CONSULTING.de
PERSONALintern.de
Oliver Frangakos, Dynata „KIs wie ChatGPT stellen eine Herausforderung für die Qualität von Marktforschungsdaten dar“
veröffentlicht am
Milena Tsvetkova, London School of Economics and Political Science "All information that we obtain online, has passed through a social filter."
veröffentlicht am
Milena Tsvetkova, London School of Economics and Political Science
marktforschung.de: Ms Tsvetkova, you are researching and teaching at the renowned London School of Economics and Political Science. What is currently the focus of your research?
Milena Tsvetkova: In general, my research focuses on studying group- and population-level phenomena such as social contagion, cooperation, and segregation. I study these phenomena with digital trace data from online communities, as well as with data from virtual lab experiments. My current research focuses on studying how inequality emerges from social interactions. I am collaborating with computer scientists and software developers to design games that will allow us to conduct experiments with volunteers online. I am also exploring different ways to study inequality and socio-economic status using digital trace data.
How does the social context influence our online behaviour?
Almost all information that we obtain online has passed through a social filter. We buy products based on what others recommend, we read news because our friends share them, and we post pictures and opinions in order to find affirmation from others. We can think of any machine learning algorithm that curates the information you receive on a website as social context, since these algorithms summarize the behavior of others.
What is the special challenge about analyzing social interactions and groups online?
The most obvious challenge is that focusing on groups dramatically reduces the "N", i.e. the number of available observations. This may sound like a preposterous statement in the era of big data but, for example, while Wikipedia may have hundreds of thousands of active contributors, it represents only one community. Another challenge is the fact that everyone is connected with everyone online. This makes it very hard to study social influence, for example, or trace causal processes in general. We need more advanced statistical techniques and algorithms to do so.
The GOR conference brings academics and practitioners together. Could you give us a sneak peek into your keynote speech: what would be of interest to academics, to practitioners, or both alike?
I know that online survey research is one of the main focuses of the conference, so I designed my talk to be somewhat provocative. For example, I will present a study that uses an online survey in which individuals judge the photos of social media users. We find that one’s judgement of another person depends not only on the two persons’ characteristics but also on how similar these two people are. In other words, researchers/practitioners cannot expect independent opinions from individuals on information that has social context. I will also present another study in which we try to do more robust research with less data, arguing against today’s mantra that the bigger the data, the better.
GOR 20 takes place in Berlin. Have you been to Berlin before? What are you expecting from the GOR conference?
I love Berlin! I spent two summers there, for a total of about 4-5 months, and I am always happy to get an opportunity to go back. I have not attended the GOR conference before so I look forward to hearing about exciting new research, forging new collaborations, and making new friends.
Kommentare (0)
Noch keine Kommentare zu diesem Artikel. Machen Sie gerne den Anfang!
Um unsere Kommentarfunktion nutzen zu können müssen Sie sich anmelden.
Anmelden