Following the assumption that trends of online queries may indicate intentions and help to predict human behavior, this study addresses the general issue of analyzing, nowcasting, and predicting migrant decisions through an analysis of Google search patterns in the case of Syrians in Turkey. Aiming to contribute to the literature on predicting migration patterns, we examine the relationship between Google search queries for province names in Turkey and the number of Syrians under temporary protection across provinces from January 2016 to December 2019 and demonstrate a positive and significant association. Then, we explore the predictive power of Google searches in predicting the stock of Syrians under temporary protection in Turkey across provinces. We exploit the alphabetical difference between Turkish and Arabic as the method of differentiation between host and migrant populations. Our findings indicate that Google searches can be good predictors for estimating refugee stocks, especially when traditional data are not available. They can also be helpful in forecasting the changing pattern of migrant stocks at frequent intervals, to which conventional socioeconomic indicators are less sensitive due to their less frequent reporting periods.

中文翻译：

---

寻找新家：难民库存和 Google 搜索


假设在线查询的趋势可能表明意图并有助于预测人类行为，本研究通过分析土耳其叙利亚人的谷歌搜索模式，解决了分析、即时预测和预测移民决定的一般问题。为了为预测移民模式的文献做出贡献，我们研究了土耳其省名的谷歌搜索查询与 2016 年 1 月至 2019 年 12 月期间跨省受到临时保护的叙利亚人数量之间的关系，并证明了一种积极且显着的关联。然后，我们探讨了谷歌搜索在预测土耳其各省临时保护的叙利亚人数量方面的预测能力。我们利用土耳其语和阿拉伯语之间的字母差异作为区分东道国人口和移民人口的方法。我们的研究结果表明，谷歌搜索可以成为估计难民库存的良好预测工具，尤其是在无法获得传统数据的情况下。它们还有助于预测频繁间隔的移民存量的变化模式，而传统的社会经济指标由于报告周期不太频繁，因此不太敏感。