One area of research that has fascinated me for the longest time is whether or not crowds’ responses to a particular question are correct. This is an issue that has been widely debated in academic circles. And the literature on this subject is far from conclusive.
While experiments in crowdsourcing are not always successful, thanks to the growing influence of social media, journalists have used these experiments to great effect. In 2015, Columbia University’s Tow Center for Digital Journalism released its Guide to Crowdsourcing, highlighting different reasons why so many news organizations have adopted these techniques. One of the report’s main takeaways is that when done right crowdsourcing empowers “people to share what they know individually so that journalists can communicate the collective information” (p. 14).
Journalists who are willing to use these techniques recognize the limits of their own knowledge. Thus their “call-outs” are designed to help them get a deeper-understanding of their news stories and hence improve the quality of their reporting. Crowdsourcing journalism changes the dynamics between the journalist and readers, giving readers an opportunity to shape how media organizations cover a particular event.
A good example of crowdsourcing journalism is CBS News Correspondent David Begnaud’s March 16, 2018 “call-out” to his social media followers for suggestions on issues to cover and places to visit in Puerto Rico six months after Hurricane Maria. His reports echoed his followers’ suggestions, as he noted in a video he recorded at the end of his trip to the island. For more information on Begnaud’s methods, read my earlier post.
Begnaud’s crowdsourcing experiment gives us an opportunity to test how knowledgeable the respondents are about Puerto Rico’s challenges post-Maria.
As I noted in an earlier post, using Pablo Barbera’s rFacebook package for R, I downloaded 2,658 responses to Begnaud’s “call-out” for information in Facebook. I utilized Julia Silge’s and David Robinson’s Tidytext package to tokenize the responses, to remove stopwords (i.e. prepositions) and to put together a corpus we can analyze. Begnaud’s followers encouraged him to visit Yabucoa, Humacao, Utuado and other towns in the southeast as well as in the mountains.
Any person following Puerto Rico’s post-Maria recovery knows that one of the top news stories is the Puerto Rico Electric Power Authority’s (PREPA) and the U.S. Army Corps of Engineers’ (USACE) troubles repairing the island’s electric grid. The lack of electricity serves as a good proxy to the many problems that have affected Puerto Ricans since September 20, 2017: lack of access to reliable healthcare, traffic lights not working, troubles with the distribution of potable water, access to banking services and so forth. And Begnaud’s followers identified electricity as the top issue of concern.
My corpus included over 30,000 words and his followers’ references to electricity represent 4% of this total – a really high number! The main words associated with the electric’s grid repair are included in the following pie chart.
How much do Begnaud’s followers know about Puerto Rico’s challenges post-Maria? As noted above, any person who has paid attention to the island’s struggles would have encouraged Begnaud to cover how the lack of electricity has affected Puerto Ricans’ lives. Thus, this piece of “collective information” by itself is not too interesting. But combining these data points with the towns Begnaud followers’ encouraged him to visit demonstrates that most of his followers do have a deep understanding of the island’s post-Maria struggles.
As detailed in the next table, PREPA groups the island’s 78 municipalities into seven administrative regions. The figure in the last column aggregates the municipalities mentioned by Begnaud’s followers. For example, PREPA’s San Juan Region aggregates all the times San Juan, Guaynabo, or Trujillo Alto were mentioned in the responses, which was 107 times.
|PREPA Region||Number of Meters||Percent of Meters||Number of Municipalities Per Region||Sum of Municipalities Mentioned By Begnaud’s Followers By Region|
For the last two months, the USACE has been publishing the percentage of meters connected to the electric grid in each of PREPA’s regions. The biggest challenge has been reconnecting the meters in the Caguas Region followed by the Arecibo Region, as demonstrated in the graph below.
If we only look at the municipalities that were mentioned 20 times or more, then we can appreciate how informed Begnaud’s followers are. Note that those who responded to his “call-out” mentioned 15 out of the 17 municipalities located in the PREPA’s Caguas Region.
|PREPA Region||Number of Municipalities Per Region||Number of Municipalities with 20 Mentions or More||
Percentage of Top Municipalities Mentioned
What can we learn from this analysis? First, it was a very good idea for Begnaud to crowdsource information from his Facebook and Twitter followers regarding his trip to Puerto Rico.
Second, Begnaud’s experiment in crowdsourcing was successful because his reporting has been closely followed by Puerto Ricans both in the island and in the mainland. As I noted in two earlier posts (here and here), Begnaud’s social media audience started to rapidly grow when he was in Puerto Rico reporting on the humanitarian catastrophe that ensued after Hurricane Maria devastated the island. Thus, many of the followers that responded to his “call-out” probably had some type of personal connection to the island and its people.
This second point is important because crowdsourcing works best when a majority of the people who responded to his “call-out” have some prior knowledge of the issues being considered. Thus, this particular analysis does not answer whether or not crowds are always right. But it does demonstrate that a majority of the followers who responded to Begnaud’s call for information knew quite a bit about Puerto Rico’s and its challenges following Hurricane Maria.