The Instagram Interview: Talking to People About Travel Experiences Across Online and Offline Spaces

Studying visitors’ experiences with cultural sites has been complicated by the availability of internet‐connected mobile devices. Simply observing visitors on site is no longer sufficient since they can interact with a site offline and online: before, during, and after their visit. Furthermore, cultural sites are as much sites of cultural heritage as they are sites of tourism. To study such complex experiences, new approaches to the study of human interactions with cultural sites must be devel‐ oped; these methods must account for the fact that the offline and online realms can no longer be considered separate. In this article, I introduce the method of the Instagram interview as applied in an Instagram ethnography, contextualized by my project on visitor experiences of a Dr. Martin Luther King, Jr. Memorial in Washington, DC, where I interviewed visitors after their visit. The Instagram interview helps study a dispersed population that engages, through Instagram posts, with one physical location and its narratives, allowing conclusions about visitor experiences of the site and the role of Instagram in this context. When constructing the Instagram interview in a manner that corresponds to platform conventions, it pro‐ duces personal, in‐depth narratives about the interviewee’s experiences. Conceptualizing the experience of a memorial as expanding beyond the space and time of the site visit, the Instagram interview is suitable for holistically studying visitors’ complex experiences: before, during, and after their visits, as it recognizes that offline and online interactions with the site are part of the same experience.


Introduction
Solemn faces carved in marble, stoically looking down from their pedestals, visitors zipping past on scooters, looking at their mobile phone screens, snapping a selfie every now and then: Memorials in Washington, DC have become more than the physical representation of public memory, a "socio-political construct" through which groups define and distinguish themselves from others by communicating specific versions of the past (Neiger et al., 2011, p. 4). They have become tourist destinations, and concomitantly, individuals' visits to these sites have become more complex. Visitors are no longer (purely) taking a pilgrimage to the nation's "monumental core" (Savage, 2009) to witness the embodiment of the nation's identity; they are also visiting must-see sites, taking tourist photographs, and judging their visits' success as much by the memorials' expressiveness, as by their personal experiences (cf. Sturken, 2007). This development is accelerated by the prevalence of mobile phones and the importance of social media to tourists. While it is obvious that individuals at these sites are using their mobile phones-aside from seeing them take picturesit remains unclear how they use their internet-connected devices to augment their visits. I, therefore, wanted to know how visitors experience such sites, including how they use their devices to engage with the memorials and the narratives they tell.
Museum scholars have long acknowledged that the offline and online realms intersect at cultural sites (e.g., Weilenmann et al., 2013). This intersection of spaces is particularly palpable at memorial sites when visitors, having traveled far to see them, can be observed looking at their mobile phones while in the presence of a unique artifact. As Couldry and McCarthy (2004) describe with their concept of mediaspace, "digital media and everyday life form part of the same spatial realities" (Pink, 2021, p. 55). Spaces overlap here in that visitors can inhabit the online realm while standing at the physical site. However, the spaces also expand: Visits to DC memorials are impacted by individuals' (online) media use and destinations for tourism are in part chosen by what individuals have seen before (e.g., Muslimah & Keumala, 2018). Furthermore, visits do not necessarily end upon leaving the location, as engagement can continue, particularly through social media. The space of the memorial experience is therefore not restricted to the physical site. In this article, I propose the Instagram interview as an adaptive method that can help researchers answer questions regarding user's complex experiences of cultural sites, both on-site and online, during and after a visit.

Literature Review
To study visitor engagement at cultural sites, different approaches have been taken. One strand of research focuses on how mobile devices (audio guides or personal mobile phones) impact offline behavior and experience (e.g., Bowman, 2010;Hillman et al., 2016). Such studies can be located within a traditional ethnographic approach centered around participant observation (cf. Clifford & Marcus, 1986). Other approaches focus on the resulting online artifacts created by visitors to cultural sites (e.g., Budge, 2017;Weilenmann et al., 2013). However, when acknowledging that online and offline experiences are not separate and impact one another, we must study them together, which few studies thus far have done. Exceptions include Hughes and Moscardo's (2017) study, which observes how mobile phone use impacts behavior through a museum exhibition and what types of photographs individuals take during their quest. Their study was conducted under experiment conditions, meaning that mobile phone use was not natural. Studies analyzing naturally occurring offline and online behavior ethnographically differ in their approach, "inflected by the theoretical and practice stances of particular disciplines and fields of study" and developed as part of specific projects and questions (Pink, 2016, p. 162). Such research combines observations and different types of participant interviews. Bareither (2020), for example, in his work on the Memorial to the Murdered Jews of Europe, conducts a digital ethnography by combining on-site and chat interviews with content analyses of Instagram and Facebook posts. This method closely resembles the one proposed in the present article. Bareither (2019) does not elaborate on his interview methods, but he posits elsewhere that ethnographic concepts in the digital context must be further developed. In order to fill this gap, I propose an elaboration on one digital interview method: the Instagram interview.
Researchers in different fields have conducted a variety of digitally enhanced interviews. Amongst them, email interviewing is the most popular asynchronous interview method (Bampton & Cowton, 2014;Dahlin, 2021;Fritz & Vandermause, 2018;Meho, 2014), while skype and other video chats are used for synchronous interviews (Lo Iacono et al., 2016;Serafinelli, 2017). To allow for almost instantaneous interactions embedded in participants' everyday routines, the mobile experience sampling method automatically prompts participants to self-report experiences (van Berkel et al., 2017). While this method mostly corresponds to a quantitative research perspective, it has been developed further, such as in Kaufmann et al.'s (2021) study about individuals' situatedness during the first Covid-19 lockdowns. Their mobile instant messaging interviews present a qualitative approach to interviewing individuals in their everyday life contexts, showing that digital methods can be suitable to study offline behavior, thereby overcoming the online/offline dichotomy. In this article, I present an addition to these methods in the form of the Instagram interview; a mostly asynchronous method, it can be employed to assess online and offline behavior about users' experiences, as well as the intersections between them.

Studying Visitors of Memorial Sites: The Project
To illustrate this method, I draw on an exploratory project in which I combine offline and online methods to holistically assess visitor experiences of memorial sites, including their (online and offline) engagement with them. I want to briefly introduce the project to contextualize the method but will only elaborate on methodological findings thereafter. In the overarching project, I focus on two unique sites: The Martin Luther King, Jr. (MLK) Memorial and the inscription for his 1963 "I Have a Dream" speech on the steps of the Lincoln Memorial, both in Washington, DC. In this article, I focus only on the Instagram ethnography studying the former. Dr. King is the only member of a marginalized community honored along the National Mall, the city's prime location for memorials. King is often tokenized as the only representation of the civil rights movement (e.g., Theoharis, 2018), and his memorial contains little information about his life, the movement, or any acknowledgment of the hardships he faced. In this context, the online realm is often considered a potential pathway for individuals to add to or challenge the official narrative told by the institutions (e.g., Florini, 2016). This means that surrounding this site, countless opportunities for engagement exist, which go beyond the possibility of looking up information about the site on its official website and sharing tourist photographs.
I understand these sites as sites of public memory and tourism, and for both types of engagement, visitors who carry their internet-connected mobile devices can be present at the site while at the same time engaging in the online realm. The fieldwork from this overarching project, however, has shown that visitors do not, in fact, use their mobile devices to look up information while on-site, so they are unable to bring information from the online realm into the physical space, nor do they frequently convey information out of the space, by sharing on social media, for example (Hugentobler, 2022). Instead, mobile phone use at the site was largely limited to taking photographs (Hugentobler, 2022). While the site itself can be considered a "hybrid space," one which is "created by the constant movement of users who carry portable devices continuously connected to the internet and to other users" (de Souza e Silva, 2006, p. 262), much of the online interaction with the memorial and its surrounding narratives does not happen in that physical location, but most often after the visit. This requires an additional method to access individuals' engagement with the site after the visit, assuming that the visitor's experience of the site expands beyond the duration of their presence in the physical location.
This project started with traditional fieldwork. Gathering data for in-depth description (cf. Geertz, 1973, pp. 9-10) in two sets of on-site fieldwork (in 2019 and 2021), I conducted observations, auto-ethnographies, and expert interviews with members of the National Capital Planning Commission and the National Park Service; I also went on commented walks with a member of the National Park Service and a participant from the Instagram ethnography. Furthermore, I conducted short, qualitative interviews with 81 visitors and two tour guides to access visitors' personal perspectives on their own experiences. During this fieldwork, participating in and observing the natural setting of the memorial sites (cf. P. Atkinson, 2017, p. 10), I focused on the atmosphere and patterns, observations that cannot be made through online methods. I also studied visitors' offline and online behavior, including how they used their mobile phones or if previous experiences in the online world impacted their offline behavior and impressions. During the first fieldwork, I found that online engagement with the site hardly happens in the physical location and that time spent on site with visitors was limited, findings which meant I had to adapt my methods. This is a common issue with explorative research, which researchers can anticipate by being flexible in their research design (cf. Dahlin, 2021). Because I could not access individuals' online engagement with the site, I created an Instagram ethnography to complete my research (following the principles suggested by Pink et al., 2016 (Goffman, 1959, p. 6). This includes social media behavior, meaning that individuals' performances on social media are not random but rather influenced by what individuals think they are supposed to post (e.g., Ditchfield, 2020). This reflects what Norman (1999, p. 39) calls "cultural constraints"learned conventions about encouraged actions. The content analysis, therefore, goes beyond showing what individuals liked on site or what information they consider worth sharing or adding: It also indicates which conventions impact their Instagram performance (Hogan, 2010). We know, however, that performances by individuals are never the full story (Clifford, 1986) and can indeed differ in their appearance from individuals' motivations, particularly on a social media platform such as Instagram, which is often used to present an idealized version of oneself (e.g., Baym, 2011;Hu et al., 2014;Humphreys, 2018). While the content analysis, therefore, allows conclusions about visible traces of behavior and cultural Instagram norms, the interviews-elaborated on in this article-access conscious motivations and fuller experiences of the site: before, during, and after the visit.

Choosing a Platform for the Digital Ethnography
To study visitor interactions with the memorial, I focus on Instagram for three reasons connected to the platform affordances, its technological possibilities, and conventions (see Norman, 1999). First, the experience I study is characterized by an inherent duality: Visiting a DC memorial is an act of tourism as well as an interaction with public memory and, thereby, US American identity. Instagram functions according to a similar logic: as the prime visual social networking site, it is ideally suited to sharing tourism images, one of the main uses of social networking sites (e.g., Christou et al., 2020). Additionally, posting on Instagram is used to construct or negotiate personal identity (e.g., Lee et al., 2015), often done by showing an ideal version of oneself, which can include the self as a traveler (Lo & McKercher, 2015) who shares recognizable images (Hugentobler, 2022) or has ideological allegiances. Second, Instagram is an inherently visual medium, as are memorials. Visitors frequently stress that much of their enjoyment stems from the site's beauty. I, therefore, study interactions on a site with a visual focus. Third, Instagram's affordances allow me to study behavior that corresponds to my research question: I study individuals' posts which are aggregated by being tagged at a specific location and thus part of a larger narrative, but at the same time, they are also showcased on users' personal profiles. These Instagram posts thereby serve two functions: They add a personal to a larger narrative, and they create and manage a personal identity through a curated profile. These posts can then be read as contributions or challenges to public memory and as personal narratives about the self. It is therefore important to choose a platform that corresponds to the research questions. Instagram is particularly suited for interactions with the world that bridge the private/public realm and engage with the world visually.

The Instagram Profile
To access Instagram posts and interact with users, one needs an Instagram account. Previous research has made a case both for using the researcher's personal profile and for creating a profile specifically for the project at hand (e.g., Urbanik & Roks, 2020). Since my project is not necessarily concerned with sensitive data and relies on comparatively short interactions, which require lower levels of trust-building, I decided to create a professional profile and mark it as such. However, when researchers rely on establishing deep, personal relationships with their subjects, it might be worth considering using one's personal Instagram. I set the privacy of my profile to "public" and used my full name as my username. For my profile picture, I used the same photograph as on my website, to which I also linked (see Figure 1; cf. Hine, 2000, p. 74). Both additions help signal the account's legitimacy. Widespread agreement on digital media research ethics posits that researchers must disclose their "professional persona when collecting data in online communities" (de Seta, 2020, p. 90). Despite not studying an online community per se but rather individual Instagram users, I chose to be as transparent as possible about my intentions without influencing the participants' responses. By adding a link to the University of Zurich's Instagram profile, I hoped to legitimize the profile further.
In the next step, I populated my profile because an empty profile appears suspicious: It might seem like a lurker or a social bot (cf. Urbanik & Roks, 2020). So, I added images to reciprocate what I would find on participants' profiles. If I learn something about them when looking at their profiles, so should they when looking at mine. The researcher role I inhabit here is, therefore, closest to sociologist Raymond Gold's (1958, p. 221) "observer-as-participant": I participate by sharing some travel pictures on Instagram but remain more of an observer than an active participant. The selection of images was not random: I chose personal photographs of DC memorials (Figure 2), thereby signaling my interest without framing the sites in any particular manner, keeping captions descriptive.
After contacting several users, I realized that my profile still looked suspicious because I had not shared any pictures of myself, which is unusual for an Instagram profile. As Hu et al. (2014, p. 597) found, "nearly half (46.6%) of the photos in [their sample] belong to Selfies and Friends categories with slightly more self-portraits." Therefore, an absence of photos of people might have contributed to several participants responding to my interview requests by asking if I was really a person or telling me they first thought my message was spam (see Figure 3). I, therefore, started including images showing me during fieldwork (Figure 4). I also playfully captioned some of those images, inviting the audience into my experience of doing fieldwork in DC, thereby allowing them a glimpse into my life, where the researcher and private person overlap. This strategy likely impacted the response rate since I no longer received messages doubting my identity. I, therefore, recommend constructing a profile that reflects the researcher's current interests, showing themselves as fully human because the site's conventions require it.

The Sample
To answer my research question, I had to find posts that were (a) about the MLK memorial and (b) engaged with the official narrative. The first condition can be met in two ways: searching for the memorial under either its hashtag or its location tag. I chose the "Martin Luther King, Jr. National Memorial" location tag because the #mlkmemorial included different memorials to Dr. King; it did not just present a collection of stories about the same site. The second requirement, posts engaging with the official narrative, can be argued in different manners. Every social media post mentioning the memorial engages with the official narrative. However, only when narratives go beyond the private realm, such as being tagged at a location, do they become searchable (and    therefore viewable by individuals searching for the site) as a collection of narratives about the same topic. They are public and can potentially create bottom-up narratives that differ from the dominant one, thereby challenging it. This sampling method excludes private profiles as only posts on public profiles appear at a location tag. While I, therefore, exclude posts by users who might exhibit different posting behaviors-in line with their privacy preferences-for my research question, focusing on public profiles, in fact, makes sense: I am interested in public, individual engagement with the official narrative. This means that, for my research question, private Instagram posts are not relevant.
Sampling began at the location tag, where I saved the most recently tagged post and followed the person who shared the post. This is helpful for two reasons. First, following an account creates a notification on that account, helping the person notice me and my message. Second, by following their account, individuals are more likely to follow me back, which can help counteract the problems that "research-only" profiles often look fake because they have no followers (Urbanik & Roks, 2020, p. 224). At the time of writing, this research profile had 59 followers. After taking a screenshot and following their account, I contacted the user with the recruitment message. I conducted continuous sampling, contacting individuals, and saving posts until someone responded to my request. I then halted sampling to allow sufficient time to conduct the interview. I stopped sampling once I reached theoretical saturation. The response rate for this first study was 19% (118 users contacted, 22 interviewed), which is rather low, but as Norman (1999) claimed, even just a few participants can be sufficient when studying how people use devices. Referring to Jakob Nielsen, he says: "Three to five people will give you enough for most purposes. But they need to be real people, doing real activities" (Norman, 1999, p. 41). Context is essential here: In line with much ethnographic research, which is not meant to be universally generalized (being deeply grounded in context; Kozinets, 2010, p. 59), the number of participants can remain low, given that saturation is reached, and the context of the study is natural.
My proposed method allows researchers to access individuals who share on only one platform, but if this platform is of specific interest to the researcher, this bias is irrelevant. In fact, talking to people on the platform where the behavior under study occurs can be an advantage (Altmann, 2011, p. 100). The sampling method does, however, present other biases. Self-selecting into a research project means that participants are more interested in the topic than the average person (Bethlehem, 2010). Furthermore, individuals who agree to be interviewed through Instagram are likely more experienced Instagram users (cf. Altmann, 2011;Fischer, 2009). However, being able to answer research questions online can also lead to more introverted people participating than in traditional research settings (Hertel et al., 2008). Researchers must consider these issues when adapting their own research questions to Instagram research. Overall, sample biases are comparable to the ones in offline interview methods. The method is, therefore, not inherently inferior, and its advantages will be shown in the following.

The First Interaction: Crafting the Recruitment Message
The first interaction is crucial since it decides if a person chooses to participate. I contacted individuals through the direct messaging function, as avoiding attachments has already been suggested by researchers conducting email interviews (e.g., Meho, 2014, p. 26). One important aspect to consider is message-length. I, therefore, carefully crafted this message, keeping it brief and clear. Because it was still a long message by Instagram standards, I split the message in two so that participants were not confronted with a big wall of text; instead, they would see the messages separately in the message preview, which is available on most smartphones. As Crystal (2001) found, questions should be readable "within a single screenview, without any need for scrolling." It can further create issues if the messages are "complicated or verbose" (Bampton & Cowton, 2014, p. 7). This is particularly true of the first message as it sets the tone for the following conversation. Therefore, I described the project briefly and clearly without giving too much information, which might create a bias in the following responses ( Figure 5). For this research project, I did not have to undergo ethics reviews. However, researchers who work with ethics boards must ensure they comply with those guidelines while also catering to the expected communication on Instagram. When using long recruitment messages, researchers should consider sending the second half only after they have received a first reaction. This caters more to the back-and-forth that is common on the platform. In this case, however, researchers must ensure that they have a positive affirmation to all parts of the recruitment message.
While the first message was purposely short, the remainder were crafted according to how the conversation developed. Since Fritz andVandermause (2018, p. 1644) found in their email interview study that "longer emails tended to elicit deeper thought, lengthier sharing of ideas, and greater revelations," I did not force short questions when interviewees were eager to respond and engage with the topic more deeply. Here, researcher flexibility is crucial as some participants are likely to quit if messages are too long or complex.
To lower the hurdle for participation, I constructed the interview environment as naturally as possible, including options for responding via voice message and encouraging natural language use. For the latter, I included emojis to show the conversational style of discussion I expected. Fritz andVandermause (2018, p. 1646) noted that their attempts to reflect participants' language led to enhanced communication and that using emojis improved the authenticity of voice and elicitation of trust. Therefore, I opened with an emoji to signal that I approved of their use, but mirrored participants' emoji use in the following conversation. When individuals did not respond after a week, I followed up once to avoid taxing participants too much and to respect their unwillingness to participate in research. An additional method to increase response rates in those willing to participate but who had simply forgotten is to be active on one's own profile: By posting an image every few days, my profile appeared in the feeds of potential interviewees who followed my account. This subtle reminder can help increase participation without having to explicitly remind people.

Crafting the Conversation
The first question was open-ended and geared toward the interviewee's experience at the site. This question was phrased based on the individual's specific post (cf. R. Atkinson, 1998, p. 42). Keeping the question open invites participants to guide the conversation in the direction of subjectively important aspects, allowing observations about how the interviewee understands their own practice (Dahlin, 2021, p. 3). It also enables access to subconscious patterns of thinking and acting (Altmann, 2011, p. 98), which is particularly interesting in routinized behavior such as traveling and Instagram posting. Due to the personal nature of the conversation, researchers can become too casual, particularly if they personally use Instagram as a means of communication. While a certain level of playfulness can be conducive to this type of interview, it is important to still consider each question we ask, so that it always serves our research. Respecting and valuing participants' time requires ensuring that easy access to interviewees does not result in their exploitation. When a more casual exchange about the topic was initiated by the participant, I always participated (cf. Fritz & Vandermause, 2018, p. 1643. However, when the conversation was about something private and irrelevant to the research, I deflected. Some participants asked about my personal opinions of the memorial after I asked them about theirs, and I answered only if I felt confident that I would not influence their following responses. While researcher involvement in this process can be risky, the medium and the topic of conversation are so clearly centered around personal opinions and experiences that it might seem unnatural if the researcher only asked questions and the participant only responded. This is not the type of exchange we are used to on this platform, and interview settings should feel as natural as possible (cf. Gläser & Laudel, 2010). Constantly re-evaluating one's role as a researcher and adapting to the specific conversation is therefore essential.
One way of engaging more without being too dominant in one's own views is to comment on individuals' responses ( Figure 6), staying in the background while providing support and encouragement (cf. R. Atkinson, 1998, p. 32). In this short conversation, I gave my perspective on something the participant had just elaborated on. This was the beginning of a long conversation, and at one point, Denaise told me: "I am very impressed with your interviewing skills. You share wonderful compliments, then come in with the next question(s)!" showing that this interview method kept her engaged. Therefore, crafting messages carefully is an important aspect of Instagram research and one we must be particularly mindful of since the platform's interface and the, at times, fast and spontaneous back-and-forth of the messages might entice us to react instantaneously. My interviews showed that finding a balance between a professional exchange and a conversation appropriate for the platform helps ensure interviewee retention and higher involvement.

Boundaries of Space and Time
The obvious advantage of leading an asynchronous interview online is that distances in terms of geography or timezone no longer matter. This also means that, as researchers, we can access populations we might not otherwise reach, and participants can respond in familiar settings, ideally putting them at ease and resulting in natural, genuine responses (Stewart & Williams, 2005). However, this must be planned diligently during the initial stages of the project to ensure that only populations of interest to the project are studied: Just because we can access far-away populations through the internet, it does not automatically mean we should study them. For my project, the method made sense because I was studying a tourism site: Individuals come from a variety of places to visit and then return to their homes where social media posting about the site most often occurs. Not being dependent on being "on" at the same time also means that researcher and participant can craft their messages when it suits them. Previous work has shown that, when taking time to construct their responses, interviewees tend to think about their answers more (Hertel et al., 2008). This can be seen in some of the messages I received where an individual acknowledged receiving the question but indicated needing more time, such as this: "Ok this is a good one! I'm a bit busy so ima get back to you when [sic] later today!!" He then responded nine days later (after one reminder) with several voice messages (Instagram restricts voice messages to one minute), which totaled five minutes and six seconds, showing that responding on their own time is helpful for interviewees and results in detailed responses.
One significant disadvantage of online interviews is a lack of commitment, and withdrawing participation is easier when the interview is done asynchronously because all participants have to do is stop responding: They do not have to justify their retreatment to a researcher (Kivits, 2005). The phenomenon of losing participants is exacerbated when the interviewer and participants have never met (Bertrand & Bourdeau, 2010). This is a disadvantage that cannot be denied and one that can be frustrating. However, there were plenty of participants who made an effort to participate. For example, I received this message from one interviewee: "Omg Soo sorry for late response. Just had surgery not too long ago and have been recovering from that," show-ing that there had been an implicit commitment. This conversation then continued for more than 4,000 words. The ease of generating new participants can balance the low response and retention rates: While it is a shame when a participant quits, it is easy to find new interviewees. I always contacted several users in a day which sometimes led to interviews being conducted simultaneously, depending on how fast people responded to my request (some did within seconds while others took several weeks). While this can be straining on the researcher (e.g., Dahlin, 2021), pausing contact with new participants once one interview starts can help keep the workload manageable.

Method-Related Insights
One advantage of the Instagram interview is the chat function and its conventions. Scholars have noted that in non-face-to-face interviews, the interviewer cannot observe body language cues (Meho, 2006(Meho, , p. 1289, meaning that the present method is only suitable for studies in which visual cues are not significant. Their absence, however, can also help the interview situation by reducing social desirability (Fischer, 2009, p. 73). While my profile picture on Instagram makes it clear that I am a white woman, individuals still receive fewer visual cues than they would in a face-to-face setting. It is nevertheless important to reflect on one's visible identity markers and how they might influence a conversation. In my project, ethnicity is a central theme which is why I often explicitly asked interviewees (after the interview) about their impression of me as the interviewer. In the conversation shown in Figure 7, the interviewee frequently brought up his black American identity and how it impacts his opinion of the memorial. By this point, we had been talking for several weeks, which is why it felt appropriate to explicitly talk about my role (cf. R. Atkinson, 1998, p. 35). It was important for Reflecting on one's own identity and its potential impact on responses is crucial but in these personal exchanges, explicitly discussing them can even present an additional value: It allowed me to gain deeper insight into the experience under study.
Another advantage of the Instagram chat is that the language used in online contexts is likely natural (Crystal, 2001). In my project, this was partly due to the familiarity with the platform: Instagram is used in private settings for interpersonal communication. The natural language can be seen in typos which indicate that the participants did not proofread their answers. Participants often used emojis and colloquialisms in their responses, indicating that they were using language "the way they do in most of their everyday interactions" (Meho, 2014, p. 41). This everyday use of the platform is also reflected in individuals sending me additional material: Several participants shared personal photographs to illustrate their visit or included links they found on the topic which they thought were interesting. This gives more context to the experience and shows that an Instagram post about the memorial site does not exhaustively represent a person's interest in the topic. In this sense, it complicates the research process, but it also shows the complexity of engagement which is never quite "done." One insight from the interviews, which has implications for adapting this method, is individuals' insistence that their images "speak for themselves." Interestingly, in their interviews, participants voiced a desire for their audience to learn things that are invisible in their posts. For example, one interviewee did not include a caption (Figure 8), but in her interview, she said she had chosen the picture because: "It makes an impact both visually & meaningfully. He's part of the whole & large like a mountain, but unfinished. There's work yet to be done. I wanted to share ALL of that with others." While this interpretation corresponds to the artist's intention (National Park Service, n.d.), it requires an in-depth reading of the image, one that viewers of the post might not make, also because the gap in the mountain-the "whole"-from which the statue of King is "removed" is invisible in all photographs in this post. Therefore, had I only looked at this post, I would not have been able to conclude what its creator's intentions were, making the interviews with creators a necessary addition to the methodological toolkit for researchers who want to understand the practices behind Instagram posting.
When asked about their experiences on site and what the memorial means to them, several interviewees gave intimate responses. This shows that interviewing through a social media platform-even after a visit-can be an appropriate method to access emotions as it helps individuals open up. One interviewee, for example, told me: "What I would like others to do is show empathy for black people in America because we are hurting and crying for help." And he later continued by writing: Seeing it [the memorial] again was very meaningful to see a black man who looks like me was an ice breaker. When I mean ice breaker I mean that a lot of black boys or girls that grew up didn't get to [see] a statue/memorial of someone black and that make us feel as if we weren't good enough or smart etc that we couldn't lead or move a nation. It served as a big inspiration forreal [sic].
While these types of conversations can also occur in face-to-face settings, the combination of anonymity and intimacy on Instagram can increase their likelihood. However, I also received a message from one interviewee, saying: "Next time you come to DC let me know in advance so we can talk more on the subject. There are certain conversations I won't have on social media." When asked to elaborate on the type of topic, he was hesitant to discuss it online; he only responded: "Let's just say all the data entered and exchanged on the internet passes through a central location for each government. While we have freedom of speech it doesn't mean that we won't [sic] have freedom of consequences." This interviewee is a black man living in America who posted from the MLK memorial saying that people must fight in difficult times (caption abbreviated to ensure anonymity). This individual apparently has critical views of the US government and the public memory of Dr. King, including how it is celebrated at the memorial. However, Instagram is not the place for him to share this in detail, neither in his post nor in an interview. This example reiterates that Instagram posts do not always encapsulate an individual's full opinion of what they show (in this case, the memorial site and, by extension, public memory of Dr. King), making interviews with their creators a necessity. However, it also shows that individuals differ in their willingness to discuss personal and controversial topics online. Researchers must therefore ensure their research topics are suitable for the Instagram interview and consider offering alternative forms of interviewing when interviewing individuals who might endanger their own safety by discussing controversial topics or conspiracy theories, particularly ones that believe that the government is monitoring all behavior. When such groups are of interest to the research, the Instagram interview, just as other online interviews, is likely unsuitable. In those cases, researchers can use Instagram to recruit individuals for offline interviews (see more on this in Section 7). Overall, however-as scholars have claimed for online conversations more broadly (e.g., Pertierra, 2018, pp. 96-97)-Instagram interviews tend to be conducive to personal and intimate interactions.
Lastly, I found that when people are invited to share their experiences and opinions, they seem to enjoy sharing, which helps retain participants. This is counterintuitive to much of what we know about online interviewing: that answers tend to be short and non-committal (cf. Altmann, 2011, p. 101;Fischer, 2009). While that was sometimes true for this study, I also led long and in-depth conversations; in one example, a conversation that spanned two months and totaled almost 6,000 words. Another interviewee switched from text messages to audio recordings, saying: "So I'm going to start sending out voice messages because the questions are getting better and require more extensive answers" showing his interest in the discussion. This further became visible when participants would follow up with me to either hear about updates on the project or to talk about new ideas they had about the topic. In one case, during my second fieldwork, I even met with one participant, Denaise Seals; we visited the memorials together and continued the discussion in person (Figure 9). During our walk, Denaise told me: "I'm grateful to have this with you today because, as I've said [whispers]: I have never seen the reflecting pool. And when you said you wanted to see Lincoln, I went 'oooh' that will be interesting to find it!" This shows that when the interview topic corresponds to individuals' personal interests, not only is their quality of response high, but interviewees themselves can profit from the conversations, which-in a way-compensates them for the time invested.

Overcoming the Online/Offline Dichotomy in Access to the Field
Denaise and I met online through her Instagram posts of the MLK memorial. Before embarking on my second fieldwork, I asked Denaise, who lives in the area, if she would like to join me on a walk around the memorial (as well as the other memorial in this project). She agreed, and we met for a one-hour-and-twenty-sevenminute walk, during which I recorded our conversation and we took photographs that we later shared with each other. Our conversations during this time were connected to what we had discussed online, but they also touched on even more personal aspects. It felt like we already knew each other and going on a walk together was conducive to even more intimate conversations and more spontaneous reactions to the sites. It can also help to gain a deeper understanding of individuals' experiences who are not as comfortable typing lengthy messages or, as mentioned above, discussing controversial topics online. I only met with one individual on site. However, this offline interaction, initiated through online contact, can be purposefully integrated into a project when researchers plan this during the initial stages. For this project, it proved easier to create offline connections through initial online interactions than vice versa. During fieldwork, I asked individuals for their email addresses to contact them after the visit (Fieldwork 1) and gave out my card with my email and Instagram, with an invitation to connect (Fieldwork 2), which resulted in almost no interactions. However, the contacts I made online were eager to meet "in real life." While I only managed to meet Denaise, several others said I should contact them when I was back in town. The Instagram interview can therefore be used as a recruiting method but only after the interview was conducted: This is to ensure that a relationship between researcher and interviewee has been established, ensuring the necessary trust and interest in a joint site visit, as well as to help manage expectations as to the types of conversations to be had.

Conclusion
In this article, I have shown that the inclusion of Instagram interviews in a (digital) ethnography provides more advantages than simply allowing for interactions when in-person interviews are impossible. The method has inherent advantages compared to in-person interviews in the context of visits to cultural sites and research questions that are (also) interested in individuals' visual engagement with the site. Taking advantage of the platform's location-based affordances, it is an online method that allows sampling through an offline location, which not only makes access to a dispersed population easier but actually enables it in situations when offline methods have proven unproductive. As part of a purely digital or a mixed-method ethnography, the Instagram interview can be used to study intersections of offline and online spaces because it is a medium that bridges that gap: Users post about their offline lives on Instagram, visit places in the physical world that they have seen on "the gram," and they use the app to talk to friends they know from online or offline contexts. Using Instagram to interview individuals also means reaching them through the platform on which they share aspects of exactly those phenomena under study: Having logged onto Instagram, participants were already in an appropriate mindset and ready to talk about their travel experiences and their understanding of public memory because the platform affordances and cultural constraints encourage engagement with such experiences.
In an increasingly connected world, few aspects of our daily lives occur exclusively offline. The online realm often impacts our experiences in the offline world to the extent that many experiences can no longer be considered purely "offline." It can be as simple as our movement through a city being guided by an online map or as complex as the potential for creating an Instagram post impacting where we go for dinner, what seat we sit in, and what we order. In order to study these types of entangled experiences-and individuals' thoughts and motivations connected to the experience-I propose the Instagram method as an additional interview method for qualitative research interested in (cultural) tourism experiences. By interviewing visitors to cultural sites through Instagram after their visit, researchers can move beyond learning about individuals' motivations for social media posts, allowing us to contextualize their online content within their experiences occurring on-site (both offline and online). This helps us see how visitors, in fact, use memorial sites and how they negotiate their meanings, including how they post about them on Instagram. Therefore, the Instagram interview should not be understood as an online method of accessing offline behavior: It is a method for interviewing individuals about entangled offline and online experiences on a platform that is perceived to bridge that gap.
Lastly, the main contribution of adding the Instagram interview to the researcher's repertoire when studying individual cultural tourism experiences lies in allowing researchers to study the complex phenomenon of experiencing a cultural site in a manner that acknowledges the complexity of the experience: It does not necessarily end once visitors leave the site. They might reflect on their visit, think about it in the context of their daily lives, reflect on it when they read an article in the newspaper, and, especially when we talk about sites of tourism and public memory, engage with the site actively on social media. The act of posting about an offline experience is not an "after the fact" accounting of an experience; it is part of the experience itself. Posting about a site and making sense of it in the context of one's personal life, one's personal (online) identity, is part of the visit and must be included in the study of the visitor's experience. When we conceptualize visits to cultural sites, or tourism sites more broadly, as spanning across the space and time of the physical presence of the visitor at the site, we must adapt our methods to study these experiences. We must consider the full experience, including its online components: before, during, and after the visit.

About the Author
Larissa Hugentobler is a research and teaching assistant at the University of Zurich, where she is pursuing her PhD in communication and media research. Her research focuses on experiences impacted by digital interactions with the physical world and centers on nonprofessional content creators and marginalized communities.