Published on 21.03.19 in Vol 5, No 1 (2019): Jan-Mar
Preprints (earlier versions) of this paper are available at http://preprints.jmir.org/preprint/11285, first published Jun 13, 2018.
Kuantim mi tu (“Count me too”): Using Multiple Methods to Estimate the Number of Female Sex Workers, Men Who Have Sex With Men, and Transgender Women in Papua New Guinea in 2016 and 2017
Background: Female sex workers (FSW), men who have sex with men (MSM), and transgender women (TGW) are at high risk of acquiring HIV in many settings, such as Papua New Guinea (PNG). An understanding of the approximate size of these populations can inform resource allocation for HIV services for FSW, MSM, and TGW.
Objective: An objective of this multi-site survey was to conduct updated population size estimations (PSE) of FSW and MSM/TGW.
Methods: Respondent-driven sampling (RDS) biobehavioral surveys of FSW and MSM/TGW were conducted in 3 major cities—(1) Port Moresby, (2) Lae, and (3) Mount Hagen—between June 2016 and December 2017. Eligibility criteria for FSW included: (1) ≥12 years of age, (2) born female, (3) could speak English or Tok Pisin (PNG Pidgin), and (4) had sold or exchanged sex with a man in the past six months. Eligibility for MSM/TGW included: (1) ≥12 years of age, (2) born male, (3) could speak English, or Tok Pisin, and (4) had engaged in oral or anal sex with another person born male in the past six months. PSE methods included unique object multiplier, service multiplier, and successive sampling-population size estimation (SS-PSE) using imputed visibility. Weighted data analyses were conducted using RDS-Analyst and Microsoft Excel.
Results: Sample sizes for FSW and MSM/TGW in Port Moresby, Lae, and Mount Hagen included: (1) 673 and 400, (2) 709 and 352, and (3) 709 and 111 respectively. Keychains were used for the unique object multiplier method and were distributed 1 week before the start of each RDS survey. HIV service testing data were only available in Port Moresby and Mount Hagen and SS-PSE estimates were calculated for all cities. Due to limited service provider data and uncertain prior size estimation knowledge, unique object multiplier weighted estimations were chosen for estimates. In Port Moresby, we estimate that there are 16,053 (95% CI 8232-23,874) FSW and 7487 (95% CI 3975-11,000) MSM/TGW, approximately 9.5% and 3.8% of the female and male populations respectively. In Lae, we estimate that there are 6105 (95% CI 4459-7752) FSW and 4669 (95% CI 3068-6271) MSM/TGW, approximately 14.4% and 10.1% of the female and male populations respectively. In Mount Hagen, we estimate that there are 2646 (95% CI 1655-3638) FSW and 1095 (95% CI 913-1151) MSM/TGW using service multiplier and successive sampling, respectively. This is approximately 17.1% and 6.3% of the female and male populations respectively.
Conclusions: As the HIV epidemic in PNG rapidly evolves among key populations, PSE should be repeated to produce current estimates for timely comparison and future trend analysis.
JMIR Public Health Surveill 2019;5(1):e11285
- Papua New Guinea;
- population size estimation;
- female sex workers;
- men who have sex with men;
- transgender women;
- key populations;
- respondent-driven sampling;
- unique object multiplier;
- service multiplier;
- successive sampling
HIV disproportionally affects marginalized and stigmatized populations . Female sex workers (FSW), men who have sex with men (MSM), and transgender women (TGW) are 3 key populations (KP) at greater risk for HIV [ - ]. Globally, FSW, MSM, and TGW are estimated to be 13.5 [ ], 19.3 [ ], and 48.8 [ ] times more likely to be infected with HIV than the general population, respectively. This risk is accentuated by critical barriers to HIV-related prevention and treatment services, such as violence, criminalization, stigma, and discrimination [ - ].
Papua New Guinea (PNG) has the largest HIV epidemic in the Pacific region, with a national prevalence estimated at 0.9% . Recent surveys of FSW and MSM/TGW estimated HIV prevalence among FSW and MSM/TGW at 14.9% and 8.5%, respectively, in the capital of Port Moresby. In Lae, the country’s economic hub, it was estimated at 12.9% and 7.1%, respectively [ ].
Population size estimates can inform resource targeting and program monitoring . The only population size estimation of FSW and MSM in Port Moresby using empirical methods was conducted in 2006 and utilized the service multiplier in conjunction with a respondent-driven sampling (RDS) survey [ ]. Social changes such as population growth and increased mobility have occurred since then [ , ]. Updated size estimates are needed in Port Moresby and other cities, such as Lae and Mount Hagen, require first-time estimates. We used unique object multiplier, service multiplier, and successive sampling-population size estimation methods, which are direct and empirical, to estimate the number of FSW and MSM/TGW in these 3 cities.
Community consultation was undertaken within each city. Population members recognized that MSM and TGW are distinct populations but TGW are too few to achieve adequate sample size as an independent RDS sample. The 2 populations thus agreed to be combined into a single sample.
Seeds were purposively selected to be diverse with respect to age, sexual and gender identity, residence, region of origin, marital status, receipt of a unique object, and affiliation with a non-governmental or community-based organization.
Separate RDS biobehavioral surveys (BBS) of FSW and MSM/TGW were conducted in the 3 cities Port Moresby, Lae, and Mount Hagen between June 2016 and December 2017. Survey eligibility criteria for FSW included: (1) born female, (2) >12 years of age, (3) able to speak English or Tok Pisin, and (4) had sold or exchanged sex with a man in the past 6 months. Survey eligibility for MSM/TGW included: (1) born male, (2) >12 years of age, (3) able to speak English or Tok Pisin, and (4) had engaged in oral or anal sex in the past 6 months with another person born male.
Sample Size and Precision
We aimed to enroll 700 people into each BBS in each city. This assumed a design effect of 2 and was sufficiently powered to estimate an assumed HIV prevalence of 20% with an absolute precision of 10% . The sample size was calculated so that the results of the present survey can be compared to anticipated follow-up studies ( , [ - ]).
Unique Object Multiplier Method
Given the general unavailability of HIV testing or organization membership data in the survey cities, we primarily used the unique object multiplier (UOM) in conjunction with RDS surveys to estimate population size [, ]. Volunteers, consisting of local KP peers and survey team members, distributed approximately 1000 keychains per population 1 week before survey rollout in each city. They were instructed to verify that each keychain recipient had not already received an object, that this person received only 1 object, and told the recipient to keep the object for the near future and not to give it to anyone else. Distributors were provided with 30 kina (US $12) for distributing keychains, irrespective of the actual number distributed—this removed any incentive to report distributing more keychains than actuality as may occur if the compensation was given per keychain distributed. They also received 5 kina (US $2) for transportation. To strengthen recall of receiving an object among BBS participants, keychain distributors wore distinctive hats that featured the survey logo “Kauntim mi tu.” During survey eligibility screening, the coupon manager asked participants if they received a unique object from a distributor. Those indicating that they had the unique object were asked to show it. If unable to do so, they were asked to select the keychain from among other keychains displayed by the coupon manager.
Service Multiplier Method
The service multiplier method was used only in Port Moresby and Mount Hagen [, , ]. Survey participants were asked during the face-to-face interview whether they had tested for HIV at specific health facilities in 2015 (Port Moresby) or 2016 (Mount Hagen). Four HIV testing providers in Port Moresby and 1 in Mount Hagen were capable of providing key population-specific testing information. None were able to do so in Lae. Survey participants in each city were also asked whether they belonged to a KP organization. Responses to these questions were paired with data from HIV testing organizations and the KP community organization to develop multiple service multiplier estimates. In Lae and Mount Hagen, no KP organization could provide unique membership data.
Successive Sampling Method: Population Size Estimation
Using Respondent-Driven Sampling Analyst (RDS-A) version 0.62 [, ] the successive sampling-population size estimation (SS-PSE) method was used to produce size estimates using routinely collected data in RDS surveys including: (1) self-reported network size, (2) number of participant’s recruits enrolled in the survey, and (3) the date order of survey enrollment. We imputed visibility using these 3 routinely collected data items in order to smooth the network size distribution, reduce the effect of outliers, and minimize heaping of values (eg, around 5,10,15) [ ]. Prior estimates were calculated using distribution of age and sex in each city, city general population sizes, proportions of FSW, MSM, and TGW in other countries, and previous knowledge of sex work in PNG.
Data were analyzed using RDS-A with the Gile SS-PSE and Microsoft Excel 2016. Standard formulas for the UOM method, service multiplier method, and SS-PSE were used . Both weighted and unweighted estimates were produced for the multiplier methods to compare results. The 95% CI were calculated around point estimates using RDS-A.
Protection of Minors
Participants <18 years were provided referrals as needed to organizations that offer counseling, protection, and other relevant services for victims of sexual exploitation and abuse.
This survey was approved by the PNG National Department of Health’s Medical Research Advisory Committee, the Research Advisory Committee of the National AIDS Council Secretariat, the PNG Institute of Medical Research’s Institutional Review Board, and the Human Research Ethics Committee at University of New South Wales Sydney, Australia. The activity was reviewed according to the Centers for Disease Control and Prevention (CDC) human research protection procedures and was determined to be research but the CDC was not engaged in research collection. Two peer-led organizations for KP (ie, Friends Frangipani and Kapul Champions) provided letters of endorsement.
Sampling and Recruitment
Among the FSW, 673 were enrolled in Port Moresby, 709 in Lae, and 709 in Mount Hagen while for MSM/TGW there were 400 enrolled in Port Moresby, 352 in Lae, and 111 in Mount Hagen (). Similarly, more keychains were distributed to FSW in Port Moresby, Lae, and Mount Hagen (N=867, N=790, N=546, respectively) compared to MSM/TGW (N=598, N=777, N=152, respectively), suggesting that FSW are easier to reach, and they are likely to be better networked. The number of keychains to be distributed was determined with a calculator found in international guidelines [ ]; we aimed to distribute more keychains than our sample size. The RDS data collection took approximately the same amount of time for each city and population (14-20 weeks).
and present population size estimates for KP in each city. For FSW ( ), population size estimates in Port Moresby ranged from 3537-35,048 (2.1%-20.7% of the adult female population), in Lae from 4482-6105 (10.5%-14.4% of the adult female population), and in Mount Hagen from 2386-6315 (15.5%-40.9% of the adult female population). For MSM/TGW ( ), population size estimates in Port Moresby ranged from 501-18,644 (0.3%-9.6% of the adult male population), in Lae from 3455-4669 (7.5%-10.1% of the adult male population), and in Mount Hagen from 1095-3625 (6.3%-20.8% of the adult male population).
|Target population in each city||Enrolled participants, (N)||Keychains distributed, (N)||Keychains not distributed, (N)||Participants receiving keychains, (N)||Survey period||Data collection period, weeks|
|Port Moresby||673||867||133||51||Jun 2016-Nov 2016||20|
|Lae||709||790||291||110||Jan 2017-May 2017||16|
|Mount Hagen||709||546||454||71||Sep 2017-Dec 2017||14|
|Port Moresby||400||598||402||35||Jun 2016-Nov 2016||20|
|Lae||352||777||223||75||Jan 2017-May 2017||17|
|Mount Hagen||111||152||848||5||Sep 2017-Dec 2017||14|
Port Moresby provided the most analytic possibilities, although there was variability in the number of unique individuals tested across these 4 organizations and only 1 KP organization could provide data on the number of members. As described above, in Lae, only the UOM and SS-PSE methods were used. In Mount Hagen, UOM, service multiplier (only for FSW), and SS-PSE estimates were derived for both KP. As many population estimates were developed in Port Moresby, meetings were held to discuss the estimates and the extent to which assumptions were met in order to identify a final estimate. In Lae and Mount Hagen, where fewer population estimates were developed, such meetings played a smaller role. Each of the 3 methods is direct and empirical, making them superior to other indirect methods such as census and enumeration . These latter simplistic methods depend on counting everyone or a large number of people, which can be costly for census and often unfeasible for both with hidden groups like KP. Finally, weighted estimates were selected over unweighted estimates in order to increase the likelihood of independence between the convenient nature of keychain distribution (the capture) and RDS recruitment (the recapture) by turning the RDS sample data into population-based data. Without weighting of estimates, key assumptions of multiplier methods are violated.
Final weighted estimates chosen for each city were (1) 16,053 (95% CI 8232-23,874) FSW and 7487 (95% CI 3975-11,000) MSM/TGW in Port Moresby approximately 9.5% and 3.8% of the female and male populations, respectively, (2) 6105 (95% CI 4459-7752) FSW and 4669 (95% CI 3068-6271) MSM/TGW in Lae, approximately 14.4% and 10.1% of the female and male populations, respectively, and (3) 2646 (95% CI 1655-3638) FSW and 1095 (95% CI 913-1151) MSM/TGW in Mount Hagen approximately 17.1% and 6.3% of the female and male populations respectively (and ).
|Population size estimation method||Multiplier number, N||Survey, %||Estimate, 95% CI||Female urban population size based on total, %a|
|Port Moresby, N=169,291a|
|Mount Hagen, N=15,430a|
aValues provided by the 2011 census of Papua New Guinea.
|Population size estimation method||Multiplier number, N||Survey, %||Estimate, (95% CI)||Male urban population size based on total, %a|
|Port Moresby, N=194,834a|
|Mount Hagen, N=17,400a|
aValues provided by the 2011 census of Papua New Guinea.
We employed several methods to develop population size estimates of FSW and MSM/TGW in PNG. We believe that the most robust population size estimates produced were with the UOM for both KP in Port Moresby and Lae, and in Mount Hagen, the service multiplier for FSW and SS-PSE for MSM/TGW. We present both unweighted and weighted estimates, the latter which adjusts for RDS recruitment. Individuals with smaller network sizes are up-weighted, whereby their responses are more valued as seen by the larger weighted estimates than unweighted estimates in our study.
Final population estimates were chosen through a series of rigorous meetings between experts in PNG, Australia, and the United States. Given the number and range of estimates produced, investigators reviewed each size estimation method, the extent to which their assumptions were met, and the resulting estimates. The investigators narrowed down the estimates to those that were most robust in each city. These estimates were then presented to key stakeholders including the National Department of Health, key population organizations, donors, and United Nation agencies who were tasked with agreeing on a single estimate. Details and results of these discussions follow below.
Several precautionary steps were taken to maximize the utility and accuracy of the unique object multiplier. We minimized the risk of participants falsely indicating that they had received a keychain when in fact they had not by asking participants to (1) show their keychain, or (2) identify the correct keychain from a group consisting of this keychain and other incorrect keychains. Distributing the unique object approximately one week before the start of the RDS survey helped to (1) limit in and out-migration of participants, (2) minimize the possibility of target population members giving their object away, and (3) reduce the possibility of someone receiving objects from multiple distributors.
It is worth noting that fewer keychains were distributed to MSM/TGW, which may be explained by their lower propensity toward gathering in public places due to stigma and discrimination and smaller network sizes, as compared to FSW. While keychain distribution is influenced partially by how well a distributor is trained, smaller network sizes pose a challenge because fewer people may be present in KP-friendly hotspots and there is greater reliance on tapping into an individual’s limited social connections to distribute these keychains. We tried to address stigma/discrimination barriers in Mount Hagen prior to study initiation by building trust and rapport. For future work, we would consider using more volunteers who each distribute fewer objects, since individuals may know fewer people due to stigma.
We encountered no problems distributing keychains to either population and leftover keychains were returned to survey sites, suggesting that keychain distribution occurred with fidelity. In Mount Hagen, UOM estimates indicated that FSW account for approximately 40% of the adult female population and MSM/TGW 20% of the adult male population. (and ). These estimates were deemed unreasonably high, leading to their exclusion in favor of the service multiplier and SS-PSE estimates, respectively, in this city. All MSM/TGW estimates in Mount Hagen were hampered by the small RDS sample recruited.
While the data provided by HIV testing organizations in Port Moresby were meant to represent unique individuals, some service multiplier results there were excluded due to very wide CI values, while other population size estimates were too low to be plausible given our understanding of sex work in this city, and MSM/TGW globally . Our results are higher than results from a previous survey in Port Moresby [ ], which reported FSW and MSM service multiplier estimates to be 4212 (95% CI 3586-4839) and 2126 (1787-2468), respectively. We believe that our results are acceptable because those population estimates are almost a decade old, Port Moresby has undergone tremendous growth since then, and we included transgender women in our estimates.
Given the age of previous PNG estimates, we lacked a recent and more accurate prior value for use in producing an estimate with the SS-PSE method, suggesting a reason why the results from this method were relatively low. Likewise, the SS-PSE method estimates the number of people that fit our survey’s time-frame eligibility criteria (ie, behaviors in the past 6 months) . Therefore, individuals who sold sex, engaged in same-sex sexual behaviors, or TGW who had sex with men more than six months ago are not included in this estimate. As these people are still KP and face multiple vulnerabilities, we felt it important to select a larger and more plausible estimate, hence the selection of the UOM for Port Moresby and Lae. In both cities, the UOM produced population size estimates that were slightly larger than SS-PSE.
Our findings are limited in several ways, beginning with the self-reported nature of the interview data. Participants may have chosen to underreport or overreport HIV testing or organization membership due to stigma or fear, resulting in overestimation and underestimation, respectively. This social desirability bias could be mitigated in the future through the use of computer-assisted self-interviews [- ].
The unique object multiplier relies on accurate reporting of keychains given out by distributors and received by KP. We cannot know whether all keychains that were said to be distributed were actually distributed correctly, that individuals received only one keychain each, or how many keychain recipients were ineligible for the survey. Additionally, though we distributed keychains immediately before survey initiation, data collection took up to four months to complete, so it is possible that KP members migrated out of the catchment area, causing an underestimation.
Few HIV testing providers in Port Moresby and Mount Hagen were able to provide data and those that were able tested only a small number of KP. HIV testing data were nonexistent in Lae, meaning the service multiplier method could not be used. There is also the concern of recall bias for participants in Port Moresby who were asked about services accessed in 2015. Furthermore, SS-PSE relies on prior estimates, which themselves may not be accurate in Port Moresby and we did not have in either Lae or Mount Hagen.
Our findings are important for describing the number of FSW and MSM/TGW in the survey cities. The direct and empirical size estimation methods used in this survey are superior to other indirect methods, such as census and enumeration. The population size estimates in this survey will inform efforts to improve resource targeting and monitoring of both existing and new services for these key populations. HIV testing providers in PNG should be encouraged to disaggregate their data by key population to facilitate the use of the service multiplier. This will also increase the utility of routinely collected data.
Much of this project’s success, which was led by the national government of PNG with technical assistance from the United States and Australia, should be credited to the Papua New Guinean “Kaumtim mi tu” survey team, which included KP members in strategic staff positions, and who nurtured trust among KP in each city. The staff was crucial in organizing keychain distribution and obtaining service provider data. In addition, the survey team was almost unchanged across all cities. This resulted in a consistency of operations and data collection, improved efficiency, and an overall increase in technical capacity to implement RDS surveys and population size estimation activities. We also used tablets to collect data electronically, which simplified data management and decreased chances of error. Furthermore, we found it valuable to start our project with the “easiest” site first, Port Moresby, because the capital city had the most visible KP, facilitating KP engagement and survey implementation. News of the survey’s benefits to individuals and KP as a whole, as well as the friendliness and professionalism of the survey team, traveled to Lae and Mount Hagen motivating participation there. Given the lack of recent population size estimates, we found it indispensable to use several methods to estimate population size because each method has its strengths and weaknesses.
Multimedia Appendix 1
Study sample size and precision calculations.PDF File (Adobe PDF File), 37KB
- WHO, CDC. UNAIDS, FHI. Biobehavioural survey guidelines for populations at risk for HIV: World Health Organization; 2017:360-392.
- Baral S, Beyrer C, Muessig K, Poteat T, Wirtz AL, Decker MR, et al. Burden of HIV among female sex workers in low-income and middle-income countries: a systematic review and meta-analysis. Lancet Infect Dis 2012 Jul;12(7):538-549. [CrossRef] [Medline]
- Baral S, Sifakis F, Cleghorn F, Beyrer C. Elevated risk for HIV infection among men who have sex with men in low- and middle-income countries 2000-2006: a systematic review. PLoS Med 2007 Dec;4(12):e339 [FREE Full text] [CrossRef] [Medline]
- Baral SD, Poteat T, Strömdahl S, Wirtz AL, Guadamuz TE, Beyrer C. Worldwide burden of HIV in transgender women: a systematic review and meta-analysis. Lancet Infect Dis 2013 Mar;13(3):214-222. [CrossRef] [Medline]
- Baral S, Holland CE, Shannon K, Logie C, Semugoma P, Sithole B, et al. Enhancing benefits or increasing harms: community responses for HIV among men who have sex with men, transgender women, female sex workers, and people who inject drugs. J Acquir Immune Defic Syndr 2014 Aug 15;66 Suppl 3:S319-S328. [CrossRef] [Medline]
- Beyrer C. Global prevention of HIV infection for neglected populations: men who have sex with men. Clin Infect Dis 2010 May 15;50 Suppl 3:S108-S113. [CrossRef] [Medline]
- Baral S, Logie CH, Grosso A, Wirtz AL, Beyrer C. Modified social ecological model: a tool to guide the assessment of the risks and risk contexts of HIV epidemics. BMC Public Health 2013;13:482 [FREE Full text] [CrossRef] [Medline]
- UNAIDS. Geneva. 2018. Papua New Guinea URL: http://www.unaids.org/en/regionscountries/countries/papuanewguinea [accessed 2018-12-21] [WebCite Cache]
- Kelly-Hanku A, Badman S, Willie B, Narakobi R, Amos-Kuma A, Pekon S, et al. Kelly-Hanku A, Badman S, Willie B, Narakobi R, Amos-Kuma A, Pekon S, Kupul M, Aeno H, Neo RB, Ase S, Nembari J, Hou P, Dala N, Weikum D, Vallely A, Hakim A. 90-90-90 and the HIV continuum of care: how well is Papua New Guinea doing amongst key populations? 9th IAS Conference on HIV Science (IAS 2017); Paris: International AIDS Society; 2017. 2017 Presented at: 9th IAS Conference on HIV Science; 2017; Paris.
- Johnston LG, Prybylski D, Raymond HF, Mirzazadeh A, Manopaiboon C, McFarland W. Incorporating the service multiplier method in respondent-driven sampling surveys to estimate the size of hidden and hard-to-reach populations: case studies from around the world. Sex Transm Dis 2013 Apr;40(4):304-310. [CrossRef] [Medline]
- TheWorldBank. Papua New Guinea Country Profile URL: http://databank.worldbank.org/data/Views/Reports/ReportWidgetCustom.aspx?Report_Name=CountryProfile&Id=b450fd57&tbar=y&dd=y&inf=n&zm=n&country=PNG [accessed 2018-09-09] [WebCite Cache]
- The World Bank. Strategic Directions for Human Development in Papua New Guinea URL: https://openknowledge.worldbank.org/bitstream/handle/10986/13543/40924.pdf?sequence=1&isAllowed=y [accessed 2018-08-10] [WebCite Cache]
- Kelly A, Kupul M, Man W, Nosi S, Lote N, Rawstorne P, et al. Askim na save (Ask and understand): People who sell and/or exchange sex in Port Moresby. Key Quantitative Findings. Sydney: Papua New Guinea Institute of Medical Research and the University of New South Wales; 2011:978.
- Maibani G, Ryan C, Lote N, Edward G, Ralai A, Gare J, et al. Papua New Guinea-Australia Sexual Health Improvement Program (PASHIP). Report on the Baseline Survey. PNG IMR 2011.
- Vallely A, Page A, Dias S, Siba P, Lupiwa T, Law G, et al. The prevalence of sexually transmitted infections in Papua New Guinea: a systematic review and meta-analysis. PLoS One 2010 Dec 23;5(12):e15586 [FREE Full text] [CrossRef] [Medline]
- Quaye S, Fisher RH, Atuahene K, Amenyah R, Aberle-Grasse J, McFarland W, Ghana Men Study Group. Critique and lessons learned from using multiple methods to estimate population size of men who have sex with men in Ghana. AIDS Behav 2015 Feb;19 Suppl 1:S16-S23. [CrossRef] [Medline]
- Holland CE, Kouanda S, Lougué M, Pitche VP, Schwartz S, Anato S, et al. Using Population-Size Estimation and Cross-sectional Survey Methods to Evaluate HIV Service Coverage Among Key Populations in Burkina Faso and Togo. Public Health Rep 2016 Nov;131(6):773-782. [CrossRef] [Medline]
- Sharifi H, Karamouzian M, Baneshi MR, Shokoohi M, Haghdoost A, McFarland W, et al. Population size estimation of female sex workers in Iran: Synthesis of methods and results. PLoS One 2017;12(8):e0182755 [FREE Full text] [CrossRef] [Medline]
- Fearon E, Chabata ST, Thompson JA, Cowan FM, Hargreaves JR. Sample Size Calculations for Population Size Estimation Studies Using Multiplier Methods With Respondent-Driven Sampling Surveys. JMIR Public Health Surveill 2017 Sep 14;3(3):e59 [FREE Full text] [CrossRef] [Medline]
- McLaughlin K, Handcock M, Johnston L. UCLA College. 2015. Inference for the Visibility Distribution for Respondent-Driven Sampling URL: http://www.stat.ucla.edu/katherine.mclaughlin/JSMpaper_mclaughlin.pdf [WebCite Cache]
- Handcock M, Fellows I, Gile K. Statistics UCLA. RDS Analyst: Software for the Analysis of Respondent-Driven Sampling Data Version 0 URL: http://hpmrg.org/ [accessed 2018-12-21] [WebCite Cache]
- Johnston LG, McLaughlin KR, El RH, Latifi A, Toufik A, Bennani A, et al. Estimating the Size of Hidden Populations Using Respondent-driven Sampling Data: Case Examples from Morocco. Epidemiology 2015 Nov;26(6):846-852 [FREE Full text] [CrossRef] [Medline]
- Handcock MS, Gile KJ, Mar CM. Estimating hidden population size using Respondent-Driven Sampling data. Electron J Stat 2014;8(1):1491-1521 [FREE Full text] [CrossRef] [Medline]
- Kelly CA, Soler-Hampejsek E, Mensch BS, Hewett PC. Social desirability bias in sexual behavior reporting: evidence from an interview mode experiment in rural Malawi. Int Perspect Sex Reprod Health 2013 Mar;39(1):14-21 [FREE Full text] [CrossRef] [Medline]
- Adebajo S, Obianwu O, Eluwa G, Vu L, Oginni A, Tun W, et al. Comparison of audio computer assisted self-interview and face-to-face interview methods in eliciting HIV-related risks among men who have sex with men and men who inject drugs in Nigeria. PLoS One 2014;9(1):e81981 [FREE Full text] [CrossRef] [Medline]
- Ghanem KG, Hutton HE, Zenilman JM, Zimba R, Erbelding EJ. Audio computer assisted self interview and face to face interview modes in assessing response bias among STD clinic patients. Sex Transm Infect 2005 Oct;81(5):421-425 [FREE Full text] [CrossRef] [Medline]
|BBS: biobehavioral surveys|
|CDC: Centers for Disease Control and Prevention|
|FSW: female sex workers|
|KP: key populations|
|MSM: men who have sex with men|
|PNG: Papua New Guinea|
|RDS: respondent-driven sampling|
|SS-PSE: successive sampling-population size estimation|
|TGW: transgender women|
|UOM: unique object multiplier|
Edited by J Neal; submitted 13.06.18; peer-reviewed by L Johnston, A Gurunga; comments to author 06.08.18; revised version received 27.09.18; accepted 06.10.18; published 21.03.19.
©Damian Weikum, Angela Kelly-Hanku, Parker Hou, Martha Kupul, Angelyne Amos-Kuma, Steven G Badman, Nick Dala, Kelsey C Coy, John M Kaldor, Andrew J Vallely, Avi J Hakim. Originally published in JMIR Public Health and Surveillance (http://publichealth.jmir.org), 21.03.2019.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Public Health and Surveillance, is properly cited. The complete bibliographic information, a link to the original publication on http://publichealth.jmir.org, as well as this copyright and license information must be included.