Katherine Stovel

Associate Professor of Sociology
206R Savery Hall
Box 353340
University of Washington
Seattle, WA 98105
206|616|3820

stovel at dot u dot washington dot edu

Adolescent Sexuality

My interest in social networks is responsible for my on-going relationship with the Add Health Study, a large NIH-funded nationally representative study of adolescent health.  My commitment to studying adolescent health is long-standing (see Timko, Stovel, et. al, 1995; 1992a 1992b 1992c), yet I find the Add Health Study particularly exciting because its unusual design and sampling structure emphasizes the important role of social context in the development of adolescents (see Bearman, Moody, Stovel, and Thalji 2004 for an overview). 

One of the strengths of the Add Health study is that it contains rich social and sexual network data from large numbers of adolescents who attend the same schools.  In a widely discussed paper published in the AJS, we provided the first documentation of the complete structure of a sexual and romantic network of interacting adolescents (Bearman, Moody, and Stovel, 2004).  We found that though the sexual network in this setting is highly connected, the connectivity depends on long chains of ties, and is therefore extremely fragile.  Critical to our observed network is the pronounced absence of cycles.  In contrast to theoretical expectation, the observed structure of the sexual network does not appear to have a core.  The absence of cycles guarantees that we are unable to observe a densely interconnected core functioning as a disease reservoir.  Rather, we observe a spanning tree, characterized by the specific absence of cycles of length four.  Through simulation given a prohibition against the formation of cycles of length four, we replicate the structural features of the observed network, with respect to size reach, centralization, density, and number of cycles.  We conclude by arguing that the micro-level processes governing the formation of romantic relationship in this setting hinder the diffusion of bacterial STDs in this population.

I continue to work with the Add Health data to address specific questions about the social context of adolescents' disease-related sexual behavior.  At the moment this work is primarily undertaken in conjunction with graduate students in the sociology department and in my role as a member of the training faculty at the University of Washington's Center for Aids Research.  With colleagues at CFAR, I have studied the extent to which condom use is ‘habit-forming' among adolescents.   In a paper published in Sexually Transmitted Diseases, we reported that using a condom at sexual debut doubles the odds of subsequent condom use, even after controlling for a wide variety of demographic, behavioral, and relational characteristics (Shafii, Stovel, Davis and Holmes 2004).  We interpret this in light of an ‘imprinting hypothesis,' and argue that experiences at first sex are crucial for the development of sexual habits.  A follow-up paper followed the same adolescents further into young adulthood, and found that after almost seven years, adolescents who used a condom at sexual debut were 36% more likely to use a condom at their most recent sex, had the same number of lifetime sex partners, and were half as likely to currently have a sexually transmitted infection.

Obviously this line of research has important implications for abstinence-only sex education, and has been recognized with awards at adolescent health and pediatric medicine meetings. 


This figure reflects the structure of adolescent romantic and sexual relationships ocurring over an 18 month period among students attending a mid-sized high school in the US.  While many students have only one partner, most students are connected into longer chains of contact.  In fact, the majority of students in the school are connected into a single massive component. However, this component is not characterized by dense clusters of highly active students; rather, it is a spanning tree, composed on long sparse chains of relationships.  While chain-like structures have the potential to expose large numbers of persons to disease, they are extremely fragile.

Bearman, Moody, and Stovel 2004