Trigger Warnings Might Not Coddle After All
Careful examination of a recent high-profile study reveals many new caveats.
Posted Aug 15, 2018
A recent study led by a Harvard Ph.D. student has weighed into the trigger-warnings debate in a high-profile way. Trigger warnings give students a heads-up that material is coming that may be disturbing for some. The big question is whether the warnings help or harm. Most discussions of this study say harm, but such claims seem overblown.
First, kudos to the Harvard researchers: A study on trigger warnings with a control group was badly needed. And the study had other strengths. The researchers concluded, cautiously, that "trigger warnings may present nuanced threats to selective domains of psychological resilience" (Bellet et al., 2018). Most popular press authors have concluded more strongly.
But at least six issues have been missed not only by popular press but, in some cases, apparently by the researchers themselves. The obvious problem that has not been missed is that none of the participants were at-risk college students. What I discuss goes beyond this concern. There are even indications in the same study that trigger warnings might be helpful.
A bottom line is not to jump to conclusions based on a single study, at least when it has such serious limitations.
1. The Participants
Beyond the fact that participants did not include any at-risk students, they completed surveys online and unproctored through Amazon Mechanical Turk (MTurk), a crowdsourcing internet marketplace. Though MTurk has produced some results very similar to more controlled studies, it still allows for several factors to skew results (DeSoto, 2016).
2. The Measures: Trigger warnings might increase students’ accuracy.
The main results were that trigger warnings made participants rate themselves and others as more vulnerable to the emotional effects of past traumas. The researchers regarded these outcomes as harmful, but we don’t know the average ratings or how accurate they were. For all we know, the participants who did not receive trigger warnings underestimated actual vulnerability, and the trigger warnings made participants more accurate. The researchers themselves referred to their participants as “trauma-naïve,” which seemed to regard them as not having much knowledge about the impact or prevalence of traumas. Even doctors underestimate the prevalence of posttraumatic stress disorder (Ehlers et al., 2009).
Improved accuracy about the mental health of self and others may be helpful and not harmful. As one instructor put it, “trigger warnings remind students who may not have suffered a trauma…that these experiences happen” (Gust, 2016).
I acknowledge that greater perception of vulnerability can also have mental health risks (even if accurate). But what if that greater perception arose only because of the particular wording of the study’s trigger warning?
3. The Wording of the Trigger Warning
The researchers’ wording of the actual trigger warning seemed more extreme than most other trigger warnings I’ve seen. The warning stated that the material “contains disturbing content,” whereas other warnings say "may contain” or "potentially disturbing.” Also, the warning stated that an anxiety response was "especially [likely] in those who have a history of trauma” as opposed to something like, "may occur for some students.”
There are many ways to word trigger warnings, depending on the course and content. But if participants were told by researchers from Harvard that the content is definitely disturbing and that anxiety is especially likely for certain people, then it’s not surprising that participants would defer to that authority and estimate greater vulnerabilities. A less extremely worded trigger warning without the Harvard logo might’ve led to different outcomes. (I’m assuming “Harvard” appeared somewhere on the consent form, but I could be wrong.)
4. Alternative Interpretation
The only other result that supported the trigger-warnings-can-harm position pertained only to participants who reported strong beliefs that words can cause harm. For them, trigger warnings led to greater self-reported anxiety from the target reading materials. But maybe the result simply showed that trigger warnings, or the authoritative statement that the materials can cause anxiety, made it easier or more acceptable for these sensitive individuals to express anxiety. That might be a good thing in reducing stigma. And again, we don’t know the average anxiety scores: They could be well below the scale midpoint.
5. Unnoticed Results: Trigger warnings might decrease anxiety.
Overall, the researchers reported that trigger warnings did not affect self-reported anxiety, neither from the target reading materials nor from later materials which had no warnings. However, careful examination of the full results reveals that trigger warnings seemed to slightly decrease anxiety from those later materials. Also, for participants who did not report strong beliefs that words can cause harm, trigger warnings seemed to decrease anxiety from the target reading materials.
For readers who want some statistical details, the two above results were reported as nonsignificant, meaning the p values were not less than .05. But based on my calculations, the p values were less than .09, which researchers sometimes call "marginally significant.” I’m not a proponent for reporting .08 as statistically significant, and the researchers were justified in ignoring those results. But in this inflamed debate and new research area, I think there is also justification at least to mention these two consistent trends that run counter to the strongest critics of trigger warnings.
6. The Trigger Warning Did Not "Coddle"
Another overlooked section from the Harvard article directly pertained to a primary talking point against trigger warnings, namely that trigger warnings “coddle” the mind (Lukianoff & Haidt, 2015). It turned out that the Harvard researchers explicitly reported that “the ‘coddling’ hypothesis” received no support—specifically, that there was no support that "trigger warnings exacerbate an expectancy of predictability that sensitizes people to less severe unexpected stressors.”
The trigger-warnings debate is broader than discussed here. I agree that some uses of trigger warnings go too far. But the claim that trigger warnings harm students based on the new Harvard study also goes too far. It’s a single study with serious limitations, and so it’s too soon to say. And those same results actually suggest that trigger warnings might be helpful in some ways. Without research, our political leanings or other biases can affect what we think of trigger warnings. Instructors need to make their best decisions, but let’s await further research before being sure about the consequences.
Benjamin W. Bellet, Payton J. Jones, and Richard J. McNally, “Trigger Warning: Empirical Evidence Ahead,” Journal of Behavior Therapy and Experimental Psychiatry 61 (2018): 134–41.
K. Andrew DeSoto, “Under the Hood of Mechanical Turk,” APS Observer, March, 2016, https://www.psychologicalscience.org/observer/under-the-hood-of-mechanical-turk.
Anke Ehlers, Nuri Gene-Cos, and Sean Perrin, “Low Recognition of Posttraumatic Stress Disorder in Primary Care,” London Journal of Primary Care 2 (2009): 36–42.
Omni Gust, “I Use Trigger Warnings: But I’m Not Mollycoddling My Students,” Guardian, June 14, 2016, https://www.theguardian.com/higher-education-network/2016/jun/14/i-use-trigger-warnings-but-im-not-mollycoddling-my-students.