Choose Wisely! – Considering the Perceived Threat in the Selection of Trauma Film Clips May Improve the Ecological Validity of the Trauma Film Paradigm
The trauma film paradigm (TFP) is the gold standard experimental psychopathology model of psychological trauma. However, different film clips with distinct effects on trauma-analogue symptoms are used across TFP studies, raising questions about the generalizability of study-specific results and the ecological validity of the TFP. Little is known about differences among film clips commonly used in the TFP regarding key features of psychological trauma (i.e., threat and anxiety) and their veridicality.
Methods
Thirty healthy participants watched eight trauma film clips in randomized order and subsequently rated each in view of perceived threat, anxiety, and realism. In addition to descriptive statistics, variance analyses and pairwise comparisons were performed to test for differences on these outcomes.
Results
The results indicated significant differences among the trauma film clips in terms of perceived threat and realism. However, nearly all trauma film clips evoked moderate anxiety levels.
Conclusions
This study stressed the importance of perceived threat as a key feature in selecting film clips for the TFP, as highly threatening film clips were also perceived as more realistic. When replicated using delayed outcomes (e.g., intrusive memories), choosing trauma film clips in view of their perceived threat might improve the ecological validity of the TFP.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Introduction
Improving our understanding of psychological trauma is key to the development of psychological interventions to alleviate symptoms following traumatic experiences (Kube et al., 2020). In addition to clinical studies and patients’ retrospective reports, research needs prospective experimental studies to identify causal mechanisms of action (Ehring et al., 2011). By understanding the basic mechanisms underlying posttraumatic symptom development (e.g., negative expectations: Herzog et al., 2022a), we can translate these findings from the laboratory to the clinic (Ehring et al., 2022) and test brief, innovative, science-driven, scalable interventions (e.g., CBM: Woud et al., 2012; Tetris: James et al., 2015; Placebos: Kube et al., 2022) that may enable optimizing treatment outcomes (Herzog et al., 2020), thereby reducing the global burden of trauma-related mental health problems.
Experimental research on the mechanisms mediating the development and maintenance of posttraumatic intrusive memories has increased considerably, and the trauma film paradigm (TFP) has emerged to the most important experimental psychopathology model of psychological trauma (Arnaudova & Hagenaars, 2017). More than one hundred studies involving the TFP and its variants have been done (For reviews, see Holmes & Bourne, 2008; James et al., 2016).
In the TFP, aversive film clips are presented to healthy people as an analogue stressor to real-world trauma (James et al., 2016), enabling assessment of immediate emotional responses as well as intrusive recollections of the clips in the days that follow the experiment (Brewin, 2015; Iyadurai et al., 2019). Interestingly, although studies highlight the heterogeneity of PTSD, 95.7% of all patients reported either intrusive memories, nightmares, or flashbacks (i.e., re-experiencing symptoms) (Bryant et al., 2022), and a recent large network analysis highlighted the centrality of re-experiencing symptoms in PTSD (Duek et al., 2021). In fact, ‘intrusive thoughts’ seem to be the most often occurring most central symptoms in PTSD (Birkeland et al., 2020; Isvoranu et al., 2021), supporting also its particular importance in the TFP. Importantly, the TFP is ideal for testing hypotheses about the emergence and persistence of the hallmark intrusive symptoms as investigators can manipulate key psychological processes before, during, and after exposure the film clips, making this paradigm an ethically acceptable tool in psychological trauma research. Participants view one or more film clips depicting stressful scenes (e.g., car accidents, physical violence, etc.) and then record the frequency of intrusive memories associated with the film clip over subsequent days in a daily diary (e.g., 7 days).
Numerous studies showed that trauma films can reliably induce trauma-analogue symptoms such as intrusive memories (Holmes & Bourne, 2008). As it is well suited to provoke intrusion-like symptoms, the TFP sometimes raises concerns about psychological harm by institutional review boards. However, Takarangi et al. (2023) found that such in-lab trauma simulations do not cause excessive or prolonged distress to participants and hence are well-tolerated by participants, aligning with IRB ethical guidelines. Moreover, in contrast to people with PTSD where intrusions often occur with a frightening sense of nowness, intrusive memories elicited by the TFP are less intense and fade within 7 days after the experiment (Holmes et al., 2004).
However, TFP studies have varied greatly in the trauma film clips exposed to participants, and these have elicited diverse emotional and other trauma-analogue responses (Weidmann et al., 2009). Such methodological diversity may impede the generalizability of findings and the ecological validity of the TFP itself. For example, Arnaudova and Hagenaars (2017) showed that different trauma films evoke different immediate (e.g., disgust, embarrassment, heart rate) and delayed effects (e.g., intrusive memories). In contrast to the different immediate and delayed effects, the authors also found that the trauma film clips did not differ in terms of valence and arousal across themes, further begging the question of what other important features of traumatic events, beyond the emotional response per se, might be in play that can explain the differences found.
One particular trauma-specific feature highlighted in predictive processing frameworks1 is perceived threat: Within this theoretical framework, reexperiencing symptoms can be understood in the context of hyper-precise predictions with regard to the anticipation of threat, thereby shaping perceptions toward the continued experience of threat, similar to a self-fulfilling prophecy (Kube et al., 2020; Linson & Friston, 2019; Wilkinson et al., 2017). In line with that idea, the emotional processing theory (Foa & Kozak, 1986; Foa et al., 1989, 1992) and the cognitive model of PTSD (Ehlers & Clark, 2000) highlight the importance of the emotional and cognitive processes associated with threat in the development and persistence of PTSD symptoms such as intrusions. For example, Baptie et al. (2021) presented a trauma film clip within a virtual reality (VR) environment to increase its perceived threat. Although participants felt more personally involved when the clip appeared in VR compared to the traditional non-VR presentation, the two presentation formats did not result in significantly different frequencies or distress levels of intrusive recollections in the days following the experiment (Baptie et al., 2021). It remains unclear why enhanced realism as triggered by the VR environment did not affect these outcomes. Hypothetically, the setup might not have increased perceived threat as intended, but more research is needed.
Building on Arnaudova and Hagenaars’s work (2017), we aimed to further provide guidance in the selection of trauma films by investigating other key features such as the perceived threat, realism and anxiety of different trauma films commonly used in TFP studies. By doing so we hoped to increase the paradigm’s ecological validity.
Methods
This study was approved by the ethics committee of the Department of Psychiatry and Psychotherapy at the University Medicine Greifswald (Reference number BB 133/21). All participants were treated in accordance with the ethical guidelines of the German Psychological Society and the guidelines for human studies of the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all participants.
This is a preliminary study in an independent sample for a larger pre-registered trauma film paradigm study experimentally manipulating threat perception by using different trauma films (AsPredicted Identifier: #87293, Link: https://aspredicted.org/5XL_QV9).
Participants
Participants were recruited through various channels, including social media platforms such as Facebook and Instagram, as well as flyers posted in the university and public areas within the city. They stated that the aim of the study was to examine how different violent film material is perceived. Prospective participants expressed their interest in participating by sending an email with their personal information, including their name, age, and field of study or occupation. The inclusion criteria were: age between 18 to 65 years and sufficient proficiency in German. Individuals were excluded who had current mental disorders, suicidal thoughts/behavior, non-suicidal self-injury behavior, or other vulnerabilities (e.g., debilitating medical conditions). To verify eligibility, we first screened potential participants by having them complete an online survey (SoSciSurvey) which included questions about traumatic experiences and depressive symptoms, assessed through the Life-Event-Checklist for DSM-5 (LEC-5) (Weathers et al., 2013), and Patient-Health Questionnaire-9 (PHQ-9) (Kroenke et al., 2001), with additional questions allowing to exclude participants with a medical history of heart disease or epilepsy, history of and tendency to faint, visual impairment, substantial experience in the medical field, as well as students of psychology and medicine. After checking the inclusion criteria, eligible participants were invited by email to a lab appointment. Prior to the experiment, inclusion criteria were verified again through clinical interviews via the Structured Clinical Interview for DSM-5 Disorders (SCID-5-CV) (First et al., 2016), and Columbia-Suicide Severity Rating Scale (C-SSRS) (Posner et al., 2011). In case any mental disorder and/or risk for suicide were detected, participants were given the opportunity to receive more information about their mental health condition and potential treatment options. Interviewers were trained in SCID-5-CV and C-SSRS, including theoretical and practical elements, to ensure proper implementation of observer-evaluated measures.
The sample size was determined via an a priori power analysis using the G*Power Version 3.1 (Erdfelder et al., 2009; Faul et al., 2007) for the primary outcome of perceived threat. Based on previous studies (Baptie et al., 2021; Dibbets & Schulte-Ostermann, 2015), the power analysis indicated that an optimal sample size of N = 24 was required to detect an expected effect size of f = 0.33 for a repeated measures, within-subject ANOVA with eight measurement time points, a significance level of α = .05, a power of 1 – β = .95, and a correlation of values across measurement time points of r = .50. Since the effect size was estimated based on previous studies that used virtual reality to manipulate perceived threat, and the effects in virtual reality may be larger due to stronger immersive involvement, we aimed to include N = 30 participants to ensure sufficient power to detect smaller effects.
In total, n = 56 people were interested in study participation, of whom n = 20 people were either excluded after the pre-screening (i.e., indicated the exposure to a traumatic event in the LEC-5 or showed increased depressive symptoms above the PHQ-9 cut-off) or were not responsive to contact attempts after expressing initial interest. Then, n = 36 potentially suitable people were invited to participate in the study, and an appointment for study participation was scheduled for n = 31 people. A total sample of N = 30 participants were included in the study, while n = 1 was excluded after the in-lab clinical interview screening (i.e., currently met the criteria for a mental disorder in the SCID-5-CV). As an incentive, people received either credit for research participation, or a financial compensation of up to 15€ for participating in the complete study and 5€ in case they withdrew early from participation.
Procedure and Material
After inclusion in the study, participants watched several aversive film clips and completed computer-based post-film measures. After the lights were turned off, all film clips were presented on a computer screen (Dell-Flat-Panel-Monitor, S2716DG, 27 inches) at a distance of approximately 1.5 m with the original soundtrack of the film clips through headphones (AKG K) and keeping the volume constant. A total of eight different film clips (see Table 1) were successively rated for perceived threat, realism, and anxiety. Each film clip lasted approximately 2–3 min, with the subsequent questionnaires taking approximately five minutes each. Thus, the total duration of the experiment, including screening for inclusion and exclusion criteria, was approximately 1.5 h for participants, with the experimental study lasting around one hour (20 min of film material, 40 min of questionnaire completion). Sequence effects of the film clips were avoided by randomizing the order for each participant. After each film clip, participants were asked to confirm that they have watched the film clip from the beginning to the end for attention check.
Table 1
Description of the different trauma film clips
No.
Theme
Title
Content
Type of the film and perspective
Length (min)
Source/Director (Country, Year)
V_1
Sexualized violence (against men)
„Antichrist”
A woman hits her husband’s genitals with a wooden peg.
Feature film from the perspective of an observing person
2:55
L. von Trier (Dänemark, 2009)
V_2*†
Car accident
Compilation of short (amateur) videos
Self-recorded/re-enacted video footage of car accidents in real life
Real-life videos/ re-enacted videos from the perspective of a person involved
2:20
Source: Youtube (e.g., "first person car crash")
V_3†
Car accident
Compilation of short amateur videos
Self-made recordings of real-life (serious) car accidents
Real-life videos from the perspective of an observing person
A violent confrontation occurs between a group of people and one person is beaten to death with a fire extinguisher.
Feature film from the perspective of an observing person
3:10
G. Noe (Frankreich, 2002)
V_5*
Physical violence
„Hardcore Henri”
In a building shell, two parties fight each other with firearms. There are scenes of chase, deaths and injuries.
Feature film from the perspective of a person involved
1:52
I. Naishuller (USA, Russland; 2015)
V_6
Sexualized violence (against women)
„Irreversible”
A man rapes a woman in an underpass.
Feature film from the perspective of an observing person
3:49
G. Noe (Frankreich, 2002)
V_7*†
Physical violence
„Maniac”
First, a violent conflict between a man and a woman in an apartment is shown, followed by a car accident.
Feature film from the perspective of a person involved
2:59
F. Khalfoun (Frankreich, USA; 2012)
V_8
Car accident
„Texting while Driving”
Three young women have a serious car accident. Both passengers die, while the driver can be rescued by the paramedics.
Feature film from the perspective of an observing person
2:45
P. Watkins-Hughes (UK, 2009)
* = First-person perspective
† = Compilation of different clips and no coherent story
The selection of film clips was based on previous studies (e.g., Herzog et al., 2022a; Rattel et al., 2019) and recommendations for TFP studies (Arnaudova & Hagenaars, 2017), as those film clips reliably induce intrusions in experimental paradigms but differed in their content (i.e., type of violence). The aversive film excerpts utilized in this study, along with their contents and features (length, narrative structure, camera perspective), are detailed in Table 1. Particularly, the trauma film clips contained scenes of physical and sexual violence as well as car accidents, ranging in length from 1:52 („Hardcore Henri”) to 3:49 min („Irreversible”) and varying in its narrative structure from compilations (“Car accidents”) to a coherent story („Irreversible”). Additionally, they differed in the camera perspective between first- vs. third-person. Although watching film material has been criticized for being too passive in TFP (Dibbets & Schulte-Ostermann, 2015), a trauma-analogue stressor in VR has not evoked significantly more intrusions, nor have they been more distressing (Baptie et al., 2021; Cuperus et al., 2017). Although these findings do not support the use of VR considering the cost–benefit ratio (Baptie et al., 2021), this study highlighted the importance to elicit stronger feelings of involvement and presence in the traditional TFP.
To increase the variance in perceived threat and realism, we balanced films on different characteristics: Assuming varying levels of perceived threat in different film clips, this phenomenon may potentially be attributed to distinct characteristics (e.g., content, perspective, coherent story) of these scenes. Studies indicate that film scenes (1) depicting some potentially traumatic events known to increase the likelihood for the development of PTSD symptoms such as interpersonal violence (e.g., trauma film content: sexualized violence > physical violence > traffic accidents) (Arnaudova & Hagenaars, 2017; Weidmann et al., 2009), (2) filmed from the first-person (> third-person) perspective (Denisova & Cairns, 2015; Gorisse et al., 2017), and (3) with a clear, cohesive narrative (Weidmann et al., 2009) would elicit a higher degree of perceived threat. Consequently, we assume that these would also be rated as more realistic due to an enhanced sense of presence (Baptie et al., 2021; Weidmann et al., 2009). Most of the selected film clips were used in previous TFP studies and all are either freely available on the Internet or feature films available for purchase (e.g., Irreversible).
In compliance with the copyright, the material used for this study (e.g., trauma film clips) will be shared with other researchers upon request by the corresponding author.
Measures
Perceived Threat Questionnaire (PTQ). The perceived threat evoked by the trauma films was the primary outcome, assessed with the PTQ – a new questionnaire specifically designed for this study consisting of seven items (e.g., “While watching the film clip, I was worried that the same thing might happen to me (e.g., car accident, physical attack).”, and “While watching the film clip, I felt like I was in great danger.”). Participants were asked to rate how threatened they felt by the film clips on a 5-point Likert-type-like scale ranging from 1 (“strong disagreement”) to 5 (“strong agreement”). A total score was computed by summing the ratings on all seven items, resulting in scores ranging from 7 to 35 with higher scores indicating higher levels of perceived threat. A principal component analysis was conducted for the 7-item PTQ across the eight films to detect the latent factor structure. In addition to content validity aspects legitimizing a single-factor solution, the Kaiser-criterion, the Scree-plot, a parallel analysis according to Horn (1965) and the MAP-test consistently supported a single-factor-solution “perceived threat,” accounting for 66.38% of the total variance. An internal consistency analysis of the 7-item PTQ in our sample revealed a Cronbach’s α = .91. For exploratory analyses, two additional 2 items measured the participant’s probability estimates of these events to happen; namely, (1) “How likely do you think it is that you could experience what you have just seen?” (PTQ8), and (2) “How likely do you think it is that you could experience what you have just seen compared to other people your age?” (PTQ9) on a slider scale ranging from 0 – 100%. The PTQ can be found in the supplements (see Supplemental Material).
ITC-Sense of Presence Inventory (ITC-SOPI; Lessiter et al., 2001). The ITC-SOPI is a 44-item questionnaire designed to measure the sense of presence and involvement in a media environment consisting of 4 subscales: sense of physical space, engagement, ecological validity, and negative effects. As a cross-media measure, it focuses on users’ experiences of media with no reference to objective system parameters. In the present study, the ecological validity subscale – a 5-item subscale measuring the believability and realism of the content, and the naturalness and solidity of the environment – was used to assess the perceived realism of the different trauma film clips. Participants were asked to rate the items (e.g., “The content seemed believable to me.”, “The scenes depicted could really occur in the real world.”) on a 5-point Likert-type-like scale ranging from 1 (“strong disagreement”) to 5 (“strong agreement”) with total scores ranging from 5 to 25. These variables indicate a tendency to perceive the mediated environment as lifelike and real. The questionnaire was translated from English into German and checked by back translation. The ITC-SOPI has shown good psychometric properties with regard to validity and reliability across languages (Lessiter et al., 2001; Vasconcelos-Raposo et al., 2019).
State-Trait-Anxiety-Inventory (STAI; Spielberger et al., 1983). The STAI is a questionnaire assessing two dimensions of anxiety; namely, anxiety as a trait (STAI-T), and anxiety as a state (STAI-S), i.e., a transient emotional state, each dimension measured with 20 anxiety-related items. In this study, participants rated the German version of STAI-T and STAI-S (Laux et al., 1981) on an 8-point Likert-scale ranging from 1 (“almost never”) to 8 (“almost always”), and 1 (“strongly disagree”) to 8 (“strongly agree”), respectively. Total scores range from 20 to 160, with higher scores indicating higher trait/state anxiety levels.
Other Measures
Sociodemographic characteristics. Participants were asked about their age, gender, highest educational degree, and current employment status.
Follow-up survey (Herzog et al., 2022). After rating all film clips, participants were asked in a follow-up interview to rate on a 11-point Likert-scale ranging from 0 (“not at all”) to 10 (“very well”) (1) how distressing it was for them to watch the film clips and how well they could identify with the victims, (2) how well they were able to concentrate on the trauma film. Further, they were asked if they used any strategy to distract themselves while watching the film clips (e.g., looking away, closing one’s eyes).
Statistical Analyses
All statistical analyses were conducted in IBM SPSS ® Statistics (Version 26). In accordance with Tabachnick and Fidell (2019), all scales included in the statistical analyses were inspected for outliers and extreme values using descriptive analyses and box-whisker plots. After descriptive examination of the data, no implausible outlier values of the individual data were identified and the final data set included N = 30 participants. In addition to computing descriptive statistics, a repeated measures, within factors ANOVA with 8 measurement time points including pairwise comparisons were performed to test whether the eight film clips differed in their perceived threat, as assessed with the PTQ. To further determine differences among the eight trauma films, we conducted ANOVAs on the secondary outcomes, i.e., realism, as assessed with the ecological validity subscale of the ITC-SOPI, and anxiety, as assessed with the STAI-S.
Results
Sample Characteristics
The mean age of our sample was M = 23.43 years (SD = 4.08), ranging from 19 – 36 years. Among all participants, 36.7% identified as female, 96.7% hold the German citizenship, whereas 100% identified as having a European cultural background and only 3.3% had a history of migration. The majority were in a partnership (60%) and no participant had children. Most participants were students (93.3%) with 13.8% already holding their first university degree; the rest were full-time employees. The trait anxiety level in this sample was moderate (STAI-T: M = 77.07 (SD = 11.07), ranging from 59 and 98).
Implementation Check
In this sample, most participants reported unfamiliarity with the film clips, whereas 23.3% have seen one or more of them. In general, they reported that they perceived watching the film clips as being unpleasant (M = 7.57 [SD = 1.87]), while the degree of distress elicited by all film clips was moderate (M = 4.57 [SD = 3.08]). The mean self-reported concentration while watching the film clips was high (M = 9.10 [SD = 1.09]), while their identification with the victims in the film clips was moderate (M = 4.90, [SD = 2.59]). Moreover, 44.8% participants reported that they did not look away at all, 48.3% looked away one to three times, and 6.8% four to six times during the film clips. Although 72.4% participants reported not to have closed their eyes while watching the film clips, 20.6% did so one to two times, and 6.8% did so three to five times. The results from the exploratory open question regarding distraction strategies indicated that the most frequently used strategy was attention shifting, cognitive soothing, and contextualizing the aversive experience (for example: “It’s just a movie.”; “It’s not real.”). To determine the representativeness of the results from the follow-up interview, we compared them to results of other TFP studies. In line, similar studies found that the distress caused by watching trauma films was moderate ranging from M = 4.30 to M = 6.50 (e.g., Holmes et al., 2004; James et al., 2015). Similarly, our findings for the attention and concentration ratings can be well compared with the results of other studies (e.g., Laposa & Alden, 2008).
Differences Between Trauma Films
The descriptive statistics for the outcome variables, separated by the eight film clips, are displayed in Table 2. Descriptively, the film clips 3 (Car accidents from Olatunji and Fan (2015)), 6 (Excerpt depicting sexualized violence from the movie Irreversible), and 8 (Car accidents from Texting While Driving) elicited the highest levels of perceived threat with M = 18.83, M = 19.40, M = 19.07, respectively. Of note, these ratings are in line with the participants’ probability estimates of these events to happen (assessed with the additional items PTQ8 and PTQ9). The film clips 2 (First person car accidents from Youtube) with M = 12.77, and 5 (Excerpt depicting physical violence from the movie Hardcore Henry) with M = 10.17 provoked the least perceived threat.
Table 2
Descriptive statistics of all scales for each of the trauma film clips
Scale
PTQ
ITC-SOPI
STAI-S
PTQ8
PTQ9
Film clip
M
SD
M
SD
M
SD
M
SD
M
SD
V_1 (Sexualized violence)
15.17
5.36
14.03
4.48
87.20
31.35
5.90
7.21
5.00
8.54
V_2 (Car accidents)
12.77
4.48
17.43
4.39
71.53
26.99
19.47
18.57
15.47
15.47
V_3 (Car accidents)
18.83
5.66
22.83
2.26
81.00
27.97
30.77
17.04
19.10
17.20
V_4 (Physical violence)
16.57
5.35
14.90
4.84
89.10
29.42
6.73
11.30
4.47
5.02
V_5 (Physical violence)
10.17
2.79
9.47
3.49
69.57
24.02
4.43
5.69
2.73
2.36
V_6 (Sexualized violence)
19.40
6.99
20.87
3.01
91.27
30.99
26.07
24.62
24.03
23.62
V_7 (Physical violence)
14.43
4.85
14.67
3.97
79.30
29.49
9.67
13.60
8.00
13.04
V_8 (Car accidents)
19.07
5.89
20.57
3.29
82.80
31.37
35.87
22.72
21.80
19.15
Note. M = mean value; SD = standard deviation. PTQ = Perceived Threat Questionnaire with scores ranging from 7 to 35, ITC-SOPI = Subscale ecological validity of the ITC-Sense of Presence Inventory (Lessiter et al., 2001) with total scores ranging from 5 to 25; STAI-S = State-Trait-Anxiety-Inventory (Spielberger et al., 1983) measuring the state anxiety with total scores ranging from 20 to 160 (German Version); PTQ8 = Additional item 8 (0–100%); PTQ9 = Additional item 9 (0–100%). V_1 = Sexualized violence (Excerpt from the movie Antichrist), V_2 = Car accidents (First person car crashes from “Youtube”), V_3 = Car accidents from Olatunji and Fan (2015), V_4 = Physical violence (Excerpt from the movie Irreversible), V_5 = Physical violence (Excerpt from the movie Hardcore Henry); V_6 = Sexualized violence (Excerpt from the movie Irreversible), V_7 = Physical violence (Excerpt from the movie Maniac; V_8 = Car accidents (Texting While Driving)
As visualized in Fig. 1, the film clips 3, 6, and 8 were also rated as the most realistic, and the film clip 5 as least realistic. However, these differences vanish to some extent in view of the post-film anxiety they elicited, with less variation between the trauma film clips.
Fig. 1
Comparisons of trauma film clips with regard to perceived threat (A), realism (B), and anxiety (C). Note. PTQ = Perceived Threat Questionnaire, ITC-SOPI = Subscale ecological validity of the ITC-Sense of Presence Inventory (Lessiter et al., 2001); STAI-S = State-Trait-Anxiety-Inventory (Spielberger et al., 1983). V_1 = Sexualized violence (Excerpt from the movie Antichrist), V_2 = Car accidents (First person car crashes from “Youtube”), V_3 = Car accidents from Olatunji and Fan (2015), V_4 = Physical violence (Excerpt from the movie Irreversible), V_5 = Physical violence (Excerpt from the movie Hardcore Henry); V_6 = Sexualized violence (Excerpt from the movie Irreversible), V_7 = Physical violence (Excerpt from the movie Maniac; V_8 = Car accidents (Texting While Driving)
×
An ANOVA indicated significant differences between the film clips in the primary outcome, i.e., perceived threat, F(7, 203) = 24.47, p < .001, ɳ2p = .46. For the secondary outcomes, the ANOVAs revealed significant differences among the film clips for realism, F(7, 203) = 53.45, p < .001, ɳ2p = .65, and anxiety, F(7, 203) = 12.88, p < .001, ɳ2p = .31. The results of the pairwise comparisons appear in Table 3. Interestingly, pairwise comparisons revealed significant and large differences between almost all trauma film clips in perceived threat and realism, but these differences among the trauma film clips diminished in eliciting anxiety.
Table 3
Pairwise comparisons between trauma film clips on different outcomes
Pairwise comparisons between film clips
V_1
V_2
V_3
V_4
V_5
V_6
V_7
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
Mean Difference (95% CI)
SE
PTQ
V_1 (Sexualized violence)
V_2 (Car accidents)
−2.40 (0.96; 0.53)
0.96
V_3 (Car accidents
3.67 (0.53; 6.81)*
0.91
6.01 (2.94; 9.20)
0.91
V_4 (Physical violence)
1.40 (−1.59; 4.39)
0.87
3.80 (0.54; 7.06)*
0.95
−2.27 (−5.57; 1.04)
0.96
V_5 (Physical violence)
−5.00 (−8.03; −1.97)***
0.88
−2.60 (−5.03; −0.17)*
0.71
−8.67 (−11.98; −5.35)***
0.96
−6.40 (−9.40; −3.40)***
0.87
V_6 (Sexualized violence)
4.23 (0.01; 8.46)*
1.30
6.63 (2.73; 10.53)***
1.13
0.57 (−3.62; 4.75)
1.22
2.83 (−1.04; 6.71)
1.13
9.23 (5.21; 13.26)***
1.17
V_7 (Physical violence)
−0.73 (−3.74; 2.28)
0.88
1.67 (−0.95; 4.28)
0.76
−4.40 (−7.65; −1.15)**
0.94
−2.13 (−4.77; 0.50)
0.77
4.27 (1.98; 6.55)***
0.66
−4.97 (−8.40; −1.54)***
1.00
V_8 (Car accidents)
3.90 (0.75; 7.052)**
0.92
6.30 (3.63; 8.98)***
0.78
0.23 (−2.62; 3.08)
0.83
2.50 (−0.5; 5.50)
0.87
8.90 (5.44; 12.36)***
1.01
−0.33 (−4.01; 3.34)
1.07
4.63 (1.73; 7.54)***
0.85
STAI-S
V_1 (Sexualized violence)
V_2 (Car accidents)
−15.67 (−29.30; 2.03)*
3.96
V_3 (Car accidents
−6.20 (−16.07; 3.67)
2.87
9.47 (−1.66; 20.59)
3.23
V_4 (Physical violence)
1.90 (−7.04; 10.84)
2.60
17.57 (4.11;31.03)**
3.91
8.10 (−2.3; 18.50)
3.03
V_5 (Physical violence)
−17.63 (−31.45; −3.82)**
4.02
−1.97 (−9.68; 5.75)
2.24
−11.43 (−23.30; 0.43)
3.45
−19.53 (−31.89; −7.18)***
3.59
V_6 (Sexualized violence)
4.07 (−3.83; 11.96)
2.30
19.73 (7.02; 32.44)***
3.70
10.27 (1.35; 19.18)*
2.60
2.17 (−8.11; 12.44)
2.99
21.70 (8.06; 35.34)***
3.97
V_7 (Physical violence)
−7.90 (−18.58; 2.78)
3.11
7.77 (−2.29; 17.82)
2.92
−1.70 (−12.48; 9.08)
3.13
−9.80 (−19.90; 0.30)
2.94
9.73 (−0.37; 19.84)
2.94
−11.97 (−23.09; −0.85)*
3.23
V_8 (Car accidents)
−4.40 (−13.07; 4.3)
2.52
11.27 (0.36; 22.18)*
3.17
1.80 (−6.01; 9.61)
2.27
−6.30 (−14.49; 1.89)
2.38
13.23 (1.73; 24.74)*
3.34
−8.47 (−18.39; 1.46)
2.89
3.50 (−4.46; 11.46)
2.31
ITC-SOPI
V_1 (Sexualized violence)
V_2 (Car accidents)
3.40 (−0.57; 7.37)
1.16
V_3 (Car accidents
8.80 (6.06; 11.54)***
0.80
5.40 (2.25; 8.55)***
0.92
V_4 (Physical violence)
0.87 (−1.98; 3.71)
0.83
−2.53 (−6.93; 1.87)
1.28
−7.93 (−11.26; −4.61)***
0.97
V_5 (Physical violence)
−4.56 (−7.77; −1.36)***
0.93
−7.97 (−11.01; −4.93)***
0.88
−13.37 (−16.32; −10.41)***
0.86
−5.43 (−8.57; −2.30)***
0.91
V_6 (Sexualized violence)
6.83 (4.51; 9.15)***
0.68
3.43 (0.41; 6.46)
0.88
−1.97 (−4.02; 0.08)
0.60
5.97 (3.45; 8.48)***
0.73
11.40 (8.80; 14.00)
0.76
V_7 (Physical violence)
0.63 (−2.40; 3.66)
0.88
−2.77 (−5.89; 0.35)
0.91
−8.17 (−11.14; 5.20)***
0.86
−0.23 (−3.25; 2.78)
0.88
5.20 (2.69; 7.71)***
0.73
−6.20 (−8.34; −4.06)***
0.62
V_8 (Car accidents)
6.53 (3.84; 9.23)***
0.79
3.13 (−0.13; 6.40)
0.95
−2.27 (−4.77; 0.23)
0.73
5.67 (2.22; 9.11)***
1.00
11.10 (8.53; 13.67)***
0.75
−0.30 ( −2.51; 1.91)
0.64
5.90 (3.47; 8.33)***
0.71
Note. PTQ = Perceived Threat Questionnaire, ITC-SOPI = Subscale ecological validity of the ITC-Sense of Presence Inventory (Lessiter et al., 2001); STAI-S = State-Trait-Anxiety-Inventory (Spielberger et al., 1983). V_1 = Sexualized violence (Excerpt from the movie Antichrist), V_2 = Car accidents (First person car crashes from “Youtube”), V_3 = Car accidents from Olatunji and Fan (2015), V_4 = Physical violence (Excerpt from the movie Irreversible), V_5 = Physical violence (Excerpt from the movie Hardcore Henry); V_6 = Sexualized violence (Excerpt from the movie Irreversible), V_7 = Physical violence (Excerpt from the movie Maniac; V_8 = Car accidents (Texting While Driving). CI = Confidence Interval; SE = Standard Error. Significance: *p < .05, **p < .01, ***p < .001
Discussion
Although the trauma film paradigm (TFP) is the gold standard experimental approach for testing hypotheses about the psychological impact of trauma, a variety of trauma films have been used that are associated with differing emotional and trauma-analogue symptoms (Arnaudova & Hagenaars, 2017; Weidmann et al., 2009). This considerable between-study heterogeneity makes it not only difficult to choose a trauma film for a specific study, compare effects across studies and generalize study-specific results (e.g., by meta-analyses), but also raises concerns about the ecological validity of the TFP itself. The present intraindividual study empirically examined several aversive film clips varying in content and often used as trauma-analogue stressors in the TFP with regard to perceived threat, perceived realism and anxiety, aiming to provide guidance for researchers using the TFP and improve its ecological validity likewise.
We found that the trauma film clips highly differed in their potential to elicit threat, realism, and anxiety. Generally, our findings align well with the prior assumptions that the specific trauma film content (i.e., sexualized violence) and a clear, cohesive plot in the film scene trigger a higher perceived threat and are perceived as more realistic (Arnaudova & Hagenaars, 2017; Baptie et al., 2021; Denisova & Cairns, 2015; Gorisse et al., 2017; Weidmann et al., 2009), while the camera perspective seems to play a minor role in determining their outcomes. Interestingly, the results showed that the content is not the only important feature as shown by the different levels of perceived threat among film clips depicting similar content. Hypothetically, the perceived threat might depend on base rate of specific events, i.e., how likely it is to experience a specific event oneself, stimulating self-referential processing. Except for the compilation of short amateur videos by Olatunji and Fan (2015), other compilations with no coherent story were rated less threatening. Contrary to our assumptions, though, the films scenes filmed from first-person perspective were rated less threatening. The realism might have been impacted by the specific design of an action.
Descriptively, the most threatening film clips were the excerpt from Irreversible (Noé, 2002) depicting a scene of sexualized violence of a man against a woman, followed by fatal car accidents from Texting While Driving and the car accidents compilation from Olatunji and Fan (2015). The least threatening film clips were an excerpt from the movie Hardcore Henry showing physical violence, and a self-made Youtube compilation depicting first person car crash accidents. The most threatening film clips were also perceived as the most realistic, natural and believable film clips, both dimensions together indicating a higher sense of presence and involvement in these film clips and making them particularly suitable for the TFP. Moreover, pairwise comparisons between trauma film clips indicated statistically significant differences on threat perception that are in line with the perceived level of realism of the trauma film clips. Although the film clip depicting sexualized violence from the movie Irreversible also evoked the highest anxiety response, all trauma film clips elicited moderate levels of post-film state-anxiety in participants and differences among film clips were less pronounced or vanished. Interestingly, some trauma film clips evoked high anxiety while not being perceived as threatening or realistic (e.g., excerpt depicting sexualized violence from the movie Antichrist).
Our findings are well in line with the study by Arnaudova and Hagenaars (2017): they found that trauma film clips did not differ in terms of valence and arousal, but differed in their potential to evoke strong unpleasant emotions (e.g., disgust, distress and embarrassment) and intrusive memories. For example, Arnaudova and Hagenaars (2017) found higher values on shame (assessed by Film Response Questionnaire) for the same sexualized violence against women scene from Irreversible compared to film clips depicting food, car accidents (Texting while driving) and physical violence (Irreversible). However, they used an almost exclusively female young sample. In contrast, only one third of the participants in our sample identified as female, indicating a more balanced gender distribution that should be considered in the interpretation of our results. This study adds to this discussion another important peritraumatic factor: the perceived threat that is closely related to the perceived realism of a trauma film clip. In fact, peritraumatic threat perception appears crucial for the development of PTSD: the predictive processing framework proposed that perceived threat positively predicts the likelihood of reexperiencing symptoms (Kube et al., 2020).
Replicating and extending previous research (Arnaudova & Hagenaars, 2017; Weidmann et al., 2009), this study demonstrated that different film clips used in TFP studies evoked distinct perceptual and emotional responses, underlining that researchers using the TFP should carefully select trauma film clips for their studies. Although anxiety during or after the exposure to a potentially traumatic event is a relevant risk factor for the development of intrusive memories (Hagenaars et al., 2010; Laposa & Alden, 2008; Laposa & Rector, 2012), researchers should also include variables such as perceived threat and realism in the selection of their trauma films to achieve a higher involvement and thus increase the ecological validity of their TFP study.
Limitations, Future Directions and Implications for Trauma Film Paradigm Research
Our study has limitations: First, we assessed only immediate effects of each film clip. Future studies should investigate also the influence of perceived threat on delayed responses such as the intrusive memories, e.g., by using our results to modulate the degree of perceived threat in a TFP study. Second, a further limitation is that we did not assess important parameters other than self-report data, unlike the study by Arnaudova and Hagenaars (2017), and future studies could integrate also physiological marker such as heart rate. Third, although the majority of other TFP studies recruited quite similar samples, i.e., young students (James et al., 2016), the selectivity of the sample is a limitation and future studies should try to include a larger variability regarding sociodemographic characteristics to have a more representative sample. Fourth, we derived trauma film clips from previous trauma-analogue studies with different lengths that potentially influenced the outcomes along other film characteristics. Descriptively, the longest trauma film clip (i.e., 6: Excerpt depicting sexualized violence from the movie Irreversible) was perceived as the most threatening and realistic, and the shortest (i.e., 5: Excerpt depicting physical violence from the movie Hardcore Henry) as the least threatening and realistic. Although no clear pattern was recognizable in the other film clips, a possible dose–response relationship should therefore be kept in mind when interpreting the results. Future studies could directly examine whether film characteristics predict different outcomes, but also whether there is a dose–response relationship. The latter would also have ethical implications of using this experimental paradigm. Fifth, while the level of perceived threat seems to be a key feature for modeling danger-based/fear-based trauma types (i.e., life threat to self or other) and its hallmark symptom of intrusive memories in the traditional TFP, our findings might not generalizable to non-danger-based traumatic events (e.g., moral injury) that are related to other outcomes (e.g., chronic guilt). Future research should consider that the equipotentiality of different traumatic events might ignore potentially unique contexts and its consequences (Stein et al., 2012), and search for the most adequate trauma films per trauma type.
When replicated in an experimental study manipulating perceived threat by using different trauma film clips, the current findings may have some implications on the selection of trauma films in the TFP. Within the context of the TFP, researchers should carefully select aversive film clips for their studies to increase the ecological validity, as different film clips seem to evoke distinct perceptual and emotional responses. When choosing a trauma film clip for a TFP study, our findings suggest to take into account not only the anxiety levels per se, but also the perceived threat and realism elicited by specific trauma film clips.
Acknowledgements
The authors are very grateful to the students who likewise contributed to the study: Elisa Schollähn, Vanessa Arndt, Stefan Wegscheider, and Nergis Mengene.
Declarations
Conflict of Interest
PH and RJM declare that they have no conflict of interest.
Informed Consent
Written informed consent was obtained from all individual participants included in the study.
Research Involving Human Participants or Animals
This study was performed in line with the ethical guidelines of the German Psychological Society and the guidelines for human studies of the World Medical Association Declaration of Helsinki. Approval was granted by the local ethics committee of the Department of Psychiatry and Psychotherapy at the University Medicine Greifswald (Reference number BB 133/21).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Met BSL Psychologie Totaal blijf je als professional steeds op de hoogte van de nieuwste ontwikkelingen binnen jouw vak. Met het online abonnement heb je toegang tot een groot aantal boeken, protocollen, vaktijdschriften en e-learnings op het gebied van psychologie en psychiatrie. Zo kun je op je gemak en wanneer het jou het beste uitkomt verdiepen in jouw vakgebied.
Predictive processing is a popular framework in neuroscience and computational psychiatry that provides a new understanding of the operating principles of the brain, shaping the way how mental disorders such as borderline personality disorder (Herzog, Kube, et al., 2022) are conceptualized. This framework assumes that the brain tries to make sense of the world by actively generating hypotheses, which are continuously refined against the background of new information resulting from sensory input. In Bayesian inference terms, a probabilistic ‘prior’ prediction is combined with the likelihood of new sensory information to compute a ‘posterior’ belief (Knill & Pouget, 2004), assuming that past experiences form predictions about future events that are then compared with new information. Mismatches between prior predictions and new sensory information are referred to as prediction error (PE). Applied to PTSD, the experience of a traumatic event in Bayesian terms can be understood as a perceptual hypothesis that is subsequently given a very high a-priori likelihood because of its (life-) threatening significance (Kube et al., 2020). As a result, this hypothesis is re-selected again and again in response to triggers, which explains re-experiencing symptoms in PTSD—although it does not fit the actual sensory input.
Choose Wisely! – Considering the Perceived Threat in the Selection of Trauma Film Clips May Improve the Ecological Validity of the Trauma Film Paradigm
