Evaluating sampling biases from third-party reporting as a method for improving survey measures of sensitive behaviors
Section snippets
Background
Sensitive and stigmatizing behaviors that put people’s health at risk are notoriously difficult to measure through surveys, due largely to social desirability biases (Cleland et al., 2004). Survey respondents, for example, misreport how much they smoke (Messeri et al., 2007; Muir et al., 1998), drink (Stockwell et al., 2004) or have sex (Cleland et al., 2004). This may affect estimates of the prevalence of such risk behaviors, as well as the design and evaluation of interventions aimed at
Data
We use data from the first round of the Likoma Network Study (LNS). Likoma is a small island located in the northern part of Lake Malawi, in East Africa. The LNS is a sociometric study (Morris, 2004) aimed primarily at documenting the dynamics of HIV transmission among the island’s population. The LNS did not implement TPR, but did collect data on friendship networks and sexual behaviors (Helleringer et al., 2009, 2014), which allow testing the validity of the two “structural” hypotheses
Analytic approach
We first describe the outdegree distribution of the friendship network, i.e., the number of friends each respondent nominated during the survey (Wasserman and Faust, 1994), by gender. We next assess if the surrogate sample of friends is representative of the population of interest by comparing the distributions of socio-economic characteristics (age, education, marital status, mobility, housing material and availability of a pit latrine in the household) between the samples of respondents and
Results
Nine women (1.8%) and 7 men (1.6%) did not nominate any friend of the same gender, whereas 294 women (58.6%) and 279 men (66.3%) nominated 4 such friends (Fig. 3). Among reported friendships, 2616 (83.9%) relationships were matched to a record in the household rosters.
There were no differences in the distribution of socio-economic characteristics between the respondent and the surrogate samples of men and women (Fig. 4). Among women, the surrogate sample of friends appeared slightly older
Discussion
Asking questions about third parties such as friends and confidants, rather than about respondents themselves, may reduce social desirability biases during surveys of sensitive behaviors. However, third-party reporting (TPR) rests on two structural assumptions about the underlying social networks within which respondents and their friends are embedded. We tested these assumptions using data from the Likoma Network Study.
We found that, among men and women, the surrogate sample produced by TPR
Acknowledgements
The authors' work on this paper was supported by grants from the National Institute of Child Health and Human Development (R03 HD071122 & P2C HD066613).
References (32)
- et al.
To tell the truth: measuring concordance in multiply reported network data
Soc. Networks
(2007) - et al.
Do tobacco countermarketing campaigns increase adolescent under-reporting of smoking?
Addict. Behav.
(2007) - et al.
Estimating clandestine abortion with the confidants method--results from Ouagadougou, Burkina Faso
Soc Sci Med
(2006) - et al.
The game of contacts: estimating the social visibility of groups
Soc. Networks
(2011) - et al.
Who knows your HIV status? What HIV + patientsand their network members know about each other
Soc. Networks
(1995) - et al.
Counting hard-to-count populations: the network scale-up method for public health
Sex. Transm. Infect.
(2010) - et al.
Monitoring sexual behaviour in general populations: a synthesis of lessons of the past decade
Sex. Transm. Infect.
(2004) Why your friends have more friends than you do
Am. J. Sociol.
(1991)Social desirability Bias and the validity of indirect questioning
J. Consum. Res.
(1993)- et al.
Death by survey: estimating adult mortality without selection bias from sibling survival data
Demography
(2006)
Informal confidential voting interview methods and temporal changes in reported sexual risk behaviour for HIV transmission in sub-Saharan Africa
Sex. Transm. Infect.
Sexual Network Structure and the Spread of HIV in Africa: Evidence from Likoma Island, Malawi
AIDS
The reliability of sexual partnership histories: implications for the measurement of partnership concurrency during surveys
AIDS
The likoma network study: context, data collection, and initial results
Demogr. Res.
Cohort profile: the likoma network study (LNS)
Int. J. Epidemiol.
Unconventional approaches to mortality estimation
Demogr. Res.
Cited by (6)
The confidante method to measure abortion: implementing a standardized comparative analysis approach across seven contexts
2023, Population Health MetricsDescribing the safety of abortion at the population level using network-based survey approaches
2022, Reproductive HealthInequities in the incidence and safety of abortion in Nigeria
2020, BMJ Global Health