Non-probability samples rely on convenience, judgment, or self-selection, then data are cheaper, faster and easier to collect compared to probability sampling methods. Nonetheless, the major concern with the use of this kind of data is their non-representativeness that may led to large bias in survey estimates. In this paper two approaches to make inference from informative non-probability samples are discussed. The first one, based on the empirical likelihood, does not require parametric specification of the population model but the probability of being in the non-probability sample needed to be modeled. The second one introduces the concept of uncertainty on data generating model resulting from the lack of knowledge of the sampling design acting in the non-probability sample. A measure of uncertainty is introduced and its reduction due to the availability of extra-sample information is evaluated.
Alternative Approaches for Addressing Selection Bias in Non-probability Samples / Marella, D.. - (2026). (SIS-FENStatS 2026 2026 roma ).
Alternative Approaches for Addressing Selection Bias in Non-probability Samples.
Daniela Marella
2026
Abstract
Non-probability samples rely on convenience, judgment, or self-selection, then data are cheaper, faster and easier to collect compared to probability sampling methods. Nonetheless, the major concern with the use of this kind of data is their non-representativeness that may led to large bias in survey estimates. In this paper two approaches to make inference from informative non-probability samples are discussed. The first one, based on the empirical likelihood, does not require parametric specification of the population model but the probability of being in the non-probability sample needed to be modeled. The second one introduces the concept of uncertainty on data generating model resulting from the lack of knowledge of the sampling design acting in the non-probability sample. A measure of uncertainty is introduced and its reduction due to the availability of extra-sample information is evaluated.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


