Have you ever shared a news story over social media? Upon sharing such a story, did you verify if the evidence presented within the article was factual? Often times, the stories easily shared throughout our favorite social media sites such as Facebook and Twitter aren’t factual, no matter how many likes and shares they have. However, the legitimacy of such articles are not often questioned, and they continue to be shared as more and more people believe the content to be factual.
It seems that fake news is a pretty hot topic within today’s political climate. The heated debates about if fake news was present in the 2016 election and what should be done about fake news have continued to develop throughout the news media and via social media these past few years. As we approach the 2020 election, I thought it’d be interesting to conduct a project about fake news for the following reasons:
- to become familiar with research and literature on the problem of fake news in various contexts,
- to practice data manipulation with a pre-existing dataset recorded from a survey about fake news,
- and to gain insight on what has been done to combat fake news since the infamous 2016 election.
The 2018 Fact Opinion Study conducted by Pew Research Center found interesting results about how the typical American views the news they consume in their daily lives. By using a web-based survey with a representative sample of about 5,000 people, the team was able to identify some unique patterns among voters of different ages and political ideologies. Although the data has been used to create articles within the Pew Research website, this project will search for other trends within the dataset.
This article will explore the continuously developing topic of fake news and take a deeper look at the dataset provided by the Pew Research Center through data manipulation methodology through the program R.
Literature Review
In recent years, media and political scholars have been conducting research in attempts to explain fake news relevancy in recent elections (Allcott & Gentzkow, 2017), gain higher insight on the way people perceive the effects of fake news (Stefanita, Nicoleta, & Buturoiu, 2018), and to better understand why fake news is so easily spread throughout the United States (Beach, 2019).
Fake news is constituted of information presented as factual news that is actually surprised of misinformation and false truths (Githaiga, 2019). Fake news often has no inherent truth within its story, but consists of information that is presented as factual to its audience. Githaiga claims that fake news is not a new phenomenon, but the channels in which fake news can be presented and spread has changed throughout the years.
Mark Beach agrees with Githaiga’s sentiment that fake news is not new to the society, claiming that “American journalism has a rich history of jumping back and forth across the line between bias and accuracy,” and believes that today’s issue of fake news is one actively fostered by the age of social media (2019). Certainly, social media provides an easy and quick way to share any type of information.
After the results of the 2016 election were announced, there was a high degree of speculation that an extreme amount of fake news had influenced the election results (Allcott & Gentzkow, 2017). Since then, many conversations about how fake news can affect the democracy of our country have taken place.
Some studies have shown that media consumers often are affected by what is known as the third-person effect when it comes to their beliefs of fake news media, meaning that they believe others are more affected by fake news than themselves (Stefanita et. al, 2018). Due to this effect, people may not take warnings about fake news seriously regarding their own news media intake.
Although much of the literature regarding fake news seems quite negative, some scholars and journalists have presented ways for audiences to notice fake news and be less passive about the media they are passing along to their friends and family. Readers should check the publication of the article to see if it was written from a reputable source, check the language of the article to be grammatically correct, use fact-checking applications and websites, and call out fake news articles they see shared among peers throughout social media (Lamb, 2018). By becoming more active in their news media intake, audiences can combat and prevent the spread of fake news.
By reviewing the dataset provided by the Pew Research Center, I hope to find relevant links between the reviewed literature and data regarding fake news attitudes within the United States.
Methodology
Once again, the data used for this article came from Pew Research Center’s Spring 2019 Survey on Factual and Opinion Statements in the News dataset. This data was collected via a web-based survey using a probability sampling method. The research sample was generated from GfK KnowledgePanel to generate a sample most representative of the US population.
Throughout the questionnaire, the participants were asked about their daily activities, which outlets they receive news from, what type of news they are interested in, how much they trust news outlets, and more. Other sections in the survey including party affiliation and if they could distinguish certain statements as facts or opinions.
The survey took place from Feb. 22-March 4 of 2018, and was available in English and Spanish. The total number of respondents ended up at 5035. The data was analyzed in two different forms during this study, a longform dataset and a shortform dataset. This article will be focusing on the shortform dataset provided.
The Pew Research Center’s study used SPSS as a tool for manipulating and analyzing the data. To learn more about their research methods regarding SPSS and view the study’s datasets and surveys, visit the Pew Research Center website.
For this project, I used R to statistically and visually analyze data presented within the survey. After cleaning up the data for variables that I was interested in studying, I began to analyze the data. I performed simple functions to get a better look on some of the variables, such as ‘NEWS_ACTION_a’, which was how likely a participant was to share a news item with someone. In terms of visualizations, I first created a barplot to represent the distribution of political affiliations of those who participated in the survey. Then, I created a pie chart to visualize the respondent’s political ideologies. Lastly, I created a stacked barplot to visualize one of the questions asked in the survey.
Results
Although there were more democrats in within the participants of the study, it seems that there was a pretty even distribution throughout republicans (1437), democrats (1633), and independent (1438) affiliations throughout the study, with a small amount of participants reporting ‘something else’ (423).

Next, after revaluing the survey data, I was able to create a pie chart with a legend that visualizes participant ideology. As the chart shows, a wide percentage of participants considered themselves to be of moderate ideology (42.7%), while smaller percentages fall under conservative (25.5%) and liberal (16.5%). Much smaller percentages fall in the ‘extremes’ of the two sides.
After making these visualizations, I ran some code to find out how likely our respondents were to share a news item via online or hardcopy. Of those who responded, 259 answered often, 1374 answered sometimes, 1482 answered hardly ever, and 1853 said Never, meaning around 37% of survey participants never share news with others.
Lastly, I made a visualization presenting a likert scale question from the survey. The question asked ‘How much, if at all, do you trust the information you get from…?’ for categories, social media sites, national news organizations, and friends, family, and acquaintances. From this visualization, we can see that participants trusted information the most from news outlets and their family and friends in comparison to social media sites.
Discussion
Firstly, I was pretty satisfied that there was such an even distribution regarding party affiliation in respondents, and a little bit surprised that there were so many independent participants, choosing either side. This may be a sign of a developing mindset that one does not have to choose one side or the other, and a fair amount of citizens are creating their political ideology by standing for various issues they believe in. Following this, it is notable that very small percentages of this representative sample fell under ‘Very Conservative’ (7.3%) and ‘Very Liberal’ (5.5%) ideologies. I decided to provide both of these visualizations to give a general idea of how representative the sample was.
What was most interesting to me was the visualization regarding how much participants trusted information from various sources. Because of the fake news epidemic described in the literature, I figured that people would be a bit more trusting of these sources. However, they may have just been reporting what they think the questionnaire is searching for. With ‘fake news’ being a term paraded throughout mainstream media, people know that it is wrong to trust information and news on social media, however, they still may do it.
I think it’s interesting that participants chose that they trusted ‘some’ information given from family and friends most compared to the other categories. This is because you don’t truly know where those people are getting their information from. The fact that social media favored less heavily results was shocking to me, mainly because I hear people use an iteration of the phrase ‘I saw this on Facebook…’ all the time.
37% of respondents said that they don’t share any news with their peers, which is interesting within itself. I wonder if they don’t share news due to the fact that they don’t know if the information they may spread is reliable, or if they aren’t interested in news in general.
Conclusion
In conclusion, the sample used by the Pew Research Center was distributed pretty well in terms of political affiliation and ideology, and provided great insight on how people spread and receive news regarding politics and other issues. Because the dataset is so representative, the results are fairly truthful to make different observations about news and fake news within the United States. The dataset is massive in the variables it covers, so feel free to check it out if you’re interested in learning about the different discoveries they’ve made. By using this data, analysts will be able to make even more inferences about fake news and in the process, help spread awareness to combat such falsities.
References
Allcott, H., & Gentzkow, M. (2017). Social media and fake news in the 2016 election. Journal of Economic Perspectives, 31(2), 211–236.
Beach, M. (2019). Fake news, truth and trust. Media Development, 65(1), 38–41. Retrieved from http://search.ebscohost.com.libproxy.siue.edu/login.aspx?direct=true&db=ufh&AN=137132938&site=ehost-live&scope=site
Githaiga, G. (2019). Fake news: A threat to digital inclusion? Media Development, 65(1), 35–38. Retrieved from http://search.ebscohost.com.libproxy.siue.edu/login.aspx?direct=true&db=ufh&AN=137132937&site=ehost-live&scope=site
Lamb, B. (2018). FALSE WITNESSES: Fact and Fiction in the Age of Fake News. Screen Education, (91), 94–99. Retrieved from http://search.ebscohost.com.libproxy.siue.edu/login.aspx?direct=true&db=ufh&AN=131888168&site=ehost-live&scope=site
Stefanita, O., Corbu, N., & Buturoiu, R. (2018). Fake News and the Third-Person Effect: They are More Influenced than Me and You. Journal of Media Research, 11(3), 5–23. https://doi-org.libproxy.siue.edu/10.24193/jmr.32.1