29 March 2019

Trigger Warnings

Trigger Warnings Are Trivially Helpful at Reducing Negative Affect, Intrusive Thoughts, and Avoidance' by Mevagh Sanson, Deryn Strange and Maryanne Garry in (2019) Clinical Psychological Science comments
Students are requesting and professors issuing trigger warnings—content warnings cautioning that college course material may cause distress. Trigger warnings are meant to alleviate distress students may otherwise experience, but multiple lines of research suggest trigger warnings could either increase or decrease symptoms of distress. We examined how these theories translate to this applied situation. Across six experiments, we gave some college students and Internet users a trigger warning but not others, exposed everyone to one of a variety of negative materials, then measured symptoms of distress. To better estimate trigger warnings’ effects, we conducted mini meta-analyses on our data, revealing trigger warnings had trivial effects—people reported similar levels of negative affect, intrusions, and avoidance regardless of whether they had received a trigger warning. Moreover, these patterns were similar among people with a history of trauma. These results suggest a trigger warning is neither meaningfully helpful nor harmful. 
The authors argue
Universities around the world are grappling with demands for trigger warnings—cautions to students about upcoming course content that may cause them distress (Medina, 2014; Palmer, 2017). The ideas that various topics may trigger distress—because the material itself is negative or reminds people of prior negative experiences—and warnings about the material’s topic can prevent this distress have long circulated online (Vingiano, 2014). But now these ideas have spread to universities: Two recent surveys of U.S. professors found over half reported using trigger warnings about their course content (Kamenetz, 2016; National Coalition Against Censorship, 2015). Some professors believe trigger warnings help decrease their students’ distress following an encounter with negative material, rather than merely allowing students to avoid that material altogether (e.g., Gust, 2016; Manne, 2015). Other professors, however, believe trigger warnings are not only an affront to academic freedom but might actually increase students’ distress either by allowing students to avoid material altogether (thereby preventing them from learning to cope effectively with reminders of prior negative experiences) or encouraging a more negative reaction to material they do encounter and contributing to the rising levels of anxiety and depression among college students (American Association of University Professors, 2014; Center for Collegiate Mental Health [CCMH], 2016; Lukianoff and Haidt, 2018; McNally, 2014). 
When it comes to the effects of trigger warnings, these conflicting positions are not simply ideological; they are also psychological. Yet when we turned to the psychological literature to find out what effects trigger warnings have, we found research suggesting they would be helpful, research suggesting they would be harmful, but no data directly addressing their effects—until we were writing this manuscript, when Bellet, Jones, and McNally (2018) published a single experiment. In this experiment, trigger warnings produced a small increase in people’s self-reported anxiety after reading negative passages—but only among people who strongly believed words can cause emotional damage. In addition, trigger warnings led people to believe they and others were slightly more susceptible to emotional harm from future, hypothetical traumatic experiences. Taken together, these results suggest trigger warnings mostly did very little. 
But there are at least three reasons Bellet et al.’s (2018) findings are not conclusive on the effects of trigger warnings. First, it is plausible that at least some of their primary findings become nonsignificant once corrected for the increased false discovery rate that arises from multiple comparisons. The rest of their findings regarding the effects of trigger warnings are nonsignificant without correction. This collection of null findings is difficult to interpret, particularly against a backdrop of a single experiment and a modest sample size. Second, anyone with a history of exposure to an extremely distressing event was not permitted to complete the experiment (comprising roughly 50% of those who started it; Jones, 2018). Considering that the majority of the population has been exposed to a potentially traumatic event (Breslau et al., 1998), this exclusion limits generalizability. Third, inasmuch as Bellet et al. tell us something about peripheral effects of trigger warnings, most crucially, they do not tell us much about trigger warnings’ putative ability to alleviate people’s symptoms of distress. 
The fact that this issue remains unresolved is a problem—one we aim to remedy in this article. We systematically and empirically examined the consequences of trigger warnings for three symptoms of people’s distress: negative affect following exposure to negative material, intrusive thoughts related to the negative material, and avoidance of reminders of the negative material. These symptoms at their extreme can constitute part of a posttraumatic stress disorder (PTSD) diagnosis (American Psychiatric Association [APA], 2013). But many people experience these symptoms following exposure to a traumatic experience or negative material without developing PTSD (Breslau et al., 1998; Rothbaum, Foa, Riggs, Murdock and Walsh, 1992). These symptoms are also therefore the very ones that negative course materials could evoke—and that trigger warnings might alleviate or exacerbate. 
They conclude We conducted six experiments investigating the effects of trigger warnings. Meta-analyses revealed that people who saw trigger warnings, compared with people who did not, judged material to be similarly negative, felt similarly negative, experienced similarly frequent intrusive thoughts and avoidance, and comprehended subsequent material similarly well. Although some measures yielded effects in a “trigger warnings are helpful” direction, these effects were so small as to lack practical significance (Ferguson, 2009). As a reference point, a Cochrane review found the standardized mean difference in self-reported symptoms between those who underwent therapy for PTSD and controls was ‒1.60, 95% CI = [‒2.02, ‒1.18] (Bisson, Roberts, Andrew, Cooper, and Lewis, 2013). Of course, trigger warnings are not intended to substitute for therapy; nevertheless, the symptom reductions we observed are minuscule in comparison. Moreover, our meta-analytic confidence intervals were narrow, suggesting high precision—yet still showed trigger warnings plausibly have no effect or might even work slightly in the direction of causing harm (Cumming, 2012). 
A critic might wonder if some subjects found the warnings helpful because they withdrew from the experiments after being warned, thereby avoiding the material and any ensuing symptoms (Gross, 2015). But when we examined responses from subjects who quit our experiments before completion, we found similar proportions quit in the warned and unwarned conditions, and the number of subjects who quit specifically after seeing the warning was very small (none in Experiment 1a, 1b, or 2a; nine in Experiment 2b; six in Experiment 3; one in Experiment 4), suggesting few if any subjects used the warning to avoid negative material (for more detail about when subjects quit, see the Supplemental Material). Moreover, because avoidance is a PTSD symptom (APA, 2013), the use of warnings to avoid material could be construed as harmful. 
There are other possible reasons to explain why trigger warnings exerted only trivial effects, and those accounts are less interesting. For example, perhaps subjects did not notice the trigger warning or the wording did not change expectations about the material to follow. But a large majority of warned subjects said they remembered seeing the warning, and in Experiment 3, we found that warned subjects expected the material to follow would be more negative than unwarned subjects. Of course, it is possible that for people’s expectations to matter, warnings need to target people’s beliefs about their symptoms rather than about the material (Kirsch, 1997). It is also possible the warnings did not constitute an obvious enough prompt that, or indeed how, people should prepare to regulate their emotions—possibilities that fit with the literature showing people often cannot optimally use warnings to adjust their behavior (Gross, 2015; Wilson and Brekke, 1994). We based our warnings on “real-world” uses, but future research could examine the effects of variations that target a range of theoretical issues. 
Our results run contrary to findings that warnings encountered before films made people feel worse afterward (Cantor et al., 1984; de Wied et al., 1997). But given our warned and unwarned subjects found the material similarly negative, it makes sense they then reported symptoms to similar degrees (Hall and Berntsen, 2008; Rubin et al., 2008; Wenzlaff and Wegner, 2000). Indeed, it could be we found little effect of warnings because of a floor effect—our materials did not make people feel sufficiently negative, such that warnings could not help much. Yet for unwarned subjects who saw material of greater negativity, the 95% CIs around their reported negative affect did not overlap with the bottom of the scale. Put another way, although these scores could have been decreased by trigger warnings, they were not. In addition, our finding that more negative materials produced more symptoms of distress fits with the idea that over-accessible memories of traumatic experiences contribute to symptoms of PTSD, and our finding that trigger warnings produced similar effects before material of greater and lesser negativity fits with the idea that the same mechanisms are at play during more and less traumatic experiences (Rubin et al., 2008). Our results also extend recent work suggesting trigger warnings had little effect on most people’s self-reported anxiety; we showed that trigger warnings have little effect on people’s distress (Bellet et al., 2018). 
Taken together, our findings show that trigger warnings are at best trivially helpful. But this conclusion comes with at least three caveats and limitations. First, we did not recruit people with a history of psychopathology (e.g., those with a diagnosis of PTSD, anxiety, or depression), and so we do not know how well our results generalize to clinical populations (although the results of Experiment 4 fit with the idea that most of our subjects—like most of the population—have had a traumatic experience; Breslau et al., 1998). Second, we did not ask our subjects their socioeconomic status or education level, which limits our ability to characterize the samples on whom we tested the effects of trigger warnings. Our samples were, however, drawn from populations for whom trigger warnings are often provided. Third, trigger warnings may have nontrivial effects we did not measure. For example, we did not ask about the phenomenology of the intrusions, yet warnings may have altered the vividness of the intrusions, for instance (Takarangi and Strange, 2010). Further, we used only self-report measures rather than taking physiological measures of hyperarousal symptoms, for instance (APA, 2013). Indeed, when people can precisely predict the timing of an unpleasant experience (e.g., an electric shock), they have lesser physiological responses to it even though their ratings of its magnitude are unaffected (Lykken, Macindoe, and Tellegen, 1972; for a review, see Lykken and  Tellegen, 1974). It is possible, therefore, that if trigger warnings allow people to predict an encounter with negative material, those warnings may reduce people’s physiological responses to the negative material. These issues constitute interesting directions for future research. 
Where do the current findings leave us? Some might wonder if professors should continue to issue trigger warnings. After all, if the warnings do not worsen distress and students believe the warnings are helpful, then why not? Put simply, people are not always good judges of the effects interventions have on themselves or others (Lilienfeld, Ritschel, Lynn, Cautin, and Latzman, 2014; Wilson and  Brekke, 1994), and the chronic effects of trigger warnings may be different from their acute effects. College students are increasingly anxious (CCMH, 2016), and widespread adoption of trigger warnings in syllabi may promote this trend, tacitly encouraging students to turn to avoidance, thereby depriving them of opportunities to learn healthier ways to manage potential distress.