Beyond Embodiment and Social Presence: Preferences for Virtual Assistant Gender and Clothing Style Jeunese Adrienne Payne 1 , Andrea Szymkowiak 1 , Graham Johnson 2 , Paul Robertson 1 and Rosemary Henderson 1 1 University of Abertay, Dundee 2 NCR, Dundee {0501726@live.abertay.ac.uk, a.szymkowiak@abertay.ac.uk, GJ162001@NCR.com, p.robertson@abertay.ac.uk, 0702066@live.abertay.ac.uk} Abstract While introducing a human-like embodied character can enhance social presence in computing technology, it can often be poorly received because user preferences for particular appearance-based attributes are not taken into account. To investigate some of these preferences, specifically gender and dress code, this paper extends the findings of a small exploratory, interview-based study, to data obtained from 390 participants aged between 16 and 82 who chose their preferred three-dimensional virtual assistant (VA) as part of an interaction with a self-service checkout (SSCO) simulation. These data were collected from visitors to an exhibition at a science center in the UK. Chi-Square tests revealed a preference for female VAs, but only by female participants. There was no preference for formality of dress. The paper discusses the appropriateness of the following psychological theories to explain these results: the stereotype content model, the behaviors from intergroup affect and stereotype (BIAS) map framework, and the shifting standards model of stereotypes. One of the most significant discussions in Human- Computer Interaction (HCI) currently is what features of an embodied interface agent can help improve the interaction between a system and its user, e.g., Haake (2009), Vogeley and Bente (2010), and Payne, Johnson, & Szymkowiak (2011). An interface agent is a computer program implemented to help users complete on-screen tasks (Dehn & van Mulken, 2000). The human-like embodiment of these agents helps provide a sense of social presence in computing technology (Vogeley & Bente, 2010; Demeure, Niewiadomski, & Pelachaud, 2011), i.e., the perceptual awareness of another person (Abeele, Roe, & Panelare, 2007) or the feeling of human- like personal contact (Verhagen, van Nes, Feldberg, & van Dolen, 2011). A sense of social presence can be induced via several social cues of an interface agent, not all of which rely on its appearance, including social dynamics (e.g. turn- taking, co-operation, and praise), psychology (e.g. humor, personality, and empathy), and language (e.g. interactive language, spoken language, and langauage recognition) (Fogg, 2003). In fact, it is argued that users will treat computers as social agents without any visual representation at all (Dehn & van Mulken, 2000; Reeves & Nass, 1996). However, evidence suggests that people tend to prefer an embodied interface agent with a dialogue box as opposed to text-only because it makes computing technology appear more personal and caring (Liu, Helfenstein, & Wahlstedt, 2008). Moreover, physical social cues not only enhance social presence (Fogg, 2003) but have been found to positively influence user performance and attitude – increasing efficiency, overall enjoyment, social collaboration, engagement, and satisfaction with a computer-based program as well as reducing learning time (Cowell & Stanney, 2005; McBreen & Jack, 2001). Physical social cues include human-like features such as a face and body, as well as behaviors such as movement and nonverbal communication (Fogg, 2003). Another related cue for enhancing social presence is social role (Fogg, 2003), and this too can be implied by physical cues (e.g. uniforms or social grouping), even when it was not the intention of the designer. Common interface agent social roles are teacher and assistant. While these roles do not rely on appearance – it is reasonable for the role of an agent to be implied based solely on the type of gudiance it provides – they are often visually implied, e.g., via a lab coat or employee branding. Social roles can also be implied from visually apparent demographics such as age, gender, or ethnicity. Men and women, for example, are expected to behave in accordance with gender stereotypes which reflect the divergent social roles they typically perform, e.g., breadwinner and homemaker respectively