Wednesday, February 10, 2016
Social media is like looking into the window of a house. You can see what is happening, but it can be difficult to interpret. If a person inside the house is waving their arms in the air, are they are excited or angry? Similarly, with social media, the da

ta is there. , but the interpretation can be more complex. The December attack in San Bernardino prompted people to ask whether or not the government should be more active in using social media data to prevent such things from happening again. Certainly, social data can be highly valuable, but it has its limitations, and agencies must be aware of how best to use it. Determining sentiment – It is very useful to understand how citizens feel about certain issues such as a new a policy. Many programs will try to determine sentiment by categorizing certain words as positive or negative. However, it is important to remember that sentiment isn’t black or white. Something can be said with sarcasm or irony and mean the opposite of what it seems at face value. True sentiment analysis must take context into consideration. As humans, we understand things that simple algorithms do not. Short of manually assessing sentiment, agencies should look for programs that incorporate natural language processing, network and demographic mapping, and additional content related to conversations. Predicting outcomes – What can social media tell us about the future? The USGS has found that real-time tweet data was able to predict earthquakes, often faster than even seismic tools could. Social media has also helped to track epidemics and provide early warnings in cases like Ebola. All of this is done through a process of identifying patterns, assessing trends, and understanding networks. However, it’s worth noting that these predictions are often of near-future events. Predicting how many earthquakes will happen or the strain of next year’s flu is much more difficult. Predicting future outcomes requires strong patterns – the more random or rare the outcome, the less likely we are able to predict it. In addition, agencies shouldn’t expect social media data alone to provide all the answers, but should look to combine it with other data sources as well. Improving delivery – Where social data is not exactly a crystal ball with every answer, it is absolutely still an opportunity to improve the citizen experience. As pointed out in DigitalGov’s Federal Social Media Analytics Toolkit, benefits include: More effective distribution of critical information, more responsive public programs, better-informed strategies, and increased use of innovative tools and services. Through analyses of social interactions, organizations can determine what’s working and what’s not, what citizens need, and how best to respond. For example, an agency may find that citizens are repeatedly asking the same question, mention difficulty with a particular service, use a hashtag related to a common issue, or seem to be unaware of a new campaign. Each piece of data provides insight to an agency on how to better serve its constituents. Organizations should look to combine a variety of metrics to get the best picture. Social data may not have all the answers; alone, it cannot predict every possible outcome, but it is an important piece of the puzzle. Understanding there are limitations to the data as well as a need to weigh a citizen’s right to privacy, there are still many ways social media metrics can help agencies better achieve their missions. Platforms’ built-in analytics, such as Twitter Analytics, Facebook Insights, or Google Analytics, are typically free and a good place to start. For more details on social media analytics, please see the additional reading below. Have you or your organization successfully used social data to improve services? Share your experiences below. Additional Reading: DigitalGov Social Media Analytics Toolkit A Manager's Guide to Assessing the Impact of Social Media Interactions