Abstract Recently, attention was drawn to the failure of two very large internet-based probability surveys to correctly estimate COVID-19 vaccine uptake in the U.S. in early 2021. Both the Delphi-Facebook COVID-19 Trends and Impact Survey (CTIS) and Census Household Pulse Survey (HPS) overestimated uptake substantially, by 17 and 14 percentage points in May 2021, respectively. These surveys had large numbers of respondents but very low response rates (<10%), thus, nonignorable nonresponse could have had substantial impact. Specifically, it is plausible that ‘anti-vaccine’ individuals were less likely to participate given the topic (impact of the pandemic on daily life). In this article, we use proxy pattern-mixture models (PPMMs) to estimate the proportion of adults (18 +) who received at least one dose of a COVID-19 vaccine, using data from the CTIS and HPS, under a nonignorable nonresponse assumption. Data from the American Community Survey provide the necessary population data for the PPMMs. We compare these estimates to the true benchmark uptake numbers and show that the PPMM could have detected the direction of the bias and provide meaningful bias bounds. We also use the PPMM to estimate vaccine hesitancy, a measure for which we do not have a benchmark truth, and compare to the direct survey estimates.

中文翻译：

---

使用代理模式混合模型来解释两项大型调查中对 COVID-19 疫苗接种估计的偏差

摘要最近，人们注意到两项基于互联网的大型概率调查未能正确估计 2021 年初美国的 COVID-19 疫苗接种情况。Delphi-Facebook COVID-19 趋势和影响调查 (CTIS) 和人口普查家庭Pulse Survey (HPS) 大幅高估了 2021 年 5 月的使用率，分别高估了 17 和 14 个百分点。这些调查有大量受访者，但答复率很低（<10%），因此，不可忽视的不答复可能会产生重大影响。具体而言，鉴于该主题（大流行对日常生活的影响），“抗疫苗”个人参与的可能性似乎较小。在本文中，我们使用 CTIS 和 HPS 的数据，在不可忽略的无反应假设下，使用代理模式混合模型 (PPMM) 来估计接受至少一剂 COVID-19 疫苗的成年人 (18 岁以上) 的比例。美国社区调查的数据为 PPMM 提供了必要的人口数据。我们将这些估计值与真实的基准采用率进行比较，结果表明 PPMM 可以检测到偏差的方向并提供有意义的偏差范围。我们还使用 PPMM 来估计疫苗犹豫（我们没有基准事实的衡量标准），并与直接调查估计进行比较。