Introduction
Gender bias in mental healthcare has emerged as a contentious topic. Despite the guidelines set forth by ethics codes of practice (e.g. American Psychological Association, 2002; World Medical Association, 2006), mental health professionals are unavoidably susceptible to being influenced by prejudices and stereotypes during their interactions with service users. As a consequence, these biases can lead to biased clinical practices.
Previous research has predominantly focused on investigating gender bias in the diagnosis of personality disorders (McCredie & Morey, 2019), However, there has been relatively less examination of prognosis and treatment recommendations. The findings from these studies have been somewhat inconsistent. In general, the evidence has indicated that women are more prone to receiving mental health diagnoses overall, with a higher likelihood of being diagnosed with conditions such as depression, histrionic personality disorder (Garb, 1997), dependent, and borderline personality disorders (Bjorklund, 2006; Brodie, 2003), whereas men are more likely to be diagnosed with antisocial (Brodie, 2003; Samuel & Widiger, 2009), paranoid, schizoid, schizotypal, narcissistic, and compulsive personality disorders (Brodie, 2003).
Setting diagnoses aside, research indicates that mental health professionals tend to perceive men as mentally healthier but less likely to have favourable prognoses (Garb, 1997). There is also a tendency to perceive men as potentially more violent than women, despite the fact that male and female risk of violence in psychiatric contexts is nearly equal (Nicholls et al., 2004). Moreover, clinicians appear to exhibit higher levels of stigmatising behaviours, such as negative stereotyping or social restrictiveness, when interacting with individuals of the opposite sex (Henderson et al., 2014).
Experimental vignette studies offer a methodological approach to effectively investigate the impact of gender bias on clinicians’ decision-making. A vignette consists of a concise narrative depicting a hypothetical situation, which allows researchers to examine respondents’ judgments and behaviours in response to the presented scenarios (Atzmüller & Steiner, 2010; Finch, 1987; Sheringham et al., 2021). These vignettes can be presented in various formats, such as text, pictures, videos, or enacted by actors, providing flexibility and realism in the experimental design (Evans et al., 2015). By manipulating key variables while controlling others, vignette studies create a controlled yet lifelike setting, bridging the gap between conventional survey studies and real-life observations (Sheringham et al., 2021).
Case vignettes have proven valuable in experimental psychology for studying various aspects, including the influence of gender bias and aggressiveness on clinical judgment. Studies examining gender bias using vignettes have shown consistent differences in the prevalence rates of personality disorders between men and women. Female subjects are more likely to receive diagnoses of Histrionic Personality Disorder (Erickson, 2001; Samuel & Widiger, 2006) and Borderline Personality Disorder (Delphin, 2001; Giacalone, 1997; Knowles, 2009), while male subjects are more commonly diagnosed with Narcissistic Personality Disorder (Erickson, 2001; Kolker, 1994) and Antisocial Personality Disorder (Samuel & Widiger, 2009).
Findings on depression (Hartman, 2001; Kales et al., 2005; Norman, 2004; Roades, 1994) and sexual disorders (Fuss et al., 2018; Klein et al., 2019) have been mixed. Meta-analytic evidence indicates that men are perceived as more physically and verbally aggressive than women, though the difference narrows when it comes to verbal aggression (Eagly & Steffen, 1986). Accordingly, vignette studies have shown that men are perceived as more dangerous and provoke a greater desire for social distance compared to women (Fuss et al., 2018). Interestingly, when women are depicted as offenders with mental problems, they are considered more deranged and at a higher risk of criminal relapse than men (Sygel et al., 2017). Additionally, men and women exhibit different reactions to various types of aggressive behaviour. Women may respond more strongly to verbal aggressiveness than physical provocation (Bettencourt & Miller, 1996). Accordingly, vignette-based studies reveal that female clinicians perceive clients in a violent context as more dangerous and rate them higher in short-term aggression and violence risk than their male counterparts (McCredie & Morey, 2019; Rojas, 2016).
Despite the existing evidence, the interplay between perceived gender-related aggressiveness and personal values remains unexplored. Addressing this gap, our study aims to assess the impact of mental health service users’ gender-related verbal aggressiveness on mental health professionals’ decision-making process, specifically during evaluation and diagnosis. We hypothesise that clinicians’ clinical judgment will be less favourable towards service users who exhibit gender-related verbal aggressiveness compared to those displaying non-gender-related verbal aggressiveness.
Methods
Design
We conducted an experimental study, which was registered at https://dx.doi.org/10.17605/OSF.IO/25VFK and approved by the University of Barcelona Institutional Review Board [IRB: 00003099] on October 8, 2015. Participants provided informed consent prior to participating in the study. All data were anonymised, and confidentiality was maintained throughout the research process.
The independent variables (IVs) in this study were the service user's gender, which was always opposite to that of the MHP, and the type of verbal aggressiveness displayed, categorised as either gender-related or unrelated to the MHP’s gender. These variables were systematically manipulated using clinical vignettes (see extended data at https://dx.doi.org/10.17605/OSF.IO/25VFK). Participants were randomly assigned to one of two vignettes, each featuring a service user of the opposite gender. Consequently, male participants evaluated a vignette involving a female service user, while female participants evaluated one involving a male service user. Both scenarios included instances of verbal aggression, either linked to gender or not.
The Dependent Variables (DVs) included the participants’ scores on the Clinical Judgment Scale (CJS; Lee, 2012; Wallach & Strupp, 1960) and their choice of the main vignette symptomatology in a nominal four-level variable (depression, psychosis, obsessions, and mania). Additionally, we considered the following covariates related to the participants: gender, age, profession, years of experience, values (Inglehart, 2008), and scores on the Beliefs and Attitudes towards Service Users’ Rights scale (BAMHS; Eiroa-Orosa & Limiñana-Bravo, 2019).
Participants
A minimum of 44 participants per group was considered necessary to detect a statistically significant difference of 0.6 standard deviations or higher. Ultimately, 155 Mental Health Professionals (MHPs) were recruited to participate in the experiment, which took place during the enrolment process of a continuing education course focused on the mental health recovery model (Eiroa-Orosa, 2023). Further recruitment became infeasible due to COVID-19 restrictions on continuing education activities.
Participants who did not complete the entire survey (29%) were excluded from the analysis, resulting in a final sample of 110 participants (59 in the control condition and 51 in the experimental condition). The average age of the participants was 44.71 years, with 87 (79.1%) identifying as women. The majority of the participants were clinical psychologists or clinical psychology residents (n=33; 30.9%), followed by 23 (20.9%) psychiatrists or psychiatry residents, 12 (10.9%) social workers, 11 (10%) social educators, 3 (2.7%) nurses, and 1 (0.9%) occupational therapist. Additionally, 26 participants (23.6%) had other professions, mostly related to direct support. The average years of professional experience among the participants were 14.49. For further details regarding demographic characteristics, please refer to Table 1.
Table 1. Sociodemographic characteristics and baseline scores on the BAMHS scale.
| Female participants | Male participants | Statistical Significance |
|---|
| Intervention (n=41) | Control (n=46) | Statistical Significance | Intervention
(n=10) | Control (n=13) | Statistical Significance | Intervention vs.
Control | Women vs. Men |
|---|
| N | % | N | % | OR, 95%CI | p | N | % | N | % | OR,95%CI | p | OR, 95%CI | p | OR. 95%CI | p |
|---|
Higher
education* | 30 | 73.2 | 38 | 82.6 | 1.742, 0.622-4.873 | .288 | 9 | 90 | 9 | 69.2 | .250, 0.023-2.696 | .231 | 1.205, 0.487-
2.981 | 0.686 | 0.994, 0.326-
3.028 | 0.992 |
| M | SD | M | SD | t | p | d | M | SD | M | SD | z | p | d | t | p | d | t | p | d |
| Age | 44.00 | 10.29 | 44.13 | 10.82 | -0.057 | .954 | -0.01 | 45.50 | 10.88 | 48.38 | 9.55 | -0.435 | .663 | -0.28 | -0.386 | .700 | -0.07 | -1.254 | .213 | -0.29 |
Years of
experience as
MHP | 13.33 | 8.77 | 13.86 | 9.51 | -0.239 | .812 | -0.06 | 17.75 | 7.89 | 18.00 | 9.31 | -0.045 | .964 | -0.03 | -0.288 | .774 | -0.06 | -1.802 | .075 | -0.48 |
| VALUES | | | | | | | | | | | | | | | | | | | | |
Equality
between men
and women | 2.41 | 1.00 | 2.22 | 0.87 | 0.986 | .327 | 0.21 | 2.30 | 0.95 | 2.54 | 1.27 | -0.295 | .768 | -0.21 | 0.559 | .577 | 0.11 | -0.545 | .587 | -0.13 |
A kid needs a
mother to be
happy | 3.12 | 1.23 | 3.50 | 1.11 | -1.507 | .135 | -0.32 | 3.70 | 1.49 | 3.46 | 1.33 | -0.543 | .587 | 0.17 | -1.101 | .273 | -0.21 | -0.851 | .397 | -0.20 |
| BAMHS | | | | | | | | | | | | | | | | | | | | |
| Beliefs | 2.47 | 0.44 | 2.35 | 0.37 | 1.344 | .183 | 0.29 | 2.43 | 0.26 | 2.48 | 0.34 | -0.5 | .617 | -0.18 | 1.083 | .281 | 0.21 | -0.582 | .561 | -0.14 |
| Coercion | 2.50 | 0.43 | 2.33 | 0.39 | 1.9 | .061 | 0.41 | 2.56 | 0.35 | 2.35 | 0.41 | -1.327 | .185 | 0.54 | 2.272 | .025 | 0.44 | -0.385 | .701 | -0.09 |
| Paternalism | 2.43 | 0.47 | 2.21 | 0.37 | 2.366 | .020* | 0.51 | 2.10 | 0.49 | 2.51 | 0.49 | -1.86 | .063 | -0.83 | 0.973 | .333 | 0.19 | -0.162 | .872 | -0.04 |
| Discrimination | 1.90 | 0.45 | 1.84 | 0.40 | 0.65 | .517 | 0.14 | 1.68 | 0.41 | 1.79 | 0.43 | -0.447 | .655 | -0.27 | 0.32 | .749 | 0.06 | 1.324 | .188 | 0.31 |
| Total | 2.38 | 0.33 | 2.25 | 0.28 | 1.94 | .056 | 0.42 | 2.27 | 0.14 | 2.39 | 0.29 | -1.338 | .181 | -0.53 | 1.322 | .189 | 0.25 | -0.364 | .716 | -0.09 |
| CJS | | | | | | | | | | | | | | | | | | | | |
| Treatment | 3.84 | 0.59 | 3.90 | 0.73 | -0.451 | .653 | -0.10 | 3.40 | 0.45 | 4.09 | 0.67 | -2.746 | .006** | -1.18 | -1.521 | .131 | -0.29 | 0.524 | .601 | 0.12 |
| Motivation | 2.80 | 0.78 | 2.79 | 0.71 | 0.043 | .966 | 0.01 | 2.93 | 1.07 | 2.97 | 0.62 | -0.535 | .592 | -0.05 | -0.048 | .962 | -0.01 | -0.921 | .359 | -0.22 |
| Severity | 5.07 | 0.73 | 4.66 | 0.81 | 2.489 | .015* | 0.54 | 4.22 | 1.05 | 4.88 | 0.72 | -1.527 | .127 | -0.75 | 1.254 | .213 | 0.24 | 1.348 | .181 | 0.32 |
| Poor skills | 4.32 | 0.47 | 4.27 | 0.52 | 0.533 | .595 | 0.12 | 3.65 | 0.75 | 4.40 | 0.48 | -2.476 | .013* | -1.23 | -1.003 | .318 | -0.19 | 1.698 | .092 | 0.40 |
| CJS total | 4.11 | 0.41 | 4.01 | 0.55 | 1.027 | .308 | 0.22 | 3.61 | 0.42 | 4.19 | 0.37 | -2.981 | .003** | -1.49 | -0.359 | .72 | -0.07 | 1.046 | .298 | 0.25 |
Instruments
Before the course, and upon obtaining written informed consent, participants were requested to complete an online questionnaire. The questionnaire consisted of the Beliefs and Attitudes towards Service Users’ Rights scale (BAMHS), followed by sociodemographic information. Subsequently, participants were presented with the randomly assigned clinical vignette and asked a question regarding the main symptomatology of the vignette (mania, depression, obsessions, or psychosis). The Clinical Judgment Scale (CJS) and two values items adapted from the World Values Survey (Inglehart, 2008) were also included in the questionnaire.
Sociodemographic questionnaire
Participants were requested to complete a concise demographic questionnaire, providing information on their birth year, gender, educational level, profession, years of clinical experience, and any personal or relative’s history of mental health problems.
Beliefs and Attitudes towards Mental Health Service Users’ Rights Scale
The BAMHS (Eiroa-Orosa & Limiñana-Bravo, 2019) is a tool designed to assess mental health professionals’ attitudes towards service users’. The scale consists of 25 items, each rated on a 4-point Likert scale. It comprises four dimensions: system criticism / justifying beliefs (α = .701), freedom / coercion (α = .596), empowerment / paternalism (α = .702), and tolerance / discrimination (α = .620). The system criticism / justifying beliefs dimension captures mental health professionals’ beliefs that justify the status quo of service users. The second dimension encompasses recurring topics in discussions about service users’ rights. The empowerment / paternalism subscale reflects beliefs concerning the perceived inability of individuals diagnosed with a mental disorder to take charge of their own lives. The final dimension collects prevailing prejudices towards mental health service users. In our study, higher scores indicate a higher level of rights violations.
Clinical Judgment Scale
The CJS (Lee, 2012; Wallach & Strupp, 1960) is an 18-item questionnaire employing a 7-point Likert scale for each item. It assesses therapists’ overall clinical impressions of a clinical case on a pessimism-optimism dimension concerning clinical prognosis. Higher scores on the CJS indicate a poorer prognosis. Throughout the questionnaire, participants are prompted to make judgments on various factors: Poor Skills (α = .711), Severity (α = .716), Motivation (α = .756) and Treatment (α = .807). The Poor Skills dimension evaluates the perception of service users’ social and relational skills, whereas Severity assesses how therapists perceive the service users’ insight into their problems, their contact with reality, and their prognosis without treatment. Motivation reflects the professionals’ level of interest in working with the service user, while Treatment measures the professional’s impression of the service users’ readiness for treatment and its potential effectiveness.
Values items adapted from the World Values Survey
Two items from the World Values Survey (Inglehart, 2008) were included in the study to assess participants’ alignment with either traditional or progressive values. These items were rated on a 6-point Likert scale, ranging from “I completely disagree” to “I completely agree”. The items were as follows: “Currently there is total equality between men and women” and “A child needs a mother to grow up happy”.
Procedure
The instruments and vignettes were administered through the Qualtrics survey platform. Four clinical vignettes (available as extended data) were specifically designed for this study. All four vignettes presented identical symptomatology, but each one varied in terms of the service user’s gender (male or female) and the type of verbal aggressiveness displayed (related to the opposite gender or unrelated to gender entirely).
In each vignette, the scenario depicted a conversation with a service user who appeared disoriented upon admission to an acute crisis unit. During the interaction, the service user exhibited verbally aggressive behaviour towards the mental health professional. In the vignettes where verbal aggressiveness was not related to gender, the service user’s family described the individual as reserved. However, in the vignette with a male service user displaying gender-related aggressiveness, the family portrayed him as a very traditional man. Conversely, in the vignette with a female service user exhibiting gender-related aggressiveness, the family described her as a very modern woman (for a comprehensive vignette description, refer to the extended data).
The creation of the vignettes adhered to Heverly et al. (1984) four-step protocol to enhance its validity: 1) to identify the experimental factors to be varied, 2) to generate descriptions of behaviour that clearly reflect concrete levels of the desired factor, 3) to empirically validate these descriptions, and 4) to construct the whole vignette from the validated pieces.
Statistical analyses
We employed percentages, means, and standard deviations to describe socio-demographic information, the values items adapted from Inglehart (2008), the BAMHS and the CJS scores. Due to the small size of the male sample, non-parametric tests were used for tests within this group. To explore differences between the experimental and control groups, as well as between male and female participants, t-tests or Mann-Whitney U tests and Cohen’s d were applied for quantitative variables. For categorical variables, chi-square and odds-ratio tests were used. Analyses involved multiple group comparisons using different combinations of the levels of the independent variable, such as gender-neutral men attacking a female MHP vs. gender-related men attacking a female MHP, gender-neutral attacking a male MHP vs. gender-related women attacking a male MHP, men vs. women, and verbal aggressiveness related to gender vs. non-gender-neutral verbal aggressiveness. Additionally, Pearson correlations were performed among the values, BAMHS, and CJS scores. For statistically significant correlations, ANCOVAs were conducted to test the main effects and interaction of the experimental condition on the CJS scores while controlling for covariates.
Results
Sociodemographic characteristics, values, and BAMHS scores are presented in Table 1. Statistically significant differences were observed in Paternalism scores between women in the experimental and control groups, as well as in Coercion scores between the experimental and control groups.
CJS scores are also provided in Table 1. A statistically significant difference was detected in the Severity score between women in the control and intervention groups. Regarding men, statistically significant differences were found in the Treatment and Poor Skills subscales, as well as in the total CJS score, between the intervention and control groups.
Data on the main symptomatology assigned to the vignettes can be found in Table 2. Our findings reveal statistically significant differences between male and female participants concerning psychosis. Female MHPs tended to assign psychosis as the main symptomatology (67.8%), irrespective of whether they belonged to the experimental or control group. In contrast, most male MHPs designated obsessions as the main symptomatology of the vignettes (56.5%), with no statistically significant differences observed between the experimental groups.
Table 2. Symptomatology assigned to the vignette.
| Female participants | | Male participants | | Women vs.
Men | Intervention
vs. Control |
|---|
| Intervention
(n=47) | Control
(n=46) | OR. 95%IC | Intervention
(n=10) | Control
(n=16) | OR. 95%IC | OR. 95%IC | OR. 95%IC |
|---|
| N | % | N | % | | N | % | N | % | | | |
|---|
| Depression | 2 | 4.9 | 3 | 6.5 | NC | 3 | 30 | 1 | 7.7 | NC | 3.453, 0.846-
14.090 | 1.495, 0.379-
5.893 |
| Psychosis | 29 | 70.7 | 30 | 65.2 | 1.289. 0.521-
3.189 | 3 | 30 | 3 | 23.1 | 1.429. 0.220-
9.262 | 0.167, 0.60-
0.471*** | 1.327, 0.617-
2.854 |
| Obsessions | 9 | 22.0 | 12 | 26.1 | 0.797. 0.269-
2.144 | 4 | 40 | 9 | 69.2 | 0.296, 0.053-
1.669 | 4.086, 1.565-
10.666** | 0.619, 0.271-
1.413 |
| Mania | 1 | 2.4 | 1 | 2.2 | NC | 0 | 0 | 0 | 0 | NC | NC | NC |
When conducting Pearson correlations between CJS, BAMHS, and values scores, statistically significant relationships were observed between the CJS Total score and the BAMHS Total and Coercion scores. Furthermore, statistically significant correlations were found between CJS Treatment and BAMHS Coercion scores, as well as between CJS Severity and BAMHS Total, Beliefs, Coercion, and Paternalism scores (see Table 3). Subsequently, the results of the ANCOVAs with the experimental condition as a factor and BAMHS subscales with statistically significant correlations with CJS scores are presented in Table 4. In all cases, the covariate was found to be statistically significant. Notably, in the model including CJS Treatment and BAMHS Coercion, the experimental condition displayed statistical significance, with this model also exhibiting a higher adjusted R-squared. Additionally, when adding the interaction between the covariate (BAMHS) and the experimental group, statistically significant results were only obtained for the BAMHS Paternalism with the CJS Severity (F = 4.485, p = .037, ηp2 = 0.041).
Table 3. Pearson’s correlation between the BAMHS and the CJS.
| Beliefs | Coercion | Paternalism | Discrimination | BAMHS
Total | Equality
between
men and
women | A kid
needs a
mother to
be happy |
|---|
| Treatment | -0.064 | 0.292** | 0.042 | 0.167 | 0.099 | -0.011 | 0.051 |
| Motivation | 0,008 | 0,11 | 0,09 | 0,114 | 0,14 | -0,024 | 0,124 |
| Severity | ,315** | ,211* | ,219* | 0,119 | ,293** | 0,064 | -0,021 |
| Poor skills | 0,1 | 0,062 | 0,046 | 0,035 | 0,069 | -0,003 | -0,149 |
| CJS Total | 0,144 | ,273** | 0,156 | 0,168 | ,234* | 0,018 | 0,008 |
Table 4. ANCOVA using CJS as dependent variable, experimental condition as factor and BAMHS scores as covariate.
| Dependent Variable | Covariate | F | Sig. | Partial Eta
Squared | Adjusted
R Squared |
|---|
| Treatment | Coercion | 13.255 | <.0001 | 0.111 | .112 |
| Condition | 5.264 | .024 | 0.047 |
| Severity | Beliefs | 10.985 | <.001 | 0.094 | .089 |
| Condition | 0.779 | .379 | 0.007 |
| Coercion | 4.061 | .046 | 0.037 | .031 |
| Condition | 0.543 | .463 | 0.005 |
| Paternalism | 4.958 | .028 | 0.044 | .040 |
| Condition | 1.130 | .290 | 0.010 |
| BAMHS Total | 9.239 | .003 | 0.079 | .076 |
| Condition | 0.822 | .367 | 0.008 |
| CJS Total | BAMHS Total | 6.572 | .012 | 0.058 | .041 |
| Condition | 0.465 | .492 | 0.004 |
Discussion
In this experimental study, we explored the potential differential impact of gender-related verbal aggressiveness on mental health professionals’ assessment of a clinical vignette. Our findings revealed that such influence exists to a certain extent, and there is a correlation between participants’ attitudes and beliefs regarding coercive practices and how the experiment affected their clinical judgments concerning treatment.
The results partially support our main hypothesis that professionals’ judgment would be less favourable for vignettes depicting service users whose verbal aggressiveness was related to gender compared to those whose aggressiveness was not. However, it is important to note that these differences can only be observed within each gender and not globally across all participants.
Aggressiveness may be viewed as a direct consequence of gender roles. Previous experimental studies have consistently shown that men tend to be perceived as more aggressive than women (e.g., Eagly & Steffen, 1986), unless gender role features are less salient in the given circumstances. Reciprocity plays a significant role in social interactions, making provocation a common justification for aggression. For women, provocation becomes particularly relevant as it allows them to break free from the limitations often imposed by gender role norms (Bettencourt & Miller, 1996). In addition, differences in aggression between provoked and neutral conditions are more pronounced for women than for men (Bettencourt & Miller, 1996; McCredie & Morey, 2019; Rojas, 2016). Numerous studies, incorporating variations in participants’ gender and levels of provocation, consistently demonstrate that women exhibit less aggression than men under relatively neutral conditions. However, when provoked, women tend to display slightly higher levels of aggression compared to men (Schuck et al., 1971; Taylor & Epstein, 1967). In our study, we found that male participants exhibited more significant differences between the experimental conditions compared to female participants.
The observation that female participants in our study exhibited more stable clinical judgments than male participants can be attributed to various factors. Firstly, previous studies have consistently found that males tend to engage in riskier behaviours compared to females (e.g. Pawlowski et al., 2008). Secondly, women are inclined to downplay hostility and conflict (Prakash & Caeli Flores, 1985) and tend to score higher in agreeableness, which encompasses traits like empathy, kindness, and the maintenance of social harmony, in comparison to men (Weisberg et al., 2011). These higher rates of agreeableness in women may lead them to respond in ways that align with socially accepted norms, while men may feel more inclined to take risks when making decisions. In our study, this could translate to female participants making less severe clinical judgments compared to male participants.
Research on gender differences in stability when making judgments has yielded mixed results. One fMRI study indicated differences in Emotional Stability between men and women, suggesting that men exhibit moderately more Emotional Stability than women across the countries included in the sample (McRae et al., 2008). Accordingly, some studies have reported that women tend to display higher neuroticism, which is associated with less emotional stability (e.g. Grijalva et al., 2015). However, in contrast to these findings, another study found no stable gender differences in individual confidence on judgments, with the stability depending on the knowledge domain or cognitive processes activated (Jonsson & Allwood, 2003). In contrast, in our study, we observed that female participants showed more stability when making clinical judgments compared to male participants.
Becker and Wagner (2009) propose that different types of gender identity can be partly predicted by varying levels of women’s endorsement of sexist beliefs. Women who strongly identify with the ‘female’ category tend to internalise more progressive identity contents and reject benevolent, hostile, and modern sexism. Additionally, their study highlights the potential interaction between gender roles and gender identity. Female participants’ responses to a text describing a woman whose gender role varied (from feminist to traditional) depended on their individual level of gender identification (Becker & Wagner, 2009). However, in our study, no statistically significant correlations were found between female participants with progressive or traditional values and the clinical judgments made for both experimental conditions.
The prevalence of specific diagnoses varied by gender in our study. These differences could be linked to the actual prevalence of these disorders, which may influence professionals’ stereotypes. Accordingly, several studies have indicated that the incidence of psychosis is higher in men compared to women. Moreover, findings on symptoms are inconclusive, but it appears that men experience more negative symptoms, while women tend to have more affective symptoms (Ochoa et al., 2012). The incidence of psychosis is greater for men when negative symptoms and a longer duration of the illness (associated with worse prognosis) are described, whereas there is an equivalent risk for both sexes when affective symptoms and a shorter duration of the illness (related to a better prognosis) are present. Consistently, our results demonstrate that more men were assigned to the psychosis symptomatology category than women. Similarly, although obsessive disorders may be more common among males during childhood, they become more prevalent among females during adolescence and adulthood (Mathes et al., 2019).
As expected in a vignette study, our results revealed that clinicians’ pre-existing beliefs had a stronger predictive power on their clinical judgments compared to the experimental condition. Specifically, our findings indicate that the BAMHS score may predict the severity of participants’ assessment of the given vignette. This suggests that the BAMHS not only serves as a valuable tool to measure the impact of recovery and anti-stigma interventions carried out with mental health professionals (Eiroa-Orosa, 2023; Eiroa-Orosa et al., 2021), but also holds the potential to predict the type of decisions a professional may make.
Furthermore, these results highlight the significance of clinicians’ attitudes and beliefs when evaluating a service user. Prior research has demonstrated a tendency among healthcare providers to hold pessimistic views regarding service users’ reality and their chances of recovery (Henderson et al., 2014; Schulze, 2007). Such pessimism can contribute to stigma and act as a barrier to recovery (Knaak et al., 2017).
Similarly, the use of coercive measures remains prevalent in mental health care. While a significant number of MHPs view coercion as a violation of fundamental rights, a considerable percentage consider it an indispensable measure in psychiatry and beneficial for service users (Morandi et al., 2021). Furthermore, the more a professional is involved in situations where coercion is employed, the more they tend to believe in its efficacy (van Doeselaar et al., 2008) and perceive it as a means of achieving better care and providing increased security (Dahan et al., 2018). Consequently, the results of this study suggest that professionals who endorse coercive practices tend to make more pessimistic judgments.
This study has certain limitations that should be considered. Firstly, the small size of the male sample has resulted in insufficient statistical power for gender-disaggregated analyses. Additionally, data on ethnic and cultural identification could not be collected due to the low ethnic-cultural diversity of mental health professionals, which could have compromised participant anonymity. Regarding the vignettes, they do not fully meet all the factors in the quality assessment framework designed by Sheringham et al. (2021) for evaluating the robustness of experimental vignette studies. Only one vignette was presented to each participant, which prevented the assessment of intra-subject judgments’ variation (number). Moreover, no validity or credibility assessment was conducted to confirm the vignettes’ similarity to real-life scenarios (assessment). However, the study did have some strengths in the design of the vignettes. A comprehensive description of the vignette construction was provided (credibility), the vignettes included variability of the experimental factors (variability), the textual modality was followed (modality), and a full description of the vignettes used in the study was available (description). Regarding the study design, although the purpose of the study was concealed from participants, no active strategies were implemented to further hide the purpose, such as including a distracting element in the vignette (concealment). Additionally, the study did not control for whether the completion of questionnaires took place in contexts that replicated aspects of healthcare (realism). However, the sampling and response rate were clearly explained (sampling and response rate). Finally, as only one vignette was provided per participant, the vignette was not included as a covariate to control for in the analysis (analysis).
It is important to note that the experiment was designed differently for male and female participants. In the experimental vignettes, we depicted a “traditional” male service user and a “modern” female service user, leading to the use of different verbal aggressiveness. However, this approach was inevitable as the vignettes were specifically designed to analyse participants’ reactions after being provoked in relation to their gender. Given the existence of real social differences, gender-related verbal aggressiveness holds distinct connotations for men and women. Moreover, considering that women generally hold more pro-feminist ideas than men (Rojas, 2016), we can expect that reactions to a progressive person would differ from those to a traditional person for both genders. Similarly, in both control conditions, a reserved person was described, but the interpretation of this trait might differ between genders. Therefore, it is essential to acknowledge that the intervention and control conditions are not directly comparable between genders as they are not entirely equivalent. These differences were necessary to examine the unique aspects of verbal aggressiveness related to gender and its impact on the participants’ judgments.
Despite its limitations, this study sheds light on a highly contentious issue. The findings underscore the need for significant changes in the approach of MHPs to provide unbiased care in line with ethical codes. Recognising the potential impact of gender role expectations can help creating a safer therapeutic environment. To address this, establishing open discussions, offering continuing education, and providing supervision on the influence of gender roles in treatment can be valuable tools. Additionally, adopting dimensional diagnostic approaches may prove beneficial. Further research is essential to expand our understanding of how gender influences decision-making in mental health care and to effectively mitigate gender bias.
Ethics and consent
We conducted an experimental study, which was registered at https://dx.doi.org/10.17605/OSF.IO/25VFK and approved by the University of Barcelona Institutional Review Board [IRB: 00003099] on October 8, 2015. Participants provided informed consent prior to participating in the study. All data were anonymised, and confidentiality was maintained throughout the research process.
Before the course, and upon obtaining written informed consent, participants were requested to complete an online questionnaire.
