October 1, 2018–October 3, 2018
We present a dataset of 250 participants’ ratings of 600 written descriptions of events on a scale ranging from “Not at all traumatic” to “Extremely traumatic.” The definition of what constitutes a trauma has increasingly become a topic of scientific study over the past few decades, as the range of events that may precipitate a diagnosis of post-traumatic stress disorder (PTSD) may be expanding.
Once limited to survivors of warfare and sexual abuse and other canonical extreme stressors, the PTSD label now applies to a much broader conceptualization of traumatic events . For example, a recent survey of 769 students at Arizona State University revealed that 25% met criteria for “clinically significant” levels of distress on par with diagnosable PTSD in response to the 2016 election . Other examples of the “conceptual bracket creep in the definition of trauma” (p. 231)  include exposure to crude jokes , giving birth to a healthy baby , and viewing televised coverage of the 9/11 terrorist attacks . This expanded view of PTSD raises questions about what types of events might be traumatic for whom, in which settings, and why.
The aim of collecting these data was to establish a stimulus set of a wide range of potentially traumatic events which could be used in further experimental research. Thus, we collected objective ratings of potential stimuli as well as demographic characteristics of the raters.
The sample consists of 250 English-speaking US participants recruited through Amazon Mechanical Turk from October 1, 2018 to October 3, 2018. Participants ranged in age from 20 to 70, with a mean age of 36 (SD = 10). Most participants identified as female (n = 138), with the remainder identifying as male (n = 108) or other (n = 4). A majority of raters self-identified their race as White/Caucasian (n = 193) and non-Hispanic (n = 221). Most participants reported no history of serious trauma (n = 173) or lifetime psychiatric diagnosis (n = 218).
The largest portion of the dataset concerns participants’ ratings of the events. Events were formulated by the research team via brainstorming exercises. Guided by the team’s expertise in trauma research, events were intended to cover a wide range of experiences with diverse potential emotional reactions and to provide a sufficient sample size for use in experimental research. Participants rated the events on a Likert scale ranging from 1 (“Not at all traumatic”) to 7 (“Extremely traumatic”).
Participants were also assessed in terms of demographic characteristics (gender, race, ethnicity, previous trauma exposure, psychiatric diagnosis, religiosity, political orientation, age).
Participants were randomly assigned to one of six possible stimulus sets in Qualtrics. Each stimulus set contained 98 unique items and 12 constant items that were used for reliability verifications across the six sets. Providing a visual example, Figure 1 shows a thresholded correlation network (r ≥ 0.3) of the 98 unique items in the first set with items ordered and colored according to their mean rating. After rating all items, participants completed demographic items and a human participant and English language verifier.
Participants were required to complete a reCAPTCHA prior to taking the survey to ensure that they were human and not automated responders. Three attention checks (“If you’re actually reading this question, please select the number five as your response”) were interspersed in the item ratings. Participants were removed from the dataset if they did not complete all three attention checks correctly. Following the survey, participants were asked to complete a human participant and English verifier (“In the space provided below, please describe your activities last weekend in exactly 3 sentences”). Participants were removed from the dataset if their response did not address the prompt, contained less than two ideas, contained less than three ideas and at least one English error, or contained more than three English errors. Of 300 participants who completed the study, 50 were excluded. An additional 25 individuals accessed the Qualtrics survey but did not complete it. Interrater reliability across the batches assessed with the set of 12 constant items was good when considering each of 250 participants as separate judges (ICC1 = 0.70), and excellent when considering the average value across each of the six sets (ICC1 = 0.99).
All identifying information has been removed from the dataset. Participants indicated their consent prior to the survey. The study was approved by the Harvard Institutional Review Board.
(3) Dataset description
Format names and versions
The data have been deposited under a CC-By Attribution 4.0 International (CC-By) License.
(4) Reuse potential
This dataset contains ratings from 250 participants on 600 potentially traumatic events. These data are suitable for a wide variety of purposes. First and foremost, the data represent validated stimuli that could be used in a wide variety of experimental settings. For example, researchers could provide an experimental manipulation and measure response via ratings of the stimuli, or researchers could display the stimuli as a manipulation while participants are measured via EEG or fMRI equipment. This is the original purpose for which we collected this data .
In addition to being used as stimuli, the data could also be analysed to study participants’ conceptualization of trauma. Researchers could examine whether ratings differ across different demographic groups or prior experiences with trauma and mental health. Researchers could use text analysis or qualitative coding to separate the items into different types of trauma (e.g., “violence”, “sexual abuse”, “natural disaster”) or different types of exposure (e.g., “experienced directly”, “witnessed”, “learned about”) and examine whether differences emerge between ratings on these categories .