BigSurv20 program
Friday 6th November Friday 13th November Friday 20th November Friday 27th November
Friday 4th December
Back
Embrace the trace? Exploring digital trace data 2Moderator: Trent Buskirk (trent.Buskirk@umb.edu) |
|
| Friday 13th November, 11:45 - 13:15 (ET, GMT-5) 8:45 - 10:15 (PT, GMT-8) 17:45 - 19:15 (CET, GMT+1) |
Combining digital traces and survey data to study media effects: A mobile lab approach
Ms Felicia Loecherbach (Vrije Universiteit Amsterdam) - Presenting Author
Mr Mickey Steijaert (Vrije Universiteit Amsterdam)
Ms Veerle Koopal (Vrije Universiteit Amsterdam)
Ms Josephine van der Erve (Vrije Universiteit Amsterdam)
Dr Mariken van der Velden (Vrije Universiteit Amsterdam)
Dr Kasper Welbers (Vrije Universiteit Amsterdam)
Dr Judith Moeller (University of Amsterdam)
Dr Wouter van Atteveldt (Vrije Universiteit Amsterdam)
Dr Damian Trilling (University of Amsterdam)
The shift to online, especially mobile, news consumption has led to the creation of unprecedented amounts of data on individual news consumption behavior. Using these digital traces, we can now know which respondents viewed (and liked, shared, and commented on) which news articles at which time, allowing a much more fine-grained study of media effects compared to the ‘classical’ method of linking media content to (panel) survey data. A more precise measurement of news exposure allows us to investigate certain features of an individual’s online (and especially mobile) news diet and the context that people encountered content (e.g. on which platform, accompanied by appraisals of others). However, the collection of these digital traces faces technological, legal and ethical challenges. Previous studies mostly relied on the use of intrusive tracking software, that participants need to install on their devices. In this study we propose a mobile lab approach, where people at different events are asked to donate their digital traces.
Of course, any project that sets out to track users’ news consumption should ensure that it only does so after receiving explicit informed consent and in compliance with high standards of privacy protection and security. To convince users, companies and policy-makers that there is a case for social scientists to obtain and retain access to digital trace data, it is imperative that we do all we can to only gather and use this data in compliance with high ethical standards.
In contrast to tracking approaches, our approach is fully based on “data donations”: We do not collect any data on the background, but respondents actively donate already existing data (that is present on their devices or at a service provider they use). This includes browsing histories and social media takeouts (Facebook, Twitter, Whatsapp). Because of our lab-based approach, the respondents are physically sitting next to us and therefore have full control about the process of collecting, extracting, and processing their data; in particular, they can exclude any possibly sensitive part of the data before donating it. As an additional benefit, our approach gives participants immediate insight into their news consumption behavior, as well as which data which companies store about them. This can be used to enhance their media literacy.
The digital traces that are gathered can then be combined with (quantitative and qualitative) survey data to better understand how the content that people consume is related to their knowledge, attitudes, perceptions, and behavioral intentions. In this project, four different aspects are more closely evaluated: (1) How does the diversity of the news diet (regarding e.g. actors, topics, and viewpoints) relate to perception of public opinion and political attitudes? (2) How prevalent are shared news appraisals and expression of partisanship on different social media, and what form do they take? (3) What knowledge do people gain from their news consumption and is this dependent on platform and personal characteristics? (4) Does people's online news consumption influence democratic attitudes, like participation in elections?
Visits to fake news sites: Who goes and how did they get there?
Dr Tom Paskhalis (NYU-SMaPP)
Professor Jonathan Nagler (NYU-SMaPP) - Presenting Author
We combine a nationally representative two-wave panel of 1500 survey respondents collected during the 2018 United States Legislative elections with detailed information on the web-browsing activity of respondents collected via a browser plug-in. Self-reported information obtained from survey responses: online social media activity - including frequency of posting about politics, offline media consumption, perceptions of the ideology of their online networks, ability to answer a series of questions about politically relevant facts correctly, and belief in the prevalence of fake news.
Using the web-tracking data we are able to examine a series of questions about how voters reach fake-news websites online. We are able to determine what types of sites voters were at immediately before going to fake news sites. And, since we have full web browsing history - we can also see the distribution of news-consumption for each individual, and determine what media diet is associated with visiting fake news sites. In other words, are readers of fake news sites also consuming quality news online, or is fake news consumed by people for whom it is a substantial part of their media diet?
We then examine several key research questions by analyzing the web browsing behavior combined with our survey data. First, we examine the types of respondents likely to visit fake news sites. In particular, we can find out if these are people who claim to be politically active online: both in posting activity on social media, and in frequency of consumption of political information. And of course we can examine the demographics of people who go to fake news sites, as well as examine their (self-reported) offline media consumption. And, since we asked questions of factual knowledge in both waves of our panel, we can examine whether visits to high quality news websites were associated with increases in political knowledge.
Integrated web tracking and surveys to study selective exposure to news by populist radical right party supporters
Dr Pascal Siegers (GESIS Leibniz-Institute for Social Sciences) - Presenting Author
Dr Johannes Breuer (GESIS Leibniz-Institute for Social Sciences)
Dr Sebastian Stier (GESIS Leibniz-Institute for Social Sciences)
Computer-mediated activities have become deeply ingrained in political life. People use digital technologies to get political information, discuss politics or advocate for political causes. The measurement of these activities poses a considerable challenge for researchers since they are distributed across multiple channels and platforms, often intertwined, and ephemeral. Researchers have attempted to study these phenomena using survey methods that, however, suffer from the unreliability of self-reported media use (due to social desirability and/or recall bias). Studies in computational social science, on the other hand, collect digital traces of human behavior in a non-intrusive way. However, these approaches oftentimes do not include relevant attributes of research subjects (e.g., demographic characteristics) and outcome variables (e.g., party identification). Our project synthesizes these two paradigms as they have the potential to compensate for their respective weaknesses when combined in a systematic way.
Leveraging such an integrated research design with various established and novel data sources, our paper investigates to what extent supporters of populist right-wing parties engage in selective exposure to news online. While experimental research has shown that anti-media rhetoric by political parties or politicians can sow distrust towards the media among their supporters, it is still unclear whether these party cues also affect the actual selection of political information, or even result in a general news avoidance. In our analysis we concentrate on supporters of the German right-wing populist party Alternative for Germany (AfD) whose politicians routinely accuses the mainstream media of political bias. For the analysis we use mediation models that allow us to disentangle the effects of party cues and media distrust.
In study 1, we use self-reported exposure measures from a large representative survey of 2,930 participants (Rolling Cross Section component of the German Longitudinal Election Study 2017). Results show that AfD supporters consume less news from legacy press and public broadcasting sources but rely more on online information sources than supporters of other German parties. Study 2 relies on a dataset that links web browsing histories (web tracking data) of 1,219 panelists to their responses in an online survey. We again find a negative relationship between AfD support and website visits for legacy press and public broadcasting sources, but a highly significant positive effect on the consumption of hyperpartisan news. In both studies, direct effects of AfD support were stronger than the mediated paths via media distrust.
Overall, the findings indicate that anti-media party cues by populist right-wing parties and politicians drive their supporters away from established media sources towards less-balanced digital sources of political information. These tendencies have implications for public opinion formation in contemporary democracies where populist radical right actors and mistrust of political elites have become entrenched. The results also demonstrate the added value of the web tracking data, as politically sensitive issues, such as visits to hyperpartisan news and the actual intensity of online news use, are hard to study with just survey self-reports.
Linking web tracking and survey data to improve the study of online pornography consumption
Dr Pascal Siegers (GESIS Leibniz-Institute for Social Sciences) - Presenting Author
Mr Maximilian von Andrian-Werburg (Julius-Maximilians-Universität Würzburg )
Dr Johannes Breuer (GESIS Leibniz-Institute for Social Sciences)
The advent of the internet has dramatically increased the availability of sexually explicit media (SEM). Existing research on SEM, however, relies on self-reported porn consumption as it is measured, for example, in the US General Social Survey since the 1970s. Several studies have looked at the relationship between pronography consumption and demographic factors. Results show that SEM use in adults tends to decrease with age (Price et al., 2016). For women there further seems to be an inverse relationship between religiosity and SEM consumption, but not for men (Martyniuk & Dekker, 2018). Findings based on self-reports, however,are likely to be biased due, for example, the high sensitivity of the questions, resulting in a lack of external validity (e.g., Ferguson & Hartley, 2009; Fisher & Kohut, 2020). As social desirability is a prominent issue in SEM research, the use of behavioral data is a promising approach to counteract bias.
In our study, we use web tracking data (URLs on domain level) from a period of one year (June 2018 to May 2019) with approximately 2,000 German participants per month to test and extend previous findings on the association between pronography use and sociodemographic attributes. N = 1323 of the panelists also completed an online questionnaire in which we asked for basic demographic attributes as well as a set of attitudes and personality traits, including ambivalent sexism, religiosity, social dominance orientation, and moral foundations. We further did a basic content analysis of the most popular SEM websites in our sample and derived categories like video sites, erotic story blogs or red light/prostitution guides. Hence, our data allows us to not only look at overall online pornography consumption but also to identify different usage patterns.
First results show a curvilinear relationship between age and SEM use for men, thus, challenging the findings from previous research. In our presentation, we will also present findings on the predictors (demographics, attitudes, personality) of (different types of) online SEM use as well as factors that determine temporal patterns in online SEM use. In addition, we will also discuss potential biases of digital behavioral data in SEM research.