In light of the impact of digitalization on healthcare, along with the scarcity of professionals and the escalating expenses associated with accessing care, digital tools such as chatbots (Woebot and Nuna) have been created to promote well-being through accessibility and continuous support, offering non-judgmental and cost-effective alternatives to professional care (Inkster et al., 2018; Lin et al., 2023). Among the factors that influence the effectiveness of these well-being chatbots, anthropomorphization-assigning human-like attributes, including gender-plays a critical role (Rapp et al., 2021). Previous studies have shown that gender stereotypes influence perceptions of chatbots. Female chatbots often appear more human and are better at meeting users' needs, potentially due to stereotypical associations of women with warmth and communal traits (Borau et al., 2021; Nass et al., 1994). These human attributes can enhance user interaction, leading to increased willingness to engage with the chatbot (Belanche et al., 2021). Despite extensive research in commercial settings, the impact of chatbot gender in non-commercial contexts, particularly well-being, remains underexplored (Borau et al., 2021).This study investigates how the gender of text-based chatbots affects users' perceptions of agency, communality, trust, and intention to use the chatbot for well-being purposes. It hypothesizes that female-gendered chatbots elicit more positive attitudes, greater trust, and higher intentions to use than male-gendered chatbots. Also, relying on the stereotype content model (SCM) (Fiske et al., 2007), the warmth and competence framework (Belanche et al, 2021;) and Novak and Hoffman's Assemblage Theory ( 2018), this research assumes that Female (male)-gendered chatbots are perceived more communal (agentic) than male-gendered chatbots. Agentic orientation involves instrumentality, dominance, and competence in the pursuit of individuating the self. Communal orientation involves cooperativeness, helpfulness, and trustworthiness. Finally, following Pitardi et al. (2022) or Zogaj et al. (2023), this research assumes that the congruence of chatbot and user genders leads to more favourable evaluations. Methodology

A total of 301 participants from the Prolific panel completed an online questionnaire, resulting in a final sample of 297 after excluding inconsistent responses. Participants were randomly assigned to interact with either a male-gendered or female-gendered chatbot. Gender was manipulated using names and avatars following Borau et al. (2021), ensuring correct gender assignment through pre-tests.

Measures for agency and communality were adapted from Eyssel and Hegel (2012), while trust, attitude, and behavioral intentions were measured using scales from Pitardi and Mariott (2021), Borau et al. (2021), and Liu and Tao (2022), respectively. Data were analysed using SPSS (version 28).