Micro-blog Social Mood Measuring And The Relationship With Societal Risk Perception And Risk Decision Making

People are now in the data world with the development of Internet and information technology. There are massive information about human psychology and behavior in the Internet, which brings unprecedented challenges and opportunities for social science. The use of ecological data makes up for the shortcomings of introspection bias and sample representation and other issues of traditional social science.The relationship of emotion and risk perception and risk decision making is a hot topic of psychology. But traditional research mainly concentrated on individual and small group using questionnaires or experimental methods. We attempt to explore the relationship of social moods and societal risk perception and risk decision making using massive data sets and get verified the classic assumption at the macro level.In fact, social scientists have been concerned on the group emotion from Durkheim and Le Bon. But the research is relatively limited because of the difficulty of emotion measurement of wide range population (i.e. social mood). Traditional measurements of social moods are sampling methods or proxy indicators, which are time consuming or scope limitingIn recent years, the use of public information on the Internet to measure social moods directly is popular in the information science. These researches divide social moods into positive and negative categories simply. Even with a more detailed classification, the dividend of social moods is from intuition and subjective experience without support of emotional structure theory. In addition, the research from information science is mainly at technology and application level, and less in testing of important hypothesis, and absent of theoretical perspective of social science obviously.On the basis of systematic summary of the emotional structure of psychology and sentiment analysis and emotion lexicon of information science, we created the micro-blog basic emotion lexicon and emotional dimension lexicon. The basic emotion lexicon can measure the happiness, sadness, fear, anger and disgust five kind of basic social moods. The emotional dimension lexicon can measure valence, arousal, dominance and avoidance of each words.With the assistance of collaborators, we get the daily time series of five social moods of more than1.6million users lasting17months. We find the relationship of five social moods is compliance with the emotional dimension theory. The validate of social moods indicators is very good, for example, the happiness is lower in weekends than weekdays. The social moods fluctuations correspond to important festivals and events, such as the happiness increased significantly on the Spring Festival while the other four negative moods decreased.Then we use Granger Causality Analysis and Liner Regression Model to test the relationship of social moods and societal risk perception (Baidu news hot search terms indicators) and risk decision making (Shanghai Composite Index and the volume), we find social moods are early indicators. The four negative moods can predict the fluctuations of societal risk perception of the next day, but different kind of moods predict risk perception differently. Societal risk perception can in turn predict social moods too, for example, the government management risk can predict anger and the resources&environments risk can predict sadness. Moreover, the lowest arousal of negative mood-sadness can significantly increase the public decision making tendency, which can predict the volume of Stock market on the next1-3days positively which support the "emotional maintenance hypothesis".Our research is an attempt to use the Internet to study the psychological phenomenon. We explore the research from description and explanation to forecast and increase the practical application of psychology. Although the research is limited on sample and variables representation, it can be a useful alternative or supplement of traditional psychological research.