1 Garb, H. N. (1997). Race bias, social class bias, and gender bias in clinical judgment. Clinical Psychology: Science and Practice, 4(2), 99-120. (1 student) Karen Savary Zimmerman 1 st Year Doctoral Student PY 870 Tests and Measurements Department of Clinical Psychology Antioch New England Graduate School Keene, New Hampshire Studies on race bias, social class bias, and gender bias are reviewed. Previous misconceptions about the presence of social class bias and gender bias in the areas of psychodiagnosis and rating level of adjustment, the description of personality traits and psychiatric symptoms, the prediction of behavior, and treatment planning are shown not to be supported by empirical evidence. Race bias is present, but only under limited conditions. Race bias, social class bias, and gender bias are said to occur when accuracy of judgment varies as a function of client race, social class, or gender. Accuracy of judgments can vary for a number of reasons. Diagnostic criteria can be biased If they are made by using a self-report inventory that is biased Clinicians may form hypotheses based in part on stereotypical beliefs held about race, social class, or gender Judges may integrate information based on biases held.
2 Focus of these studies is on bias and the integration of information. All clinicians integrated information. In some studies clinicians gathered information. Judgements may still be biased even if integration is not biased. Diagnostic material was biased Judgments based on self-report inventory that was biased Judgments for the same task may be unbiased in the study, but biased in clinical practice. Method Studies are included in this review only if level of psychopathology was controlled for. In analogue studies, the level of psychopathology was controlled for by experimentally manipulating the reported race, social class, and gender of the subjects. In field studies, clinicians work was analyzed. Measures of client psychopathology were used to predict clinicians judgments, and then race, social class, and/or gender were used to predict the variance that was unaccounted for by the measures of psychopathology. Other measures used to determine inclusion this review: Ratings must be done by mental health professionals or graduate students in mental health field Ratings had to be made for clients or psychiatric patients, including prison inmates if there was contact with a mental health professional. Assessment information could not be atypical of that used in clinical practice Judgments had to be similar to types made in clinical practice
3 Significant differences were not reported if not verified by statistical tests. Race Bias Psychodiagnosis African-American and Hispanic (Puerto Rican) patients were less likely than White patients to be diagnosed as having a psychotic disorder and more likely to be diagnosed with schizophrenia, even when measures of psychopathology do not indicate that a diagnosis of schizophrenia is justified. In three studies described Black patients were: more likely to be diagnosed with schizophrenia than with an affective disorder misdiagnoses of schizophrenia were more common among Black and Hispanic patients than among White patients being Black was predictive of being diagnosed with schizophrenia Black patients were not more likely to be diagnosed with schizophrenia when clinicians were asked to differentiate between schizophrenia and drug-induced psychosis Race bias was not reported for other diagnostic tasks including: educably mentally retarded/mildly mentally retarded, normal, neurotic, psychotic situational disorder, psychosis, alcoholism, personality disorders, drug dependence, paranoid psychosis, or major depressive disorder. Level of Adjustment When clinicians made ratings of severity of mental disturbance or level of adjustment, the effect of race was generally not
4 statistically significant. When race was significant, White patients were rated as having a more severe mental disturbance than Black patients. Personality Traits and Psychiatric Symptoms Children Race bias was not common in the assessment of children, and when it does occur, ratings may be more positive for Black children than for White children. Ratings of intelligence, classroom behavior, personality, and social relations did not vary significantly by race for Black and White children. Ratings for minority children have not been less favorable than for White children. Adults The effect of race was not frequently significant when assessing for traits and symptoms. When there was a significance noted it just as likely to be favorable for Black clients as for White clients. White clinicians may harbor negative attitudes toward African Americans, but when faced with individual clients, their ratings of impulse control may not be affected by race. When differences did occur, though significant, the average differences were quite small. In studies with Asian-American and White American clinicians, results suggested that when one is more familiar with a client s culture, one may be less likely to infer that psychopathology and maladaptive behaviors are present. Behavioral Prediction Race of a client did not affect prediction of academic, social, adaptive, or vocational behavior. In studies involving
5 adult clients, when prognostic ratings by Black and White clinicians were pooled, prognoses were more favorable for Black clients than for White clients in 3 studies, and equally favorable in one study. Patients with a psychotic affective disorder are more likely to be misdiagnosed as having schizophrenia if they are African American rather than White. Race bias was reported when clinicians predicted the occurrence of violence in psychiatric hospitals or prisons, but was not detected when predicting suicide, compliance with medication, or occurrence of violence in the community. Black and American Indian inmates were predicted to be more violent than White inmates, but race was not significantly correlated with the occurrence of violence. Treatment Planning Children and Adolescents The effect of race was not statistically significantly in analogue studies when the task was to indicate a referral to a learning disability remedial program, advise students on choosing a college and a career path, and to decide if child abuse should be reported. in field studies results suggest that planning decisions are as appropriate or more appropriate for black children than for White children. Race of a child was a significant predictor of failure to report abuse: Cases of child abuse were less likely to be reported if the child was White. Adults Although African Americans are over represented among adults hospitalized in, and committed to, public psychiatric hospitals,
6 research indicates that clinicians are not biased against Blacks when they decide whether to hospitalize and/or commit patients. The decisions admit a person or commit some one to the hospital was not biased against Blacks at least there was no bias to treat Blacks in a more restrictive manner. Race bias has been well documented in the prescription of antipsychotic medications. In the U. S., African Americans, compared to other patients, received a significantly larger number of psychiatric medications, number of doses of antipsychotic medicine, and number of injections of antipsychotic medication. The effect of race was generally not significant in decisions about psychotherapy. Client race generally had no effect when judging: whether clients were likely to benefit from psychotherapy whether clients should be seen for short- or long-term therapy whether clients should be individual, family, couple or group therapy whether therapy should be behavioral, nondirective, or insight-oriented. Social Class Bias Psychodiagnosis and Ratings of Adjustment Mental illness occurs most frequently, and with greater severity, among the poor. several explanations for the relationship between poverty and mental illness; stressors associated with poverty may contribute occurrence of mental illness can contribute to a downward decline in SES
7 diagnoses and ratings of adjustment may be biased. Research has not established that social bias occurs when clinicians make diagnoses and ratings of levels of adjustment. In most studies published in the last 20 years, lower-class subjects have not been diagnosed as having more severe mental disorders than middle- or upper-class clients. In one study, using DSM-III criteria for drug abuse, clinicians were more likely to make the correct diagnosis when clients were described as lower SES than when clients were described as being upper class. They under detected the disorder for upper class clients. Personality Traits and Psychiatric Symptoms Social class symptoms were usually not significant in ratings of traits and symptoms were made for children, and when social class was significant, ratings were more favorable for lower social class children. When ratings were made for adults, the effect of social class was never significant when clinicians described a client s psychiatric symptoms, but was somewhat significant when describing personality traits. Behavioral Predictions Prognostic ratings were never more favorable for lower-class clients than for middle- or upper-class clients. The occurrence of social class bias depends in part on how clients socioeconomic backgrounds are described. Ratings between a client described as a commercial artist and one described as a bulldozer operator showed no significant difference. Prognostic ratings were significantly worse for a client described as an unemployed welfare recipient with a seventh grade education.
8 Treatment Planning Children Social class effects were either not significant or, when present, favorable in the opposite direction one would expect. Referrals to remedial programs were made more often to middleclass than to lower-class children. The effect of social bias was significant in reporting cases of child abuse. Child abuse was more likely to be reported if the child was lower class. A disproportionate number of child abuse cases were not reported when families were of a higher income. Adults Social class bias did not occur when decisions were made for hospitalization. Bias did occur when making recommendations for psychotherapy. Middle-class clients were more likely to be recommended for therapy. When lower-class clients were recommended it was more likely for supportive rather than insight-oriented psychotherapy. Explanations for social class bias regarding psychotherapy: clinicians assume middle-class clients are more intelligent clinicians believe cognitive functioning varies with social class clinicians may be able to empathize more readily with middle-class clients. Gender Bias Psychodiagnosis Females are more likely to be diagnosed with histrionic personality disorder, and males are more likely to be diagnosed
9 with antisocial personality disorder, even when displaying the same symptomology. Biases were reported for both male and female clinicians. there is some speculation that females are overdiagnosed for depression. Males were more likely to be diagnosed with an organic mental disorder while females with the same symptoms were more likely to be diagnosed with a severe depressive disorder. Gender bias was not reported when diagnoses were made for other Axis I and Axis II disorders. Level of Adjustment Early research indicated that mental health professionals view males as being psychologically healthier than females. The early study by Broverman et al. was widely criticized for including more positive male stereotype traits than positive female stereotype traits. When the study was revised to contain more items that described socially desirable healthy female stereotype traits, ratings of healthy females were similar to ratings of gender-unspecified healthy adult, and healthy males were not similar to the gender-unspecified adult. In clinical judgment studies, ratings of adjustment have generally been as high for females as for males. The effect for client gender was not significant when clients were described as being: anxious or depressed either psychotic or phobic either passive or aggressive or parents of children who were mildly defiant of parents rules. The effect of gender was significant in six studies. In four, ratings of adjustment were better for female than for male
10 clients, even though clients were described by identical case histories except for the designation of gender. Politically conservative clinicians found female clients as more disturbed than male clients when both clients were described as being politically active and liberal and having significant emotional problems. In an another study, males were found to be less well adjusted when described as being passive, and females were rated as less well adjusted when described as being aggressive. Personality Assessment The effect of gender was not significant on a regular basis when rating personality traits and psychiatric symptoms. Behavioral Predictions For several prediction tasks, the effect of gender was not significant. Gender bias was not present for predicting: who would be rehospitalized within 2 years suicide compliance with treatment. Gender bias was present when predicting violence. Estimates for the likelihood of violence were higher for males than for females. False positive predictions were more frequent for males, and false negatives were more frequent for females. Treatment Decisions Children and Adolescents There is practically no evidence that gender bias occurs when clinicians make treatment decisions for children. The effect of client gender as not significant when mental health professionals read case studies and rated appropriateness of
11 hospitalization for suicidal adolescents. When deciding if child abuse should be reported, gender effect was not statistically significant. Failure to report abuse could be predicted by race, social class, and nature of abuse, but not by gender. Gender bias did not occur when children were referred to special education programs. Adults Gender bias does occur under some circumstances when making decisions about hospitalization. Gender bias was not present in analogue studies, but was present in clinical practice and varied with diagnostic category. The effect of gender was generally not significant in treatment decisions related to psychotherapy. The one exception: Clinicians who conducted intake interviews often referred clients to therapists of the same gender. Gender bias does occur in some circumstances: Males are more often diagnosed with antisocial personality disorder females are more often in individual psychotherapy males are more often in group therapy goals of balancing vocational and domestic roles are judged to be more important for females clients are more likely to be assigned to therapists of the same sex. Discussion For many important tasks related to integration and collection of information there appears to be little bias, and when it is present, seems to be in the direction that favors
12 groups that have been historically discriminated against. Limitations of the research on race bias, social class bias, and gender bias are: bias has not been studied in the areas of neuropsychological assessment and case formulation in this review, results were only presented when level of psychopathology was controlled for use of analogue studies is problematic in that clinicians could discern the purpose of the study and make socially desirable judgments. Recommendations for reducing bias include: clinicians need to sensitive to biases that have been reported attendance to diagnostic criteria would reduce bias biases can be reduced by the use of statistical prediction rules.