Section 2. What are the sources of migration data, their strengths and limitations?

In this section we will explore traditional and new data sources (Figure 3.3) as well as discuss the challenges and limitations of each type of migration data.

2.1 What are the traditional migration data sources?

Among the main traditional migration data sources are (Allen and Vargas‐Silva 2024, p. 86):

• censuses collected by national statistical offices every five or ten years. They usually collect information about the country of birth, citizenship, sometimes family history of migration, and year of arrival in the country. In some countries population registry data serve similar purposes;
• administrative sources, such as data registers of health records, tax-related data, educational statistics, social benefits, criminal records, police and border registers, visas and work permits which allow monitoring of changes to the situation of groups of migrants and estimating migration flows into the country;
• household surveys that can be national and international. These provide detailed information on the socio-economic characteristics of people, their attitudes and experiences of migration, and measure access to services.

2.2 What are the new data sources?

Innovative data sources not directly related to migration are increasingly used to help estimate migration-related phenomena and to study different groups of migrants.

New migration data sources include (Tjaden 2021, p. 15, IOM-UN Migration 2023):

Mobile phone data (call detail records [CDRs], GPS location, phone location history, also information about the time and associated cell tower of users’ text messages, calls, and data exchanges, which provide information about movements in almost real time at local, regional, and national levels)

Social media data (Facebook, Twitter, Instagram, LinkedIn, TikTok, Google) provides information about user activities and positions based on GPS location data, geotagged activities, Internet Protocol addresses and Wi-Fi network locations. Social media data might be useful in estimating changes in friendship/communication networks, in estimating migration stocks and flows, as well as attitudes towards migrants or providing information about migration roots or smuggling services advertised via social media platforms

Ukrainian-language Twitter data was also used in a combination with the news and events database (Armed Conflict Location and Event Data Project) to develop predictive models of forced displacement of Ukrainians to neighbouring countries (Wycoff et al. 2025)

Finally, in this video, based on 117 videos posted between January 2023 and April 2024, you can learn how Luigi Achilli presents his analysis of how migrant smugglers use TikTok to advertise irregular migration services (Migration Policy Centre 2024)

Online search engines, such as Google or Bing, can be monitored by governments to look for the intentions and destinations of potential migrants from specific countries. Although such predictions might approximately indicate aspirations but not necessarily migration plans or their implementation, they are also used as one of the tools in diaspora mapping through the analysis of searches in a particular language or about a particular country from another location

Satellite data and drone images can be used to collect information about surface changes, light emission at night, the level of construction/destruction of built structures or infrastructure, estimating local traffic. These types of data can potentially help to estimate the number of residents, seasonal migration/displacement, and infrastructural changes on the ground after a disaster

Air passenger data help to estimate international migration flows and measure the differences between arrivals and returns between countries. They are increasingly used by the authorities to monitor potential terrorist and criminal activities

Online news platforms and databases of events can provide us with data about migration events, disasters, migration aspirations or attitudes. They might be especially useful in finding data on new proximate causes of migration, early indicators of possible trigger events (e.g. emerging conflicts, disasters) or be used in combination with other traditional or big data sources (Henningsen 2025).

Among the most notable databases are:

1. Armed Conflict Location & Event Data (ACLED)
2. Global Data on Events Location and Tone Project (GDELT)
3. University of Uppsala’s conflict databases
4. Event Registry
5. Integrated Crisis Early Warning System (ICEWS)
6. The U.S. National Oceanic and Atmospheric Administration databases on weather and natural disasters (NOOA databases)
7. The Europe Media Monitor (EMM) project

Artificial intelligence and machine learning should also be included in data innovation in migration research because they can help us analyse large amounts of data from a variety of sources, predict migration flows from specific countries, and evaluate attitudes towards migrants and their experiences.

Both traditional and new migration data sources have strengths and limitations we must be aware of when using them.

2.3 What are the challenges and gaps in migration data sources?

With respect to traditional migration data sources, challenges can occur throughout the whole data pathway process (see Figure 3.5) starting from the choice of definition as to who is a migrant and what is migration; through the choice of measurement methodology; the methods of data collection; and subsequently due to the challenges of sharing, accessing, interpreting and disseminating migration data.

Migration researchers agree that, despite UN recommendations on the standardisation of definitions, some countries still prefer to use their national definitions of ‘international migrant’, especially with regard to the duration of stay abroad. This impacts on the global comparability of data.

Information on the drivers of migration, geographic distribution, migrants’ socio-demographic characteristics is not always available.

Moreover, the timeliness of migration data, especially derived from censuses, might make such data more difficult for policymakers and practitioners to use efficiently in real time (Allen and Vargas‐Silva 2024, p. 84).

This explains the major gaps we still find in global migration data (see Figure 3.6).

Furthermore, some groups of migrants are difficult to reach and measure. These include:

• people with dual citizenship who might be doubly counted;
• seasonal workers or circular migrants who might not be included in household surveys or censuses;
• persons with irregular legal status who will most likely not be included in data collected by the authorities due to their fear of being detected or deported.

We only have scattered data about the missing migrants.

A further challenge with migration data is fragmentation and poor coordination between the different agencies producing migration data and those using them.

For example, national statistical offices might not have the data needed by the ministries of health or education to evaluate aspects of migrants’ integration or lack the staff and resource capacity to use available data.

Finally, we should also consider the capacities of different actors and agencies to generate, evaluate, exchange and use migration data, as well as being able to assess its quality, comparability and relevance for their activity and policy priorities (Laczko and Vidal 2024). As a result, some data might be underused or not used at all, especially if the data is considered not relevant, not trustworthy or not suitable for specific political agendas.

Migration researchers argue (Laczko, Mosler Vidal, and Rango 2024, p. 74) that gaps in migration data “hinder the ability of policymakers and practitioners to:

1. understand past and current migration trends (descriptive analysis);
2. identify the drivers of shifts in migration and mobility patterns, as well as their impact on individuals and communities at large (causal analysis);
3. make predictions about how these patterns will evolve in the short or long term, or anticipate future shifts (predictive analysis);
4. assess the impact of policy and programmatic measures on migration and related areas, as well as monitor the effectiveness of specific policy interventions over time (impact analysis).”

With regard to the new migration data sources, numerous challenges derive primarily from their digital nature and the ethical challenges it poses ((IOM-UN Migration 2023; Dodevska 2024; Laczko, Mosler Vidal, and Rango 2024, Donato et al. 2025, Yildiz et al. 2025, Zook et al. 2017), in particular:

• concerns about ownership, privacy, security and ethical use of data for research or policy collected without explicit informed consent for such purposes. As new data sources are often collected by private entities, concerns about the reliability and integrity of information from the new data sources should be taken into consideration as well as its potential impact on migrants in vulnerable situations;
• trustworthiness and the extent of regulatory frameworks on the use of new types of data may vary significantly across countries. For example, the European General Data Protection Regulation includes very strict rules for user profiling used to make automated decisions that target individuals and may have legal or other types of personal consequences;
• availability and capacities (analytical and technical) to use and interpret new types and the amount of data by different actors and agencies, the risk of misuse, problems of accessibility and costs;
• selection bias when using social media tools (e.g. the use of different platforms is uneven across countries and different migrant groups; it also changes over time), concerns about continuity and sustainability of using this type of data if platforms cease to function;
• the timeliness of new migration data sources does not always mean it is useful for longer-term forecasting, and there is a risk of misuse or politicization of data;
• combination of traditional and new migration data sources might be difficult due to difficulties in harmonizing data across spatial and temporal scales,
• presumed ‘neutral’ automated screening and analysis of data may lead to real-life decisions that can be biased, discriminatory or wrong because of selection bias, misinterpretation of data or technical failures.

Even though new migration data tools, especially artificial intelligence-driven tools, could speed up and improve the analysis of large amounts of various types of data, we should be mindful about its potential limitations, especially:

• which migrant groups are either not covered or are underrepresented in the data;
• which information might be missing;
• the time and geographical limitations of the data collected.

2.3 Are migration data and evidence the same thing?

The two terms ‘data’ and ‘evidence’ are often used interchangeably, but do they mean the same thing? Not really.

Different types of data can be used as part of evidence together with stakeholder input, academic research, and many other forms of information that allow policymakers to assess whether policies are effective in improving integration outcomes.

Data does not speak for itself; it needs analysis and interpretation that will contextualize the findings and make them part of a broader picture.

For instance, regarding migration, migration researchers study various aspects of migrant integration from multiple angles combining the use of migration data and other types of information (B. J. Slootjes and Zanzuchi 2023) to evaluate:

Policy inputs, namely:

1. what kind of migrant integration policies and practices exist;
2. how did they change over time and across countries;
3. how were they implemented and how effectively?

Policy outcomes, namely:

1. what are the outcomes of migrants’ integration?
2. what gaps need to be addressed?

Policy impacts, including:

1. how are migrant integration policies connected to migrant integration outcomes?
2. What are the intended and unintended consequences of policies?
3. What makes certain policies effective, for which migrant groups?

To sum up, migration data is important but not sufficient to analyse complex and multi-dimensional social phenomenon of migration and integration. Yet data serves as building blocks for testing research hypotheses, evaluating policies and making more informed decisions. Without quality data you can’t have enough evidence.