Meta-analysis and evidence synthesis for migration and demography studies

Evidence synthesis in migration and demography

Demography measures and explains population change through births, deaths and migration. Migration studies examine why people move, how they fare in new places, and how receiving and origin communities change. Synthesis questions include whether migrants differ from non-migrants in health, how discrimination in hiring varies by group and country, how fertility changes after migration, how return migration affects origin communities, and which integration policies are associated with outcomes. Some of these topics have large numbers of comparable studies, such as audit studies of hiring discrimination.

The evidence presents specific issues. Definitions of who is a migrant (by birthplace, nationality, ethnicity or length of stay) vary and change over time. Data sources capture some groups much better than others, with undocumented migrants and people who move frequently often missing. Findings depend on the origin and destination countries and the period. And the topic is sensitive and politically charged. Our methods follow systematic review and meta-analysis practice, adapted to these features. This page builds on the general guidance for social sciences. The service provides research synthesis only.

Definitions and data sources

A review should state its definition and map each study's definition to it. For example, studies of the health of migrants should record whether the comparison is with the population of origin or of destination, since "healthy migrant" selection and later health changes lead to different patterns by comparison group. Registry data may exclude people who leave the country, which can bias mortality rates, and survey data may under-sample some groups. These limits apply to prevalence and rate pooling as well as to comparative effect sizes.

Correspondence experiments on discrimination

Correspondence (audit) experiments send matched applications that differ only in a signal such as the name of the applicant to employers or landlords, and measure differences in callback rates. They give causal evidence of differential treatment at the first stage of hiring. Meta-analyses have found consistent evidence of lower callback rates for several minority and migrant groups, with large variation across countries, occupations and over time. The effect measure is usually the ratio of callback rates or an odds ratio. A review should extract the signal used, the occupation, the country and year, the number of applications and the callbacks in each group. These experiments measure a particular stage, not final hiring or earnings, and the signal used (for example, a name) can carry other information, such as perceived class. Their strength is clear causal identification, their limit is scope.

Health and wellbeing of migrant populations

Studies compare migrants and non-migrants on mortality, mental health, chronic disease and use of health care. Findings include lower mortality among some migrant groups at arrival in some destination countries, the so-called healthy migrant effect, which can fade with time, and higher rates of some mental health problems among refugees. Prevalence of conditions among refugees is pooled with methods for prevalence, and varies greatly with the screening tool, sampling frame (camp, community, clinic) and time since displacement. A review should report prevalence with prediction intervals, record the instrument and cultural validation, and avoid describing screen-positive rates as diagnoses. Clinical aspects are covered under public health.

Labor market, fiscal and local effects

Studies of the effects of immigration on wages and employment of the host population use different methods: variation across regions, natural experiments from sudden inflows, and structural models. Results differ with the method, the skill group and the definition of the labor market. Meta-analyses, like those on other economics topics, have reported small average effects with variation and some evidence of selective reporting. Studies of migrants' own outcomes examine earnings assimilation, which depends on the years since arrival, the skill level, language and legal status. A review should show results by method and group, and should apply the methods for economic meta-analysis described on the labor economics page.

Fertility, ageing and population change

Demographic syntheses address questions such as the relation between education and fertility, the effect of family policies on birth rates, and the association between fertility and later health. Period fertility measures can mislead when the timing of births changes. Cross-country analyses have few units, and effects of policies on fertility are modest and hard to separate from other changes. Population projections are model outputs based on assumptions about fertility, mortality and migration, and are not pooled as observations. A review should describe assumptions and uncertainty ranges in projections it summarizes and avoid presenting them as predictions.

Qualitative and participatory research

Qualitative studies document migrants' experiences of journeys, settlement, work, discrimination and belonging. They are essential for understanding mechanisms and for including the voices of the people studied. Qualitative evidence synthesis can combine such studies, and the qualitative evidence synthesis page describes the methods. Reviews that involve researchers from migrant communities or origin countries improve interpretation, and we describe how involvement was arranged.

Publication bias, language and coverage

Research on migration is published in many languages and regions, and searching only English-language sources misses large parts of the literature. We search regional databases where relevant, report language coverage and check small-study effects within comparable sets of studies. Studies that find large effects of discrimination may be more likely to be published, and correspondence experiments are increasingly preregistered, which allows comparison with registered plans.

Integration policy and education outcomes

Studies of integration policy examine language courses, civic programs, reception conditions, access to work and housing allocation. Many use administrative data and exploit changes in rules or quasi-random allocation of refugees to locations, which gives credible evidence on a few questions, such as the effect of early labor-market access on employment. Education studies compare the achievement of children of migrants with that of other children, often finding gaps that shrink after accounting for socioeconomic background and language, with differences by origin, age at arrival and school system. A review should record which covariates were controlled, the age at arrival, and the structure of the school system, since tracking and early selection affect outcomes.

Integration is a two-way process, and measures that only focus on migrants' behavior leave out the actions of receiving societies. Reviews should note which perspective the measures take, who defined success, and whether the people studied were consulted. Return and circular migration complicate the picture further, because people who leave are not observed in destination data, and conclusions about long-run integration depend on who stayed, a selection that a careful review states openly.

Common pitfalls we look for

• Combining different definitions of migrant without a moderator.
• Using registry data that exclude people who leave.
• Treating screen-positive rates as diagnoses.
• Treating callback differences as measures of final hiring or earnings.
• Treating population projections as observations.
• Generalizing from a few destination countries.

Planning a migration or demography synthesis

We help define the population, comparison and outcomes with precise definitions, plan searches in Sociological Abstracts, Scopus, Web of Science, PopLine, EconLit, MEDLINE and regional sources, and set up coding of origin and destination, period, definition, data source, design and outcome. See the systematic review service for scope and process.

An invented example of a callback ratio

Suppose a review of 30 invented correspondence experiments finds that applicants with a minority-group name receive callbacks 30 percent less often on average than majority-name applicants. If the majority callback rate is 20 percent, the minority rate would be about 14 percent, a gap of six percentage points. The ratio is about 0.70, and a prediction interval from 0.45 to 1.10 would show that in some settings the gap is absent. If the ratio is closer to 0.85 in occupations with labor shortages and 0.55 elsewhere, the context matters. These results describe first-stage callbacks in the markets tested, and they do not describe final hiring or wages.

Coding and transparency

Coding frames record origin and destination countries, definition of migrant, period, data source, sampling frame, design, outcome and measure, signal used (for audit studies), occupation or sector, and funding. Two coders extract data independently on a sample and the coded data and code are shared with the final report.

Ethics and responsible language

Migration is a sensitive topic. Reviews use published aggregate data, avoid identifying individuals or communities, use accurate and respectful terms and make clear that findings describe groups and not individuals. We take no position on immigration policy.

How we support research projects in this area

Boundaries of this service

A migration and demography synthesis describes published evidence on populations and movement. It does not provide immigration or legal advice, assess individual cases, forecast populations for governments or take policy positions. Definitions and data sources differ by country, and some groups are poorly covered, so results should be read with those limits in mind.

Frequently asked questions