The Use of Sampling and Weighting in Quantitative Social Research

Proceedings of The 6th International Conference on Future of Teaching and Education

Year: 2022



The Use of Sampling in Quantitative Social Research

Dr. Tamar Tako Doreuli, Prof. Dr. Nino Durglishvili, Vano kechakmadze, Guguli magradze



Generalizing results based on sampling population research about the general population represents one of the most significant issues of quantitative social research. The models of relevantly formed sampling represent a necessary precondition for overcoming such a complicated and demanding challenge. The review of the sampling model and its appropriate weighing is provided in the article, based on which a quantitative sociological survey was conducted in 2020 in the capital of Georgia – Tbilisi, using face-to-face interviews. When formulating the sampling model, the following analytical tools were employed: stratification, cluster sampling, PPS methodology, systematic random sampling, probability weighting, and calibration. All full-aged citizens (18+) of Tbilisi have been specified as a target group; the database of population census has been used as a sampling frame, which encompasses all current households in Tbilisi. The sample size comprised 1,000 respondents. The two-phase, stratified-cluster model represented the type of sampling. Districts of Tbilisi presented strata, while cluster sampling was done by census-taking districts. The weighting model is constructed according to the sampling stages. The calculation of the standard error (5%) relied on the utilization of Thompson’s formula (Thompson, 2012), and the design effect of cluster sampling  (Kish, 1995). To grasp the conceptual essence of the design effect, the emphasis was placed on estimating distribution density for dependent observations  (Kvatadze & Pharjiani, 2019). The central outcome of this research involves the representative sampling model of the total population, with a predefined standard error. However, it should be emphasized that such a model and individual techniques and approaches integrated within it can be successfully used in other similar research. The process of sampling modeling underscored the significant importance of the theoretical sources and analytical tools used. In addition, exploring density estimation for dependent observations could prove advantageous for extending the research in relevant contexts.

Keywords: Quantitative research, stratified-cluster sampling, PPS, design effect, weighting