Zhong Zheng

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



First Advisor

Robert Hornik


Social influence is an important theoretical factor when studying how an individual's thoughts and behaviors are formed. However, empirical evidence demonstrating its effects is difficult to obtain due to the nature of social influence. One problem is that its exact source is often unclear. Also, because social influence is an ongoing process, its effects are cumulative and therefore difficult to measure directly.

An analytic model based on the contextual analysis approach is developed in this study and applied to survey data that was collected in Zaire using a cluster sampling method. As respondents from the same cluster (community) live geographically close or in the same neighborhood, it is assumed that if social influence is affecting individual respondents, their behavior will be associated with their community of residence. This model first measures the associations between community of residence and the outcome variables, which represent the total effects of community of residence. After community structural variations are controlled for, it is argued that any resulting residual effects are likely due to social influence.

Eleven outcome variables (which measure thoughts and behaviors related to diarrhea and its treatment) were found to be significantly associated with community of residence, and are analyzed in this study. Bivariate relationships between these outcome variables and various community compositional variables (such as community levels of education and wealth) and community infrastructural variables (such as access to health facilities) are examined. The community variables that are significantly related to outcome variables are controlled. Residual community of residence effects are found with respect to some outcome variables, and thus the social influence hypothesis is partially supported.

The weakness of this approach is that evidence supporting the social influence hypothesis is gathered by eliminating all alternative hypotheses, without directly measuring social influence as a variable. Since it is very difficult to specify and eliminate all possible alternative variables, the conclusions about the presence and effect of social influence are tentative.