The Loop Trail “Quest”: Use of a Choice-based Digital Simulation, An Interactive Video, and a Booklet to Communicate and Analyze Decision-making of Park Visitors

ABSTRACT

Depreciative park visitor behaviors in protected areas can alter the relationship between humans, wildlife, and ecosystems and provide new opportunities for pathogens to cause illness in people. We created communication interventions that illustrate to park visitors the impact of off-trail walking, dogs off-leash, and littering. We aimed to uncover reasons behind park visitor decision-making and to compare differences in responses between messaging treatments. Our study in Great Falls Park (Potomac, MD) exposed 1207 park visitors to one of three messaging treatments: a choice-based digital simulation, a video, or a booklet. Participants either experienced or observed the role of park visitor behaviors in disease transmission to humans. All participants answered Likert-type questions that gauged their perceptions of the link between humans, animals, and disease. Participants in simulation and video treatments also ranked reasons for non-compliant behaviors. We compared changes in Likert scores by Kruskal-Wallis and Dunn's tests, and compared rank data with Chi square analysis. Key findings suggest that participants selected reasons for non-compliant behaviors that aligned with their “self-interests.” This study applies a social marketing lens to a One Health problem and demonstrates the need for educational interventions that employ game-based learning to convey associations between human and wildlife health.

KEYWORDS:

• Communicating one health
• social marketing
• conservation game-based learning
• emerging technologies
• conservation decision-making

Introduction

Traversing 184.5 miles along the Potomac River from Cumberland, Maryland to Washington, DC, the Chesapeake & Ohio Canal National Historic Park (CHOH) is a national park unit popular for outdoor activities, such as hiking, running, biking, and kayaking. Great Falls Park, a protected parkland within the CHOH with approximately 500,000 annual visitors, provides a living laboratory to explore the interconnectedness of humans with nature, particularly the link between human, wildlife, and ecosystem health.

As confirmed by interviews with park officials, non-compliant behaviors such as littering, off-trail walking, and letting dogs off-leash are significant park management challenges in Great Falls because of their natural resource impacts (Landsman, personal communication, December 4, 2017). These include pollution, damage to vegetation, and disturbance to wildlife (Manning et al., 2017). Not widely researched is the role that these same behaviors can play in altering the relationship between humans, wildlife, and ecosystems, providing new opportunities for pathogens to cause disease in people (Kahn et al., 2012). For each of these common non-compliant behaviors, there is a balance between choices and consequences that may impact the link between human and ecosystem health (Kahn et al., 2012). Going off-trail can disrupt the predator-prey relationship, while staying on the marked trail can protect the predator-prey relationship and thus reduce the park visitor’s risk for disease. Disturbing predators can result in the preponderance of white-footed mice, which are carriers of the bacterium that causes Lyme disease, potentially increasing exposure to humans (Miller et al., 2001; Muhly et al., 2011). Letting dogs off-leash may increase risk of human encounters with ticks and potentially exposure to tick-borne diseases (Jones et al., 2018). Littering provides an opportunity for water to collect in trash. Some species, such as the Asian Tiger Mosquito, breed in small container habitats (cups, bottles, cans) (Bartlett-Healy et al., 2012). These mosquitoes can carry the diseases that cause Zika or dengue (Pereira Horta et al., 2013). Cleaning up trash can reduce mosquitoes and, therefore, disease risk (Little et al., 2017).

Encouraging citizens to take personal responsibility for care of the environment are the key principles of education interventions, such as Leave No Trace (LNT). These principles have become prominent education intervention approaches in US protected areas. Posted signage and web-based messages encourage “correct behavior” and discourage depreciative behavior (Harmon, 1997; Marion & Reid, 2001, 2007; Vagias & Powell, 2010). Messages as they relate to the three common compliance issues—off-trail walking, litter, and dogs off-leash – include “travel on durable surfaces,” “dispose of waste properly,” “respect wildlife,” and “be considerate of other visitors” (Marion & Reid, 2007). While these messages are designed to instill an ethic of conservation stewardship among visitors (Marion & Reid, 2007), their effectiveness on behaviors has yet to be determined.

In an extensive literature review of non-compliant behaviors, Dietsch et al. (2016) categorize certain behaviors as unavoidable, uninformed, unskilled, careless, and deliberate/willful. Information and education approaches such as the LNT interventions are grounded in the assumptions that if people only knew of the damage their actions caused (informed), cared a little more, were concerned about what others thought of their actions (appropriate behavior), or were motivated by a sense of moral responsibility in protecting wildlife and ecosystems, they would change their behavior. The role of common non-compliant behaviors in disease dynamics and the opportunity to shift compliance messaging and interventions to focus on protecting human health are not widely studied (Decker et al., 2010).

Innovative approaches are needed to communicate the potential for each of these “non-compliant” management behaviors to impact park visitors, wildlife, and ecosystem health through a “causal loop of self-inflicted harm” (Lovejoy, 2012, p. xi). There is an opportunity to apply principles of game-based learning (GBL) through simulations and videos, to visually communicate the link between a healthy ecosystem and healthy people in order to foster park visitor behavior compliance. The NPS Advisory Board Education Committee advocates deepening and extending “in-park technology experiences that add value to the interactive physical experience” (NPS, 2012, p. 2). Listed among top priorities, the committee recommends “develop[ing] easy-to-use [technology-based] tools to support park-based learning” (NPS, 2012, p. 10).

Learning through technology has seen an increase in both industry and academia (Plass et al., 2015). Games, game-based learning and simulation environments offer opportunities to break the “tell and test” instructional paradigm prevalent in education today (FAS, 2006; Prensky, 2007) and improve the motivation to learn, enhance the outcomes from learning experiences, and transfer what is learned to practical application. GBL facilitates “stealth learning,” where players do not realize they are learning embedded content because they are enjoying the experience of “playing” (Annetta, 2010).

Application of technology-based learning in conservation

The motivational power of games has been applied to conservation education. In the first study to examine the effects of games on conservation teaching, Tan et al. (2018) reviewed seven hands-on activities and role-play games, including topics on conservation biology, conservation genetics, and human wildlife conflict. This study concluded that conservation games can play an important role in making conservation more immersive. While these studies discuss the potential of games and gamification¹ to foster pro-environmental behavior and support of biodiversity conservation (Fletcher, 2016), as with “traditional” messaging (e.g. trail signage, bulletin boards), the focus is on “park” or “conservation” self-interests (Rothschild, 1999) as opposed to self-interests of the park visitor.

Abrams et al. (2020) recommend that “parks use messages emphasizing the visitor experience gained by engaging in the desired behavior rather than messages that only highlight the importance of wildlife protection” (p. 255). The presence of non-compliant behaviors is widely documented. However, the reasons why visitors engage in these behaviors are not widely studied, with the exception of convenience and avoidance to explain off-trail behavior (Wimpey & Marion, 2011), habituated behavior to explain dogs off-leash (Bowes et al., 2018), or personal morals and social norms related to littering (Hughes et al., 2009). Elements of GBL, such as digital simulations, may provide an opportunity to identify park visitor beliefs and salient themes that resonate (Marion & Reid, 2007) with park visitors and more specifically discover reasons why park visitors behave the way they do related to their own “self-interest”. These approaches and technologies can be viewed through a lens of social marketing in order to assess beliefs and themes. Social marketing is the application of marketing concepts and techniques to create, communicate, and deliver value to influence behavior and benefit the target audience and society (Lee & Kotler, 2011). “Value” is assessed similar to the manner in which “commercial exchanges” are weighed: there are “good deals” and there are “bad deals” (Maibach, 2002). Effective social marketing approaches demonstrate to the targeted audience why implementing the target behavior would align with their self-interest (Rothschild, 1999).

Application of simulations in non-compliant behaviors

Simulations are a type of GBL where participants control and “see” a graphic representation of their movements and the consequences of their actions (Klopfer et al., 2004). We created communication interventions that illustrate the hypothesis that off-trail walking, dogs off-leash, and littering initiate the “causal loop of self-inflicted harm” (Lovejoy, 2012, p. xi). The specific purpose of this study was threefold: to (1) uncover using three different media treatments why park visitors make the decisions that they make related to common park compliance issues, (2) examine changes in perception of disease causation when presented with consequences and benefits of their choices, and (3) compare differences between digital simulations, videos, and booklets. We applied Prensky’s (2007) principles of learner engagement as a theoretical framework in designing both the simulation and the video (Figure 1). Elements of the framework include goal orientation such as a player’s desire to be perceived as good stewards of the environment; in other words, characteristics of a “hero” (Prensky, 2007). Decisions, choices in confronting specific problems, are critical to reaching the goals. Decisions, according to Prensky, are at the heart of the “learning loop” of decision-action-feedback-reflection. Decisions in a park setting include staying on-trail or going off-trail; keeping dogs on-leash or letting dogs off-leash; and picking up trash or leaving trash behind. “Good deals and bad deals” (Maibach, 2002) related to park behaviors can be presented in the form of opposing choices, or competition. Simulations can demonstrate the “learning loop” (Prensky, 2007) by illustrating the decision (e.g. go off-trail) and the action that follows (disruption of predator-prey relationships), and then offering feedback on the consequences of park visitors’ chosen action (increased risk for transmission of Lyme disease) (Ostfeld & Holt, 2004). The simulation can then be personalized with the learner’s input followed by an opportunity for reflection and a reconsideration of the learner’s goal orientation (e.g. heroic role, villain, or victim). Finally, an emotional connection can be created through animation, sound, and motion.

Figure 1. Loop Trail Quest experimental design model. The figure shows how each treatment incorporates elements of Prensky’s framework. The shaded boxes represent additional steps that we added to reflect social marketing principles to reveal value-based “reasons why” visitors do what they do in order of priority, and to assess any changes in “ethic” from their original goal orientation.

Videos, while visually engaging through sight and sound, do not typically enable the requisite elements of engagement described by Prensky (2007), including setting goals, making decisions, handling conflict, and managing competing choices, nor do videos traditionally provide personalized feedback. [The experiment described herein elicited feedback from the learner after viewing the video.] Similarly, more traditional education materials such as booklets are limited by design in their ability to do any of the above, with the exception of illustrating through words and photography, the cause and effect of behaviors.

Experimental design methods: the “Loop Trail Quest”

Study site

The location of this study was Great Falls Park, situated along the Chesapeake & Ohio (C&O) Canal in Maryland (Lat. 38.99, Long. −77.25) (hereinafter referred to as “the park”). Located approximately 15 miles from downtown Washington, DC, the park is a suburban yet urban-proximate park (NPS, n.d.). Study locations include (1) the Great Falls visitor center near restrooms, (2) the entrance/exit to the Billy Goat Trail Section A (popular hiking trail), and (3) the entrance to Olmstead Lookout (scenic view of the falls) (Figures 2 and 3).

Figure 2. Study Sites of Great Falls, Maryland, USA.

Note: Stars indicate study points. NPS, n.d.

Figure 3. View of Great Falls from Olmstead lookout, Maryland, USA. 2019.

Study design

Stakeholder engagement and development of research question

As a first step in the design of this experiment, we conducted stakeholder interviews at the CHOH Headquarters in Hagerstown, Maryland with NPS officials to understand non-compliant behaviors. Senior CHOH officials from divisions including interpretation, safety, conservation biology, and resource management validated the need for behavior change-focused research and shared common park visitor infractions including littering, walking/hiking off-trail, and letting dogs off-leash (Landsman, personal communication, December 4, 2017; Adams, personal communication, December 4, 2017).

Media treatments: the “Loop Trail Quest” interventions

We designed an experimental study using three treatments: a digital simulation, a video, and a booklet. For all three interventions pre- and post-test assessments were conducted. The interactive digital simulation enabled respondents to choose and experience the outcome of non-compliant behaviors in an immersive digital setting and experience the impact those decisions have on wildlife health, ecosystem health, and potentially their own health. Alternatively, players could also experience the benefits that compliant behaviors have on ecosystem health and therefore their and their community’s health. For the video, characters were shown engaging in non-compliant behaviors and the consequences of those behaviors. For both the video and the simulation, respondents were asked to rank their reasons for compliant or non-compliant behaviors. The booklet was a didactic presentation of words and photographs with similar content to that of the video and digital simulation.

Design of educational interventions

All interventions were designed to be completed within 6–8 min, including the consent process and pre- and post- assessment. The video and digital simulations were hosted on a Kindle 8 tablet. Ranking of reasons, or projected ranking of reasons, was only enabled in the video and simulation.

Study design: simulation flow

The simulation flow diagram is shown in Figure 4. The simulation gameplay “trailer” is available through this link:

Figure 4. Simulation flow and decision points for Loop Trail Quest simulation.

   https://www.youtube.com/watch?v=pE2OdMeWqaY&feature=youtu.be

Each scenario presented a “relatable conflict” (FAS, 2006). For instance, in the case of making a choice between staying on the marked trail or going off, the narrative included a “conflict” such as the trail being crowded. In the case of litter, the visitor had inadvertently left the bottle and cap and only realized it after leaving the area, and retrieving it required them to turn back. In the case of advice to a friend, the conflict is that they may feel uncomfortable telling a friend to keep their dog on leash. Without a conflict, players are likely to choose the socially desirable choice (a response that is seen as favorable) (Grimm, 2010). At the end of each scenario respondents ranked the reasons for each of their decisions (e.g. convenience, lack of awareness, others did the behavior, adventure, lack of concern for their health or safety or ecosystem health, etc.). This list of choices was based on a combination of a review of literature as well as stakeholder interviews.

After making the choices for the simulation character, respondents then received feedback on the impact that their choices may have: going off-trail may disrupt predator-prey relationships, leaving mice, reservoirs for Lyme disease to reproduce; leaving behind a bottle and bottle cap may create a breeding ground for mosquitoes (also vectors of disease including dengue and Zika); and letting a dog off-leash may provide an opportunity for vector-borne disease transmission to humans via an infected tick carried on the dog’s fur.

All respondents for each of the three treatments were asked post-treatment exposure questions to see any changes in their regard for disease causation and their One Health ethic as a result of experiencing, seeing, or reading about these associations.

Booklet flow

The booklet was developed in the form of a “flip-book” with “tabs” and headings of “if, then” that showed static photographs and graphic depictions with short narrative descriptions describing what might happen “if a park visitor goes off trail,” “if a park visitor litters,” and “if a park visitor lets their dog off leash.” This didactic approach focused solely on knowledge transfer and was designed to serve as the non-interactive information and communication intervention. The supplementary material provides the booklet pages.

Video flow

The video used the same visuals as the interactive simulation and presented each scenario with the negative outcomes only; showing a hiker going off-trail, leaving litter behind, and advising their friend to let their dog off-leash. Players interacted with the video by providing feedback as to why the character made the decisions they observed. The video is available through the following link:

https://www.youtube.com/watch?v=Q3ZNsrhQMig&feature=youtu.be

Participant selection and research protocol

A visitor intercept approach (Figure 5) was used to select participants over the summer months of June, July, and August 2019 who were exposed to one of three treatments: a video, an interactive simulation, and a booklet. Each researcher alternated the order of the treatments. For instance, if the researcher started with the booklet, the next visitor would receive the video, and the following visitor would receive the interactive simulation. The criteria for participation in the research were that the participants must be at least 18 years of age, are recreational visitors to the C&O Canal, and are not employees or volunteers of the NPS. Demographic – age, gender, education, residence – as well as recreational use data were collected, including reason for use of the park, frequency of visiting the park, amount of time spent in the park, distance of residence in relation to the park, and whether or not the visitor paid to enter the park. Our rationale for assessing recreational, park use, and residence data was to aid the NPS in designing interventions that target select demographics. Segmenting interventions by age and gender may lack specificity; tailoring messages to groups, such as local residents or hikers, may allow the NPS to develop more effective educational messaging.

Figure 5. A George Mason University researcher engaging participants. Photo: Susan Howard, 2019.

Participants were first asked to consent to participation. The consent form was digitally embedded within the electronic tablet. The researcher followed approved guidelines for recruitment, privacy and confidentiality, risks, benefits, advertising, etc. per IRB guidelines (#1175240-2, George Mason University), as well as stipulations per the NPS permit (#CHOH-2018-SCI-0001). Participants also answered pre-test questions that asked them to rank their evaluation of statements using a 1–5 Likert scale (strongly disagree, disagree, neutral regard/no opinion, agree, strongly agree). These statements were designed to gauge respondents’ “victim, villain, hero” mentality regarding disease causation. The terms “victim, villain, hero” were not used in the pre- and post-test, but rather describe the mentality that we aimed to assess (Figure 6).

Figure 6. Likert item questions from pre- and post-tests during Loop Trail Quest study.

Participants then engaged with their assigned treatment: interactive simulation, the video, or the booklet. Respondents for the interactive simulation were asked to make choices within the simulation and then asked to rank the reasons for each of their choices/behaviors. Ranking, or choice ordering, is a technique used in market research that enables researchers to evaluate factors and attributes to better understand reasons for choices/behaviors (McFadden, 1986). Viewers of the video were asked to rank why the characters shown in the animation might have made the choices they made for each non-compliant behavior. Readers of the booklet were not presented with rank options as there was no decision-making. At the conclusion of each treatment, participants then answered the same post-test questions and answered questions relating to demographics and use of the park.

Statistical analysis

We performed statistical analysis in R 3.6.2 (R Core Team, 2013). We used an analysis of variance (ANOVA) to compare pre-test responses among the intervention types. We also conducted Chi square tests and post-hoc tests to test the distribution of demographic variables among the treatment groups. We used a Kruskal–Wallis test to compare differences in distributions of pre-test Likert scores between demographic groups and participants in various test treatments. When the Kruskal–Wallis test yielded a significant result (p < 0.05), we then used a Dunn’s test to further analyze differences between levels of each category, using a Benjamini & Hochberg p-adjustment (significance at p < 0.05) (Dinno, 2015; Ogle et al., 2020).

We also used ordinal logistic regression to determine the influence of demographic variables on change in Likert score. We included all demographic and park use variables, including the intervention type (video, simulation, or booklet), as factors in the model. We ordered variables for “Frequency of visit,” “Education,” and “Time Spent in the Park,” as well as the outcome variable of change in Likert score, to indicate increasing value. If an answer for these ranked categories included “I prefer not to answer,” we removed that response from analysis to avoid confounding the ranked data. We used a stepwise regression to determine the model with the lowest Akaike information criterion (AIC) value. We then used Variance Inflation Factor (VIF) values to select the best fitting model, with values over “4” being eliminated.

To test differences in ranking, data was subset into Video and Simulation treatments (Booklet treatments did not contain a ranking task). We then created a new category, tallying if participants selected each rank as their first, second, or third choice. We then used Chi square analysis to compare each of these values by each demographic variable. Results were considered significant at p = 0.01 to reduce Type I Error. For the simulation treatment, we conducted Chi square analysis to determine if there was a significant association between each demographic variable and the option selected for each rank during simulation play (p < 0.01).

Results

Participant demographics

A total of 1207 participants were recruited through intercept interviews. A total of 409 park visitors were exposed to the video, 402 to the simulation, and 396 to the booklet. Given that the boundaries of Great Falls border the neighborhood of Potomac, Maryland and are easily accessible by foot or bike from Bethesda (two miles) and Cabin John (five miles), respondents were asked to designate if they lived in these neighborhoods specifically. A total of 257 (21%) respondents were from these neighborhoods. In terms of reasons for use of the park, 508 (42%) visited the park to hike one of the park’s 18 formal trails. Fifty percent of respondents were male and 49% female (1% selected “no answer”). Fifty-three percent had a Master’s degree or higher. An ANOVA comparing pre-test score by intervention type revealed no significant difference in score among the treatment groups (p > 0.1) for all four statements. A Chi square post-hoc test revealed little difference between demographic variables represented among treatment groups. The only significant differences revealed that more participants in the Simulation treatment reported “Option 1: 18–24 years old” for “Age Group” (p = 0.047), and “Option 3: Some college, no degree” for “Education” (p = 0.03).

Respondents’ conservation orientation: Likert data

Table 1 shows the median Likert scores for the pre- and post-tests for each educational intervention. For Likert #1 (Wildlife causes disease) the median score changed from a 2 to a 3, from “disagreeing” that wildlife causes disease to “neutral” after reviewing the booklet. For this same question, the median score for respondents exposed to the video and simulation treatments changed from a 2 to a 4, from “disagreeing” that wildlife causes disease to “agreeing” with this statement. A majority of respondents (50–60%, depending on the treatment) did not change their pre- and post-test score for Likert #3 (Humans cause disease). For Likert #3 (Humans cause disease) 80% stated they “agree” or “strongly agree” in the pre-test, increasing to 93% in the post-tests.

Table 1. Median scores for Likert test in Booklet, Video, and Simulation treatments.

Median Likert Score by Item and Treatment
	Booklet		Video		Simulation
	Pre-Test	Post-Test	Pre-Test	Post-Test	Pre-Test	Post-Test
Likert #1 Wildlife causes disease	2	3	2	4	2	4
Likert #2 No connection	2	2	2	2	2	2
Likert #3 Humans cause disease	4	4.5	4	5	4	4
Likert #4 Wildlife plays protective role	5	5	5	5	5	5

There were significant associations between intervention types and changes in pre–post score for Likert #1 (Wildlife causes disease, p = 0.003) and Likert #2 (No connection between human-wildlife disease, p = 0.02), revealed by the Kruskal–Wallis Test. After a post-hoc Dunn’s test, the only significant difference existed between Intervention type and Likert #1 and #2 (Table 2). For Likert #1, there was a significant difference in distribution of change in response between the Booklet and Simulation Game (p = 0.02) and between the Booklet and Video (0.003). For Likert #2, there was a significant change between the Booklet and Video (p = 0.04) and between the Simulation Game and Video (p = 0.04) (Table 2). Histograms (Figure 7) reveal the difference in distribution indicated by the Dunn’s test. Approximately 60% of participants in Simulation Game and Video treatments exhibited positive changes (towards “agree”), compared to approximately 45% of Booklet participants.

Figure 7. Histograms showing the distribution of changes in score from Likert pre-post-tests.

Table 2. Significant results of Dunn’s test comparing demographics and test mode by change in Likert score.

	Comparison	z-score	p-value
Likert 1: Wildlife cause disease	Booklet-Simulation	−2.4	0.02 *
	Booklet-Video	−3.26	0.003*
	Simulation-Video	−0.8	0.4
Likert 2: No connection between human and wildlife health	Booklet-Simulation	0.2	0.8
	Booklet-Video	−2.2	0.04*
	Simulation-Video	−2.4	0.04*

Note: *Indicates significance <0.05, with Benjamini Hochberg p-adjustment.

Table 3 shows the odds ratio, confidence interval, and p-value of each variable that best predicts the change in pre–post Likert score. Educational intervention was the only variable that was present in each model. The stepwise AIC did not remove any variables for Likert item #3.

Table 3: Best fitting models produced by Ordinal Logistic Regression and Stepwise Regression showing variables that predict the outcome variable of “change in Likert score”

	Variables	Odds Ratio	Confidence Interval	p-value
Likert #1: Wildlife causes disease	Intervention: Simulation	1.33	1.12–1.90	0.03
	Intervention: Video	1.55	1.23–2.07	0.0008
	Gender (Male)	0.83	1.02–1.18	0.09
Likert #2: No connection between wildlife and human health	Intervention: Simulation	0.96	0.74–1.25	0.8
	Intervention: Video	1.37	1.04–1.77	0.02
Likert #4: Wildlife plays a protective role	Simulation	0.8	0.56–1.05	0.09
	Video	1.1	0.77–1.44	0.7

Choice-ordering of reasons for park visitor behaviors

Video treatment

The majority of all of the respondents who watched the video (n = 409) chose either adventure (258, 63%), because they observed others doing the behavior (227, 56%), or to bypass crowds (199, 49%), to explain the primary reasons why the hiker in the video went off-trail. The top three reasons given for explaining why the character in the video did not pick up litter were as follows: the majority gave the reason that a park volunteer would pick it up (251, 61%); there is no harm to them personally (220, 54%); or because others had left litter behind (198, 48%). The “dog will come when called” was given as the primary reason to explain why the character in the video advised the friend to let her dog off-leash (265, 66%). “No harm to their health or safety” was reported as the second reason (233, 57%); and “they see others do it” was the third most reported reason (203, 50%) (Table 4).

Table 4. Most commonly ranked options for Video treatment. Options ranked as 1st, 2nd, or 3rd choice.

Rank option: Video (n = 409)	Rank Option	Frequency ranked 1st	Frequency ranked 2^nd	Frequency ranked 3^rd	Total
Going off-trail	Adventure	158 (39%)	66 (16%)	34 (8%)	258 (63%)
	Bypass crowds	67 (16%)	82 (20%)	50 (12%)	199 (49%)
	Others did it	60 (15%)	84 (21%)	83 (20%)	227 (56%)
Littering	Volunteer will pick up	141 (34%)	67 (16%)	43 (11%)	251 (61%)
	Others left litter	47 (11%)	82 (20%)	69 (17%)	198 (48%)
	No harm to them	86 (21%)	77 (19%)	57 (14%)	220 (54%)
Dogs off-leash	Other dogs are off leash	79 (19%)	72 (18%)	55 (13%)	206 (50%)
	Dog comes when called	115 (28%)	75 (18%)	75 (18%)	265 (66%)
	No harm to their health or safety	84 (21%)	83 (20%)	66 (16%)	233 (57%)

People who lived in the neighborhood were less likely to state that the character went off-trail because they saw others do it as a “top three” reason, compared to visitors who lived outside of the neighborhood (RR = 0.7, p-value  < 0.01, CI = 0.54–0.92). Respondents who were male were more likely to list “because others did it” as a reason for littering when compared to females (RR = 1.36, p < 0.01, CI = 1.11–1.67). Park visitors who paid admission were more likely to state that litter poses no harm to personal health or safety as a “top three” reason for why the character in the video might litter, compared to those who did not pay to enter (RR = 1.37, CI = 1.15–1.63, p < 0.001). Park visitors with a park pass were less likely to state that litter poses no harm to personal health or safety (RR = 0.7, CI = 0.58–0.86, p < 0.001) (Table 5).

Table 5. Results of Chi Square test: demographics associated with choice-ordering for Video treatment.

Behavior Choice: Rank Option (Video)	Demographic variable	Count (%) reporting the option	p-value	Confidence Interval	Risk Ratio
Went off trail: Others did it	Don’t live in neighborhood	191 (59%)	–	–	–
	Live in neighborhood	36 (42%)	0.006	0.54–0.92	0.7
Littered: Others did it	Female	85 (41%)	–	–	–
	Male	113 (56%)	0.003	1.11–1.67	1.36
Littered: No harm to personal health/safety	Did not pay admission	118 (47%)	–	–	–
	Paid admission	102 (65%)	0.0008	1.15–1.63	1.37
Littered: No harm to personal health/safety	No park pass	141 (62%)	–	–	–
	Park pass	79 (43%)	0.0004	0.58–0.86	0.7
Littered: No harm to personal health/safety	Other reason for visit	198 (54%)	–	–	–
	Visited to walk path	35 (77%)	0.005	1.19–1.72	1.43

Simulation treatment

Table 6 shows the options most commonly ranked as 1st, 2^nd, or 3^rd by participants in order of perceived importance. Of the 402 respondents that played the interactive simulation, 196 (49%) chose to go off-trail; and 206 (51%) chose to stay on-trail. Of the 196 that chose to go off-trail, “adventure/exploration” was a top three reason given among 114 respondents (58%), followed by “others have gone off trail” for 79 respondents (40%), and “bypass crowds” for 69 respondents (35%).

Table 6. Most commonly ranked options for Simulation treatment. Options ranked as 1st, 2nd, or 3rd choice.

	Rank option	Frequency ranked 1st	Frequency ranked 2nd	Frequency ranked 3rd	Total
Simulation Rank Choice: Went off trail (n = 196)	Adventure/Explore	81 (41%)	24 (12%)	9 (5%)	114 (58%)
	Bypass crowds	19 (10%)	31 (16%)	19 (10%)	69 (35%)
	Others have taken it – must be hikeable	18 (9%)	36 (18%)	25 (13%)	79 (40%)
Simulation Rank Choice: Stayed on Trail (n = 206)	Prevent from getting lost	68 (33%)	37 (18%)	37 (18%)	142 (69%)
	Health/safety risks to me or my group	24 (12%)	50 (24%)	27 (13%)	101 (49%)
	Harm to the environment	73 (35%)	24 (11%)	38 (18%)	135 (66%)
Simulation Rank Choice: Left litter (n = 19)	Volunteer will pick it up	7 (37%)	3 (16%)	1 (5%)	11 (58%)
	Didn’t know “trash free” meant to carry trash	1 (5%)	4 (21%)	3 (16%)	8 (42%)
	None of these apply	5 (26%)	0	1 (5%)	6 (32%)
Simulation Rank Choice: Picked up litter (n = 383)	Know we have to carry trash	149 (39%)	93 (24%)	53 (14%)	295 (77%)
	It isn’t the right thing, even if others do it	48 (13%)	119 (31%)	84 (22%)	251 (66%)
	Harm to natural environment	162 (42%)	71 (19%)	54 (14%)	287 (75%)
Simulation Rank Choice: Let dog off leash  (n = 33)	Others let dogs off leash	7 (21%)	5 (15%)	3 (9%)	15 (45%)
	Dog returns when called	13 (39%)	4 (12%)	5 (15%)	22 (67%)
	There is no harm to us or dog	3 (9%)	6 (18%)	6 (18%)	15 (45%)
Simulation Rank Choice: Kept dog on leash (n = 369)	Signs & volunteers	104 (28%)	44 (12%)	53 (14%)	201 (54%)
	Wildlife encounter – safety concern for dog	112 (30%)	96 (26%)	47 (13%)	255 (69%)
	Wildlife encounter – safety concern for us/families	72 (20%)	93 (25%)	56 (15%)	221 (60%)

Of the 206 that chose to stay on the trail, 142 (69%) chose the reason “to prevent from getting lost” as one of their top three rankings, followed by “harm to the environment” (135, 66%) and “concern for health and safety risks to self” (101, 49%). Only 21 (10%) of the 206 who chose to stay on-trail chose “reprimand by a park ranger” as their top three reasons for compliance. Of the 196 who went off-trail, 52 (27%) ranked “no apparent risk to their health or safety” as their first, second, or third choice, and 54 (28%) chose “no apparent harm to the environment” as their first, second, or third choice.

Most participants (383, 95%) chose to pick up the bottle. The following choices were listed as a “top three” reason for picking up the bottle: “knowing that they have to carry trash out” (295, 77%); “harm to the natural environment” (287, 75%); and “not the right thing, even if others have done it” (251, 66%). Only 105 (27%) chose as their top three rankings “harm to my health or safety.”

Of the 369 respondents who advised their friend to keep their dog on-leash, 255 (69%) ranked harm to the dog’s safety as a result of a wildlife encounter, 221 (60%) ranked concern to humans as a result of a wildlife encounter, and 201 (54%) ranked signs and reprimand from park officials/volunteers as one of their top three reasons to advise their friend to keep their dog on-leash.

Table 7 shows that males were more likely to rank adventure as a “top three” reason for going off-trail (RR = 1.48, CI = 1.17–1.89, p = 0.001). People who visited the park at least once a week were less likely to rank concern for the environment in choosing to pick up the bottle and cap (RR = 0.57, CI = 0.37–0.89, p < 0.001). Park visitors who do not live in the neighborhood are more likely to keep their dog on-leash due to concern for the dog’s health and safety. Similarly, visitors who are not a regular visitor (less than monthly) are more likely to state their dog’s health and safety as a reason to keep their dog on a leash. People who do not live in the neighborhood are more likely to state “concern for their own health and safety” for keeping their dog on-leash.

Table 7. Results of Chi Square test: demographics associated with choice-ordering for Simulation treatment.

Behavior Choice: Rank Option	Demographic variable	Count (%) reporting the option	p-value	Confidence Interval	Risk Ratio
Went off trail: Adventure	Female	47 (48%)	–	–	–
	Male	69 (72%)	0.001	1.17–1.89	1.48
Went off trail: Adventure	No Park Pass	79 (69%)	–	–	–
	Park Pass	38 (47%)	0.004	0.52–0.89	0.68
Picked up bottle: Concern for environment	Frequency: Other	276 (77%)	–	–	–
	Frequency: Once a Week	11 (44%)	0.0006	0.37–0.89	0.57
Kept dog on leash: Concern for dog’s health/safety	Does not live in neighborhood	220 (72%)	–	–	–
	Lives in neighborhood	36 (55%)	0.01	0.61–0.96	0.77
Kept dog on leash: Concern for dog’s health/safety	Not a regular visitor	204 (73%)	–	–	–
	Regular visitor (few times/month)	50 (57%)	0.005	0.63–0.94	0.77
Kept dog on leash: Concern for personal health/ safety	Does not live in neighborhood	193 (63%)	–	–	–
	Lives in neighborhood	29 (45%)	0.007	0.53–0.93	0.7

Discussion

Drivers of park visitor decisions

This study revealed that park visitors act out of “self-interest” when either complying or not complying, accepting “good deals” and rejecting “bad deals” similar to commercial exchanges as described by Rothschild (1999). It is the nature of park management to advance organizational interests, yet often the behaviors that the park managers promote are not perceived by the individual to be of self-interest. We identified an opportunity for park management to promote compliance with park rules and regulations by shifting away from “organizational self-interests” to “visitor self-interests.” In the case of off-trail, dogs on-leash, and litter behaviors, the visitor chose reasons to comply or not based to some extent on “self-interest.” Choice-ordering responses were similar for the Simulation Game and Video treatments, which gives further credibility to this projective technique.

In the case of off trail behaviors, “adventure” and “bypassing” crowds were selected as the top two reasons for the behavior in both Simulation Game and Video treatments, yet reasons to stay on the trail included to “keep from getting lost,” “concern for health and safety risks,” as well as “concern for the environment.” The primary reasons were aligned with visitor self-interest. In the case of going off-trail, “adventure” and “bypassing crowds” outweighed concern for the environment. Similarly, “keeping from getting lost” and “health and safety” risks outweighed the inconvenience of “crowds or mud.” Advising a friend to keep a dog on-leash was chosen for a large majority of players of the simulation. The top two reasons given were for “safety concerns of the dog in regard to wildlife encounters” and “safety concerns” for self. In each of these examples, self-interest was a primary consideration.

This study further corroborates the “commercial exchanges” idea of behaviors embraced by social marketing (Rothschild, 1999); if “adventure” is perceived as the primary self-interest of the park visitor, then staying on the trail to protect wildlife or ecosystem health is not as good a “deal.” However, keeping from getting lost was a better deal for some than protecting the environment for those who chose to stay on-trail. Interestingly, those who chose to stay on the trail were concerned not only for the health of the environment, but for their own health. Those who chose to keep their dog on-leash chose safety of the dog or self over the trade-off of letting the dog run free and roll around on the grass. Self-interest considerations may have guided respondents who overwhelmingly chose not to litter, as the reason given was because they “know to carry out their trash.” The self-interest benefit may be a favorable perception by others (i.e. social norms).

Changes in perception of disease causation

We saw respondents regard humans as “villains” in disease causation in all media treatments, with little median change in perception from pre- to post-test assessments for Likert #3 (Humans cause disease). However, we saw a shift in “blame” for disease causation toward wildlife in the interactive media (video and simulation). In the video and simulation, more participants had a positive change in score toward “agree” as it relates to wildlife as causing disease, compared to the booklet. While there was a shift in “blame” toward wildlife as “causing disease,” respondents’ regard for wildlife’s “protective” or heroic role, as stated above, remained overwhelmingly constant.

The shift in perception toward wildlife as a cause of disease was more prominent in Video and Simulation treatments than for the Booklet. This result may be influenced by animation in both the video and simulation to illustrate things “unseen” that result in human actions, such as mosquitoes (wildlife) breeding in standing water in a bottle or bottle cap as a result of litter, and the preponderance of white-footed mice (can be reservoirs of disease) as a result of hiking off-trail and disrupting predators (Levi et al., 2012; Muhly et al., 2011). These results corroborate previous studies showing that GBL elements promote engagement and facilitate learning (Prensky, 2007). There was less of a shift in perceptions of humans as causing disease because this was already high from the pre-test.

For 684 out of 1206 (57%) respondents, there was a positive change from pre- to post-test (towards “agree”) in the perception of “wildlife as cause of disease” (Likert 1). However, while “blame” tipped toward wildlife as “villain,” the perceptions of “humans as villain” in disease causation (Likert 3) and “wildlife as heroic” (Likert 4) remained constant. Moreover, perceptions increased with regard to the connection between wildlife health and ecosystem health. This is encouraging in that Decker and colleagues (2010) caution “blaming” wildlife for disease emergence. However, in this study, we see a recognition of the connection between wildlife health and human health; we see an increase in the perception that wildlife is in fact a reservoir of disease, balanced by the perception that humans can play a role in triggers of disease through their behaviors. Overall, the study findings demonstrated that the link between wildlife health and human health (central to One Health principles) can be created while still maintaining their regard for the “heroic” or protective role of wildlife.

Differences between treatments: digital simulation, video, and booklet

We have some evidence that simulation and video are more effective in shifting perception than static media. The only variable that was significantly associated with change in Likert score was the “intervention type” per the ordinal logistic regression. For the Likert #1, the greatest change in wildlife as a “cause” of disease was seen in the video and simulation compared to the booklet. Yet there is little difference between the video and simulation. The impact of both the video and the simulation compared to the booklet can be attributed to the game-based elements applied to both media. Principles of engagement were employed in both media, including goal orientation, emotional connection, feedback, and reflection. The simulation was the only treatment that allowed decision-making by the player based on competing choices. However, the video, using the same visuals from the simulation, showed the character doing all the non-compliant behaviors with the ensuing consequences and then asked the viewer to reflect and explain why the character may have made the decisions that they observed. Even though the viewers did not make decisions, they were engaged in reflecting and rationalizing (through the ranking exercise) and were provided feedback.

This study addresses the need to develop a technology-based learning strategy, as recommended by the NPS Advisory Board (2012). The report states that “technology can be an important path to connecting with the natural resources of the national parks” (NPS, 2012). Our findings support this recommendation, in that we observed the appeal of novel and interactive uses of technology as a result of shifts in knowledge and perceptions among participants of the video and digital simulation.

The findings of this study potentially demonstrate how GBL may foster learner engagement over more traditional forms of learning as a result of applying a game-based framework (Prensky, 2007). The Likert assessment provided “goal orientation” based on desires and values in relation to disease dynamics (e.g. villain, victim, hero, or no regard). Decisions were designed to give learners choices in confronting or resolving a specific problem. Competition was presented in the form of choices that served either the interests of the “organization” (natural resource protection) or “self-interests” (health and safety). Further, our research conveys the importance of interactivity in messaging. We suggest that GBL could be integrated into existing campaigns, such as LNT. These interventions could be implemented in a visitor center kiosk through tablets or other touch screen displays. Also, interventions could be integrated into a website under a “Plan Your Visit” section, allowing visitors to interact with this material before and after the park experience. The ability to identify reasons for why park visitors make the decisions that they do – to comply or not comply with park rules – is lacking in the literature. We designed the simulation intervention to enable learners to reveal motivators and drivers of their choices and behaviors.

Limitations

We recognize that the intercept methodology may bias the study results, as respondents who agreed to participate may differ from those who refuse.

Two of the questions were “socially biased” toward answering, as the respondent would be expected to state that they “Agree” (Likert #3 “humans cause disease” and Likert #4 “wildlife serves in a protective role”). Regardless, interesting results emerged from the first two questions regarding the role of wildlife in disease causation and wildlife serving a protective role – where the “answer” is not clear.

The education level of the sample size was very high, with over half of the respondents having master’s degrees or higher. This region, home to the National Institutes of Health and Walter Reed National Medical Center, has a highly educated demographic with high income. The location of the study likely contributed to the demographics of the sample population.

The study design did not include a “no message” control, where park visitors could rank potential reasons for non-compliant behaviors without observing an educational treatment. This study cannot assess the impact of message content outside of its treatment group.

For future studies, we suggest that similar research could be conducted on different sections of the C&O canal path, which could yield different findings of visitor perceptions based on park use, regional, or recreational characteristics. These demographics of interest could be modified to fit different contexts, for example, parks that cater to specific recreational groups (rock climbers, campers, etc). Future research could also integrate interactive, gamified elements to existing campaigns, such as LNT messaging. As this study was conducted prior to the emergence of COVID-19, follow-up studies could now reveal different attitudes toward human, animal, and environmental health among the general population. Our study reveals changes in attitudes following interaction with gamified interventions. Further studies could assess the potential for gamified messaging as a behavior change tool.

Conclusion

The results of this study suggest an encouraging opportunity for park management to leverage the link between ecosystem health and human health in communicating conservation goals. This study underscores the need to view compliance issues from the lens of park visitor “self-interests” and explores the opportunity to position “compliant behaviors” as mutually benefitting “the park” through a social marketing framework that assesses exchanges and trade-offs between “good deals” and “bad deals.” This can be achieved by first identifying visitor self-interests and then accommodating those interests. Park managers need to recognize that there is always “competition” and that for every choice, there are competing benefits. The list of “self-interest” reasons for doing the behaviors in this study provides insights that can be researched further through concept testing with different messaging that conveys different benefits and “costs.” Park managers need to apply this understanding of social marketing principles to “tip the scales” toward one choice and away from competing choices, and in doing so achieve both conservation and self-interest benefits.

Finally, the study provides an opportunity for park managers or other organizations to “target” park visitors by activity and residence. The “use of park” and “distance of residence” data can allow park managers or other organizations to identify common behaviors among segments of the population so that audiences can be targeted for information and communication intervention purposes. For instance, park management or volunteer organizations could subsequently target recreation (e.g. hiking, kayaking) and neighborhood associations with information and communication interventions based on findings with an opportunity to engage them in the solutions.

The combination of game-based engagement and social marketing principles employed in this study points to the need for park service management personnel to explore best practices from other disciplines beyond conservation and the environmental sciences. New approaches should involve sectors including public health, education, and the behavioral sciences, as well as new innovations including digital simulations and interactive videos to foster conservation behaviors that align the interests of both the park and park visitor. National parks provide a living laboratory to continue to study the link between human, wildlife, and ecosystem health. With their mission of protecting all species, the NPS can be a place to understand the interconnectedness of all living things.

Acknowledgements

We would like to thank Howard Delafield International LLP for their financial support of this research. We would also like to thank the team of researchers who conducted these surveys, including Abby Butler-Cefalo, Marcy Delos, Natasha Trock, and research coordinator Dylan Donlon-Moyer from George Mason University. We appreciate the work of Margie Joyce in translating storyboards into interactive media for the treatment groups. We also would like to thank Fly Guy Interactive, an initiative at George Mason's Virginia Serious Game Institute, who created the animation and architecture for the simulation and the video. Special thanks to Orin Adcox, who conducted the visual animation for the simulation and video and served as overall producer of the game and video production.

Disclosure statement

Susan Howard, the PI of this project, is a co-owner of Howard Delafield International, a consulting company which designs and develops public health and environmental game-based learning platforms. HDI does not anticipate any financial benefits, including licenses from products, as a result of this study.

Data availability statement

The data that support the findings of this study are openly available in Zenodo at https://zenodo.org/badge/latestdoi/283313029, reference number 10.5281/zenodo.3964833.

Notes

1 Gamification: “The use of game design elements in non-game contexts” (Deterding et al., 2011).