The Most Followed Accounts on Twitter
By Joshua BoydJun 13
Published June 3rd 2019
(Quick warning: This blog post looks at a range of causes of death, including cancer, heart disease, overdose, and suicide. Reader discretion is advised).
Death is a grizzly subject, but it’s one of the few things experienced by all human beings, regardless of our backgrounds.
With this collective fate, it’s almost surprising how many misconceptions there are around the subject.
I recently came across an excellent piece by students at UCSD, based on a previous study by Paul Slovic and Barbara Combs, that looked at disparities between causes of death and how the media covers those causes. The students at UCSD then looked at Google Trends data alongside the media coverage, to gauge whether people’s worries seem to be led by actual threats or by the media instead.
One of the things I found interesting is that in spite of some correlation between the data sets, disparity still existed between them. This suggests that peoples’ world views aren’t shaped entirely by what they’re told or what’s around them.
This got me interested in exploring how people talk about death on social media – do these disparities exist in human conversation? Or, when we post publicly online, are we more outwardly attuned to the facts and data?
Four data sources are included in this report, three of which (CDC mortality rates, Google Trends and news coverage) were retrieved from Owen Shen’s Github (Owen is one of the UCSD students).
Meanwhile, social media data was retrieved by me via Brandwatch Analytics. A query (a search of online conversation) was created following the same keyword breakdown as in the aforementioned data sets to make the data comparable. I used a six year date range (2010 – 2016) again, to make the data comparable. A 10% sample of mentions was used due to large volumes.
The below charts include the ten largest causes of mortality (according to the The CDC’s WONDER database for public health data) alongside homicide, terrorism, and overdose, chosen due to the high media attention given to each.
|Cause of death||Average % of deaths|
|Lower respiratory disease||7%|
|Pneumonia and influenza||2%|
As we can see, heart disease and cancer are the biggest killers, accounting for more than half of all deaths on average between 2010-2016. Conversely, homicide, terrorism, and overdose combined account for about 3% of fatalities on average.
Now take a look at how these causes of death are discussed on social media.
|Cause of death||Average % of social media conversation relating to these causes of death|
|Pneumonia and influenza||9%|
The disparity between actual volumes of fatal causes and volumes of causes written about in the media (as found in previous studies) was prevalent on social media, too.
Despite being the largest cause of mortality, heart disease was one of the least talked about, while terrorism, accounting for 0.00004% of all fatalities appeared in 12% of the social media posts we studied.
Let’s see how the averages look across all the datasets we have access to – our own (social media data), CDC data, and the Google Trends and news coverage data.
|Cause of death||% of actual deaths||% of social conversation around the cause||% of Google searches around the cause||% of news coverage around the cause|
|Pneumonia and influenza||2%||3%||2%||3%|
The above chart shows the emphasis that media outlets give to causes that are underrepresented in real life, such as homicide and terrorism. Meanwhile, the correlation between Google Trends data and social media is interesting, suggesting people’s private and public concerns have a degree of concordance.
In fact, when we look at the factor of difference between the social numbers and the other numbers here, Google Trends had the lowest, while the CDC data had the highest. This means that Google and social had the highest correlation, while social and actual causes of death had the weakest connection. Terrorism was the biggest driver here, since the difference is so great.
It’s surprising to me that there were such disparities between news coverage and social media when it’s estimated that nearly 64.5% of internet users receive their news via social media. You’d expect one to be reflective of the other. This may be an issue with the date range, as news consumption via social media has been rising steadily, only overtaking news websites in 2018.
The purpose of this research was to discover if the disparities that exist between the real causes of death and the perceived importance of each (as demonstrated by the articles and Google searches previous studies have looked at), also exist on social media.
The data above shows this to be true, suggesting there is a disconnect between what kills people and what they’re worried about.
This isn’t particularly groundbreaking – the studies done prior suggested that this would be the case. But I do think there are some broader points about social media data that could be inferred from this study:
Full credit to Owen Shen, Hasan Al-Jamaly, Maximillian Siemers, and Nicole Stone for their study which this piece is based on as well as the open source data they provided.