Enhancing estimation methods for integrating probability and nonprobability survey samples with machine‐learning techniques. An application to a Survey on the impact of the COVID‐19 pandemic in Spain

Abstract

Web surveys have replaced Face‐to‐Face and computer assisted telephone interviewing (CATI) as the main mode of data collection in most countries. This trend was reinforced as a consequence of COVID‐19 pandemic‐related restrictions. However, this mode still faces significant limitations in obtaining probability‐based samples of the general population. For this reason, most web surveys rely on nonprobability survey designs. Whereas probability‐based designs continue to be the gold standard in survey sampling, nonprobability web surveys may still prove useful in some situations. For instance, when small subpopulations are the group under study and probability sampling is unlikely to meet sample size requirements, complementing a small probability sample with a larger nonprobability one may improve the efficiency of the estimates. Nonprobability samples may also be designed as a mean for compensating for known biases in probability‐based web survey samples by purposely targeting respondent profiles that tend to be underrepresented in these surveys. This is the case in the Survey on the impact of the COVID‐19 pandemic in Spain (ESPACOV) that motivates this paper. In this paper, we propose a methodology for combining probability and nonprobability web‐based survey samples with the help of machine‐learning techniques. We then assess the efficiency of the resulting estimates by comparing them with other strategies that have been used before. Our simulation study and the application of the proposed estimation method to the second wave of the ESPACOV Survey allow us to conclude that this is the best option for reducing the biases observed in our data.

Calibration and XGBoost reweighting to reduce coverage and non-response biases in overlapping panel surveys: Application to the Healthcare and Social Survey

“…[20] showed that Gradient Tree Boosting can lead to selection bias reductions in situations of high dimensionality or where the selection mechanism is Missing at Random (MAR). [25,26,27,28,29,30,31] have applied boosting algorithms in propensity score weighting showing better results than conventional parametric models.…”

Discriminative machine learning for maximal representative subsampling

“…The most commonly used method for PSA is logistic regression, but recent research showed that different machine learning methods can be advantageous in estimating the propensity score. At this point, it is not yet clear which algorithm is preferred for different distributions 18 – 20 .…”