- Email: [email protected]
#Schizophrenia: Use and misuse on Twitter
Schizophrenia Research 165 (2015) 111–115
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
#Schizophrenia: Use and misuse on Twitter Adam J. Joseph a, Neeraj Tandon a, Lawrence H. Yang b, Ken Duckworth c, John Torous a, Larry J. Seidman a, Matcheri S. Keshavan a,⁎ a b c
Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, United States Department of Epidemiology, Columbia University, New York, NY, United States National Alliance on Mental Illness, Arlington, VA, United States
a r t i c l e
i n f o
Article history: Received 8 December 2014 Received in revised form 1 April 2015 Accepted 9 April 2015 Available online 30 April 2015 Keywords: Stigma Social media Twitter Schizophrenia
a b s t r a c t Background: The role and prevention of stigma in mental illness is an area of evolving research. Aims: The present study is the first to examine the use and misuse of the word ‘schizophrenia’ on Twitter.com in comparison with another illness (diabetes) by analyzing Tweets that use the adjective and noun forms of schizophrenia and diabetes. Method: Tweets containing one of four search terms (#schizophrenia, #schizophrenic, #diabetes, #diabetic) were collected over a forty-day time period. After establishing inter-rater reliability, Tweets were rated along three dimensions: medical appropriateness, negativity, and sarcasm. Chi square tests were conducted to examine differences in the distributions of each parameter across illnesses and across each word form (noun versus adjective). Results: Significant differences were seen between the two illnesses (i.e., among “schizophrenia”, “schizophrenic”, “diabetes”, and “diabetic”) along each parameter. Tweets about schizophrenia were more likely to be negative, medically inappropriate, sarcastic, and used non-medically. The adjective (“schizophrenic”) was more often negative, medically inappropriate, sarcastic, and used non-medically than the noun “schizophrenia.” Schizophrenia tweets were more likely to be negative and sarcastic when used non-medically and in a medically inappropriate manner. Conclusions: Our findings confirm the presence of a great deal of misuse of the term schizophrenia on Twitter, and that this misuse is considerably more pronounced by the adjectival use of the illness. These findings have considerable implications for efforts to combat stigma, particularly for youth anti-stigma efforts. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Stigma – a highly discrediting attribute that reduces the stigmatized person “from a whole and usual person to a tainted, discounted one” – is an area of evolving research in the schizophrenia literature (Goffman, 2009). Stigmatizing attitudes and beliefs about schizophrenia (and other mental illnesses) decrease treatment seeking behaviors and medication adherence (Corrigan, 2004). Survey, vignette, and interview studies have found that community respondents perceive those with schizophrenia as unpredictable and dangerous. Respondents generally view schizophrenia as more negative and stigmatizing than other mental illnesses, such as depression or anxiety (Angermeyer and Dietrich, 2006; Pescosolido et al, 2010). Early identification of schizophrenia is critical since treatment delays are associated with worse
⁎ Corresponding author at: Room 610, Massachusetts Mental Health Center, 75 Fenwood Road, Boston MA 02115, United States. Tel.: +1 617 754 1256; fax: +1 617 754 1250. E-mail addresses: [email protected], [email protected] (M.S. Keshavan).
http://dx.doi.org/10.1016/j.schres.2015.04.009 0920-9964/© 2015 Elsevier B.V. All rights reserved.
clinical outcomes (Perkins et al, 2005). Therefore, stigmatizing beliefs held among youth are particularly important for early detection and treatment of schizophrenia. The formation of stigmatizing beliefs deserves consideration due to stigma's role in diverting treatment. Illnesses with complex etiology such as schizophrenia are particularly likely to be stigmatized (Sontag, 2013). Studies suggest that various medically inappropriate causal explanations for mental illness can increase stigmatizing perceptions of schizophrenia (Lincoln et al., 2008). Increased mental health literacy may also reduce negative, stigmatizing behaviors and beliefs (Rüsch et al., 2005). Stigma also results from labeling, which entails oversimplification of the characteristics of groups and association of the label with negative attributes (Link and Phelan, 2001). Consequently, the adjective form of schizophrenia (“schizophrenic”) generates greater negative emotions, including perceptions of dangerousness and greater desire for social distance from the person with the illness, than does the noun “schizophrenia” (Penn and Nowlin-Drummond, 2001; Reynaert and Gelman, 2007; Jorm and Oh, 2009). The adjective “schizophrenic” is associated with perceptions of the permanence of the illness, and facilitates the cognitive separation between “us” and “them” central to the
112
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
process of stigma (Link and Phelan, 2001; Jorm and Oh, 2009). Thus, distinguishing the adjective and the noun forms of the illness is critical for alleviating stigma and advancing treatment (Link et al, 1989). Further, the adjective may facilitate internalization of one's illness, thus leading to self-stigmatizing attitudes, social withdrawal, and harmful psychological outcomes (Livingston and Boyd, 2010). Finally, humor and jokes in social interactions can propagate stereotypic imagery and marginalization of individuals with schizophrenia. Humor may reinforce hostility towards a targeted group and further macrosociological prejudice that maintains power imbalances between social groups (Ford and Ferguson, 2004). Scheff argues that humor indicates references to deviance from social norms, creating the moral ‘other’ that is stigmatizing to those with mental illness (Scheff, 1971). Humor may denigrate people suffering from schizophrenia, making their suffering less legitimate than the suffering of those with other medical conditions. Previous newspaper and survey studies about stigma do not capture the dynamic nature of stigma or the naturalistic use of the stigmatizing labels in day-to-day conversation as opposed to formalized print media (Duckworth et al, 2003; Chopra and Doody, 2007; Magliano et al., 2011; Vahabzadeh et al., 2011; Roberts et al., 2013; Thornicroft et al, 2013; Athanasopoulou and Välimäki, 2014; Cain et al, 2014). Previous studies contain a number of methodological problems including response-bias, anchoring effects (whereby participants rely too heavily on the first piece of information they observe to answer future questions), potential problems with the wording of questions for responses, problems with responders' comprehension of survey questions, and social desirability (the tendency for subjects to present themselves positively) (Podsakoff et al, 2003). We aim to address these shortcomings in previous stigma research through the social media technology. Social media technology offers a novel opportunity to examine perceptions and social attitudes about schizophrenia in a low cost and reproducible way. Twitter.com is a real-time microblogging website with over 241 million active users (Twitter, 2014). Users broadcast “Tweets” of 140-character length to their social network of other “Tweeters”. Utilizing this novel technology instead of traditional tools, such as surveys, to study stigma offers several potential advantages. First and foremost, Twitter allows examination of the colloquial use (and misuse) of the illness through naturalistic rather than formalized sampling observed in other media and vignette studies. Second, people may expect to receive responses to their Tweets, making messages dynamic and interactive with others. Third, Twitter allows users to not report their names or to report false names. This anonymity or quasi-anonymity may make people more willing to report sensitive issues on Twitter, including risky health behaviors that track closely to other measures of these behaviors (Prieto et al, 2014; Young et al., 2014). An agent's anonymity is perhaps more salient to the agent when there is greater physical distance from other agents, suggesting that users may be more willing to express stigmatizing beliefs online. Fourth, Twitter offers a platform for hard-to-reach minority groups such as African Americans and Hispanic Americans, who Tweet more frequently than other ethnic groups in certain parts of the United States. Finally, it is widely used among teens and young adults, a population at a higher risk for schizophrenia onset and for whom there remains a need to understand stigmatizing attitudes (Mislove et al, 2011). Twitter has been successfully applied in medical research, to track cholera outbreaks in Haiti, and in psychological research, to reliably predict personality traits (Chunara et al., 2012; Hughes et al, 2012). More recently, it has been utilized as a tool to study mental health conditions: a recent study demonstrated the ability of Twitter to track suicide risk across the United States and another helped estimate and track the incidence of depression (Jashinsky et al, 2013). It would be instructive to contrast stigma in complex nonpsychiatric medical conditions to stigma of schizophrenia. Diabetes is another severe chronic illness that has been used in this context (Lee et al, 2005; Georg Hsu et al., 2008). Both diabetes and schizophrenia are leading causes of global disability, are often undiagnosed, and have
high mortality (National Center for Health Statistics, 2012). They are both treatable medical illnesses of complex etiology with a number of misconceptions and negative stereotypes (for diabetes, being “fat”, “lazy”, or “lacking control”) (Mann et al, 2009; Browne et al, 2013). Both illnesses require self-management, which might be affected by public conceptions of the illness. Schizophrenia, compared to diabetes, might be viewed as having a more complex public manifestation due to its salient, and frequently more public, behavioral symptomatology of delusions, hallucinations, and sometimes more erratic and high risk behavior. Due to its complex etiology and variable public manifestation, schizophrenia may not only be associated with negative stereotypes of the illness specifically, but also be more likely to be associated with negative stereotypes of all sorts and variations (Fig. 1). We examined use of the word “schizophrenia” on Twitter compared to that of diabetes, analyzing Tweets that use the adjective and noun form of schizophrenia and diabetes. We ascertained three dimensions of each illness's linguistic use — namely medical appropriateness (as either medically appropriate, medically inappropriate, or non-medical references to the illness), negative valence, and humor sentiment (which is frequently negative, and therefore we use the term “sarcasm”). We hypothesized that both adjective and noun forms of schizophrenia would be more frequently used incorrectly, non-medically, more negatively, and more sarcastically compared to diabetes. We also hypothesized that negativity and sarcasm would be higher in medically inappropriate uses of the illness, and that these effects are particularly salient for the adjective form of schizophrenia. 2. Methods After receiving an Exemption from the ethics committee of the Beth Israel Deaconess Medical Center, a total of 1838 Tweets containing one of four hashtags (#schizophrenia, #schizophrenic, #diabetes, #diabetic) were randomly sampled from a data set containing Tweets over a forty day time period (12 September 2013 to 22 October 2013). We determined forty days to be long enough to account for any short-term changes in discussion on Twitter about the illnesses. Tweets were collected using the Twitter Application Programming Interface (API), which provides free programmatic access to Twitter data. Each Tweet was given one of three ratings under Medical appropriateness: a) medically appropriate, b) medically inappropriate, or c) non-medical. Medically appropriate Tweets include a medically valid reference to the illness or a feature of the illness (e.g. “people with schizophrenia may hallucinate”), or a person with the illness (e.g., “my brother has schizophrenia”). Medically inappropriate Tweets make direct reference to inaccurate facts about the illness, such as confusing multiple personality disorder with schizophrenia, confusing the symptoms of diabetes, or confusing Type I and Type II diabetes. Nonmedical references are Tweets that do not reference illness (e.g. using the illness to describe the weather). Tweets with negative sentiment were rated based on the presence of negative valence words within the Tweet — Tweets with the words “die”, “hate”, or “pain”, for instance, were flagged as negative. Sarcasm was evaluated on the basis of the presence of emoticons or direct references to laughing, sarcasm, or humor. For instance, Tweets with positive emoticons (e.g. “:)”) were flagged as containing sarcasm. Definitions for each category are summarized in Table 1. Of the Tweets with original user-input content, 79 were excluded on the basis of incomprehensibility (incomplete spellings or non-English language). 441 ‘Retweets’ – where one person writes a message and another can respond to the original message with or without additional content – were excluded from the analysis if without additional content. A total of 1318 Tweets were rated (354 for “#schizophrenic”, 331 for “#schizophrenia”, 284 for “#diabetes”, and 349 for “#diabetic”). Three inter-rater reliable raters (two research associates, AJ and NT, and one psychiatry resident, JT) rated Tweets for medical appropriateness, negative sentiment, and sarcasm (Table 2).
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
113
Fig. 1. Comparison of ratings across illness type (main effects).
To test reliability, raters reviewed an initial subset of 100 Tweets (not included in this analysis) to apply initial classifications of each category. Classifications were then clarified and refined to reflect experience in rating the initial classifications. Raters independently rated a new training set of 300 Tweets to which they were previously blinded. Reliability was .80 for negative sentiment, .93 for medical appropriateness, and .96 for sarcasm, thus establishing very high agreement. A rater blind to the purposes of the study scored a kappa of .77, .79, and 1.0 for medical appropriateness, negative sentiment, and sarcasm respectively, indicating good reliability. Chi square tests were conducted across illness type (schizophrenia vs. diabetes) for medical appropriateness, negative sentiment, and sarcasm. Differences in frequency between illnesses for each measure (“main
Table 1 Rater's criteria for the categorization of tweets. Medical Medically appropriateness appropriate
Negativity
Sarcasm
Schizophrenia: Hallucinations, delusions, cognitive decline, paranoia, anhedonia, medication use, a person with the illness Diabetes: Obesity, insulin, blood sugar, diet, exercise, type I vs. type II, neuropathy, foot ulcers, a person with the illness Medically Schizophrenia: inappropriate Confusion of schizophrenia with bipolar disorder or dissociative identity disorder Diabetes: Confusion of diabetes types (type I vs. type II), or inaccurate causes of diabetes (“drinking too much water”). Non-reference Using schizophrenia or diabetes to describe something besides a person with the illness or symptoms. Not Negative Lack of words with negative valence Negative Words with explicit negative valence, such as “hate”, “fear”, “die”, “terrible”, and “horrible”. Not sarcastic Lack of humorous phrases or emoticons. Sarcastic Containing: Positive emoticons: :), ;), :D, etc. “LOL”, “LMAO”, “haha”, etc.
effects”), and differences within each part of speech (noun vs. adjective) in negativity and sarcasm measures between each level of medical appropriateness (“dual effects”) are reported at p b .05 (two-tailed). 3. Results 3.1. Schizophrenia vs. diabetes comparisons (main effects) As hypothesized, chi squared tests showed statistically significant differences between the two illnesses (i.e., among “schizophrenia”, “schizophrenic”, “diabetes”, “diabetic”) along negative sentiment, medical appropriateness, and sarcasm. Proportions endorsed for negative sentiment across the four conditions significantly differed, with the proportion of #schizophrenic and #schizophrenia, appearing to be higher when compared with #diabetic and #diabetes. Likewise, proportions of medical appropriateness ratings across the four search terms were significantly different, with #schizophrenic and #schizophrenia appearing more often medically inaccurate than #diabetic and #diabetes. #Schizophrenic and #schizophrenia are more likely to be used non-medically to the illness compared to #diabetic and #diabetes. Proportions across sarcasm were likewise significantly different, with #schizophrenic Tweets showing higher rates of sarcasm than #schizophrenia, #diabetic, and #diabetes (Table 2) (Fig. 1). 3.2. Noun vs. adjective comparisons (dual effects) Chi square tests also showed statistically significant differences in proportions of negative sentiment for different medical appropriateness ratings (i.e., when statements were “medically inappropriate” vs. “medically appropriate” vs. “non-medical”) for schizophrenia and schizophrenic. #Schizophrenic appears most likely to be negative when medically inappropriate than when appropriate or non-medical. #Schizophrenia appears more likely to be negative when not a reference to the illness than when medically inappropriate or appropriate. Proportions of sarcasm were significantly different along each medical appropriateness rating for #schizophrenia, which appears most likely to be sarcastic when used non-medically than when inappropriately used or correctly used. Proportions of sarcasm significantly differed
114
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
Table 2 Comparison of ratings for schizophrenia and diabetes (main effects).
Negative Medically inaccurate Without reference Sarcastic
Schizophrenic (n = 354)
Schizophrenia (n = 331)
Diabetic (n = 284)
Diabetes (n = 284)
χ2
p Value
33% 30.1% 33.3% 14.4%
21.1% 6.6% 14.8% 1.8%
17.3% 3.2% 4.6% 2.5%
12.6% 2.9% 2.6% 2.6%
49.05 403.73 40.29 64.20
b0.001⁎ b0.001⁎ b.001⁎ b0.001⁎
⁎ p b .01 after Bonferroni correction.
across various negativity ratings for #Schizophrenic, with sarcasm more likely to appear in non-negative tweets than in negative tweets (Table 3). 4. Discussion Our findings confirm the presence of significant misuse of the term schizophrenia on Twitter, and that this misuse is considerably more pronounced by the adjectival form of the illness (“schizophrenic”). Schizophrenia was more often used without non-medically to the illness and more negatively than diabetes, confirming our hypotheses and previous observations about the use of schizophrenia in media (Link et al, 2004; Vahabzadeh et al., 2011). Our analysis suggests that negative, inappropriate, non-medical, and negative sentiments about schizophrenia are prevalent in social media. Our study is the first to examine sarcasm associated with use of the word schizophrenia in a naturalistic setting. A significantly higher rate of sarcasm was associated with the adjective “schizophrenic” compared to diabetes. Additionally, sarcasm was more frequent in the schizophrenia Tweets that used the term non-medically, suggesting that nonmedical usages may be associated in some circumstances with sarcastic/humorous, deviant behaviors. Sarcasm in #schizophrenic was more likely to be present in non-negative Tweets, also suggesting that sarcastic uses of the illness are distinct from negative references. Schizophrenia is thus doubly burdened by negative perceptions and denigrating humor. Humor can have an important role in stigmatizing processes, particularly in interpersonal and colloquial interactions. Humor and sarcastic jokes – used colloquially and formally – may contribute to stigma by belittling the suffering associated with schizophrenia, and also by stereotypic imagery and prejudice, which marginalize those with schizophrenia (Scheff, 1971). Future research will need to place a greater focus on such mechanisms. Our observations of the stigmatizing use of schizophrenia (i.e. negative, sarcastic, inappropriate use, and use without reference to the illness) being highest in the adjective form have important implications for labeling theory in mental health stigma. Labeling requires oversimplification of stereotypes associated with complex groups that may dynamically shift over time (Link and Phelan, 2001). Our findings suggest that current colloquial use of the adjective (one form of the label) may lead to misunderstanding, as observed in the higher frequency of
medical inappropriateness in the adjective than in the noun. Similarly, the adjective is more closely associated with negative sentiment than the noun, suggesting that the negative associations are strongest when schizophrenia is used as an adjective. Labels are theorized to create cognitive separation between a group and another marginalized group (Link and Phelan, 2001). As noted previously, humor can propagate stereotypic imagery. The adjective is selectively responsible for the association of schizophrenia with sarcasm. Ostensibly, the adjective – through humor mechanisms that carry stereotypic imagery – may facilitate the cognitive separation of “us” vs. “them” (“schizophrenics”). The inappropriate and non-medical uses of schizophrenia are connected to both negativity and sarcasm. #Schizophrenic was more likely to be negative when medically inappropriate than when appropriate or not a reference, suggesting that the illness is more stigmatizing when poorly understood, or that negative beliefs about the illness lead individuals to misuse schizophrenia as something other than the medical diagnosis. Our findings suggest that the illness is more stigmatizing when used non-medically than when referring to the illness itself. #Schizophrenia was most likely to be sarcastic when not a reference than when inappropriately used or correctly used. Overall, these findings suggest that use of the word is drifting away from appropriate medical usages, perhaps an indication of stigma in that it trivializes the debilitating symptoms of major mental illness, and that such trivialization is unlikely in another major illness (diabetes). The drift away from medical uses may obscure the actual symptoms of schizophrenia, furthering the disconnect between the linguistic use and the medical meaning. Confusion about the symptoms of schizophrenia may impede treatment-seeking behaviors and diagnosis. A key difference between schizophrenia and diabetes is the tradition of using the word “schizophrenia” to reference a “madness” of some kind as opposed to the disease entity that it denominates in psychiatry (Schomerus et al, 2007). Schizophrenia carries more meaning than a strictly medical term like diabetes. Therefore, the finding in this study that schizophrenia on Twitter drifts away from medical uses more so than diabetes reflects the dual tradition in the word “schizophrenia” that is absent in “diabetes.” Our findings on sarcasm may also be alternatively explained. Humor, in some instances, reduces stigmatizing beliefs (Corrigan et al, 2014). The sarcastic sentiments that we noted in #schizophrenic may be an
Table 3 Comparison of ratings for noun and adjective forms of schizophrenia and diabetes (dual effects).
Negative
Sarcastic
Sarcastic
Schizophrenic (n = 118) Schizophrenia (n = 70) Schizophrenic (n = 51) Schizophrenia (n = 6)
Schizophrenic (n = 51) Schizophrenia (n = 6)
⁎ p b .05 after Bonferroni correction.
Medically inappropriate
Appropriate
Without reference
χ2
p Value
41.7%
33.6%
25.4%
6.70
b0.05
27.3%
18.1%
34.7%
7.36
b0.05
9.4% 0%
17.6% 1.2%
16.9% 6.1%
NS 6.15
NS p b .05
Negative
Non-negative
χ2
p Value
6.8%
18.2%
7.45
b0.01⁎
0.7%
5.7%
5.07
b0.05⁎
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
attempt to relieve cognitive dissonance while discussing a stigmatized illness. One limitation is the generalizability of our study due to the Twitter demographic, which is not representative of the general population. Twitter users are more likely to come from populous counties, are more likely male, computer literate, and younger (Mislove et al, 2011). However greater insight is needed to understand the attitudes and beliefs of high-risk youth, making the Twitter population ideal for stigma research (Link et al, 2004). We are unsure if the individuals with mental illness or unaffected community members are overrepresented among the Tweeters. Furthermore, stigma is a cultural process that may be shaped by local cultural views (Yang et al., 2007). Possibly, Twitter users form their own micro-culture that affects stigmatizing use of the illness label unique from offline interactions. While the two raters were not blind to the purposes of the study, a blind rater established agreement with the two raters on a subset of Tweets, suggesting that the stringent rating criteria prevented bias. Finally, we analyzed the hashtags specifically (rather than using a more general query on Twitter), though there is a possibility that this may have led to a sampling bias that does not fully capture the variation among Twitter users. The findings in this study have several contributions and implications for stigma research. Our study suggests that Twitter data reflects public misinformation about schizophrenia in comparison to diabetes. Future studies can examine location and user data embedded in Tweets to inform the deployment of anti-stigma campaigns by examining geographical differences in stigmatizing perceptions at a low cost. Further, the terms “schizophrenia” and “schizophrenic” are often used nonmedically in a population that is at high risk for mental illness and that such non-medical usage is associated with negativity and sarcasm. Overall, the results reflect the dual meaning of “schizophrenia,” poor education, and lack of public awareness about the seriousness of mental illness, and that this lack of awareness could perhaps persist in a younger, online, generation. Role of funding source None. Contributors Authors AJ, NT, and MK designed the study. Author AJ wrote the first draft of the manuscript. Three authors, AJ, JT, and NT, conducted the rating of Tweets. All authors contributed to the analysis and interpretation of the data, and have approved the final manuscript. Conflict of interest The authors have no conflicts of interest to report. Acknowledgments We would like to thank Ian Mathew, Ewan Foster, Jai Shah, Ned Ohringer, and Anne Harrington for their contributions to this paper.
References Angermeyer, M.C., Dietrich, S., 2006. Public beliefs about and attitudes towards people with mental illness: a review of population studies. Acta Psychiatr. Scand. 113, 163–179. Athanasopoulou, C., Välimäki, M., 2014. ‘Schizophrenia’ as a metaphor in Greek newspaper websites. Stud. Health Technol. Inf. 202, 275–278. Browne, J.L., Ventura, A., Mosely, K., Speight, J., 2013. “I call it the blame and shame disease”: a qualitative study about perceptions of social stigma surrounding type 2 diabetes. BMJ Open 3, e003384. Cain, B., Currie, R., Danks, E., Du, F., Hodgson, E., May, J., N., et al., 2014. “Schizophrenia” in the Australian print and online news media. Psychosis 6 (2), 97–106. Chopra, A.K., Doody, G.A., 2007. Schizophrenia, an illness and a metaphor: analysis of the use of the term “schizophrenia” in the UK national newspapers. J. R. Soc. Med. 100, 423–426. Chunara, R., Andrews, J.R., Brownstein, J.S., 2012. Social and news media enable estimation of epidemiological patterns early in the 2010 Haitian cholera outbreak. Am. J. Trop. Med. Hyg. 86, 39–45.
115
Corrigan, P.W., 2004. How stigma interferes with mental health care. Am. Psychol. 59, 614–6253. Corrigan, P.W., Powell, K.J., Fokuo, J.K., Kosyluk, K.A., 2014. Does humor influence the stigma of mental illnesses? J. Nerv. Ment. Dis. 202, 397–401. Duckworth, K., Halpern, J.H., Schutt, R.K., Gillespie, C., 2003. Use of schizophrenia as a metaphor in US newspapers. Psychiatr. Serv. 54, 1402–1404. Ford, T.E., Ferguson, M.A., 2004. Social consequences of disparagement humor: a prejudiced norm theory. Personal. Soc. Psychol. Rev. 8, 79–94. Georg Hsu, L.K., Wan, Y.M., Chang, H., Summergrad, P., Tsang, B.Y.P., Chen, H., 2008. Stigma of depression is more severe in Chinese Americans than Caucasian Americans. Psychiatry Interpers. Biol. Process. 71, 210–218. Goffman, E., 2009. Stigma: Notes on the Management of Spoiled Identity. Simon and Schuster. Hughes, D.J., Rowe, M., Batey, M., Lee, A., 2012. A tale of two sites: Twitter vs. Facebook and the personality predictors of social media usage. Comput. Hum. Behav. 28, 561–569. Jashinsky, J., Burton, S.H., Hanson, C.L., West, J., Giraud-Carrier, C., Barnes, M.D., Argyle, T., 2013. Tracking suicide risk factors through Twitter in the US. Crisis 35, 51–59. Jorm, A.F., Oh, E., 2009. Desire for social distance from people with mental disorders. Aust. NZ J. Psychiatry 43, 183–200. Lee, S., Lee, M., Chiu, M., Kleinman, A., 2005. Experience of social stigma by people with schizophrenia in Hong Kong. Br. J. Psychiatry 186, 153–157. Lincoln, T.M., Arens, E., Berger, C., Rief, W., 2008. Can antistigma campaigns be improved? A test of the impact of biogenetic vs psychosocial causal explanations on implicit and explicit attitudes to schizophrenia. Schizophr. Bull. 34, 984–994. Link, B.G., Phelan, J.C., 2001. Conceptualizing stigma. Annu. Rev. Sociol. 363–385. Link, B.G., Cullen, F.T., Struening, E., Shrout, P.E., Dohrenwend, B.P., 1989. A modified labeling theory approach to mental disorders: an empirical assessment. Am. Sociol. Rev. 54, 400–423. Link, B.G., Yang, L.H., Phelan, J.C., Collins, P.Y., 2004. Measuring mental illness stigma. Schizophr. Bull. 30, 511–541. Livingston, J.D., Boyd, J.E., 2010. Correlates and consequences of internalized stigma for people living with mental illness: a systematic review and meta-analysis. Soc. Sci. Med. 71, 2150–2161. Magliano, L., Read, J., Marassi, R., 2011. Metaphoric and non-metaphoric use of the term “schizophrenia” in Italian newspapers. Soc. Psychiatry Psychiatr. Epidemiol. 46, 1019–1025. Mann, D.M., Ponieman, D., Leventhal, H., Halm, E.A., 2009. Misconceptions about diabetes and its management among low-income minorities with diabetes. Diabetes Care 32, 591–593. Mislove, A., Lehmann, S., Ahn, Y.Y., Onnela, J.P., 2011. Understanding the Demographics of Twitter Users. ICWSM. National Center for Health Statistics (US): Health, United States, 2012: With Special Feature on Emergency Care. Hyattsville (MD): National Center for Health Statistics (US); 2013. Penn, D.L., Nowlin-Drummond, A., 2001. Politically correct labels and schizophrenia: a rose by any other name? Schizophr. Bull. 27, 197–203. Perkins, D.O., Gu, H., Boteva, K., Lieberman, J.A., 2005. Relationship between duration of untreated psychosis and outcome in first-episode schizophrenia: a critical review and meta-analysis. Am. J. Psychiatry 162, 1785–1804. Pescosolido, B., Martin, J., Long, J., Medina, T., Phelan, J., Link, B., 2010. “A disease like any other”? A decade of change in public reactions to schizophrenia, depression, and alcohol dependence. Am. J. Psychiatr. 167 (11), 1321–1330. Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., Podsakoff, N.P., 2003. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88, 879–903. Prieto, V.M., Matos, S., Álvarez, M., Cacheda, F., Oliveira, J.L., 2014. Twitter: a good place to detect health conditions. PLoS One 9, e86191. Reynaert, C.C., Gelman, S.A., 2007. The influence of language form and conventional wording on judgments of illness. J. Psycholinguist. Res. 36, 273–295. Roberts, E., Bourne, R., Basden, S., 2013. The representation of mental illness in Bermudian print media, 1991–2011. Psychiatr. Serv. 64 (388–39). Rüsch, N., Angermeyer, M.C., Corrigan, P.W., 2005. Mental illness stigma: concepts, consequences, and initiatives to reduce stigma. Eur. Psychiatry 20, 529–539. Scheff, T.J., 1971. Being Mentally Ill. Transaction Publishers. Schomerus, G., Kenzin, D., Borsche, J., Matschinger, H., Angermeyer, M.C., 2007. The association of schizophrenia with split personality is not an ubiquitous phenomenon. Soc. Psychiatry Psychiatr. Epidemiol. 42 (10), 780–786. Sontag, S., 2013. Illness as Metaphor and AIDS and its Metaphors. Penguin, U.K. Thornicroft, A., Goulden, R., Shefer, G., Rhydderch, D., Rose, D., Williams, P., Thornicroft, G., Henderson, C., 2013. Newspaper coverage of mental illness in England 2008–2011. Br. J. Psychiatry 202, s64–s69. Twitter, 2014. About Twitter, Inc. [Online]. Available at:, https://about.twitter.com/ company (Accessed 16 Jul. 2014). Vahabzadeh, A., Wittenauer, J., Carr, E., 2011. Stigma, schizophrenia and the media: exploring changes in the reporting of schizophrenia in major US newspapers. J. Psychiatr. Pract. 17 (6), 439–446. Yang, L.H., Kleinman, A., Link, B.G., Phelan, J.C., Lee, S., Good, B., 2007. Culture and stigma: adding moral experience to stigma theory. Soc. Sci. Med. 64, 1524–1535. Young, S.D., Rivers, C., Lewis, B., 2014. Methods of using real-time social media technologies for detection and remote monitoring of HIV outcomes. Prev. Med. 63, 112–115.
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
#Schizophrenia: Use and misuse on Twitter Adam J. Joseph a, Neeraj Tandon a, Lawrence H. Yang b, Ken Duckworth c, John Torous a, Larry J. Seidman a, Matcheri S. Keshavan a,⁎ a b c
Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, United States Department of Epidemiology, Columbia University, New York, NY, United States National Alliance on Mental Illness, Arlington, VA, United States
a r t i c l e
i n f o
Article history: Received 8 December 2014 Received in revised form 1 April 2015 Accepted 9 April 2015 Available online 30 April 2015 Keywords: Stigma Social media Twitter Schizophrenia
a b s t r a c t Background: The role and prevention of stigma in mental illness is an area of evolving research. Aims: The present study is the first to examine the use and misuse of the word ‘schizophrenia’ on Twitter.com in comparison with another illness (diabetes) by analyzing Tweets that use the adjective and noun forms of schizophrenia and diabetes. Method: Tweets containing one of four search terms (#schizophrenia, #schizophrenic, #diabetes, #diabetic) were collected over a forty-day time period. After establishing inter-rater reliability, Tweets were rated along three dimensions: medical appropriateness, negativity, and sarcasm. Chi square tests were conducted to examine differences in the distributions of each parameter across illnesses and across each word form (noun versus adjective). Results: Significant differences were seen between the two illnesses (i.e., among “schizophrenia”, “schizophrenic”, “diabetes”, and “diabetic”) along each parameter. Tweets about schizophrenia were more likely to be negative, medically inappropriate, sarcastic, and used non-medically. The adjective (“schizophrenic”) was more often negative, medically inappropriate, sarcastic, and used non-medically than the noun “schizophrenia.” Schizophrenia tweets were more likely to be negative and sarcastic when used non-medically and in a medically inappropriate manner. Conclusions: Our findings confirm the presence of a great deal of misuse of the term schizophrenia on Twitter, and that this misuse is considerably more pronounced by the adjectival use of the illness. These findings have considerable implications for efforts to combat stigma, particularly for youth anti-stigma efforts. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Stigma – a highly discrediting attribute that reduces the stigmatized person “from a whole and usual person to a tainted, discounted one” – is an area of evolving research in the schizophrenia literature (Goffman, 2009). Stigmatizing attitudes and beliefs about schizophrenia (and other mental illnesses) decrease treatment seeking behaviors and medication adherence (Corrigan, 2004). Survey, vignette, and interview studies have found that community respondents perceive those with schizophrenia as unpredictable and dangerous. Respondents generally view schizophrenia as more negative and stigmatizing than other mental illnesses, such as depression or anxiety (Angermeyer and Dietrich, 2006; Pescosolido et al, 2010). Early identification of schizophrenia is critical since treatment delays are associated with worse
⁎ Corresponding author at: Room 610, Massachusetts Mental Health Center, 75 Fenwood Road, Boston MA 02115, United States. Tel.: +1 617 754 1256; fax: +1 617 754 1250. E-mail addresses: [email protected], [email protected] (M.S. Keshavan).
http://dx.doi.org/10.1016/j.schres.2015.04.009 0920-9964/© 2015 Elsevier B.V. All rights reserved.
clinical outcomes (Perkins et al, 2005). Therefore, stigmatizing beliefs held among youth are particularly important for early detection and treatment of schizophrenia. The formation of stigmatizing beliefs deserves consideration due to stigma's role in diverting treatment. Illnesses with complex etiology such as schizophrenia are particularly likely to be stigmatized (Sontag, 2013). Studies suggest that various medically inappropriate causal explanations for mental illness can increase stigmatizing perceptions of schizophrenia (Lincoln et al., 2008). Increased mental health literacy may also reduce negative, stigmatizing behaviors and beliefs (Rüsch et al., 2005). Stigma also results from labeling, which entails oversimplification of the characteristics of groups and association of the label with negative attributes (Link and Phelan, 2001). Consequently, the adjective form of schizophrenia (“schizophrenic”) generates greater negative emotions, including perceptions of dangerousness and greater desire for social distance from the person with the illness, than does the noun “schizophrenia” (Penn and Nowlin-Drummond, 2001; Reynaert and Gelman, 2007; Jorm and Oh, 2009). The adjective “schizophrenic” is associated with perceptions of the permanence of the illness, and facilitates the cognitive separation between “us” and “them” central to the
112
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
process of stigma (Link and Phelan, 2001; Jorm and Oh, 2009). Thus, distinguishing the adjective and the noun forms of the illness is critical for alleviating stigma and advancing treatment (Link et al, 1989). Further, the adjective may facilitate internalization of one's illness, thus leading to self-stigmatizing attitudes, social withdrawal, and harmful psychological outcomes (Livingston and Boyd, 2010). Finally, humor and jokes in social interactions can propagate stereotypic imagery and marginalization of individuals with schizophrenia. Humor may reinforce hostility towards a targeted group and further macrosociological prejudice that maintains power imbalances between social groups (Ford and Ferguson, 2004). Scheff argues that humor indicates references to deviance from social norms, creating the moral ‘other’ that is stigmatizing to those with mental illness (Scheff, 1971). Humor may denigrate people suffering from schizophrenia, making their suffering less legitimate than the suffering of those with other medical conditions. Previous newspaper and survey studies about stigma do not capture the dynamic nature of stigma or the naturalistic use of the stigmatizing labels in day-to-day conversation as opposed to formalized print media (Duckworth et al, 2003; Chopra and Doody, 2007; Magliano et al., 2011; Vahabzadeh et al., 2011; Roberts et al., 2013; Thornicroft et al, 2013; Athanasopoulou and Välimäki, 2014; Cain et al, 2014). Previous studies contain a number of methodological problems including response-bias, anchoring effects (whereby participants rely too heavily on the first piece of information they observe to answer future questions), potential problems with the wording of questions for responses, problems with responders' comprehension of survey questions, and social desirability (the tendency for subjects to present themselves positively) (Podsakoff et al, 2003). We aim to address these shortcomings in previous stigma research through the social media technology. Social media technology offers a novel opportunity to examine perceptions and social attitudes about schizophrenia in a low cost and reproducible way. Twitter.com is a real-time microblogging website with over 241 million active users (Twitter, 2014). Users broadcast “Tweets” of 140-character length to their social network of other “Tweeters”. Utilizing this novel technology instead of traditional tools, such as surveys, to study stigma offers several potential advantages. First and foremost, Twitter allows examination of the colloquial use (and misuse) of the illness through naturalistic rather than formalized sampling observed in other media and vignette studies. Second, people may expect to receive responses to their Tweets, making messages dynamic and interactive with others. Third, Twitter allows users to not report their names or to report false names. This anonymity or quasi-anonymity may make people more willing to report sensitive issues on Twitter, including risky health behaviors that track closely to other measures of these behaviors (Prieto et al, 2014; Young et al., 2014). An agent's anonymity is perhaps more salient to the agent when there is greater physical distance from other agents, suggesting that users may be more willing to express stigmatizing beliefs online. Fourth, Twitter offers a platform for hard-to-reach minority groups such as African Americans and Hispanic Americans, who Tweet more frequently than other ethnic groups in certain parts of the United States. Finally, it is widely used among teens and young adults, a population at a higher risk for schizophrenia onset and for whom there remains a need to understand stigmatizing attitudes (Mislove et al, 2011). Twitter has been successfully applied in medical research, to track cholera outbreaks in Haiti, and in psychological research, to reliably predict personality traits (Chunara et al., 2012; Hughes et al, 2012). More recently, it has been utilized as a tool to study mental health conditions: a recent study demonstrated the ability of Twitter to track suicide risk across the United States and another helped estimate and track the incidence of depression (Jashinsky et al, 2013). It would be instructive to contrast stigma in complex nonpsychiatric medical conditions to stigma of schizophrenia. Diabetes is another severe chronic illness that has been used in this context (Lee et al, 2005; Georg Hsu et al., 2008). Both diabetes and schizophrenia are leading causes of global disability, are often undiagnosed, and have
high mortality (National Center for Health Statistics, 2012). They are both treatable medical illnesses of complex etiology with a number of misconceptions and negative stereotypes (for diabetes, being “fat”, “lazy”, or “lacking control”) (Mann et al, 2009; Browne et al, 2013). Both illnesses require self-management, which might be affected by public conceptions of the illness. Schizophrenia, compared to diabetes, might be viewed as having a more complex public manifestation due to its salient, and frequently more public, behavioral symptomatology of delusions, hallucinations, and sometimes more erratic and high risk behavior. Due to its complex etiology and variable public manifestation, schizophrenia may not only be associated with negative stereotypes of the illness specifically, but also be more likely to be associated with negative stereotypes of all sorts and variations (Fig. 1). We examined use of the word “schizophrenia” on Twitter compared to that of diabetes, analyzing Tweets that use the adjective and noun form of schizophrenia and diabetes. We ascertained three dimensions of each illness's linguistic use — namely medical appropriateness (as either medically appropriate, medically inappropriate, or non-medical references to the illness), negative valence, and humor sentiment (which is frequently negative, and therefore we use the term “sarcasm”). We hypothesized that both adjective and noun forms of schizophrenia would be more frequently used incorrectly, non-medically, more negatively, and more sarcastically compared to diabetes. We also hypothesized that negativity and sarcasm would be higher in medically inappropriate uses of the illness, and that these effects are particularly salient for the adjective form of schizophrenia. 2. Methods After receiving an Exemption from the ethics committee of the Beth Israel Deaconess Medical Center, a total of 1838 Tweets containing one of four hashtags (#schizophrenia, #schizophrenic, #diabetes, #diabetic) were randomly sampled from a data set containing Tweets over a forty day time period (12 September 2013 to 22 October 2013). We determined forty days to be long enough to account for any short-term changes in discussion on Twitter about the illnesses. Tweets were collected using the Twitter Application Programming Interface (API), which provides free programmatic access to Twitter data. Each Tweet was given one of three ratings under Medical appropriateness: a) medically appropriate, b) medically inappropriate, or c) non-medical. Medically appropriate Tweets include a medically valid reference to the illness or a feature of the illness (e.g. “people with schizophrenia may hallucinate”), or a person with the illness (e.g., “my brother has schizophrenia”). Medically inappropriate Tweets make direct reference to inaccurate facts about the illness, such as confusing multiple personality disorder with schizophrenia, confusing the symptoms of diabetes, or confusing Type I and Type II diabetes. Nonmedical references are Tweets that do not reference illness (e.g. using the illness to describe the weather). Tweets with negative sentiment were rated based on the presence of negative valence words within the Tweet — Tweets with the words “die”, “hate”, or “pain”, for instance, were flagged as negative. Sarcasm was evaluated on the basis of the presence of emoticons or direct references to laughing, sarcasm, or humor. For instance, Tweets with positive emoticons (e.g. “:)”) were flagged as containing sarcasm. Definitions for each category are summarized in Table 1. Of the Tweets with original user-input content, 79 were excluded on the basis of incomprehensibility (incomplete spellings or non-English language). 441 ‘Retweets’ – where one person writes a message and another can respond to the original message with or without additional content – were excluded from the analysis if without additional content. A total of 1318 Tweets were rated (354 for “#schizophrenic”, 331 for “#schizophrenia”, 284 for “#diabetes”, and 349 for “#diabetic”). Three inter-rater reliable raters (two research associates, AJ and NT, and one psychiatry resident, JT) rated Tweets for medical appropriateness, negative sentiment, and sarcasm (Table 2).
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
113
Fig. 1. Comparison of ratings across illness type (main effects).
To test reliability, raters reviewed an initial subset of 100 Tweets (not included in this analysis) to apply initial classifications of each category. Classifications were then clarified and refined to reflect experience in rating the initial classifications. Raters independently rated a new training set of 300 Tweets to which they were previously blinded. Reliability was .80 for negative sentiment, .93 for medical appropriateness, and .96 for sarcasm, thus establishing very high agreement. A rater blind to the purposes of the study scored a kappa of .77, .79, and 1.0 for medical appropriateness, negative sentiment, and sarcasm respectively, indicating good reliability. Chi square tests were conducted across illness type (schizophrenia vs. diabetes) for medical appropriateness, negative sentiment, and sarcasm. Differences in frequency between illnesses for each measure (“main
Table 1 Rater's criteria for the categorization of tweets. Medical Medically appropriateness appropriate
Negativity
Sarcasm
Schizophrenia: Hallucinations, delusions, cognitive decline, paranoia, anhedonia, medication use, a person with the illness Diabetes: Obesity, insulin, blood sugar, diet, exercise, type I vs. type II, neuropathy, foot ulcers, a person with the illness Medically Schizophrenia: inappropriate Confusion of schizophrenia with bipolar disorder or dissociative identity disorder Diabetes: Confusion of diabetes types (type I vs. type II), or inaccurate causes of diabetes (“drinking too much water”). Non-reference Using schizophrenia or diabetes to describe something besides a person with the illness or symptoms. Not Negative Lack of words with negative valence Negative Words with explicit negative valence, such as “hate”, “fear”, “die”, “terrible”, and “horrible”. Not sarcastic Lack of humorous phrases or emoticons. Sarcastic Containing: Positive emoticons: :), ;), :D, etc. “LOL”, “LMAO”, “haha”, etc.
effects”), and differences within each part of speech (noun vs. adjective) in negativity and sarcasm measures between each level of medical appropriateness (“dual effects”) are reported at p b .05 (two-tailed). 3. Results 3.1. Schizophrenia vs. diabetes comparisons (main effects) As hypothesized, chi squared tests showed statistically significant differences between the two illnesses (i.e., among “schizophrenia”, “schizophrenic”, “diabetes”, “diabetic”) along negative sentiment, medical appropriateness, and sarcasm. Proportions endorsed for negative sentiment across the four conditions significantly differed, with the proportion of #schizophrenic and #schizophrenia, appearing to be higher when compared with #diabetic and #diabetes. Likewise, proportions of medical appropriateness ratings across the four search terms were significantly different, with #schizophrenic and #schizophrenia appearing more often medically inaccurate than #diabetic and #diabetes. #Schizophrenic and #schizophrenia are more likely to be used non-medically to the illness compared to #diabetic and #diabetes. Proportions across sarcasm were likewise significantly different, with #schizophrenic Tweets showing higher rates of sarcasm than #schizophrenia, #diabetic, and #diabetes (Table 2) (Fig. 1). 3.2. Noun vs. adjective comparisons (dual effects) Chi square tests also showed statistically significant differences in proportions of negative sentiment for different medical appropriateness ratings (i.e., when statements were “medically inappropriate” vs. “medically appropriate” vs. “non-medical”) for schizophrenia and schizophrenic. #Schizophrenic appears most likely to be negative when medically inappropriate than when appropriate or non-medical. #Schizophrenia appears more likely to be negative when not a reference to the illness than when medically inappropriate or appropriate. Proportions of sarcasm were significantly different along each medical appropriateness rating for #schizophrenia, which appears most likely to be sarcastic when used non-medically than when inappropriately used or correctly used. Proportions of sarcasm significantly differed
114
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
Table 2 Comparison of ratings for schizophrenia and diabetes (main effects).
Negative Medically inaccurate Without reference Sarcastic
Schizophrenic (n = 354)
Schizophrenia (n = 331)
Diabetic (n = 284)
Diabetes (n = 284)
χ2
p Value
33% 30.1% 33.3% 14.4%
21.1% 6.6% 14.8% 1.8%
17.3% 3.2% 4.6% 2.5%
12.6% 2.9% 2.6% 2.6%
49.05 403.73 40.29 64.20
b0.001⁎ b0.001⁎ b.001⁎ b0.001⁎
⁎ p b .01 after Bonferroni correction.
across various negativity ratings for #Schizophrenic, with sarcasm more likely to appear in non-negative tweets than in negative tweets (Table 3). 4. Discussion Our findings confirm the presence of significant misuse of the term schizophrenia on Twitter, and that this misuse is considerably more pronounced by the adjectival form of the illness (“schizophrenic”). Schizophrenia was more often used without non-medically to the illness and more negatively than diabetes, confirming our hypotheses and previous observations about the use of schizophrenia in media (Link et al, 2004; Vahabzadeh et al., 2011). Our analysis suggests that negative, inappropriate, non-medical, and negative sentiments about schizophrenia are prevalent in social media. Our study is the first to examine sarcasm associated with use of the word schizophrenia in a naturalistic setting. A significantly higher rate of sarcasm was associated with the adjective “schizophrenic” compared to diabetes. Additionally, sarcasm was more frequent in the schizophrenia Tweets that used the term non-medically, suggesting that nonmedical usages may be associated in some circumstances with sarcastic/humorous, deviant behaviors. Sarcasm in #schizophrenic was more likely to be present in non-negative Tweets, also suggesting that sarcastic uses of the illness are distinct from negative references. Schizophrenia is thus doubly burdened by negative perceptions and denigrating humor. Humor can have an important role in stigmatizing processes, particularly in interpersonal and colloquial interactions. Humor and sarcastic jokes – used colloquially and formally – may contribute to stigma by belittling the suffering associated with schizophrenia, and also by stereotypic imagery and prejudice, which marginalize those with schizophrenia (Scheff, 1971). Future research will need to place a greater focus on such mechanisms. Our observations of the stigmatizing use of schizophrenia (i.e. negative, sarcastic, inappropriate use, and use without reference to the illness) being highest in the adjective form have important implications for labeling theory in mental health stigma. Labeling requires oversimplification of stereotypes associated with complex groups that may dynamically shift over time (Link and Phelan, 2001). Our findings suggest that current colloquial use of the adjective (one form of the label) may lead to misunderstanding, as observed in the higher frequency of
medical inappropriateness in the adjective than in the noun. Similarly, the adjective is more closely associated with negative sentiment than the noun, suggesting that the negative associations are strongest when schizophrenia is used as an adjective. Labels are theorized to create cognitive separation between a group and another marginalized group (Link and Phelan, 2001). As noted previously, humor can propagate stereotypic imagery. The adjective is selectively responsible for the association of schizophrenia with sarcasm. Ostensibly, the adjective – through humor mechanisms that carry stereotypic imagery – may facilitate the cognitive separation of “us” vs. “them” (“schizophrenics”). The inappropriate and non-medical uses of schizophrenia are connected to both negativity and sarcasm. #Schizophrenic was more likely to be negative when medically inappropriate than when appropriate or not a reference, suggesting that the illness is more stigmatizing when poorly understood, or that negative beliefs about the illness lead individuals to misuse schizophrenia as something other than the medical diagnosis. Our findings suggest that the illness is more stigmatizing when used non-medically than when referring to the illness itself. #Schizophrenia was most likely to be sarcastic when not a reference than when inappropriately used or correctly used. Overall, these findings suggest that use of the word is drifting away from appropriate medical usages, perhaps an indication of stigma in that it trivializes the debilitating symptoms of major mental illness, and that such trivialization is unlikely in another major illness (diabetes). The drift away from medical uses may obscure the actual symptoms of schizophrenia, furthering the disconnect between the linguistic use and the medical meaning. Confusion about the symptoms of schizophrenia may impede treatment-seeking behaviors and diagnosis. A key difference between schizophrenia and diabetes is the tradition of using the word “schizophrenia” to reference a “madness” of some kind as opposed to the disease entity that it denominates in psychiatry (Schomerus et al, 2007). Schizophrenia carries more meaning than a strictly medical term like diabetes. Therefore, the finding in this study that schizophrenia on Twitter drifts away from medical uses more so than diabetes reflects the dual tradition in the word “schizophrenia” that is absent in “diabetes.” Our findings on sarcasm may also be alternatively explained. Humor, in some instances, reduces stigmatizing beliefs (Corrigan et al, 2014). The sarcastic sentiments that we noted in #schizophrenic may be an
Table 3 Comparison of ratings for noun and adjective forms of schizophrenia and diabetes (dual effects).
Negative
Sarcastic
Sarcastic
Schizophrenic (n = 118) Schizophrenia (n = 70) Schizophrenic (n = 51) Schizophrenia (n = 6)
Schizophrenic (n = 51) Schizophrenia (n = 6)
⁎ p b .05 after Bonferroni correction.
Medically inappropriate
Appropriate
Without reference
χ2
p Value
41.7%
33.6%
25.4%
6.70
b0.05
27.3%
18.1%
34.7%
7.36
b0.05
9.4% 0%
17.6% 1.2%
16.9% 6.1%
NS 6.15
NS p b .05
Negative
Non-negative
χ2
p Value
6.8%
18.2%
7.45
b0.01⁎
0.7%
5.7%
5.07
b0.05⁎
A.J. Joseph et al. / Schizophrenia Research 165 (2015) 111–115
attempt to relieve cognitive dissonance while discussing a stigmatized illness. One limitation is the generalizability of our study due to the Twitter demographic, which is not representative of the general population. Twitter users are more likely to come from populous counties, are more likely male, computer literate, and younger (Mislove et al, 2011). However greater insight is needed to understand the attitudes and beliefs of high-risk youth, making the Twitter population ideal for stigma research (Link et al, 2004). We are unsure if the individuals with mental illness or unaffected community members are overrepresented among the Tweeters. Furthermore, stigma is a cultural process that may be shaped by local cultural views (Yang et al., 2007). Possibly, Twitter users form their own micro-culture that affects stigmatizing use of the illness label unique from offline interactions. While the two raters were not blind to the purposes of the study, a blind rater established agreement with the two raters on a subset of Tweets, suggesting that the stringent rating criteria prevented bias. Finally, we analyzed the hashtags specifically (rather than using a more general query on Twitter), though there is a possibility that this may have led to a sampling bias that does not fully capture the variation among Twitter users. The findings in this study have several contributions and implications for stigma research. Our study suggests that Twitter data reflects public misinformation about schizophrenia in comparison to diabetes. Future studies can examine location and user data embedded in Tweets to inform the deployment of anti-stigma campaigns by examining geographical differences in stigmatizing perceptions at a low cost. Further, the terms “schizophrenia” and “schizophrenic” are often used nonmedically in a population that is at high risk for mental illness and that such non-medical usage is associated with negativity and sarcasm. Overall, the results reflect the dual meaning of “schizophrenia,” poor education, and lack of public awareness about the seriousness of mental illness, and that this lack of awareness could perhaps persist in a younger, online, generation. Role of funding source None. Contributors Authors AJ, NT, and MK designed the study. Author AJ wrote the first draft of the manuscript. Three authors, AJ, JT, and NT, conducted the rating of Tweets. All authors contributed to the analysis and interpretation of the data, and have approved the final manuscript. Conflict of interest The authors have no conflicts of interest to report. Acknowledgments We would like to thank Ian Mathew, Ewan Foster, Jai Shah, Ned Ohringer, and Anne Harrington for their contributions to this paper.
References Angermeyer, M.C., Dietrich, S., 2006. Public beliefs about and attitudes towards people with mental illness: a review of population studies. Acta Psychiatr. Scand. 113, 163–179. Athanasopoulou, C., Välimäki, M., 2014. ‘Schizophrenia’ as a metaphor in Greek newspaper websites. Stud. Health Technol. Inf. 202, 275–278. Browne, J.L., Ventura, A., Mosely, K., Speight, J., 2013. “I call it the blame and shame disease”: a qualitative study about perceptions of social stigma surrounding type 2 diabetes. BMJ Open 3, e003384. Cain, B., Currie, R., Danks, E., Du, F., Hodgson, E., May, J., N., et al., 2014. “Schizophrenia” in the Australian print and online news media. Psychosis 6 (2), 97–106. Chopra, A.K., Doody, G.A., 2007. Schizophrenia, an illness and a metaphor: analysis of the use of the term “schizophrenia” in the UK national newspapers. J. R. Soc. Med. 100, 423–426. Chunara, R., Andrews, J.R., Brownstein, J.S., 2012. Social and news media enable estimation of epidemiological patterns early in the 2010 Haitian cholera outbreak. Am. J. Trop. Med. Hyg. 86, 39–45.
115
Corrigan, P.W., 2004. How stigma interferes with mental health care. Am. Psychol. 59, 614–6253. Corrigan, P.W., Powell, K.J., Fokuo, J.K., Kosyluk, K.A., 2014. Does humor influence the stigma of mental illnesses? J. Nerv. Ment. Dis. 202, 397–401. Duckworth, K., Halpern, J.H., Schutt, R.K., Gillespie, C., 2003. Use of schizophrenia as a metaphor in US newspapers. Psychiatr. Serv. 54, 1402–1404. Ford, T.E., Ferguson, M.A., 2004. Social consequences of disparagement humor: a prejudiced norm theory. Personal. Soc. Psychol. Rev. 8, 79–94. Georg Hsu, L.K., Wan, Y.M., Chang, H., Summergrad, P., Tsang, B.Y.P., Chen, H., 2008. Stigma of depression is more severe in Chinese Americans than Caucasian Americans. Psychiatry Interpers. Biol. Process. 71, 210–218. Goffman, E., 2009. Stigma: Notes on the Management of Spoiled Identity. Simon and Schuster. Hughes, D.J., Rowe, M., Batey, M., Lee, A., 2012. A tale of two sites: Twitter vs. Facebook and the personality predictors of social media usage. Comput. Hum. Behav. 28, 561–569. Jashinsky, J., Burton, S.H., Hanson, C.L., West, J., Giraud-Carrier, C., Barnes, M.D., Argyle, T., 2013. Tracking suicide risk factors through Twitter in the US. Crisis 35, 51–59. Jorm, A.F., Oh, E., 2009. Desire for social distance from people with mental disorders. Aust. NZ J. Psychiatry 43, 183–200. Lee, S., Lee, M., Chiu, M., Kleinman, A., 2005. Experience of social stigma by people with schizophrenia in Hong Kong. Br. J. Psychiatry 186, 153–157. Lincoln, T.M., Arens, E., Berger, C., Rief, W., 2008. Can antistigma campaigns be improved? A test of the impact of biogenetic vs psychosocial causal explanations on implicit and explicit attitudes to schizophrenia. Schizophr. Bull. 34, 984–994. Link, B.G., Phelan, J.C., 2001. Conceptualizing stigma. Annu. Rev. Sociol. 363–385. Link, B.G., Cullen, F.T., Struening, E., Shrout, P.E., Dohrenwend, B.P., 1989. A modified labeling theory approach to mental disorders: an empirical assessment. Am. Sociol. Rev. 54, 400–423. Link, B.G., Yang, L.H., Phelan, J.C., Collins, P.Y., 2004. Measuring mental illness stigma. Schizophr. Bull. 30, 511–541. Livingston, J.D., Boyd, J.E., 2010. Correlates and consequences of internalized stigma for people living with mental illness: a systematic review and meta-analysis. Soc. Sci. Med. 71, 2150–2161. Magliano, L., Read, J., Marassi, R., 2011. Metaphoric and non-metaphoric use of the term “schizophrenia” in Italian newspapers. Soc. Psychiatry Psychiatr. Epidemiol. 46, 1019–1025. Mann, D.M., Ponieman, D., Leventhal, H., Halm, E.A., 2009. Misconceptions about diabetes and its management among low-income minorities with diabetes. Diabetes Care 32, 591–593. Mislove, A., Lehmann, S., Ahn, Y.Y., Onnela, J.P., 2011. Understanding the Demographics of Twitter Users. ICWSM. National Center for Health Statistics (US): Health, United States, 2012: With Special Feature on Emergency Care. Hyattsville (MD): National Center for Health Statistics (US); 2013. Penn, D.L., Nowlin-Drummond, A., 2001. Politically correct labels and schizophrenia: a rose by any other name? Schizophr. Bull. 27, 197–203. Perkins, D.O., Gu, H., Boteva, K., Lieberman, J.A., 2005. Relationship between duration of untreated psychosis and outcome in first-episode schizophrenia: a critical review and meta-analysis. Am. J. Psychiatry 162, 1785–1804. Pescosolido, B., Martin, J., Long, J., Medina, T., Phelan, J., Link, B., 2010. “A disease like any other”? A decade of change in public reactions to schizophrenia, depression, and alcohol dependence. Am. J. Psychiatr. 167 (11), 1321–1330. Podsakoff, P.M., MacKenzie, S.B., Lee, J.-Y., Podsakoff, N.P., 2003. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88, 879–903. Prieto, V.M., Matos, S., Álvarez, M., Cacheda, F., Oliveira, J.L., 2014. Twitter: a good place to detect health conditions. PLoS One 9, e86191. Reynaert, C.C., Gelman, S.A., 2007. The influence of language form and conventional wording on judgments of illness. J. Psycholinguist. Res. 36, 273–295. Roberts, E., Bourne, R., Basden, S., 2013. The representation of mental illness in Bermudian print media, 1991–2011. Psychiatr. Serv. 64 (388–39). Rüsch, N., Angermeyer, M.C., Corrigan, P.W., 2005. Mental illness stigma: concepts, consequences, and initiatives to reduce stigma. Eur. Psychiatry 20, 529–539. Scheff, T.J., 1971. Being Mentally Ill. Transaction Publishers. Schomerus, G., Kenzin, D., Borsche, J., Matschinger, H., Angermeyer, M.C., 2007. The association of schizophrenia with split personality is not an ubiquitous phenomenon. Soc. Psychiatry Psychiatr. Epidemiol. 42 (10), 780–786. Sontag, S., 2013. Illness as Metaphor and AIDS and its Metaphors. Penguin, U.K. Thornicroft, A., Goulden, R., Shefer, G., Rhydderch, D., Rose, D., Williams, P., Thornicroft, G., Henderson, C., 2013. Newspaper coverage of mental illness in England 2008–2011. Br. J. Psychiatry 202, s64–s69. Twitter, 2014. About Twitter, Inc. [Online]. Available at:, https://about.twitter.com/ company (Accessed 16 Jul. 2014). Vahabzadeh, A., Wittenauer, J., Carr, E., 2011. Stigma, schizophrenia and the media: exploring changes in the reporting of schizophrenia in major US newspapers. J. Psychiatr. Pract. 17 (6), 439–446. Yang, L.H., Kleinman, A., Link, B.G., Phelan, J.C., Lee, S., Good, B., 2007. Culture and stigma: adding moral experience to stigma theory. Soc. Sci. Med. 64, 1524–1535. Young, S.D., Rivers, C., Lewis, B., 2014. Methods of using real-time social media technologies for detection and remote monitoring of HIV outcomes. Prev. Med. 63, 112–115.