Measuring interpersonal influence

How much influence over political behaviors and beliefs do family members and neighbors exert over one another? Past studies of this question ignore unobserved heterogeneity as a confounding factor and are therefore subject to omitted variable bias. The statistical validity of several observational and experimental techniques are evaluated and through the use of three randomized field experiments unbiased estimates of inter-personal influence are obtained.; Chapter 3 randomly assigns subjects a position in a social network. The particular network studied is college freshman randomly assigned to dorm rooms. A survey was conducted at the beginning and end of the school year. Roommate and suitemate attitudes at the beginning of the year are used to predict student attitudes at the end of the year. Peer effects are detected for especially salient topics such as homosexuality and marijuana, but not found for ideological questions.; Chapter 4 controls communication within a defined social network. Volunteers were asked to list people they would be willing to encourage to vote. Each volunteer's list was then divided into a treatment group, whose members were received blandishments to vote, and a control group, which received no communication about the election. After the election, official voter turnout records were matched against the list of subjects. No mobilization effect was detected as a result of this peer-to-peer contact. This result is curious given the robust effectiveness stranger-to-stranger get out the vote efforts have exhibited. Possible explanations for the null finding are discussed.; Chapter 5 provides an exogenous shock to an existing social network and traces the ripple effects. The specific networks analyzed are households with two registered voters. A placebo-controlled experiment conducted where households are assigned to receive encouragement to either vote or recycle. The people answering the door in the voting condition are found to be 9 percentage points more likely to vote than the people answering the door in the recycling condition. Roughly 60% of the increased propensity to vote is passed onto the other voter in the household. The conclusion reached is that voting is a highly contagious behavior.