Transcript
You're listening to RadioEd, the ÌÒ×ÓÖ±²¥ podcast. I’m your host, Jordyn Reiland.Ìý
Think back to how you made decisions growing up. Whether it was the clothes you wore, where you hung out, who your friends were, and even how you coped with the struggles you faced—who did you confide in?ÌýÌý
Was it a friend, a sibling, a parent, a mentor or other trusted adult? It’s likely that peer influence—or how much your personal circle can affect what you do and think— had a big impact.ÌýÌý
Peer influence has the power to shape nearly every decision a young person makes, and social media often plays an outsized role in how those everyday conversations occur.Ìý
Approximately 95 percent of youth ages 13 to 17 report using a social media platform—according to recent data reported by the Office of the Surgeon General.ÌýÌý
Social media can also help researchers understand the context of certain health-related behaviors by offering real-time insights into trends, interactions and peer influences within online communities.ÌýÌý
For example, it can shed light on the challenges faced by vulnerable groups, such as youth experiencing homelessness, who may turn to digital platforms for support, connection, or expression of their struggles.Ìý
Anamika Barman-Adhikari (01:12):ÌýÌýÌý
Essentially these networks are, what kind of people these young people are connected to, how these connections provide support and encouragement and discouragement from different kinds of behaviors, and also where that support might come from.Ìý
Jordyn Reiland (01:31):ÌýÌý
That’s Anamika Barman-Adhikari, an associate professor of social work at the ÌÒ×ÓÖ±²¥. She researches how the social networks of young people experiencing homelessness influence behaviors like substance use.Ìý
By using large language models like AI, hundreds of thousands of Facebook posts and survey data, Barman-Adhikari and her colleagues investigated connections between the content of people’s Facebook posts and substance use.Ìý The goal of the research was to use the Facebook posts to identify and address substance use before they enter adulthood.Ìý
Barman-Adhikari identifies two types of peer groups that young people experiencing homelessness are connected to: what she calls street peers or those who are also unhoused, and home-based peers, or the friends or family members they had before they became unhoused.ÌýÌý
Home-based peer connections are often maintained online.Ìý
Anamika Barman-Adhikari (02:18):ÌýÌýÌý
And one of the things that I noticed in my research, the more I talk to these young people, is that a lot of these home based connections were maintained via technology, so via Facebook, via Instagram, cell phone, email, phone calls, and that told us that the traditional conventions that we hold about these young people not connected to technology, really, you know, is inaccurate. And we can use these social media forums to better understand their more organic interactions and also understand the context of those interactions. So how are they posting and what kind of responses are they getting from their friends? Which is not possible with traditional survey research or even qualitative research.Ìý
Jordyn Reiland (03:12):ÌýÌý
Barman-Adhikari says the idea that phones are not a daily part of life is one of the biggest misconceptions about youth experiencing homelessness. In fact, her data found that roughly 80 percent of unhoused youth reported using social media weekly—and only 9 percent did not have a social media profile.Ìý
When it came to the type of platform, approximately 90 percent of youth who used social media preferred to use Facebook.ÌýÌý
Barman-Adhikari says the use of Facebook specifically may have to do with the amount of time and limited access to Wi-Fi unhoused teens have, compared to using other platforms like Instagram or TikTok that may take more time on the backend.Ìý
Anamika Barman-Adhikari (03:46):Ìý
A lot of their cell phones, they have to access Wi-Fi in grocery stores or at a Starbucks, at a library. So they might not have as much time to make a video for TikTok, you know, do all the editing. So it might just be a quick Facebook post. So that might be more convenient, rather than doing video and audio editing.Ìý
Even though these young people are homeless, they're still digital natives. They're still plugged in. They grew up with computers. They grew up with social media. And, you know, a lot of their lives are spent and are built and their lives are lived on social media. So I think you can really, you know, understand those naturally occurring interactions, and then use natural language processing or AI to find patterns in those interactions and how they're related to different kinds of health outcomes, and in my case, substance use.Ìý
Jordyn Reiland (04:43):ÌýÌý
Barman-Adhikari knew digital peer influence played a role in the factors that impacted substance use among young people experiencing housing instability or homelessness.Ìý
However, it wasn't until she met one particular youth experiencing homelessness that she realized the subtilties and nuances of peer influence online. Barman-Adhikari uses the alias Marcus to protect the youth’s identity.ÌýÌý
Anamika Barman-Adhikari (05:04):ÌýÌý
Marcus had been experiencing homelessness since he was 16. He was couch surfing and bouncing between shelters, sometimes sleeping on the streets. And, you know, we started talking about his substance use. And when we talked about his substance use, he didn't describe it as a simple choice. It was not like, ‘Oh, I just decided to start using,’ right? It was deeper than that. He told me how the people around him, which he called his survival network, all use substances to stay awake at night, to suppress hunger and also to numb their emotional experiences, emotional needs and the instability on the streets. And essentially, what that told me was that saying no to substance use was not just about willpower, right? It was about survival. And for Marcus especially, it was about being with the people who had his back. It was being around the few networks that he had access to.ÌýÌýÌý
For most people, substance use is not a choice. They do it because it meets some kind of need. It could be an emotional need, or, as I said before, it could be a survival need. And so instead of stigmatizing substance use, instead of penalizing substance use, we need to provide better supports and resources. And what my data has shown is that as long as we are able to meet their basic needs and their emotional needs and provide them with quality resources. We can find a solution to substance use among this group of young people, and I'm hoping that more people will believe in that message and be willing to contribute to this cause.Ìý
Jordyn Reiland (07:14):ÌýÌýÌý
What surprised you most when you first started looking at these Facebook posts, and were there any patterns or findings that you didn't expect to see?Ìý
Anamika Barman-Adhikari (07:23):Ìý
Yes. So one of the most interesting things that we found was that a lot of times when these young people were talking about financial problems, that's when they were more likely to use substances, which didn't come up with our survey data. Because we were looking at emotional responses. We were looking at trauma, we were looking at other indicators of substance use, and through our AI based analysis, we found that financial stress is a huge trigger beyond all these other factors.ÌýÌý
Jordyn Reiland (07:59):ÌýÌý
Barman-Adhikari also identified Facebook posts that showcased the ways in which some unhoused teens dealt with stress.Ìý
Anamika Barman-Adhikari (08:06):ÌýÌý
This person's friends or peers, whoever they were interacting on Facebook, instead of suggesting mental health resources, were suggesting alcohol, or get a joint, or come hang out with us and we are gonna chill out, which is code for we will use drugs together. So even that tells us a lot about their environment. And it's not that these young people want to intentionally encourage bad behavior, what we see as bad behavior. These young people also don't have other resources. So all they can think about is what's available to them.Ìý
ÌýJordyn Reiland (08:44):Ìý
You mentioned in your research that social media can provide valuable digital trace data. What does that mean?Ìý
Anamika Barman-Adhikari (08:51):ÌýÌý
What we did through our study is that we scraped data from Facebook feeds of young people experiencing homelessness who gave us consent to collect that data. So we tried to be as ethical and make sure that youth knew what we were doing, and also fully consented to their participation in the study. And one of the reasons why we did this is because survey data is imperfect because, first of all, there are recall biases. You ask young people something and they might forget, right? And the second is, their social desirability issue. They might not respond honestly. On the other hand, social media allows us to study these real time interactions more organically and also understand the emotional tones of those interactions and multiple interactions, right? It's not just interpersonal one-on-one interactions, it's multiple interactions. And we collected data across one year, and we collected almost 60-70,000 posts and responses.ÌýÌý
I was very naive when I first started this study. I'm like, ‘Oh, I'm gonna collect all these textual data, 60-70,000 posts, and I'm gonna find a way of analyzing them.’ And pretty soon, I figured that, no, I needed advanced technology to sift through all this data and make sense of it. And traditional statistical analysis needs structured data, and we had all this unstructured data, volumes and volumes of unstructured data. That's when I connected with faculty at Daniel School of Business, and they suggested natural language processing, and that's how we were able to convert all this unstructured data into structured data for analysis. And some of the findings were really, really interesting. Yeah, so that's digital trace data for you all this Facebook, all the things that we post online can be scraped and actually analyzed, and that's how we get our Netflix algorithms, right? Your feed, everybody's feed is unique. Your Facebook feed is unique because what the computer does is every click you make, every like, everything that you click on, it keeps track. And then it builds an algorithm that is unique to your needs and is based on your previous behavior.Ìý
Jordyn Reiland (11:34):ÌýÌý
Now I know you spoke about Marcus. Do any other conversations with teens from your research stand out to you?Ìý
Anamika Barman-Adhikari (11:41):Ìý
So many. But you know one of the ones that really just pulled at my heartstrings is this young woman, young person who was trying to get housing. She was on a million wait lists for different types of housing. She had a job, she had her life together. And one day, she brings this picture to me, and it's of this abandoned warehouse with used condoms, used syringes, and in the middle of the room there is this mattress where she sleeps. You know, after looking at that, it was just so difficult to imagine myself when I was that young, at 19,20, to imagine myself in those circumstances. I have children now, and I don't think a lot of people could survive those circumstances. And this young person was still getting up every day, going to her job, finding her basic necessities, getting a shower, trying to find housing, while still maintaining a job and her sanity. And if that's not resilience, I don't know what resilience is. But what is still disheartening is we shouldn't be treating young people like that in the most, richest, developed country in the world.Ìý
Jordyn Reiland (13:02):ÌýÌý
You talked a little bit about some of the collaborators in the work that you've been doing, and I'm curious if you have any more examples of how working with the experts from various fields like computer science, mental health, public policy, all the different ones that you've also spoken about, how has that expanded your understanding of this problem and influenced the direction of your research?Ìý
Anamika Barman-Adhikari (13:23):Ìý
It's been very interesting to do interdisciplinary research and get to know how different people view problems through their own professional lens. So with computer science folks, data science folks, everything is about precision. It's all about optimizing some kind of outcome. And so for me, as a social work scholar, what I cared about is more of an in-depth understanding, more of a substantive, qualitative understanding. And for a lot of computer scientists, very rightfully, that's not enough. Like, how accurately are we predicting this outcome?Ìý
With traditional statistical models, we are able to accurately predict something probably 15% of the time. With these models, so for example we predicted Marijuana use with 81% accuracy. And when we also accommodated, incorporated other features, like emotions and survey questions, demographics, that increased to almost 85% accuracy. I couldn't have even imagined that if I was working in my own professional silo. So that's something I have really learned from them. How do we predict things with precision?Ìý
Jordyn Reiland (14:47):ÌýÌý
Why should people care about your research?Ìý
Anamika Barman Adhikari (14:50):Ìý
First of all, AI is everywhere. I mean, look at our own lives. If you used a GPS to get here today, or you used your phone, you used your car, everything relies on AI. And I'm not saying AI is a perfect solution, or technology is a perfect solution for everything, but when it comes to vulnerable groups of people, not just young people experiencing homelessness, but vulnerable groups of people, our understanding of how to use technology and AI is really lagging behind. We are in the dark ages when it comes to marginalized groups of people. And technology is not the solution, but it can be a way to better first provide resources. Especially for this group of young people, we have to meet them where they are.
Jordyn Reiland (15:41):Ìý
The dark ages, as Barman Adhikari puts it, is not a place to be when dealing with an issue that impacts so many Americans.ÌýÌý
Results from the 2023 National Survey on Drug Use and Health show that of people aged 12 or older, 59 percent used tobacco, vaped nicotine, used alcohol, or used an illicit drug in the past month.
Those numbers can be even higher for unhoused youth, according to Barman-Adhikari. Lifetime rates of use can range from 24 to 69 percent for methamphetamines and 12 to 15 percent for heroin among the population.
Barman-Adhikari says that communities need to have conversations at the intersection of technology and fields like social work, and about how technology can be used to understand and help vulnerable populations better.
If you could leave one message for policy makers, social workers or community leaders listening today, what would it be in terms of addressing substance use among youth experiencing homelessness in a more effective, data driven way?Ìý
Anamika Barman-Adhikari (16:43):ÌýÌý
Can you imagine if all of these young people had cell phones and smartphones, and we could connect with them and collect all this data and then send them resources and supports? I'm not saying that everybody can be helped with that, but there's a segment of the population that we could reach very efficiently and very effectively. Let's find money to provide them. And a lot of people say, ‘Oh, what if they sell it?’ These young people, their cell phones are even more important to them than the clothes on their back. It's the last thing they'll sell. And that's what most of these young people have told us. So, yeah, we can give them this very basic resource. It's not a luxury item anymore.Ìý
Jordyn Reiland (17:21):Ìý
Yeah, and like you said, it becomes a lifeline. It's a way to communicate with their networks, and really just be aware of the world around them.
Anamika Barman-Adhikari (17:28):Ìý
Even finding a bus route, finding an HIV test, finding the nearest emergency room. Can you imagine living without your phone? I cannot.Ìý
Jordyn Reiland (17:40):Ìý
Ultimately, at the heart of Barman Adhikari’s research is relationships—and the ways in which even one person can make a difference in someone’s live.Ìý
Anamika Barman-Adhikari (17:49):Ìý
Even one person, one trusting adult, can completely change the trajectory of their lives. Yeah, if you can do that, and you can be compassionate and look out for these young people and try to really combat a lot of the stigma that comes with homelessness, I think these young people would have so many more positive outcomes. They have a lot of potential.Ìý
Jordyn Reiland (18:21):ÌýÌý
A big thanks to our guest, ÌÒ×ÓÖ±²¥ associate professor Anamika Barman-Adhikari. More information on her work can be found in the show notes. If you enjoyed this episode, I encourage you to subscribe to the podcast on Apple Music or Spotify—and if you really liked it, leave us a review and rate our work. It really helps us reach a larger audience—and grow the pod.ÌýÌýÌý
Emma Atkinson is my co-host. Joy Hamilton is our managing editor. Madeleine Lebovic is our production assistant and musical genius, and James Swearingen arranged our theme. I'm Jordyn Reiland and this is RadioEd.Ìý
