ABSTRACT
Does exposure to news affect what people know about politics? This old question attracted new scholarly interest as the political information environment is changing rapidly. In particular, since citizens have new channels at their disposal, such as Twitter and Facebook, which increasingly complement or even replace traditional channels of information. This study investigates to what extent citizens have knowledge about daily politics and to what extent news on social media can provide this knowledge. It does so by means of a large online survey in Belgium (Flanders), in which we measured what people know about current political events, their so-called general surveillance knowledge. Our findings demonstrate that unlike following news via traditional media channels, citizens do not gain more political knowledge from following news on social media. We even find a negative association between following the news on Facebook and political knowledge. We further investigate why this is the case. Our data demonstrate that this lack of learning on social media is not due to a narrow, personalized news diet, as is often suggested. Rather, we find evidence that following news via social media increases a feeling of information overload, which decreases what people actually learn, especially for citizens who combine news via social media with other news sources.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Data Availability Statement
The data described in this article are openly available in the Open Science Framework at https://doi.org/10.7910/DVN/D0COF1
Notes
1. Note, however, that in the end, due to a higher attrition rate between the two waves among respondents below 30, our final sample is not fully representative on age and on average slightly older. To test whether this potentially affects the results we interacted the main effects with age, but we found no different effects for different age groups.
2. Given that some items were more challenging than others, we also used a jack-knife procedure -each time leaving out one of the political knowledge items- to check whether results remain robust when dropping an item. This was the case, indicating that our results are not driven by one single item.
3. A principal component analysis assumes that all items have the same difficulty. Given that this is not the case for our knowledge items, a Mokkenscale procedure that takes this difference in difficulty into account is recommended. Items can be included in the scale when the Loevinger’s H is above.40. We refer to Mokken (Citation1971) for the technical details.
4. We mainly opted for this categorization in order to have a sufficient large percentage of Facebook reliant users (7%) in our analysis. Moreover, since most respondents are likely to over report on how many news sources they use (due to the social desirability related to self-reporting (Araujo et al., Citation2017)) the actual number of sources used will most likely be lower for most of these respondents. We also realize that the number of traditional news sources does not necessarily say anything about the attention respondents give to these sources. Therefore, to be more certain about the validity of the categorization, we checked whether Facebook reliant users and users with a low news diet score significantly lower on our items that measure the frequency with which they use traditional news channels (TV, radio, newspapers and online news websites). A regression (Appendix A) shows that this is the case.
5. Given the lower factor loading of the third item, we also conducted each analysis with a scale based solely on the first two items. In the end, the results remain similar with this alternative scale.
6. We use a measure of information overload from wave 1 because of item availability. However, given the strong correlation between the type and amount of media use in the first and second wave, we do not expect this to strongly impact the results.
7. We tested several other model specifications. Given the somewhat skewed distribution of our dependent variable we ran models with a log transformation of political knowledge and we tested a Poisson model. Both alternative model specifications yielded similar results as the linear regressions. For reasons of interpretability we report the latter.
8. Due to the runtime of our survey (June 5–14), it is possible that early survey participants may have a better recall of the answers to the questions than those who completed the survey in the last few days. When we compare the two groups (early: June 5–10; late: June 11–14), we indeed notice that the early survey participants score slightly higher on political knowledge than the late participants (early:3.07, late:2.82). However, this difference does not drive the results as they remain similar when controlling for the date of the completion of the survey. Moreover, the negative effect of Facebook news holds for both early and late survey participants.
9. Also if we purely look at the descriptives of the personalized news environment variable (Appendix C), the idea that people live in filter bubbles should be nuanced.
10. Note that these results also hold when we use the media frequency measures from wave 1.
Additional information
Funding
Notes on contributors
Patrick F. A. van Erkel
Patrick F. A. van Erkel is a postdoctoral researcher at the University of Antwerp (Belgium). His research focuses on electoral and political behavior, political communication and political polarization.
Peter Van Aelst
Peter Van Aelst is a professor of political communication at the University of Antwerp (Belgium) and a founding member of the research group Media, Movements and Politics (www.M2P.be). His current research focuses on the relationship between politicians, journalists and citizens in the digital age. He has published extensively on agenda-setting, election campaigns and the interactions between journalists and politicians in comparative perspective.