Teaching about Marginalized Groups Using a Digital Human Library: Lessons Learned

Abstract

This paper presents lessons learned from a project inspired by digital storytelling and the human library to reduce prejudices against marginalized groups. By comparing the outcomes of similar participants in different settings over the same period, the study explored which types of activities might be pivotal when influencing the perspective-taking attitudes of participants. The study used a case study approach, with data from the digital human library project, and selected participants from three different engagement contexts: participants in group A were involved in reading story abstracts online, having short face-to-face meetings regarding human books, and engaging in editorial activities; participants in group B were involved in extended face-to-face sharing provided by human books, followed by question-and-answer interaction; and participants in group C were involved in the reading of stories online without interaction. Convenience sampling was used and included 250 registered participants who completed pre-test and post-test questionnaires. The study found that merely reading stories online (group C) did not significantly reduce prejudice, and face-to-face contact on its own (group B) was also not the most effective in changing attitudes. Group A participants who combined short face-to-face meetings and story-retelling activities showed the most significant changes in perspective-taking attitudes. These findings imply that dialogic cognitive processes in narrative activities, rather than the mode of contact, may be pivotal in enhancing perspective-taking attitudes. This paper calls for further research into the scalability of digital human library hybrids and more rigorous experimental research designs. It underscores the potential of these interventions to foster more inclusive societies, mitigate social biases, and support equity.

1. Marginalized Groups in Hong Kong

Marginalized groups are individuals at risk of discrimination based on attributes such as ethnicity, religion, sex, gender, age, health status, disability, or education level (David 2013; Matheson et al. 2019). The term “marginalized” rather than “marginal” reflects a shift that occurred during the social revolutions of the 1970s, highlighting the experiences of those pushed to society’s fringes (Bhopal and Deuchar 2015). These individuals are systematically denied full societal participation, lacking the resources to improve their conditions. Despite the diverse challenges these groups face, they share a common social identity shaped by their experiences of marginalization.

In Hong Kong, as is the case in many developed regions, marginalized groups face various forms of discrimination. For instance, mainland Chinese immigrants often encounter hostility from some locals (Chou 2012), Muslims must negotiate to cope with prejudice (O’Connor 2010), transgender individuals face derogatory terms (King 2009), southeast Asians experience clear societal exclusion (Hong Kong Unison 2012), and foreign domestic workers are stereotyped due to policy frameworks (Cheung et al. 2019). Current efforts in Hong Kong focus on helping marginalized groups assimilate into mainstream society through skill development and protective institutional measures, including several anti-discrimination ordinances. Educational initiatives are also in place to integrate non-Chinese speaking students and support those with special educational needs (Hong Kong Unison 2012; The Hong Kong Council of Social Service 2014).

However, critics argue that these measures are inadequate because they fail to address the root prejudices within the majority population (Lau 2015; Paluck and Green 2009). Educating the public about marginalized groups involves several challenges (Kumagai and Lypson 2009; Kumashiro 2013). Educators must foster a deep understanding and sensitivity toward the issues faced by these groups, which can be difficult due to diverse personal backgrounds and training levels. Additionally, facilitating direct interactions between the public and marginalized groups can lead to uncomfortable situations if not properly managed. Moreover, curricular constraints and time limitations often prevent these critical topics from being explored in depth, hindering the public’s thorough understanding and appreciation.

Against this background, we initiated a digital Human Library project in Hong Kong to address these gaps and to attempt to influence public attitudes toward marginalized groups. This paper discusses what we learned from the project’s outcomes in order to encourage further discussion and research into more effective strategies.

2. Digital Storytelling (DST)

The project was a fusion of two practice ideas commonly used to serve marginalized groups. One of the background ideas is digital storytelling (DST), a storytelling method interwoven with selected digitized images, texts, sounds, and other digital elements (Chan and Sage 2021; Chan and Yau 2019). Partly because of the powerful distribution capability of the internet, human service practitioners have become aware of the potential of using DST to enable marginalized groups to have their voices heard (Botfield et al. 2017; de Jager et al. 2017; Emert 2013; Fokides 2016; Guse et al. 2013; Mnisi 2015; Teti et al. 2016). Some studies have shown that such activities help decrease participants’ closed-mindedness and essentialist views (Chan and Holosko 2020) and also enhance participants’ critical openness and perspective-taking attitude (Chan 2019a, 2019b; Chan et al. 2022). Despite the growing popularity of DST, some reviews highlight its current limitations and identify several research gaps.

First and foremost, most DST practices focus on enabling protagonists to express their views, and current research leaves unclear how these stories will impact the audiences (the majority, the public) or whether the public will equally engage with them. Indeed, practitioners have noted that little is known about how digital stories are used once circulated in the public domain (Lenette et al. 2015).

Chan and Sage (2021) have suggested that the term DST sometimes refers to a specific genre; for example, some writers refer to the definition of the StoryCenter, seeing DST as a production genre of a video lasting a few minutes (Lambert 2010). Some use it as an umbrella term covering different digital production activities; these may be referred to by various terms, such as “youth media production” or “photovoice.” They have identified diverse DST models and theoretical bases that serve different purposes. Additionally, many DST initiatives are event-based and episodic; they do not constitute sustainable long-term practices. This means that the format of digital storytelling (DST) does not have a universal consensus, and its effectiveness also depends on the specific types of DST design being referred to.

Meyerhofer-Parra et al. (2024) have provided maybe the most innovative review of DST thus far. The article discusses the evolution of storytelling within the digital and transmedia fields, highlighting the transition from digital to post-digital storytelling. The discussion implies that “digital” has lost its currency when everything has already gone digital. With the rise of transmedia storytelling, which involves spreading a narrative across multiple platforms and media, the practice has become more complex and diversified. In the post-digital era, storytelling practices have further evolved to seamlessly integrate digital, analog, face-to-face, and other media forms. This hybrid approach transcends the simple digital/non-digital dichotomy, incorporating diverse communication modes and activity forms to create richer, more engaging narratives. Thus, the term “digital” alone no longer captures the multifaceted nature of modern storytelling practices, necessitating broader concepts like “post-digital” and “transmedia” to better describe this landscape.

3. Human Library (HL)

Another practice idea informing the project is the Human Library (HL), which was developed in Denmark in 2000. In brief, HL practice addresses people’s prejudices by helping them talk to those they would not usually meet. The organizers used a library analogy to leverage the idiom “don’t judge a book by its cover”, used the term human books to name people from marginalized groups, and posited public participants as readers. Such authentic contact and conversations were intended to help reduce social prejudice. The first HL event ran for several days, with eight hours of conversation daily, and more than 1000 people took part (Little et al. 2011; Zhai et al. 2012). This practice idea has evolved into an organization (https://humanlibrary.org/ (accessed on 6 May 2024)) and has inspired similar practices in various regions across the globe (Bagci and Blazhenkova 2020; Garbutt 2008; Kinsley 2009; Kudo et al. 2011; Kwan 2020; Stewart and Richardson 2011; Wong and Lin 2023).

The Human Library is also called Living Library. The two terms refer to the same concept. Initially, all activities were conducted under the name Living Library. In 2010, the organization was renamed the Human Library Organization and, since then, activities in the UK and the US have tended to use the term Human Library (Little et al. 2011). Despite the difference in names, both terms describe the same methodology and share the same aims and values of promoting intercultural dialogue and reducing prejudice through personal conversations between human books and readers. The human books involved in the project include diverse types of marginalized groups.

Little et al. (2011) provided the earliest official documentation introducing the HL method in their guide “Don’t Judge a Book by Its Cover!” However, while this guide outlines the concept and purpose of the Living Library, it does not provide specific theorizations or detailed evaluation methods for assessing its effectiveness. This omission means that, while the foundational principles and goals are clear, organizers are left without concrete steps by which to conduct and measure the success of the HL interventions.

Orosz et al. (2016) conducted one of the earliest evaluations of the use of the HL method to reduce prejudice toward Roma and LGBT communities in Hungary. The study involved high school students who interacted with trained volunteers (books) from these communities. These volunteers shared personal stories of discrimination and social exclusion with the students (as readers). The study’s results indicate that the HL intervention effectively reduced participant prejudice. However, the study did not detail the micro-level processes involved in the HL intervention or clarify how their findings could be generalized to other HL methods.

Lam et al. (2023) have provided the most comprehensive current review of implementation, processes, and outcomes of human library (HL) practices. Their review covers diverse aspects, including the variations in format, venue, scale, preparation, and recruitment of HL programs. It highlights the wide range of social identities represented by the books and the predominant educational backgrounds of the readers. The review also discusses the reported outcomes of HL events, such as reduced prejudices, improved attitudes, and personal growth for both readers and books.

The review includes a total of 23 publications, consisting of 21 journal articles and 2 book chapters. These publications cover a range of study types, including descriptive studies, case studies, post-event surveys, pre-/post-evaluation studies, and qualitative research. Specifically, 14 (61%) of the studies are descriptive, 5 (22%) use pre-/post-evaluations, and others employ systematic qualitative designs like interviews and phenomenological case studies. This variety in research methodologies helped capture a comprehensive view of HL practices and their impacts.

The review concludes that HL lacks systematic documentation and rigorous empirical evidence. Furthermore, its conceptual analysis and process of change are underexplored. The study emphasizes the need for theorization, structured guidelines, and rigorous research methodologies to better understand and enhance the effectiveness of HL initiatives.

In sum, although the HL method is inspiring and involves an interesting analogy, reviews indicate research gaps. First, there is scant empirical research on the effectiveness of the Human Library in changing readers’ attitudes. Second, the literature has not explained how meeting a human book differs from an ordinary conversation. Third, participants and advocates of HL emphasize face-to-face contact but have not adequately proven why this method is more effective than other modes of communication. This emphasis on face-to-face interactions also limits the potential for broader application and development of the practice, as it relies heavily on physical presence and personal interaction.

4. A Project Inspired by HL and DST

Although they differ in methods and rhetoric, DST and HL practices share a common mission: amplifying the voices of disadvantaged groups. Both approaches prioritize live stories; DST emphasizes digital and online elements, while HL focuses on authentic face-to-face interactions. However, their rhetoric does not always align with actual practices, and their shared goals appear more substantial than their differences. Reviews indicate that DST is not entirely digital or online (Chan and Sage 2021; Meyerhofer-Parra et al. 2024), and, with regard to HL, it has not been convincingly theorized as to how and why face-to-face contact leads to positive changes (Lam et al. 2023; Wong and Lin 2023). Thus, we see these concepts as inspirations and have considered how specific, digitally enhanced storytelling activities using authentic live stories (human books) can achieve intended outcomes. This paper reports on a project that leverages the strengths of these two practice ideas.

The project was a pilot social service program in Hong Kong named zhenren tushu wang in Chinese (literally translated into English as “the web of human books”). It was funded by the Hong Kong Jockey Club between January 2019 and Aug 2022 (see https://lib2022.humans.asia/ (accessed on 6 May 2024)). The operation of the project was straightforward. First, it engaged public participants to work with individuals from marginalized groups willing to share their life stories. Second, it presented these stories on a website (i.e., the library). Finally, it used such stories to engage public participants in workshop activities.

This online human library is a digital “storytelling library”, meaning that the library itself tells stories. It is an online platform offering diverse storytelling formats, encompassing digital content, offline activities, and various methods. In essence, it is a digital platform that provides a broad spectrum of live stories from marginalized groups.

Discussion in library science and information studies not only focuses on the technical management of information but also emphasizes the importance of humanizing information to cater to personal stories and motivations related to information use (Aharony and Shonfeld 2015; Akinbobola and Adeleke 2016; Ryan et al. 2010). The idea of a “storytelling library” partly aligns with this line of discussion, integrating personal narratives into the library’s fabric and making it a more responsive resource. Using a library to offer a space where community, culture, and shared experiences come together enhances the relevance and accessibility of information.

The library collected over 150 stories from marginalized groups and organized more than 100 workshop events; it included the stories of ethnic minorities, people with special education needs, transgender people, people with physical disabilities, and people with mental illness (see Figure 1, Figure 2 and Figure 3). It attracted visitors and users from different backgrounds, but most were from Hong Kong. Google Analytics showed that the website accumulated 450,845 pageviews and had 45,236 unique visitors in the project period, of which 9032 were returning visitors. About 81.75% of the website users were in Hong Kong. The website also invited the registration of participants, with registered users being able to use more functions, such as reading the details, joining workshops, and leaving comments. Among those 45,236 unique visitors, 5800 became registered users. Among the 5800 registered users, 917 participated in some structured activities and provided data at multiple time points.

5. The Theoretical Base

Contact theory is a well-recognized approach to reducing prejudice, suggesting that prejudice can be diminished through equal-status contact between majority and minority groups working toward common goals (Allport et al. 1954; Fibbi et al. 2021). Traditionally, this theory emphasizes the importance of face-to-face interactions, which is also one of the theoretical foundations referenced by HL practices (Groyecka et al. 2019; Kwan 2020; Lam et al. 2023).

However, practical challenges often make face-to-face interactions between marginalized individuals and the broader public difficult. Such interactions do not consistently lead to positive outcomes and can sometimes reinforce existing prejudices if not carefully managed (Williams 1964). Additionally, contact alone may not be the sole factor in inducing perception changes. Research suggests that the effectiveness of these encounters often hinges more on the context, such as the setting, collaborative activities, and communication styles, rather than just meeting in person (Holmes and Butler 1987; McClendon 1972). Last but not least, the project was approved before the COVID-19 pandemic and was implemented during the pandemic, marked by prolonged social distancing and limited social contact (Chan and Au-yeung 2021; Ku et al. 2021). Such considerations and social circumstances forced us to rethink the value of face-to-face contact and explore alternative paradigms.

Considering these limitations, our study acknowledges the insights of contact theory while trying to explore a different lens. We propose that, rather than authentic contact or merely providing adequate information, the dialogic cognitive processes embedded in narrative activities may explain the ontology, epistemology, and methodology of prejudice reduction. Additionally, we suggest using media resources in narrative activities to strengthen these dialogic processes further.

From a theoretical standpoint, understanding is seen as a dialogic process where individuals may either revise or reaffirm their prejudices. This view is informed by hermeneutics, which argues that “prejudices” per se are not problematic but become so when interpretative processes malfunction (Gadamer 1996). Dominant misconceptions lead to stereotypical generalizations, and rigid preconceptions can turn understanding into a dogmatic, deductive process that reinforces confirmation bias (Nickerson 1998). A healthy cognitive process involves a dialectic relation between preunderstanding and new insights, facilitating a spiral of understanding rather than a linear progression.

Storytelling, therefore, serves as a tool with which to engage these interpretive circles, allowing individuals to reconsider overlooked life events and deconstruct dominant narratives. This approach aligns with narrative practice theories (White 2007), which describe narrative as a journey between the landscapes of consciousness (concepts, values) and actions (concrete events). Effective storytelling strategies help individuals reassess the links between their perceptions and actions, challenging entrenched beliefs.

Moreover, incorporating media resources (e.g., texts and photos) into storytelling can enhance these reflective processes. Media education theories suggest that symbolic mediators, such as signs and symbols, are crucial for meaning-making (Vygotsky and Luria 1994). Teachers can help students articulate their knowledge through media texts, by encouraging systematic thinking and by questioning underlying assumptions (Buckingham 2003). Using symbols and texts allows participants to visualize their thoughts and potentially revise or refine them.

Building on this theoretical assumption, we designed the project to facilitate interactions between marginalized groups (as human books) and the public (as readers) through narrative activities. The strategy aimed to facilitate public attitude changes through intrapersonal and interpersonal dialogues embedded in structured narrative processes. Participants engaged in dialogues with human books, other public participants, and themselves through activities like writing and rewriting stories and selecting photos.

6. A Quantitative Case Study Comparing Three Groups of Participants

6.1. The Context and Research Question

This paper presents a quantitative case study based on available data extracted from the digital storytelling–library project. The case study method enables a deep exploration of complex issues within our project implementation contexts, offering detailed insights. Using quantitative measures in case studies further enables researchers to systematically compare different cases and enhance the replicability of studies across various contexts (Dwyer and Walsh 2020; Shareia 2016).

The project involved diverse marginalized groups and the undertaking of eclectic activities, such as workshops, sharing experiences, lectures, mini concerts, media productions, and regular online chat programs. The project was a service and was not intended to be a laboratory experiment, so participants were not randomly assigned to participate in those activities. Background conditions limited the possibility of a rigorous experimental design and the extrapolation of the results toward a more general theory. Nevertheless, because the system tracked registered users’ online and offline participation records, it was possible to distinguish between the participants and obtain an overview of the activities in which they participated and the outcomes. Participants completed the consent form and a questionnaire when they registered for the platform, and they filled in the same questionnaire again after completing some organized activities. Therefore, the study could assess some psychological conditions across time, compare the differences among groups, and draw conclusions concerning the project’s outcomes.

The study was approved by the Hong Kong Polytechnic University’s Human Subjects Ethics Subcommittee (file number HSEARS20180302004). All participants were aged 18 or above, indicated their consent to provide personal data via self-reported questionnaires, and understood that their activities on the project website were recorded.

This study did not aim to prove the project’s effectiveness, as the design and available data did not support such straightforward generalizations. However, by comparing the outcomes of similar participants in different settings over the same period, the study explored which types of activities might be more pivotal in influencing the perspective-taking attitudes of participants.

6.2. Selection of Reader Participants from Three Types of Activities

Project participants were recruited through social media pages and personal networks, beginning with students in social work classes and members of social work organizations. Consequently, most participants are social service volunteers, junior social workers, students at the undergraduate level or above, and their friends. They were not all from the same schools, and some were not necessarily students or social service practitioners, but they had connections to friends with social work and service backgrounds.

This study used convenience sampling, a non-probability method where units are selected based on their availability to the researcher (Simkus 2022). We were fully aware of the limitations of convenience sampling (Farrokhi and Mahmoudi-Hamidabad 2012). For example, participants are not randomly selected, and those who are easily accessible may not represent the entire population accurately. As the sample may not represent the broader population, the generalizability of findings is limited. Despite these limitations, convenience sampling remains valuable for exploratory research, pilot studies, and situations where practical constraints prevent more rigorous sampling methods.

Convenience sampling is commonly used in social service settings, as social service research often faces practical limitations such as time constraints or logistical challenges (Holosko et al. 2013; Thyer et al. 2003). Researchers sometimes use convenience sampling for pilot testing, gathering data from a small sample to refine the instruments or programs before conducting a more extensive study. A more fundamental reason for using convenience sampling in our case study were the social circumstances in which we undertook our study. The digital human library project in Hong Kong was initiated during the COVID-19 pandemic, a period marked by prolonged social distancing and limited social contact (Chan and Au-yeung 2021; Ku et al. 2021). Initially conceived as a service rather than a rigorous experiment, the project aimed to enhance understanding of and reduce prejudices against marginalized groups. Between January 2019 and August 2022, we conducted a mix of limited face-to-face, blended, and online sessions. Frequently, our program sessions were fragmented and disrupted due to sudden changes in arrangements.

Given this background, we chose a period where we had relatively stable program implementation and more intact data collection. We identified the period between September 2019 and August 2021, during which we ran three modes of social contact in parallel: situation A—participants read brief story abstracts online, met human books face-to-face for less than 30 min in medium-sized groups (20–30 participants), and engaged in story retelling and editorial activities (about 45 min) in order to develop articles about the protagonists further. Situation B—participants met human books face-to-face for about 90 min in medium-sized groups (20–30 participants). The process involved extended personal sharing from the human books and question-and-answer interactions. Although they did not have digitally mediated contact, public participants in this group had the longest face-to-face contact time with the story protagonists of all of the three groups. Situation C—participants occasionally read the stories on the project website over a 3-month period; they did not have any face-to-face or online dialogues with the story protagonists.

The research team selected three groups of participants from these three contrasting contact situations: group A (from situation A), group B (from situation B), and group C (from situation C). Inclusion criteria included the following: (i) participants were above 18 years old; (ii) the participants participated in the activities between September 2019 and August 2021, when the three modes of social contact were run almost in parallel; (iii) they participated in one activity type (A, B, or C); and (iv) they completed both pre-tests and post-tests. Exclusion criteria included the following: (i) participants who were story characters on our project website and (ii) participants who engaged in more than one activity type during that period.

Based on these criteria, we eventually formed three groups of participants for comparison: group A (45 participants), group B (58 participants), and group C (147 participants). We examined whether these three groups were similar enough for meaningful comparisons. To assess gender differences, we employed the chi-square test, a suitable method for analyzing nominal or categorical variables like gender. This test revealed no significant gender differences among group A (45 participants: 24 females, 21 males), group B (58 participants: 43 females, 15 males)or group C (147 participants: 82 females, 65 males, 13 unknown), with χ2 = 5.07 and p = 0.08. For evaluating age differences, we used one-way ANOVA, an effective technique for numerical variables. This analysis showed no significant age differences between the groups (group A average age 33.6, SD = 9.3; group B average age 30.4, SD = 8.4; group C average age 34.1, SD = 11.9), with F(2, 243) = 2.44, p = 0.09, and η² = 0.02. These statistical approaches indicated that the groups were not unevenly matched in terms of gender and age distributions at the start of the study, ensuring a fair and balanced comparison across the groups.

In brief, we compared three social contact situations based on the structured activities of the project and identified eligible participants from these contexts. These participants had similar backgrounds and motivations and engaged in one of the three modes of social contact with the human books.

6.3. The Human Books Contacted by the Participants Covered in This Study

The human books were people with one or more of the following marginalized group identities: ethnic minorities, people with special education needs, transgender people, people with physical disabilities, and people with mental illness. These groups face separate challenges but also have a collective social identity. Within the project context, they knew they were human books, which helped them talk about their stories on behalf of the project. They received only an introductory briefing, which included information about the purpose of the sharing session, time limits, and the nature of the audience, but no structured training. In addition, the structures of their story pages were consistent, their offline workshops followed compatible formats, and all of the individuals from marginalized groups addressed the same theme in their sharing: a key challenge in life and how to overcome it. As such, in the context of this study, we take these story protagonists to be members representing the same set. The short profiles of the main characters in this study are as follows:

Mimi, born as a male in a conservative era, conformed to societal expectations by marrying and having a family. Deep down, she always felt like a woman. In her fifties, Mimi discovered she had gender identity disorder. This revelation led to the end of her marriage and the loss of her job. Embracing her true self, Mimi underwent sex reassignment surgery to become the woman she always knew she was (https://lib2022.humans.asia/web/book-intro.php?book_id=158 (accessed on 6 May 2024)).

Liu Mian, transformed from a healthy young man to a person with quadriplegia due to an accident, chose singing as his new path. Despite initial struggles, he improved through practice and became vice chairman of the Association for the Physically Handicapped. His journey led him to confidently participate in public singing events and a beauty contest, highlighting his inner beauty (https://lib2022.humans.asia/web/book-intro.php?book_id=155 (accessed on 6 May 2024)).

Jeffrey Andrews, of Indian–Hong Kong descent, is Hong Kong’s first social worker of Indian heritage. Despite limited Chinese education, resulting in poor writing skills, he overcame early legal troubles and pursued education to become a social worker. Now, he works with Christian Action, primarily aiding refugees in Hong Kong (https://lib2022.humans.asia/web/book-intro.php?book_id=14 (accessed on 6 May 2024)).

6.4. Measuring Perspective-Taking

This study focused on the concept of a perspective-taking attitude (PTA). The choice of focusing on PTA in this study was based on several reasons. First, PTA is negatively correlated with prejudice formation (Shih et al. 2009; Vescio et al. 2003) and enhancing PTA is supposed to help reduce prejudice. Second, the project purposively avoided wordings on the project website or questionnaire that implicitly demonized the public or victimized individuals belonging to marginalized groups (the human books). Reader participants were recruited simply because they were curious and wanted to know more; it was at no point suggested that they were to be in an experiment testing the level of prejudice. This decision followed studies indicating that asking direct questions about whether or not an individual holds a prejudice could be counterproductive because people tend to answer such questions following concepts of political correctness (Iyengar and Hahn 2009). Finally, PTA is a construct that is directly relevant to the theoretical base of the project and which emphasizes intrapersonal and interpersonal dialogues. Under such considerations, this study measured a critical construct that does not sound deficient or negative yet is proven to be closely correlated to prejudice.

In this study, PTA was measured by the perspective-taking attitude (PTA) sub-scale (consisting of seven rating questions) of the Interpersonal Reactivity Index (Davis 1980). The Chinese version of the index (Siu and Shek 2005) exhibited acceptable psychometric properties among Chinese participants in Hong Kong (Cronbach’s Alpha 0.7) (Siu and Shek 2005). A higher score indicates a stronger tendency to consider other people’s perspectives. The questions involved in the scale are listed in

Table 1. The questions asked in the PTA scale.

(1) I sometimes find it difficult to see things from the “other guy’s” point of view (reverse item); (2) I try to look at everybody’s side of a disagreement before I make a decision; (3) I sometimes try to understand my friends better by imagining how things look from their perspective; (4) If I am sure I am right about something, I do not waste much time listening to other people’s arguments (reverse item); (5) I believe that there are two sides to every question and try to look at them both; (6) When I am upset at someone, I usually try to “put myself in his shoes” for a while; (7) Before criticizing somebody, I try to imagine how I would feel if I were in their place.

.

The response set was configured on a 10-point scale where “1 = extremely disagree” and “10 = extremely agree.” These were adjusted to fit participants’ frames of reference and enhance instrument reliability (Coelho and Esteves 2007). The targeted participants were mostly Hong Kong citizens who had been educated in the local school system, which uses “100 marks” to represent full marks for an assignment. Therefore, all participants were assumed to be familiar with percentile-based assessment frameworks. As such, a 10-point scale was assumed to be mostly consistent with participants’ prior experiences and preferable to other numeric scale ranges.

7. Results

Group A (n = 45) consisted of participants who read brief story abstracts online, met the human books face-to-face, and had retelling and editorial activities that helped further develop those articles; group B (n = 58) consisted of participants who only met the human books face-to-face in extended sharing sessions, and group C (n = 147) consisted of participants who only read the stories on the project website.

At baseline, the perspective-taking scores of the three groups were similar: the one-way ANOVA at the baseline time point indicated no significant differences between the three groups, PTA: F(2, 247) = 1.27, p = 0.28, η² = 0.01 (see

Table 2. Differences of PTA scores among the three groups (A, n = 45; B, n = 58; C, n = 147) at baseline (T1).

Group	M (T1)	SD	df (between, within)	F	p	η2
A	7.09	1.09	2, 247	1.27	0.28	0.01
B	7.12	0.88
C	6.88	1.17

).

At post-activity time points the groups under study demonstrated notable and significant differences in their outcomes, as detailed in

Table 3. Differences of PTA scores among the three groups (A, n = 45; B, n = 58; C, n = 147) at post-activity (T2).

Group	M (T2)	SD	df (between, within)	F	p	η2
A	7.56	1.14	2, 247	6.80	0.001	0.05
B	7.23	0.99
C	6.88	1.17

below:

The PTA score of group A was significantly higher than that of either of the two comparison groups with medium effect size (A = 7.56, B = 7.23, C = 6.88, p = 0.001, η² = 0.05). This finding shows that group A participants had a stronger willingness to consider other people’s perspectives (i.e., stronger cognitive empathy) than the other two groups.

8. Interpreting the Results

Results indicate that participants in group A demonstrated better psychological outcomes (i.e., more willingness to take other people’s perspectives) than participants who only occasionally read the project website or only contacted the story protagonists face-to-face. Group A participants read brief story abstracts online, met human books face-to-face for a relatively shorter period, and spent longer in story retelling and editorial activities.

As mentioned at the start of this article, contact theory focuses on reducing prejudice through equal-status group contact but has limitations, such as impracticality and inconsistent outcomes. Some research suggests that face-to-face interactions can even reinforce power imbalances. Findings from this study show that just face-to-face contact (group B) was not the most effective in changing attitudes among the three groups.

Likewise, merely reading stories online does not appear to reduce prejudice. Psychological studies have long demonstrated the phenomenon of confirmation bias, which refers to a cognitive bias that leads a person to misinterpret new information as supporting previously held hypotheses (Nickerson 1998). Confirmation bias is easily activated when people must assess correlations and make judgments (Rabin and Schrag 1999). Learners with insufficient prior knowledge to provide internal guidance tend to follow their limited (or even biased) preconceptions in their reading (Kirschner et al. 2006). In other words, when participants read website materials alone, they may select and process only the information supporting their current convictions. Such perspectives align with this study’s finding that participants who only read website materials (group C) did not indicate attitude changes.

These outcomes result from specific forms of face-to-face sharing and story-retelling activities. Therefore, we do not and cannot generalize these findings to other human library activities. However, by comparing these outcomes from similar participants in contrasting settings over the same period, we reasonably speculate that structured narrative activities based on live stories are more pivotal in reducing prejudice than face-to-face contact or the amount of information.

9. Limitations

This case study has limitations. First, participants were not randomly assigned. Second, the story protagonists were from diverse backgrounds, and they were also not randomly assigned. Third, the psychological measure only addressed participants’ perspective-taking but did not measure their acceptance of specific marginalized conditions. Fourth, it is likely that the public participants were generally not heavily biased or hateful but curious or seeking engagement. Our project engaged participants through NGOs, colleges, and social media, employing a recruitment strategy that, while effective in snowballing engagement, unavoidably introduced a selection bias toward individuals already somewhat open to engaging with marginalized narratives. Notably, participants who joined the activities that focused on proactive narrative activities may be interested in learning about marginalized groups or even having a personal connection with the story protagonists. These interests and intentions might have influenced how we interpret the outcomes. These limitations notwithstanding, it is possible to suggest various forms of future project development and research.

10. Implications for Research and Practice

This paper contends that public participants’ attitude changes were induced more by their own reflective and narrative activities than their face-to-face contact with the story protagonists. Among participants in the three groups, group A participants did not have the longest contact/conversation time with the story protagonists, but they did have much more time to retell, summarize, compare ideas, and suggest how to develop the articles about those story protagonists. These intrapersonal and interpersonal dialogues among the participants appear to have changed their attitudes. Such observations inform research and practice in several aspects:

10.1. Narrative Activities

Instead of prioritizing face-to-face contact, we should focus more on designing narrative activities that explore live stories. Using genuine face-to-face contact to reduce social prejudice, as conventional HL practices proposed, seems like a good idea, but it is not easy to achieve. One reason is the size difference between marginalized minorities and the majority group, making real interaction challenging. Additionally, the daily routines and limited time of marginalized groups, due to such things as long working hours, make face-to-face meetings difficult, and the time cost for these meetings might be too high. Furthermore, in places like Hong Kong, where the majority speaks Chinese and minorities might speak different languages, language barriers can pose significant challenges during such authentic interactions. To effectively tackle social prejudices against marginalized groups, looking beyond simple physical contact between groups may be essential. Leveraging participants’ own narrative activities may be a viable option.

The project demonstrates that engaging participants with stories from marginalized groups can enhance their ability to adopt the perspectives of others. This finding is crucial for developing a perspective-taking attitude and reducing prejudices. It is possible to view understanding as an internal dialogue, and that prejudice is a result of that dialogue being stopped, whereas storytelling can activate dialogues. The results of this study partly illustrate this theoretical position. Writing or narrating is a dialogue per se; a narrator enacts the dialogue with imagined readers, carries on the dialogue with themselves, and the eventual circulation of texts will also speak to actual readers and produce interpretation or feedback. The discussion and editorial activities required participants to discuss different ideas and to rewrite story contents or titles based on the human books’ first-person sharing. The format of the activity discussion required participants to address perspective differences in a relatively safe space. As they were encouraged to brainstorm and revise their ideas, such retelling activities allowed participants to play with different voices. When participants knew that they could develop different versions of the same story they thereby became aware of the significance of different perspectives. This implicit deconstruction process embedded in narrative processes supports a pedagogy that may reduce prejudice, and such observations echo the results of earlier studies using similar methods (Chan 2019a, 2019b). This insight opens avenues for future research into the specific elements of narrative activities that most effectively influence attitudes and suggests practices that prioritize reflective storytelling processes.

10.2. Resources That Enhance Understanding

Instead of focusing solely on social interactions, we should explore resources that enhance interactions and interpretations. Digital resources, for instance, can facilitate internal dialogues. The widespread use of smartphones has made an offline to online (O2O) strategy more feasible. For example, participants can search online for additional information about the protagonists before or during their meetings. Online articles and images help readers recall the story and elaborate on their ideas. This inference is supported by recent studies, which have shown that participants achieve better outcomes in blended learning experiences than in either fully online or fully face-to-face learning experiences (Ayala 2009; Wang 2010).

Additionally, using photos can significantly enhance reflective thinking. Visual stimuli, such as images and photos, can evoke emotional responses and personal connections, making the learning experience more impactful. Photos can serve as powerful storytelling tools that help participants visualize and relate to the experiences of marginalized groups, fostering perspective-taking and understanding. When participants engage with these images, they are prompted to think critically about the context and content, leading to deeper reflection. Research has shown that incorporating visual elements into learning can aid memory retention and comprehension (Chan et al. 2012; DeCoster and Dickerson 2014). In the context of this project, photos not only help participants recall specific details of the stories but also encourage them to explore the underlying themes and messages. By using visual aids, participants can better grasp complex concepts and see different perspectives, which enhances their ability to analyze and interpret the information.

10.3. Connecting Communities

Instead of relying heavily on members of marginalized groups to advocate for themselves, we should also focus on engaging participants who are interested in or supportive of these groups.

An online library can construct a context that defines its members and users, thereby creating a communal context. Recent research studies have generally noted that online communities can shape members’ imagined audiences and self-perceptions (Chan 2006; Dahya 2017; Kedzior and Allen 2016). An online library of marginalized groups provides a reference by which to determine who was speaking to whom; therefore, when the public participants were involved in activities organized by that library, they were also recruited to join a community practice that aligned with the interests of marginalized groups. These inferential connections suggest that future research could focus on optimizing digital platforms for reflective learning and on exploring how online communities influence individual and collective understanding, with implications for educational and social practices.

By highlighting stories of marginalized groups, the project fosters a sense of community and inclusivity within educational settings. Participants learn the value of diversity and the importance of including all voices in societal dialogues, which can translate into more inclusive school environments. Our recruitment strategy likely attracts participants who are already interested in learning about or have a personal connection to marginalized groups, which could be viewed more as an opportunity. The transition from opposition to support is a continuous journey, and different people’s levels of support or opposition also exist along the spectrum. Beginning with individuals who are at least somewhat open to marginalized voices could be a logical and effective strategy. This approach provides a solid foundation from which to expand support more broadly. Such a library may support marginalized groups by uniting like-minded participants and nudging individuals from casual interest to more resounding support. Further research is necessary to explore the effects on participants who do or do not share similar experiences with the human books. It is also helpful to expand the testing of this storytelling library to include specific groups with clear stances, such as religious communities or those from various ethnic backgrounds. This will deepen our understanding, bridge engagement gaps, and promote inclusive conversations.

11. Conclusions

We began our exploration with a project that was inspired by digital storytelling (Chan and Sage 2021; Meyerhofer-Parra et al. 2024) and human library practice (Lam et al. 2023), cumulating in an online human library focused on marginalized groups. Based on the observations from this case study, this paper argues that, rather than authentic contact or the amount of information, the dialogic cognitive processes embedded in narrative activities may explain the ontology, epistemology, and methodology of prejudice reduction. Additionally, it is suggested that storytelling, enhanced by media resources, may help strengthen these dialogic processes.

Referring back to the discussion on HL and DST (Chan and Sage 2021; Lam et al. 2023; Meyerhofer-Parra et al. 2024), it seems that a constructive discussion should not focus on debating the effectiveness of HL or DST, nor on looking for “orthodox” HL or DST. Instead, we may posit these practice concepts as inspirations, examining how specific technology-enhanced storytelling activities using authentic live stories (human books) can result in specific outcomes. By focusing on the dialogic processes and narrative activities embedded in these approaches, we may explore new methodologies and create impactful storytelling experiences that address and reduce prejudice, foster empathy, and highlight the voices of marginalized groups.

As previously mentioned, this case study has limitations. Its use of convenience sampling poses limitations that could impact the findings’ generalizability. This method may introduce selection bias, as those who participated might already be more open to engaging with marginalized groups, skewing the results towards more positive outcomes. Thus, while the study provides some insights, its case study nature necessitates caution in interpreting and applying its conclusions to broader contexts.

Despite these limitations, it is possible to propose future project developments and research directions. We have formulated speculations that may hold validity based on our observations. One idea suggested by the findings of this case study is to develop and examine an online storytelling library to facilitate narrative activities that foster understanding and empathy toward marginalized groups. Rather than merely conceptualizing the library as a metaphor, it is worthwhile to consider creating a fully operational digital library by, with, and for marginalized groups. This library could be dynamic, incorporating features that encourage interaction and narrative activities.

Instead of relying solely on face-to-face interactions with marginalized groups to correct public misconceptions, it is suggested that we can have a library provide reading resources and narrative activities that help the public re-examine their preconceptions and attitudes. The proposed model is neither a moving-image-based digital production nor a static archive of stories. Instead, it emphasizes a dynamic storytelling process enriched with digitally mediated resources, blending storytelling’s fluidity with digital media’s interactive capabilities to create a comprehensive and engaging experience.

For individuals harboring prejudices toward marginalized groups, such a library could serve as a transformative tool, offering alternative perspectives and supporting NGOs that work with marginalized groups by promoting their activities and resources. For those curious about marginalized groups, the library could facilitate involvement in volunteer work or employment opportunities focused on serving these groups. This includes tasks like reporting, rewriting, volunteering, other support activities, and continuing the active participation vital for fostering perspective-taking.

This library could also support diverse narrative activities such as inquiry-based learning, project-based learning, case studies, and class presentations in educational settings. As students often find real-life stories more engaging than textual data, teachers could utilize these online narratives in various learning activities, enhancing lessons with quality references, interactive quizzes, and links to relevant organizations.

Furthermore, while commercial resources exist that compile notable historical figures’ biographies, a similar repository exists for stories of disadvantaged groups. Such a platform would provide a vast and varied selection of materials for educational use, accessible whenever suitable.

In conclusion, the digital era offers new storytelling and story engagement possibilities. Today, public participants can interact with narratives like never before—searching for information, viewing photos, or revisiting video segments—either remotely or in blended settings, all via smartphones. This accessibility allows involvement from any distance, a capability unheard of in the pre-digital age. These advancements suggest the potential for more active engagement with stories from marginalized groups. However, realizing such an infrastructure demands substantial resources, thorough evaluation, and a strategic approach. Continued dialogue, debate, and research are essential to refine and implement this model effectively.