Share Abstract
Rapid population size estimation of online men who have sex with men using Facebook marketing data in India
Abstract Content:
Background: Population size estimates (PSE) are used by key
population (KP) HIV programs to plan implementation and measure progress toward
targets. However, it''s challenging
to estimate sizes of KPs not linked to known networks or physical venues.
Programs need a way to quantify the reachable/accessible
population, including those not present in physical venues.
Methods: Between September and October 2017, the USAID- and PEPFAR-supported LINKAGES project in India used a new method to rapidly enumerate online men who have sex with men (MSM) leveraging publicly available marketing data on Facebook and short online surveys on Survey Monkey to refine estimates useful for programming.
Results: Using Facebook''s ad manager, there are 16.5 million adult male profiles in the program''s catchment areas of Mumbai, Thane, Pune, and Vijayawada and 268,000 profiles if restricted to profiles indicating “interested in” men. A survey advertised to those profiles acquired 29 respondents, 34% reporting MSM behavior. A similar survey more widely distributed to MSM through ads and social media postings acquired 318 MSM respondents and found 72% do not specify “interested in” men on their Facebook profile. These data were used to estimate 325,428 Facebook profiles used by MSM in the 4 geographies, representing 234,308 individual MSM (after de-duplicating for 28% of MSM reporting two or more profiles on the survey). This estimate represents 1.42% of the adult male population on Facebook in the same geographies. Sixty-one percent of these MSM reported exclusively meeting partners for dating or sex online and not at physical locations where existing PSE and outreach efforts focus.
Conclusions: Leveraging Facebook''s marketing data is a promising approach for assessing the size of online markets for KP HIV services - a new way to complement existing PSE approaches quickly and affordably. The approach can be adapted for other settings/populations by identifying the attributes on people''s social media profiles which predict whether they are KP members and then refine based on the accuracy of that attribute and its expression among the target audience. The method is still being developed and needs to be tested rigorously with better quality surveys to establish proof of concept.
[Figure 1: Estimating the population of online MSM in areas of India using Facebook marketing data]
Methods: Between September and October 2017, the USAID- and PEPFAR-supported LINKAGES project in India used a new method to rapidly enumerate online men who have sex with men (MSM) leveraging publicly available marketing data on Facebook and short online surveys on Survey Monkey to refine estimates useful for programming.
Results: Using Facebook''s ad manager, there are 16.5 million adult male profiles in the program''s catchment areas of Mumbai, Thane, Pune, and Vijayawada and 268,000 profiles if restricted to profiles indicating “interested in” men. A survey advertised to those profiles acquired 29 respondents, 34% reporting MSM behavior. A similar survey more widely distributed to MSM through ads and social media postings acquired 318 MSM respondents and found 72% do not specify “interested in” men on their Facebook profile. These data were used to estimate 325,428 Facebook profiles used by MSM in the 4 geographies, representing 234,308 individual MSM (after de-duplicating for 28% of MSM reporting two or more profiles on the survey). This estimate represents 1.42% of the adult male population on Facebook in the same geographies. Sixty-one percent of these MSM reported exclusively meeting partners for dating or sex online and not at physical locations where existing PSE and outreach efforts focus.
Conclusions: Leveraging Facebook''s marketing data is a promising approach for assessing the size of online markets for KP HIV services - a new way to complement existing PSE approaches quickly and affordably. The approach can be adapted for other settings/populations by identifying the attributes on people''s social media profiles which predict whether they are KP members and then refine based on the accuracy of that attribute and its expression among the target audience. The method is still being developed and needs to be tested rigorously with better quality surveys to establish proof of concept.
[Figure 1: Estimating the population of online MSM in areas of India using Facebook marketing data]