The hyperpersonal effect in online dating: effects of text-based CMC vs. videoconferencing before meeting face-to-face

ABSTRACT

This study investigated the impact of (the lack of) audiovisual cues during conversations preceding a first face-to-face meeting among prospective daters on daters’ perceptions of partners’ social and romantic attraction. Additionally, the study examined the effect of modality switching, from online to offline interaction. Thirty-nine individuals participated in a round-robin speed dating event, resulting in 95 unique conversation pairings. For their first conversations they were randomly assigned to meet via text-based CMC or videoconferencing. The dyads then had a second encounter, which was face-to-face. Results showed more social attraction between interactants who used text-based CMC than videoconferencing, supporting the hyperpersonal model of CMC. Furthermore, after a modality switch to a face-to-face encounter the hyperpersonal effect persisted for social attraction, while romantic attraction declined.

Today, computer-mediated communication (CMC) plays a significant role in the formation of romantic relationships. Due to new dating apps like Tinder, dating sites and applications are very popular among younger people. In 2017 in the U.S., 30 percent of the internet users between 18 and 29 years old, were using online dating (Statisticbrain.com, 2017). Online date-seeking refers to the practice of using web-based dating sites for finding a romantic partner. Generally, online dating involves building a personal profile, using a searchable database of potential partners or getting recommendations for likely partners based on matching algorithms (see for a review Tong, Hancock, & Slatcher, 2016), and then communicating with a prospective date in order to make a final decision about meeting for a face-to-face encounter. According to one review of online dating processes, one of the most critical aspects of online dating sites is the communication they support that allows prospective dates to interact in the interval before they meet face-to-face (Finkel, Eastwick, Karney, Reis, & Sprecher, 2012).

One potential complication during this interval is that getting acquainted with a prospective date online can distort the levels of attraction among partners compared to meeting face-to-face. Although some suggest that multimodal interfaces, including numerous nonverbal cues, are necessary for daters to gather sufficient information to like one another (e.g., Kotlyar & Ariely, 2013), other theories such as the hyperpersonal model of CMC (Walther, 1996), empirical studies, and anecdotal research indicate that impressions and social attraction developed online through text-only CMC may be exceptionally positive (e.g., Antheunis, Valkenburg, & Peter, 2007; Whitty & Carr, 2006).

However, tests of the theoretical model that may help explain this phenomenon generally occurred in non-romantic relationships (e.g., Antheunis, Schouten, Valkenburg, & Peter, 2012; Antheunis et al., 2007; Sprecher, 2014; Walther, 1997). Few studies of interactive online impression-formation and attraction involved online dating settings specifically (with exceptions of Kotlyar & Ariely, 2013; Ramirez Jr, Sumner, Fleuriet, & Cole, 2015; Sharabi & Caughlin, 2017). While the development of impressions and social attraction has many commonalities across a variety of relationship contexts (e.g., Antheunis, Valkenburg, & Peter, 2010; Gibbs, Ellison, & Lai, 2011), people evaluate prospective romantic partners using additional characteristics and evaluations, namely, in terms of their romantic attraction, which includes such qualities as physical and sexual attraction (Sprecher & Regan, 2002).

For this reason, the ability for a communication medium to support visual cues that can convey physical characteristics may be important in judgments of physical attraction, during the communication that occurs leading up to decisions about the pursuit of an offline romantic relationship. Because physical attraction is an important determinant of romantic attraction (Byrne, London, & Reeves, 1968), online daters often seek and employ visual information to form impressions about others. In that respect, videoconferencing offers potential advantages to online date-seekers. Ultimately, if physical information is important in date-selection, the potential benefits of videoconferencing in this context cannot be discounted.

In sum, although text-based CMC (i.e., text-based chat) may stimulate more attractive hyperpersonal impressions compared to videoconferencing in many social contexts, when it comes to online date-seeking users may nevertheless avoid text-based CMC in their belief that videoconferencing offers greater access to critical attraction-related characteristics. The question of which medium actually confers greater attraction, across multiple dimensions, remains to be investigated. Therefore, the first aim of this study is to investigate the impact of audiovisual cues in a first online date-seeking encounter on social and romantic attraction by comparing media with only textual cues to media with audiovisual cues.

Furthermore, we want to investigate the effect of modality switching on attraction. What are the cross-media implications of the conversations between prospective daters before they meet offline, on their reaction to one another when they do meet offline in a first face-to-face encounter? Following the exchange of messages with a prospective date online, a face-to-face meeting provides additional information that can improve or diminish the impression that prospective dating partners tentatively formed online. If daters have formed over-idealized impressions of each other in online chat, the face-to-face meeting is likely to be disappointing, as the physical encounter might not live up to expectations (Ramirez Jr et al., 2015; Walther & Parks, 2002). However, the potential difference between expectations that prospective daters develop online and their subsequent face-to-face reactions might be mitigated when people use videoconferencing in their online encounters prior to a face-to-face meeting. There should be less of a discrepancy between the overall level of attraction developed during first encounters supported by audiovisual media and a subsequent face-to-face meeting than there may be between an initial text-based CMC encounters and a subsequent face-to-face encounter, especially regarding the qualities of romantic attraction with its partial basis on physical characteristics.

Despite the popularity of online dating, there is little research concerning what happens when online daters meet face-to-face generally, and none that has examined differences in their reactions depending on whether their preliminary online conversations took place using textual CMC or videoconferencing. One survey study investigated the effect of modality switching in online dating, in terms of the relational communication participants experienced. However, that study involved only text-based CMC before a prior face-to-face meeting (Ramirez Jr et al., 2015) and offered no videoconferencing comparison. Another study investigated online dating success after initial face-to-face interaction and found that perceptions of attraction declined after online dating site users met each other face-to-face (Sharabi & Caughlin, 2017). Other studies on modality switching’s impact on attraction took place in non-dating contexts, and have yielded inconsistent results (Ramirez Jr & Wang, 2008; Ramirez Jr & Zhang, 2007; Sprecher, 2014). Therefore, the second aim of this study is to investigate the effect of modality switching on attraction, but unlike previous studies among daters, using alternative forms of online interaction before an offline encounter takes place.

To achieve these aims, the research employed a real-life speed dating event, a format which has been described as a useful research setting for investigating actual dating interactions (Finkel & Eastwick, 2008). In a single speed dating session, a participant had meetings with five different prospective dates. Half of participants used text-based CMC for their first encounters with a prospective date, whereas the other half used videoconferencing for their first encounter. All participants also had a second encounter with the same persons in a face-to-face meeting.

Attractiveness in romantic relationships

Both social and romantic attraction are important in predicting whether one desires a romantic relationship with someone. This study takes both dimensions into account. Social attraction, or liking, implies that a one perceives that another person is pleasant to be with, could become a friend, and would fit in one’s existing network of friends (McCroskey & McCain, 1974). However, the success of a romantic relationship depends not only on social attraction but also, to a greater extent, on romantic attraction (Sprecher & Regan, 2002). Romantic attraction refers to the extent to which someone positively evaluates a person as a potential romantic partner and intends to persist in pursuing future contact. Romantic attraction consists of both physical and sexual attraction (Sprecher & Regan, 2002). Physical attractiveness is an important aspect of romantic attraction (e.g., Finkel & Eastwick, 2008; Lou & Zhang, 2009; Sprecher & Duck, 1994). Based on fitness-related evolutionary theories and ample empirical evidence, attractive people are considered more desirable dating partners (e.g., Gangestad & Scheyd, 2005; Langlois et al., 2000). Moreover, people make inferences about an individual’s interpersonal qualities based on those individuals’ level of physical attractiveness: Individuals perceive physically attractive people to be warmer, more responsible, more interesting, and kinder than less physically attractive people (Dion, Berscheid, & Walster, 1972). Moreover, the success of a date also has been shown to depend on sexual attractiveness which is strongly related to physical attractiveness (Graziano, Jensen-Campbell, Shebilske, & Lundgren, 1993; Regan & Berscheid, 1995).

Communication effects in online dating

One of the most critical aspects of online dating is the communication they support that allows prospective dates to interact before they meet face-to-face (Sharabi & Caughlin, 2017). According to Finkel et al.’s review of online dating processes (2012), the communication facilities provided by online dating sites range from impersonal signifiers of interest such as “winks,” to in-house asynchronous CMC systems that resemble email, synchronous text-based CMC; some dating systems even offer “live interaction via webcams that allows users to see and hear each other” (p. 6). Most dating site users do not go directly from identifying a match to an offline meeting. Rather they tend to engage in communication using the instant or asynchronous messaging system in a site, or text messaging, before arranging to meet face-to-face.

The literature offers different accounts of the effect of online communication that precedes an initial face-to-face meeting in romantic contexts. One of the earliest reports on the subject indicated that individuals who met online and developed romantic interests conversed extensively before exchanging photos or meeting face-to-face. Baker (1998) interviewed couples who met online and eventually continued a long-term romantic relationship face-to-face. The couples indicated that they had gotten to know and love their partners through text-based interaction so well that the eventual exposure to their physical likenesses was superficial and unimportant. Other studies indicate that online, text-based interaction allows people to express and perceive one another’s “true selves” in text-based CMC, a version of the self that is more positive than offline day-to-day interaction affords (Bargh, McKenna, & Fitzsimmons, 2002).

The hyperpersonal model of CMC (Walther, 1996) offers a conceptual framework for understanding and predicting differences in interpersonal interaction and attraction between communication modalities, that has been used to understand communication in online dating situations (e.g., Ramirez Jr et al., 2015; Toma, Hancock, & Ellison, 2008). In contrast to more traditional models of communication, it suggests that media with fewer, rather than greater, cue systems afford users more control over their message construction. When individuals use text-only communication channels such as texting, email, and instant messaging, they take advantage of the lack of nonverbal cues and engage in greater selective self-presentation than people can usually manage when they communicate with a fuller range of channels such as facial expressions, vocalics, and physical appearance (Walther, 1996; see for review Burgoon, Buller, & Woodall, 1989). Rather than experience decrements to their expressiveness, users easily express themselves favorably using the content and style of language and typographic features. They employ the affordances of the channel to craft their messages deliberately and edit them before sending them on, allowing them to create more favorable and intimate messages than would be likely in comparable face-to-face encounters (Walther, 2007).

Research also supports the hyperpersonal model’s assertions that individuals idealize one another when forming impressions based on textual messages (e.g., Antheunis et al., 2012, 2007; Bargh et al., 2002; McKenna, Green, & Gleason, 2002). If initial message cues, or the context, suggest that a partner may be attractive in some way, CMC users experience exaggerated perceptions of the partner’s appeal. That is, if a partner initially seems to be smart, funny, similar in some way, or even a member of the desired age and sex, people may overgeneralize from some of these characteristics into a more global positive perception, in the manner of a halo effect. The reciprocal interaction among CMC users enhances the dynamic between selective self-presentation to and idealization even more favorable levels. Online text-based conversations prompt also a greater proportion of question-asking and self-disclosure to get information about a partner, in a way that is less obvious and undesirable than it is in face-to-face interactions (Antheunis et al., 2012, 2007; Tidwell & Walther, 2002), and which enhances interpersonal attraction online (Antheunis et al., 2012, 2007; Dai, Shin, Kashian, Jang, & Walther, 2016).

Although the hyperpersonal model has garnered empirical support in a variety of contexts, including dating, contrary findings have also appeared in an experiment that directly compared text-based CMC versus videoconferencing in initial interactions. Sprecher (2014) hypothesized that systems supporting multiple nonverbal cue systems lead to more favorable interpersonal impressions and liking than media with fewer cue systems. Her experiment had non-romantic dyads engage in two brief conversations with each other. In the first interaction, dyads used text-only CMC, voice-only CMC, or videoconferencing. The second interaction always employed videoconferencing. After the first interaction, partners felt more closeness when their medium supported more cue systems: Any medium with audio cues was superior to text-based CMC (including videoconferencing, which did not differ from voice).

While Sprecher’s study provides an important contrast to the hyperpersonal model’s prediction, its potential applicability raises questions. First, anticipating a second conversation via an audiovisual Skype call impels the kind of positive online relational dynamics that have been associated with the anticipation of continued text-based interaction (Walther, 1994) or with an anticipated modality switch from online to face-to-face interaction (Gibbs, Ellison, & Heino, 2006). Second, the social context in which the experiment took place did not involve any particular motivation to cultivate liking. In contrast, other research finds that, in online contexts where social goals are salient, text-based CMC users’ express and achieve hyperpersonal levels of affection rather quickly (Roberts, Smith, & Pollock, 1996). In dating contexts, the social goal of finding a romantic partner impels individuals to optimize their self-presentations and relational orientation even more strongly (see e.g., Ellison, Heino, & Gibbs, 2006; Whitty, 2008), which should implicate hyperpersonal processes and greater levels of attraction. Therefore, we posit the following hypothesis:

H1:

Initial interactions in an online dating context via text-only CMC engender greater (a) social attraction, and (b) romantic attraction than initial interactions via videoconferencing.

Modality switching

Once the initial contact is established online, and people are still interested in each other, partners decide if they want to pursue a face-to-face meeting. If the impressions are indeed idealized in an online chat, one would expect a drop in attraction, as the positive impression might not live up to expectations based on the chat.

A number of studies have shown that the hyperpersonal relations that develop using text-based CMC actually diminish when CMC users switch to other media that offer more nonverbal cue systems. A study of virtual student groups found that members who worked together entirely online using only text-based messaging over the course of a semester experienced greater affection and social attraction for each other than did groups who worked together all semester but saw a photo of one another near the end of their association (Walther, Slovacek, & Tidwell, 2001). Photographs of CMC partners show that some people are more attractive and that others are less so, which normalizes interpersonal impressions, in comparison to the positive bias that CMC users accrue when using only text.

Studies showing non-romantic dyads’ disappointment when switching from CMC to face-to-face interaction reflect this phenomenon. Ramirez and colleagues’ (Ramirez Jr & Wang, 2008; Ramirez Jr & Zhang, 2007) studies on modality switching show that people who start to communicate via text-based CMC form more favorable relationships, initially, than do people who start off with face-to-face communication. When these people switch modalities from CMC to face-to-face interaction, depending on how much time they spent using CMC, the effect is even more stark. People who began their relationships using CMC are disappointed with their partners when they meet them face-to-face, and even more so the longer they interacted online before making the switch. Face-to-face interaction does not afford the idealized and exaggerated impressions and intimacy that getting acquainted via CMC allows.

These effects also take place in the dating context, as shown in two recent studies. In a retrospective survey among online daters, Ramirez Jr et al. (2015) found that people who get acquainted with a prospective date through text-based CMC are more often disappointed when they experience their first face-to-face encounter than do people who started off face-to-face (depending on how long they waited to meet). In a longitudinal study among online daters, Sharabi and Caughlin (2017) asked participants to report perceptions about a partner’s social and physical attractiveness, and their likelihood of having a continuing relationship, both before and then again after their first face-to-face meeting. Both social and physical attraction declined after the face-to-face meeting, but the declines were moderated by the amount of attitude similarity daters perceived that they had with their partners, and the amount of uncertainty reduction that the intervening communication had provided. The level of precision this study adds to understanding the modality switching effect in online dating is valuable. However, the moderators that it explored involved only psychosocial characteristics. They did not examine the possible moderating effect of exposure to the physical nonverbal characteristics that daters claim to value and which videoconferencing could provide. The potential of audiovisual versus text-based CMC to moderate the difference between levels of attraction after communicating online versus communicating face-to-face prompts the second hypothesis:

H2:

The difference between perceptions after an initial CMC interaction and those occurring after face-to-face communication is moderated by CMC modality as follows: When initial interactions occur using text-only CMC, a) social attraction, and b) romantic attraction decline after face-to-face interaction more so than when initial interactions occur using videoconferencing.

In other words, perceptions of partners formed through initial text-based interaction become more negative after a face-to-face meeting, but perceptions formed via videoconferencing do not.

Method

Sample

To investigate these hypotheses in an online dating context, our experiment took the form of a speed dating event in which participants interacted with each of their prospective dates twice: The first interaction involved either text-based CMC or videoconferencing, and the second took place face-to-face. Participants were recruited on a university campus via posters and flyers. Heterosexual students that were single and looking for a relationship could sign up via a website for one of four sessions. In order to keep the proportion of men and women the same, each session involved five men and five women, creating 25 unique pairings per session. One participant failed to attend. Hence, a total of 39 individuals, between 18 and 27 years of age (M = 21.53, SD = 2.58) participated in the experiment, resulting in 95 unique pairs each of whom met twice and evaluated their partners each time, yielding 380 evaluations.

Procedure

To conduct the experiment in its most natural way, a speed dating event was organized in a manner similar to how people meet each other during actual speed date sessions. In a typical speed date event, individuals interact with multiple others in brief (typically 5 minutes or less) dates and for each date – anonymously – indicate their willingness to see the other again. Afterwards, all matches (i.e., dates in which both individuals indicated to be willing to see each other again), are provided their mutual contact information (Finkel & Eastwick, 2008)

Speed dating has several advantages as a research venue (see also Finkel, Eastwick, & Matthews, 2007). Regardless of allowing experimental control, a speed dating event enables us to collect data in a natural setting, resulting in higher external validity compared to the frequently used lab experiments. Consequently, participants who volunteered for the event may be assumed to have some interest in meeting prospective dates, presumably more so than student subjects who might volunteer solely to receive course credit or payment. Second, even though in survey and other quasi-experimental procedures one can ask for – and control for – all the communication modes used, the speed dating venue that the present study employs also allows researchers to be confident about what modes of communication the participants used in each phase of the event, without constraining them.

Participants completed a questionnaire prior to the event which asked some demographic information and control questions, such as age, relationship status and sexual orientation. Upon arrival, female participants reported to a different location than the male participants, in order to preclude them from meeting each other prior to the experimental sessions. To optimize the locations for this event, we made the rooms convenient and fun by providing drinks and snacks, and the rooms were only accessible for participants of the event. Participants were randomly assigned to one of the experimental conditions for their first meeting with prospective dates: text-based CMC (i.e., chat; 100 unique dates) or videoconferencing (i.e., Skype; 90 unique dates). In the text-based CMC condition, participants interacted exclusively via an instant messaging-type software designed for the experiment. In the videoconferencing condition, participants interacted through Skype, using a headset with a microphone and a webcam. Researchers confirmed after each session that none of the paired partners knew each other beforehand and did not see each other before the interaction started.

Before each session started, the participants got a brief instruction about the event. They were told that the event consisted of two parts, and that in each part they would have five conversations for three minutes¹ each with prospective dates. In the first part, the conversation would take place via text-based chat or Skype. After each conversation, the participants evaluated their prospective dating partner by filling out a short questionnaire measuring social and romantic attraction on Time 1.

After the five different initial conversations, the male and female participants were guided to a single location for the second part of the experiment, in which all the participants met the same dating prospective dating partners again, but in a face-to-face meeting. For each unique dyad this was their second interaction. The order in which each individual met with each partner face-to-face was the same as for the online conversations in the first part of the experiment. Moreover, after each face-to-face conversation, participants completed the same short questionnaire in order to measure social and romantic attraction on Time 2. The total duration of the event was less than two hours, which is optimal in order to stay focused and interested the whole time (Finkel et al., 2007). After the event was finished, they were debriefed.

Measurement

Social attraction

Three items measured social attraction, based on McCroskey and McCain’s social attraction dimension of the interpersonal attraction scale (1974). This subdimension differs from the task attraction and physical attraction subscales, and focuses on generalized liking. The items were: “This person could be a friend of mine,” “I could have a pleasant conversation with this person,” and “This person could not be a friend of mine.” The response options ranged from 1 (completely disagree) to 7 (completely agree). The three items formed a unidimensional scale (explained variance 66% at T1 and 71% at T2), with a Cronbach’s alpha reliability of .73 at T1 (M = 5.00, SD = 1.17) and .79 at T2 (M = 4.94, SD = 1.21).

Romantic attraction

The measure of romantic attraction comprised seven items, measuring physical attraction, romantic attraction, and sexual attraction based on existing scales by McCroskey and McCain (1974), Townsend and Levy (1990), Eastwick and Finkel (2008). The items were: “This person is physically attractive,” “I feel attracted towards this person,” “This person is handsome,” “I would like to go out with this person,” “Romantically speaking I am interested in this person,” “I would like to have a relationship with this person,” and “I would like to have sex with someone like this person.” The 7-interval response options were again 1 (completely disagree) to 7 (completely agree). Despite the potentially multidimensional scope of the measure, the seven items formed a unidimensional scale (explained variance 77% at T1 and 79% at T2), α = .95 at T1 (M = 3.42, SD = 1.30) and .96 at T2 (M = 3.09, SD = 1.32).

Results

Hypotheses tests involved multilevel linear analysis using the SPSS MIXED procedure using maximum likelihood estimation. Each of the 39 participants interacted with five different partners at two different times, filling in the survey after each interaction. This yielded 195 observations after T1 and another 195 observations at T2, resulting in 390 observations. As each participant contributed 5 cases to the data at T1 and five at T2, observations are not independent from each other. Moreover, as participants interacted in pairs, it was important to factor in possible correlation between the dyad members. We therefore conducted multilevel analyses using the MIXED procedure in SPSS. Both participant and dyad were included as subject factors in the model and we allowed the intercepts to vary randomly in order to correct for the nonindependence between observations. Because all hypotheses involved directional predictions, results are reported using 1-tailed p values.

Hypothesis 1 posited that (a) social attraction and (b) romantic attraction toward a partner are greater after initial interaction via text-only CMC than after initial interaction via videoconferencing. For this analysis, we used only the T1 data. We conducted two analyses, one for social attraction and one for romantic attraction. In each analysis, condition was included as a factor, while dyad and participant were included as random factors allowing the intercept to vary randomly to correct for non-independence. Results confirmed that social attraction was greater in the text-only CMC condition (M = 5.25, SD = 1.00) than in the videoconferencing condition (M = 4.71, SD = 1.29), t(44.78) = 1.953, p = .029. However, the effects of communication condition were not significantly different for romantic attraction, t(37.91) = 0.226, p = .412. See Table 1 for means and standard deviations for all dependent variables across media conditions for Time 1 and Time 2. Therefore, H1a was supported while H1b was not.

Table 1. Means and standard deviations for social and romantic attraction at T1 and T2 for gender and both text CMC and videoconferencing conditions

		T1		T2
		Men	Women	Men	Women
Social Attraction	Text-based CMC (T1)	5.18	5.32	5.35	4.97
		(0.96)	(1.05)	(0.85)	(1.35)
	Videoconferencing(T1)	4.98	4.44	4.95	4.44
		(1.27)	(1.26)	(1.27)	(1.20)
Romantic Attraction	Text-based CMC (T1)	3.76	3.17	3.33	2.90
		(1.36)	(1.01)	(1.15)	(1.24)
	Videoconferencing (T1)	3.62	3.15	3.34	2.78
		(1.41)	(1.35)	(1.50)	(1.38)

Hypothesis 2 posited that after initial interaction via text-only CMC, subsequent interactions via face-to-face communication reduce social and romantic attraction, but this effect is not predicted for those who had their initial online interaction using videoconferencing; those who initially used videoconferencing will have been inoculated about their partners’ physical appearance. This hypothesis test involved a similar procedure as for H1 but it involved the observations from both T1 and T2, using time (i.e., T1 vs. T2) as a repeated measure in the analyses. The analysis also controlled for the interdependence between T1 and T2 measures by including participant and dyad at T1 and T2 as a random factor and allowing the intercept to vary randomly.

For social attraction there was a significant effect of condition, t(55.26) = 1.732, p = .044. Participants who communicated via text-based CMC at T1 still were more socially attracted to each other after T2 than participants who communicated via videoconferencing at T1, indicating that hyperpersonal effects persisted after T1. For social attraction, there was no significant effect of time, t(189.56) = 0.369, p = .357, nor was there a significant interaction effect between time and condition, t(189.42) = 0.401, p = .345. Results did show a significant effect of time on romantic attraction, t(186.07) = − 3.678, p < .001: Romantic attraction declined after the second interaction. However, no effects were found for communication condition, t(42.27) = 0.303, p = .382, or for the interaction effect between communication condition and time, t(185.93) = − 0.132, p = .448. Univariate analysis showed that the effect of time was significant for both those whose initial interaction occurred via text-only CMC, t(185.77) = − 3.703, p < .001, as well as those who had an initial audiovisual interaction, t(186.07) = − 3.688, p < .001. The moderation aspect of the hypothesis was not supported. Romantic attraction also declined after the second meeting, but this change was not moderated by the initial communication condition.

Gender effects

We also checked for possible differences between men and women. For H1, there were no main effects of gender, t(34.58) = 1.263, p = .215, or interaction effects between gender and condition, t(33.85) = 0.237, p = .407 for romantic attraction and there were no main effects of gender, t(34.82) = 1.615, p = .058, or interaction effects between gender and condition, t(34.37) = 1.454, p = .077 for social attraction. Pairwise comparisons for the interaction between gender and condition, however, showed that for women, there was a significant difference between the videoconferencing and the text-only CMC condition, t(43.02) = 2.413, p = .010, while for men, there was no such difference, t(38.02) = 0.525, p = .302.

For H2, there was no main effect of gender on romantic attraction, t(41.58) = 1.380, p = .087, nor any interaction effects of gender with time or condition, t’s < 0.60, p’s > .25. For social attraction, there was a significant 3-way interaction effect of gender, time, and condition, t(189.43) = 2.200, p = .015. Pair-wise comparisons showed that these results mirrored those for H1. At T1, there was a significant difference between the videoconferencing condition and the text-only CMC condition for women, t(47.33) = 1.380, p = .008, but not for men, t(49.19) = 1.380, p = .301. There was no main effect of gender, t(40.05) = 1.465, p = .076, and no further interactions between gender, time and condition, t’s < 0.50, p’s > .35.

Discussion

This research examined the effects of different initial CMC systems on attraction both before and after a face-to-face meeting in a speed dating event. The initial conversation systems differed in the kinds of communicative cues they supported, from text-based CMC with only language and typographic cues, to videoconferencing with physical appearance, voice, and facial expression accompanying language. The second conversation took place face-to-face.

Different theoretical approaches and previous studies have lent conflicting predictions and empirical precedents about the superiority of one type of system over another for initial conversations and for modality switching over time. The hyperpersonal model of CMC, for instance, argues that certain affordances of text-based CMC lead users to develop idealized perceptions of one another and enhance relationships. From this perspective, modality-switching from CMC to face-to-face interaction can lead to disappointing interpersonal outcomes. Other studies suggest that audiovisual communication more fully supports interpersonal processes and the presentation of physical features that are likely to be critical to online daters’ evaluations of prospective partners. From that perspective, modality-switching from videoconferencing to face-to-face would mitigate the disappointment that seems often to accompany a modality-switch from text-based CMC to face-to-face interaction.

The results provided general support for the hyperpersonal model’s predictions for the difference between text-based and audiovisual initial conversations’ effects on initial levels of social attraction – the kind of attraction that accompanies wanting to be friends and socialize with someone. In this case, text-based initial interactions provided greater social attraction than did audio-visual CMC. It indeed seems that the reduced cue systems offered by text-based CMC allow users to optimize their self-presentations, leading to idealized impressions of the attraction of each other in a speed dating session. Post-hoc analysis showed that this hyperpersonal effect can be attributed to women and not to men.

The hypothesis had also predicted that romantic attraction – attraction based partly on impressions of physical characteristics – would be greater following an initial text-based CMC conversation than it would following one using videoconferencing. The rationale for this prediction was that text-based CMC users would idealize their partners’ physical characteristics if they saw no contradictory visual indications, and that those who used videoconferencing would experience a range of physical attractiveness among their 5 speed dating partners, the average of which would be unremarkable. However, the results did not support this prediction, as there was no difference between text-based CMC and videoconferencing on romantic attraction.

The second hypothesis predicted that initially using text-based CMC versus videoconferencing would moderate the effects of modality-switching, attenuating the disappointment daters experience when moving from online to offline. This should have been the case particularly with respect to romantic attraction, since that construct is based in part upon physical attraction, cues to which should be more abundant in videoconferencing. In partial support of our hypothesis, romantic attraction did decline after the subsequent face-to-face interaction, although social attraction did not.

Overall, initial conversations via text-based CMC yielded greater social attraction, which persisted even after a face-to-face meeting, in contrast to general modality-switching syndromes reported in the literature. Romantic attraction did not initially differ due to CMC variations, and it declined after a face-to-face meeting regardless of the initial CMC variation.

The unanticipated persistence of the social attraction-enhancing effect of text-based CMC deserves further consideration. One possibility is that text-based CMC of a romantic nature shifts the basis of attraction from physical to something more psychological. There is abundant literature suggesting that romantic couples get to know and value one another’s in text-based CMC (e.g., McKenna et al., 2002). In non-romantic contexts, research on the use of social networking sites shows that people reduce uncertainty about each other, leading to greater liking, through interactive text-based CMC discussions and not by looking at each other’s profile photos that reveal physical characteristics (Antheunis et al., 2010). Sharabi and Caughlin (2017) also found that uncertainty reduction, accrued through online interaction, was negatively associated with reduced attraction daters experienced after meeting face-to-face. It seems likely that couples who conversed using text-based CMC had nothing to do but reduce uncertainty and seek similarity, without physical distraction, and it worked positively for them. In contrast, being exposed to one another physically using videoconferencing may have introduced a physical appearance heuristic that lead videoconferencing users to be blinded by physical characteristics and distracted from more meaningful forms of uncertainty reduction. Such distraction may have short-circuited conversational efforts to discover attitude similarity, a condition that remains among the most potent predictors of attraction in the long run, both conventionally (Byrne et al., 1968) and in online dating (Sharabi & Caughlin, 2017). This explanation could account for the persistence of elevated social attraction across both initial and subsequent conversations that began with textual CMC.

The unanticipated finding that the hyperpersonal effect was particularly found for women in this study also deserves further consideration. Though several studies found gender differences in the amount and depth of self-disclosure (e.g. McKenna et al., 2002), this is the first study that found a difference in the hyperpersonal effect between women and men. As we found this difference for social attraction, it can be explained by evolutionary psychology. Men value physical attractiveness in women more than women do in men. Hence, men tend to seek more for physical attractive partners, whereas women seek more for emotional bonding (e.g., Toma et al., 2008). Hence, for men the physical features are likely to be more critical to their evaluations of prospective partners.

While the principles articulated in past research aspired to generalize across settings, few previous studies took place in a date-seeking context. Yet the online date-seeking context provides a real-life setting in which it is common to get acquainted in two stages, the first using restricted modalities, leading to a second, face-to-face encounter. Therefore, research looking at different cue affordances and comparisons over several episodes provides the potential not only to address theoretical conflicts but to help understand a contemporary communication problem as well. A speed dating setting provided the means by which to test competing theoretical predictions and to provide a realistic setting in which to do so.

Implications of the study

Our study has several theoretical and societal implications. First, our results provide support for the hyperpersonal model and modality switching in the context of online dating. There was more social attraction after a first meeting via text-only CMC compared to videoconferencing. Furthermore, after a modality switch to a face-to-face encounter the hyperpersonal effect persisted for social attraction, while romantic attraction declined. So, in a dating setting, in which physical attractiveness is more important compared to initial interactions in the context of friendship formation, people develop idealized impressions of one another using text-based CMC. Furthermore, contrary to other studies (Ramirez Jr et al., 2015; Sharabi & Caughlin, 2017), this idealized impression persisted when they met in person. Our results showed that the hyperpersonal effect persisted when couples in the text-based CMC condition met each other in person.

Second, this study also provides some support for sociobiology and evolutionary psychology, as it shows that physical appearance of the potential date is of greater importance for men then for women. In line with studies on online self-presentation (e.g., Hancock & Toma, 2009; Toma et al., 2008), this study proves that gender differences in online dating and hyperpersonal effects can be predicted by evolutionary theories. Even though our gender effect in the hyperpersonal effect nicely fits theories in evolutionary psychology, future research should examine gender differences in hyperpersonal effects more thoroughly as in earlier research this difference was either not found or was not examined.

This study also yields implications for online daters. Ironically, the results of this study bear out the benefits of using videoconferencing rather than text-based CMC prior to meeting a prospective date face-to-face, but for reasons that are diametrically opposed to those which most often appear in the literature. Most studies comparing textual CMC with videoconferencing (e.g., Sprecher, 2014), or even comparing textual CMC with the use of avatars (e.g., Kotlyar & Ariely, 2013), claim that the additional nonverbal cues that the latter media support raise the level of attraction that prospective daters want and need when choosing a date they wish to meet face-to-face. The present results also recommend that videoconferencing is beneficial, but not because it raises attraction. Quite the contrary: Videoconferencing may help people develop the more realistic perception of prospective partners’ social attractiveness, and their (un)desirability for dating, that a subsequent face-to-face encounter would also be likely to yield. In very stark contrast to a predominant view in the literature, our results suggest that videoconferencing and its representation of nonverbal cues is apparently very good at making clear the unattractive qualities of prospective dates and helping daters to decide to screen them out, rather than audiovisual communication fanning the flames of desire. Text-based CMC, on the other hand, leads individuals to hold out hope, and perhaps kindle an otherwise unlikely dating trajectory.

Limitations and suggestions for future research

Some readers may find limitations in the current work in terms of potentially hidden effects of individual differences that might affect the kind of attraction person conveys to others or that affect the kind of attractiveness judgments an observer may make. For instance, variations in participants’ innate physiognomy can affect physical attractiveness judgments, which can, in many cases, have important impacts in prospective dating decisions (e.g., Toma & Hancock, 2010). Yet we generally assume that physical characteristics are randomly distributed among participants, meaning that media generated greater systematic differences than random variation in physiognomy. For physical characteristics to have had an effect on the present results – reduced attraction after videoconferencing – could only suggest that the participants in this study were physically uglier than a normal random sample would yield, which is possible but quite unlikely. The possible interaction effect of individual differences in physiognomy and communication mode, however, is an interesting possibility that deserves future research. The larger point to remember is that, it once again appears, that communication modality affects perceptions of attractiveness despite natural variations in appearance. In terms of other potential factors, such as dating experience, may have had unknown effects; whether and how these factors interact with modality differences also deserve future research.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. A 3-minute speed date might seem short but is common in speed dating events (i.e., most of them are between 3- and 5-minutes; Finkel et al., 2007) and gives the participants enough time to form an impression. It is known from impression management research that impressions are commonly formed based on minimal contact and/or on little information (Allport, 1937; Carney, Colvin, & Hall, 2007; Eastwick, Finkel, Mochon, & Ariely, 2007; Stecher & Counts, 2008). Based on minimal exposure, people make broad generalizations about another person, in this case their date. Allport (1937) described this as Thin Slices, which are short chunks of social behaviour from which people can form impressions. In 2007, a study of Carney et al. (2007) that people are able to form accurate impressions in 5 seconds, though this depends on the sort of impression. Negative accurate impressions were formed after 5 seconds, while for positive impressions it took a bit longer. They found that 60 seconds generated the optimal ratio between accuracy and slice length. Also, Walther’s hyperpersonal model (1996) states that people tend to rely on limited information that is available in their impression formation. Moreover, in online contexts where social goals are salient, text-based CMC users’ express and achieve hyperpersonal levels of affection rather quickly (Roberts et al., 1996).