A new research paper has been published that looks at how representative social media really is. Social media has completely changed people’s lives. Even now the buzz around social media is still growing. In the UK, more than 50% of the population uses it on a regular basis. There are 50 different social media platforms used there with over a million users each.
It has not only changed everyday life for regular people. It has also completely changed many occupations such as journalism, and has completely changed the nature of politics and elections. One area that is less talked about though, is research. Research done at universities, marketing agencies, and throughout government has completely changed due to social media and the web. The articles author points out that during the time taken to read the article. More data will have been produced via social media than could be collected in a decade long census. People use social media to discuss everything from personal issues, to politics, and consumer products.
The view held by many researchers, is that social media has the potential to make their craft stronger and more useful than ever. New analytics companies and technologies have arisen to take advantage of and measure all the new metrics brought about by social media. The social sciences in particular can make use of far greater data, and many new types of it. Despite the new opportunities social media presents researchers. There are also dangers, particularly as many important principles of social science research have been lost in social media research. They have instead been replaced by the need for larger numbers and more data, gathered as quickly as possible. It is because of this that although social media research methods have far greater numbers and potential. They are still not as highly regarded as the traditional methods of social science research.
The authors of the paper have identified several areas where social media research falters, and address ways to improve it. Their biggest area of concern is representivity. Making sure your research correctly reflects a defined group (particular voter, geographical area etc). Although it isn’t always possible, it is vital for researchers to understand how reflective a dataset is of their target population.
They outlined six different areas where social media based research needed improving. The first was in the collection method, as sometimes you miss the data you want and instead get data that is of no use. The second is prolific accounts; much of the data online comes from only a small percentage of high posting users. This makes it hard to know if what you are finding is really representative of an entire group, or just a minority of very vocal users. Other issues include the use of corporate and institutional accounts that reflect the needs of the company, not the individual. The use of bots also creates issues, as does the difficulty in knowing the exact geographical location of the user. The final issue was how social media tends to over represent the views of certain groups and ages, i.e. younger people who are more likely to be using it. The authors then go on to highlight ways to improve these, most of which involve new technologies.
This was an interesting article that highlights some of the flaws in social media data collection. It is sometimes easy to get carried away with new technologies or ways of doing things, and completely forgot about traditional methods. It will be interesting to see how these methods improve over time.
If researchers are able to remove these issues with social media data collection and fine tune the process. Do you think social media based data collection will completely replace the traditional methods of data collection such as the census? Will there still be any need for the traditional methods of data collection?