Modeling and analyzing respondent-driven sampling as a counting process

Authors

• Yakir Berchenko,

• Jonathan D. Rosenblatt,

• Simon D. W. Frost

---

Summary

Respondent-driven sampling (RDS) is an approach to sampling design and analysis which utilizes the networks of social relationships that connect members of the target population, using chain-referral. RDS sampling will typically oversample participants with many acquaintances. Naïve estimators, such as the sample average, will thus be biased towards the state of the most highly connected individuals. Current methodology cannot estimate population size from RDS, and promotes inverse probability weighted estimators for population parameters such as HIV prevalence. We propose to use the timing of recruitment, typically collected and discarded, in order to estimate the population size via a counting process model. Once population size and degree frequencies are made available, prevalence can be debiased in a post-stratified framework. We adapt methods developed for inference in epidemiology and software reliability to estimate the population size, degree counts and frequencies. A fundamental advantage of our approach is that it makes the assumptions of the sampling design explicit. This enables verification of the assumptions, maximum likelihood estimation, extension with covariates, and model selection. We develop large-sample theory, proving consistency and asymptotic normality. We further compare our estimators to other estimators in the RDS literature, through simulation and real-world data. In both cases, we find our estimators to outperform current methods. The likelihood problem in the model we present is separable, and thus efficiently solvable. We implement these estimators in an accompanying R package, chords, available on CRAN.

Continue reading

Using Respondent-Driven Sampling to Recruit Illegal Drug Purchasers to Evaluate a Drug Market Intervention

SOURCE

Allison I. Ober 1

Jesse Sussell 2

Beau Kilmer

Jessica Saunders 1

Douglas D. Heckathorn 3

1. ¹RAND Corporation, Santa Monica, CA, USA
2. ²Precision Health Economics, San Francisco, CA, USA
3. ³Cornell University, Ithaca, NY, USA

Allison J. Ober, RAND Corporation, 1776 Main Street, Santa Monica, CA 90407, USA. Email: ober@rand.org

Abstract

Background Violent drug markets are not as prominent as they once were in the United States, but they still exist and are associated with significant crime and lower quality of life. The drug market intervention (DMI) is an innovative strategy that uses focused deterrence, community engagement, and incapacitation to reduce crime and disorder associated with these markets. Although studies show that DMI can reduce crime and overt drug activity, one perspective is prominently missing from these evaluations: those who purchase drugs.

Objectives This study explores the use of respondent-driven sampling (RDS)—a statistical sampling method—to approximate a representative sample of drug users who purchased drugs in a targeted DMI market to gain insight into the effect of a DMI on market dynamics.

Methods Using RDS, we recruited individuals who reported hard drug use (crack or powder cocaine, heroin, methamphetamine, or illicit use of prescriptions opioids) in the last month to participate in a survey. The main survey asked about drug use, drug purchasing, and drug market activity before and after DMI; a secondary survey asked about network characteristics and recruitment.

Conclusions Our sample of 212 respondents met key RDS assumptions, suggesting that the characteristics of our weighted sample approximate the characteristics of the drug user network. The weighted estimates for market purchasers are generally valid for inferences about the aggregate population of customers, but a larger sample size is needed to make stronger inferences about the effects of a DMI on drug market activity.

FULL TEXT PDF

 

Random Walks on Directed Networks: Inference and Respondent-Driven Sampling

Jens Malmros¹ / Naoki Masuda² / Tom Britton³

¹ Department of Mathematics, Stockholm University, SE-106 91 Stockholm, Sweden.

² Department of Mathematical Informatics, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japan.

³ Department of Engineering Mathematics, University of Bristol, Merchant Venturers Building, Woodland Road, Clifton, Bristol BS8 1UB, United Kingdom.

 

Download full text PDF

Lessons learned from respondent-driven sampling recruitment in Nairobi

• Jerry Okal¹Email author,
• Henry F. Raymond²,
• Waimar Tun³,
• Helgar Musyoki⁴,
• Sufia Dadabhai⁵,
• Dita Broz⁶,
• Joan Nyamu⁷,
• David Kuria⁸,
• Nicholas Muraguri⁹ and
• Scott Geibel³

BMC Research Notes20169:158

DOI: 10.1186/s13104-016-1965-y

©  Okal et al. 2016

Received: 21 November 2015

Accepted: 26 February 2016

Published: 11 March 2016

Abstract

Background

Respondent-driven sampling (RDS) is used in a variety of settings to study hard-to-reach populations at risk for HIV and sexually transmitted infections. However, practices leading to successful recruitment among diverse populations in low-resource settings are seldom reported. We implemented the first, integrated, bio-behavioural surveillance survey among men who have sex with men, female sex workers and people who injected drugs in Nairobi, Kenya.

Methods

The survey period was June 2010 to March 2011, with a target sample size of 600 participants per key populations. Formative research was initially conducted to assess feasibility of the survey. Weekly monitoring reports of respondent characteristics and recruitment chain graphs from NetDraw illustrated patterns and helped to fill recruitment gaps.

Results

RDS worked well with men who have sex with men and female sex workers with recruitment initiating at a desirable pace that was maintained throughout the survey. Networks of people who injected drugs were well-integrated, but recruitment was slower than the men who have sex with men and female sex workers surveys.

Conclusion

By closely monitoring RDS implementation and conducting formative research, RDS studies can effectively develop and adapt strategies to improve recruitment and improve adherence to the underlying RDS theory and assumptions.

Keywords

Men who have sex with men (MSM) Female sex workers (FSW) People who injected drugs (PWID) Respondent-driven sampling (RDS) Field experiences

more

Behavioral Surveillance Analyst Fellowship (CDC/ORISE)

Position Description: A fellowship opportunity is available with the Behavioral and Clinical Surveillance Branch of the Division of HIV/AIDS Prevention (DHAP) within the National Center for HIV/AIDS, Viral Hepatitis, STD, and TB Prevention (NCHHSTP) at the Centers for Disease Control and Prevention (CDC) in Atlanta, GA.

This fellowship offers the opportunity to work on a high-priority, high-impact issue in domestic HIV surveillance. With 20 participating metropolitan project areas throughout the United States, the National HIV Behavioral Surveillance (NHBS) system collects and reports data on three populations at increased risk for HIV infection: men who have sex with men, persons who inject drugs, and heterosexuals at increased risk in the United States. NHBS data are used to monitor issues affecting these populations, which include describing racial disparities, reporting HIV prevalence and awareness, exploring the prevalence of HIV- related risk behaviors, such as drug use and sex, and examining access to HIV testing, care and prevention. NHBS collects data through personal interviews and HIV testing and is the only national source of data on HIV- negative populations and HIV- positive individuals unaware of or not in care for their infection.

This fellowship will focus on conducting data analysis, project management of data processes, and assisting with dissemination of NHBS data. Specific opportunities during the fellowship may include: • Assisting in the analysis of data collected using respondent driven sampling (RDS) and venue- based sampling (VBS) methods • Assisting in the development of data documentation, reports, analysis, and questionnaire data quality for NHBS, which incorporates the use of RDS and VBS methods • Writing SAS programs to conduct data analyses and manage data • Aiding in the development of analysis guides, data documentation, and data management procedures for a national surveillance program • Researching technical issues that arise from data collection and management processes • Tracking data management and analysis processes and updating related guidance documents • Participating in data analysis and dissemination using surveillance data including co-authoring a surveillance report and a peer-reviewed journal article • Conducting end-user testing on Questionnaire Development System (QDS) coded questionnaire • Travel to state and local health department jurisdictions to support CDC in providing technical assistance. This program, administered by ORAU through its contract with the U.S. Department of Energy to manage the Oak Ridge Institute for Science and Education, was established through an interagency agreement between DOE and CDC. The initial appointment is for one year, but may be renewed upon recommendation of CDC contingent on the availability of funds. The participant will receive a monthly stipend commensurate with educational level and experience. Proof of health insurance is required for participation in this program. The appointment is full-time at CDC in the Atlanta, Georgia, area. Participants do not become employees of CDC or the program administrator, and there are no fringe benefits paid.

<strong>Qualifications:</strong> A Master’s degree or higher in public health, behavioral science, statistics or related field earned within the past five years is required A doctoral level degree earned within the past five years is preferred. Experience using statistical software is required (SAS preferred). Knowledge of HIV prevention is preferred. Strong oral and excellent writing skills are required.
Contact Name: Melissa Cribbin,  Behavioral Surveillance Analyst Fellowship (CDC/ORISE)
Contact Location:

Contact Location:
Contact Phone:
Contact Fax:
Contact Email: mwc4@cdc.gov

Web Address:
How to Apply: Email resume/CV to mwc4@cdc.gov. Please include the reference code CDC- NCHHSTP-2015-072 in your email.

An Introduction to Respondent-Driven Sampling

Institution: Medical Research Council South Africa

Venue: Cape Town, South Africa

Date: 25/05/2015 – 29/05/2015

Respondent-driven sampling (RDS) is a type of sampling increasingly used to sample from ‘hidden’ populations, such as men who have sex with men (MSM),injection drug users, the homeless, and immigrants. RDS involves peers recruiting their peers, which can result in rapid study recruitment, but also presents a number of logistical and statistical challenges. This event will provide an introduction to the basic theory of RDS, how best to conduct RDS studies in the field, and current methods for analysing data from RDS studies.

Organiser

Dr Loraine Townsend
Bryan Hansen

Level

Entry (no or almost no prior knowledge)

Cost

The fee is:

1. ZAR 1000 – For South African registered postgraduate students

2. ZAR 1900 – For staff at South African academic institutions, funded researchers and registered charity organisations

3. ZAR 5000 – For all other participants

All fees include event materials, lunch, morning and afternoon tea. They do not include travel and accommodation costs.

Website and registration

Contact Bryan Hansen
Training Administrator
BRYANT Research Systems
Cape Town, South Africa
Tel: +27 83 248 2476

More information and to register email: info@bryantresearchsystems.com

HIV Prevalence and Risk Behaviours Among Foreign Migrant Women Residing in Cape Town, South Africa.

Loraine Townsend, Maggie Giorgio, Yanga Zembe, Mireille Cheyip, Catherine Mathews

[
ABSTRACT: HIV prevalence and risk behaviour among foreign migrants in South Africa has not been explored. This paper describes the effectiveness of respondent-driven sampling (RDS) to recruit foreign migrant women residing in Cape Town, reports HIV prevalence, and describes key characteristics among them. We conducted a biological and behavioural surveillance survey using RDS. After written informed consent, participants completed an audio computer assisted self-interview and provided a dried blood sample for HIV analysis. HIV prevalence was estimated to be 7 % (CI 4.9-9.5) among 935 women. HIV sero-positivity was associated with older age (p = 0.001), country of origin (p < 0.000), being unmarried (p < 0.000), having lived in South Africa for 3-5 years (p = 0.023), sexual debut at ≥15 years (p = 0.047), and having used a condom at last sex with a main partner (p = 0.007). Few women reported early sexual debut, or multiple sexual partners. RDS was successful in recruiting foreign migrant women.

AIDS and Behavior 04/2014; · 3.49 Impact Factor

http://www.researchgate.net/profile/Loraine_Townsend/publications

Is formative research important before implementation of a survey using respondent driven sampling (RDS) among MSM in three South African cities?

A. Cloete¹, T. Duda¹, Y. Ntsepe¹, D. Naidoo², K. Jonas³, L. Simbayi¹

¹HIV/AIDS, STIs & TB (HAST) Research Programme, Human Sciences Research Council, Cape Town, South Africa, ²HIV/AIDS, STIs and TB Research Programme, Human Sciences Research Council, Durban, South Africa, ³Population Health, Health Systems & Innovation (PHHSI), Human Sciences Research Council, Cape Town, South Africa

Background: We conducted pre-surveillance formative research in order to inform the implementation of a HIV biological and behavioural survey using respondent driven sampling (RDS). This study was conducted in the three urban centres of South Africa, namely Cape Town, Durban and Johannesburg.
Methods: As part of formative activities, we established a Community Advisory Board (CAB) comprising largely of representatives from service providers for MSM in each of the three study cities. We conducted 53 semi-structured interviews with MSM from diverse backgrounds in each of the study cities. Information was gathered focusing on the selection of seeds, acceptability of RDS, level of incentives, staffing and coupon design. For the purposes of RDS analysis, questions on each participant’s network size are essential. Questions on network size were formulated but final decisions regarding whether or not we will obtain accurate responses regarding network size was determined in this activity. As part of formative activities, we conducted structured observations to find appropriate sites for RDS purposes in each of the three study cities.
Results: Overall MSM interviewed demonstrated interest to take part in our study. We found that MSM were socially networked however networks tended to be isolated within race and class categories. In addition, where stigmatisation of same sex behaviours are still apparent, MSM networks remained closed. Hence the selection of seeds is important in order to bridge race and socio-economic status divides. Moreover, concerns were raised regarding issues of confidentiality, anonymity, protection from further stigmatization, in particular for MSM who remain closeted. As a result, discreet and accessible interview sites were established in each of the three study cities.
Conclusions: We found that pre-surveillance formative research is not only important but is an essential step in the successful implementation of any RDS study, in particular in contexts where MSM remain stigmatized.
Download the e-Poster

The South African Marang Men’s Project: HIV bio- behavioural surveys conducted among men who have sex with men in Cape Town, Durban and Johannesburg using respondent driven sampling

The main findings of the Marang Men’s Project

To date, the Marang Men’s Project is the largest HIV prevalence study conducted among MSM in South Africa. The main findings of the study were as follows: 

HIV prevalence was high among MSM sampled in each of the three study cities.

Self-reported condom use at last sex with a man was high in all three cities.

The majority of MSM in each of the three studied  cities reported having three or more male sex  partners in the last six months.

Limitations of the study
• The use of RDS to sample MSM presents as a limitation.

• This, method does not sample a population directly, but via a connected social network (Kerr et al.,  2013).

• Hence, each network selected in the three cities sampled different sectors of the population category MSM (see Kerr et al., 2013).

• Crude samples in each of our three study cities were an over- representation of particular sub-categories or social networks of MSM.

• Cape Town – sex workers, previously incarcerated MSM

• Durban – students • Self-reported information: Under-and over reporting on sensitive topics such as sexual behaviours (ie condom use at last sex with a man) and alcohol use and sexual intercourse might have occurred.

www.hsrc.ac.za/en/research-data/ktree-doc/13782

[RDS] UCSF Surveillance Training: Methods for Sampling Key Populations – September 22-26, 2014 – Registration Open

The University of California, San Francisco (UCSF) Global Health Sciences is offering a course for implementing integrated bio-behavioral surveillance surveys among key populations using respondent driven sampling and time location sampling and integrating population size estimation. This course will be based on the recently released IBBS Toolbox available on line at globalhealthsciences.ucsf.edu/IBBStoolbox. This one week course will be at UCSF Global Health Sciences located in downtown San Francisco September 22-26, 2014 and will cover:

• Overview of HIV Surveillance and Sampling Methods

• Formative Assessment for IBBS

• Questionnaire Design

• Respondent Driven Sampling Nuts & Bolts

• Population Size Estimation

• Time Location Sampling Nuts & Bolts

• Using Results

Participants will leave with a CD or flash drive of the Toolbox.

The cost of the course is $1,200. To register, please use this registration link:

Register here