How Self-Disclosure in Online Postings Affects Female Readers’ Stigmatisation Towards Mothers with a Disability

Self-disclosure of a health problem or a disability in online environments can be helpful to reduce stigmatisation and to empower marginalised individuals. Although stigmatisation leads to adverse health outcomes, it is still unclear which factors reduce readers’ stigmatising attitudes. This 2 × 2 × 3 × 2 online-experiment with 715 female participants examined how the depiction of self-disclosure of a mother’s disability, her coping with a stressful situation, and the child’s type of disability and sex in a fictional blog post affect female readers’ stigma-related attitudes. The role of readers’ perspective taking is also studied in this regard. A MANOVA yielded two main effects: Disclosure of the mother’s disability reduced social distance compared to the control condition (no disability). Female respondents with a high ability for perspective taking reported less stigmatising attitudes than female respondents scoring low in this regard. Both effects are also reflected in a four-way interaction with mother’s coping and child’s sex. Overall, results indicate a high potential of self-disclosures for anti-stigma communication. Implications for blogging about a disability and future health communication research are discussed.

Self-disclosure about health problems can be helpful, at least in some situations, for reducing public stigmatisation and for empowering marginalised individuals (e.g., Rains, 2014). Especially mothers with a disability, as well as mothers of children with a disability, appear to use blogs to connect with each other (Peterson-Besse et al., 2019), but also for self-representation and selfdisclosure about their situation (Litchman et al., 2019;Loukisas & Papoudi, 2016). Since both groups are subjected to public stigmatisation (Eaton et al., 2016;Werner & Shulman, 2015), they also use their online postings for sharing anti-stigma messages (Litchman et al., 2019;Swanke et al., 2009). Research indicates that self-disclosure of a disability in online postings yields positive effects regarding an individual's perspective on life, social inclusion, and knowledge about the disability (Litchman et al., 2019;Loukisas & Papoudi, 2016). While health communication scholars so far focused primarily on the impact of such postings on the authors' well-being, empowerment, and perceived social support, especially in the context of health blogging Rains & Keating, 2011, 2015, effects on readers' stigmatising attitudes remained unclear. Being stigmatised greatly increases the risks for adverse health effects and delayed treatment (Evans-Lacko et al., 2015). These negative health consequences and the fact that people with health problems or a disability are particularly affected by stigmatisation, emphasise the importance of examining factors that shape such stigmatisation processes.
Building on Bos and colleagues' (2013) conceptualisation of stigmatisation processes, the present study mostly relates to public stigma and family stigma, as they are known to affect both mothers with a disability as well as mothers of children with a disability (e.g., Litchman et al., 2019;Werner & Shulman, 2015). Public stigma refers to prejudiced and discriminating attitudes of the general population towards persons with a stigmatised condition (e.g., a disability). As a subcategory of stigma by association, family stigma "relates to the stigma experienced by family members as a consequence of being associated with a stigmatised relative" (Werner & Shulman, 2015, p. 272). Following the three components of attitude (Breckler, 1984), stigmatising attitudes are commonly conceptualised as either affective (e.g., emotions towards a person, group, or issue), cognitive (e.g., knowledge about a person, group, or issue), or conative (e.g., behavioural intentions toward a person, group, or issue). Accordingly, social distance is the most commonly used operationalisation of public stigma with the purpose "to assess (expected) discriminatory behaviour" (Baumann, 2007, p. 132). Hence, in the context of this study, stigmatising attitudes encompass both public negative (stigmatising) attitudes (i.e., social distance) and individual positive (destigmatising) attitudes (i.e., motivation to unprejudiced behaviour). Such attitudes can be formed, but also changed, by media or social media representations of affected individuals (e.g., Clement et al., 2013).
The concept of mediated intergroup contact (Park, 2012), which is grounded on Allport's (1954) contact hypothesis, presumes "that people form their ideas about different social groups based on what they read, hear, and/or see from both fictional and factual media sources" (Park, 2012, p. 137). Mediated (in-vitro) contact is considered to be as effective as real (in-vivo) contact (Reeves & Nass, 1996) and has shown to successfully reduce stigmatisation (Clement et al., 2012;Wojcieszak & Azrout, 2016). Yet, such attitude changes are dependent on further influencing factors. A review of 50 years of research on Allport's influential contact hypothesis (Kenworthy et al., 2005) indentified particularly self-disclosure as well as perspective taking as important for succesful encounters. Both are applicable for mediated contacts, but demand further exploration (Park, 2012). The present experimental study therefore examines how (1) self-disclosure of a person with a disability and (2) further stigma-related exemplar characteristics in online postings affect stigmatising attitudes of a predominantly female audience, as well as (3) which role female readers' ability for perspective taking plays in this process.

Self-Disclosure and Family Stigma of Mothers of Children with a Disability
Self-disclosure, the coming out with one's own disability or health condition, is known to positively affect attitudes and reduce stigmatisation (Bos et al., 2009;Smith et al., 2008). A metaanalysis by Smith and colleagues (2008) of 21 studies showed that self-disclosure of an individual's HIV-positive status increased social support and decreased stigma towards affected individuals. The coming out proud approach (Corrigan & Matthews, 2003) aims to diminish public stigma and self-stigma by self-disclosure of persons with mental health problems, and has been successfully applied in various settings (Bos et al., 2009;Rüsch et al., 2014). Since this approach is mainly building on Allport's contact hypothesis, it can be adopted for mediated intergroup contacts (Park, 2012). So far, this has only been tested for individuals with mental health problems, and not for online postings. Moreover, previous experimental research on the destigmatising effects of media representations (e.g., vignettes or news portrayals) of people with a disability focused primarily on the role of different characteristics of media exemplars by utilizing a thirdperson perspective (e.g., Hastall et al., 2016;Oliver et al., 2012;Röhm et al., 2018), instead of self-disclosure, which is commonly used in blogs and other forms of online postings to provide first-person information (Joinson, 2001;. Using a firstperson perspective in health-related text-based narratives has been identified as particularly promising for attitude changes, compared to third-person narratives (de Graaf et al., 2016). Additionally, self-disclosure generally conveys intimacy (Laurenceau et al., 1998) and liking (Collins & Miller, 1994). Thus, sharing of personal information about a health problem or a disability can help readers to relate with the person (Corrigan & Matthews, 2003), even when he or she is unknown and when no direct social or para-social relationship is established (Park, 2012). Taken together, self-disclosure of a mother with a (physical) disability should have a destigmatising effect for female readers: Hypothesis 1: Self-disclosure of a mother with a disability reduces female readers' stigmatising attitudes towards people with a disability, compared to portrayals of a mother without a disability. Corrigan and colleagues (2013) also identified successful coping with daily challenges as a key aspect in the coming out proud approach. Accordingly, effective anti-stigma "messages need to include 'on the way up' stories, discussing strengths and successes" (Corrigan et al., 2013, p. 176) in situations of daily life, which recipients can relate to. This is different to the approach of using media representations of individuals with a disability for destigmatisation, who achieved extraordinary accomplishments, as, for example, Bartsch and colleagues (2018) examined in the context of the Paralympics. Accordingly, this study aims to shed light on the effect of descriptions of an either positive or negative coping experience in a stressful everyday situation. Building on the assumptions from the coming out proud approach (Corrigan & Matthews, 2003;, we presume that depictions of positive coping reduce stigmatisation (compared to depictions of negative coping): Hypothesis 2: The depiction of an exemplar's positive coping with a stressful situation reduces female readers' stigmatising attitudes towards people with a disability, compared to depictions of negative coping.
In contrast to the hypothesised positive effects of disability self-disclosure on public stigma, mothers of children with a disability are also very likely affected by family stigma (Eaton et al., 2016;Werner & Shulman, 2015). This stigma experience includes bullying or social exclusion by others (Benson et al., 2016), as well as psychological distress or low quality of life (van der Sanden et al., 2016), and even self-stigma (Eaton et al., 2016). Regarding individuals with an intellectual disability, a systematic review of 20 studies on family stigma (Ali et al., 2012) shows that both individuals and family members experience stigma, and that it can negatively affect their psychological well-being. Still, little is known about stigmatisation and especially family stigma effects in the context of self-disclosure in online postings of mothers of children with a disability. Findings regarding the effects of exemplars with disabilities on audiences' stigma-related attitudes suggest, however, that such attitudes are likely affected by disability-related cues (Röhm et al., 2018;von Sikorski & Schierl, 2014). Priming processes are likely responsible for this formation of attitudes (Cho et al., 2006;von Sikorski & Schierl, 2014), and have been comprehensively examined in the context of mass media effects (Roskos-Ewoldsen et al., 2009). Priming is generally conceptualised as an often unintentional and unaware "activation of social representations (e.g., traits, stereotypes, or goals) by exposure to different types of information, and the application of these activated representations in social judgements and behaviors" (Molden, 2014, p. 3). Priming of disability-related information (in comparison to sports, politics, or no cues) yielded decreased positive attitudes towards people with an amputation (von Sikorski & Schierl, 2014). Building on these findings, we assume that a priming of disability-related cues activates stigma-related attitudes. More specifically, the salience of a child's disability should negatively affect readers' stigma-related attitudes towards the mother: Hypothesis 3: The depiction of a child with a disability increases female readers' stigmatising attitudes towards people with a disability, compared to the depiction of a child without a disability.
Findings in the context of Weiner's (1986) attributional theory indicate that intellectual disabilities are generally more stigmatised than physical disabilities (Barr & Bracchitta, 2015;Hernandez et al., 2000;Miller et al., 2009;Weiner et al., 1988;Werner, 2015). Likewise, Werner and Shulman (2015) observed that family carers of children with an intellectual disability feel more embarrassed for their child's behaviour than family carers of children with a physical disability. We therefore expect this phenomenon to also materialize on the level of a family stigma effect: Hypothesis 4: The depiction of a child with an intellectual disability increases female readers' stigmatising attitudes towards people with a disability, compared to a portrayed child with a physical disability.

Influence of Readers' Perspective Taking on Stigmatisation
According to Cohen's (2001) definition of identification, perspective taking is an essential part of cognitive empathy, and important for identifying with media characters (Cohen, 2001;Igartua & Vega, 2016;Moyer-Gusé, 2008). In contrast to affective empathy, the sharing of another person's feelings (Igartua & Vega, 2016;Moyer-Gusé, 2008), cognitive empathy is commonly understood as the ability to see and rate a situation from another person's perspective (Chung & Slater, 2013;Cohen, 2001). Therefore, perspective taking can be considered as a personality trait that "helps viewers recognize the humanizing complexity of others and highlights their individuality" (Chung & Slater, 2013, p. 899), which, in turn, should promote positive attitude changes and reduce stigmatisation. Although research shows that, in general, high empathy is strongly related with reduced stigmatisation (Bartsch et al., 2018;Chung & Slater, 2013;Comello & Farman, 2016;Oliver et al., 2012), identification, empathy, as well as perspective taking are often operationalised as the same unidimensional state variable. Chung and Slater (2013), for example, studied how identification in the form of perspective taking with an either low-stigmatised or high-stigmatised media character affects the social acceptance of this person. On the one hand, a highly stigmatised character evoked less perspective taking than a low-stigmatised character. On the other hand, perspective taking as a mediator increased social acceptance, which was, in turn, moderated by the level of stigmatisation (low or high). A reanalysis of the same data by Comello and Farman (2016) suggests that perspective taking functioned as both mediator and moderator of the observed effects on social acceptance. This leads to the conclusion that integrating affective and cognitive empathy into one dimension is (a) not in line with the underlying theoretical concepts (e.g., Cohen, 2001) and (b) limits the validity and interpretability of the reported effects. In accordance, Chung and Slater (2013) acknowledge that "it may prove useful to separate the cognitive and emotional components of perspective-taking" (p. 902). For this reason, and since media effects on stigmatisation involving (affective) empathy as state are well documented (Bartsch et al., 2018;Oliver et al., 2012) and perspective taking is considered just as important for attitude changes (Kenworthy et al., 2005;Park, 2012), the present study focuses on the investigation of the influence of cognitive empathy in the form of perspective taking as trait. While studies in the context of affective empathy examine if a certain (media) stimulus has the potential to elicit recipients' empathic feelings towards another person or group (e.g., Bartsch et al., 2018;Oliver et al., 2012), in our experiment, an individual's general ability for perspective taking is considered as a precondition for being empathetic with another person, which, in turn, can reduce stigmatisation (Comello & Farman, 2016). Thus, we assume that high levels of perspective taking lead to decreased stigmatisation, compared to low levels of perspective taking: Hypothesis 5: Female readers reporting high levels of perspective taking ability show less stigmatising attitudes towards people with a disability after reading the stimulus material, compared to female readers reporting low levels of perspective taking ability.

Design and Procedure
A 2 × 2 × 3 × 2 online experiment was conducted, in which participants read a fictional blog post of a mother with a young child. The stimulus text was manipulated regarding the mother's disability (no disability vs. physical disability), the mother's coping in a conflicting situation (positive vs. negative), the child's disability (no disability vs. physical disability vs. intellectual disability), and the child's sex (female vs. male) as control. Before the stimulus material was presented, participants' perspective taking was measured. Participants' social distance and motivation to behave unprejudiced were measured as dependent variables after the stimulus material was presented. Finally, socio-demographic data was collected and all participants were debriefed and thanked for participating in the study.

Stimulus Material
The stimulus material was designed as a typical blogging post (see Appendix A). A blogging mother describes a conflict between her own child and another child at a garden party, and mentions the reactions of other mothers. The mother describes herself as either having no disability or using a wheelchair (physical disability). Her coping with and solution of the conflict situation is depicted as successful resulting in positive reactions from other mothers (positive coping), or as unsuccessful with negative reactions (negative coping). The child of the blogging mother has either no disability, uses a wheelchair (physical disability), or has Down syndrome (intellectual disability), and is either a girl (named Lea) or a boy (named Leon).

Stimulus Check
A stimulus check with N = 100 participants (M = 34.7 years; SD = 14.1; 62% female) was conducted using a paper and pencil questionnaire. The participants were recruited via personal contacts and were presented with six of the 24 manipulated blog posts. For each article, they were asked to indicate which version of the mother's disability (no disability vs. physical disability), the mother's coping in a conflicting situation (positive vs. negative), the child's disability (no disability vs. physical disability vs. intellectual disability), and the child's sex (female vs. male) they noticed. Conformation of the successful manipulations was obtained by calculating the agreement among all participants as coders (intercoder reliability; Hayes & Krippendorff, 2007) for each stimulus feature (Krippendorff's alphas: mother's disability = .94; coping = .84; child's disability = .83; child's sex = .99). The stimulus manipulations were correctly identified by 97.7% (no disability) and 99.3% (physical disability) of our respondents for mother's disability, 99% (positive) and 93% (negative) for coping, 92.5% (no disability), 96% (physical disability), and 89.5% (intellectual disability) for child's disability, as well as 99.3% (female) and 100% (male) for child's sex.

Sample
Seven hundred and fifteen female respondents fully completed the study (M = 24.4 years; SD = 4.6). The sample was mainly recruited via the most followed German family-or disability-related accounts on Instagram. They were asked to share the link to the survey on their profile or in their stories to reach as many potential recipients as possible. In addition, the link was shared on further German social media sites, message boards, and mailing lists. Seventeen respondents (2.4%) indicated to have a physical disability. One hundred and ten respondents (15.4%) reported to have one or more children, of which nine had a physical and five had an intellectual disability. In addition, 22.8% (n = 163) stated to have no or little contact with people with disabilities, while 39.8% (n = 284) had more or less steady encounters and 37.4% (n = 267) had contact with people with a disability on a regular basis.

Measures
Perspective Taking. Respondents stated their ability for perspective taking as trait on four items (Cronbach's alpha = .76; e.g., "In an argument I try to understand both sides before making a decision.") from the German adaption of the Interpersonal Reactivity Index (IRI-S D; Paulus, 2016) on a five-point Likert scale (1 = never; 5 = always). High scores indicate a high self-reported ability for perspective taking.
Social Distance. Social distance was measured using an adaption of Angermeyer and Matschinger's (1995) seven-item scale (Cronbach's alpha = .86) to operationalise public stigma towards people with a disability in general (Ouellette-Kuntz et al., 2010). Respondents were asked to indicate on a five-point Likert scale (1 = in any case; 5 = in no case at all) if they would accept "someone with a disability" as a colleague, a neighbour, a friend, a work recommendation, a subtenant, a marry-in, or a child carer. High scores indicate a high tendency for social distance towards people with a disability.
Motivation for Unprejudiced Behaviour. Participants' motivation for unprejudiced behaviour was assessed with eight items (Cronbach's alpha = .77; e.g., "I'm angry about prejudiced behaviour against minorities.") adapted from Banse and Gawronski's (2003) subscale behaviour control, using a five-point Likert scale (1 = does not apply at all; 5 = fully applies). High scores indicate a high motivation for unprejudiced behaviour, which reflects an individual's intention to show anti-discriminatory behaviour (i.e., destigmatisation) towards minorities. Table 1 shows means, standard deviations, and intercorrelations between the trait variable and dependent measures included in this study.

Results
A multivariate analysis of variance (MANOVA) with all four experimental manipulations (1. mother's disability, 2. coping, 3. child's disability, 4. child's sex) as well as readers' perspective taking (median-split) as factors was computed with both dependent measures. To protect subsequent univariate analyses of variance (ANOVAs) against type I error, only effects of the MANOVA with p < .05 using Pillai's trace (V) are reported (Field, 2018). The significance of differences between the estimated marginal means was determined through pairwise comparisons using Sidak-corrected simple effect post-hoc tests, which compare all levels of one experimental factor with each level of all other factors (instead of comparing all means with each other). Effect sizes are indicated by both eta-squared (η 2 ) and partial eta-squared (η p 2 ) for the subsequent ANOVAs based on the SPSS output. While eta-squared measures "the proportion of total variance that a variable explains" (Field, 2018, p. 780) and is recommended for reporting in communication and social sciences (Levine & Hullett, 2002), partial eta-squared (a) is more consistent with the Vvalues of the MANOVA, (b) accounts for the intercorrelation of the dependent variables, (c) is "the proportion of variance that a variable explains that is not explained by other variables in the analysis" (Field, 2018, p. 780), and (d) allows to compare effect sizes among different studies (Ellis, 2010;Lakens, 2013).

Main Effects of Experimental Manipulations
A main effect of mother's disability emerged on the interaction of both dependent measures, V = .013, F(2, 659) = 4.410, p = .013, η p 2 = .013. Subsequent ANOVAs, however, only showed a significant effect for social distance, F(1, 660) = 7.341, p = .007, η 2 = .001, η p 2 = .011, but not for motivation to behave unprejudiced. Depictions of mothers with a physical disability (M = 1.52; SE = 0.03) evoked significantly less social distance than depictions of mothers without a disability (M = 1.62; SE = 0.03; p = .01), which supports Hypothesis 1, but only shows very small effect sizes. Since the MANOVA yielded no significant effects of coping or child's disability, Hypotheses 2, 3, and 4 cannot be supported.

Main Effect of Readers' Perspective Taking
Readers' perspective taking produced a main effect on the combination of both dependent measures, V = .037, F(2, 659) = 12.756, p < .001, η p 2 = .037. Subsequent ANOVAs indicated significant effects on social distance, F(1, 660) = 4.809, p = .029, η 2 = .001, η p 2 = .007, and motivation to behave unprejudiced, F(1, 660) = 24.456, p < .001, η 2 = .001, η p 2 = .036. As presumed in Hypothesis 5, readers who scored low on perspective taking reported significantly more social distance and less motivation for unprejudiced behaviour, compared to readers with high perspective taking scores (see Figure 1). Again, effect sizes are rather small and readers' perspective taking only accounts for about 3.7% of the variance not explained by other variables.

Figure 1. Estimated Marginal Means for the Main Effect of Readers' Perspective Taking on Social Distance and Motivation for Unprejudiced Behaviour
Note. Pairwise comparisons using Sidak-corrected simple effect post-hoc tests. *** p < .01, * p < .05.

Higher-Order Interaction
A mother's disability × coping × child's sex × readers' perspective taking four-way interaction emerged on the interaction of both dependent variables, V = .015, F(2, 659) = 5.166, p = .006, η p 2 = .015. Subsequent ANOVAs revealed a significant effect of this interaction on social distance, F(1, 660) = 4.226, p = .04, η 2 = .001, η p 2 = .006, but not on motivation to behave unprejudiced. Figure 2 shows a rather complex effect pattern, indicating particularly moderating effects of (a) mother's disability, (b) readers' perspective taking, and (c) the depicted coping success on reported social distance towards mothers of male children. On the one hand, depictions of mothers without a disability and a male child in a negative coping-condition evoked most social distance from readers reporting low perspective taking. On the other hand, depictions of mothers with a physical disability and a male child in a positive coping-condition yielded least social distance from readers, who scored high on perspective taking. In contrast to Hypothesis 2, the depiction of a positive coping experience increased social distance of readers reporting low perspective taking towards mothers with a disability and a male child, compared to a negative coping experience. All in all, the effect patterns reflect the already reported main effects of mother's disability and readers' perspective taking for very specific combinations of exemplar characteristics, which limits to some extent the generalisability of the reported main effects. However, since (1) there is no significant interaction pattern that contradicts the reported main effects (Field, 2018), (2) all main effects show a larger effect size than the interaction effect, and (3) the main effect of perspective taking does not solely emerge on social distance, the main effects are still considered as valid.  Note. Means sharing the same symbol indicate significant mean differences between readers' perspective taking, whereas means sharing the same capital letter indicate mean differences between the mother's disability, and means sharing the same small letter indicate mean differences between the depicted coping success with p < .05 from pairwise comparisons using Sidak-corrected simple effect post-hoc tests.

Discussion
The aim of this study was to shed light on the interplay between different features of online postings by mothers of children with a disability and female readers' ability for perspective taking on stigmatisation towards this group. In detail, disclosure of the mother's disability reduced social distance compared to the control condition (no disability), as presumed by Hypothesis 1. This effect underlines the stigma-reducing potential of self-disclosure and blogging, as reported by other scholars (Corrigan & Matthews, 2003;Rains, 2014). In addition, this adds to the present body of research on the stigma-reducing impact of mediated contacts with individuals with a disability or other health issues (e.g., Clement et al., 2012). Because all participants were female and rather young, similar to the portrayed mother, the mediated contact could have induced a high perception of similarity with the depicted blogging mother. Which, in line with Allport's (1954) original considerations, leads to reduced stigmatisation towards the portrayed minority group (mothers with a disability). However, these effects deserve further investigation, since the impact of non-real contacts on attitude changes is currently questioned by some scholars (Firat & Ataca, 2020). Regarding the influence of the depicted successful versus unsuccessful coping with a challenging situation on readers' stigma-related attitudes (Hypothesis 2), a four-way interaction effect emerged instead of the postulated main effect. The observed effect pattern implies that a negative coping situation, compared to a positive coping situation, of a mother with a physical disability and a male child reduces social distance reported by readers with low perspective taking. This finding is conflicting with the key assumptions for anti-stigma messages, according to which the depiction of successful coping with daily challenges is advised Corrigan & Matthews, 2003). Since this effect emerged only in this specific combination, it could be an accidental finding, or an indication of gender-specific discrimination patterns. Nonetheless, Hypothesis 2 cannot be supported.
Results also do not support the assumed family stigma effect (Hypothesis 3). Depictions of mothers of children with a disability did not evoke more stigmatisation than depictions featuring children without a disability. Hypothesis 4 must be likewise rejected, as no significant differences in readers' reactions towards mothers with children with a physical or intellectual disability were found. In contrast to previous findings in the context of priming processes (Röhm et al., 2018;von Sikorski & Schierl, 2014), disability-related cues did not activate the proposed attitude patterns. Besides, it is possible that this manipulation was not obvious enough for readers. The disability condition of the child could have been less salient than the disability condition of the mother, and therefore less relevant for attitude formation. Likewise, the manipulation of the child's sex, which was included for control, did not yield any significant effect. Concerning the influence of readers' perspective taking on stigmatisation, we found support for our fifth hypothesis. Readers who indicated a high ability for perspective taking before reading the stimulus were generally less social distancing and more unprejudiced than readers with low perspective taking ability. This is reflected in the observed interaction effect. Overall, this result is in line with findings by other scholars (Comello & Farman, 2016), and adds to the presumed relevance of perspective taking for mediated intergroup contacts (Park, 2012). In addition, this finding can help to shed light on the conceptual understanding and relevance of both empathy as trait (perspective taking) and empathy as state in the context of stigmatisation (c.f., Chung & Slater, 2013). Thus, it is not only important to get emotionally affected by a stimulus to be less stigmatising, as reported by some scholars (e.g., Bartsch et al., 2018;Oliver et al., 2012), but also to have some kind of prerequisite to assume another person's point of view in order to react less prejudiced. Considering Comello and Farman's (2016) findings, highly empathic respondents are also likely to become emotionally affected. In future studies, however, these relationships need more attention and ideally an operationalisation that distinguishes between cognitive and emotional aspects of empathy. Overall, the results signal self-disclosure and perspective taking as key factors for better understanding stigmatisation processes, since both were able to affect behaviour-related attitudes. In particular, readers' tendency to agree with or to reject public stigma and their intention to behave less stigmatising is influenced by personality traits such as perspective taking. With that in mind, processes of stigmatisation as well as destigmatisation, especially on the individual level, not just

Limitations
Only young female readers were invited to participate in the study, which was decided due to the nature of this topic (motherhood) and related audience topic preferences. Thus, it cannot be ruled out that the participants were positively pre-influenced regarding the presented topic due to their recruitment via family-or disability-related accounts on Instagram. Therefore, the generalisability of the findings might be limited and should be verified with a more diverse sample. While the results can be considered as relatively representative for this specific population, future studies should also assess, for instance, attitudes of older as well as male readers towards motherhood or generally parenthood with a disability or children with a disability. It is also likely that the readers' educational background has an effect on stigmatisation, which was not assessed in the context of this study and should therefore be taken into account in future studies. Then again, readers' disability status as well as previous encounters with people with a disability were surveyed, but not included in the final data analysis, because of an unequal distribution among the cells of the MANOVA. Thus, future research should more systematically consider recipients' disability and previous contact as control variables for media effects. Moreover, state variables like identification, empathy, or pity, which could have shed light on further effects of the rather emotional stimulus material and helped to explain the lower social distance towards mothers with a disability, compared to mothers without a disability, were not measured in this study. The selection of the applied attitude measures only accounts for negative (stigmatising) attitudes (i.e., social distance) and positive (destigmatising) attitudes (i.e., motivation to unprejudiced behaviour) on the behavioural dimension of attitudes (Breckler, 1984). Future studies should therefore examine if the observed effects also emerge on the emotional and cognitive dimension, as well as employ more salient message manipulations, in order to clarify and uncover further stigmatisation effects. Thus, generalisability of the results is reduced, because of (a) the unidimensional attitude measures, (b) the one-time presentation of the stimulus manipulations in a single online posting, and (c) the rather small effect sizes (Ellis, 2010). Further investigations should consider a more powerful approach to examine the impact of each manipulation in more detail, and, in addition, should also account for possible long-term effects across multiple online postings. While our analyses strictly controlled for type 1 errors, we cannot fully exclude that some findings became European Journal of Health Communication 2021, Vol. 2(2) 1-20 CC BY 4.0 13 significant by accident and therefore need to be tested regarding their replicability in future research.

Conclusion
Self-disclosure about certain health conditions and disabilities in online postings, and the direct and indirect effects on stigmatisation and health outcomes, are important but still largely neglected topics in health communication. In light of the widespread use of blogs and other social media platforms, as well as their accessibility and importance for people with disabilities, this study indicates a high potential of written self-disclosures in online postings for anti-stigma communication purposes.