2.1 Story creation and delivery

An extensive research and literature review on elephant behavior, the ecological impact of elephants on ecosystems, HEC issues, and conservation strategies was conducted to understand pertinent issues (Appendix SA), which then informed the development of the interactive CYOA story. Complex scientific concepts were simplified into accessible and engaging narratives while ensuring that the material remained scientifically accurate and relevant. This process involved first identifying core ideas, translating technical terms into everyday language, and then synthesizing the topics into a coherent storyline. The resulting story was both understandable and engaging for the public while preserving its educational value. The story was delivered in two modes: online animation via YouTube and live performances. Both formats allowed for participant interaction at fixed decision-making intervals, leading to one of four possible endings, which were framed either positively or negatively. The YouTube version of the story was disseminated through social media platforms from 14 March to 14 August 2022. At the end of the activity, participants could choose to replay the story, and if they did, they were asked in the postactivity survey how many endings they had witnessed. The live performances were conducted at two events with different audiences. The first was at the “World Elephant Day” (WED) carnival in FELDA Lok Heng Barat, Johor, on 13 August 2022, where the story was performed three times (10:00 AM, 1:00 PM, and 3:00 PM) for members of the plantation and Lok Heng community, who frequently face conflicts with elephants due to crop-raiding. The second event was part of the “Merdeka Lecture” series at the University of Nottingham Malaysia (UNM) campus on 1 September 2022, where the story was performed once for high school students to spark their interest in STEM education. During the live performances, a collective decision-making process was implemented through audience voting, with the majority vote determining the story's direction. The stories for the online and live versions differed slightly, with the live performances being extended to increase the duration of audience engagement (Appendices SB and SC). Figure 1 illustrates the research concept, focusing on the evaluation of the CYOA's impact on audience perception, learning, and engagement in conservation activities related to elephant conservation. Table 1 summarizes the parameters measured and the research questions tackled.

To evaluate audience experiences of the CYOA activities, postactivity surveys were designed and administered at the end of each activity. The survey for the YouTube activity gathered participants' opinions on the most effective methods to protect elephants and curb poaching, their understanding of the ecological role of elephants, and their likelihood of engaging in a range of conservation activities. It included sections where respondents rated (on a Likert scale of 1–6) their engagement in specific behaviors, expressed their agreement or disagreement (on a Likert scale of 1–6) with statements related to conservation issues, and reported on actions taken while traveling, as well as their volunteer and donation behaviors (Appendix SD). Additionally, the survey sought feedback on a conservation-themed activity, evaluating participants' intrinsic motivation, overall experience, the most important lesson learned from the activity, and the most important question they had after the activity. The item on the most important lesson learned provided insights into the audience's understanding of the complexities of HEC and the need for sustainable solutions. The item on the most important question captured participants' curiosity and potential areas for further exploration, which could help to identify knowledge gaps for continued learning. The survey also collected sociodemographic details, including Gender, Age, Region, Socioeconomic level, and Occupation, with categories provided in Appendix SD. A QR code linked to the survey was provided at the end of the final YouTube video.

In the live version of the activity, the audience was asked three key questions: “Why, how, and when did the elephant cross the road?” These questions were designed to assess their understanding and engagement with the complexities of elephant behavior. The “why” question explored their interest in the motivations behind the elephant's behavior, “how” gauged their curiosity about the biological mechanisms and challenges involved, and “when” measured their awareness of the importance of timing in human–elephant interactions. Responses were collected both pre- and postactivity for the UNM Merdeka performance, and only postactivity for the Lok Heng Barat performance. Further details on the surveys are presented in Appendix SD.

The postactivity surveys differed between the online and live versions for several reasons. First, the time constraints during live events only permitted a brief survey to be administered after the activity, which also meant that sociodemographic details were not collected. Second, to assess the effectiveness of the online activity in delivering a conservation message, we administered a similar online survey to a control group that had not participated in the activity. This control group was recruited via social media, in the same way as the “online activity” group. Third, the differing audiences for each activity made it challenging to directly compare the responses. Hence, we adjusted the survey for each activity based on the research question we would like to address. The details of the survey designs for the online and live versions are outlined in Appendix SD.

2.2 Statistical analysis

We used a Wilcoxon signed-rank test to compare the ordinal ratings of the activities against a neutral score (3). We also compared each pair of ratings of the three activities using Mann–Whitney U tests. These nonparametric tests are particularly suitable for analyzing ordinal data. To determine if the lessons learned postactivity differed among the three activities, we first categorized the open-ended responses into three types: knowledge, understanding, and application. A response was categorized as “knowledge” if it involved a fact or information about elephants. If the response provided a reason or mechanism explaining elephant behavior, it was categorized as “understanding.” Responses indicating what the participant intended to do to benefit elephants were classified as “application.” We then used Fisher's exact test to assess whether the types of lessons learned differed across the activities.

For the live performance at UNM, we first assessed the accuracy of participants' responses to the questions “Why, how, and when did the elephant cross the road?” by marking the correctness of each response. We then used Pearson's chi-squared test to determine whether the accuracy of responses differed between the preactivity and postactivity phases.

It is pertinent to note that the UNM event boasted a notably larger sample size of 259 participants, compared to the 39 participants at the WED event. While both events shared similarities, our focus primarily centered on the UNM event due to its direct relevance to the study objectives and the availability of more comprehensive data. We emphasize this context to provide transparency regarding data exclusion, ensuring clarity and rigor in our reporting of methodologies and results.

To examine whether the responses to the YouTube version differed according to socioeconomic factors, we employed a generalized linear model (GLM). The survey responses (statements 4–8 and 16–18) served as the response variable, with “gender,” “age,” “region,” “socioeconomic level,” and “occupation” as explanatory variables. Given the small number of participants in certain categories, we grouped them as follows: age (below 25 or above 25 years old), region (Southeast Asia [SEA], others), socioeconomic level (nonworking, working, middle, or upper-middle), and occupation (education, others). Additionally, to examine the effects of the activity's message on behavioral intentions, we also applied a GLM, using the responses to each of the 12 “behavioral intention” statements (question 4) as the response variable. The explanatory variables included “gender,” “age,” “region,” “socioeconomic level,” and “occupation,” along with “ending type” and “message type.” The “ending type” was categorized as “direct,” “indirect,” or “multiple.” A “direct” ending relates to the behavioral intention statement in the survey. For example, if the video's final message was “Slow down to save elephants,” it aligns with the survey statement “Be mindful and slow down when driving on roads with animal crossings.” An “indirect” ending refers to an end statement that is unrelated to the behavioral intention statements in the survey. “Multiple” means that the participant was exposed to both “indirect” and “direct” messages because they chose to replay the story. The study categorized “message type” as “positive,” “negative,” or “multiple.” A “positive” message highlights the beneficial, optimistic, or constructive aspects of a situation or idea. For example, one of our endings features a positively framed message: “Slow down, and you can help save elephants and your life.” Conversely, a “negative” message emphasizes the potential loss or adverse consequences, such as: “Slow down, or you may injure elephants and yourself.” When participants were exposed to both types of messages, it was categorized as “multiple.” For analyzing the survey data, different statistical models were employed based on the type of response variable. Binary response variables were analyzed using the “Binomial” error family, numerical (counts) used the “Poisson” error family, and ordinal data were analyzed with a cumulative link model.

3.1 The efficacy of delivery methods in CYOA activities

To assist in interpreting the results, numbers in brackets are displayed in the following format: estimate of the difference in means ± standard error of the difference, significance of the p-value (*p < 0.05, **p < 0.01, ***p < 0.001), test statistics.

The study found that all activities—Merdeka, WED, and YouTube—received significantly higher overall ratings compared to the neutral score of 3 (Table 2). Specifically, theatrical activities were rated higher than the online activity (see Table 2). WED received notably higher ratings than YouTube, while Merdeka's ratings, though higher than YouTube's, were lower than those for WED (Table 2, Figure SE2). Participants across all activities primarily gained insights into elephant conservation, with slight variations in focus: WED emphasized elephant behavior and factors contributing to HEC, Merdeka highlighted coexistence, protection, and conservation, and YouTube concentrated on threats like poaching and habitat fragmentation. Fisher's exact test indicated no significant difference in the lessons learned among the three activities (p > .05) (see Table 2, Figures SE2 and SE3 for details).

In the Merdeka activity, participants' postactivity responses to the three questions “Why, how, and when did the elephant cross the road?” were more consistent compared to their preactivity responses (Figure 2). The majority of preactivity responses to “why” focused on the motivations of elephants to seek shelter and sustenance, as well as habitat loss, while postactivity responses centered on the elephants' need to forage and for social interaction, particularly to find a mating partner. For “how,” prevalent answers from preactivity were on physical mechanisms used by elephants in road-crossing movement, for example, “crossed,” “walked,” and elephants' social behavior, that is, traveling in social groups. Postactivity responses were similar to the preactivity but included additional mentions of the paths elephants take, such as wildlife crossings, bridges, and overpasses. Responses to “when” revolved around the need to crossroads for food, shelter, or due to habitat loss, and the time that elephants would make these crossings, that is, midnight. The postactivity had similar themes to the preactivity, where participants mentioned the exigencies that result in road-crossing behavior exhibited by elephants, that is, for food and finding a partner, and the time these crossings were made, that is, anytime and night.

The chi-square test output revealed significant differences in the correctness of responses to the questions “why” (χ² = 80.623, df = 1***, Figure 3a) and “how” (χ² = 38.386, df = 1***, Figure 3b) pre- versus postactivity. However, there was no significant difference for the question “when” (χ² = 0.25502, df = 1, Figure 3c).

3.2 Effects of the online activity on learning, behavior, and perception

The online activity was analyzed in greater detail, offering a comprehensive examination of its impact on learning, behavior, and perception. This format allowed for an in-depth survey that gathered both quantitative data on audience engagement and qualitative insights into participants' perceptions and learning outcomes regarding elephant conservation. A control survey was also implemented, providing a basis for comparison (Table 1). The wide range of topics and demographics covered in both surveys provided a detailed understanding of how exposure to a YouTube video on conservation influences audience perception, learning, and engagement.

3.2.1 Perception

Participants' overall perception of the activity was positive, with all statements receiving ratings above average (≤3.5) (Figure SE4; Table S7). On average, participants strongly agreed that the activity allowed engagement with both the character and the topic. This enjoyment was reflected in high ratings for the following statements, that is, “time flew by,” “enthusiastic and excited,” “enjoyment,” and “overall satisfaction level of experience.” Furthermore, the activity helped participants deepen their understanding of conservation and attain positive feelings, as evidenced by the average ratings for statements like “strengthen my understanding of conservation” and “think I can help elephant achieve a good outcome.” Many participants found that their prior understanding of conservation was helpful during the activities. Details on (1) the best and most memorable features, (2) suggested improvements, and (3) participants' feelings are detailed in Figure SE6.

The responses to the three open-ended questions in Figure 4 reveal a wide diversity of answers. Both groups emphasized habitat protection and education as key strategies for saving elephants. However, there were notable distinctions: the treatment group mentioned law enforcement activities and wildlife bridges, while the control group suggested coexistence strategies.

When asked about the best methods to stop poaching, both groups focused on the importance of proper governance, stricter law enforcement, punishments for poachers, and education. The responses from the treatment group were more diverse when discussing the ecological role of elephants, but the themes were consistent. Participants recognized the role of elephants as keystone species, ecological engineers, and seed dispersers, and acknowledged the importance of elephant feces for soil fertility.

3.2.2 Learning

The online survey responses comparing the control and treatment groups showed similar levels of learning across statements on poaching, development, governance, and behavioral change (Figure 5). The responses only differed for one statement (Table S1): selling ivory to generate income for elephant conservation (−1.183 ± S.E. 0.461 ***, LR = 15.750).

Regarding the correctness of eight statements on elephants and forests, the treatment group showed slightly higher average correctness (Treatment: x̄ = 2.811 ± 0.183; Control: x̄ = 2.362 ± 0.198). However, the differences were not statistically significant (0.704 ± 0.409 S.E., LR = 2.984) (Table S2).

3.2.3 Engagement

Significant differences were observed between the control and treatment groups in their likelihood to engage in conservation activities, as shown in Figure 6. Specifically, the treatment group exhibited a higher likelihood of participating in five key actions: reporting poachers (1.214 ± S.E. 0.429**, LR = 8.257), paying attention to wildlife and environmental issues in the media (0.956 ± S.E. 0.423*, LR = 5.230), sharing wildlife and environmental issues on social media (−1.399 ± S.E. 0.438**, LR = 10.806), recommending the activity to others (1.535 ± S.E. 0.427***, LR = 13.500), and supporting the conservation of wild elephants in Malaysia (0.827 ± S.E. 0.422*, LR = 4.074) (Table S3).

However, participants from the treatment group ranked lower for statements related to actions taken when encountering wildlife on the road (Figures SE5): “Stop when I see an animal crossing the road” (Treatment: x̄ = 2.453 ± 0.190; Control: x̄ = 3.149 ± 0.199), “Slow down when I see ‘animal crossing’ signs” (Treatment: x̄ = 2.585 ± 0.178; Control: x̄ = 2.809 ± 0.154), and “Look out for signs on speed limit and animal crossings” (Treatment: x̄ = 3.000 ± 0.148; Control: x̄ = 3.319 ± 0.258).

Conversely, the treatment group ranked higher for statements regarding the timing of driving (Treatment: x̄ = 3.528 ± 0.204; Control: x̄ = 3.043 ± 0.164) and the use of public transport (Treatment: x̄ = 3.434 ± 0.208; Control: x̄ = 2.681 ± 0.234). Significant differences were found between the control and treatment groups for three statements: taking public transportation (1.067 ± S.E. 0.413*, LR = 6.874), driving during the day instead of at night (0.758 ± S.E. 0.403*, LR = 9.457), and stopping when encountering an animal crossing the road (−1.159 ± S.E. 0.427*, LR = 7.620).

3.3 Factors affecting learning, behavior, and perception in the online activity

3.4 Ending and message type on behavior and perception

The ending type in the CYOA activity did not significantly affect participants' responses to statements related to driving mindfully, reporting poachers, recommending the activity, planting trees, decreasing car usage, or supporting wild elephant conservation (Table S6).

Moreover, the difference in message type (positive or negative) did not affect participants' likelihood to engage in future conservation activities, but certain socioeconomic factors did (Table S6): Participants aged above 25 years were less likely to recommend the activity (−1.208 ± S.E. 0.655*) or be involved in tree planting (−1.479 ± S.E. 0.654*); nonconservation jobs had a negative influence on involvement in tree planting (−1.469 ± S.E. 0.678*); and being a Southeast Asian participant had a positive effect on driving mindfully on roads with animal crossings (2.00 ± S.E. 0.981*).

We found that message type affected participants' feelings. Specifically, positive messages were associated with increased positive feelings toward the activity (−1.289 ± S.E. 1.313*), while the opposite was found for negative messages (Table S7). Participants' overall satisfaction was also influenced by the ending type (1.575 ± S.E. 1.209*), with ending types 3 and 4 showing a more positive satisfaction level relative to other endings (Table S7).

4.1 The efficacy of delivery methods in CYOA activities

Previous studies have explored various forms of CYOA activities, including visual novels, texts, and questionnaires. In our study, we tested two distinct delivery modes—live theater performances and online videos—to disseminate messages on elephant conservation. Our findings indicate a difference in participants' overall satisfaction between the two approaches, with live theater (WED and Merdeka performances) receiving higher ratings compared to the YouTube video format. The preference for live theater suggests that physical activities may foster more active participation and higher audience engagement than online activities. This aligns with existing literature that emphasizes the importance of a sense of peer community and engaging instruction in successful online learning environments (Norman, 2020). In contrast, our online activity lacked narration, potentially resulting in lower audience engagement levels. Narration in storytelling can enhance engagement by incorporating vocal elements such as tone, loudness, accents, and emotion, which may increase the attention span of the audience. Given these findings, future studies should explore whether the inclusion of narration in online CYOA activities could improve audience engagement and learning outcomes.

Regardless of the delivery method, participants from both the online and theatrical activities successfully captured the central message of elephant conservation and protection. The WED and Merdeka activities managed to convey themes of HEC and coexistence, while the YouTube activity focused on anthropogenic threats that elephants face and the human responsibility to conserve and protect these sentient beings. The results from the Fisher's exact test confirmed the similarities between the messages received by participants across the three activities, verifying the effectiveness of each delivery method in conveying the story's objectives.

The uniformity of responses in the postactivity surveys, compared to the preactivity surveys, suggests that participants were able to clearly encapsulate the overarching theme of the events. For all questions, participants' responses mirrored our hypotheses. They demonstrated an understanding of (1) the intrinsic reasons behind elephant behavior, (2) the general physical mechanisms involved in elephant movement and social behavior when crossing roads in groups, and (3) the time elephants normally cross the roads and behaviors resulting in road crossings.

These responses were consistent with existing research, such as Wadey et al. (2018), which describes road-crossing behavior in elephants as predominantly nocturnal, and Yamamoto-Ebina et al. (2016), which notes that elephants are often attracted to highways due to food availability. Postactivity responses were generally more accurate than preactivity responses; for instance, participants refined their answers to “when,” answering “midnight” and “night” in the preactivity survey and “night” for postactivity. Some of the postactivity answers were new, with participants noting the utilization of wildlife bridges for road-crossing. Similarly, when asked about the reasoning for elephant movement, participants added “social behavior,” that is, looking for mates, as one of the motivations behind road-crossing.

These patterns suggest that the activity helped correct misconceptions about elephant movement behaviors and deepened participants' understanding of the underlying mechanisms and elements involved. An increase in the number of correct responses was exhibited by participants in the postactivity survey for the “why” and “how” questions but not for “when,” possibly due to participants' awareness of the timing of elephant movement before the activity. These findings are consistent with previous studies by Melcer et al. (2020) and Panjaitan (2016), which showed that participants scored higher on postactivity surveys after being exposed to CYOA stories. CYOA activities provide appropriate contextual frameworks that help learners correct misconceptions and may improve the long-term retention of new knowledge.

4.2 Effectiveness of the online activity for learning, perception, and engagement

The perception of the online activity was positive, with participants expressing general satisfaction with the experience. The activities allowed participants to engage with the characters and topics of the story while also strengthening their understanding of elephant conservation. The concept of “online CYOA” offers an alternative narrative style of teaching, similar to educational video games. The CYOA activity available on YouTube functions much like a point-and-click video game in its mechanics. Research indicates that students are more motivated to learn ecological concepts through video games, with learning performance comparable to traditional methods (Hwang et al., 2013; Kuo, 2007). This suggests that our online CYOA can be an effective tool for enhancing ecological education and engaging learners. Additionally, the positive attribution toward the CYOA activity is consistent with findings from previous studies, with participants finding the CYOA format to be engaging, enabling a deeper understanding of the subject due to its interactive nature (Jenkins, 2014; Panjaitan, 2016; Scott et al., 2021; Stanfill & Martin, 2023; Thomas et al., 2022). Qualitative responses from participants highlighted the flexibility to explore alternative endings, the engaging storyline, animation, and messages as the best features of the activity. By allowing the exploration of multiple endings, the participants were able to understand the cause and effect of different actions. This reiterative method of understanding a new concept also contributes to strengthening learning outcomes.

To confirm the effectiveness of the online activity for individual learning, perception, and engagement in conservation, we measured the differences in audience responses toward different statements between the control and treatment groups.

4.3 Influence of sociodemography, ending, and message type on perception, learning, and engagement in the online activity

4.4 Synthesis

Our results show that the CYOA story approach is both engaging and effective in enhancing comprehension, fostering engagement, and encouraging behavioral and perception changes concerning HEC. For example, participants showed significant improvement in the accuracy of their answers to the question of why elephants crossroads after engaging with the story. This suggests that interactive learning methods, such as CYOA activities, can facilitate learning and foster positive attitudes toward elephants by actively involving participants in the narrative (Bruner, 1986; Young & Monroe, 1996). However, the impact of CYOA activities on actual conservation engagement was negligible, possibly due to limitations in the storyline. The story did not adequately highlight the consequences of HEC on other key stakeholders, which could have limited its effectiveness in motivating participants to take conservation action (Niemiec et al., 2020). More research is needed to test the effects of CYOA on the intention to engage in conservation by including multistakeholder perspectives into the storyline to enable an understanding of the multidimensional impacts of HEC. Additionally, improvements to the online version of the CYOA story, such as adding voice narration, could further improve the interactive experience and potentially increase its effectiveness.

Implementing the CYOA activity approach in other topics and contexts presents several challenges. For instance, translating complex conservation information into interactive elements requires the educator to have a deep understanding of the learning outcomes associated with different narrative paths. For example, in our study, each decision path led to a specific learning outcome, but participants had to restart the story to explore alternate scenarios, and the effect of this restarting process was not measured. In addition, participants following different paths might experience inconsistent or incomplete learning outcomes. To address this, future CYOA activity designs could include a learner profile that tracks lessons learned and maps them to the story's educational objectives. Another challenge is to create a story with decision-making points that accurately reflect real-life situations. In our study, the decision tree for elephant behaviors was informed by extensive research on HEC and elephant movement ecology, but such detailed information may not be available for all species or ecosystems where conservation is urgently needed. Nonetheless, this gap highlights the need for future research to focus on the importance of understanding animal behavior and ecology in conservation efforts. Furthermore, crafting survey questions that accurately capture the lessons learned is also challenging. We suggest that future studies incorporate periodic questionnaires to assess memory retention over short, medium, and long terms. While this study highlighted the benefits of the CYOA teaching method through a paired test—comparing participants' perception, learning, and conservation engagement with and without the CYOA activity—these findings should be supplemented by quantitative comparisons of the efficiency of CYOA against other educational methods.

Storytelling is a powerful tool in conservation education, balancing the accuracy of information with mechanisms of engagement, especially for nonexpert audiences (Avraamidou & Osborne, 2009; Dahlstrom, 2014). The CYOA approach has the potential to promote engagement in conservation and build awareness of complex conservation issues by reinforcing the consequences of actions. This method also encourages a perspective shift through the exploration of different paths, promoting creative problem-solving while learning about alternative decisions that lead to different outcomes (Dincelli & Chengalur-Smith, 2020). By simulating direct experiences, CYOA storytelling can better prepare communities that are prone to conflicts with wildlife for potential encounters, increasing their willingness to coexist in shared landscapes.

Some questions arise from the results of this study that must be addressed before CYOA stories can be effectively replicated in other educational contexts, especially to foster engagement in conservation: (1) How can scientific language and evidence be simplified in a story without compromising the rigor and accuracy of information? (2) How can information and learning outcomes be standardized across stories created by different educators and between online and physical settings? (3) How well would participants communicate what they learn from the activity? Tackling these questions requires an in-depth analysis that incorporates pedagogical, social, and environmental perspectives, ensuring that CYOA stories can be a reliable and impactful tool in conservation education.

4.5 Further studies

Future research could explore the cultural adaptability of storytelling approaches, examining how different narratives and storytelling techniques resonate with diverse audiences, including those from various ethnic, linguistic, cultural, and socioeconomic backgrounds. Social factors exert a significant influence on how people perceive human–wildlife conflicts (Dickman, 2010), making it imperative to adopt more innovative and interdisciplinary approaches to wildlife mitigation strategies. This study highlights the potential of CYOA storytelling in conservation education. Another avenue for future study could involve longitudinal research to assess the long-term impact of CYOA storytelling on attitudes, engagement, and decision-making related to conservation. This would also include measuring the sustainability of engagement over time. Furthermore, investigating the potential scalability of CYOA storytelling through digital platforms, social media, or mobile applications could help reach larger and more diverse audiences. Exploring the effectiveness of incorporating interactive technologies, such as virtual or augmented reality, into CYOA storytelling for conservation education is another promising area. Such research could determine whether immersive experiences further enhance engagement and behavioral change. Finally, examining the transferability of the CYOA storytelling method to other domains beyond conservation, such as public health, disaster preparedness, or environmental policy, could offer insights into its broader applications and effectiveness.

In conclusion, our study demonstrates the engaging and effective nature of the CYOA storytelling approach in enhancing education about elephant movement ecology, behavior, and HEC. While our findings highlight the potential benefits of interactive teaching styles and the positive outcomes of participant engagement, several challenges remain that warrant further exploration. Replicating this approach across diverse topics and contexts will require addressing issues such as simplifying scientific language without compromising accuracy, assessing the impact on understanding complex issues, standardizing learning outcomes, and evaluating the effective communication of knowledge gained. Further studies could expand the application of the CYOA method to a wider range of educational disciplines, adapt narratives for diverse audiences, evaluate long-term impact, explore immersive technologies, and investigate its applicability beyond conservation. By addressing these considerations, we can continue harnessing the art of storytelling to balance accuracy and engagement, ultimately fostering greater engagement and awareness in environmental education.