|
Non-Sampling Errors in Respondent Driven Sampling Method
|
| Arezoo Bagheri *, Mahsa Saadati |
| National Population Studies & Comprehensive Management Institute |
|
|
Abstract: (2541 Views) |
| Many populations which are in the most focus of policy-makers are hidden or hard to reach populations that are formed by people with certain condition such as HIV infection. A careful study of these populations in order to their identification is very important, since their high risk behavior jeopardizes public health. Thus, defining a reliable and valid sampling method of them is also very vital. Applying conventional sampling methods (probability) to take samples from these populations since there is rarely a list or sampling frame of them are not practical and these methods are associated with coverage error. Non-probability sampling methods such as chain referral sampling also do not produce statistically valid estimates. One of the valid methods to take samples from these populations is respondent driven sampling method. In this sampling method, one of the most important sources of non-sampling errors is the bias which results from homophily. To check the validity of samples taken from this sampling method, in this article homophily is introduced which is a tool to assess the level of bias and the representative of samples. The main objective of this article is to examine the consequences of homophily in the production of unbiased estimators in this sampling method. |
|
| Keywords: Respondent driven sampling, bias, Homophily, degree, equilibrium |
|
|
Full-Text [PDF 259 kb]
(885 Downloads)
|
Type of Study: Applicable |
Subject:
Special
Received: 2016/10/23 | Accepted: 2018/07/7 | Published: 2018/09/25
|
|
|
|
|
|
