TLDR

1. ​Finding: Users' perceptions about algorithms are an important aspect of their experience, especially in conjunction with their levels of interest in a potential partner/match. Playing a significant role in how useful the algorithms and systems are perceived, which could impact retention rates. 
   Insights: Good PR meant to influence users' and potential users' attitudes, specially those related to trust in the usefulness of algorithms to deliver consumer expectations, is important and should be considered a critical marketing strategy. Furthermore, if a dating application has restrictions on how many potential partners a user can be shown during engagement, a strategy to support increased users' perceptions about algorithm usefulness could be ensuring the last potential partner option shown to a user during each engagement is one that has a a high likelihood of the user being interested in them.

   ​

2. Finding: Increased amounts of feedback about users' self-presentations impacts how useful they perceive the applications to be. This study demonstrated that the feedback can be provided by either human or technological information sources. 

   Insights: To help ensure the application is positively received, the system should be designed so that users provide potential partners specific feedback about how their self-presentation is received (ie. liking a specific aspect of a profile rather than the holistic profile) and the system should in turn provide users with increased amounts of information. Strategies to address this would involve the system collecting data about what aspects of a users' self-presentation are more positively received by potential partners than others and providing these insights to the consumer in manners that give them senses of support, through guidance by the system, and autonomy, in how they choose to self-present being given said feedback.

​

Summary

     The introduction of technologies into everyday life has impacted society in numerous ways - often, silently until uncovered. Their influence is able to permeate in part due to seamless user experiences, a product of polished platform designs and the veiling of the system mechanisms that enable their functionality like algorithms. However, most people are unaware of the presence of algorithms in the systems they use [1 & 2] and overlook the influencing powers designs can carry. Illuminating the need for investigation into the impacts of design affordances and algorithm mediation on human attitudes and behaviors, explicitly as they pertain to interpersonal relationship development. 

     The formation of interpersonal relationships, specifically romantic relationships, has been altered by the expansion and persistent presence of social networking platforms, like online dating applications. The use of dating applications has risen exponentially in the last decade [3] and prior research has shown that the increasing ubiquity of dating applications in today’s networked society has altered the architecture of romantic relationship seeking and formation; especially in regard to users’ self-presentations [4 & 5] and in turn the feedback received about their self-presentations’ reception by others. In the context of dating applications, feedback can be provided by other users and/or the technology itself [6]. Thus, highlighting a need for increased understanding about how varying degrees of combined source multiplicity feedback can affect users’ attributions about potential partners and if user recognition of technology’s influence affects their perceptions, behaviors, and decisions related to application use. 

     The Source Multiplicity, Attributions, Recognition, and Transformation (SMART) Model of Online Dating [6] has been proposed to address these topics (more information about this model below), however little scholarship utilizing it has been produced. Thus, the objectives of this study were to examine how perceptions about dating application algorithms and system usefulness are affected by source multiplicity, attribution, and the recognition of technology - utilizing the SMART Model as a lens for consideration. This study utilized an experimental survey methodology in which the stimulus was a web application that mimicked a dating application and tested three degrees of feedback; implicit, somewhat explicit, and explicit. More information about the stimulus can be found below. 

     This study found that participants had above-average self-esteem and confidence in dating applications algorithms for providing them relevant matches, interest in liker influences and predicts how useful participants find the algorithms, and variations in the degree of feedback about how participants’ self-presentations were received plays a role in how useful they perceive the application’s system for attaining their goals. Highlighting that perceptions about algorithms are important, interest in a liker plays a significant role in how useful algorithms and systems are perceived, and increased information provided by human and technological information impacts how useful applications are perceived. Theoretically, the findings provide support for the applicability of the SMART Model and practically, providing evidence about the potential effects that system designs pertaining to feedback can have on users’ perceptions of usefulness. 

​

SMART Model

   The Source Multiplicity, Attributions, Recognition, and Transformation (SMART) Model of Online Dating sets out to explain how technology influences the perceptions, attributions, and decisions people make in computer mediated dating environments [6]. This paper focuses on the source multiplicity, attributions and recognition components of the model. Dating systems combine information from multiple sources into a single mediated space, thus source multiplicity reflects the two classifications information sources: human sources and technological sources [6]. The model predicts that that both sources will influence users’ attributions about partners and decisions regarding mate selection [6], i.e., interest in matching with a liker. Furthermore, users’ level of conscious recognition of technology during online relationship formation is a subprocess predicted to affect their information processing [6]. To better understand how these components are interrelated, see figure 1 below.

 

 

 

 

 

 

 

 

 

 

 

​

Figure 1. The Source Multiplicity, Attribution, Recognition, and

Transformation (SMART) Model of Online Dating.

​

Stimulus

     The stimulus’ interface design (Figure 2) mimics popular dating applications on the market and the web application was coded by a software engineer. It was engaged with by participants in a tab or browser external to the survey and once opened, participants were prompted to select their gender preference, in terms of how their “liker” would identify, and upload a photo of themselves to function as their profile image. Neither their selected gender preference nor image was recorded in association with participants’ survey responses. Participants were then shown one of the three stimuli, in the form of “cards,” through random distribution. Analysis revealed a relatively uniform split between the three, with 38.2% receiving Card1 – implicit feedback, 34.4% receiving Card2 – somewhat explicit feedback, and 27.4% receiving Card4 – explicit feedback. 

​

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

​

Figure 2: Stimulus interface architecture and interaction design

​

     Each card contained essentially the same interface design with variations pertaining to how much feedback participants were provided about how their self-presentation was received by a fictional “liker.” The “likers” were depicted by free-for-use stock images collected from Pexels.com. These six images were selected for a myriad of reasons, most notably to include a variety of “liker” ethnicities and to illicit feelings that they could be a realistic and authentic user of the platform. Implicit feedback was measured through M_Card1 and F_Card1 (Figure 3). These cards contained minimal human and technology provided feedback, as the fictional potential match “liked” a participant’s holistic profile. 

​

​

 

 

 

 

 

 

 

 

 

 

 

​

Figure 3: Cards1 - Implicit Feedback

​

Somewhat explicit feedback was measured through M_Card2 and F_Card2 (Figure 4). These cards contained more human feedback than the implicit feedback cards, as the fictional potential match “liked” a specific aspect of the participant’s profile – their uploaded image. 
 

 

 

 

 

 

 

 

 

 

 

 

​

​

Figure 4: Cards2 - Somewhat Explicit Feedback

 

Finally, explicit feedback was measured through M_Card4 and F_Card4 (Figure 5). These cards the most amount of human and technology provided feedback, as the fictional potential match “liked” a specific aspect of the participant’s profile, their uploaded image, and were provided statistical information related to fictional previous engagement with this aspect of their profile. On these cards it was stated that the image engaged with was their “most liked” photo – with 20 users having liked the photo, of which the participant had matched with 10 of them. The ratio of likers to matches was intentionally set as 50% to maintain a sense of neutrality in how well it was received and avoid potential framing  of how useful participants felt the application was. 

 

 

​

​

 

 

 

 

 

 

 

 

​

​

Figure 5: Cards3 - Explicit Feedback