FEELING THE FEAR: USING HAPTIC FEEDBACK TO MANIPULATE 
EFFICACY AND COGNITIVE PROCESSING OF A FEAR APPEAL ADVERTISEMENT 
IN VIRTUAL REALITY 

By 

Alexander David Lover 

A THESIS 

Submitted to 
Michigan State University 
in partial fulfillment of the requirements   
for the degree of 

Media & Information – Master of Arts 

2024 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ABSTRACT 

With the rise of virtual reality technology, traditional media can be now consumed in a 

more immersive modality. Furthermore, the inclusion of haptic feedback (i.e., physical responses 

within the modality) brings a new sense to media that otherwise rely on visual/auditory senses. 

Traditional media (e.g., advertisements) need to be re-examined as our cognitive processing can 

potentially change within the new modality and accompanying technology. The present study 

investigates the role of haptic feedback on the efficacy and cognitive processing of fear appeal 

advertisements in virtual reality. Participants watched a short, low-threat safe-driving 

advertisement in virtual reality, feeling haptic feedback be played either at the beginning, middle, 

or end of the stimulus. Afterwards, participants answered questionnaire items pertaining to their 

self-reported fear, self-reported perceived threat, attitudes and behaviors/intentions towards safe 

driving, and their memory of the stimulus. Results provide support for perceived threat having a 

positive effect on attitudes and behaviors/intentions, but not for memory. There is no support for 

the positive influence of haptic feedback on perceived threat. Furthermore, there is only support 

for a significant difference in one section of memory between conditions, but no support for the 

other sections, potentially due to haptic manipulations or the chosen stimulus. In summary, this 

study suggests that haptic feedback might work better as an orienting eliciting structural feature 

for the perceptual resource pool rather than cognitive. Implications are provided as well as 

directions for future research studies. 

 
ACKNOWLEDGEMENTS 

I would like to express my heartfelt appreciation and sincere gratitude for my advisor Dr. 

Rabindra Ratan and committee members Dr. David Ewoldsen and Dr. Nancy Rhodes for their 

continued mentorship and advice throughout my master’s program. An extra thank you to Dr. 

Edward Downs as well, who inspired me to pursue graduate school back in undergrad. I also 

want to thank all members of SPARTIE Lab and SCAMP Lab for providing me with a 

supportive environment in my academic journey. My journey wouldn’t have been possible 

without the support of all these individuals above and more. Thank you. 

iii 

TABLE OF CONTENTS 

INTRODUCTION  ……………………………………………………………………………….1 

LITERATURE REVIEW  ………………………………………………………………………..3 

METHOD  ………………………………………………………………………………………11 

RESULTS  ………………………………………………………………………………………17 

DISCUSSION  ………………………………………………………………………………….21 

CONCLUSION  …………………………………………………………………………………26 

BIBLIOGRAPHY  ………………………………………………………………………………27 

APPENDIX A: SELF-REPORT FEAR ITEM………………………………………………….30 

APPENDIX B: PERCEIVED THREAT ITEMS  ………………………………………………31 

APPENDIX C: ATTITUDES TOWARDS SAFE DRIVING ITEMS  …………………………32 

APPENDIX D: BEHAVIORS & INTENTIONS TOWARDS SAFE DRIVING ITEMS  ……..33 

APPENDIX E: MEMORY RECALL ITEMS  ………………………………………………….34 

APPENDIX F: GENERAL INFORMATION ITEMS  …………………………………………36 

iv 

 
 
INTRODUCTION 

Over the past decade, virtual reality development has skyrocketed. A headset that costs 

tens of thousands of dollars now can cost as low as $300 (numbers courtesy of vr.space). These 

headsets aim to provide the most immersive experience possible, and new consumer 

technologies have been designed to accomplish this. Haptic feedback equipment, for example, 

brings physical touch to virtual reality. The company bHaptics has designed a series of 

equipment that brings haptic feedback to various parts of the body (e.g., face, hands, and torso). 

This haptic feedback strengthens feelings of presence while in virtual reality (García-Valle et al., 

2017). Virtual reality has traditionally been used for gaming, barring a few academic scholars 

that have pushed towards incorporating the tech into the classroom (e.g., Liou & Chang, 2018). 

However, companies such as Meta have pushed towards implementing the Metaverse. Coined by 

Stephenson (1992), the Metaverse describes a digital reality that mimics our own reality – a 

parallel universe of sorts, with new forms of media. This new reality has already created 

advertisements of all sorts, encouraging participants to spend real money on virtual designer 

clothing for their avatars (Mcdowell, 2023). Apple has now developed their version of a headset, 

releasing the Vision Pro. This headset is rather interesting, as it seems to mimic the functionality 

of a smartphone. Both cases seem to imply a push away from gaming contexts, now towards 

media consumption. 

Media consumption has typically been viewed as a two-dimensional interaction, the 

salient example being advertisements. An individual would watch an advertisement on their 

television, for example, then either accept or reject the message. This becomes rather interesting 

in fear appeal advertisements, that is, advertisements that aim to elicit cognitive/physiological 

fear responses in order to gain message acceptance. Frameworks such as the Extended Parallel 

1 

 
Processes Model (Witte, 1992) describe a component of perceived threat. This construct is built 

upon perceived severity and perceived susceptibility, describing the perceived danger of the 

threat and the perceived likelihood of experiencing the threat, respectively. Haptic feedback, a 

physical response, can potentially strengthen these feelings by tying this strong, physical 

sensation to the danger described. A fear appeal advertisement that was originally designed for 

television might now be processed differently in virtual reality, which already is innately 

immersive. The inclusion of these consumer technologies such as haptic feedback could then 

potentially strengthen the ads further. This issue motivates the need to re-examine media 

consumption, now in the context of virtual reality and accompanying tech. The present study 

aims to shed light on media in virtual reality by incorporating new consumer technology 

(bringing physical sensations) to existing media artifacts. All of this is done to answer the 

guiding research question of how haptic feedback affects our cognitive processing of 

advertisements. 

In this paper, I will overview the Limited Capacity Model of Motivated Mediated 

Message Processing (LC4MP; Lang, 2017; Fisher, 2020) and the Extended Parallel Processing 

Model (EPPM; Witte, 1992) to provide a theoretical framework. Alongside the EPPM, I will 

discuss haptic feedback research to supplement the accompanying hypotheses. I will then discuss 

existing research on fear appeals in virtual reality. Afterwards, I will summarize the hypotheses 

posited, and outline a series of hypotheses to build a model of the role of haptic feedback in 

cognitive processing. I conducted a quantitative experiment to test these hypotheses, in which 

participants watched a safe-driving fear appeal advertisement in virtual reality while feeling 

haptic feedback at a given time. I will finish by presenting the results found, provide a discussion 

for the theoretical importance of these results, and their implications for future research. Through 

2 

this novel study and approach, I hope to expand our theoretical understanding of perceived threat 

and memory through haptic feedback. 

LITERATURE REVIEW 

In this section, I will review the Limited Capacity Model of Motivated Mediated Message 

Processing (LC4MP; Lang, 2017; Fisher, 2020) to explain the cognitive processing component 

of the study. I will also discuss the literature on fear appeals, using the Extended Parallel 

Processes Model (EPPM; Witte, 1992) as a framework. I will also describe research on haptic 

feedback alongside fear appeals to provide holistic justification for the accompanying 

hypotheses. Afterwards, I will also briefly discuss the relevant research on fear appeals (and 

more broadly, fear arousal) in virtual reality to provide a practical exigency for the present thesis. 

Finally, I will summarize the hypotheses posited, and will put forward a hypothesized mediated 

model that will be tested in the experiment proper. 

THE LIMITED CAPACITY MODEL OF MOTIVATED MEDIATED MESSAGE 

PROCESSING 

To best explain our cognitive processing, I will also be using the Limited Capacity Model 

of Motivated Mediated Message Processing (LC4MP; Lang, 2017; Fisher, 2020). This model 

assumes that humans have a limited capacity when it comes to processing a message, that we 

process information through natural motivation, cognitive resource allocation can be measured 

through psychophysiological reactions, and finally that all communication media is mediated 

(Lang, 2017; Fisher et al., 2018). Generally speaking, these resources are measured through 

cognitive measures such as memory, or physiological measures such as heart rate, attention, and 

secondary-task reaction times (Fisher, 2020). Furthermore, there has been research on the 

inclusion of a perceptual resource pool, a secondary pool of resources that drives perception 

3 

 
(Fisher et al., 2018). This perceptual pool drives visual and audio identification of unique 

components in the stimulus (Fisher, 2020).  

An interesting component of the LC4MP is the orienting response, which is simply a 

physiological and cognitive shift in the viewer to start paying attention to the message. Haptic 

feedback is assumed to act as an orienting response driver–typically called an orienting eliciting 

structural feature (OESF)–which cues a viewer to start paying attention to the upcoming content 

(Fisher, 2020). Haptics could be a sort of dual-purposed OESF, potentially cueing the viewer 

either for the cognitive pool, perceptual pool, or both. Therefore, the goal with this experiment is 

to first investigate the cognitive resource pool by using a higher-order function like memory.  

The research on the LC4MP has been applied to several different contexts. While this 

review won’t be exhaustive, there are a few key studies to note as they have helped frame and 

design the general research protocol of the present study. A study found that the structure and 

content of a health communication message on cancer directly influences how effectively it is 

encoded, stored, and retrieved at a later time (Lang, 2006). Lang’s study prompted the idea that 

haptic feedback can also act as an added structure to the content, potentially supplementing the 

narrative or message being shown. In a study pertaining to safe driving public service 

announcements, results suggested that message processing decreased as the intensity of the 

advertisement increased - presumably due to the limited capacity of resources we can allocate 

(Rhodes, 2017). This study has been helpful in two ways. First, it inspired the idea that a safe-

driving advertisement, within the framework of the LC4MP, would work well with haptic 

feedback. If the participant watched a car crash on their television, they would imagine themself 

in that situation and narrative, similarly to how they would via the Extended Transportation-

Imagery Model (Van Laer et al., 2014). However, with the inclusion of haptic feedback, I argue 

4 

that there is less imagination as there is now a physical response directly tied to the crash, further 

supporting haptics as a new layer of structure to media. This claim will be further examined in 

the fear appeals section later in the paper. Rhodes (2017) study also suggests a sort of ceiling 

effect, where an intense advertisement decreased message processing. This finding is further 

supported by Plant et al. (2017), which suggests that graphic imagery leads to non-effects in 

outcomes. These studies help guide the choice for stimulus, as I hope to avoid any ceiling effect 

where message processing stops. Hence the decision to use a lower-threat advertisement on safe 

driving. 

Using the LC4MP, a number of hypotheses can be made relating to general memory of 

the advertisement. The primary assumption in the present study is that haptic feedback can act as 

a forced orienting eliciting structural feature (OESF) of the stimulus. Therefore, the hypotheses 

relate to specific sections of memory of the advertisement, as I hope to first target the cognitive 

resource pool. For example, if haptic feedback is played in the middle of the ad, then we should 

see stronger memory of the second half of the advertisement rather than the first, as the haptics 

would elicit an orienting response for the remainder of the advertisement. This also illustrates the 

primary manipulation of the study, as participants will either feel haptic feedback at the 

beginning of the stimulus (an introductory OESF), in the middle right before the car crash of the 

stimulus (a threat-based OESF), or at the end of the stimulus (a concluding OESF). A general 

research question is also put forward, as there isn’t enough information to safely hypothesize that 

the memory of the second half of the advertisement will differ between two specific conditions. 

Hypothesis 1a: When haptic feedback is used as an introductory OESF, memory of the 

advertisement will be stronger than when haptic feedback is used as a concluding OESF.  

5 

 
Hypothesis 1b: When haptic feedback is used as a threat-based OESF (i.e., haptics in the middle 

of the advertisement), memory of the second half of the advertisement (five recall questions on 

the last 30 seconds of the stimulus) will be stronger than the first half of the advertisement (five 

recall questions on the first 30 seconds of the stimulus). 

RQ: When haptic feedback is used as an introductory OESF, how will the participant’s memory 

of the second half of the stimulus compared to when haptic feedback is used as a threat-based 

OESF.  

FEAR APPEALS & HAPTICS 

I will be discussing both fear appeals and haptics in this section, overviewing the 

Extended Parallel Processes Model and salient research. The hypotheses view haptic feedback as 

a manipulation of perceived threat (a prominent component of fear appeals), so it serves well to 

discuss both simultaneously. While there is extensive research on both subjects individually, 

there is seemingly no research on the intersection of the two. After covering each topic, I will 

posit a few relevant hypotheses. 

I will be utilizing the Extended Parallel Processes Model (EPPM; Witte 1992). The 

EPPM includes components such as perceived threat and perceived efficacy that ultimately lead 

to message acceptance or rejection. Perceived threat is the primary focus for the present study 

and is built upon perceived threat (i.e., how severe the individual perceives the threat to be) and 

perceived susceptibility (i.e., how likely the individual perceives they could be affected by the 

threat). In tandem with perceived threat, the model describes fear as an emotion, which acts as a 

feedback loop for perceived threat (Witte, 1992). Too much fear might lead to fear control 

processes, ultimately leading to message failure. The model does well to explain why a fear 

appeal might succeed or otherwise fail (Witte, 1996). Witte (1992) also puts forward the 

6 

 
proposition that when the perceived threat of the stimulus is low, no message processing will 

occur. This proposition has traditionally been tested indirectly through message acceptance 

measures such as attitudes or behaviors/intentions (Popova, 2012; Lewis et al., 2007). The 

propositions of the EPPM have seen mixed empirical support (for review, see Popova, 2012), 

though testing the proposition via direct message processing outcomes like memory should prove 

to be beneficial. The LC4MP is a useful tool in adding more layers of message processing when 

testing the outcomes of perceived threat, rather than relying only on attitudinal or behavioral 

change, hence the inclusion in the present study.  

Salient to this thesis, research involving safe driving campaigns are important to note. 

When attempting to reduce the number of risky driving behaviors, many advertising campaigns 

have utilized fear appeals as a way to scare viewers into the desired behavior. The EPPM is a 

beneficial framework to utilize in designing and evaluating these campaigns (Cismaru, 2014), 

studying contexts such as texting and driving (e.g., Cismaru & Nimegeers, 2017), speeding (e.g., 

Rhodes, 2017), or drunk driving (e.g., Ngene, 2023). While safe driving as a topic seems to work 

well with haptic feedback, which specific context to use becomes a focus in choosing the 

stimulus for this study. The stimulus chosen relates to driving over the speed limit, though this 

choice was more due to the requirements of the stimulus than the topic itself. For example, a car 

crash needed to be depicted halfway through the ad to create more even timings between the 

different haptic feedback for each condition, in which a specific public service announcement 

about speeding was executed well. Future research should investigate these extra topics as well 

to further define the boundaries of haptics in cognitive processing. 

It serves well to discuss haptic feedback here as well, given the initial claim of haptics 

influencing perceived threat. While research has been conducted in fear-appeal and fear-arousal 

7 

 
 
fields, seemingly no research has been done investigating the role of physical sensations (i.e., 

haptic feedback) in fear appeals. This isn’t terribly surprising, given traditional media 

consumption only relies on sight and hearing, not touch. Generally speaking, haptic feedback 

strongly influences virtual presence and immersion (Kreimeier et al., 2019; Lee & Kim, 2008; 

García-Valle et al., 2017). In a fear appeal advertisement, viewers typically have to imagine 

themselves in that situation with the depicted threat, relying on a cognitive effort to make a 

perception about the danger and likelihood of the threat. I argue that the added feelings of 

presence and immersion alongside external physical responses to the stimulus content acts as a 

realized threat, which would then elicit stronger senses of susceptibility to the danger, rather than 

relying on the viewer to imagine themselves in the situation. Given that haptic feedback will be 

more accessible by these VR companies (as they have control of those environments and 

accompanying tech), there are far reaching implications within advertising and media 

consumption. 

Therefore, I hypothesize that depending on when the haptic feedback is played during the 

advertisement, there will be an influence on the perceived threat and fear of the stimulus, as 

haptic feedback could directly influence the perceived severity and perceived susceptibility of 

the threat. 

Hypothesis 2: When haptic feedback is used as a threat-based orienting eliciting structural 

feature (OESF), participants’ perceived threat will be greater than when haptic feedback is used 

as an introductory OESF (H2a), and greater than when haptic feedback is used as concluding 

OESF (H2b). 

8 

 
Hypothesis 3: When haptic feedback is used as a threat-based OESF, participant’s fear towards 

the advertisement will be greater than when haptic feedback is used as an introductory OESF 

(H3a), and greater than when haptic feedback is used as a concluding OESF (H3b). 

I have also put forward a few hypotheses to examine the first proposition of the EPPM 

more closely. The hypotheses towards attitudes and behaviors/intentions aren’t novel like the 

hypothesis towards memory, though all aim to expand the boundaries of said proposition. 

Hypothesis 4: The perceived threat of the advertisement will positively influence the overall 

attitudes towards safe driving (H4a), behaviors/intentions towards safe driving (H4b), and 

memory of the advertisement (H4c).  

VIRTUAL REALITY & FEAR APPEALS 

Within the context of VR, little research has been done investigating fear appeals. One 

study had participants watch a safe driving ad – which was either a fear appeal or a positively-

framed ad – on either a computer screen or in VR. This study suggests that positively framed ads 

decreased risky driving, which was further supported specifically in VR; however, they did find 

that when participants watched the fear appeal in VR, it actually increased their albeit self-

reported risky driving behaviors (Cutello et al., 2021). As discussed previously, Rhodes (2017) 

study suggested a similar effect. Fear overload is certainly important to avoid in the present study 

to not rely on individual heuristics. Another study employed a more immersive experience, 

having participants work through a construction safety program. The experience aimed to elicit 

fear arousal within the participants, ultimately finding that it improved their attitudes towards 

workplace safety (Hoang et al., 2021). Using fear to lead viewers to message acceptance is 

definitely a salient finding, and as it helps support the practical goal of using fear for positive 

9 

 
 
outcomes. The advent of virtual reality headsets designed for media consumption could elicit 

positive change but needs to be examined carefully. 

SUMMARY OF HYPOTHESES AND OVERALL MODEL 

Using the literature discussed, I have put forward a number of hypotheses that describe 

the different interactions of haptic feedback. In summary, I hypothesized that haptic timing (i.e., 

when in the stimulus haptic is used), there will be an influence on perceived threat, fear, and 

memory. Through these hypotheses, a general, mediated model detailing the relationship 

between haptics, perceived threat, attitudes, behaviors/intentions, and memory can be built.  

Hypothesis 5: The relationship between haptic timing and attitudes (H5a), behaviors/intentions 

(H5b), and memory (H5c) will be mediated by the perceived threat of the advertisement. 

These hypotheses work together to build a general model illustrating the role of haptic 

feedback in our cognitive processing of fear appeal advertisements. See figure 1 below for the 

full hypothesized model. The feedback loop between perceived threat and fear is described 

explicitly within the EPPM, and the interaction between perceived threat and message outcomes 

(attitudes, behaviors/intentions, and memory) is based on the first proposition of the EPPM. The 

present study intends to add another layer to this interaction by incorporating haptic feedback. 

 Figure 1 - Proposed model 

10 

 
RESEARCH DESIGN AND PARTICIPANTS 

METHOD 

Undergraduate students from Michigan State University (N = 101, 46 cisgender men, 53 

cisgender women, 2 preferred not to disclose) participated in a between subjects experiment in 

which they viewed a safe-driving fear appeal advertisement in a virtual reality headset. 

Participants were randomly assigned to one of three conditions: haptic feedback was either 

played at the beginning (i.e., an introductory OESF), the middle (creating a threat-based OESF), 

or at the end (creating a concluding OESF). Afterwards, participants responded to survey items 

on a 2022 MacBook Pro. On average, the entire study lasted about 10 minutes. Participants 

ranged from 18 to 32 years old (8 freshmen, 18 sophomores, 52 juniors, 20 seniors, 3 senior+). 

Within the 101 participants, 71 are white, 10 are Asian and/or Asian American, 8 are black 

and/or African American, 5 are Hispanic/Latinx, 1 is Middle Eastern, and 6 reported dual 

ethnicities (3 Hispanic/white, 1 white/Asian, 1 black/white, 1 black/Hispanic). Students were 

recruited from the SONA participant pool on campus – undergraduates signed up in exchange for 

credit, which either was counted as a letter grade or as extra credit depending on different 

classes’ syllabuses. 205 time slots were generated, 108 participated in the study for credit (7 

students were excluded), around 50 students did not show up for the study, around 40 time slots 

were not assigned, and 10 participants had to be canceled due to technical issues (Meta servers 

being down stopped headsets from functioning for a day). 

For data analysis, it was determined a priori to exclude any incomplete responses from 

analysis, low standard deviation of responses (evidence of straight lining), or any unverifiable 

manipulation. The emphasis on unverifiable is placed as 23 participants failed the manipulation 

check, either reporting that they felt the haptic feedback at an unassigned time, or that they didn’t 

feel any haptic feedback. The researcher had listened for the haptic feedback during each 

11 

 
experiment, so the manipulation for the remaining sample is confirmed and will be tested as if 

they subconsciously recognized the haptic. However, this does introduce a new data point, which 

will be called ‘manipulation awareness’, given the interesting interactions with memory and 

physiological responses. Of the 108 final participants, six were excluded from analysis due to 

unverified manipulation because of the lack of cross checking with the data sheet, and one 

participant was excluded for low standard deviation (SD < .50 was the cutoff for exclusion), 

leaving 101 participants for analysis (N = 101). 

For data cleaning, prior VR experience included an open-ended question asking how 

many hours the participant has used VR within the past year, which included response types such 

as “2 hours”, “2 hours and 30 minutes”, or “7-8 hours”. These responses were changed to “2”, 

“2.5”, and “7.5” respectively. The median of the range of hours should act as a better 

representation of the range, rather than relying on either end of the spectrum. In total, there were 

15 cases that converted text and minutes to hours, and three cases that converted a range to a 

median. Participants who self-described as man or woman were grouped with cisgender man or 

cisgender woman (two instances of this case), respectively, and the individuals who preferred not 

to disclose their gender was excluded from any gender-based analysis. The sample also was 

predominantly white, so analysis categorized race/ethnicity by white versus non-white for 

statistical power.  

EQUIPMENT AND STIMULUS 

Participants used a Meta Quest 2 headset for the study. This was a factory default 

headset, using a third-party backstrap for comfort. The stimulus experience was programmed in 

Unity version 2022.1.23f1. The haptic feedback was administered via the bHaptics TactVisor, a 

third-party cover that covers the cushion around the user’s eyes. There are four motors evenly 

placed around the top of the rim of the visor and would play the haptic feedback for 1.5 seconds 

12 

at the appropriate time (depending on the condition). No controllers were needed from the study, 

and the researcher controlled the transitions of scenes in the headset on their end. Participants 

completed the post-test survey on a 2022 Macbook Pro with a 2018 Apple mouse. 

Participants viewed an eye test within the headset to determine if the lens were placed 

respective to the participants’ inter-pupillary distance. The main stimulus included a floating 

movie theater overhanging a simple road. See figure 2 below for a screenshot of the experience. 

Participants stood at the side of the road, looking across towards the video. The video itself was a 

minute-long safe-driving public service announcement in which a narrator describes the 

difference between a car crash at 60 km/hour and 65 km/hour. This stimulus was used in a 

previous study (Rhodes, 2017) and was deemed low threat. A low-threat stimulus was chosen to 

avoid a ceiling effect in the interaction.  

Figure 2 - A screenshot of the primary stimulus 

MANIPULATIONS 

There are three potential conditions for participants: either haptic feedback was played at 

the beginning (condition A; n = 34), the middle (condition B; n = 33), or the end (condition C; n 

= 34), all for 1.5 seconds. The stimulus is exactly 60 seconds long, so this haptic feedback played 

at the 0 seconds, 30 seconds, or 58.5 seconds marks. 58.5 seconds was chosen as no further 

information in the stimulus is provided and the haptic feedback bookmarks the end of the 

stimulus perfectly. 

13 

 
Haptic feedback being played at the beginning is assumed to represent a traditional 

orienting eliciting structural feature (OESF) in which the rumbling cues the participant in to start 

paying attention (which is labeled as an introductory OESF). Haptic feedback being played in the 

middle happens right before the car crash scene in the stimulus, which was hypothesized to 

further increase perceived threat and fear of the stimulus, becoming a structural component of 

the advertisement alongside the general manipulations of timing. Due to this added structural 

component, condition B was labeled as a threat-based OESF. Haptic feedback being played at 

the end (labeled as a concluding OESF) was hypothesized to not have any effect on the measures 

but was included to provide a comparison to the other conditions. 

PROCEDURE 

After participants arrived, their anonymous ID was confirmed so course credit was given 

to the correct individual. After this confirmation, they were instructed to sit at the computer and 

read through the consent form after silencing their phone. After the consent process, the 

researcher helped participants put on the headset and adjusted as needed. Participants first 

viewed an eye test to see if the lens were adjusted properly respective to their inter-pupillary 

distance. Adjustments were made as needed, followed by a quick tutorial of the haptic feedback. 

This haptic feedback was played for 1.5 seconds to prepare the participant; this was also chosen 

so that the self-reported fear item of the survey wasn’t exacerbated by the shock and/or surprise 

of the haptic feedback. The participant was then taken to the side of a virtual road, where they 

watched the video play above the road. After the stimulus finished playing, the researcher 

instructed participants to take off their headset, and return to the computer to finish the survey. 

After completing the survey, participants received their credit. 

14 

 
 
MEASURES 

Participants answered a number of survey items relating to their self-reported fear, 

perceived threat of risky driving, attitudes towards safe driving, their self-reported behaviors and 

intentions towards driving safely, their overall memory of the first half and second half of the 

stimulus, their prior VR experience and their demographic information. Measures will be 

reported in order of the study procedure.  

Self-Reported Fear. A 7-point Likert scale measuring the participant’s self-reported fear was 

adapted from an existing study on fear appeals (Carey & Sarma, 2016; α = .88). Participants 

reported their level of agreement to feeling various emotions such as afraid, panicked, worried, 

nervous, etc. See Appendix A for all seven items. 

Perceived Threat. This measure employed a previous study’s 7-point Likert scale (Shi & Smith, 

2016) to gauge the participants perceived threat of the stimulus (α = .62). This scale includes 

three items for perceived severity (α = .60) and three items for perceived susceptibility (α = .59), 

as per the EPPM. Participants reported their level of agreement on items such as “Getting into a 

car crash could be fatal for me” for severity and “I’m at risk for getting in a car crash while I’m 

on the road” for susceptibility. See Appendix B for all six items. 

Attitudes. Attitudes towards safe driving were also measured. The scale was adapted from 

Roskos-Ewoldsen et al. (2004) and included six 7-point semantic differential items. Participants 

reported their attitudes on spectrums such as negative/positive, unsafe/safe, and boring/exciting 

(α = .68). See Appendix C for a full list of all 6 items. 

Behaviors & Intentions. The behaviors/intentions scale from Roskos-Ewoldsen et al. (2004) was 

also adapted to fit the present study. Participants reported their level of agreement on five items, 

including “I plan to encourage my friends or family to drive the speed limit” and “I do not intend 

15 

 
to drive the speed limit in the near future” (reverse coded; α = .73). See Appendix D for the full 

list of five items. 

Memory. Ten memory recall questions were created for the stimulus. This included a mix of 

purely visual, purely audial, or both visual/audio questions. Given that the stimulus is exactly 60 

seconds, five questions asked about the first 30 seconds of the stimulus, and five questions asked 

about the last 30 seconds. The goal in splitting the stimulus in half is to test if haptic feedback is 

strong enough as an OESF to target memory of a specific section. For example, I hypothesized 

that the second half memory will be stronger than the first half if the orienting response happens 

in the middle. These responses were graded, which created three new variables: first half score 

(skewness = -.54, kurtosis = -.17, mean = 3.59, SD = 1.05), second half score (skewness = -.44, 

kurtosis = -.79, mean = 3.52, SD = 1.20), and total score (skewness = -.52, kurtosis = -.31, mean 

= 7.12, SD = 1.81). See Appendix E for the full list of ten items used in the present study. 

Novelty & Demographic Information. Participants ended the survey by reporting their prior VR 

experience (both in number of times used and a total hour count), and their demographic 

information. Participants were asked about their gender identity, age, race/ethnicity, and 

academic year (e.g., freshmen vs. sophomore). The manipulation check of the study was placed 

here as to not clue the participants into the exact manipulations of the study immediately. This 

item asked participants when they felt the haptic feedback, which included each of the three 

conditions, and a ‘no haptic felt’ option as well. Gender, race/ethnicity, prior VR experience, 

their glasses being on or off (if applicable) will all act as potential control variables. For a full list 

of items, see Appendix F. 

16 

RESULTS 

In order to test the hypotheses and the research question, a number of statistical tests were 

conducted. Simple independent samples t-tests were conducted to compare specific sections of 

memory between the three conditions. Analysis of variance (ANOVA), alongside a Tukey post-

hoc test comparing means, was used to compare the means of each condition to the dependent 

variables. An analysis of covariance (ANCOVA) was also used to better control for gender, race, 

age, and prior VR experience. Haye’s PROCESS Model (2012) helped test the mediation model 

built from the hypotheses as well. These tests were conducted on SPSS version 29 on a 2022 

MacBook Pro. 

Of the 101 responses, 23 individuals failed the manipulation check (22.8% of sample). 

This, of course, prompts the question why did so many of them fail? Feeling the haptic, but at the 

wrong time (na = 6 cases, nb = 5 cases, nc = 3 cases) is another data point for memory. However, 

not feeling any haptic at all (na = 6 cases, nb = 1 cases, nc = 2 cases), can act as a data point for 

physiological measures alongside cognitive memory. The haptic feedback was hypothesized to 

be a tool to elicit stronger perceived threat (directly relating to fear, per the EPPM). A tutorial 

was played before the experience proper so as to not affect fear responses due to shock or 

surprise. The haptic manipulation for each final participant was confirmed, so it serves well to 

investigate if their awareness of the haptic feedback influenced these cognitive and pseudo-

physiological variables in any way. Alongside the test for each hypothesis regarding perceived 

threat, fear, and memory, I will also be analyzing the relationship using manipulation awareness 

to further examine this anomaly. For analysis, this variable was broken into three levels – they 

either felt the haptic at the correct time, felt the haptic at the ‘wrong’ time, or felt ‘no’ haptic.  

17 

Hypothesis 1a stated that when haptic feedback is played at the beginning of the stimulus 

(condition A), memory will be stronger than when haptic feedback is played at the end 

(condition C). The memory recall questionnaire wasn’t pre-tested, so each section of memory 

will be compared here to roughly gauge if the items were too easy or too challenging to 

complete. There were no significant differences found between the total scores and first half 

scores, but there was a significant difference in the second half scores (t(66) = 2.24, p <.05). I 

confirmed this finding by running an ANCOVA, using haptic timing as the fixed factor, glasses 

being off or not applicable as a nominal covariate, and second half scores as the dependent 

variable (F(2, 65) = 3.94, p < .05, ηp

2 = .11). The memory recall items are a mix of visual/audio 

information, so incorporating glasses being taken off for select students is necessary. While this 

result should be viewed cautiously, there are some potential implications about the nature of 

haptic feedback in the context of safe driving. However, such implications should be viewed as 

conjecture until a more thorough pre-test of the items can be conducted. See figure 3 below for 

differences in second half score memory. 

Figure 3 - Differences in Second Half Scores Between Conditions 

18 

 
Hypothesis 1b stated that when haptic feedback is played in the middle, memory of the 

second half of the stimulus will be stronger than the first half of the stimulus. This hypothesis 

was made as haptic feedback is generally hypothesized to act as an OESF, which would in turn 

positively affect memory of the stimulus following the haptic feedback. This relationship is non-

significant, but trends in the opposite direction – the first half scores (mean = 3.71, SD = 1.03) 

are stronger than the second half scores (mean = 3.47, SD = 1.16) when haptic is played in the 

middle. 

The research question asked how the memory of the second half of the stimulus when 

haptic feedback is played at an introductory OESF (condition A) compared to haptic feedback as 

a threat-based OESF (condition B). This was posed as a research question due to the lack of 

understanding in how long an OESF occurs (i.e., if the introductory OESF is still in effect when 

the other hypothetical OESF starts). The second half scores for the introductory OESF (mean = 

3.88, SD = 1.07) are stronger than the second half scores for the threat-based OESF (mean = 

3.47, SD = 1.16), but there was no significant difference found. 

Hypothesis 2 detailed that when haptic feedback is used as a threat-based OESF 

(condition B), perceived threat will be greater than when haptic feedback is used as an 

introductory OESF (condition A; H2a) and concluding OESF (condition C; H2b). Hypotheses 

2a/2b are not supported, as an ANOVA test suggests the differences between condition B (mean 

=  4.33, SD = .55), condition A (mean = 4.33, SD = .47), and condition C (mean =  4.20, SD = 

.46) are non-significant (F(2, 98) = .639, p = .53, ηp

2 = .01) using haptic timing as a fixed factor 

and perceived threat as a dependent variable. An ANCOVA test also suggests that when 

manipulation awareness as a covariate is added to the previous test, no significant differences 

were found (F(2, 98) = 1.095, p = .36, ηp

2 = .03).  

19 

 
Hypothesis 3 detailed that when haptic feedback is used as a threat-based OESF 

(condition B), self-reported fear will be greater than when haptic feedback is used as an 

introductory OESF (condition A; H3a) and concluding OESF (condition C; H3b). Hypotheses 

3a/3b are not supported, as the ANOVA test suggests that the differences between condition B 

(mean = 2.97, SD = .75), condition A (mean = 2.90, SD = .79), and condition C (mean = 2.61, 

SD = .88) are non-significant (F(2, 98) = 1.68, p = .191, ηp

2 = .03) using haptic timing as a fixed 

factor and self-reported fear as a dependent variable. However, an ANCOVA test, when adding 

manipulation awareness as a covariate, shows results much closer to significance (F(2, 98) = 

2.422, p = .07, ηp

2 = .07). This approach to significance should be taken with caution, as it might 

not become significant with better sample size power. These results trend in the hypothesized 

direction implying that the manipulation awareness is affecting this particular measure.  

Hypothesis 4 stated that the perceived threat of the stimulus will positively influence 

attitudes (H4a), behaviors/intentions (H4b), and memory (H4c). These are broad tests of the first 

proposition of the EPPM. I further hypothesized that perceived threat acts as a mediating 

variable between haptic timing and attitudes (H5a), behaviors/intentions (H5b), and memory 

(H5c). To test all these hypotheses, I used the PROCESS model to conduct a regression analysis 

for each part of the mediated relationship at once. This hypothesized model was tested as model 

#4 in SPSS. In this test, haptic timing was used as the independent variable, perceived threat was 

used at the mediator variable, and the dependent variable was attitudes, behaviors/intentions, and 

memory, respective to each hypothesis. Hypothesis 4a is supported (β = .196, SE = .085, p < 

.05), alongside hypothesis 4b (β = .23, SE = .10, p < .05), providing support for the impact of 

perceived threat on message outcomes. However, hypothesis 4c is not supported (β = -.03, SE = 

.37, p = .93), suggesting that perceived threat does not influence memory of the advertisement. 

20 

 
 
An ANCOVA adding manipulation awareness as a covariate further confirms this lack of 

support. Hypotheses 5a, 5b, and 5c are all unsupported, as there was no significant relationship 

found between haptic timing and perceived threat, breaking the first part of the mediated 

relationship (β = .07, SE = .10, p = .49).  

DISCUSSION 

The present study examined the role of haptic feedback in cognitive processing by using 

the context of fear appeals in virtual reality. To do so, both the Limited Capacity Model of 

Motivated Mediated Message Processing (LC4MP; Lang, 2017, Fisher, 2020) and the Extended 

Parallel Processes Model (EPPM; Witte, 1992) were used to provide a theoretical framework. 

The study had participants view a short safe-driving advertisement in virtual reality, and felt 

haptic feedback at the beginning, middle, or end of the video. Afterwards, they responded to 

items measuring their fear, perceived threat, attitudes towards safe driving, behaviors/intentions 

towards safe driving, and memory of the stimulus. Consistent with previous research, perceived 

threat was found to affect attitudes, behaviors, and intentions towards safe driving (Popova 2012; 

Lewis et al., 2007). However, the timing of haptic feedback was not found to have any 

significant effect on the perceived threat of the advertisement. Haptic timing only showed 

significant differences in memory of the second half of the stimulus, not the first half, nor the 

entire stimulus together. 

An interesting finding of this study was the variable manipulation awareness. This new 

data point was created as 23 of the 101 participants had reported feeling haptic at an otherwise 

unassigned time. For example, some participants reported that they felt the haptic feedback play 

during the middle of the advertisement, when their random condition assignment has haptic 

feedback playing at the end of the advertisement. The researcher verified the manipulation 

21 

during each experiment, so the data was treated as the manipulations being correct and 

subconscious, while ‘manipulation awareness’ became another variable to explore in the data. 

While manipulation awareness didn’t have a significant effect on memory, it drastically 

increased the differences in self-reported fear (p = .07 rather than the initial p = .191). The 

tutorial sequence exists to get the participant used to the haptic feedback, rather than the shock or 

surprise affecting their levels of fear. However, this shift might suggest that the tutorial did not 

achieve the desired effect. Therefore, future iterations of the experiment should refine the tutorial 

further, potentially adding repeated exposures, or a longer break between the tutorial and the 

experience proper. Failing to remember the manipulation timing is indicative of both memory 

and physical feelings, which is salient to the nature of this study, so it is worth exploring further.  

The research question was placed in lieu of a hypothesis as there is not a strong 

understanding in how long an orienting response occurs. The second half scores being stronger 

in the first condition (haptic being used as an introductory OESF) compared to the second 

condition (haptics being used as a threat-based OESF) isn’t terribly surprising, as both conditions 

aim to elicit an orienting response to the remainder of the stimulus. However, the first hypothesis 

(H1b) becomes a rather interesting comparison to the research question. Hypothesis 1b states that 

when haptic feedback is used in the middle of the advertisement (condition B; threat-based 

OESF), memory of the second half of the ad would be stronger than memory of the first half of 

the ad. This hypothesis was an outcome of haptic feedback being viewed as a stronger OESF, 

which would in turn cue the participant to pay attention to the remainder of the stimulus. Not 

only was this hypothesis not supported, but data trended in the opposite direction (i.e., first half 

scores being stronger than second half). This finding is presumably due to the innate OESF that 

is present within the stimulus already, excluding any presence of haptic feedback. Repeated 

22 

exposure to OESFs were found to diminish any effects (Fisher, 2020), so although haptic 

feedback is seen as a stronger OESF, that could explain the results seen here. 

There are a few explanations for the unsupported hypotheses. The first proposition of the 

EPPM states that when perceived threat is low, no message processing will occur. This was 

traditionally measured through message acceptance measures like attitudes and 

behaviors/intentions (Popova, 2012). While perceived threat was found to positively affect 

attitudes and behaviors/intentions, it is plausible that perceived threat was still low enough to halt 

all forms of direct message processing like memory storage and recall. This indicates a potential 

floor effect to avoid in future research. It is also possible that the current study isn’t using an 

optimal configuration of haptic feedback. A longer duration of haptic feedback (e.g., 3 seconds 

instead of 1.5 seconds), a different location of haptic feedback (e.g., on chest rather than face), or 

repeated exposure to haptic feedback (e.g., played consistently through the ad rather than once) 

all might have different effects on perceived severity and perceived susceptibility. While 

repeated OESFs diminish effects of processing (Fisher, 2020), the potential increases to 

perceived threat could mitigate this. Given that the stimulus was low threat innately, it seems that 

we have hit a floor effect with the hypotheses regarding haptic timing, perceived threat, and 

direct message processing outcomes. While haptic feedback is viewed as a tool for eliciting 

orienting responses, it’s entirely plausible that haptic feedback consumes available resources, 

which would then run into the same issue in halting message processing. A secondary-task 

reaction time would help illustrate the consumption and distribution of available resources to 

examine this claim (Fisher, 2020). 

Haptic feedback might also work better as the type of orienting eliciting structural feature 

(OESF) that would prime attention and secondary-task reaction times, rather than memory. The 

23 

perceptual resource pool of the LC4MP is fairly subconscious (Fisher et al., 2018), further 

suggesting that haptic feedback might be better aligned with that pool of resources given the 

differences in manipulation awareness. However, that claim will remain conjectured until a full 

study including physiological measures, measures of attention, and secondary-task reaction times 

is conducted. While some support was found in significant differences in sections of memory, 

other sections trended in the opposite direction from the hypotheses. This could be attributed to 

poor memory recall items, as the entire quiz was fairly left skewed. 

LIMITATIONS 

There are limitations in this study that should be addressed. A power analysis indicated 

that 120 participants were needed for the sample size (per the effect size of the LC4MP). 

Therefore, the final sample size of N = 101 was insufficient for proper analysis. The participant 

recruitment issues posed to be a challenge, so the hypotheses that approached significance could 

have potentially been swayed. The decision to group participants into white versus non-white in 

race-based analyses is not ideal either, as it is rather reductionist in identifying differences. 

The lack of physiological measures such as heart rate, and more physical measures such 

as attention or secondary-task reaction times also is a prominent limitation of the present thesis. 

The unsupported hypotheses regarding memory suggest that haptic feedback might work as a 

manipulation for the perceptual resource pool, influencing measures like secondary-task reaction 

times more than cognitive memory. Physiological measures like heart rate would determine if 

haptic feedback is eliciting an orienting response at all (Fisher, 2020). The low reliability of the 

perceived threat also is a limitation, as the general hypothesized model relies on perceived threat 

as a mediation between haptic timing and message processing outcomes.  

24 

 
 
FUTURE WORK 

Future iterations of this research should aim to include a number of extra measures. 

Measuring attention to the content and secondary-task reaction times would help illuminate the 

perceptual resource pool of the LC4MP (Fisher et al., 2018; Fisher et al., 2019). It is possible that 

haptic feedback works better to prepare these types of responses rather than purely cognitive 

measures. Furthermore, measuring physiological responses such as heart rate and skin 

conductance will help determine if haptic feedback is eliciting an orienting response at all 

(Fisher, 2020). These added measures should better gauge the cognitive resource pool capacity, 

allocation, and distribution. 

A pre-test of the stimulus, perceived threat, and memory recall questionnaire should also 

be conducted. Such a pre-test would provide a potential comparison group of no haptic, no 

virtual reality to see if the presence of haptic feedback is influencing perceived threat or memory 

in the next study proper. A pre-test can also help refine the memory recall questionnaire to create 

a normal distribution in the scores, rather than the current left skewed data, and help create a 

higher reliability scale for measuring perceived threat. A higher threat stimulus, with similar 

criteria to the current video, should also be chosen to avoid the floor effect of the first EPPM 

proposition.  

With these added measures, a pre-test conducted to refine memory and perceived threat 

items, and a higher threat stimulus, a study of similar design could create a more holistic picture 

to better determine the cause for the manipulation awareness issue of the present thesis. These 

limitations addressed and implemented can also better test the hypotheses put forward in this 

thesis, as they will examine both the cognitive and perceptual resource pools. 

25 

 
 
CONCLUSION 

The present study provided more support for the interaction between perceived threat and 

attitudes/behaviors. While there was no support found for the interaction between the timing of 

haptic feedback and perceived threat, the anomaly of manipulation awareness begs future 

research to investigate further as there seems to be some form of manipulation within the 

perceptual resource pool of the LC4MP. Haptic feedback has promise in influencing perceived 

threat and cognitive resources but requires more data to answer the question more holistically. 

This study acts as a preliminary report to inform my future program of research during my 

doctoral journey. 

26 

 
 
 
BIBLIOGRAPHY 

Cismaru, M. (2014). Using the extended parallel process model to understand texting while 
driving and guide communication campaigns against it. Social marketing quarterly, 
20(1), 66-82. 

Cutello, C. A., Gummerum, M., Hanoch, Y., & Hellier, E. (2021). Evaluating an intervention to 
reduce risky driving behaviors: taking the fear out of virtual reality. Risk analysis, 41(9), 
1662-1673. 

Cismaru, M., & Nimegeers, K. (2017). “Keep your eyes up, don’t text and drive”: a review of 

anti-texting while driving Campaigns’ recommendations. International Review on Public 
and Nonprofit Marketing, 14, 113-135. 

Fisher, J. T., Keene, J. R., Huskey, R., & Weber, R. (2018). The limited capacity model of 

motivated mediated message processing: Taking stock of the past. Annals of the 
International Communication Association, 42(4), 270-290. 

Fisher, J. T., Huskey, R., Keene, J. R., & Weber, R. (2018). The limited capacity model of 
motivated mediated message processing: Looking to the future. Annals of the 
International Communication Association, 42(4), 291-315. 

Fisher, J. T., Hopp, F. R., & Weber, R. (2019). Modality-specific effects of perceptual load in 

multimedia processing. Media and Communication, 7(4), 149-165. 

Fisher, J. T., & Weber, R. (2020). Limited capacity model of motivated mediated message 

processing (LC4MP). International Encyclopedia of Media Psychology. Hoboken, NJ: 
Wiley Blackwell. 

García-Valle, G., Ferre, M., Breñosa, J., & Vargas, D. (2017). Evaluation of presence in virtual 
environments: haptic vest and user’s haptic skills. IEEE Access, 6, 7224-7233. 

Hoang, T., Greuter, S., Taylor, S., Aranda, G., & Mulvany, G. T. (2021, October). An Evaluation 
of Virtual Reality for Fear Arousal Safety Training in the Construction Industry. In 2021 
IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-
Adjunct) (pp. 177-182). IEEE. 

Kreimeier, J., Hammer, S., Friedmann, D., Karg, P., Bühner, C., Bankel, L., & Götzelmann, T. 
(2019, June). Evaluation of different types of haptic feedback influencing the task-based 
presence and performance in virtual reality. In Proceedings of the 12th acm international 
conference on pervasive technologies related to assistive environments (pp. 289-298). 

Lang, A. (2000). The limited capacity model of mediated message processing. Journal of 

communication, 50(1), 46-70. 

Lang, A. (2006). Using the limited capacity model of motivated mediated message processing to 

design effective cancer communication messages. Journal of communication, 56, S57-
S80. 

27 

Lang, A. (2017). Limited capacity model of motivated mediated message processing (LC4MP). 

The international encyclopedia of media effects, 1-9. 

Lee, S., & Kim, G. J. (2008). Effects of haptic feedback, stereoscopy, and image resolution on 
performance and presence in remote navigation. International Journal of Human-
Computer Studies, 66(10), 701-717. 

Lewis, I., Watson, B., Tay, R., & White, K. M. (2007). The role of fear appeals in improving 

driver safety: A review of the effectiveness of fear-arousing (threat) appeals in road 
safety advertising. International Journal of Behavioral Consultation and Therapy, 3(2), 
203. 

Liou, W. K., & Chang, C. Y. (2018, February). Virtual reality classroom applied to science 

education. In 2018 23rd International Scientific-Professional Conference on Information 
Technology (IT) (pp. 1-4). IEEE. 

McDowell, M. (2023, July 14). Valentino to dress meta avatars in digital fashion. Vogue 

Business. https://www.voguebusiness.com/technology/valentino-to-dress-meta-avatars-
in-digital-fashion  

Ngene, A. H. Toward Sustainable Cities and Communities: A Qualitative Study of Motorists’ 

Response to an Anti-drunk Driving Media Campaign in Southeast, Nigeria. 

Popova, L. (2012). The extended parallel process model: illuminating the gaps in research. 

Health Education & Behavior, 39(4), 455-473. 

Rhodes, N. (2017). Fear-appeal messages: Message processing and affective attitudes. 

Communication research, 44(7), 952-975. 

Roskos‐Ewoldsen, D. R., Yu, J. H., & Rhodes, N. (2004). Fear appeal messages affect 

accessibility of attitudes toward the threat and adaptive behaviors. Communication 
Monographs, 71(1), 49-69. 

Shi, J., & Smith, S. W. (2016). The effects of fear appeal message repetition on perceived threat, 
perceived efficacy, and behavioral intention in the extended parallel process model. 
Health communication, 31(3), 275-286. 

Stephenson, N. (1992). Snow Crash. Bantam Books. 

Van Laer, T., De Ruyter, K., Visconti, L. M., & Wetzels, M. (2014). The extended 

transportation-imagery model: A meta-analysis of the antecedents and consequences of 
consumers' narrative transportation. Journal of Consumer research, 40(5), 797-817. 

When did VR become popular? key milestones and breakthroughs uncovered. VR.Space. (2024, 

May 10). https://vr.space/news/equipment/vr-headsets-throughout-history/  

Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process model. 

Communications Monographs, 59(4), 329-349. 

28 

Witte, K. (1996). Fear as motivator, fear as inhibitor: Using the extended parallel process model 
to explain fear appeal successes and failures. In Handbook of communication and 
emotion (pp. 423-450). Academic Press. 

29 

 
 
APPENDIX A: SELF-REPORT FEAR ITEMS 

I found the advertisement to be scary (Carey & Sarma, 2016):  

-3 - Strongly Disagree -> 3 Strongly Agree 

Please rate the extent to which you felt the following emotions while watching the advertisement 
(-3 strongly disagree -> 3 strongly agree; Carey & Sarma, 2016): 

Afraid 
Panicked 
Scared 
Worried 
Nervous 
Tense 

30 

 
 
 
APPENDIX B: PERCEIVED THREAT ITEMS 

Please rate the following beliefs towards safe driving (-3 strongly disagree -> 3 strongly agree; 
Shi & Smith, 2016) 

Car crashes are severely dangerous. 
Getting into a car crash could be fatal for me. 
Car crashes are harmful to everyone involved. 
I am more likely to get into a car crash while speeding. 
I’m at risk for getting in a car crash while I’m on the road. 
Car crashes are a threat whenever I am driving.  

31 

 
 
APPENDIX C: ATTITUDES TOWARDS SAFE DRIVING ITEMS 

Please rate your views on safe driving in the following scales (Roskos-Ewoldsen et al., 2004): 

-3 Negative -> 0 Neutral -> +3 Positive 
-3 Worthless -> 0 Neutral -> +3 Valuable 
-3 Dislike -> 0 Neutral -> +3 Like 
-3 Bad -> 0 Neutral -> +3 Good 
-3 Safe -> 0 Neutral -> +3 Unsafe (Reverse Coded) 
-3 Boring -> 0 Neutral -> +3 Exciting 

32 

 
 
APPENDIX D: BEHAVIORS & INTENTIONS TOWARDS SAFE DRIVING ITEMS 

Please rate the following statements (-3 strongly disagree -> 3 strongly agree; Roskos-Ewoldsen 
et al., 2004) 

In the next two weeks, I will drive the speed limit while on the road. 
I do not intend to drive the speed limit in the near future. 
I plan to encourage my friends or family to drive the speed limit. 
I am now more motivated to be safe and drive the speed limit. 
I will make safe driving decisions a habit of mine. 

33 

 
 
APPENDIX E: MEMORY RECALL ITEMS 

1. What is the name of the professor that narrated the video? 

*A. Ian Johnston 
B. Daniel Smith 
C. Isaac Brahm 
D. Oliver Jones 

2.  What was the difference in speed between the two cars? 

A. 2 km/hr 
*B. 5 km/hr 
C. 8 km/hr 
D. 10 km/hr 

3. What was the color of the two cars shown in the video? 

A. Red/Gold 
B. Blue/Silver 
C. Blue/Black 
*D. Silver/Black 

4. What was the speed of the *black* car initially? 

A. 55 km/hr 
B. 50 km/hr 
C. 70 km/hr 
*D. 60 km/hr 

5. What was in the road ahead of the two cars? 
*A. A semi-truck passing through 
B. Pedestrians crossing the street 
C. A parked car 
D. A tree fell onto the road 

6. In the last 5 meters of breaking, how much speed do you wipe off? 

*A. Half 
B. A quarter 
C. Two thirds 
D. All of your speed 

7.  What was the speed of the *silver* car upon collision? 

A.  0 km/hr 
B. 10 km/hr 
*C. 32 km/hr 
D. 5 km/hr 

8. What was the speed of the *black* car upon collision? 

A.  0 km/hr 
B. 10 km/hr 
C. 8 km/hr 
*D. 5 km/hr 

9. What was the difference in speed between the two cars upon collision? 

*A.  27 km/hr 
B. 15 km/hr 
C. 8 km/hr 

34 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
D. 2 km/hr 

10.  What was the motto shown at the end of the video? 

A. Shave off 10 
B. Don’t speed 
*C. Wipe off 5 
D. Texting can wait 

35 

 
 
 
 
 
 
 
 
APPENDIX F: GENERAL INFORMATION ITEMS 

When did you feel the haptic feedback play? 

At the beginning of the ad 
During the middle of the ad 
At the end of the ad 
I didn't feel any haptic feedback 

How many times have you used virtual reality in total? 

Never (0 times) 
A few times (1-10) 
More than a few times (11-25) 
A handful of times (25-50) 
Many times (50+) 

How many hours have you spent using virtual reality in the past year? 
________________ 

What year are you in school? 

Freshman 
Sophomore 
Junior 
Senior 
Senior+ 
Graduate Student 

What is your gender? 

Cisgender Man 
Cisgender Woman 
Non-binary, Genderqueer, and/or Gender Fluid 
Transgender Man 
Transgender Woman 
Prefer to self describe: _________ 
Prefer not to say 

What is your ethnicity and racial background? 
Asian and/or Asian American 
Black and/or African American 
Hispanic, Latino, Latina, and/or Latinx 
Indigenous/Native American, Alaska Native, and/or First Nations 
Middle Eastern or Northern African 
Native Hawaiian and/or Other Pacific Islander 
White 
Prefer to self-describe: _________ 
Prefer not to say 

36 

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
How old are you? 
________________ 

37