Disentangling Mental Illness Labeling Effects from Treatment Effects on Well-Being

Abstract

The emerging field of Mad Studies has returned attention to deficiencies of the medical model, refocusing scholars on social causes of mental health problems and on consumers’/survivors’ experiences of labeling and stigma. These themes echo issues addressed in traditional and modified labeling theories. A fundamental labeling premise is that professional categorization as “mentally ill” is a major determinant of individuals’ poorer psychological well-being. However, this relationship has not been tested appropriately because past studies frequently measured formal labeling by a person’s involvement in treatment. Treatment involvement can indicate the receipt of potentially beneficial services or harmful categorization with a stigmatizing label. Independent measures of these constructs in the National Comorbidity Survey-Replication enable reexamining traditional and modified labeling hypotheses for individuals with (N = 1,255) and without (N = 4,172) a recurrent clinical disorder. Supporting labeling theory’s central proposition, formal labeling was linked to more negative affect and disability days in both groups. These relationships were not spurious products of preexisting serious symptoms, refuting a psychiatric explanation. Treatment involvement effects differed noticeably between the groups, underscoring the need to keep treatment and labeling measures distinct.

Keywords

mental illness labeling theory modified labeling theory well-being

[A]ctually what I think that [medical] model does is . . . it puts all of the blame on to you, there’s something wrong with you, that’s the message that you get . . . yet I would say society also plays a role in our individual experience because we’re part of society.

I think it’s the one thing that is still very stigmatized, and people make all sorts of assumptions about people who’ve got mental health problems. (Beresford, Nettle, and Perring 2010:14)

The expanding field of Mad Studies (e.g., Kirk, Gomory, and Cohen 2013; Menzies, LeFrançois, and Reaume 2013; Rimke 2016) has revived criticisms of the biomedical model of psychiatry, pointing to the social construction of psychiatric disorders, the medicalization of individuals’ emotional and mental states by powerful pharmaceutical companies, and the disregard of consumers’/survivors’ accounts of oppressive treatment and stigma. These themes reprise some of the central concerns raised by labeling theory and modified labeling theory in earlier decades of work in the sociology of mental health (Becker 1963; Link 1987; Link et al. 1989; Scheff 1963). Resonating with Mad Studies and mad activism, traditional labeling theory held that psychiatric categories are social constructions, that those categories get imposed by authorities on persons who breach behavioral or feeling norms, and that inequalities in social status and power are sustained as a consequence. Traditional and modified labeling theorists also argued that acquiring an official mental illness label triggered inter- and intrapersonal dynamics that funneled “deviant” individuals into long-term “careers” of coping with adverse life circumstances and chronic mental health problems. Consumers’/survivors’ experiences with mental health professionals, public stereotyping, and stigma echo these arguments (e.g., Beresford et al. 2010).

Given these themes in common, past mental illness labeling research could presumably substantiate consumers’/survivors’ critiques of the medical model and their strong concerns about the consequences of psychiatric labeling and stigma. To date, however, labeling theories’ fundamental claim—that a mental illness label is harmful to well-being—has not been appropriately tested. This is because researchers most often used individuals’ involvement in mental health treatment to indicate having been formally labeled. Treatment involvement is an ambiguous measure: It can mean categorization with a stigmatizing label (from a labeling perspective) or access to beneficial therapeutic services (from a psychiatric perspective).

To better examine consumers’/survivors’ concerns, this article retests traditional and modified labeling theories’ propositions with a more fitting indicator of formal labeling. The article traces the evolution of explanations of individuals’ poorer well-being from traditional labeling, modified labeling, and psychiatric perspectives; describes problems created by previous use of treatment involvement to indicate formal labeling; and tests competing labeling versus psychiatric explanations of poorer well-being. The focus will be on four theorists who have developed, critiqued, and modified traditional labeling theory—Scheff, Gove, Link, and Rosenfield. These theorists originated the central arguments, conducted key studies, and supplied detailed, updated reviews of the research literature over the years.

Labeling Theory’s Evolution, with Hypotheses Needing Reexamination

Traditional Labeling Theory: Scheff

Traditional labeling theory proposed that formal categorization as “deviant” by agents of social control (e.g., police, judges, psychiatrists) was a crucial event setting off a cascade of unfortunate consequences—stigma, stereotyping, and discrimination—that would keep labeled individuals trapped in a deviant career (e.g., Becker 1963; Lemert 1967). Thomas Scheff (1963, 1966, 1974, 1984) applied labeling theory specifically to mental illness to explain individuals’“careers” of persistent mental health problems.

In his application of the theory, Scheff (1963, 1966, 1984) presumed that almost all persons experience one or more psychological symptoms at some time in their lives, describing symptoms as “residual norm violations.” Typically, these are transient, explained away, and unlabeled. However, when the frequency, seriousness, or visibility of symptoms cannot be excused or ignored by other people, the person will be brought to the attention of authorities who will officially classify this deviant behavior as “mental illness” and the person as “mentally ill” (Scheff 1963, 1966, 1984). This formal labeling brings culturally shared mental illness stereotypes to the fore in others’ minds. In American culture, persons with a mental illness have persistently been viewed as unpredictable, incompetent, irrational, incomprehensible, dangerous, and unable to engage in self-care (Link et al. 1999; Martin, Pescosolido, and Tuch 2000; Pescosolido et al. 1999, 2010; Phelan et al. 2000; Wahl 1995). These beliefs are stigmatizing; specifically, mental illness is viewed as a condition that is highly devalued and discrediting (Goffman 1963). Mental illness stereotypes cause other people to perceive and treat the labeled individual with suspicion, rejection, insults, hostility, and withholding of social and employment opportunities (Scheff 1963, 1984). According to Scheff, faced with others’ bias and discrimination, labeled individuals experience greater upset and will eventually accept the deviant identity of “mentally ill person” or “mental patient,” continuing to perform this deviant role because it now defines who they are. Scheff (1963:451, emphasis in the original and later publications) concluded, “labeling is the single most important cause of careers of residual deviance [mental illness].”¹ In short, formal categorization with a stigmatizing label sets an interpersonal process in play that results in chronic disorder and, by implication, further emotional distress and disadvantaged life circumstances. For reexamination here with a more apt measure of formal labeling, Hypothesis 1 is that individuals who have been formally labeled will report greater negative affect and higher mental health disability days in the past month, the two well-being outcomes employed in this study.

A Psychiatric Critique and Alternative Argument: Gove

Walter Gove initiated an intense debate with Scheff by taking a psychiatric perspective, criticizing labeling theory, faulting the studies that Scheff viewed as supportive (Scheff 1963, 1966, 1974), and marshaling evidence substantiating a psychiatric perspective (Gove 1970a, 1970b, 1975, 1980, 1982). Gove’s argument was straightforward and consistent: Labeling is not the cause of individuals’ continued mental health problems; having a severe illness is. The stigma and stereotypes attached to mental illness labels do not produce differential treatment by other people; instead, ill persons’ severe symptoms do. The more unusual or unpredictable individuals’ feelings or behaviors seem, the more likely other people will reject and discriminate against them. For Gove, serious psychiatric symptoms were the primary cause of all else that might follow: treatment entry, official labeling, stigma and discrimination, and chronic mental health problems. In essence, Gove was insisting that any relationship found between a formal label and consequent mental health problems was spurious, easily explained away by the severity or disruptiveness of a person’s symptoms.

This point is critically important and has gone untested in decades of labeling research (see reviews in Gove 1970a, 1970b, 1975, 1980, 1982; Link and Phelan 1999a, 1999b, 2010, 2013; Scheff 1963, 1966, 1974 1984). It pits a medical explanation of mental illness and its consequences directly against a social explanation of illness and its consequences. Furthermore, the argument can be broadened beyond the labeling–well-being relationship. From a psychiatric perspective, persons’ problematic symptoms occur prior to their treatment entry, formal diagnosis, and stigma experiences, with serious symptoms as the primary reason for all of these events’ occurrences. Thus, all of these events’ associations with well-being should be spurious. This broader implication can be summarized as Hypothesis 2, that the seriousness of individuals’ preexisting symptoms will explain away relationships of treatment involvement, formal labeling, and perceived stigma with current well-being (i.e., negative affect and mental health disability days in the past month). If confirmed, Hypothesis 2 would undermine evidence for labeling theory’s basic premise, stated in Hypothesis 1.

Modified Labeling Theory: Link

As originally formulated, labeling theory viewed deviant individuals as victims of bias and discriminatory behaviors by other people, triggered by an authority figure’s stigmatizing label (Becker 1963; Scheff 1963). Norm breakers were essentially forced or hemmed into a permanent deviant role by others’ negative attitudes and actions. In the late 1980s, Bruce Link (1987; Link et al. 1989) proposed “modified labeling theory,” a revised process that shifted research attention away from other people’s discriminatory behaviors to labeled persons’ own fears of stigma instead—an intrapersonal rather than interpersonal process.

Link argued that when a person has been officially labeled mentally ill, widely shared cultural stereotypes of mental patients suddenly become personally relevant and threatening. The labeled individual becomes concerned about possible devaluation and rejection by others. These fears can have direct and indirect effects on well-being. Because of the stigma attached to the mental illness label, individuals can feel different and ashamed, leading directly to lower self-esteem, heightened distress, lessened quality of life, and greater vulnerability to subsequent episodes of disorder (Link 1987; Link et al. 1989). Both experimental and survey studies confirmed that among patients, perceived stigma was linked to such negative consequences (Link and Phelan 1999a, 1999b, 2010, 2013). Additionally, Link argued, fear of stigma can operate indirectly through labeled individuals’ attempts to protect themselves from anticipated rejection. Protective coping strategies included keeping illness a secret, withdrawing from social interaction, educating others about mental illness, and challenging prejudice (Link 1987; Link et al. 1989; Thoits and Link 2016). Studies showed that patients’ most frequent coping strategies, especially secrecy and withdrawal, produced lower self-esteem, greater social isolation, lower employment, and higher levels of depression and distress (Link et al. 1989, 1997, 2001; Link, Mirotznik, and Cullen 1991; Thoits and Link 2016). However, in this research, formal labeling was most often indicated by patients’ treatment participation, not by their receipt of an official label. To reexamine the central claim of Link’s modified theory with a more suitable labeling measure, Hypothesis 3a is that individuals who have received a formal label and fear stigma will report greater negative affect and mental health disability days in the past month.

An Integrative Approach: Rosenfield

Link’s modified labeling theory and accompanying studies moved attention in the literature away from the original “labeling versus serious symptoms” debate between Scheff and Gove to a focus on stigma as the primary cause of individuals’ difficulties and poor well-being. For Link, acquiring a formal label was necessary to activate concerns about stigma, but fear of stigma—not the label itself—was the cause of negative outcomes.

Rosenfield (1997) noted the incompatibility between Gove’s psychiatric perspective and Link’s argument. Gove (1970a, 1980, 1982) had insisted that serious symptoms were the primary determinant of individuals’ eventual well-being, dismissing the relevance of stigma fears. He contended that symptoms lead to official labeling, labeling leads to treatment entry, and treatment is typically effective in restoring positive well-being. Rosenfield observed that Link stressed the harmful impacts of stigma fears due to formal labeling, whereas Gove emphasized the benefits gained from treatment received due to formal labeling.

Rosenfield (1997) proposed and tested a theoretical integration of these conflicting arguments. She argued that both perspectives were likely valid given that both had bodies of supportive evidence. She hypothesized that becoming formally labeled as mentally ill simultaneously provides access to helpful therapeutic services and creates potential exposure to damaging stigma. These positive and negative consequences of formal labeling should have opposite, counterbalancing effects on individuals’ subsequent well-being, assessed in her study as subjective satisfaction with life. Rosenfield examined these competing effects by first holding formal labeling and serious symptoms constant: All respondents in her study were in a model psychiatric rehabilitation program and had a serious chronic disorder. She then measured, as distinct constructs, the number of therapeutic program services that participants used (services received) and their perceptions of stigma. Received services and perceived stigma were positively and negatively related to participants’ ratings of their life satisfaction, respectively, consistent with Rosenfield’s integrative argument (see also Link, Castille, and Stuber 2008; Link et al. 1997; Link and Phelan 2013).

Importantly, Rosenfield’s study did not examine the effects of formal labeling per se. Her formal labeling indicator was a constant: being a patient in a model treatment program. Because of this, the effect of labeling on services received, perceived stigma, and life satisfaction could not be assessed. Instead, clients’degree of treatment involvement was measured with the number of rehabilitation services they used. The study fell short of addressing the heart of Rosenfield’s integrative insight, that labeling leads to stigma concerns (the modified labeling view) and to helpful treatment (the psychiatric view). The analyses for Hypotheses 2 and 3a will allow a reexamination of these relationships.

An Alternative Modified Labeling Theory: Gove

Near his retirement, Gove (2004) revisited the labeling versus psychiatric debate, offering his own integration of the two perspectives, viewed here as an alternative modified labeling theory. Elaborating his earlier psychiatric stance, Gove argued that most people who experience symptoms of disorder have acute transitory conditions, see themselves as having a stress-induced “nervous breakdown,” and seek usually effective mental health services for help with their problems. Thus, they do not continue on to a career of recurrent disorder and lessened life chances. However, he proposed, the process is different for persons with serious conditions. Such persons are more likely to be hospitalized, thus officially labeled as mentally ill by professionals. This label intensifies the negative well-being consequences of having severe symptoms. Specifically, formal labeling heightens the likelihood of seriously ill persons experiencing discrimination and a general sense of failure, which in turn augment their ongoing distress or disorder (Gove 2004:370).² By implication, serious symptoms coupled with formal labeling should be especially harmful to well-being.

Link’s and Gove’s modified labeling theories therefore differed regarding the combination of circumstances that should raise the risk of poor well-being. For Gove, serious symptoms that have been officially labeled place individuals at higher risk; for Link, official labeling accompanied by fear of stigma raises risk. Although Gove (2004:370) claimed his argument was more “complete and parsimonious” than Link’s, he acknowledged that stigma fears could produce problematic consequences too. Given this, both Link’s and Gove’s modified theories might find empirical support. Gove’s alternative modified labeling hypothesis becomes Hypothesis 3b, that individuals with serious symptoms who have been formally labeled will report greater negative affect and mental health disability days in the past month. (Scheff’s, Gove’s, and Link’s intervening explanatory variables—discrimination experiences, sense of failure, ineffective coping strategies—cannot be examined due to a lack of available measures in the data used here.)

The Problem: Measures of Formal Labeling

As noted earlier, these hypotheses have not been tested properly because of an unnoticed, persistent measurement problem in the labeling literature. Formal labeling has often been measured as persons’ involvement in mental health treatment.³ Older studies mainly used mental hospitalization as their formal labeling indicator (see reviews in Gove 1975, 1980, 1982; Scheff 1974, 1984); later studies included participation in outpatient care (Link et al. 1989; Rosenfield 1997; Thoits 2016). Most researchers presumed that involvement in treatment conveys both to patients and other people that patients have been professionally categorized as ill. Sometimes this presumption was explicit. Gove (2004:370) repeatedly equated hospitalization with formal labeling, stating, for example, that discrimination is especially harmful if persons “have already been labeled by being hospitalized” (emphasis added). Similarly, Link and colleagues (1989:404) clarified the status of subgroups in their classic modified labeling study: “[U]ntreated cases and ‘well’ respondents have not been labeled,” whereas current patients and former patients “have been labeled” (emphasis added).

There are two problems with equating formal labeling with being in treatment. As mentioned before, the most important problem is that treatment participation carries two possible meanings: formal categorization with a stigmatizing label (from a labeling perspective) and access to professional therapeutic assistance (from a psychiatric perspective). It is obviously inadvisable for one measure to convey two different meanings. The term formal labeling should mean receipt of an actual label from an authority figure, namely, an explicit classification of the person as mentally ill or as having a specific condition such as panic disorder. Treatment involvement should denote a person’s receipt of mental health services such as hospitalization, psychotherapy/counseling, or doctor-prescribed psychotropic medications. To date, formal categorization and treatment participation have not been separated operationally, leaving labeling theory’s foundational premise unexamined: A formal or official label is damaging to an individual’s subsequent well-being.

A second problem is created when labeling is not disentangled conceptually and operationally from treatment involvement. Persons in treatment are convenient to sample, so they have been frequently studied. A patient sample means that formal labeling is held constant by study design, so the effects of having versus not having a formal label cannot be assessed (as in Rosenfield’s study). Furthermore, because having a formal label is a constant, one cannot examine a label’s effects on individuals who do not actually have a mental disorder. This leaves a major implication of labeling theory untested: Formal labels will have negative consequences even when those labels are untrue (Becker 1963). Labeling theories essentially describe self-fulfilling prophecy effects, namely, what patients or other people believe to be true will become true in its consequences. In essence, whether or not persons truly have a clinical disorder does not matter. What matters is that someone with professional authority says they do. It follows that among persons who are in fact clinically well (meaning they do not meet diagnostic criteria for a psychiatric disorder), those who have acquired a formal label should exhibit poorer well-being compared to others who have not. To see whether the central labeling proposition generalizes to clinically well persons who have acquired a mental illness label, separate measures of treatment involvement and official labeling again are necessary.

How might persons with no current clinical disorder acquire a formal label? Some people who are clinically well have sought professional help for personal problems and obtained a label during treatment—perhaps a misdiagnosis, perhaps a diagnosis supplied by a provider for insurance purposes only. Others who are clinically well have had a disorder in the past, entered treatment, and were formally categorized then. If traditional and modified labeling theories generalize to people who have acquired a misdiagnosis or an earlier label that no longer applies because of recovery or effective symptom management, then support should be found for Hypothesis 4: Among individuals who are clinically well, persons who have been formally labeled will report greater negative affect and mental health disability days in the past month than those who have not been labeled.

These observations raise the question of how formal labeling can be disentangled operationally from treatment involvement. In this study, formal labeling is indicated by respondents’ knowledge of their diagnosis (e.g., “I have major depression”). There are two rationales for this measure, other than its availability in the data. First, both traditional and modified labeling theories view receiving a formal label as an event activating all else that follows. Persons must learn how an authority has categorized their symptoms for the label to have consequences—it is the authority figure’s naming that carries weight. Second, knowing one’s diagnosis is correlated with treatment experience but not perfectly so. Providers differ considerably in their willingness to supply this information to their patients (e.g., Clafferty, McCabe, and Brown 2001; McDonald-Scott, Machizawa, and Satoh 1992; Sisti et al. 2016). Patients who have been in treatment may or may not actually know the name of their condition. Due to widespread and persisting cultural devaluation of mental illness (Pescosolido et al. 2010; Phelan et al. 2000), patients who are aware of their diagnosis should be at risk of poorer well-being regardless of whether they eventually internalize the label as a personal identity or not. Conversely, when unaware of their diagnosis, patients can see themselves as simply stressed or having a nervous breakdown (Gove 2004)—not mentally ill—with little adverse effect on their well-being (Thoits 2016; Thoits and Link 2016). Treatment involvement and knowledge of one’s diagnosis are thus distinguishable empirically. This allows a more exacting test of labeling theory’s primary claim that a formal label leads to persisting or recurring mental health problems and, by implication, poorer life quality and well-being (Scheff 1963, 1984).

This study takes advantage of the National Comorbidity Survey-Replication (NCS-R; Kessler et al. 2004), which provides a nationally representative sample of adult respondents who have and have not had a diagnosable disorder in their lifetimes either with or without having had mental health treatment. These data enable separation of labeling from treatment and tests of the four hypotheses described previously.

Methods

Sample

The NCS-R, fielded between 2001 and 2003, drew a multistage clustered area probability sample of households of noninstitutionalized English-speaking adults residing in the continental United States.⁴ The participation rate was 71 percent (Kessler et al. 2004). Its purpose was to reexamine the prevalence of mental disorders and co-occurring disorders in the population. A total of 9,282 respondents were interviewed in person with Part 1 of the NCS-R survey. Part 1 assessed whether and when respondents had one or more mental disorders in their lifetimes. A subsample (N = 5,692) of all respondents with at least one lifetime disorder and a random subset of respondents with no lifetime disorder were interviewed with Part 2 of the survey. It assessed risk factors for mental health problems and respondents’ use of mental health services. Data from the Part 2 sample are employed here.

Measures

Clinical disorder

Because this study examined whether formal labeling influences well-being for individuals who do and do not have a clinical disorder, respondents were divided into two groups on this basis.

The NCS-R interview assessed all core mental disorders measured by the World Health Organization-Composite International Diagnostic Interview (Kessler and Ustun 2004). This was based on symptom criteria outlined in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV; American Psychiatric Association 1994). Dichotomous variables indicated the occurrence of each disorder during the respondent’s lifetime and in the past 12 months.

For this study, a variable indicating had any disorder in the past 12 months was constructed (1 = yes, 0 = no) on the basis of 15 conditions: bipolar I and II, major depressive disorder, dysthymia, panic disorder, agoraphobia, social phobia, simple phobia, generalized anxiety disorder, posttraumatic stress disorder, eating disorder, alcohol abuse and dependence, and drug abuse and dependence.⁵

Exploratory analyses found that 94 percent of respondents with a current 12-month disorder had an initial onset of that particular disorder more than a year ago. Thus, almost all respondents with a 12-month disorder had either a chronic or relapsing condition, here termed recurring 12-month disorders. For a crisp comparison between respondents with a recurring 12-month disorder and those with no 12-month disorder, 86 participants with new onsets were omitted. Separate analyses were conducted for these two groups. After listwise deletion for missing variables (N = 179), 1,255 respondents had a recurring 12-month disorder, and 4,172 respondents had no 12-month disorder (i.e., they were clinically well). Among the clinically well, 36 percent had been in treatment at some point in their lives, and 15 percent had treatment during the past year, so acquiring a formal label was a viable possibility in the no-disorder group.

Symptom seriousness

Gove contended that the severity of individuals’ preexisting symptoms accounted for their current well-being. Severity or symptom seriousness refers to the degree to which a condition impairs an individual’s daily functioning, measured here with the K6 scale. The K6 was developed to discriminate efficiently in population surveys between individuals who do and do not have a severe mental illness, defined by its developers as having any DSM-IV disorder accompanied by significant impairments in role performance (Kessler et al. 2003).

The K6 scale instructs respondents, “Think of the one month in the past 12 months when you were at your worst emotionally in terms of being anxious, depressed, or emotionally stressed. If there was no month like this, think of a typical month in the past 12 months.” Six items followed: “How often did you feel: nervous, hopeless, restless or fidgety, so depressed that nothing could cheer you up, that everything was an effort, worthless?” Responses were 0 = none of the time to 4 = all of the time. Summed responses ranged from 0 to 24. For persons with a recurring disorder, Cronbach’s alpha was .87; for those with no disorder, Cronbach’s alpha was .77.

Treatment experience

Individuals may have had one or more types of treatment involvement in the past 12 months: mental hospitalization, psychotherapy, or taking a prescribed psychoactive medication. Questions about these types of service use were always tied to the qualifier “for problems with your emotions or nerves or your use of alcohol or drugs.” Services included admittance for an overnight stay in a hospital, going to a hospital emergency room, seeing any of a list of professionals, and taking any of a list of prescription medications under a doctor’s supervision. The timing of respondents’ first and most recent treatment experiences was also queried. Respondents were coded as had treatment experience in the past 12 months (1 = yes, 0 = no) if they were hospitalized, saw a professional, and/or took prescribed psychotropic drugs in the past year for their emotions, nerves, or alcohol/drug use.

This measure includes more types of mental health care than were examined in the earliest labeling studies, which focused almost exclusively on mental hospitalization. However, mental hospitalization has become rare over the decades, and treatment has shifted to outpatient sites even for very serious conditions (Gove 2004). Because hospitalization is indeed rare in this analytical sample (7.8 percent had ever been hospitalized for mental health problems, only .9 percent in the past 12 months), treatment experience is indicated by the use of a broader range of mental health services.

Formal labeling

Formal labeling is measured by respondents’ knowledge of their diagnosis. Early in the NCS-R interview, all respondents were asked, “Do you have any of the following conditions?” Last on a list of health conditions (e.g., blindness, learning disability, physical disability) was, “Any emotional or mental disability?” (1 = yes, 0 = no), followed by, “What is that condition?” Ninety-nine percent of respondents who acknowledged this disability (N = 353) supplied a diagnosis. Formal labeling is coded 1 = respondent reports a diagnosis, 0 = otherwise. The vast majority reporting a diagnosis (96 percent) had also been in treatment at some point in their lives, so their diagnoses very likely came from a provider.

There are drawbacks to this formal labeling indicator. First, it probably undercounts the number of respondents who know their diagnosis. Although the phrase “emotional or mental disability” was in line with preceding questions about learning and physical disabilities, the term disability implies serious and/or permanent impairment. Individuals who knew they had a mild or moderate disorder likely said they did not have such a disability and were coded as unlabeled. Respondents with less serious symptoms (K6 scores <13) were in fact significantly less likely to report a formal diagnostic label than respondents with more serious symptoms—4.6 percent and 36.2 percent, respectively.

Second, from a psychiatric perspective, people with a disorder who said they did not have an emotional/mental disability may have been in denial or lacking insight about the state of their mental health. However, 75 percent of respondents who had a recurring disorder but said they did not have an emotional/mental disability recognized they had psychological problems: On questions asked later in the interview, they agreed they had felt a need to see a mental health professional in the past year because of emotional or substance use problems or said they had such problems but preferred handling them on their own. Thus, a large majority of people who had a disorder by clinical criteria but who did not acknowledge having an emotional/mental disability were not in denial about or unaware of their problems.

Third, from a labeling perspective, respondents who said they did not have an emotional/mental disability might have been giving a socially desirable answer by withholding information about their mental health. However, on a 10-item scale of socially desirable responding included in the survey (Zuckerman et al. 1993), the social desirability scores of participants who said they did not have this disability were equivalent to those who said that they did.

Thus, the formal labeling variable is a conservative indicator of respondents’ knowledge of their official diagnosis, but respondents’ seeming lack of knowledge cannot easily be attributed to denial, lack of awareness, or social desirability.

Stigma concerns

No conventional measures of mental illness stigma (Link et al. 1999; Livingston and Boyd 2010) were available in the NCS-R data. One question in the interview tapped the stigma that individuals associate with seeking help for mental health problems. This question was asked of all survey participants, both persons who did and did not have a disorder or treatment experience. Respondents were asked, “How embarrassed would you be if your friends knew you were getting professional help for an emotional problem?” (1 = not at all embarrassed, 4 = very embarrassed). Possible embarrassment about treatment suggests that individuals fear stigma from others or personally view treatment for emotional problems as stigmatizing.

Well-being

Two measures of psychological well-being served as dependent variables. The first is negative affect, reverse-coded from a positive affect scale to align with the direction of the second measure, number of mental health disability days in the past month.

Positive affect was assessed with four questions: “During the past 30 days, how often did you feel: confident, optimistic, happy, full of life?” Responses were 5 = all of the time to 1 = none of the time. Responses were back-coded, summed, and divided by the number of answered items, so scores for negative affect ranged from 5 (high) to 1 (low). Cronbach’s alphas were .82 for the recurring-disorder group and .75 for the no-disorder group.

Mental health disability days was constructed from three questions about respondents’ health during the past month: “Beginning yesterday and going back 30 days, how many days out of the past 30 were you totally unable to work or carry our your normal activities?”; “ . . . were you able to work and carry out your normal activities, but had to cut down on what you did or not get as much done as usual?”; and “ . . . did it take an extreme effort to perform up to your usual level at work or at your other normal daily activities?” The number of these days that were “due to your emotions, nerves, mental health, or your use of alcohol or drugs” was ascertained. Mental health disability days was coded as the largest number of days cited for experiencing total inability, having to cut down, or exerting extreme effort, ranging from 0 to 30 days.

Note that negative affect and mental health disability days were reported for the past month, in contrast to variables for recurrent disorder, serious symptoms, and treatment experience, which were reported for the past year. These differing time frames allow disorder, symptoms, and treatment experiences to be viewed as possible antecedents to the well-being dependent variables.

Control variables

Female is coded 1, male 0. Age and age squared are measured in years (age squared is included because age and symptoms have a curvilinear relationship). Race/ethnicity is categorized as black (0, 1), Hispanic (0, 1), and other race (0, 1), with non-Hispanic whites serving as the omitted comparison group. Education is measured in years of schooling, top-coded at 17 years or more. Annual household income is in thousands of dollars divided by 1,000, with households earning $200,000 or more top-coded as 200. Having private or public health insurance was coded no = 0, yes = 1, to control for a major determinant of treatment access.

Analysis Strategy

In all analyses, data were weighted to adjust for biases due to selection within households, nonresponses, differences between the sample and the 2002 Current Population Survey, and overrepresentation of the Part 1 respondents in Part 2 of the survey (Alegria et al. 2008). To correct variances and standard errors for complex survey design effects, the survey procedures for regression in Stata/SE 10.0 (StataCorp 2007) and the Taylor series linearization method for variance estimation were used.

Results

Table 1 summarizes all measures for respondents with a recurring 12-month disorder and for those with no 12-month disorder. Not surprisingly, the recurring-disorder group has significantly more negative affect, mental health disability days, serious symptoms, treatment experience, and knowledge of their diagnosis (formal labeling) than the no-disorder group. Unexpectedly, the two groups do not differ on stigma concerns (possible embarrassment). Respondents with a recurring disorder are more likely to be female, less likely to be black, are younger, lower in family income, and less likely to have health insurance, consistent with the epidemiological literature.

Table 1.

Descriptive Statistics for Respondents with and without a Recurring 12-Month Disorder.

	Recurring 12-Month Disorder(n = 1,255)	No 12-Month Disorder(n = 4,172)	p
Negative affect (1–5)	2.9 (.8)	2.3 (.7)	.001
Mental health disability days (0–30)	2.8 (7.0)	.3 (2.0)	.001
Symptom seriousness (0–24)	7.6 (5.6)	2.7 (3.2)	.001
Treatment experience (0, 1)	45.8%	14.8%	.001
Formal labeling (0, 1)	19.6%	2.5%	.001
Possible embarrassment (1–4)	2.0 (1.0)	2.0 (1.0)	ns
Female (0, 1)	60.6%	50.6%	.001
White (0, 1)	73.1%	73.2%	ns
Black (0, 1)	10.4%	12.7%	.05
Hispanic (0, 1)	11.8%	10.6%	ns
Other race (0, 1)	4.7%	3.6%	ns
Age, in years	40.2 (15.1)	46.6 (18.3)	.001
Education, in years	13.0 (2.3)	13.1 (2.6)	ns
Family income ($0–$200,000)	54.4 (47.4)	61.3 (49.4)	.001
Health insurance (0, 1)	82.2%	88.4%	.001

Note. Percentages, means, and standard deviations are based on the weighted data.

Hypothesis 1

Hypothesis 1 examines the central proposition of labeling theory: Having a formal diagnostic label should be associated with more past-month negative affect and days disrupted by emotional or substance use problems. Respondents’ negative affect scores and mental health disability days were regressed on formal labeling (knowledge of their diagnosis), with background variables controlled. Results are reported in Table 2 for respondents in the two groups.

Table 2.

Regressions of Well-Being on Formal Labeling by Respondents’ 12-Month Mental Health Status.

	Recurring 12-Month Disorder (n = 1,255)	No 12-Month Disorder (n = 4,172)
	Negative Affect	Negative Affect
Formal labeling (0, 1)	.58***	.42***
Constant	2.61***	2.40***
	Mental Health Disability Days	Mental Health Disability Days
Formal labeling (0, 1)	4.97***	2.63***
Constant	−.68	.00

Note. Each equation controls for female, black, Hispanic, other race, age, age squared, education, family income, and possession of health insurance. Unstandardized regression coefficients are reported.

***

p < .001.

Clearly, having a formal label is significantly related to poorer well-being in both groups. This supports labeling theory’s central proposition as well as the generalizability of this proposition to persons who have no current clinical disorder, thus also supporting Hypothesis 4.

Hypothesis 2

Hypothesis 2 examines the psychiatric perspective, which expects serious preexisting symptoms to explain away the labeling–well-being relationship. It also predicts spurious associations of treatment experience and of stigma concerns with well-being. From this point of view, respondents who have been in treatment, acquired a formal label, and have fears of stigma have poorer well-being because they have more severe preexisting symptoms. To test this hypothesis in both groups, negative affect and mental health disability days were first regressed on treatment experience, formal labeling, and stigma concerns, with background variables controlled. Then symptom seriousness was added to the equations in a second step. Results are shown in Table 3. If significant coefficients found in the first models become near zero in size and nonsignificant in the second models, spuriousness would be indicated.

Table 3.

Regressions of Well-Being on Treatment Experience, Formal Labeling, Stigma Concerns, and Symptom Seriousness by Respondents’ 12-Month Mental Health Status.

	Recurring 12-Month Disorder (n = 1,255)		No 12-Month Disorder (n = 4,172)
	Negative Affect		Negative Affect
	(1)	(2)	(1)	(2)
Treatment experience (0, 1)	.20***	.03	.24***	.06*
Formal labeling (0, 1)	.50***	.30***	.31*	.17+
Possible embarrassment	.02	−.00	.09***	.08***
Symptom seriousness	—	.06***	—	.08***
Constant	2.56***	2.05***	2.24***	1.94***
R²	.11	.25	.06	.19
	Mental Health Disability Days		Mental Health Disability Days
	(1)	(2)	(1)	(2)
Treatment experience (0, 1)	1.28***	.12	.54***	.30**
Formal labeling (0, 1)	4.43***	3.06***	2.34**	2.15**
Possible embarrassment	−.09	−.19	−.06*	−.08**
Symptom seriousness	—	.43***	—	.11***
Constant	−.57	−4.13**	.15	−.26
R²	.12	.21	.06	.08

Note. Each equation controls for female, black, Hispanic, other race, age, age squared, education, family income, and having health insurance. Unstandardized regression coefficients are reported.

p < .10. *p < .05. **p < .01. ***p < .001.

Considering the effects of formal labeling first, each Model 2 shows that the associations of formal labeling with poorer well-being are not spurious in either group. Serious symptoms of disorder only partially account for these relationships, explaining 31 percent to 45 percent of the strength of the labeling–well-being associations (excepting the labeling–disability days link in the no-disorder group where symptoms explain little of that association). In short, having a formal mental illness label is associated with harm to individuals’ well-being independent of the seriousness of their symptoms, again supporting labeling theory’s central proposition (Hypothesis 1) and its generalizability to persons without a current disorder (Hypothesis 4). With respect to there being a spurious relationship between formal labeling and well-being, the psychiatric prediction (Hypothesis 2) is disconfirmed.

Examining treatment effects in Table 3, it is clear for individuals with a recurrent disorder that the relationships of treatment experience with poorer well-being are fully explained by the severity of respondents’ preexisting symptoms. This supports the psychiatric perspective. For persons with no disorder, these treatment–well-being relationships are noticeably reduced in size (by 75 percent for negative affect and 44 percent for disability days) but remain significant, inconsistent with the psychiatric view.

Concerning the effects of stigma concerns on well-being, there are three noticeable patterns. First, possible embarrassment about seeking treatment is significantly related to well-being only in the no-disorder group. Second, in this group, possible embarrassment is linked to greater negative affect but—unexpectedly—fewer mental health disability days.⁶ Third, the size and significance of the stigma coefficients were not altered appreciably when individuals’ symptom levels were controlled in the no-disorder group’s Model 2. Thus, these relationships are not spurious, inconsistent with the psychiatric perspective.

In sum, the psychiatric perspective, which predicted spurious relationships caused by individuals’ preexisting serious symptoms, received mixed support. Serious symptoms did not explain away the link between formal labeling and lower well-being in either group and did not account for the association between stigma concerns and well-being in the no-disorder group. Prior symptoms explained away only the relationship of treatment involvement with lower well-being in the recurrent-disorder group, the sole finding supporting a psychiatric argument.

Hypothesis 3

Hypothesis 3 focuses on two plausible theoretical possibilities: (a) Formal labeling may intensify the negative well-being impacts of individuals’ stigma concerns (according to Link’s modified labeling theory) and/or (b) labeling may exacerbate the harmful well-being effects of having severe symptoms (according to Gove’s alternative modified labeling argument). For these hypotheses, interaction terms were constructed and added to the Table 3 equations: Formal Labeling × Possible Embarrassment and Formal Labeling × Serious Symptoms. Interactions were entered singly in separate equations to avoid multicollinearity among terms. The interaction coefficients appear in Table 4. If both Link’s and Gove’s arguments are valid, all coefficients should be positive and significant.

Table 4.

Interactions of Formal Labeling with Stigma Concerns (“Link Interactions”) and with Serious Symptoms (“Gove Interactions”) by Respondents’ 12-Month Mental Health Status.

	Recurring 12-Month Disorder (n = 1,255)	No 12-Month Disorder (n = 4,172)
Link interactions
	Negative Affect	Negative Affect
1. Formal Labeling × Embarrassment	.08+	−.16
	Mental Health Disability Days	Mental Health Disability Days
2. Formal Labeling × Embarrassment	.54	−1.34**
Gove interactions
	Negative Affect	Negative Affect
3. Formal Labeling × Serious Symptoms	−.00	.00
	Mental Health Disability Days	Mental Health Disability Days
4. Formal Labeling × Serious Symptoms	.43***	.30+

Note. Each interaction term was added singly to the group’s Table 3, Model 2 equation. Unstandardized regression coefficients are reported.

p < .10. **p < .01. ***p < .001.

Examining the Link coefficients in Rows 1 and 2, only two coefficients are significant, and only one supports Link’s reasoning. For individuals with a recurrent disorder, knowing one’s formal label and anticipating embarrassment over seeking mental health treatment is associated with higher negative affect, although only marginally so (Row 1), supporting Link’s hypothesis. However, among persons with no disorder, formal labeling accompanied by anticipated embarrassment is related to significantly fewer mental health disability days (Row 2), contrary to Link’s hypothesis.

Examining the Gove interactions in Rows 3 and 4, in both analytical groups, the combination of formal labeling with serious symptoms is unrelated to negative affect (Row 3), failing to support Gove’s modified labeling argument. However, for both analytical groups (Row 4), the co-occurrence of a formal label with serious symptoms connects to more mental health disability days in the past month, supporting Gove’s hypothesis, although in the no-disorder group this coefficient is marginally significant.

In sum, Table 4 results provide only partial support for both Link’s and Gove’s modified labeling arguments. Furthermore, all significant coefficients in Table 4 added no more than .01 percent to explained variance in the well-being variables (not shown).

Because knowing one’s diagnosis most likely comes from contact with mental health professionals and because treatment involvement was the most frequent indicator of formal labeling used in prior studies, Link’s and Gove’s modified labeling hypotheses were reexamined by substituting treatment participation for the formal labeling variable used here. Interaction terms for Treatment Experience × Possible Embarrassment and Treatment Experience × Serious Symptoms were added singly to the Table 3 equations. If treatment experience were the primary source of formal labeling, the Table 4 results should be replicated in Table 5.

Table 5.

Interactions of Treatment Experience with Stigma Concerns (“Link Interactions”) and with Serious Symptoms (“Gove Interactions”) by Respondents’ 12-Month Mental Health Status.

	Recurring 12-Month Disorder (n = 1,255)	No 12-Month Disorder (n = 4,172)
Link interactions
	Negative Affect	Negative Affect
1. Treatment Experience × Embarrassment	.08*	−.11*
	Mental Health Disability Days	Mental Health Disability Days
2. Treatment Experience × Embarrassment	−.06	−.40**
Gove interactions
	Negative Affect	Negative Affect
3. Treatment Experience × Serious Symptoms	−.03***	.00
	Mental Health Disability Days	Mental Health Disability Days
4. Treatment Experience × Serious Symptoms	.17*	.14*

Note. Each interaction term was added singly to the group’s Table 3, Model 2 equation. Unstandardized regression coefficients are reported.

p < .05. **p < .01. ***p < .001.

Indeed they are, with two additional significant effects appearing in Row 1 for negative affect in the no-disorder group and Row 3 for negative affect in the recurrent-disorder group. Consistent with Link’s modified labeling theory, persons with recurring disorder who have treatment experience and who also anticipate embarrassment reported higher negative affect (Row 1). Counter to Link’s theory, no-disorder participants who have had treatment involvement and who worry about embarrassment reported lower negative affect (Row 1) as well as fewer disability days (Row 2). Thus, partial and mixed support for Link’s modified labeling hypothesis is again obtained.

Mixed support also appears in Table 5 for Gove’s argument. In the recurrent-disorder group, respondents in treatment with very serious symptoms have significantly less negative affect (Row 3), disconfirming Gove’s hypothesis, but significantly more disability days (Row 4), confirming his hypothesis. In the no-disorder group, the combination of treatment experience with serious symptoms is related to more mental health disability days (Row 4), affirming Gove’s argument.

In short, even when treatment involvement is substituted as a measure of formal labeling, neither Link’s nor Gove’s modified labeling hypotheses obtained consistent support. Significant coefficients in Table 5 added no more than .01 percent to explained variance in the well-being variables (not shown).

Discussion and Conclusions

Some themes in contemporary Mad Studies literature echo issues that were raised in past decades by traditional and modified labeling theorists. In particular, Mad Studies scholars emphasize consumers’/survivors’ critiques of the medical model of mental disorder and highlight consumers’ concerns about the labeling and stigmatization they face (Beresford et al. 2010). Consistent with the views of survivors, traditional and modified labeling studies were grounded in a social model of mental health problems and were intended to confirm the harmful effects of psychiatric categorizing on individuals’ lives. But past studies did not properly assess labeling theories’ propositions because investigators routinely used an ambiguous measure, treatment involvement, to signify formal labeling. This study employed a more apt measure of having been formally labeled—individuals’ knowledge of their diagnosis—to help disentangle the influences of having an official label from treatment involvement itself.

Of particular relevance to scholars in Mad Studies, findings showed that having a formal label was indeed related to poorer well-being, specifically to greater past-month negative affect and mental health disability days. This was true both for individuals with a recurring disorder and those with no current disorder. These findings support the foundational proposition of labeling theory: Stigmatizing labels do matter for a person’s well-being regardless of whether that person meets criteria for a clinical disorder or not. Results further showed that the formal label–lower well-being association was not a spurious product of individuals’ preexisting severe symptoms, undermining a psychiatric explanation for that relationship, even though, not surprisingly, serious symptoms are inversely related to current well-being. The results clearly support the utility of a social/labeling model of mental illness experience. The social/labeling model does not supplant the medical/psychiatric model but instead is its complement.

Link’s and Gove’s expectations that formal labeling would exacerbate the negative well-being effects of stigma concerns and serious symptoms, respectively, received only partial and mixed support. Moreover, these combinations of factors added minimal amounts to explained variance in respondents’ well-being. These findings reinforce a simpler main takeaway: Although formal labeling is certainly not the “single most important” factor influencing well-being, contrary to Scheff’s (1963, 1966, 1984) repeated theoretical claims, labels are clearly associated with lower well-being even net of treatment participation, stigma concerns, and preexisting serious symptoms.

The analyses additionally uncovered both expected and unexpected effects of treatment experience. Respondents’ involvement in treatment was consistently linked to poorer well-being when their preexisting symptoms were not held constant. Once serious symptoms were controlled, however, this relationship was found spurious among individuals with a recurring disorder. In other words, severe symptoms fully explained why these persons were recently in treatment, supporting a psychiatric view. Among individuals with no current disorder, however, treatment involvement remained associated with poorer well-being even after the seriousness of their symptoms was controlled. From a psychiatric perspective, this was unexpected because treatment should be related to benefit rather than harm. There are a number of possible reasons for this finding in the no-disorder group: Its members may have subclinical levels of distress that are linked to poorer well-being and treatment entry, recently begun therapy or psychotropic medications may not have reached their effectiveness, or treatment may carry different meanings for individuals who are clinically well compared to persons with histories of recurrent illness and more service-use experience.

The fact that formal labeling was associated with harm in both the recurrent-disorder and no-disorder groups while the well-being effects of treatment differed noticeably between the two groups underscores again the importance of keeping labeling and treatment measures distinct from one another. If researchers adopted this practice, a more elaborate process integrating both labeling and psychiatric approaches becomes testable: Serious preexisting symptoms should more often lead to treatment entry, which in turn should simultaneously raise the likelihood that persons (a) learn their formal label and (b) obtain access to professional assistance. Knowing one’s label (path a) should evoke personal concerns about stigma, followed by coping attempts to ward off possible devaluation and discrimination by other people, which may be more or less effective in forestalling these stressors. Professional assistance (path b) should lead to therapy, medications, or both, which can help resolve personal problems and reduce clinical symptoms. Because perceived stigma and discriminatory experiences more often hurt people’s self-regard, emotional stability, and life chances (Link and Phelan 2013) whereas professional services more often help to avert such outcomes (Gove 2004; Rosenfield 1997), treatment entry itself may result in mixed or no net effects on persons’ long-term well-being. Examining this process requires treatment entry, formal labeling, and the types and duration of specific received services to be distinguished operationally. In future work, formal labeling might be better ascertained by asking, “Has anyone said that you have a mental health problem?” and, if that information were conveyed by a professional, “Do you know the diagnosis?” The types and duration of services received might be obtained from official records or by requesting details from individuals themselves.

This study’s limitations must be kept in mind. First, these are cross-sectional data, so the ordering of the relationships presumed here could be reversed: Respondents’ negative affect and day-to-day functional difficulties may have motivated them to seek treatment and thereby acquire a formal label rather than treatment and a label preceding their present well-being. Fortunately, experimental and longitudinal studies support the direction of effects assumed here (Clement et al. 2015; Link and Phelan 1999a, 1999b, 2013), and the variables’ timing (past year vs. past month vs. time of interview) help too.

Second, evidence shows that persons who are presently distressed recall past episodes of mental health problems reliably; those who are not presently distressed tend to forget episodes that they had reported four years earlier in a prior interview (Aneshensel et al. 1987). Thus, respondents with positive well-being at the time of the NCS-R interview may have underreported their previous serious symptoms, failed to remember past treatment experiences, and perhaps forgotten their formal diagnoses too. Such recall biases would strengthen the relationships of symptoms, treatment, and labeling with poorer well-being found here. However, this study’s recall period was one year, a relatively short time that would lessen such effects.

Third, the formal labeling measure in this study was not ideal, relying as it does on questions about a mental health–related “disability” rather than a mental “illness,”“disorder,” or “condition.” As detailed earlier, participants’ reports of “no disability” despite actually meeting criteria for a DSM-defined disorder were not a function of denial, lack of awareness, or social desirability. Nevertheless, this formal labeling measure almost certainly undercounted persons with knowledge of their official label, lowered the chances of observing a significant relationship between labeling and well-being, and thus provided a very conservative test of the study’s hypotheses.

Fourth, whether respondents saw their diagnosed condition as a personal identity could not be ascertained with the available data. Traditional labeling theory posits that other people’s biases and stereotyping cause individuals eventually to accept their labeled condition as an identity, causing chronic distress or disorder. Identifying as “mentally ill” or, say, “agoraphobic” is likely to strengthen that label’s negative impact on a person’s well-being. But because stereotypes of people with mental illness are pervasive, simply knowing that one has a formal psychiatric diagnosis should relate to poorer well-being—as the findings indeed showed.

It is important to add that whether respondents have personal knowledge of their diagnosis or not captures only a portion of a formal label’s possible effects. Social network members who know that respondents are in treatment may believe respondents have been categorized as mentally ill by professionals. Mental illness stereotypes might then influence network members to behave toward respondents in ways that communicate devaluation and rejection. The NCS-R did not ask participants whether their network members knew about their mental health service use, so this interpersonal aspect of having a formal label is unassessed.

Fifth, the stigma measure is a single item and focuses on the shame or undesirability of being in mental health treatment rather than the stigma of having a mental illness per se. This indicator is not comparable to other commonly used stigma measures, although it has face validity as a concern about mental health service use. More established indicators were unavailable in the survey.

Finally, the NCS-R data were collected in 2001–2003. In the almost 20 years since the survey, mental health care shifted even further from hospitals to outpatient sites, psychotherapies and medications now dominate treatment modalities due to managed care, direct-to-consumer advertising of drugs encourages self-diagnosis and treatment-seeking, mental health information and support groups are accessible online, and consumer and recovery movements combat professional and public misperceptions of mental health problems, among many other changes. These have probably altered the prevalence of clinical disorders and treatment-seeking behaviors in complex ways but are unlikely to have altered the existence of a relationship between having a formal psychiatric label and lower well-being. The American public’s stigmatizing views of persons with disorder have remained stable over the past 60 years (Link and Phelan 2013; Pescosolido et al. 2010). Regardless of whether labeled individuals personally accept or reject these views, they know as members of the culture that they are at risk of encountering bias and discrimination. This is an ongoing strain that can undermine well-being, as studies of discrimination stress show for other minority group members.

Despite these limitations, several lessons can be gleaned from this study. First, the previous practice of using treatment involvement to indicate formal labeling clearly seems inadvisable. Second, formal labeling is in fact associated with poorer well-being for individuals who do as well as for those who do not have a current clinical disorder. This confirms the generalizability of the fundamental labeling proposition to individuals who are simply seeking treatment for personal troubles (the “worried well”) as well as those who have recovered from a past labeled disorder. Third, treatment experience may have very different meanings for persons with histories of chronic illness compared to those who are recovered or among the worried well. Qualitative investigation of those meanings could be enlightening. Differing meanings may also relate to the amount of stigma that consumers attach to mental health service use and might help to unravel the interplay between treatment experience and changes in stigma concerns over time that could not be examined here. Finally, because neither Link’s nor Gove’s modified labeling hypotheses obtained consistent support in this study, the simpler integration of psychiatric and labeling processes suggested here might be theoretically preferable, a possibility that invites further empirical examination.

Footnotes

ORCID iD

Peggy A. Thoits

Notes

Bio

Peggy A. Thoits is professor emerita of Sociology at Indiana University - Bloomington. Her interests are in physical and mental illness; stress, coping, and social support processes; self and identity; and emotion. Her research focuses on the determinants and mental health consequences of holding multiple role-identities, the role of social support from experientially similar others in reducing distress and ill health, and the conditions under which individuals label themselves, are labeled by others, or resist being labeled as mentally ill.

References

Alegria

Margarita

Jackson

James S.

Kessler

Ronald C.

Takeuchi

David

. 2008. “Collaborative Psychiatric Epidemiology Surveys (CPES), 2001-2003.” [United States]. Ann Arbor, MI: Inter-University Consortium for Political and Social Research [distributor], 2016-03-23. doi:10.3886/ICPSR20240.v8.

American Psychiatric Association. 1994. Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association.

Aneshensel

Carol S.

Estrada

Antonio L.

Hansell

Mary Jo

Clark

Virginia A.

1987. “Social Psychological Aspects of Reporting Behavior: Lifetime Depressive Episode Reports.” Journal of Health and Social Behavior 28(3):232–46.

Becker

Howard

. 1963. Outsiders: Studies in the Sociology of Deviance. Glencoe, NY: Free Press.

Beresford

Peter

Nettle

Mary

Perring

Rebecca

. 2010. Towards a Social Model of Madness and Distress? Exploring What Service Users Say. York, UK: Joseph Rowntree Foundation.

Clafferty

Robert A.

McCabe

Elaine

Brown

Keith W.

2001. “Conspiracy of Silence? Telling Patients with Schizophrenia their Diagnosis.” Psychiatric Bulletin 25(9):336–39.

Clement

Schauman

Graham

Maggioni

Evans-Lacko

Bezborodovs

Morgan

Rüsch

Brown

J. S. L.

Thornicroft

2015. “What Is the Impact of Mental Health-Related Stigma on Help-Seeking? A Systematic Review of Quantitative and Qualitative Studies.” Psychological Medicine 45(1):11–27.

Goffman

Erving

. 1963. Stigma: Notes on the Management of a Spoiled Identity. Englewood Cliffs, NJ: Prentice Hall.

Gove

Walter R.

1970a. “Societal Reaction as an Explanation of Mental Illness: An Evaluation.” American Sociological Review 35(5):873–84.

10.

Gove

Walter R.

1970b. “Who Is Hospitalized: A Critical Review of Some Sociological Studies of Mental Illness.” Journal of Health and Social Behavior 11(4):294–303.

11.

Gove

Walter R.

1975. “Labeling and Mental Illness: A Critique.” Pp. 35–81 in The Labelling of Deviance, edited by Gove

W. R.

New York: John Wiley and Sons.

12.

Gove

Walter R.

1980. “Postscript.” Pp. 99–109 in The Labelling of Deviance. 2nd ed., edited by Gove

W. R.

Beverly Hills, CA: Sage.

13.

Gove

Walter R.

1982. “The Current Status of the Labeling Theory of Mental Illness.” Pp. 273–300 in Deviance and Mental Illness, edited by Gove

W. R.

Beverly Hills, CA: Sage.

14.

Gove

Walter R.

2004. “The Career of the Mentally Ill: An Integration of Psychiatric, Labeling/Social Construction, and Lay Perspectives.” Journal of Health and Social Behavior 45(4):357–75.

15.

Kessler

Ronald C.

Barker

Peggy R.

Colpe

Lisa J.

Epstein

Joan F.

Gfroerer

Joseph C.

Hiripi

Eva

Howes

Mary J.

Normand

Sharon-Lise T.

Manderscheid

Ronald W.

Walters

Ellen E.

Zaslavky

Alan M.

2003. “Screening for Serious Mental Illness in the General Population.” Archives of General Psychiatry 60(2):184–89.

16.

Kessler

Ronald C.

Berglund

Patricia

Chiu

Wai Tat

Demler

Olga

Heeringa

Steven

Hiripi

Eva

Jin

Robert

Pennell

Beth-Ellen

Walters

Ellen E.

Zaslavsky

Alan

Zheng

Hui

. 2004. “The US National Comorbidity Survey Replication (NCS-R): Design and Field Procedures.” International Journal of Methods in Psychiatric Research 13(2):69–92.

17.

Kessler

Ronald C.

Ustun

T. Bedirhan

. 2004. “The World Mental Health (WMH) Survey Initiative Version of the World Health Organization (WHO) Composite International Diagnostic Interview (CIDI).” International Journal of Methods in Psychiatric Research 13(2):93–121.

18.

Kirk

Stuart A.

Gomory

Tomi

Cohen

David

. 2013. Mad Science: Psychiatric Coercion, Diagnosis, and Drugs. New Brunswick, NJ: Transaction Publishers.

19.

Krohn

Marvin D.

Akers

Ronald L.

1977. “An Alternative View of the Labeling Versus Psychiatric Perspectives on Societal Reaction to Mental Illness.” Social Forces 56(2):341–61.

20.

Edwin

Lemert

. 1967. Human Deviance, Social Problems and Social Control. Englewood Cliffs, NJ: Prentice Hall.

21.

Link

Bruce G.

1987. “Understanding Labeling Effects in the Area of Mental Disorders: An Assessment of the Effects of Expectations of Rejection.” American Sociological Review 52(1):96–112.

22.

Link

Bruce G.

Castille

Stuber

2008. “Stigma and Coercion in the Context of Outpatient Treatment for People with Mental Illnesses.” Social Science & Medicine 67(3):409–19.

23.

Link

Bruce G.

Cullen

Francis T.

Struening

Elmer

Shrout

Patrick E.

1989. “A Modified Labeling Theory Approach to Mental Disorders: An Empirical Assessment.” American Sociological Review 54(3):400–23.

24.

Link

Bruce G.

Mirotznik

Jerrold

Cullen

Francis T.

1991. “The Effectiveness of Stigma Coping Orientations: Can Negative Consequences of Mental Illness Labeling Be Avoided?” Journal of Health and Social Behavior 32(3):302–20.

25.

Link

Bruce G.

Phelan

Jo C.

1999a. “Labeling and Stigma.” Pp. 481–94 in Handbook of the Sociology of Mental Health, edited by Aneshensel

C. S.

Phelan

J. C.

New York: Kluwer Academic/Plenum.

26.

Link

Bruce G.

Phelan

Jo C.

1999b. “The Labeling Theory of Mental Disorder (II): The Consequences of Labeling.” Pp. 361–76 in A Handbook for the Study of Mental Health, edited by Horwitz

A. V.

Scheid

T. L.

Cambridge, England: Cambridge University Press.

27.

Link

Bruce G.

Phelan

Jo C.

2010. “Labeling and Stigma.” Pp. 571–87 in Handbook for the Study of Mental Health: Social Contexts, Theories, and Systems. 2nd ed., edited by Scheid

T. L.

Brown

T. N.

New York: Cambridge University Press.

28.

Link

Bruce G.

Phelan

Jo C.

2013. “Labeling and Stigma.” Pp. 525–41 in Handbook of the Sociology of Mental Health. 2nd ed., edited by Aneshensel

C. S.

Phelan

J. C.

Bierman

Dordrecht, The Netherlands: Springer.

29.

Link

Bruce G.

Phelan

Jo C.

Bresnahan

Michaeline

Stueve

Ann

Pescosolido

Bernice A.

1999. “Public Conceptions of Mental Illness: Labels, Causes, Dangerousness, and Social Distance.” American Journal of Public Health 89(9):1328–33.

30.

Link

Bruce G.

Struening

Elmer L.

Neese-Todd

Sheree

Asmussen

Sara

Phelan

Jo C.

2001. “The Consequences of Stigma for the Self-Esteem of People with Mental Illnesses.” Psychiatric Services 52(12):1621–26.

31.

Link

Bruce G.

Struening

Elmer L.

Rahav

Michael

Phelan

Jo C.

Nuttbrock

Larry

. 1997. “On Stigma and Its Consequences: Evidence from a Longitudinal Study of Men with Dual Diagnoses of Mental Illness and Substance Abuse.” Journal of Health and Social Behavior 38(2):177–90.

32.

Livingston

James D.

Boyd

Jennifer E.

2010. “Correlates and Consequences of Internalized Stigma for People Living with Mental Illness: A Systematic Review and Meta-analysis.” Social Science & Medicine 71(12):2150–61.

33.

Martin

Jack K.

Pescosolido

Bernice A.

Tuch

Steven A.

2000. “Of Fear and Loathing: The Role of ‘Disturbing Behavior,’ Labels, and Causal Attributions in Shaping Public Attitudes toward People with Mental Illness.” Journal of Health and Social Behavior 41(2):208–23.

34.

McDonald-Scott

Patricia

Machizawa

Shizuo

Satoh

Hiroyuki

. 1992. “Diagnostic Disclosure: A Tale in Two Cultures.” Psychological Medicine 22(1):147–57.

35.

Menzies

Robert

LeFrançois

Brenda A.

Reaume

Geoffrey

. (2013). “Introducing Mad Studies.” Pp. 1–22 in Mad Matters: A Critical Reader in Canadian Mad Studies, edited by LeFrançois

B. A.

Menzies

Reaume

Toronto, ON: Canadian Scholars’ Press.

36.

Pescosolido

Bernice A.

Martin

Jack K.

Long

J. Scott

Medina

Tait R.

Phelan

Link

Bruce G.

2010. “‘A Disease Like Any Other’? A Decade of Change in Public Reactions to Schizophrenia, Depression, and Alcohol Dependence.” American Journal of Psychiatry 167(11):1321–30.

37.

Pescosolido

Bernice A.

Monahan

John

Link

Bruce G.

Stueve

Ann

Kikuzawa

Saeko

. 1999. “The Public’s View of the Competence, Dangerousness, and Need for Legal Coercion of Persons with Mental Health Problems.” American Journal of Public Health 89(9):1339–45.

38.

Phelan

Jo C.

Link

Bruce G.

Stueve

Ann

Pescosolido

Bernice A.

2000. “Public Conceptions of Mental Illness in 1950 and 1996: What Is Mental Illness and Is It to Be Feared?” Journal of Health and Social Behavior 41(2):188–207.

39.

Rimke

Heidi

. 2016. “Introduction - Mental and Emotional Distress as a Social Justice Issue: Beyond Psychocentrism.” Studies in Social Justice 10(1):4–17.

40.

Rosenfield

Sarah

. 1997. “Labeling Mental Illness: The Effects of Received Services and Perceived Stigma on Life Satisfaction.” American Sociological Review 62(4):660–72.

41.

Scheff

Thomas

. 1963. “The Role of the Mentally Ill and the Dynamics of Mental Disorder: A Research Framework.” Sociometry 26(4):436–53.

42.

Scheff

Thomas

. 1966. Being Mentally Ill: A Sociological Theory. Chicago: Aldine.

43.

Scheff

Thomas J.

1974. “The Labeling Theory of Mental Illness.” American Sociological Review 39(3):444–52.

44.

Scheff

Thomas

. 1984. Being Mentally Ill: A Sociological Theory. 2nd ed. New York: Aldine de Gruyter.

45.

Sisti

Dominic

Segal

Andrea G.

Siegel

Andrew M.

Johnson

Rebecca

Gunderson

John

. 2016. “Diagnosing, Disclosing, and Documenting Borderline Personality Disorder: A Survey of Psychiatrists’ Practices.” Journal of Personality Disorders 30(6):848–56.

46.

StataCorp. 2007. Stata Statistical Software: Release 10.0. College Station, TX: Stata Corporation.

47.

Thoits

Peggy A.

2016. “‘I’m Not Mentally Ill’: Identity Deflection as a Form of Stigma Resistance.” Journal of Health and Social Behavior 57(2):135–51.

48.

Thoits

Peggy A.

Link

Bruce G.

2016. “Stigma Resistance and Well-Being among Persons in Treatment for Psychosis.” Society and Mental Health 6(1):1–20.

49.

Wahl

Otto F.

1995. Media Madness: Public Images of Mental Illness. Piscataway, NJ: Rutgers University Press.

50.

Zuckerman

Marvin

Kuhlman

D. Michael

Joireman

Jeffrey

Teta

Paul

Kraft

Michael

. 1993. “A Comparison of Three Structural Models for Personality: The Big Three, the Big Five, and the Alternative Five.” Journal of Personality and Social Psychology 65(4):757–68.