Ignore the Affordances; It's the Social Norms: How Millennials and Gen-Z Think About Where to Make a Post on Social Media

How do Millennial and Gen-Z young adults decide between competing social media when choosing where to post? Previous research argues that decisions can be modeled based on these users' Goals, and the Affordances, Features1, and Social Norms of those media. To evaluate this model, 19 participants were given different self-presentation scenarios and asked to choose between Twitter, Snapchat, Instagram, and TikTok. Participants also constructed a flowchart to represent their mental model of factors influencing their choices. Data suggest revisions to the model; rather than reflecting Affordances and Features, we found that Social Norms and Goals primarily drove participants' choices. Participants choose media based on a direct match between Goals and existing Social Norms. Only when they failed to detect such a match did participants consider Affordances and Features. We present a revised conceptual model based on these results and discuss social media design and theory implications.


INTRODUCTION
Many people use multiple social media in their daily lives.In 2021, Pew research found that 72% of American adults have used at least one social media site, while younger adults are often active on multiple social media sites concurrently, such as Instagram (76%), Snapchat (75%), Twitter (42%), and TikTok (55%) [2].But given the multiplicity of media, how do these young adults, consisting of both Millennials (born early 1980s-mid 1990s) and Gen-Z (born mid-late 1990s-mid 2010s), who are using multiple media decide exactly where to post?According to previous research, these users consider multiple factors [7,69].Decisions often involve balancing Goals [8,42,66], media Affordances and Features [9,16,24,25], and Social Norms [3,15,71].This paper derives a media choice model from previous literature and evaluates it using scenarios, flowcharts, and interviews to examine how young adults from these generations interpret these different factors when choosing where to post.
Prior work argues the following factors influence media choice: Goals, Affordances, Features, and Social Norms.There are many potential user Goals for social media use [8,39,62,65] but these often relate to self-presentation [4,14,29,35,54,67].Given the prevalence of self-presentation as a Goal, this paper evaluates social media choices for two specific self-presentation Goals, Idealized and Authentic self-presentation.Both these Goals are common on social media but have different outcomes [14,31,46,54,61,71,74,77]. Affordances and Features are also important determinants of media choice [4][5][6]9,21,70].Affordances are how people perceive a social medium in terms of interacting with it, i.e., what users perceive as possible on a given medium [26,28,56].In contrast, Features are the immutable, objective properties of the technology that cannot be changed without transforming the medium [52,65,70].For example, speech technologies like the phone call have different Features from video technologies such as Zoom because video provides information about participants' gaze, facial expression, gestures, and posture is unsupported by the phone [27,76].Social Norms are another determinant of media choice.Social Norms are defined as general, often implicit group rules for what is acceptable on a specific social medium [15,19,55,71].Social Norms help regulate behavior and manage user expectations [36].Success in a social medium necessitates following Norms due to the penalties for Norm violations [11,60].
The research described above generally investigates the impact of a single factor on a single social medium rather than how multiple factors might influence medium posting choice [3,4,14,39,42,45,53,62,66,73].However, a few exceptions consider a media ecology perspective, choosing between multiple media depending on underlying factors [7,69].These studies argue that users choose a social medium to fulfill a particular Goal while navigating the Features, Affordances, and Social Norms of their social media ecology.For example, given the Goal of presenting oneself Authentically, Snapchat is a better medium to choose than Facebook, as it allows posting to a small Audience of known intimates [7].However, these works do not evaluate the relative importance of these factors when people make media choices.We developed an initial conceptual model based on existing literature using these factors to address this.We call this the Features, Affordances, Norms, and Goals (FANG) model (see Fig 1).Based on prior research, this model assumes that Goals drive media choice.Given a Goal, users evaluate where to post by comparing the Features, Affordances, and Social Norms of available media.Furthermore, the model implicitly assumes that all three factors are equivalently weighted.This paper aims to deepen our understanding of how young adults who use multiple social media choose between media, using scenarios to build a conceptual model of that choice process.
Authentic scenarios reflected Goals where the protagonist desired a genuine connection with others.Idealized scenarios reflected Goals where the protagonist wanted to express their best self.Our research questions are: 1. Does the FANG model reflect how participants actually choose between social media?
2. What factors influence participants' posting choices on social media?
3. Are any factors more important than others?
4. Are participants consistent in their media choices?
2 LITERATURE REVIEW

Goals and Self-Presentation
Goals of social media use broadly incorporate different motivations and motives, such as social connection, sharing identities, entertainment, etc. [8,39,62,66].We use the uses and gratifications theory to help explain why people might have specific Goals on social media [8,39].Early research on uses and gratifications characterized media such as television or books [43], while more recent research has expanded to include social media [8].Examples of uses on social media include Goals such as entertainment [62], social connection [57], and self-presentation [13].However, little research uses this framework to examine how users navigate multiple media to fulfill gratifications.One exception is a recent study by Tandoc et al., who looked at "poly-media swinging," or how people move (swing) from one medium to another and for what [69].In particular, it revealed that users could and do swing to navigate different barriers to fulfilling their Goals.Users media swing to fulfill self-presentation needs more holistically, such as by sharing opinions on Twitter and activities on Facebook.Swinging also allows them to overcome audience or Social Norm barriers to achieve their Goals.Their research also identifies two main Goals that media swinging can address: self-presentation and relationship management.As self-presentation is a common social media Goal, we now discuss self-presentation, specifically Authentic versus Idealized.Self-presentation is a complex process for people, such that sometimes they use different accounts for different types of self-presentations.[4,14,29,35,54,67].For example, studies of Instagram usage show people present a more Idealized version of themselves on one Instagram account while using another to present a more Authentic version [37,42,47,68].Because of this, a major factor in determining self-presentation is whether the Goals appear Authentic or Idealized.In this study, using scenario prompts and flowchart building, we explore two general selfpresentational Goals that users, especially young adults who are attending college, often attempt to achieve, Idealized and Authentic self-presentation.
Idealized self-presentation is often driven on social media by social desirability or a bias to present oneself through positive and socially acceptable content [14,71,74,77].For example, on Facebook, Twitter, Instagram, and WhatsApp, there are Social Norms for presenting positive emotions over negative or neutral emotions.These Norms are stronger for Instagram than for Facebook and Twitter [75], indicating Idealized self-presentations can vary.Authentic selfpresentations are also desirable at certain times and in specific contexts.Previous research has found positive outcomes for those who can Authentically express themselves online such as less negative and more positive affect [31,46,54,61].However, expressing such Authenticity on social media is often fraught [3].For example, a recent study comparing Finsta vs. main Instagram accounts found that the more Authentic Finsta accounts were less satisfying and useful due to receiving fewer comments [37].Authentic self-presentations allow users to fulfill needs alongside Idealized self-presentations with different accounts or media, allowing users to achieve their selfpresentational Goals.However, outcomes can be complex and reliant on multiple factors; another study found greater negative affect from imagining criticism for Instagram users who present a false self on Instagram than users who present a more Authentic self [38].We chose to evaluate Authentic and Idealized self-presentations because they represent drastically different Goals, with different potential presentations related to them.

Affordances, Features, and Media
Affordances are a useful lens to understand how and why people use different social media [4][5][6]9,21,70].To reiterate, Affordances relate to how people understand how to interact with something, in this case, a particular social medium [56].By contrast, Features are the technology's basic, immutable, objective properties that cannot be changed without fundamentally changing the medium.Affordances allow people to understand the possibilities of interaction with a technology [16,24,25].For example, anonymity is an affordance that describes the degree to which a user perceives they can hide their personal information in public or private media [25].This affordance of anonymity is constructed from a set of features.For example, users may feel that Snapchat has greater Anonymity than Facebook due to differences in sets of Features (i.e., using a real name on Facebook vs. not on Snapchat, snaps disappearing by default instead of staying by default, less vs. more control over who sees a post, etc.) [4,24,50,58].This relationship between Features and Affordances is not one-to-one; a set of Features (i.e., all of the Features in Instagram Stories) may lead to an Affordance rather than a single Feature (i.e., posting photos).
There are a variety of Affordance frameworks that describe Affordances specifically on social media [16,24,25].We use Fox and McEwan's Perceived Social Affordance framework [25] in this study to help describe the Affordances people use on social media.For example, Personalization (the ability to focus messages to a specific person or group of people) or Anonymity (the ability to keep oneself anonymous when interacting with others through the medium).Such Affordance frameworks are useful conceptually as they allow for comparison across media, helping to support research that examines multiple media.
The current paper looks at four media (Instagram, Twitter, TikTok, and Snapchat) with the following Features.Instagram is image-based, focused on creating albums of pictures or videos that go to a large group of people [1,68,77].Twitter is primarily text-based, although around 2015, short video posting was added as a Feature [34].Tweets go out to large groups of followers and can easily be retweeted to others, allowing a post to have a vast potential audience [53].Snapchat is image and video-based, but its key Feature is ephemerality.Posts made to it by default disappear after a certain period, and posts generally go to a smaller audience, with the poster having more control over who sees what posts [4,58,73].Finally, TikTok is the newest social medium of this group.As of 2020, it has recently exploded in popularity in the US.The content is primarily short form (15-60 sec) videos where users can add videos or audio of themselves to an existing video as a response or "duet" [3,45].Also of note is the "For You" page, where users can view a stream of algorithmically selected content based on their viewing preferences [64].Next, we will discuss an important element of the social space of social media: Social Norms.

Social Norms
Social Norms are broad, often unstated rules for what is acceptable on a specific social medium [15,19,55,71].Social Norms are vital for communities as they help regulate behavior and manage expectations in social spaces [36].However, Social Norms can be complex, especially in social PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023.media, and sanctions for violating Social Norms might be implicit.For example, on TikTok, it is socially acceptable to present a wider range of emotions, even negative ones that would be selfcensored on other media.This allows users to express different Social Norms around Authenticity compared to other media [3].However, a similar post on another medium such as Facebook might be ignored or sanctioned subtly [60].Social Norm sanctions are important for maintaining communities; pushing back against non-normative behavior helps reinforce existing Social Norms [49].Another study found that people have complex and often implicit strategies for seeking social support, as overtly seeking support is a Social Norm violation on Facebook [11].
Due to this complexity, there may be issues with navigating a complex web of Social Norms across different media.The theoretical concept of profile work [63,71] helps capture this issue through conflicting Goals, where people's desires run against the Social Norms of a medium.As discussed, an example of one's Goal conflicting with Social Norms can take the form of sharing overly emotional content in a place where other users will not accept it.The person posting may know they are violating a Social Norm but still proceed because the support they could receive is worth it.Another issue is that Social Norms are dynamic, and it can be difficult to navigate the variety of contextual Social Norms layered into a medium.This issue can lead to context collapse [53], where the unclear boundaries between communities can unintentionally lead to Social Norm violations; a college student sharing photos of drinking with friends on Facebook might get disapproving messages from family members [11].
Another factor that could lead to differences in media choice is tie strength.Social networks (both online and offline) have differences in the strength of a person's connections to others they know, also known as tie strength [10,20,30,48].Social networks on social media sites can differ depending on the site, such as having a larger group of weak and strong ties on Facebook while having a smaller group of strong ties on Snapchat [4,67,72].Although this could be a factor that leads to differences in use and social media choice, we chose to focus on Social Norms as these encapsulate broad sets of behaviors related to Goals.
Overall, prior work identifies important theoretical and empirical factors in how social media users make choices between different social media.It also suggests that Goals, Affordances, Features, and Social Norms potentially influence different choices in a user's social media ecology.Our work adds to this by identifying multiple important factors in social media choice and empirically evaluating them through posting decisions for different self-presentation Goals.We examine how young adults' evaluations of these different factors lead to different posting choices, which factors our participants use more in their decision-making process and evaluate the initial FANG model based on participant experiences to better capture how Millennials and Gen-Z make their decisions.

Participants
To evaluate the FANG media choice model, we recruited participants using Prolific [78], an online participant recruitment site.All participants met the following criteria: graduate or undergraduate student, users of at least three of the following four media (Snapchat, Instagram, TikTok, Twitter), aged 18+, and currently in the US.We chose the four social media as they are currently being actively used by Millennial and Gen-Z young adults.Although we discuss Facebook in the literature review, we did not include this as one of the social media of interest.This was due to two reasons: previous research noted that people from these generations were less likely to actively use Facebook [67].Since we wanted to learn about posting behavior, we wanted to include social media that current young adults, particularly college students, were likely to be using.Additionally, we wanted to expand social media research beyond Facebook as a "default" social media while exploring additional popular examples.
Participants were required to be in the US by IRB, and participants were students to help with sample consistency.Participants received $14 compensation upon completion and completed the interview in ~1 hour.The final sample was 19 participants.They self-identified as 12 women, 5 men, and 2 non-binary, 42.1% were White/Caucasian, 21.1% Asian/Asian American, 15.8% Mixed Race/Ethnicity, 10.5% Hispanic/Latinx, 5.3% Black/African American, and 5.3% Native American, Alaska Native.Their ages ranged from 18 to 37 (M= 24.1, SD= 5.4) indicating that our participants were members of the Millennial and Gen Z generations.The average duration of the study sessions for participants was 1:09:30, with a range of 46:11 to 1:46:24.Participants were recruited until we received saturation on themes within our sample.We stopped recruitment at 19 for this reason.Additionally, we wanted to recruit participants who used at least three of the four social media of interest, which took more time.We wanted to make sure that results from later participants were still comparable to those earlier.
We asked participants about their posting and viewing frequency for each social medium.Table 1 shows participants were frequent media users and, as expected, more likely to view media than post.All participants should therefore be familiar with typical posts on each medium, with some higher-intensity users posting and viewing very frequently.Additionally, participants did not describe themselves as business owners or influencers, but average users of the media.While all almost all users were viewing each of the social media studied at least once or twice a month, few posted more frequently than once or twice a month, which does not support the idea of participants being power users such as business owners or influencers.They did not mention in interviews that they had an unusual number of followers or followees on any of the social media discussed.The study was reviewed by the first author's university IRB and was approved as exempt.

Procedure
After completing a consent form, participants were interviewed online following a structured protocol.Four phases followed: (1) four choice scenarios in which participants were asked to choose where to post between four target media, (2) creating a flowchart to represent decision processes, (3) presentation of four further scenarios, (4) revisiting and modifying the flowchart.First, we presented participants with 4 hypothetical posting scenarios in one of two different categories: Authentic or Idealized self-presentation.The overall objective of the scenarios was for participants to choose a medium or multiple media, provide a justification for their choice, and explain why they did not choose other media.We designed the scenarios over multiple rounds of revision based on examples from each medium (see Appendix A).Scenarios were agnostic about media formats as pilot testing revealed that explicitly describing posts as text/photos/video biased participants' choices.We deployed a within-subjects design.Participants were randomly assigned to receive four Authentic or Idealized scenarios first, and we randomly presented scenarios within these categories to eliminate order as a possible confound.The protagonist's gender in the scenario was matched to the gender of the participant to increase participant engagement and to avoid potential cross-gender effects [51].Participants read each scenario and were asked, "Where do you think [the scenario protagonist] should post this?".Responses took the form of a conversation, allowing both participant and interviewer to clarify.The interviewer verbally listed potential social media at the start of each phase.Participants were encouraged to choose one social medium but could choose multiple if they made an argument for this.They then answered follow-up questions, such as "Why do you think [protagonist] would choose to post to that medium?" (See Appendix B for list of questions).Participants were given four scenarios before constructing a flowchart, as we wanted them to first experience making decisions and thinking about media.
In the second phase, participants created a flowchart that identified and organized elements influencing their media choices (See Fig. 2).Elements represented concepts written on virtual cards, such as audience*, going viral, or present a perfect self (*italics indicate the text of an element).Participants responded to the prompt "What do people consider when making a post" by creating a flowchart.They generated their flowchart using a Miro board [79] (a shared whiteboard space where participant and interviewer can simultaneously write and move virtual cards freely) to write elements and draw connections between them.The interviewer first created an initial set of unorganized elements, describing concepts the participant mentioned when discussing the scenarios during phase 1.The interviewer then encouraged the participant to edit, supplement and organize these initial elements as they saw fit.We explicitly told participants to place more important elements at the top, with less important elements below.Following similar studies [22,23], participants were asked to think aloud while constructing the flowchart, and to review their completed flowchart.This process follows existing work to construct conceptual hierarchical flowcharts (see [17,40,44]).
The third and fourth phases repeated the first two phases.In the third phase, participants answered scenarios from the category they had not seen previously (i.e., those who first saw Authentic would now judge Idealized).In the fourth phase, participants revisited the flowcharts generated in phase two.We encouraged them to add, change, or remove any elements from their existing flowchart based on the new scenarios that they had just discussed.We also presented participants with additional elements to consider based on Affordances drawn from previous work [25].As before, we asked participants to think aloud while they arranged the elements and explain the construction of their flowchart once completed.Finally, they answered some demographic questions and finished by debriefing.
All interviews were automatically transcribed using Otter.ai.The first two authors independently read through or listened to each interview and conducted a deductive independent thematic analysis of both the scenarios and flow charts; they collaborated on proposed and overarching themes.First and last authors revisited and finalized the themes best represented in the data.

Classifying the Elements Influencing Participants' Decisions
We first analyzed our participants' conceptual models of media choice, represented in their final flowcharts.Our process for categorizing flowchart elements was top-down, using concepts drawn from previous literature: Goal, Social Norm, Feature, and Affordance over multiple rounds.We aligned each element in the flowchart with the interview discussion of that element and any related elements.This allowed us to categorize the element into one of the four categories.As described above, we asked participants to put the most important flowchart elements at the top.If a flowchart contained multiple top elements, we determined the most important element based on how they ranked those elements in the interviews.
After identifying the most important elements in our sample's hierarchical flowcharts, we confirmed the importance of Goals and Social Norms in the FANG model.Participants overwhelmingly identified Goal or Social Norm based elements as the most critical determinants of their media choice.Participants frequently felt that a user's overall Goal is a critical determinant of where to make a post.However, despite this strong focus on Goals, many others mentioned Social Norms as their most important element.Out of the 19 participants, 13 of the most important elements were Goal-and 6 were Social Norm-based.Additionally, flowcharts were complex, with nearly twice the number of elements when participants revisited the flowchart in phase four compared with the flowchart they created in phase two.Flowcharts when first created contained M=12.3,SD=4.4 elements, whereas flowcharts when revisited had M=22.9,SD=8.7 elements.The following sections will first clarify the importance of Goals and Social Norms and then elaborate on the role of Affordances and Features in media choice.

Goals Are Important, And Drive What Factors Young Adults Attend
The FANG model argues for the importance of Goals, and evidence from the flowcharts and interviews confirms this.As discussed earlier, a Goal is a broad motive for why a user might use a particular social medium, usually to achieve some sort of gratification [8,43].In Fig 2, they represent their Goal of "influencing specific audiences" in the element Who I want to reach.Participant P14 explicitly discusses their choice process; they describe how they subclassify their broad Goal depending on the specific post context.The underlying Goal is more complex than the simple textual labels of their flowchart elements, and the interviews provided disambiguating context.For P14, their primary underlying Goal is to influence specific audiences to achieve different gratifications.Each audience they mention identifies a sub-Goal, which invokes different subelements.If P14 wants to reach people they know, then the sub-elements of quick posts, showing emotions, and having fun become important.These sorts of Goals fulfill a gratification of social connection.In contrast, if P14 wants to reach a wide audience, monetization, building a brand, and connectivity become useful for achieving said Goal.The gratification here would be making money or gaining influence.This nuanced presentation of Goals confirms their importance in the FANG model.By showing that our participants break Goals into sub-elements, it affirms that Goals are a strong motivator in their media choice.Subdividing a Goal indicates that Goals are a flexible factor that can drive choice towards different media.In particular medium best supports, young adults can navigate the flowchart to decide which medium best support their overall Goals.
For more support of Goals' importance, let's examine another example.A different participant identifies the Goal of communicating Perfection and Good Times, where the main motivation relates to self-presentation.Although we focused scenarios on self-presentation, some participants considered other Goals when making their flowcharts.Compared to the previous example, this participant felt that it was critical to use media to offer a specific, highly positive, Idealized selfpresentation.
"[...] I was like 'perfection'.So that probably would be my top.Because, you always want to show the best side of yourself on social media.Just because nowadays, social media is everywhere and you always want to make sure it doesn't really bite you in the butt afterwards.[...] And then probably the second one [in the flowchart hierarchy of elements] would be good times because I mean, it also ties along with perfection, because you always want to seem like 'hey, I'm having the time of my life'.It's always like… to me with social media, you're kind of bragging about your accomplishments, or anything that's good happening in your life" (P3).
As in the previous example, the participant clarified the Goal during the interview.Their most important element was to give off a specific self-presentation, that of perfection.Showing their Idealized, perfect side also protects them from online criticism or scrutiny.This example again supports the FANG model's claim that Goals are an important determinant of media choice.For this participant, if a social medium did not support their Goal of showing perfection, they would not want to post to it.

Social Norms are Important and Sometimes Drive Media Choice
We also found evidence that Social Norms (the general unstated rules for what is acceptable on a given medium) were important and also drove media choice.For example, the flowchart element Ability to show emotions describes how comfortable P14 was with sharing their emotions on different media (see the lower center of Fig 2).As discussed in the Introduction, different social media have different Social Norms regarding posting emotional content, especially negative emotions.P14 in the figure uses this element to represent how Reaching the people you know is associated with Social Norms for expressing emotions on this social medium.Another participant (P7) also discusses Social Norms, arguing that these imply direct consequences for suddenly changing the type of content one is posting: "You know, it's like somebody one day, you know, being a nonstop comedy blogger, or Vlogger, some sort.And then out of nowhere, they're posting really deep, really emotional stuff, crying all the time, when you have a whole bunch of followers who follow them to be laughing all the time.And now they're just sad when they watch everything, which is fine.It's up to you, if you want to post that kind of stuff.But if you're wanting people to stay with you, you can't just out of nowhere, poof, and switch everything" (P7).media and knew that violating them could lead to consequences such as negative feedback or suddenly losing followers.Our participants mentioned needing to carefully negotiate Social Norms, so much so that Norms sometimes seem to be the main determinant of media choices.
Recall that in addition to creating flowcharts, participants also reflected on hypothetical scenarios involving media choices.These scenario choices also affirmed the importance of Social Norms.Scenario 1 (see Appendix A for the full text) featured a protagonist who enjoys dancing and wants to show off their skills.Participants were extremely consistent in their responses to this scenario; all stated that the protagonist should post on TikTok (see Table 2 for more details).One participant notes that, due to Social Norms about what content is encouraged, TikTok is the best medium to post dancing: "Oh, that's definitely TikTok.[...] Dancing videos are all the rage on Tik Tok.So, you would definitely get the most engagement and shares and views there" (P19).
When participants were asked why they think this post is most appropriate to TikTok, they generally discussed Social Norms.For example, P19 noted that Social Norms on TikTok encourage posting fast, kinetic content, with slower-paced content seen as violating Norms.P19 explicated other media Social Norms but noted that since the scenario so clearly matched TikTok, TikTok would be the best fit.Although it's possible to post videos to other media such as Instagram, Snapchat, and Twitter, they are definitely more common in TikTok.This automatic matching based on Social Norms suggests a modification to the original FANG model, suggesting higher precedence for Social Norms.When scenarios matched existing media Social Norms, participants made that choice directly without stopping to consider Affordances and Features.However, when no such direct Social Norm match existed, their decision processes were more complex and incorporated those factors.
For many of the Authentic scenarios, participants did not see an exact match between the scenario and existing Social Norms, which led them to reflect on the scenario in more detail.Scenario 6 features a protagonist whose ex-partner has started seeing other people, and the protagonist wants to post about their feelings (see Appendix A).P13 gravitated towards Twitter in their response based on Social Norms around what people post there: "I feel like Twitter because I see a lot of people posting their feelings on Twitter.Versus Instagram because [...] it's more of a picture place.If I wanted to see written posts about something I'd go to Twitter.Snapchat is the same thing, I kind of go to Snapchat for pictures and happy experiences.
For Tik Tok, I mean, it could be appropriate, I guess.Um, I just feel like it's just a bunch of different people will see it… maybe you just want your close friends to see it" (P13).
This participant wrestled with where this post could fit and used Social Norms to help resolve their conundrum.They first noted that Twitter could be appropriate based on prior Norms about people posting their negative feelings there.Next, they evaluated Snapchat and Instagram.Either could potentially be appropriate, except that Norms for both tend to emphasize positive picture posts.TikTok could work as a medium to discuss negative emotions, but for this participant, TikTok tends to target a large group of strangers, whereas such messy, negative feelings would be better for close friends.P13's reaction first invokes respective media Norms, thinking about what people typically post on each.They finally opt for Twitter; even though it's not a perfect match, it doesn't egregiously violate P13's understanding of the Norms there.
P13's quote further supports the need to modify the original FANG model.When the scenario fails to elicit the Social Norms for a specific medium directly, the participant reverts to considering the Goal (here, posting feelings).They then evaluate Social Norms and Affordances to find the medium that best matches that Goal.
In this example and many others, we found that Affordances and Features were accorded less importance in the choice process than the original FANG model proposes.Participants evaluated Affordances only insofar as they supported Social Norms but did not view Affordances as primary determinants of media choice.Features had an even more minor role in their decisions.We now describe this.

Affordances Are Less Important Than Goals or Social Norms
The FANG model (see Fig 1) proposes a key role for Affordances.Personalization (see middle left of Fig 2) is an example of an Affordance, reflecting a participant's perception that the medium can be configured to restrict messages to specific people.For example, for some participants, Snapchat offers more Personalization than Twitter because they can restrict a particular message to a particular audience on Snapchat.However, participants making choices did not generally imbue Affordances with equal importance as Goals or Social Norms.Instead, participants described Affordances as parasitic on the Social Norms they support.For example, P5 describes how Snapchat does not have certain Affordances because of the Social Norms in place there: "Yeah.I think for Snapchat privacy isn't really an issue because only people are your friends can see it.I don't think you can be anonymous on Snapchat as well.That's not something that you can be.I think with Snapchats usually you are the one that's dictating the conversation.If you're the one that's posting your story, people usually respond back to your story, usually not the other way around" (P5).
P5 describes how for them, Snapchat doesn't possess certain Affordances related to Privacy and Anonymity due to Social Norms around its usage.Suppose a Gen-Z young adult is posting to their story.In that case, because "only people who are your friends can see it", it eliminates Privacy or Anonymity concerns because these intimate people can be trusted.This participant's view of Snapchat is typical of how participants saw the role of Affordances in media choice.Participants mentioned Affordances, but only in service to Social Norms or the Goal of the post.These observations suggest the current FANG model needs revision, with the role of Affordances being downgraded.We now discuss Features, which were again only considered in terms of how they affected Social Norms.

Features are Necessary but Not Sufficient to Determine Media Choice.
When participants referred to Features (i.e., the objective technical properties of a medium), they typically focused on the type of content found in different media.Many media allow posting in multiple formats, e.g., while users typically see Twitter as being text-based, it also supports video and pictures.Nevertheless, our participants noted they tend to post the type of content that Social Norms dictate typically appears on that medium.The example below shows that Features are necessarily considered when people use a medium.However, they are not sufficient to explain media choice, instead being subsumed by Social Norms.For the following participant, the Type of Content element is related to whether the post is a photo, video, or text.This element appears to be a Feature as it is related to what people can technically do on the medium.However, the interview clarifies that rather than their choice being based on what is objectively possible on a given medium (i.e., Features), that choice is driven by how others typically use that medium (Norms).In other words, their choice is actually driven by Social Norms: In a further deviation from the original FANG model, participants often revisited the Goal of a scenario when there was no clearly matching Social Norm.As in the example from P13 discussed in Social Norms, we saw other participants enacting this process.For example, in the Upset scenario, the protagonist of the scenario is disturbed and wants to honestly share their feelings so that a friend will comfort them.In this case, P12 is trying to choose between TikTok and Snapchat, thinking about how to contact a specific person without having others interfere.However, because there is no direct match based on Social Norms, they revisited the original Goals and broadened their choice processes to include other factors: "I'm not sure if you can make an individual TikTok just for your friends to see.But if it did that feature where you can you can just set a video to just a certain friend group then maybe TikTok?Snapchat...I know you can send it specific friends, but she didn't mention a specific friend.So … because she just wants any friend, [...] so I wouldn't necessarily choose Snapchat maybe TikTok "(P12).
P12 concluded that TikTok or Snapchat were viable choices after evaluating the Social Norms around sharing emotional content and other people seeing potentially sensitive information.However, P12 could not decide between these choices, as neither was an exact match.P12 then returned to the initial scenario to confirm with the interviewer that it involved "wanting a friend".P12 then revisited their choice, now considering the Affordances of TikTok and Snapchat.In other words, only after the participant determined that Social Norms were not directly helpful to did the participant revisit the protagonist's specific Goal.This then informed the participant's evaluation of which medium was most appropriate.
This data again suggests a modification of the original FANG model, indicating that when there wasn't a Social Norm match, participants needed to revisit the Goal after narrowing their choices based on potential media Affordances and Social Norms.Participants seem to use all the factors to some extent, but Social Norms and Goals were more important than Affordances and Features.We now discuss the scenarios in more detail, examining consistency and inconsistency in media choice.Participants' responses to each scenario are shown above in Table 2.Note that for some scenarios, such as Dancing, Identity, and College, participants were highly consistent, with the majority choosing the same medium.However, there was less consistency for other scenarios such as Expartner, Fitness, or Upset.Some participants felt that a particular scenario could be posted to multiple social media.We present the most common social media combinations and the number of participants who gave that answer for each scenario.Few participants chose all for or any combinations of three media for the scenarios; there were three instances of selecting all four media and two instances of selecting three media from all participants.The most popular choice for multiple social media across scenarios was Instagram + TikTok (18), with 15 of these coming from Idealized scenarios.These results indicate that participants did not see all media as relatively equivalent, with most scenarios featuring a majority of one social medium selected.However, when participants chose multiple social media, they saw Instagram and TikTok as being somewhat equivalent, especially for Idealized self-presentations.

Scenario Results
As with flowcharts described above, Social Norms seemed to be the driver of consistency or inconsistency in selection.Participants described explicitly thinking about these other factors when a scenario did not match an existing Social Norm for a medium, thus reducing consistency.Consistency of selection was also influenced by scenario type, with Idealized scenarios being generally more consistent.We now discuss the different reactions to these types of scenarios.4.7.1 Idealized Scenarios.Scenarios that fit existing media Social Norms showed far greater consistency.For example, scenario 1 (see Appendix A for the full text) featured a protagonist who enjoys dancing and wants to show off their skills.Almost all the participants chose TikTok.As P19 noted earlier, due to Social Norms about popular content, TikTok is the best place to post such content.Such consistency driven by Social Norms provides further evidence that Norms are more important for choice than Affordances and Features.For another example, consider scenario 2. The protagonist was a college student taking videos and pictures of a friend with a nice backdrop, which fits a typical Social Norm for Instagram.P6 describes their reaction to the College scenario as perfectly matching Social Norms for Instagram over other social media: PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023."It's the type of content that works well on Instagram, people like to see a pretty person, nice background, specifically post[ing] photos, like it's perfect for the platform.[...] And it [Instagram] has now gotten to the point where your newsfeed is just… you just keep scrolling and you want photos to catch someone's eye if you want them to stop and like it, and look at your profile" (P6).
The scenario description directly matched P6's understanding of Instagram's Social Norms, where a post must have an attractive background and subject to get someone's attention.Other participants seemed to agree, with 15 participants selecting Instagram or Instagram and another social media.These examples further substantiate Social Norms' importance compared with Affordances and Features.Participants first evaluated the scenario to determine whether it reflected an existing Social Norm for a particular medium.When such direct matches occurred, selections were consistent across participants.4.7.2Authentic Scenarios.However, scenarios that did not directly match social media Social Norms made choices harder.Authentic scenarios often did not appear to fit existing Social Norms.Some participants even mentioned feeling uncomfortable with these scenarios, e.g. when they violated Social Norms about only posting positive emotional content.
In one Authentic scenario, Ex-Partner, participants utilized both Social Norms and Goals to understand what social medium would least violate Social Norms.Recall that in this scenario, the protagonist has an ex-partner who has started seeing other people and wants to make a post related to their feelings (see Appendix A).When P7 heard this scenario, they chose Snapchat: "I would think that be private...So I would say Snapchat.[...] Yeah, you can post it on TikTok, but usually the other party doesn't really appreciate it.[...] Also it's audience tends to have a bit of an issue when things get a little too real.[...] And, to a point you can do it to relate to people and be a little more down to earth.But when it comes to [...] letting yourself out there… unless you've built the audience support.It's good to keep it close to home, because that's where you're going to get the most genuine support "(P7).
This participant chose Snapchat as they felt it was important to keep the sensitive topic of relationship drama to a smaller supportive group rather than show something "too real" to a larger group who might not appreciate it.Incorrectly choosing a group could violate a Social Norm around what topics are appropriate to post for other media such as Instagram.
As we saw before, P7's discussion of Snapchat first identifies the Goal, then evaluates Social Norms and Affordances of different media and finally revisits the scenario Goals.The example again shows that if there is no clear Social Norm match, the selection process becomes more challenging and involves revisiting Goals.In P7's view, Snapchat is a place where people can have the Affordances of closeness and Privacy with reduced risk of violating Social Norms around posting negative content.Previous research identifies Snapchat as a place where users can be Authentically vulnerable [4,73] due to the ephemerality of posts and the ability to control who sees what.P7 then revisits the Goal of the scenario as a motivation to seek support.Since the Goal is to seek support, this is best supported by the Social Norms and Affordances of Snapchat.
As discussed above, Authentic scenarios often invoked negative emotions and vulnerability, violating existing Social Norms for some media.However, there were some discrepancies in how participants viewed the scenarios.For example, scenario 6 featured a protagonist who wants to post about how they struggled in college but was eventually able to get their degree (see Appendix A).We had initially created this as an Authentic scenario since the protagonist is openly sharing a painful time, but multiple participants noted that it could also induce Idealized self-presentation.The specific narrative of a "success story" was seen as matching Norms on Instagram, with others indicating that the scenario was Idealized enough to post to multiple 'positive' media (i.e., Instagram and Snapchat).This adds further support for the importance of Social Norm matching.If a post matches existing Social Norms, especially if they are Idealized, then participants were more consistent in rating it.
We found participant responses were generally more consistent with Idealized scenarios than Authentic due to varying views of Authenticity related to Social Norms.For Authentic scenarios, the inconsistency with Social Norms required participants to consider more aspects, such as where violations would be less likely to be sanctioned or where the benefits of posting might outweigh the drawbacks.

DISCUSSION
Overall, the data reveals that media choice processes are quite complex.Specifically, we need to revise the original FANG model to emphasize the importance of Social Norms while deemphasizing Features and Affordances.Given the participant sample, graduate/undergraduate students, likely non-influencers, middle-upper class, young adults (average age 24), this model may have limited applicability outside of this population.The users in our sample were very aware of Social Norms, and at times their concern for Social Norms was prioritized over the explicit scenario Goal, especially when participants felt a strong match between a scenario and a specific Social Norm.For example, the Dancing scenario was most frequently matched to TikTok even though Instagram would be an acceptable alternative.The decision process was more complex in cases where the Social Norms did not exactly match the scenario.In these cases, our participants revisited the scenario's Goal, then determined which media best fit based on their perceived Social Norms and Affordances.Our participants also accorded Features less consideration than the FANG model proposes.Instead, participants seemed to consider Features to the extent that they primarily supported Social Norms and secondarily Affordances.The interviews indicate the pre-eminence of Social Norms compared with Affordances or Features; violating a Social Norm could lead to sanctioning, negative feedback, or a loss of followers.Our participants only revisited the Goal of the scenario after they had failed to match Social Norms directly.At that point, they would then consider both Norms and Affordances, and occasionally Features.We now discuss our revised model as well as implications for social media research and systems design.

Revised FANG Model
Our interviews and scenarios make it apparent that the original FANG conceptual model failed to capture the complexity of participants' media choice processes.To address RQ 1, we revised the FANG model in important ways (see Fig. 3).First, the revised model reflects that Social Norms often trump Goals in driving decisions.Second, Affordances mostly influenced choice by supporting Social Norms with Features hardly being discussed separately from Social Norms.Overall, Norms are more important to posting decisions than Affordances and Features.Features were even less influential than Affordances, and only mentioned in the context of other factors.Another major modification concerns when Goals are invoked.If participants failed to find a direct match for the scenario and a social medium based on Norms, they then revisited Goals to reevaluate Norms and Affordances to determine how they supported that Goal.We revised the FANG model to rework the role of Social Norms significantly.The revised model now involves an initial choice based on Social Norms.If the scenario triggers Social Norms for a particular medium, then that directly drives media choice.For example, in the TikTok dance scenario, Abby wants to make a post featuring herself showing off a dance.According to the revised model, Abby does not consider the post's Features, Affordances, or Goals.Performing a kinetic dance directly fits the Social Norms of TikTok, so she should just post there.Although historically, Norms may have arisen because of Affordances and Goals, these are not considered in the revised decision process.Using Social Norms here is like a heuristic that allows one to navigate posting decisions without needing much conscious thought.However, users in our sample can evaluate the reasoning behind such decisions.When a potential post does not match existing Social Norms, a user in our sample would need to evaluate how to make the post more consciously.
Let's illustrate another aspect of the revised model by examining a contrasting example using the Ex-Partner scenario.Bob is feeling down, seeing that his ex is dating again.His Goal is to reach out, discuss his feelings, and receive social support or advice.This Goal does not match any Social Norms on any social medium he uses, so he must think more consciously about where to post.He, therefore, revisits his Goal, clarifying that he wants to ensure that his relatives and ex do not see the post while targeting a large enough group to respond with advice or support.Considering this Goal causes him to reflect on more detailed media Affordances to determine which social media options have better Personalization and Connection Affordances.He also needs to choose a medium that does not have Norms that preclude Authentic posts, as he plans to be very honest about his feelings.Revisiting his Goal of eliciting advice and support confirms his potential choices, TikTok and Snapchat.He decides on TikTok, which allows him to reach a broad audience, which people he knows are unlikely to see (his ex has no idea what TikTok is).He's also seen people post Authentically on TikTok while receiving support, suggesting this would not violate a Social Norm.In this example, Bob makes his choice after carefully reflecting on his options.Since this is also a potentially vulnerable post, he must deliberate more consciously about which medium would allow him to avoid potential embarrassment or Social Norm violations.The two pathways of the FANG model are similar to Kahneman's concept of Type 1/Type 2 thinking [41].Type 1 thinking relies on heuristics and is much quicker (being almost automatic) than Type 2 thinking.Type 2 thinking is slower and more conscious but often solves problems that Type 1 thinking cannot.Consistent with this, our participants made some selections extremely quickly and confidently (e.g., the Dancing scenario).Although we do not argue that established Social Norms lead posting decisions to be entirely automatic, we note there was a high level of consistency for such scenarios.This pattern suggests opportunities for future research, potentially testing our tentative model in more controlled settings to enhance our understanding of media choice processes.

Implications for Social Media System Design and Theory
The results suggest implications for design of social media systems, in particular the introduction of new Features and new social media technologies.Given the importance of existing Norms, it may be hard for new Features to impact media usage.Rather than exploring new Features, users in our sample indicated they may be content to continue with those supporting existing Norms.New Features that replicate existing Features on other media must compete with existing Social Norms.For example, in late 2020/early 2021, Twitter began to roll out a new ephemeral Feature ("fleets"): tweets that would expire after 24 hours.Twitter intended fleets to be a way to allow people to tweet more frequently without making a full post, like Snapchat or Instagram stories.However, this Feature failed to gain traction, as Twitter removed it in August 2021, replacing it with Features that would encourage people to tweet more [32].The revised FANG model could explain this in the following way.In this case, if users are similar to users in our sample and have the Goal of making an ephemeral post, they already have at least two other ephemeral options with existing Social Norms (Instagram stories and Snapchat).In this case, assuming our model applies, Twitter must overcome users' pre-existing perceptions of Social Norms within people's social media ecologies if they are to exploit this new Feature.It may be difficult to break down existing Social Norms.People already understand how Twitter fits into their social media ecology, so adding a new Feature replicating other social media may not be enough to change their understanding of the Social Norms and subsequent posting behavior.
Conversely, the influence of Social Norms may make it difficult to introduce entirely new social media into a user's social media ecology, assuming they are similar to the population studied in this work.There are no pre-existing Norms for new media, as people usually do not start with a clear understanding of how to use those media.This problem was finessed in the case of TikTok, which is an excellent example of launching with a clear set of Social Norms.TikTok had already launched in the Chinese social media market as Douyin in 2016, where it had millions of users exploring its Affordances and building Social Norms around its use before it launched in Western markets.This allowed TikTok to launch in Western markets with clear expectations of Social Norms.The Western launch included examples of Social Norms provided through advertising and influencers, allowing new users to straightforwardly integrate TikTok into their social media ecologies.New users could instantly see people dancing, using the "duets" Feature, sharing audio, etc., and could build Social Norms quickly, especially since these Social Norms did not compete with existing social media within their ecologies.
Such successful adoption contrasts with launches of different products, where potential users are presented with a new set of Features and must organically derive Affordances and Social Norms.For example, when Facebook first launched to general audiences in 2006, it only had one direct competitor (MySpace).Although college students had already been using it, Facebook had few developed Social Norms, so users and companies explored different behaviors exploiting PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023.
Facebook's Features and Affordances [10].Since parents and authority figures often did not use it, high school and college students would post themselves doing illicit things or baring their souls semi-publicly.Now, ossified Social Norms mean that 14 years later, young users would never think to post anything similar on Facebook.Not only have Social Norms shifted around what is appropriate to post, but Facebook's role in users', particularly young adults, social media ecologies has solidified, allowing other social media to address different Goals and Social Norms.
We also present implications for social media ecology theory.This study provides evidence for the importance of Social Norms and Goals in making social media posting choices.Previous research examining social media ecology points to Goals as part of the reason people use different media but doesn't highlight the potential importance of Social Norms [7,69].It's not just Goals that affect how users in our population use social media but also how they interpret its Social Norms.Social Norms are a powerful tool for communities to shape member behavior, as noted in other research [3,15,71].Social Norms appeared to be challenging for people in our population to overcome, even when inconsistent with their Goals.This reluctance may be due to fear of sanctions Furthermore, our findings motivate new research evaluating user social media ecologies and how these affect posting choices.Many prior studies have focused on individual media [3,4,14,39,42,45,53,62,66,73], which oversimplifies key aspects of social media use, including how users contrast and navigate between multiple media.Multiple social media enhance opportunities for achieving Goals while offering different Social Norms around what is appropriate and where.Future research could examine how a broader group of users choose among media, extending our largely qualitative findings using more controlled methods.

Limitations
Our participants were all Millennial and Gen-Z college students in the US.While we selected this sample to fit within the expected age range of people who use at least three of the four social media of interest, there may be differences in how people make posting decisions between those who are at college and those who are not.Selecting a less WEIRD [33] sample could lead to different results.For example, young adults in college might have different self-presentation challenges compared to those who are not, such as trying to keep friends and family updated about their life while attending college far from where they grew up.Another limitation is that our sample size limits the generalizability of our model.We note a few ways in which this model could be expanded with different groups of participants.We end with another limitation, that the perception of Authentic and Idealized presentations may not be universal.
Our sample size of participants leads to external validity concerns about our conceptual model.Recruiting 19 participants from a specific sub-population (college students who were members of Millennial and Gen-Z generations) of all users of multiple social media means that it is likely that this model does not generalize to all users.For example, other sub-populations in different stages of life (adults post-collage, teenagers, seniors, etc.) may have different considerations about how they choose where to make a social media post based on the ecology of media they use.Additionally, it could be that college student users of this particular group of social media evaluate social media choice differently than equivalent users of a different group of social media.Future work could evaluate this model with additional participants from this subpopulation of current Millennial and Gen-Z college students who use the same set of social media, compare and contrast media choice between groups of participants who use different sets of social media, and/or compare to other groups of multiple social media users at different life stages to see how they consider where to make a social media post.

Fig. 2 .
Fig. 2. The top-left section of P14's flowchart.The entire flowchart represents a participant's conceptual model of factors determining where to make a social media post, with the most important elements shown at the top.At the top is Who I want to reach, indicating that is the most important element for this participant.Flowcharts often had high degrees of interconnection, containing many elements.Color has no meaning, they were chosen at random."I think primarily you have to figure out who you want to reach.And if you're posting for social media you can have multiple reasons, but if you're just posting for people you know, you're gonna have different motivations than if you're trying to reach a wider audience.So I would think for people you know, you want to be doing things like having fun, showing emotion, and even getting attention from that.The wider audience I would think you're going to be building brand, monetizing" (P14).
P14's example, "influencing different audiences" breaks down into different Goals for different audiences.By choosing which sub-elements a Ignore the Affordances, It's the Social Norms 311:11 PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023.
While this participant discussed the ability to show emotions in the context of being a consistent poster, others shared concern about suddenly changing one's posting by describing their experiences.Even though a poster might in reality feel emotionally negative, their posts on a medium with positive Norms such as Instagram should not reflect this unless that poster no longer cares about having followers.This and other examples confirm the importance of Social Norms in the FANG model.Participants were aware of a variety of Social Norms on different PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023.

Fig. 3 .
Fig. 3. Revised FANG Model: Millennial and Gen-Z young adults start by evaluating if their intended post matches an existing Social Norm.If there is an exact match, they post it on that social medium.If not, they evaluate their Goal, check Social Norms and Affordances (influenced by Features), revisit their Goal based on their thoughts, then choose a medium.

Table 1 .
Reported Frequency of Social Media Posting and Viewing PACM on Human-Computer Interaction, Vol. 7, No. CSCW2, Article 311, Publication date: October 2023.
[...] I think the type of content would be the primary thing people take into consideration.Just because of what's most normally seen on those things, I think people have a tendency to fall into kind of what other people are doing.Don't want to stick out necessarily" (P11).The critical part of this statement is that the participant feels usages tend to "fall into what other people are doing".Even though the Type of Content sounds like a Feature from the flowchart element's text, the interview clarifies that Social Norms actually determine choice.Instead of representing what someone can technically post on a medium, this element represents general rules about what is appropriate to post to avoid "stick[ing] out necessarily".In other words, Features are a necessary but insufficient step in decision processes about where to post.These examples suggest that as with Affordances in the FANG model, Features should be downgraded as they seem to be dependent on Social Norms.4.6 Goals Revisited When No Matching Social Norm.

Table 2 .
Medium Choices in Scenarios Note: The 'Multiple' column header refers to participants who said the scenario could elicit posts for multiple listed media, with the most common combination listed in parentheses