Sunbelt XXXI International Network for Social Network ... - INSNA

Voting Power CentralityCarnegie, Jeffrey A.Centrality Measures in Social NetworksPolitics, Social Influence, Centrality, Power, VotingWED.PM1Banzhaf (1965) argued that voting power was not proportional to voting weights, nor even linear. He did not consider that different coalitions might be moreor less likely, or that different people might be able to make use of structural power differently. A priori power indices assign an advisor to a dictator nopower. These measures offer no insight into the political reality. An advisor has power if he can influence the dictator. Few votes are taken without discussionof the alternatives. Voters try to convince others to vote their way. A social influence network determines the likelihood of each outcome. Network centralitymeasures are designed to measure actor importance in a social network, but they do not account for a particular voting rule. This problem is the opposite ofthe power indices: centrality measures consider only preferential information. I combine the two concepts by using an influence network as a way todetermine the likelihood of a coalition. Voting power centrality is a measure of power that accounts for the preferences and likely actions of other voters in aparticular situation. It is not a function of the voter's own preference, but only of the preferences of other voters. It uses an influence network to determinethe likelihood of each coalition, so it is comparable to centrality measures as well as power indices. I use several examples to illustrate the measure and showits usefulness, in particular the US Supreme Court and Council of the European Union.’We Just Clicked’ – Conversational Features Of Social Bonding In Romantic DatesMcFarland, Daniel A.; Jurafsky, Dan; Rawlings, CraigWords and Networks ‐ Natural Language Processing, ConflictSpeech, Tie decisions, Social bonding, Computational Linguistics, Dating relations, Dyadic AnalysisWED.PM2Most research on social bond formation has argued that status‐attraction and self‐selection (homophily) narrow the pool of suitable persons, and then withinthe remaining set, the quality of social interaction is such that some persons “click” and form bonds while others do not. But what does it mean for persons to“click” and how is that accomplished? This paper uses techniques from computational linguistics to study how different features of speech – e.g., lexical (wordusage), prosodic (pitch), and dialogue acts (backchannel) ‐ are used differently by men and women to establish a sense of social bonding. To study datingconversations we collected original data on multiple speed dating events using surveys on dating decisions and interpersonal perceptions and audio recordingsfor information on speech acts. We then analyze these data using dyadic data analysis (Kenny, Kashy and Cook 2006) and discern which speech styles are moreor less associated with a sense of "clicking" and tie formation (i.e., date selection).