Introduction

Online dating has emerged as a predominant method for establishing interpersonal relationships, with dating applications pioneering a paradigm shift through the integration of synchronous and asynchronous communication modalities. Contrary to conventional dating websites, these applications markedly reduce the time between online interactions and offline meetings by leveraging mobile geolocation features, thereby enabling users to connect with nearby individuals (Rochat et al., 2019). Apps such as Tinder have accumulated over 50 million users worldwide, with at least 10 million daily active users (Freier, 2015). The accelerated integration of these technologies into interpersonal interactions signifies a paradigm shift in how relationships are initiated, thereby challenging established models of relationship development.

Conventional relationship models, exemplified by Knapp et al.’s (2014) two-step framework, delineate relationship progression as a gradual process of escalation and de-escalation predicated on social exchange principles (Fox et al., 2013). However, the advent of dating apps has been shown to expedite this process by modifying user behavior in partner selection and early-stage interactions (Wu and Trottier, 2022). Users commonly tend to fulfill relational needs with minimal effort, leading to interactions that are rapid but potentially shallow. This structural change has sparked academic interest in how dating apps influence socialization. Studies have found differences in usage patterns across demographic groups, including gender, marital status, and psychological factors (Rochat et al., 2019).

The user-friendly design of dating applications, which facilitates rapid access to potential partners, may lead to problematic usage patterns (Orosz et al., 2018). Internet addiction, originally conceptualized using the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders criteria for substance abuse (Armstrong et al., 2000), has been linked to psychological factors such as low self-esteem, boredom, and loneliness (Fioravanti et al., 2012; Kim et al., 2009). Further research indicates that males have a higher prevalence of internet addiction than females (Çelik and Odacı, 2013; Shotton, 1991), suggesting that gender differences play a significant role in the motivations for using dating applications. For example, males have been found to prioritize casual encounters, while females often use applications to build social connections or seek self-validation (Ranzini and Lutz, 2017). These differences, combined with design elements that promote compulsive utilization, necessitate further examination of the underlying factors that predispose individuals to dating app dependency, particularly in the context of demographic and psychological variables.

The motivation for utilizing dating applications has been identified as a significant predictor of problematic use. However, there is a paucity of empirical research examining how design-level user experience (UX) factors translate this motivation into addiction. Guided by Brand et al.’s Interaction of Person–Affect–Cognition–Execution (I-PACE) model (Brand et al., 2019), Griffiths’ six-component syndrome of behavioral addiction (Griffiths, 2005), and classical human–computer interaction (HCI) theory (Fogg et al., 2002), this study categorizes dating apps and their UX properties. Ultimately, a three-level UX framework was constructed, including personal factors (casual sex, self-fulfillment, curiosity, identity, loneliness), social factors (traditional ethics, social barriers), and functional UX factors (self-optimality, usability, selectivity, agglomerativity, difference of initiative), and tested through regression analysis to determine its explanatory power in converting psychological tendencies into compulsive participation.

The application classification method and comprehensive model established based on UX differences not only reveal the quantitative impact of specific design features on user behavior but also fill the theoretical gap in existing research regarding the associative mechanisms between UX factors and addictive behavior. The academic value of this theoretical framework lies in establishing a complete explanatory pathway linking design features, psychological mechanisms, and addictive behavior, providing a diagnostic tool for auditing existing dating apps, and offering a roadmap for preventive regulation, ethical design, and personalized digital health interventions. These interventions aim to ensure user safety while maintaining platform sustainability. Future research could adapt this framework to suit emerging AI-driven recommendation environments and dynamic interfaces.

This study consists of three parts. Firstly, this study classified existing dating apps into two broad categories based on differences in UX and summarized functional experience factors. Secondly, this study attempted to develop a complete UX model for dating apps. Thirdly, the developed UX model is utilized to pinpoint the pivotal factors that heighten the propensity for addiction across various categories of dating apps. Subsequently, an exploration of potential causal mechanisms underpinning addiction to distinct types of dating apps is undertaken.

Dating app category and theoretical framework

This study classifies dating apps into location- and group-based ecosystems, then weaves functional, personal, and social UX drivers into an integrated theoretical scaffold linking the I-PACE addiction cycle, Griffiths’ symptom criteria, and HCI design heuristics, thereby constructing a 12-factor model to guide subsequent analyses.

UX category

Smartphone apps, represented by the dating apps, offer enhanced portability and accessibility, as well as geolocation capabilities, compared to “traditional” online dating sites (Schrock, 2015). The opportunity to meet and connect with other people provided by a dating app may be closely related to its user’s threshold of dating experience. As dating apps have evolved, they have gradually differentiated into different types with a focus on experiential features. This study divides dating apps into location-based dating apps, which emphasize geolocation and regional scope, and group-based dating apps, which emphasize social group categories such as job type, education, hobbies, and personality type.

Location-based dating apps are strongly associated with geo-location and regional scope, and representative apps include Tinder, Tantan, and Aloha. Including the “Find Nearby” function in some social networking apps is also a kind of location-based dating app. Upon registration, users are prompted to create an account by uploading a photo, optionally providing a brief description, and specifying search preferences such as age, gender, and geographic distance. Subsequently, the location-based dating apps allow users to quickly and anonymously match individuals within a specified radius of their geographic location.

The app then displays the profiles of individuals close to the participating users. If both users indicate a positive interest by swiping on each other’s profiles, they are designated as a “match,” thereby gaining chat access and potentially initiating interactions with the prospect of pursuing either a short-term or long-term relationship, as detailed by Rochat et al. (2019). At the core of Tinder or Grindr, for instance, is this mechanism, which utilizes geographic distance between users as a critical variable according to which possible partners can be found, reinforcing the notion that location-based dating apps are conducive to casual sex and transient relationships (Ranzini and Lutz, 2017).

Group-based dating apps focus on dividing people from different levels, with representative apps including Soul, Tashuo, QingtengLove, and Blue. Additionally, some dating apps based on dating websites are group-based. For instance, Grindr promotes a sense of community and positions itself as a social networking app for people with particular interests and sexual orientations (Wu and Trottier, 2022). Research suggests that participants are more likely to maximize action when matched with group members (Chen and Li, 2009). Individuals seek membership in groups based on hobbies, personality, and character, as simply identifying with a group of people who work or study in real life may not provide optimal uniqueness or sufficient interpersonal connections.

Group-based dating apps offer a range of features that require users to verify the authenticity of the information they provide, within the limits of the law in order to realize the different levels of categorization of groups of people. In China, for example, users on Tashuo can upload documents for verification if they fill in their graduation school or degree. The verified information has a unique logo. Meanwhile, studies have demonstrated that individuals engaging in dating apps are generally reluctant to engage in deliberate deception with their dating app counterparts (Zytko et al., 2014). This reluctance stems from believing that any falsehoods would likely be exposed upon subsequent face-to-face encounters. Consequently, users are more likely to perceive group-based dating apps as having greater reliability.

A critical distinction is drawn between location-based applications, which facilitate geosocial interaction, and group-based applications, which are characterized by identity-based communities. This classification provides a novel empirical framework for observing diverse UX configurations, thus activating different addiction pathways. Location-based applications accelerate the Person–Affect–Execution loop through the provision of instant proximity rewards, whereas group-based applications amplify identity validation and social belonging (Wang et al., 2025), thereby prolonging mood regulation and the addiction process.

It is worth noting that these two typological divisions represent differences in the underlying logic and UX of the app’s operation but not in absolute functionality. For instance, linking Tinder accounts to external platforms like Facebook and third-party services like Instagram and Spotify requires verifiability. This regulation extends to self-presentation, exemplified by the limitation that Tinder users can only choose profile pictures from their Facebook accounts. Consequently, this adherence to authenticity enhances the appeal of location-based dating apps as tools for seeking potential romantic partners in close proximity (Timmermans and Courtois, 2018). In contrast, most group-based dating apps also have geo-location features. However, they only allow users to see the approximate geographic location of the city and district to filter partners in the same city and region, rather than having precise locations like location-based dating apps.

Factors of UX in dating apps and the theoretical framework foundation

The availability of dating apps can shape users’ needs, generating new and unique satisfactions. The existing body of literature pertaining to traditional media, social media, and dating sites underscores the presence of motivations that are common across these platforms (Sundar and Limperos, 2013), as well as motivations that are distinct and specific to individual platforms (Ryan et al., 2014). According to use and gratification studies, users will continue to use a medium as long as it fulfills their needs (Ruggiero, 2000). Adopting new technological approaches and channels is driven by intent, and external factors may influence these perceptions (Wang et al, 2025). Accordingly, researchers have not only considered function and personal factors, but have also referred to social and environmental factors.

The majority of existing research literature analyzes the motivations for using Tinder, such as Table 1, which lists some research results. Table 1 shows that sexuality, socialization, self-validation, and entertainment are among the more commonly identified motivations for using Tinder. There is a certain amount of overlap between the motivations for using Tinder and the motivations for using other dating apps. Therefore, there is a need to find some commonality in the motivation studies based on dating apps. Although these findings can be used to assess the motivational references behind dating app use, they do not have strong psychometric properties or predictive properties of addiction in terms of factor structure.

Table 1 Analysis of motivations for using Tinder.
Full size table

The objective of this study is to establish a comprehensive model that describes users’ experiences on dating applications. To this end, the study emphasizes and summarizes functional experience factors, which are then combined with personal and social factors. These personal and social factors are summarized in Table 1 and proposed by Castro and Barrada (2020) based on a review of existing literature as the main basic elements. Therefore, this study uses theoretical triangulation to strengthen the theoretical foundation and identify the sub-factors of these three main elements. This can contribute to understanding users’ experiences with dating apps, as illustrated in Fig. 1.

Fig. 1
Fig. 1
Full size image

Cross-framework mapping.

To be specific, Fig. 1 is based on the I-PACE model of internet use disorder, which posits that addictive behavior is caused by the interaction of individual tendencies, emotional and cognitive responses to triggers, and reduced executive control (Brand et al., 2019). Integrating UX factors serves to expand the human-centered insights of the I-PACE model by incorporating design and contextual elements that have the potential to shape emotional and cognitive responses. Secondly, the model aligns with Griffiths’ behavioral addiction component model, which explores core addiction criteria-including salience, mood modification, tolerance, withdrawal, conflict, and relapse-in the specific context of dating app use (Griffiths, 2005). This ensures that the UX factors examined in this study correspond to explicit manifestations of addictive behavior. Thirdly, the developed framework integrates principles from HCI and UX theory, including enhancing usability and engagement, along with carefully designed persuasive design strategies (Fogg et al., 2002), which can significantly influence user behavior. The integration of psychological addiction models with UX design principles establishes a robust interdisciplinary foundation, building on prior theory while highlighting innovative contributions linking UX design to addiction behavior outcomes.

The research model was constructed and guided by the context of dating app research and insights from these three complementary theoretical frameworks. As shown in Fig. 2, the model proposed here includes 12 factors identified by the research theory to represent users’ overall experience with dating applications.

Fig. 2
Fig. 2
Full size image

UX model for dating apps.

Specifically, Fig. 2 illustrates the environment in which users engage with dating apps, highlighting the interactions among users, other users, app content, and the social environment. This theoretical triangular model was developed by mapping each UX sub-variable to components of three theoretical frameworks. The I-PACE model-guided research considers factors such as personality, contextual design, and cognitive–emotional states. Griffiths’ components ensure these factors correspond to established addiction symptoms, and HCI/UX principles emphasize design practices that promote them. Table 2 summarizes the correspondence between each sub-variable in the model and the I-PACE categories, Griffiths’ addiction components, and UX design principles. It is also a reinterpretation of Fig. 1. This integrated approach strengthens the model’s theoretical foundation and relevance, enabling identification of factors leading to dating app addiction and explanation of its causes by demonstrating connections between these factors, psychological addiction mechanisms, and design features that influence behavior.

Table 2 Sub-variable theoretical relationship.
Full size table

Developing a UX model for dating apps

Based on the UX model for dating apps, motivational factors at the functional, personal, and social levels are introduced separately. Together, these three levels outline how design, psychology, and context intertwine to shape user engagement and addictive behavior in dating apps.

Functional experience factor

Self-optimality

Self-optimality refers to a user’s ability to create an ideal digital self by selectively disclosing information and managing their image strategically. This concept integrates impression-management theory (Goffman, 1949), hyperpersonal communication (Walther, 1996), and the empirical finding that online daters present their “best possible selves” while omitting undesirable traits (Bargh et al., 2002; McKenna et al., 2002). Swipe-based platforms provide malleable contexts that invite meticulous self-presentation through profile templates, image filters, and editable bios (Whitty, 2008; Ellison et al., 2012). Tinder, for instance, demonstrates that users craft aspirational identity collages, thereby signaling desirability (MacKee, 2016). Apps for niche communities encourage emphasizing culturally valued attributes (Wu and Trottier, 2022).

In I-PACE terms, curating an enhanced self elicits positive affect (ego boost) and reinforcing cognitions (“I am desirable”), fostering repeated execution of profile tweaking. Griffiths’ mood-modification and salience components emerge when users seek “match feedback” to validate their optimized persona. Authenticity constraints and anticipated offline meetings temper exaggeration, yet cross-cultural norms modulate how far users push idealization. In image-centric subcultures, hyper-curation is expected. The construct thus captures the design-enabled optimization loop by which users continuously refine self-presentation, receive intermittent social rewards, and integrating UX, media-psychology motives, and behavioral-addiction processes within a single, theory-grounded factor.

Usability

Usability refers to the efficiency, ease of learning, and satisfaction provided by a dating app’s interface. It incorporates ISO 9241 usability criteria (effectiveness, efficiency, and satisfaction) (Heinold et al., 2025) and Nielsen’s (1994) heuristics, which emphasize reducing cognitive load and minimal interaction cost in swipe-based environments. Design choices such as one-tap registration, left/right swipe gestures, icon-based menus, and real-time match notifications drastically reduce friction (Couch and Liamputtong, 2008; Wiederhold, 2015). Location auto-fill and social media sign-ins further streamline the onboarding process, while haptic or visual feedback confirms each interaction, sustaining a rapid “swipe–reward” cadence (Licoppe et al., 2016). Within I-PACE, high usability lowers the execution threshold, making approach behavior almost automatic. Quick feedback elicits positive emotions and the expectation of further rewards. According to Griffiths, this ease contributes to salience (habitual checking) and mood modification (microbursts of pleasure with each match).

Selectivity

Selectivity refers to a dating app’s ability to expose users to a vast pool of potential partners for filtering, extending choice architecture theory (Iyengar and Lepper, 2000) and online matching research (Finkel et al., 2012). Selectivity captures the abundance of profiles and the precision with which users can accept, reject, or filter matches (Thomas et al., 2022). Features such as swipe queues, distance sliders, age filters, and interest tags allow users to quickly sort through hundreds of profiles per session. Algorithmic ranking continuously reshuffles candidates, providing a dynamic sense of limitless choice (Lefebvre, 2018). Premium boosts and rewind features further enhance control over selection, reinforcing the perception of an inexhaustible potential partner pool.

In I-PACE terms, abundant selectivity functions as a situational cue that heightens cognitive expectations (“there’s always someone better”) and affective arousal, fueling repeated execution (swiping). According to Griffiths’s framework, intermittent ideal matches yield variable-ratio rewards, fostering tolerance (the need for more swipes) and salience (preoccupation with optimizing choice). However, overabundance may also trigger conflict via decision fatigue and fear of missing out, echoing the effects of “choice overload” (Illouz, 2012). While selectivity democratizes partner access, it can also lead to compulsive swiping and decreased satisfaction.

Agglomerativity

Agglomerativity is defined as the process of platform-driven clustering of users into affinity-based niches. This concept can be understood as an online-dating extension of homophily (McPherson et al., 2001) and in-group identity theory (Gaertner and Dovidio, 2014). This phenomenon highlights the role that profile filters, interest tags, and niche platforms play in fostering communities of perceived similarity, thereby enhancing perceived safety and mutual attraction among members. The phenomenon of shared identity has been demonstrated to reduce intergroup anxiety and potentially mitigate the occurrence of discrimination (Dovidio et al., 2000). Within the I-PACE framework, agglomerative affordances function as situational cues that evoke positive affect (i.e., feelings of being understood) and cognitions (i.e., trust), thereby encouraging continuous use. Conversely, the potential for limitless intra-cluster exploration, facilitated by filters and scroll functions, may contribute to the cultivation of tolerance and the exacerbation of relapse through the dynamics of variable-ratio social rewards.

Design elements algorithmically deliver “people like me,” thereby producing perceptual congruence in values, lifestyle, or orientation. The evolution of Blued from a hookup tool to a more expansive gay social network (Miao and Chan, 2020) exemplifies the potential of agglomerativity to transform a marginalized group’s need for a secure space into long-term engagement. The phenomenon of agglomerativity has the potential to diminish exposure to diversity and perpetuate the formation of echo chambers. This is due to the possibility that users who possess intersectional identities that encompass multiple groups may experience marginalization even within niche applications (Wu and Trottier, 2022).

Difference of initiative

The gender-linked asymmetry in who initiates communication after a match on dating platforms is known as the initiative difference. This concept integrates two established research streams: courtship-script research showing that offline and online heterosexual interactions remain male-initiated and female-selective; and computer-mediated-communication research documenting male proactive efforts and female gate-keeping on swipe-based apps (Zytko et al., 2014; Timmermans and Courtois, 2018). Mutual-match architectures reduce approach costs for men by requiring only a swipe, while notification queues and profile-visibility controls allow women to filter incoming overtures. In contrast, applications such as Bumble reverse the conventional paradigm: heterosexual matches become inactive unless the female participant initiates the interaction, a feature designed to redistribute power and reduce the risk of female harassment (Aljasim and Zytko, 2023; Ganito, 2010).

This goal discrepancy stems from men prioritizing short-term or sexual motivations, whereas women prioritize relational safety (Sumter et al., 2017). Griffiths (2005) suggests that high “pursuit density” combined with variable matching ratio rewards enhances salience and tolerance within the addiction classification framework. The behavior of repeatedly checking new messages aligns with the person–affect–execution loop in I-PACE (Brand et al., 2019). Cultural context moderates initiative: in traditional societies, women initiate less due to reputational risks (Chen and Liu, 2021; Wu and Trottier, 2021).

Personal factors

Casual sex

Casual sex is defined as the motivation to seek short-term, non-committed sexual encounters via dating apps (Fischer, 2007). It adapts the sociosexuality construct (Simpson and Gangestad, 1991) and the sex motive in the Tinder Motives Scale (Orosz et al., 2018) by situating it within the evolutionary psychology literature on short-term mating and the media psychology framework of sexual gratification needs. Geographic location filtering, sexual innuendos in the information, and simplified registration processes reduce the cost and effort required for casual sexual encounters (Licoppe et al., 2016). Swipe mechanics enable rapid partner triage, facilitating opportunistic encounters (Timmermans and Courtois, 2018). In the I-PACE model, the desire for casual sex resides in the person layer as a dispositional or situational motive that heightens the affect of sexual arousal and the expectancy of gratification cognition (Andreassen et al., 2018). Intermittent matching results in successful sexual encounters and maintains excitement, consistent with Griffith’s theory (Meerkerk et al., 2006). However, not all invitations to engage in sexual activity are accepted, which can trigger conflict.

Self-fulfillment

Self-fulfillment is defined as the pursuit of enhanced self-worth, validation, and personal growth through interactions on dating apps. This concept builds upon Maslow’s (2013) idea of self-actualization and incorporates self-affirmation theory (Steele, 1988) as well as the motive of seeking validation of one’s self-worth, as identified in Tinder research (Orosz et al., 2018). Matches, likes, and supportive chats serve as micro-affirmations that increase users’ perceived social value (Blackwell et al., 2015). Continuous swipe queues and real-time like counters provide frequent, low-effort chances for external validation (Ranzini and Lutz, 2017). Push notifications for popularity verification further reinforce the sense of accomplishment. In I-PACE, the self-fulfillment motive falls under the Person domain as a need for esteem. Griffiths’ mood modification manifests as transient boosts in self-esteem, while tolerance emerges when users require greater quantities of likes to maintain the same level of satisfaction. While self-fulfillment can promote confidence and authentic self-expression, relying on external app feedback can lead to cyclical dependency.

Curiosity

Curiosity is the intrinsic desire to seek new information about potential romantic partners and one’s own attractiveness. It incorporates information-gap theory (Loewenstein, 1994) and the “curiosity/entertainment” motive identified in Tinder motives research (Timmermans and Caluwé, 2017). Empirical studies show that many users swipe not to secure dates, but rather to sample the dating “market” and assess their attractiveness (Timmermans et al., 2018). Endless swipe queues and blurred “likes” that are revealed only after reciprocation entice exploration. Gamified elements, such as match animations, discovery mode, and daily “Top Picks,” sustain novelty by continually refreshing unknown profiles. In I-PACE, curiosity acts as a person-level exploratory trait that heightens arousal and anticipatory cognition, aligning with Griffiths’s mood modification (momentary excitement) and fostering salience and tolerance (increasing swipes to satisfy novelty cravings). Although curiosity-driven browsing can satisfy exploratory needs and provide market insight, it may also induce choice overload and distract from relational goals.

Identity

Identity is defined as the pursuit of self-verification and identity expression through dating app profiles and algorithmic matching. This concept synthesizes self-verification theory (Swann, 2012) and research on online identity affirmation. This research shows that displaying one’s authentic traits yields higher relational satisfaction (Ranzini and Lutz, 2017). Algorithms that recommend “best matches” based on shared values or lifestyles reinforce the sense that one’s true self is recognized (Comunello et al., 2021). Structured profile fields, interest tags, and social media integrations enable multidimensional self-expression. Verification badges further cement the coherence between one’s online and offline identities. In I-PACE, identity affirmation functions as a need for self-coherence at the Person level. Within Griffiths’s framework, each validating match modifies mood, and the desire for congruent feedback escalates, fostering tolerance. Thwarted affirmation can generate feelings of withdrawal. While identity affirmation can enhance well-being and foster authentic connections, reliance on algorithmic mirroring can lead to echo-chamber effects and dependency on validation.

Loneliness

Loneliness is defined as the distressing discrepancy between desired and actual social connections. According to the Evolutionary Theory of Loneliness (Cacioppo and Cacioppo, 2018), it distinguishes enduring trait loneliness from transient state loneliness (Roddick and Chen, 2021). Since single adults report the highest emotional loneliness scores (Bucher et al., 2019), they disproportionately use dating apps to reconnect socially (Chin et al., 2019). Specifically, “online now” indicators and instant messaging create a constantly available social space (Coduto et al., 2020). Low-threshold sign-ups and group filters enable immediate perceived proximity. In I-PACE, loneliness operates as a person-level vulnerability that intensifies affective sensitivity to inclusion cues (Castro et al., 2020). Every notification or match provides a momentary social reprieve, which aligns with the mood modification aspect in Griffiths’ framework. A sudden lack of feedback can trigger withdrawal symptoms and prompt relapse behaviors. While dating apps can mitigate loneliness by expanding weak social networks (Green et al., 2001), excessive reliance on them may result in digital interactions supplanting meaningful relationships.

Social factors

Traditional ethics

The concept of traditional ethics can be defined as the set of culturally rooted moral norms that govern behaviors such as premarital sex, gender roles, filial piety, and sexual orientation. These norms significantly influence how individuals adopt and behave on dating applications. Integrating theories of social norms and stigma from media psychology (Raney et al., 2020) and cultural script theory from relationship sociology, this concept explains how traditional ethics shape user actions. In Griffiths’ model, norm-generated urgency has the potential to amplify both salience and conflict. These pressures act as contextual moderators, affecting both affect and cognition, and thereby influencing execution. The reinforcement of usage is achieved through the provision of variable-ratio rewards, which are occasional and culturally suitable matches.

In the realm of technological applications, users often devise strategies to either conform to or navigate around established norms. For instance, Chinese women residing abroad utilize mobile applications to seek international partners, a practice that enables them to evade conventional expectations regarding female chastity (Chen and Liu, 2021). Similarly, homosexual users in conservative regions employ anonymity in their profiles to circumvent the potential disclosure of their sexual orientation (Wu and Trottier, 2021). Furthermore, privacy features and selective filters operationalize responsiveness to traditional ethics. In contexts where societal norms tend to stigmatize certain forms of self-expression, applications such as Blued offer a platform for anonymous interaction, facilitating a more discreet exploration of personal inclinations. Consequently, the design of the system has the capacity to either mitigate or amplify the existing normative friction. Despite the variability of ethical standards across different cultural contexts, the fundamental analytical core of this construct remains consistent.

Social barrier

The dual role of dating applications in both integrating and segregating users’ online dating activities from their offline social circles manifests as a social barrier. The present construct draws on the theoretical frameworks of computer-mediated communication (Bazarova and Choi, 2014) and privacy calculus theory (Najjar et al., 2021) to elucidate the strategic use of app affordances by users. This strategic use can be delineated in two aspects: the expansion of social networks beyond existing ties and the compartmentalization of dating activity from personal or professional contacts. Within the I-PACE framework, these affordances act as situational cues, influencing affect (e.g., security vs. anxiety) and cognition (e.g., perceived privacy). These factors, in turn, influence engagement patterns. The potential for external factors to disrupt these boundaries may result in the onset of compulsive behaviors. This phenomenon aligns with Griffiths’ conflict and withdrawal components.

The direction of boundary management is contingent upon the cultural context and the user’s motives. In collectivist societies, the prevailing social attitudes and norms tend to create a stigma around online dating, thereby reinforcing behaviors that establish barriers to social interaction. Conversely, in liberal contexts, individuals may prioritize the reduction of barriers to facilitate the diversification of their social connections (Clemens et al., 2015; MacKee, 2016). In instances of conservative regional contexts, members of special groups often engage in dual strategies, seeking to establish a sense of community while simultaneously concealing their activities from local networks. Network expansion is facilitated by features such as location filters and mutual-match designs, while blocking tools and incognito modes enable compartmentalization (Licoppe et al., 2016). By enabling dynamic boundary adjustment, dating apps create a distinct “dating domain” that regulates engagement intensity.

Methods

The research methodology framework is outlined, including participant recruitment and screening, scale adaptation and psychometric validation, and the statistical analysis methods used to determine which UX dimensions predict addiction to location- and group-based dating apps. These methods include correlation analysis, exploratory–confirmatory factor analysis, and multiple regression analysis.

Participant recruitment and sample characteristics

From January to April 2024, this study disseminated questionnaires via three recruitment channels: (a) three social media communities related to dating apps for English-speaking users; (b) forum groups and two large social media discussion groups for Asian users; (c) recruitment posts on dating apps, such as Tinder and Tantan. All recruitment efforts adhered to the same eligibility criteria: age of at least 18 years and having used any dating app an average of at least once per week in the past three months. All participants read the Participant Information Sheet and electronically signed the Informed Consent Form, confirming their understanding of the study’s purpose. A two-stage screening process ensured data quality. First, a pre-survey confirmed weekly usage. Second, participants were required to complete three attention tests and a CAPTCHA-style verification to prevent bot participation. The initial link yielded 885 completed questionnaires. After excluding quick responses and incomplete records, 632 questionnaires (71.4% of the original sample) were retained for analysis.

The final sample exhibited a balanced gender distribution (324 males [51%] and 308 females [49%]). The group was predominantly young, with 35.1% aged 18–25 years and 38.9% aged 26–33 years; a combined 74% were under 34 years. This demographic skew aligns with prior research showing that young individuals frequently engage in serial online relationships and navigate overlapping romantic involvements (Cohen et al., 2003). Educational attainment was high, with 71.9% holding at least a bachelor’s degree. Smartphone applications may play a significant role in explaining changes in sexual norms on university campuses (Bersamin et al., 2014). Participants were distributed across 29 countries and regions, with the highest proportions coming from the United States (27%), China (26%), the United Kingdom (18%), and South Korea (15%). This reflects the similarities in UX across different cultural contexts and dating apps. Pilot testing revealed that issues related to ethnicity and household income significantly increased the survey dropout rate. Since the theoretical focus of the study was on UX predictors rather than socioeconomic differences, these factors were excluded. It is worth noting that the questionnaire specifies that gender statistics refer to biological sex.

Measurement variables and models

Firstly, by reviewing relevant literature and existing validated scales, this study developed a measurement instrument tailored to the multi-dimensional behaviors of dating app users—encompassing selection, conversation initiation, and offline meeting arrangements. The developed scales include the Twelve-Factor UX Model Scale and the Dating Apps Addiction Related Scale (DAARS). Table 3 outlines the scale sources, reliability metrics, and item counts. Additionally, 20 long-term dating app users (with an average of 2 years’ experience and in-depth use of at least three platforms) assisted in scale adaptation during a focus group session, grounding the modifications in real-world UXs.

Table 3 Variable adaptation, reference sources, and reliability.
Full size table

Secondly, all UX issues are measured using a five-point Likert scale, ranging from one (strongly disagree) to five (strongly agree). Specific measurement items can be found in Appendix A. To ensure the validity of the measurement of UX issues, Cronbach’s alpha reliability coefficient is used to check the consistency of the research variables of the survey questionnaire across the measurement items. Specifically, the questionnaire includes 36 questions related to 12 aspects of UX. The results show that the reliability coefficients for all variables are above 0.7, indicating good questionnaire reliability and sufficient internal consistency of these factors (Churchill and Iacobucci, 2006). Additionally, the DAARS proposes 20 example questions based on dimensions such as tolerance, loss of control, withdrawal, and escape (see Appendix B). The rating scale is based on research by Armstrong et al. (2000), with a total of 10 points ranging from “Not true at all” (1 point) to “Extremely true” (10 points). This study calculated the average of these 20 questions to represent participants’ addiction levels. Furthermore, the validity analysis of the addiction assessment scale proved its reliability (Cronbach’s alpha = 0.977).

Thirdly, the sampling adequacy and sphericity diagnostics strongly support the use of the correlation matrix in latent-variable modeling. The Kaiser–Meyer–Olkin value of 0.920 exceeds the marvelous threshold of 0.90 suggested by Kaiser (1974), indicating that the common variance is high enough for reliable factor extraction. Bartlett’s test of sphericity further corroborates factorability. The observed chi-squared value (χ2) is 24426.52 (df = 1540), and p < 0.001. This rejects the null hypothesis that the matrix is an identity matrix. This confirms the presence of substantial inter-item correlations. Together, these indices justify proceeding with principal-axis factoring or confirmatory factor analysis. They also suggest that any extracted factors will likely be stable and interpretable within this dataset.

Fourthly, a principal component analysis was performed and 13 factors with eigenvalues greater than 1 were extracted, satisfying Kaiser’s criterion. These factors jointly explained 73.6% of the total variance, which is well above the 60% benchmark usually deemed adequate for multifactor behavioral scales. Before rotation, the general factor accounted for 30.1% of the variance, indicating a substantial, though not dominant, common core. Varimax rotation redistributed this influence, reducing the first component to 25.4% and yielding a flatter eigen-spectrum. The next seven factors each capture approximately 4% of incremental variance, suggesting a set of multidimensional constructs rather than a single latent trait. The steep drop from an eigenvalue of 16.86 to 5.51 and the elbow at factor 13 support the retention of these 13 components. Overall, the solution balances parsimony with explanatory power and provides an empirical foundation for the 12 UX dimensions plus the addiction construct specified in the confirmatory model.

Fifthly, the Varimax-rotated matrix exhibits a clean, simple structure that corroborates the theorized 13-factor measurement model (see Appendix C). Component 1 captures the 20 DAARS items (Y1–Y20) with saturations of at least 0.795. This confirms that behavioral addiction symptoms form a single latent factor, independent of UX motives. Components 2–13 each align with one of the 12-factor UX model constructs. All salient loadings exceed 0.70, while cross-loadings remain below 0.30, attesting to discriminant validity. The rotation converged in seven iterations, further indicating factorial stability. Together, these results confirm that each experiential facet is psychometrically distinct and that addiction emerges as a higher-order construct rather than a mere aggregate of UX preferences. This justifies including them separately in subsequent structural analyses.

Finally, this research conducted a comprehensive reliability assessment, encompassing goodness-of-fit testing and confirmatory factor analysis (CFA), in accordance with the guidelines provided by D’Agostino (2017) and Cole (1987), respectively. The measurement model analysis was performed using AMOS. The results from the goodness-of-fit test indicated a CMIN/DF ratio of 2.037, which falls below the conventional threshold of 3. Furthermore, construct-level diagnostics (see Appendices A and B) support convergent reliability. All standardized loadings surpass 0.65, with most exceeding 0.75, attesting to strong item coherence. Composite reliabilities range from 0.753 for curiosity to 0.977 for user addiction, which is well above the 0.70 criterion. Average variances extracted exceed 0.50 for every latent variable, which establishes adequate shared variance and demonstrates psychometric soundness across the adapted constructs.

Data analysis

The data analysis in this study encompassed two primary domains. The first domain entailed evaluating the influence of the addiction measure, where the dependent variable (addiction) and the criterion variable (the 12 experiential factors) were assessed using subjective self-report measures. Consequently, it was imperative to investigate the impact of the recorded addiction scores from an objective perspective. For this purpose, correlation analyses were employed to elucidate the association between a subjective indicator of addiction (addiction) and an accurate behavioral assessment (degree of experience with dating apps) to establish the validity of the addiction measure in the study.

Subsequently, our investigation aimed to identify key UX factors contributing to the accuracy of dating app addiction predictions. To achieve this objective, linear regression analyses were employed to formulate models delineating the association between the 12 experiential factors and the phenomenon of addiction. It allows for the modeling of continuous data to determine how changes in the predictors influence the outcome. Furthermore, the variance inflation factors (VIFs) were computed to ascertain the extent of covariance among the 12 predictor variables, following the methodology outlined by O’Brien (2007), ensuring that the regression model is reliable and the derived insights are valid. Hence, linear regression provides a clear, interpretable framework for analyzing and understanding the dynamic relationship between UX factors and addiction in dating apps.

Results

This section sequentially validates model measures, compares addiction scores by gender, and tests hypothesized UX addiction pathways. Regression analysis and diagnostic visuals identify distinct sets of predictors and sound model assumptions for location- and group-based dating apps.

Assessing the impact of addiction measurement

Correlation analysis, as elucidated by Fox (1997), assesses the relationship between variables, with the correlation coefficient ranging from −1 to 1, signifying the strength of the association. The discussion in Appendix D will expound upon the correlations among the variables in this study, adhering to this criterion. The outcomes of the correlation analysis revealed a significant positive correlation between each variable and addiction. Furthermore, addiction level exhibited a notable and positive correlation with metrics such as the weekly time spent on dating apps, usage frequency, chat frequency, and offline meeting frequency. Consequently, greater engagement with the dating apps corresponded to elevated addiction levels. This observation underscores the consistency of addiction measurement from subjective and objective perspectives, validating the addiction measure employed in the present study.

Results of the gender t test

This study used independent t tests (Kim, 2015) to compare the mean differences in gender and level of experience with dating apps across all samples. As shown in Table 4, the results revealed that men reported significantly higher scores for dating app addiction than women did (M_(male) = 3.73, SD = 1.02; M_(female) = 3.43, SD = 0.94; t = 3.82, p < 0.001). The corresponding Hedges g of 0.30 indicates that this effect size is small yet statistically significant (Kim, 2015). Correlational evidence places this difference within a broader behavioral pattern. Negative correlations indicate greater engagement in core dating app activities among men. They reported higher weekly match counts (r = −0.10, p < 0.05) and greater participation in offline meetings (r = −0.21, p < 0.01). Conversely, these behaviors are positively correlated with addiction (r = 0.38 and 0.41, respectively), suggesting a behavioral pathway from typical male usage patterns to increased dependency.

Table 4 Gender differences in Dating-App Addiction Scores.
Full size table

Determining predictors of dating app addiction

Having confirmed measurement validity and gender differences, this study next identifies which UX factors drive addictive use. Separate regressions for location- and group-based apps quantify explanatory power, assess multicollinearity, verify model assumptions through residual diagnostics, and isolate the specific UX dimensions that significantly elevate addiction risk.

Firstly, in model testing, location-based dating apps, R2 = 0.279 (adjusted R2 = 0.251), indicating that the UX module explains 27.9% of the variance in addiction scores. The corresponding F2 = 10.09 (p < 0.001) confirms that this model outperforms the baseline model that uses only the intercept. For group-based dating apps, the explanatory power is nearly identical (R2 = 0.271, adjusted R2 = 0.242, F2 = 9.10, p < 0.001). According to Cohen’s (2013) benchmarks, R2 values between 0.26 and 0.33 represent a moderate effect size, indicating that the selected UX predictors are practically significant.

Subsequently, Fig. 3 depicts normal probability–probability (P-P) plots for the regression-standardized residuals predicting user addiction. In a P-P plot, observed cumulative probabilities are plotted against the cumulative probabilities expected under a perfectly normal error distribution. Thus, the closer the points lie to the 45-degree reference line, the better the residuals conform to the normality assumption that underpins the validity of t and F-tests.

Fig. 3
Fig. 3
Full size image

Normal P-P plot of regression standardized residuals on location-based dating app addiction.

Then, the relationship between the model’s standardized residuals and predicted values is illustrated by Fig. 4. The results suggest that the two samples are linear and have equal variances. There are also no significant outliers, which indicates that outliers are unlikely to bias the parameter estimates (Osborne and Waters, 2002). These visual diagnostics complement the P-P plot (Fig. 3) and show that both regression models satisfy the key assumptions necessary for valid inference.

Fig. 4
Fig. 4
Full size image

Scatterplot on dating app addiction.

Finally, the regression analysis was performed to evaluate the explanatory capability of each variable, utilizing the addiction score as the dependent variable and the 12 UX factors as independent variables (Table 5). The resulting formula is presented below, with the omission of usability, difference of initiative, curiosity, loneliness, and traditional ethics endorsement due to their limited predictive significance.

$${\rm{Location}}\mbox{-}{\rm{based}}\;{\rm{dating}}\;{\rm{app}}\;{\rm{addiction}}\;=\;0.109\;{\rm{Self}}\mbox{-}{\rm{optimality}}\;+\;0.146\;{\rm{Usability}}\;+\;0.172\;{\rm{Selectivity}}\;+\;0.128\;{\rm{Casual}}\;{\rm{sex}}\;+\;0.237\;{\rm{Self}}\mbox{-}{\rm{fulfillment}}$$
$${\rm{Group}}\mbox{-}{\rm{based}}\;{\rm{dating}}\;{\rm{app}}\;{\rm{addiction}}\;=\;0.139\;{\rm{Selectivity}}\;+\;0.139\;{\rm{Agglomerativity}}\;+\;0.142\;{\rm{Identity}}\;+\;0.197\;{\rm{Social}}\;{\rm{barrier}}$$
Table 5 Model validation and regression results.
Full size table

Specifically, the regression analysis results (see Table 5) indicated that the VIF values for the empirical factors were below 1.902, signifying the absence of covariance among the factors considered in this study (Kutner et al., 2005). In the sample of location-based dating apps, self-optimality (β = 0.109, p < 0.05), usability (β = 0.146, p < 0.05), selectivity (β = 0.172, p < 0.05), casual sex (β = 0.128, p < 0.05), self-fulfillment (β = 0.237, p < 0.05) all have significant positive effects on addiction; in the sample of group-based dating apps, selectivity (β = 0.139, p < 0.05), agglomerativity (β = 0.139, p < 0.05), identity (β = 0.142, p < 0.05), and social barrier (β = 0.197, p < 0.05) all had significant positive effects on addiction. Table 3 shows the regression model of addiction for location-based dating apps and group-based dating apps.

Discussion

This study identified a positive correlation between users’ addiction levels and their frequency of use of dating apps. High levels of addiction were associated with more extensive time spent on dating apps. Notably, male users exhibited a higher susceptibility to addiction compared to female users. These findings substantiate the existence of dating app addiction.

Gender-differentiated predictors of addiction

The t-test indicates that men exhibit more pronounced addictive behaviors than women, aligning with the results of previous multivariate studies. In the regression model, the individual pathways were linked to the risk of addiction to dating apps. Regression analysis reinforces this interpretation. In the location-based subsample, where chance encounters and rapid feedback are most prominent, predictors with higher loadings among men (self-fulfillment, selectivity, and casual sex) exhibited significant positive beta values (0.237, 0.172, and 0.128, respectively). In contrast, predictors of group app dependency (Agglomerativity, Social Barrier, and Identity) were unrelated to gender, reflecting a lack of significant gender differences in this segment. Group-oriented platforms seem to cancel out the effects of gender, which is consistent with previous findings that women value controlled social expansion and recognition.

These results suggest that higher addiction scores among men are not merely a general gender effect, but rather, are driven by the higher emotional arousal experienced during geographical matching. This emotional arousal amplifies reward expectations and offline conversion rates, thereby accelerating the cycle proposed in the I-PACE framework. Ultimately, this cycle leads to greater salience and tolerance. These gender-related preferences for availability are consistent with prior evidence (Sumter et al., 2017; Shotton, 1991), suggesting that men seek novelty and sexual opportunities while women use apps to gain recognition and controlled social expansion. Gender differences are not a product of unequal screen time, but rather, they reflect distinct combinations of motivation and UX that lead to differing addiction tendencies.

Factors and mechanisms predicting addiction

This study identifies crucial UX factors that researchers can utilize to predict addiction to dating apps. The primary UX factors predicting addiction to location-based dating apps encompass self-optimality, usability, selectivity, casual sex, and self-fulfillment. The key UX factors that predict addiction to group-based dating apps are selectivity, agglomerativity, identity, and social barrier. The current study proceeds to elucidate the reasons behind the dominance of these factors in influencing the level of addiction.

In utilizing location-based dating apps, self-optimization and self-fulfillment are mutually reinforcing. Users formulate optimization strategies within specific online self-presentation profiles, strategically managing their impressions, which arise from the inherent functionality of the dating apps itself, as elucidated by Ellison et al. (2012). This dynamic engenders a feedback loop of enhanced user satisfaction and self-fulfillment compared to offline interactions. To be specific, self-optimality refers to the ideal self that users deliberately construct on dating apps. Based on self-differentiation theory (Higgins, 1987), this concept represents a controllable feedback system in which each profile adjustment immediately triggers social recognition. According to Sherman et al. (2016), neuroimaging studies indicate that this social feedback activates the ventral striatum, a reward center associated with substance addiction and behavioral addiction. In the I-PACE sequence, this feedback shortens the emotional-cognitive phase, rapidly enhancing mood and increasing the app’s salience. Consistent with the spiral of seeking validation (Nadkarni and Hofmann, 2012), participants with higher self-optimization scores (Appendix D) opened the app more frequently (r = 0.13**) and arranged more offline meetings (r = 0.16**).

Subsequently, another factor that is positively correlated with addiction is usability. Specifically, usability manifested as interface efficiency and minimal cognitive load emerged as a significant predictor of addiction to location-based dating apps (β = 0.146, p < 0.01). However, this was not the case in the location-based model. Psychologically, high usability’s power stems from “frictionless design” (Baumeister, 2002), wherein interfaces that require minimal cognitive effort reduce self-regulatory resistance and enhance impulsive engagement (Fogg, 2002). Location-based apps compress the match-chat-meet process into a few swipes, creating a low-investment, high-return environment that enhances lead reactivity and weakens executive control. These are central mechanisms to the I-PACE model (Brand et al., 2019). Empirical research shows that usability is associated with both weekly matching (r = 0.19**) and offline meetings (r = 0.20**). These two behaviors, in turn, strongly predict addiction. Thus, the technical fluency of location-based applications acts as a “situational amplifier,” converting potential motivation into compulsive behavior (Alter, 2017).

This frictionless architecture facilitates a rapid transition from online interaction to casual sex. This motivation is associated with an increased risk of addiction among the study participants (β = 0.128*). Previous research has shown that geolocation features normalize “instant” contact and reinforce reward expectations through variable rates of success (Orosz et al., 2018). While this efficiency may satisfy the desire for novelty, it undermines the depth of the romantic relationship. According to Miles (2017), users report that when the pace of interaction accelerates, intimacy decreases and partners become more replaceable. Therefore, the synergistic effect of technical fluency and sexual immediacy intensifies the tolerance and emotional regulation aspects of behavioral addiction. This explains why usability remains a moderate, practical, predictive indicator, even when controlling for other UX factors (Gignac and Szodorai, 2016). This effect reshapes time expectations, lowers behavioral thresholds, and amplifies reward salience. Thus, it provides a clear psychological pathway from interface design to addictive engagement.

Moreover, selectivity significantly predicts addiction in both location-based dating apps and group-based dating apps in the analysis of key UX factors, reflecting the technical ability to scroll through hundreds of profiles at minimal additional cost. This phenomenon aligns with dating apps’ inherent design logic and UX mechanisms. These platforms enable users to rapidly assess numerous potential partners in a continuous stream, a process that can be inherently exhilarating (Iyengar and Lepper, 2000). Occasionally but unpredictably displaying an extremely attractive match serves as an enticing motivator (Cummings and Mays, 2021), replicating the reinforcement structure of slot machines (Alter, 2017). Selectivity, therefore, also ensures the continuity of the excitement of becoming addicted to a dating app.

On the other hand, based on group-based dating apps that emphasize homogeneity, agglomerativity (β = 0.139*) and identity (β = 0.142*) jointly predict user addiction to this type of app. This indicates that users return for social recognition rather than novel experiences. According to social identity theory, feedback within groups can enhance self-esteem (Tajfel et al., 1979). The data in Appendix D show a similar effect: agglomerativity and identity are closely related (r = 0.42**), as are both with addictiveness (r = 0.31**, 0.24**). Therefore, community tags and the algorithm’s “best match” recommendations can serve as personalized identification tools to enhance the UX. Additionally, dating apps filter out people who do not align with predefined criteria. Online vetting is essential for users due to the problem of false personal information leading to misidentification of users as well as the presence of sexual objectification in online encounters (Hess and Flores, 2018). Further, the different tags on the personal information page verified by group-based dating apps also exacerbate the user’s identity.

Furthermore, in group-based dating apps, the coefficient for social barriers is positive (β = 0.197**), which highlights the contradictory appeal of private groups. Users seek new connections while keeping their app activity separate from their existing networks. According to boundary management theory (Bazarova and Choi, 2014), this controlled visibility is appealing. Compared to the effort required to find a dating partner socially or physically, dating apps easily break the user’s pre-existing social circle and access many potential matches that match their identity. At the same time, due to the closed nature of particular groups, group-based dating apps serve as a medium for users to build intimate relationships with the same type of users in a relatively uninvolved space. Consistent with this, higher social inhibition coefficients and higher matching rates (r = 0.16**) coexisted, indicating that perceived anonymity lowered inhibition thresholds and maintained engagement.

Analysis of less effective predictors of addiction

In contrast, four factors that were found to be less effective predictors of addiction were difference of initiative, curiosity, loneliness, and traditional ethics.

First of all, the reason that the difference of initiative does not constitute a significant addiction is twofold. On the one hand, female users appear to be more rational in the face of the tilt of the initiative. The content of the gap between current and desired female online activity focuses on mate selection needs. These factors are best digested through visual language rather than the ceding of rights. On the other hand, skewing the initiative towards male users would result in the loss of more female users of dating apps (Tyson et al., 2016). With a reduced choice space, male users will be lost as well. At the same time, there is only a weak and inconsistent correlation between the difference of initiative and gender (r ≈ 0.04–0.10), indicating that users have adapted to the bilateral matching architecture. Therefore, differences in initiative no longer translate into sustained, reward-driven engagement.

Next, while curiosity sparks the initial download, its motivational half-life is short (Loewenstein, 1994). Longitudinal studies demonstrate a steep decline in novelty after the first month of use. In the data sample, curiosity correlates with adoption age (r = 0.10*) but not with high-frequency behaviors such as matches and meetings. This indicates that, once epistemic uncertainty is resolved, usage becomes goal- rather than curiosity-driven. Therefore, curiosity adds no incremental variance to addiction.

Then, loneliness is an unpleasant and distressing psychological state that refers to a person’s subjective deficiencies in social relationships (Peplau and Perlman, 1982). However, loneliness was not highly correlated with offline meetings (r = 0.05). This suggests that daters who were still lonely either chose to disengage or turned to richer social channels (Liu et al., 2020). This is likely because users are unable to establish a sense of social connection (Goldenberg, 2019). Thus, the affective deficit underlying loneliness is neither reinforced nor alleviated by the swift, low-commitment exchanges typical of dating apps, explaining the null regression effects.

The final societal factor is traditional ethics, which has an insufficient correlation with addiction to dating apps, probably because of the complexity of ethics in dating apps (Timmermans and Courtois, 2018). It motivates entry but also motivates exit once a socially approved relationship is secured. The resulting bidirectional forces cancel out in cross-sectional models, yielding non-significant coefficients. In addition, cultural heterogeneity (29 countries) dilutes any single normative signal, further weakening predictive power.

Collectively, these findings suggest that the motivational energy of each factor is unstable, context-dependent or offset by opposing forces, making it an ineffective long-term driver of compulsive engagement.

Conclusions

This research identifies vital predictors of dating apps addiction. These critical factors engender long-lasting effects and mechanisms for user engagement, increasing user motivation to extend their usage and engage in more profound interactions within the dating apps. In contrast, four less significant factors exert a weaker influence on long-term user behavior. Figure 2 illustrates the possible mechanisms of addiction in a dating app among the user, the app, and other users. Theoretically, the I-PACE model is enriched by specifying the availability of specific designs that transform personal motivation into compulsive participation; Griffiths’ criteria are operationalized through measurable UX prediction indicators; and HCI academic research is advanced by revealing how exemplary usability inadvertently reproduces the full addiction syndrome.

In Fig. 5, the distinction between addiction in location-based and group-based dating apps is attributable to their unique engagement strategies and target user behaviors. Location-based apps like Tinder utilize proximity and usability to facilitate quick, casual encounters, promoting a cycle of immediate gratification that can heighten addiction risks. Features like self-optimality and selectivity allow users to manage their interactions efficiently, reducing social failures and encouraging continuous use. Conversely, group-based apps like Grindr focus on creating a sense of community and identity among specific user groups. They offer features that bolster group cohesion and privacy, allowing users to explore and express their identities within a supportive environment. This leads to a different kind of addictive engagement, driven by emotional connections and a sense of belonging, contrasting with the more transactional nature of location-based apps.

Fig. 5
Fig. 5
Full size image

Addiction-causing mechanisms in dating apps.

In order to reduce the risk of addiction associated with dating apps, it is not enough to reduce the addictive factors highlighted in the research. This occurs because these unpredictable factors not only elevate the addiction risk but also influence the overall attractiveness of the application. Hence, an unattractive dating apps may need more matched users to avoid a shortened life cycle. Moreover, it is recognized in the field of online dating development that the core of design is to engage users and increase the commercial viability of the business (Bonilla-Zorita et al., 2021). From a business perspective, exploring the delicate balance between promoting favorable attraction and avoiding addiction is vital. Therefore, future research may employ the provided UX framework and addiction prediction model to achieve a balance between business objectives and addiction prevention. Designers face the ethical task of balancing abundant choice with debiasing tools to mitigate overload without curtailing inclusivity.

This study evaluates various application designs to clarify the implications of the findings. From the perspective of addiction-preventive design, it adopts a strategy that mitigates addiction risks while ensuring a positive, engaging UX. To translate these insights into responsible practices, the study situates them within debates on digital well-being and persuasive technology (Roffarello and De Russis, 2023; Fogg, 2002). Location-based apps exploit proximity and intermittent-reward loops; group-based apps leverage identity affirmation. Both patterns can be redirected through “humane design” principles—e.g., soft caps on daily swipes, default break reminders, and opt-in transparency dashboards that visualize cumulative usage. Such measures align with emerging ethical UX standards that prioritize autonomy, beneficence, and non-maleficence (Gray et al., 2018). This research also stress co-design with diverse user communities so that inclusivity and harm-reduction goals are embedded from inception.

Future research can build upon this study’s limitations. First, the cross-sectional design precludes causal inference, necessitating longitudinal or experience-sampling methods to establish temporal precedence. Second, while respondents hailed from 29 countries, 53% resided in the U.S. or China, limiting cultural generalizability. Third, to minimize attrition, sensitive variables like household income and sexual orientation were omitted, precluding analysis of socioeconomic and sexual orientation moderators. Fourth, the sample focused on adults aged 18+ years, while adolescent usage requires equal attention—sex-related UX motivations for addictive behaviors may heighten online sexual harassment risks among teen users (Sumter et al., 2017). Finally, this study only analyzes UX as a predictor of dating apps addiction; future research could explore potential interventions for at-risk users.