June 8, 2017

The Digital Footprint of Europe’s Refugees

Methodology

This report relies on data from Google Trends, a publicly available, web-based tool that permits users to explore online search volume by language and location over a specified time period. In this report, Google Trends is matched with migration data on refugee arrivals from the United Nations High Commissioner for Refugees (UNHCR) and asylum applications from Eurostat, the European Union’s statistical agency. The data analysis in this report compares online searches with migration flows among Arabic-speaking (mainly Syrian and Iraqi) refugees from the Middle East to Europe in 2015 and 2016.

Online search data

Trends in online search data volume used in this report are based on data from Google Trends, a publicly available web-based tool that measures search volume of a specific search term across time, in specific languages and in specific countries. In this report, search term data are for the Arabic terms for “Greece” (searched in Turkey) and “German” (searched in Germany). Data were downloaded for the time span of Jan. 1, 2015 through Dec. 31, 2016, except for Google Trends showing hourly searches in March 2016.

Google is the most widely used search engine globally and is the top search engine used in Turkey and Germany. Google also is the search engine of choice in Syria and Iraq. Consequently, it is assumed that Google is the top search engine used by refugees coming from these countries.

Google standardizes online searches on a 0 to 100 scale for the time period selected by the user. Lower numbers indicate that terms were searched less frequently while higher numbers represent more frequently searched terms during the same time period. Google Trends data from Jan. 1, 2015 through Dec. 31, 2016, are in weekly estimates. These figures change slightly between downloads as they are a sample of searches.9 To smooth out the variation, most charts in this report show monthly averages taken from weekly search volume.

Internet search data analysis like that available from Google Trends has limitations. Internet search data is an indicator of interest in a topic, not always resulting from or leading to actual behavior. Actual motivation for searches or behavior related to searches cannot be authenticated using Google Trends alone. When combined with migration data, however, internet search patterns can be linked to actual movement.

Narrowing in on Syrian and Iraqi refugees

The analysis in this report assumes that most Syrian and Iraqi refugees use Arabic as their primary search language on the internet. This assumption is important since Arabic is not widely spoken in Turkey (about 1% of the Turkish resident population speaks Arabic).10 Consequently, sharp shifts in online searches are assumed to represent new Syrian and Iraqi arrivals and not other Arabic-speaking minorities within Turkey. To support this assumption, the Arabic term for “Greece” when searched in Turkey often occurred along coastal regions, places where Turkey’s resident Arabic population are not located in large numbers. Syrian refugees were largely concentrated in these regions in 2015 and 2016, waiting to go to Europe via Greece or taking refuge from the Syrian civil war on the Turkish side of the Turkey-Syria border. Arabic searches for “Greece” also were high in areas where many refugee camps are located in Turkey.

It is possible that some searches for “Greece” in Arabic could come from Turkey’s sizeable Kurdish population. However, searches for “Greece” in Arabic were not prominent in regions of Turkey where Kurdish is widely spoken. (The Kurdish population mostly resides in the southeastern region of Turkey). Nor does it appear that the small, non-refugee, Arabic-speaking population in Turkey was searching for “Greece” as a potential tourist location. If this had been the case, there most likely would have been upticks in online searches for “Greece” in the summer of 2016 or in earlier years, something that did not occur.

Arabic is also a minority language in Germany. This analysis assumes that fluctuations in Arabic searches largely represent the online searches of new refugee arrivals in Germany. Thousands of Syrians and Iraqis also settled in other European countries, including Hungary, Sweden and Austria. However, Google Trends data in Arabic were either unavailable for other European countries or when available, provided a less consistent direction in the trend.11

Selecting the best search terms

Selection of the online search terms used in the analysis of this report do not rely on top search terms in Arabic in Turkey or Germany as Google does not publicly release top search terms by language. Instead, several terms that may be linked to migration were explored. These words include “smuggling,” “asylum” and geographic destinations in Europe such as “Greece,” “Germany” and “Sweden.”

In many cases, geographic and migration terms were linked. For example the Arabic word – اليونان – for “Greece” among internet searches in Turkey was often searched with the Arabic word for “smuggling,” according to Google Correlate, a tool that presents statistics on relationships between search terms. Within Turkey, “smuggling” in Arabic was also searched alongside “seek refuge,” further supporting the notion that “Greece” was used alongside other search terms of potential migrants. And, it appears that these searches for assistance in migration from Turkey to Greece are unique to the Arabic language in Greece. Online search trends for “smuggling” and “asylum” in Turkish within Turkey, for example, did not show a similar trend for the same time period.

Searching for map terms like “Greece” for the purposes of migration is also consistent with a media report on Google search activity for Syrians within Syria in 2015 when “immigration to Germany,” “asylum in Germany,” “map of Europe,” “Izmir” (primary coastal town in Turkey for departure to Greece) and “Greece” were listed as some of the top searches as the migration crisis in Europe began.

Analysis of the Arabic term for “Greece” in Turkey from Google AdWords shows that monthly search volume averaged 100 to 1,000 hits during 2015 and 2016, except for peak migration months between June and September of 2015 when “Greece” searches were between 1,000 and 10,000 monthly. This is a lower number of searches than actual number of people migrating each month from Turkey to Greece. A lower-than-expected number of searches for “Greece” each month suggests that Arabic-speaking internet users in Turkey were not searching for “Greece” out of curiosity or for news-related events.

The term “Greece” in Turkey was selected above other geographic terms because of its physical proximity for migration to Europe. Also, according to Google AdWords, “Greece” had a far greater number of monthly searches than other words like “smuggling” and “asylum” during the peak migration months. Consequently, it is assumed that “Greece” is less of an aspirational search, but a practical one for those searching for assistance with actual travel. The peak timing of such searches in the middle of the night is evidence for this assumption. At the same time, trends of the Turkish word for “Greece” in Turkey were mostly flat for the same time period with single week peaks in July and November 2015. Again, this suggests the searching of “Greece” in Turkey is specific to Arabic speakers in Turkey.

At the same time, the Arabic term for “Greece” within Turkey appears to be intended for migration purposes and not part of broader Arabic searches for the same term worldwide. The Arabic term for “Greece” worldwide, for example, peaks for a single week in July 2015 and subsides within a few weeks.

Google searches for the term “Greece” in English as well as Turkish within Turkey also surged in a few occasions during 2015, presumably as English and Turkish speakers in Turkey were following the news of the refugee surge or events in Greece, such as the referendum, bailout and capital controls. These peaks in searches, however, only lasted for a week or two and were not as sustained as the same term in Arabic.

In Germany, the Arabic word – الألمانية – for “German” may indicate interest in learning the German language or performing online translation. Other terms in Arabic indicating transition into Germany society such as “integration course,” “school,” “jobs” were also examined, but showed less notable trends with the number of asylum applications. At the same time, there was no noticeable trend for the search term “German” in the German language in either Germany or worldwide. Also, there was no identifiable trend in Arabic searches for the Arabic word for “German” worldwide. This suggests that the trend observed in Arabic searchers for the word “German” in Germany is uniquely representative of Arabic speakers in Germany.

Migration data

Monthly migration data comes from two sources: (1) the United Nations High Commissioner for Refugees (UNHCR), for new arrivals into Greece and (2) Eurostat figures for first-time asylum applications for Germany. Since both Syrian and Iraqi asylum seekers have had fairly high acceptance rates in Europe, it is assumed that most Syrian and Iraqi migrants entering during 2015 and 2016 applied for asylum, and did not try to evade authorities.

Monthly arrivals by country of origin on Greece’s shores from UNHCR since 2015 are unavailable. Consequently, comparisons of online searches with UNHCR data are for all refugee arrivals, not just Syrians and Iraqis. For 2015 and 2016 as a whole, however, nearly two-thirds (65%) of all sea arrivals into Greece recorded by UNHCR were Syrian and Iraqi citizens.

Every asylum seeker entering Europe, whether a child or an adult, must complete an asylum application. Data used in this report include only first-time asylum applicants, not those who have appealed an unsuccessful application. Although the European Union’s digital system (EURODAC) seeks to minimize repeat applications across member states, some asylum applicants may have applications across multiple countries as refugees may move between European countries.

  1. For more information on sampling of Google Trends, see the methodology of the Pew Research Center’s report, “Searching for News: The Flint water crisis.”
  2. Arabic is the majority language of Syrians and Iraqis in their home countries, but only about 1% of the non-refugee population in Turkey speaks the language. Similarly, an estimated 0.5% (or about half a million) of Germany’s population spoke Arabic prior the arrival of most Syrian and Iraqi refugees in 2015 and 2016, based on the number of foreign-born population from Arabic-speaking countries from United Nations migrant estimates  for mid-year 2015.
  3. One exception was Sweden where the Arabic term for “Swedish” (السويدية) tracked with first-time monthly asylum applications from Syrians and Iraqis. Sweden was not presented in the report because numbers of Syrian and Iraqi refugees to Sweden (about 80,000) were not as numerous as Germany (more than half a million).