Social Network Analysis and Sexual Assault on College Campuses (Trigger Warning)

Max Bevilacqua - Not taking second module this year

Background

Incidents of sexual assault on American college campuses are horrifying. Rate of incidents aside, college administrations’ responses and the national conversation surrounding sexual assault lag woefully behind the needs of victims of sexual assault. Social network analysis, however, can and has offered meaningful insights to the conversation. In 2015, Dr. Emily Dworkin, Dr. Samatha Pittenger, and Dr. Nicole Allen used social network analysis to understand the disclosure decisions of survivors of sexual assault. Specifically, “results suggest[ed] that characteristics of survivors, their social networks, and members of these networks are associated with disclosure decisions[1].”

Research Questions

I am interested in the inclusion of men in social network analysis confronting sexual assault, in particular attitudes about sexual assault and its disclosure. With the collection of specific attribute data network measures (detailed below), I would endeavor to answer the question: Do attitudes towards sexual assault and its disclosure correspond to the makeup of a social network?  

Methodology

First, I would create a single mode network including men and women with ties based on frequency of time spent together. Based on experiences working with class data, I would like to see a scale of 1-5 with the following wording:

1: I have never seen this person

2: I spend time with this person rarely

3: I spend time with this person occasionally

4: I spend time with this person often

5: I spend time with this person very often

Likely, it will be most helpful to dichotomize the data at greater than or equal to 4, but the full set of responses allow flexibility in playing with the data (i.e. I might dichotomize at greater than or equal to 2 to see isolates, or greater and equal to 3 to have a comparative set of network measurements to see how meaningful the difference between “rarely” and “occasionally” are).

Next, I would form a two-mode network between participants and statement about sexual assault with responses to generating weighted ties. Here the re-application of a Likert scale [Strongly Disagree (1), Disagree (2), Neutral (3), Agree (4), Strongly Agree (5)] for answering questions that search for attitudes about sexual assault will provide nuance and flexibility for dichotomizing data. A binary measurement of “Agree” or “Disagree” for a question such as: “When I first hear of an incident of sexual assault I believe the claim” will not provide much room for analysis, and might muddy the data for those who feel a “Disagree” would be too callous and therefore choose “Agree”. Additionally, one may feel “Neutral” or indeed “Disagree” with the presence of a “Strongly Disagree” option.

Basic attribute data:

Sex (Male/Female), Gender (Male/Female/Other), Sexual orientation (Gay, Lesbian, Bi-sexual, or Heterosexual) Sexual activity (1-5 “Never had to sex” – “Very Frequent Sexual Activity”)

Additional attribute questions to get a sense of family networks could include:

Do you have siblings? If you only have one sibling, Male or Female? If you have multiple siblings are they all Male? Are they all female? Were you raised primarily by your Mother or Father? (Mother/Father/Both/Neither)

Affiliation Checklist:

Fraternity

Sorority

Single-Sex Sports Team

Mixed-Sex Sports Team

Single-Sex Music Group

Mixed-Sex Music Group

Other Single-Sex Group

Other Mixed-Sex Group

Sample statements regarding sexual assault (1-5 “Strong Disagree – Strongly Agree”:)

-       When I first hear of an incident of sexual assault, I believe the claim.

-       If my friend told me they had been sexually assaulted I would believe their claim.

-       If my friend told me they had been sexually assaulted I would try to help them.

-       I have many friends of the opposite sex.

-       I believe that sexual assault is widespread in society.

-       I believe that sexual assault is widespread on this campus.

-       Generally speaking, I believe men and women are treated equally in America.

-       I would/have disclose(d) an incident of sexual assault to a friend.

-       I would/have disclose(d) an incident of sexual assault to a family member.  

-       I would/have disclose(d) an incident of sexual assault to a member of the opposite sex.

-       I would/have disclose(d) an incident of sexual assault to campus administration.

-       I would/have disclose(d) an incident of sexual assault to the police.

-       I would/have disclose(d) an incident of sexual assault to a mental health professional.

(Follow up for each “I would/have disclose(d)” questions - “Their reaction was/would be supportive”

-       Consent is not meaningful if someone is drunk.

-       Sexual assault is never the victim’s fault.

-       Etc.

Analyzing subgroups from frequency-of-time-spent-together ties, against the backdrop of attribute data and in comparison to responses to the statements above will hopefully yield identifiable patterns. Cohesion measures for the entire network related to disclosure decisions/attitudes for Dworkin, Pittenger, and Allen. Additionally for this project, I would be looking at those with high Eigenvector and Betweenness values to determine if their attitudes are influential in their networks. Prior to seeing the network data, I would also be curious about the ways in which attitudes might relate to In/Out Degree measures with respect to the opposite sex compared to the same sex in addition to how this lines up with attribute data. Significant differences should be expected amongst different networks within a college campus not to mention on different college campuses (thus I would plan for this project to be conducted on several college campuses concurrently). However, there may be patterns that transcend these differences with potential applicability beyond college campuses.

Feasibility of Data Collection and Ethical Considerations

            With such a sensitive topic, anonymity is paramount. I would consider expanding attribute data such that individual nodes have a lower likelihood of being identified in the event that network information is published. Beyond protecting identities, recruitment for participants carries ethical implications. Dworkin, Pittenger, and Allen navigated the potential coercive nature of incentivizing participation if money was given to the head of an organization, a sorority president for instance, dependent on a certain percentage of the group's participation [2]. For both the purpose of obtaining contact information and avoiding this potential coercion, I would eschew group incentives for individual incentives and partner with an administration’s sexual assault prevention program/mental health and counseling center to consult on the most sensitive and effective framing of the survey for their campus. For the purposes of greater anonymity for participants, alleviating any pushback from college administrations, and enhancing the quality of insights via comparison, I would articulate that this project would be conducted on several college campuses.    

Sources

1.     Dworkin, E. R., Pittenger, S. L. and Allen, N. E. (2016), Disclosing Sexual Assault Within Social Networks: A Mixed-Method Investigation. Am J Community Psychol, 57: 216–228. doi:10.1002/ajcp.12033. Web 16 Oct. 2016. 

2.     "Case Study: IRB Deals with Social Network Analysis Issues." AHC Media Continuing Medical Education Publishing. Web. 16 Oct. 2016.

3.     Sexual Minority Assessment Research Team (SMART). “Best Practices for Asking Questions about Sexual Orientation on Surveys.” The Williams Institute, UCLA School of Law. http://williamsinstitute.law.ucla.edu/wp-content/uploads/SMART-FINAL-Nov-2009.pdf. Web. 17 Oct. 2016.

---

[1] Dworkin, E. R., Pittenger, S. L. and Allen, N. E. (2016), Disclosing Sexual Assault Within Social Networks: A Mixed-Method Investigation. Am J Community Psychol, 57: 216–228. doi:10.1002/ajcp.12033

[2] "Case Study: IRB Deals with Social Network Analysis Issues." AHC Media Continuing Medical Education Publishing RSS. N.p., n.d. Web. 17 Oct. 2016.