Social Media Research Might Not be Scientifically Sound

In terms of social media statistics, there are sometimes spectacular sounding metrics that upon closer scrutiny may be padded or unimportant. Equally, there seems to be weak methodology in scientific research of customers on social media which will result in the info being unsound and unscientific.

Sampling the inhabitants by way of social media is problematic due to its ubiquity and lack of requirements. “[M]ounting proof means that most of the forecasts and analyses being produced misrepresent the actual world,” in line with Juergen Pfeffer and Derek Ruths, authors of a current paper titled “Social media for large studies of behavior,” which appeared in Science.

One of many foremost issues recognized within the paper was sampling bias. Sampling bias happens when customers are polled or noticed with out bearing in mind the consumer demographics. Not all social networks are patronized equally: LinkedIn customers are typically extra educated, Twitter’s base is beginning to swing towards extra male customers, Fb skews considerably feminine and Pinterest customers are generally wealthier.

One other drawback is how the info is collected. In sure circumstances, the info is handed via proprietary filters put in place by the community earlier than it reaches researchers. The ubiquity of bots, PR accounts, pretend followers and different non-real accounts also can skew outcomes. This drawback could also be partially solved by accessing Twitter’s fire hose of data, for instance, however eliminating bias and false positives continues to be troublesome.

The difficulty with amassing knowledge with out making use of correct scientific rigor is that we find yourself with dangerous knowledge and false positives. Choice alone can serve up a desired end result that’s unlikely to be true, as we noticed when Princeton and Facebook dueled with knowledge.

One other concern with giant-scale knowledge evaluation like that is the query of ethics. …read more

The post Social Media Research Might Not be Scientifically Sound appeared first on DICKLEUNG DESIGN 2014.