,
9 tweets,
3 min read
Read on Twitter
A #tweetorial on my latest publication 👇 with @mlipsitch, @yhgrad, and others.
When modeling STI spread some of the hardest parameters to find REAL and ACCURATE data on are sexual behavior parameters. In particular we have little info on HOW people choose their partners. (1/)
When modeling STI spread some of the hardest parameters to find REAL and ACCURATE data on are sexual behavior parameters. In particular we have little info on HOW people choose their partners. (1/)
Most studies aiming to understand HOW people choose partners ask about preferences... but these answers can be biased - and so is the resulting info. So we sought to objectively observe partner selection in an online setting to inform partner selection in the real world. (2/)
Additionally, STI incidence rates among Black MSM in the US are higher than other racial/ethnic groups. Previous work hypothesized this is due to differences in sexual network composition of Black MSM. We aimed to observe preferences of Black MSM in online settings. (3/)
We worked with a diverse social networking app used by men who have sex with men (MSM) and gathered data on user behavior on app in 7 US cities. In particular, we were interested in which profiles people looked at (with respect to AGE and RACE) and who people chatted with. (4/)
We calculated the assortativity coefficient from these data that summarizes how similar searchers were to the profiles they examined. For example, if a 25 year old Black man only looks at profiles of 25 year old Asian men he would be assortative on age but not race. (5/)
Men LOOKED at profiles in a highly assortative manner by age but profiles of men between 21-29 of age were looked at often by all age groups. Men CHATTED with men in the 21-29 age categories regardless of searcher age. (6/)
Men LOOKED at profiles in a highly assortative manner by race with black men being highly assortative. We see that Asian and Black MSM CHATTED with other Asian and Black men in most cities. But each city was different with regard to racial assortativity. (7/)
So can we generalize online behavior to real life behavior? Maybe but maybe not. But this is a unique objective observational study that is not subject to the traditional biases of studies of sexual behavior. (8/)
The heat maps allow each city's patterns to be examined in detail in order to create targeted and localized interventions to reduce STI incidence. We hope that this data will be used to parameterize STI models in the future. (9/)
