Social media provides us with mountains of data about the activities, proclivities, and habits of huge numbers of people using social platforms like Facebook and Twitter. The conventional wisdom holds that mining that data provides a treasure trove for economists, social scientists, marketers, and others who seek to quantify and predict future human behavior.


Not so fast, say researchers at Montreal's McGill University and Carnegie Mellon University in Pittsburgh.


In a paper published in the current issue of Science , computer scientists Derek Ruths and Jürgen Pfeffer of Carnegie Mellon outline a number of problems with academia's current usage of social media data to study human behavior.


These days, "thousands of research papers each year are now based on data gleaned from social media," so there are far-reaching implications to the poor methodology detailed by the researchers, Ruths told McGill University.


"Many of these papers are used to inform and justify decisions and investments among the public and in industry and government," he said.


Ruths and his colleagues listed several "challenges" for researchers using social media data sets, per McGill: