Effect of mood and emotion on a person's behavior and his/her interactions with other people has been studied for a long time. Positivity and negativity of a person are two important attributes of emotion and mood. Social media is a very important platform from which we can glean the positivity and negativity attributes of a user based on his/her message postings and interactions with other users. In this paper, we study and analyze a Twitter dataset of more than 130,000 users to understand the nature of their positivity and negativity attributes. We measure behavioral attributes by sentiment analysis relating to social personal concern and psychological process. We observe that social media contains useful behavioral cues to classify users into positive and negative groups based on network density and degree of social activity either in information sharing or emotional interaction and social awareness. We believe that our findings will be useful in developing tools for predicting positive and negative users and help provide the best recommendation towards helping negative users through online social media.