.080) .335	−.064 (.095) −.067	—	—	—	—	.22
Rule violator	.025*b (.011) .099	.138*b (.069) .115	−.036 (.053) −.030	.004 (.071) .003	−.028* (.011) −.125	−.135** (.047) −.129	−.104 (.112) −.069	−.046 (.057) −.059	.270** (.080) .103	−.064a (.095) .140	—	—	—	—	.15
Delinquency
Delinquency	.003**b (.001) .153	.002a (.005) .019	.007 (.004) .075	.015** (.005) .140	.000 (.001) .003	−.009** (.003) −.116	.002 (.008) .027	.004 (.004) .066	−.004 (.006) −.069	.022** (.007) .166	.016b (.008) .141	−.007 (.004) −.099	−.002a (.006) −.031	.016*b (.007) .212	.16

Note: Standard errors are in parentheses; standardized coefficients are boldfaced. N = 586; λ² = 731.62, df = 434, root-mean-square error of approximation = .030.

^aSignificant group difference p < .05.

^bSignificant group difference p < .01.

*p < .05. **p < .01. ***p < .001 (two-tailed test).

Conclusions

Our article advances research on labeling and delinquency in several ways. First, we provide evidence that rule-violating appraisals by significant others and reflected appraisals as a rule violator produce future delinquency among adolescents who had been delinquent and among those who had not been delinquent prior to being appraised as such. In other words, delinquent appraisals prove relevant for understanding both continuity in delinquency among true delinquents and the onset of delinquency among those falsely appraised. Thus, the concession that labeling processes can only explain secondary deviance to the exclusion of delinquency onset may be worth reconsidering.

Although rule-violating appraisals influence delinquency among both delinquents and those whose labels do not align with prior behavior, our second contribution to the literature is in reporting notable differences in how the reflected appraisals process operates across groups of delinquents and nondelinquents. These differences are relevant to both the secondary deviance hypotheses and the status characteristics hypothesis. With respect to the former, reflected appraisals as rule violators are a more robust predictor of continuity in delinquency than of delinquency onset and are relevant for linking parental appraisals to delinquency among those with delinquent histories but irrelevant for linking parental appraisals to delinquency among the falsely appraised. As such, future work examining the processes through which false appraisals influence delinquency onset is warranted. In particular, future research should consider the possibility that false appraisals influence delinquency onset through processes articulated in general strain and defiance theories. This would entail consideration of negative emotions—including anger, frustration, pride, and anxiety—alongside reflected appraisals as conduits through which misplaced rule-violating appraisals influence delinquency onset. Unfortunately, measures of respondent’s emotions are unavailable in the NYS. There is a measure of depression beginning in the wave 6 data, but the parent appraisals measures are available only at wave 1. Effectively linking false parental appraisals to emotional responses requires a time span much shorter than five years as well as a more comprehensive range of emotions.

A third contribution is in our findings regarding the status characteristics hypothesis. Consistent with prior research, we find that reflected appraisals of self are shaped by race and sex. Youth’s sex operates similarly in influencing the reflected appraisals model across groups of delinquents and nondelinquents, with delinquent and nondelinquent males being more likely than their female counterparts to be appraised as rule violators and more likely to develop reflected appraisals as rule violators. This is consistent with prior research and theorizing but goes beyond prior findings by showing that differential involvement in delinquency is only one explanation for why males are more likely to be labeled as troublemakers and absorb these labels into their identities (Bartusch and Matsueda 1996; Heimer 1996). Our finding that previously nondelinquent males are more likely than previously nondelinquent females to be appraised as rule-violators and form reflected appraisals as such provides evidence for the claim that cultural definitions of gender shape labeling and identity processes above and beyond explanations grounded in actual behavioral differences among individual male and female adolescents.

We find also that race shapes reflected appraisals and does so differentially across groups of previously delinquent and nondelinquent adolescents. Our results suggest that Black adolescents who engage in delinquency are relatively more protected than their nondelinquent Black peers from troublemaking self-identities. Although unexpected, this finding may be indicative of a number of social and cultural processes worthy of further attention and scrutiny. For instance, it may be that delinquent Black adolescents are especially guarded against powerful institutions of social control and negative cultural images of African Americans that adversely shape the everyday experiences of Black individuals and predominantly Black communities. Some researchers suggest that formal labeling, including police contact and arrests, often does not influence self-identities among Black adolescents because these adolescents are likely to reside in communities of disadvantage that have been the target of concentrated surveillance and incarceration, rendering labels irrelevant because they are an everyday part of most community residents’ lives (Hirschfield 2008). It remains to be seen if self-identity immunity from law-violating labels applies equally to those in the community who are involved in crime as compared to those who actively work to avoid criminal entanglements. Future research on race, false accusations, and labeling is required in order to understand how informal labels—appraisals as troublemakers by teachers, school administrators, criminal justice actors, community members, peers, and parents—and formal labels—police contacts, arrest, and conviction—coalesce to differentially influence self-identities, emotional responses to labeling, and behaviors across groups of true delinquent and falsely accused Black adolescents. It may be the case that angry defiance serves as a neutralization that helps preserve self-identities while also producing continuity in law violation among Black adolescents. These, of course, are speculations derived from intriguing findings in our study that require further empirical scrutiny.

Further research might also consider the implications of our findings for broader areas of inquiry focused on formal labels and life-course development. Life-course research focuses on the long-term effects of formal labeling and incarceration, emphasizing that both social exclusion from conventional institutions and self-identities are central to understanding continuity and desistance in crime over time (Laub and Sampson 2003; Maruna 2001; Maruna et al. 2004; Massoglia and Uggen 2010; Sampson and Laub 1997). Consistent with the general tenets of defiance theory, our research suggests the importance of considering that there may be additional variability in life-course trajectories rooted in whether individuals were falsely accused by the justice system or perhaps in individuals’ perceptions of the extent to which they were falsely or rightfully targeted for arrest and punishment. Future research on the life-course might benefit by differentiating between falsely accused and true criminals when considering life trajectories as well as between more refined categories of individuals who perceive their punishments to have been just or unjust with emphasis on how these perceptions change over time and influence criminal trajectories.

It would be important, of course, for life-course researchers to consider that the categories of true criminals and falsely accused may be more fluid than our research suggests. For instance, it is possible that some of the true delinquents in our study were falsely appraised as troublemakers during childhood, which led them to engage in delinquency as a form of defiance during the time period under study. Although these adolescents are categorized as true delinquents on account of the snapshot we were able to capture with the data available, it is worth considering that those who start their offending trajectories as a consequence of false appraisals or accusations may differ from those whose trajectories begin without unwarranted labels. In addition, there were adolescents in our sample who had been delinquent the year prior to when we captured parental appraisals but had desisted from offending during the year in which we captured parental appraisals. We defined these adolescents as nondelinquents because their labels stuck even though their behavior had changed, which was likely to have been experienced by adolescents as unjust or unwarranted. Sensitivity analyses confirmed our decision to treat these youths as nondelinquents, as models including only the adolescents who desisted from offending were similar to models including only nondelinquents and not to the models of delinquents whose informal appraisals aligned with their recent behaviors. Of course, our need to grapple theoretically and empirically with where to place desisters in our model underscores the importance of future research that explicitly captures the extent to which people experience informal and formal labels as unwarranted or justified based on their actual behaviors over time.

Finally, continued research would benefit from considering the processes studied herein with more recent data. Our findings regarding race differences in how the reflected appraisals process operates within and across groups of falsely appraised and true delinquents reveal patterns worthy of further consideration. Although we offer no hypotheses regarding time and place, it is conceivable that the processes we highlight are informed by shifts in the racial climate over time and by the degree of structural marginalization and legal cynicism across communities. Testing these empirical possibilities requires data on the reflected appraisals process from more recent years and/or multilevel data that allows the reflected appraisals process to be embedded within the broader racial and structural environments in which identity processes unfold.