The impact of exposure to social explanations for depression on public stigma Open Access

Members of the public are commonly exposed to causal explanations for mental illness. Casual attributions frequently appear in news media (Huggard and O’Connor, 2023) and in awareness campaigns and anti-stigma interventions (Walsh and Foster, 2021). Moreover, clinical interactions often involve formulating aetiological explanations of an individual’s mental health difficulties (Larkings et al., 2017). Given the documented implications of causal explanations for depression for lay attitudes (Elliott and Ragsdale, 2023; Haslam and Kvaale, 2015; Lebowitz and Appelbaum, 2019), it is imperative to identify the types of causal information associated with inclusive attitudes towards people with depression, and those that risk heightening stigma.

Meta-analytic research indicates that, when compared to social attributions, biological attributions are associated with increased perceptions of dangerousness, desire for social distance, and pessimism around prognosis (Haslam and Kvaale, 2015). This has contributed to growing concern around the biomedical model of mental illness, and a turn towards acknowledging the evidenced social determinants of mental illness (Huggard et al., 2023). Recent advocacy encourages mental health awareness campaigns to be more reflective of current empirical evidence by emphasising the diverse social factors known to impact mental health (Lebowitz and Appelbaum, 2019). However, “social determinants” encompasses a very broad range of factors, from personal relationships to macro-economic trends (Huggard et al., 2023). It remains unclear whether some social explanations are more effective than others at mitigating stigma, or whether certain social explanations even risk intensifying stigma. For instance, survivors of sexual or intimate partner abuse face specific forms of stigma (Kennedy and Prock, 2018), which may differ from other documented causes of depression, such as relationship difficulties or employment insecurity.

Currently, limited evidence enlightens how distinct social explanations might be associated with distinct attitudes, since prior research has typically collapsed social factors into a singular category, considered only as comparator or control for biological explanations (Kvaale et al., 2013). One prior cross-sectional analysis suggests that attributions to socio-political factors are associated with the most inclusive attitudes (Huggard and O’Connor, 2025), with stigmatising attitudes unrelated to tendencies to attribute mental illness to life circumstances, violence/abuse or relational challenges. Given that both clinical and public communications often directly supply causal attributions for mental health difficulties, experimental work is necessary to understand how providing distinct social explanations to laypeople impacts stigmatising attitudes.

Given that both clinical and public communications often directly supply causal attributions for mental health difficulties, experimental work is necessary to understand how providing distinct social explanations to laypeople impacts stigmatising attitudes. The current study uses experimental survey methodology to examine the impact of exposure to different explanations for depression on stigmatising attitudes. Informed by prior research suggesting lay attributions for mental illness can be decomposed into five factors such as adverse life circumstances, experiences of violence/abuse, relational challenges, socio-political turmoil, and biological factors (Huggard et al., 2025), the analysis considers whether exposure to the various explanations influences desire for social distance and perceived dangerousness of people with depression.

The study was hosted via an online survey platform (Qualtrics) in March–April 2024. Participants were randomly assigned to view one of the five fabricated news articles, followed by measures assessing attitudes towards “Alex,” who was described in the articles as having a diagnosis of depression. Ethical approval was granted by University College Dublin’s Research Ethics Committee, approval number HS-24-10-Huggard-OConnor.

The study was advertised through the participant recruitment service Prolific, with 211 participants compensated at Prolific’s minimum pay rates. Responses were collected anonymously, and participants were required to be over 18 and based in Ireland/UK. A priori power analysis indicated 200 participants were required to detect a moderate effect size (α = 0.05, 80% power). Participant information is available in Table 1. Supplementary material contains participant information for each experimental group.

To maximise external validity, the style and general format of the fictitious articles were adapted from real articles containing causal explanations for mental illness found in popular news publications (Huggard and O’Connor, 2023). Each article contained one empirically supported causal explanation for depression (Huggard et al., 2023), alongside a definition for depression adapted from DSM-5 diagnostic criteria (APA, 2013). The articles also included a quote from a fictional person “Alex,” described as having depression, with no other distinguishable characteristics such as gender or ethnicity. Aside from the different causal explanations, articles were similar in length and content. Supplementary material contains a sample vignette.

The causal information provided was chosen based on prior research, which identified four types of social factors that laypeople perceive as categorically distinct: attributions to life circumstances, violence/abuse, relational challenges and socio-political turmoil (Huggard et al., 2025). A biological causal explanation condition was also included as a reference point. The style and general format of the fictitious articles were adapted from real articles on the topic of causal explanations for mental illness found in popular news publications (Huggard and O’Connor, 2023). Each article contained the same image and a header of a fictional publication “Science Today”, which were added to increase the ecological validity and believability of the articles. Supplementary material contains all vignettes.

Mean scores were calculated for each of the following measures:

Social distance scale (Link et al., 1987). Respondents indicated their willingness to form different relationships with Alex on a scale 1–7 (e.g. “how would you feel about renting a room in your home to someone like Alex?”, α = 0.90).

Attribution questionnaire (Corrigan et al., 2003), Fear and Danger subscale. Participants rated five statements measuring Alex’ perceived dangerousness on a scale of 1–7 (e.g. “I would feel threatened by Alex”, α = 0.94). Participants also rated the three statements of the Personal Responsibility subscale of the Attribution Questionnaire (Brown, 2008; Corrigan et al., 2003). However, this measure was excluded from the analysis due to low internal consistency (α = 0.66), below the recommended minimum of 0.70 for acceptable reliability (DeVellis and Thorpe, 2021).

Social attributions for mental illness (SAMI; Huggard et al., 2025). This measure of lay attributions for depression was included as a manipulation check. Subscales include life circumstances (e.g. “poor housing conditions”, α = 0.93), violence/abuse (e.g. “domestic violence”, α = 0.90), relational challenges (e.g. “negative influences from social groups”, α = 0.77), and socio-political turmoil (e.g. “political instability”, α = 0.85). Participants rated the importance of each factor in causing depression on a scale of 1–5.

Mental illness attribution questionnaire (MIAQ; Knettel, 2019). Also as a manipulation check, the heredity/biological subscale (e.g. “genes or heredity”, α = 0.86) of this instrument was included. Participants rated the importance of seven biological factors on a scale of 1–5.

The study ended by collecting sociodemographic characteristics, including gender, age, highest level of education and ethnicity (Table 1). Participants also rated the following single-item measures:

• Political beliefs (“What best describes your political beliefs?”). Response options included “very left wing”, “somewhat left wing”, “centrist”, “somewhat right wing” and “very right wing”.

• Prior understanding of depression (“How would you rate your level of understanding of depression prior to taking this survey?”). Response options included “never heard of it” “heard of it”, “knew a little”, “knew a lot” and “expert”.

• Personal experience with depression (“Do you personally know someone with a diagnosis of depression?”). Participants indicated “yes” or “no”, and if they selected yes were asked if this person was an acquaintance or a close friend/family member/themself.

The survey also included attention checks to ensure data quality. Participants who failed more than two of these checks were excluded from the analysis.

Manipulation checks were conducted to assess whether the news articles significantly impacted participants’ attributions. One-way, non-repeated ANOVAs investigated differences in each attribution measure (SAMI and MIAQ subscales) across conditions. All sociodemographic variables in Table 1 were incorporated as covariates.

One-way, non-repeated measures ANOVA analyses were conducted to test for significant differences in perceived dangerousness and desire for social distance across vignette conditions. All sociodemographic variables in Table 1 were incorporated as covariates. Post hoc comparisons using Tukey honestly significant difference tests determined the source of significant differences across conditions.

All data was approximately normally distributed, with minimal missing data (<0.1%). Assumptions of homogeneity of variance (p > 0.05) were met for all analyses. Table 2 contains mean scores across conditions. Figure 1 shows mean perceptions of dangerousness and desire for social distance scores across conditions.

The manipulation check did not reveal any significant differences across conditions on any of the five attribution measures (Table 2). However, descriptive results showed that attribution ratings were consistently highest for the causal explanation to which participants were exposed, excepting violence/abuse. For instance, participants in the relational challenges condition rated attributions to relational challenges as more important than participants in any other condition, and participants in the life circumstances condition rated attributions to life circumstances as more important than participants in any other condition. While this was not observed for violence/abuse, this may reflect ceiling effects; participants consistently rated violence/abuse as a highly important determinant, with mean scores ranging from M = 4.34 to M = 4.51 on a scale of 1–5.

The analysis identified a significant difference in perceived dangerousness across experimental conditions, $F(4,198)=3.00,p=0.019, η2=0.06$⁠. Post hoc comparison tests indicated participants in the violence/abuse condition rated Alex as significantly more dangerous than participants in the life circumstances (p = 0.027), or relational challenges (p = 0.033) conditions. There were no other statistically significant differences across conditions.

There were no significant differences in desire for social distance from someone with depression across experimental conditions, $F(4,198)=2.17,p=0.074$⁠.

This study investigated the attitudinal impact of providing distinct social explanations for depression to members of the public. Exposure to a causal explanation pertaining to violence/abuse fostered greater perceived dangerousness of somebody with depression, compared with explanations invoking life circumstances or relational challenges. As the first experimental comparison of different social explanations’ effects on stigmatising attitudes, this study has important implications for informing clinical and public communications about depression’s aetiology.

The association between explaining depression as resulting from violence/abuse and greater perceived dangerousness is consistent with prior research. Childhood adversity explanations have been found to induce lower acceptance of persons with depression (Schomerus et al., 2014), and the tendency to attribute one’s own depressive symptoms to childhood trauma has been linked to greater stigma (Stolzenburg et al., 2018). While the character portrayed in the news article was not described as violent, they were described as raised within a context of violence and abuse. This possibly elicited a belief that the character was socialised to be violent. With moves to incorporate trauma-informed approaches to mental healthcare underway in many jurisdictions (Sweeney et al., 2018), clinicians and spokespeople should be aware that emphasising violence or abuse as causes of depression risks further stigmatising this vulnerable population.

Aside from this finding, the various social explanations did not produce significant differences in stigmatising attitudes. Prior cross-sectional research found that attributions to socio-political factors were correlated with lower stigma (Huggard and O’Connor, 2025), yet the explanation pertaining to socio-political turmoil in the present study did not significantly reduce stigma. The fact that cross-sectional relationships were not reproduced experimentally suggests these social attributions may not causally affect stigmatising attitudes. Further research is required to identify whether such correlations may reflect confounds (e.g. political ideology) or reversed causal directionality (i.e. more inclusive attitudes promoting greater awareness of socio-political determinants of mental illness).

At face value, these findings may indicate that promoting different social explanations has minimal effects on public attitudes. However, methodological caveats should be considered when interpreting findings. The non-significant manipulation checks suggest the experimental manipulation may have been insufficiently strong, i.e. the fictional articles did not influence participants’ causal beliefs. Notably, the study did not reproduce the well-established finding (albeit with some exceptions [e.g. Breheny, 2007; Phelan, 2005]) that biogenetic explanations induce greater stigmatising attitudes than social explanations (Elliott and Ragsdale, 2023; Haslam and Kvaale, 2015; Larkings et al., 2017; Lebowitz and Appelbaum, 2019; Schroder et al., 2020). Many participants self-reported as having high understanding and personal experience of depression, which may have made their attributions less malleable. Furthermore, while the news-format of the articles offered greater external validity, growing public distrust in news media (Hanitzsch et al., 2018) may have undermined endorsement of the explanations to which participants were exposed. Yet while the manipulation check fell short of statistical significance, descriptive causal attributions ran in predicted directions; moreover, prior research shows stigma attitudes may be impacted by causal explanations even in the absence of significant manipulation of causal beliefs (Dittrich et al., 2021).

Several study limitations warrant consideration. First, while the sample had good diversity of gender and age, participants were all from Ireland and the UK and were predominantly White and highly educated. As cultural factors affect attribution tendencies and stigma (Furnham and Swami, 2018), further research is needed to assess whether effects generalise to other groups and contexts. Additionally, as only 43% of the sample reported high levels of understanding of depression, yet 68% had personal experience with depression, research should investigate whether results would differ in samples with higher baseline knowledge of depression. Second, restriction to just two measures of stigma may have overlooked other effects; for instance, prior research suggests causal explanations may also impact prognostic expectations and treatment preferences (Angermeyer and Schomerus, 2017). Further research is needed to infer whether distinct social explanations may influence engagement with mental healthcare. An additional limitation is that the study did not include a comparator condition, such as a causal explanation for an individual without a diagnosis of depression or alternative diagnosis. Future research should examine whether the impact of exposure to causal explanations differs across various mental health conditions. Finally, the use of a contrived experimental vignette design may limit applicability to real-world contexts. However, adapting real news articles from within participants’ cultural context rendered the articles more naturalistic and reflective of real-world encounters with such information.

On the basis of prior research (Haslam and Kvaale, 2015; Lebowitz and Appelbaum, 2019), it remains imperative that news media, public information/awareness campaigns and clinical communications emphasise the evidenced social determinants of depression, avoiding the promotion of a solely biological model. Though preliminary, the current study suggests that the specific social explanations foregrounded may not greatly affect stigma, at least in populations with high mental health literacy. However, caution is advisable when presenting any association between depression and violence or abuse, as this could promote harmful stereotypes of dangerousness. Strategies for mitigating such counterproductive effects, for instance by pairing violence/abuse explanations with evidence of the low incidence of violence within depressed populations, should be a focus for future research.

The data associated with this research is available at: Link to FigshareLink to the website of Figshare.

The supplementary material for this article can be found online.