Cross-national comparisons based on the Generations and Gender Survey (GGS) show large differences across European countries in the prevalence of short-term fertility intentions and in the share of intentions that are realised as births — but the countries fielded the underlying survey questions with different wording, filters, ordering and omissions, so measurement artefacts and substantive country-context effects are observationally confounded. The open problem is to quantify how much of the observed cross-national variation in intention prevalence and intention-to-birth realisation is attributable to questionnaire implementation heterogeneity rather than to real country differences, to use that decomposition to settle the puzzle left by Brzozowska and Beaujouan (2021) of whether the elevated realisation of 'definitely intending' French women reflects a substantive French advantage or the narrower selection produced by the French 'trying now' filter question, and to address their accompanying unresolved observation that, among women aged 35 and over who want or are trying to have a child, the share who actually give birth remains far below the estimated biological fecundity at that age. The harmonised GGS Round 2 (fielded from 2020 with a centrally coded questionnaire including the behaviourally specific 'trying to get pregnant' item) is still accruing the required panel follow-up waves and no published analysis has performed the decomposition.

Background

Demographers study childbearing behaviour partly through survey questions on fertility intentions: whether respondents want or intend to have a (another) child, over what time horizon, and with what degree of certainty (e.g. 'definitely yes' versus 'probably yes'). The intention that is most predictive of subsequent behaviour is the short-term intention, typically 'Do you intend to have a/another child within the next three years?'. The proportion of people with a given intention who actually have a child within that horizon is called the realisation rate, and comparing realisation rates across countries is a standard way to assess how institutional, economic and cultural contexts help or hinder people in carrying out their reproductive plans.

In Europe, the main data source for such comparisons is the Generations and Gender Survey (GGS), a cross-national panel survey coordinated by the Generations and Gender Programme (GGP), in which the same respondents are re-interviewed about three years apart. In the first round (GGS-I, fielded from the early 2000s in around 20 countries), participating national teams were allowed to modify the core questionnaire, and the fertility-intention items were particularly affected. Brzozowska and Beaujouan (2021) documented the consequences in Austria, France, Italy and Poland — four countries whose main three-year-intention question was almost identical; the other ten countries that fielded both GGS-I waves were excluded from that comparison because their short-term intention items departed from the core questionnaire (no time frame in the question, binary answers) or presented data problems such as small samples, large panel attrition or inaccurate childbearing histories. Their analysis covered women and men — panel respondents aged between 18 and 45 who reported having a partner (cohabiting or not) at wave 1, with post-stratification weights at all stages. The key differences concerned the preceding question in the fecundity section of the questionnaire: Austria and Poland used the standard 'Do you yourself want to have a/another baby now?'; Italy did not ask it at all; and France replaced it with the behaviourally specific 'Are you trying to have a baby now?' and additionally used it as a filter question, so that a positive answer automatically classified the respondent as 'definitely intending' a child within three years, and a strongly negative answer as 'definitely not' intending. France also changed the wording of the longer-horizon question and reordered the items.

These implementation differences turned out to matter. Among French respondents who said they were trying to have a baby, over 90% were not using contraception (a so-called proceptive behaviour, i.e. behaviour actively directed at conception), whereas only about half of Austrian and Polish respondents who said they 'wanted a baby now' were not using contraception — showing that the 'want now' item measures a desire rather than an active attempt, and that the two items select very different people into the intention categories. Consequently, the cross-country realisation comparisons are contaminated: French women below age 35 with definite three-year intentions realised them substantially more often than their peers elsewhere, but this may reflect either genuinely higher fertility, better access to assisted reproduction, or more favourable norms in France (a substantive country-context explanation), or simply that the French filter selected a narrower, more committed subgroup into the 'definitely intending' category (a measurement explanation). The authors could not decide between these explanations. The unresolved contrast concerns women below age 35; for men the source reports the opposite pattern at late reproductive ages, with French men aged 35 and over realising definite short-term intentions more often than their peers in Austria and Italy.

A second, partly separable puzzle concerns late reproductive ages. Among women aged 35 and over who reported wanting (Austria, Poland) or trying (France) to have a child, the share who actually gave birth within the panel interval remained far below the estimated biological fecundity at that age (Leridon 2008) — even in France, where over nine in ten women 'trying now' were not using contraception. Leridon (2008) estimates the proportion of couples permanently sterile, that is unable to conceive, by age of the woman, on the basis of the risk of conceiving independently of the fate of the pregnancy; unless medical assistance is used, that age is the upper bound of the fecund period. It is therefore not an age-specific schedule of probabilities of conception and birth over a panel interval, and Brzozowska and Beaujouan (2021) state no conversion between the two quantities. The explanations they leave open — more fecundity problems among older definite intenders, adaptation of intentions as women become more aware of possible age-related sterility, and the prevailing social norms regarding age limits for having children, to which adjacent work on pregnancy attempts adds the prevalence of partnerships and access to assisted reproduction (Beaujouan and Singh 2026) — could not be separated with the available data.

The methodological core of the problem predates the paper: Gauthier, Cabaço and Emery (2018) described the limits of post-hoc harmonisation of GGS-I, and survey-methodology evidence has long shown that answers to reproductive-intention questions are sensitive to wording, sequencing and answer options. What no study has done is quantify the bias that this measurement heterogeneity induces in cross-national estimates of intention prevalence and intention-to-behaviour links, and thereby decompose observed cross-country differences into a measurement component and a substantive country-context component.

The data infrastructure for a definitive answer is now being built. GGS Round 2 (GGS-II, fielded from 2020 onwards) uses a centrally coded questionnaire with standardised national data collection (Gauthier et al. 2025), including the behaviourally specific item 'Are you or your current partner trying to get pregnant?' administered in a common format across participating countries; the fertility histories of the first web-based wave have been validated against external population benchmarks (Leocádio et al. 2023). However, measuring realisation requires follow-up interviews after the three-year horizon, and the second panel waves of GGS-II have only begun to be released in 2025-2026. Comparative studies of realisation continue to rely on GGS-I panel data, whose measurement heterogeneity is precisely the confound in question (e.g. Spéder and Bálint 2024, who study macro-level determinants of realisation across eleven countries without disentangling the measurement component). Using GGS-II wave-1 data from 14 countries, Beaujouan and Singh (2026) show that from around age 40 the share of women trying to conceive exceeds the share reporting definite short-term intentions, that long-running conception attempts and perceived infertility accumulate with age, and that contradictory answers (trying while no longer clearly intending) increase with age — but they explicitly leave the mechanisms open and note that even the harmonised 'trying' item may be interpreted differently across cultural contexts.

Problem Statement

How much of the observed cross-national variation in the prevalence of short-term fertility intentions and in their realisation as births, as measured in the Generations and Gender Surveys, is attributable to measurement artefacts — differences in item wording, filter logic, question order and item omission across national questionnaires — rather than to substantive country-context differences, and what do the substantive residual differences consist of? The problem is to: (1) quantify, with uncertainty bounds, the bias that the documented Round 1 implementation heterogeneity (the French 'trying now' wording and automatic filter, the Austrian/Polish 'want now' item, the Italian omission, reordering and filter differences) induces in cross-national estimates of intention prevalence and of the intention-to-birth link; (2) use this quantification to establish what substantive residual differences remain once that bias is removed, and in particular to determine whether the higher realisation rates of French women below age 35 classified as 'definitely intending' a child within three years reflect substantive French country context (higher baseline fertility, differential access to medical assistance, more permissive norms on age and childbearing) or the narrower selection produced by the French questionnaire implementation; and (3) address the source's accompanying unresolved observation that, among women aged 35 and over who report wanting or trying to have a child, the share who actually give birth within the panel interval remains far below the estimated biological fecundity at that age (Leridon 2008), for which the source leaves its candidate explanations open. Since Leridon (2008) estimates proportions of couples permanently sterile by age of the woman rather than a probability of conception and birth over a panel interval, any quantification of that shortfall must state the conversion it uses — the horizon, the reference population (couples, or non-sterile or actively trying women) and the treatment of medically assisted conceptions, the bound on the fecund period applying only where medical assistance is not used.

References

  1. Zuzanna Brzozowska and Eva Beaujouan (2021), 'Assessing Short-Term Fertility Intentions and Their Realisation Using the Generations and Gender Survey: Pitfalls and Challenges', European Journal of Population 37(2), 405-416. DOI: 10.1007/s10680-020-09573-x. https://doi.org/10.1007/s10680-020-09573-x (source paper)
  2. Eva Beaujouan and Shalini Singh (2026), 'Fertility intentions and pregnancy attempts across reproductive ages', Vienna Yearbook of Population Research 24, 1-33, online first 5 August 2026. DOI: 10.1553/p-535f-56a4. https://doi.org/10.1553/p-535f-56a4
  3. Anne H. Gauthier, Siyang Kong, Olga Grünwald, Martin Bujard, Arianna Caporali, Vytenis Juozas Deimantas, Tom Emery, Wojciech Jablonski, Judith C. Koops, Arieke Rijken and Almut Schumann (2025), 'The Generations and Gender Survey: a cross-national longitudinal resource', European Sociological Review 41(6), 983-1001. DOI: 10.1093/esr/jcaf005. https://doi.org/10.1093/esr/jcaf005
  4. Zsolt Spéder and Lajos Bálint (2024), 'Realization of Short-Term Fertility Intentions in a Comparative Perspective: Which Macro-Level Conditions Matter?', Population Research and Policy Review 43(5), article 72. DOI: 10.1007/s11113-024-09913-3. https://doi.org/10.1007/s11113-024-09913-3
  5. Anne H. Gauthier, Susana Laia Farinha Cabaço and Tom Emery (2018), 'Generations and Gender Survey study profile', Longitudinal and Life Course Studies 9(4), 456-465. DOI: 10.14301/llcs.v9i4.500. https://doi.org/10.14301/llcs.v9i4.500
  6. Victor Antunes Leocádio, Anne Gauthier, Monika Mynarska and Rafael Costa (2023), 'The quality of fertility data in the web-based Generations and Gender Survey', Demographic Research 49(3), 31-46. DOI: 10.4054/demres.2023.49.3. https://doi.org/10.4054/demres.2023.49.3
  7. Henri Leridon (2008), 'A new estimate of permanent sterility by age: Sterility defined as the inability to conceive', Population Studies 62(1), 15-24. DOI: 10.1080/00324720701804207. https://doi.org/10.1080/00324720701804207
  8. Zuzanna Brzozowska, Isabella Buber-Ennser and Bernhard Riederer (2021), 'Didn't Plan One but got One: Unintended and sooner-than-intended Parents in the East and the West of Europe', European Journal of Population 37(3), 727-767. DOI: 10.1007/s10680-021-09584-2. https://doi.org/10.1007/s10680-021-09584-2
  9. Ross Barker and Isabella Buber-Ennser (2024), 'Uncertainty and flexibility of fertility intentions', Advances in Life Course Research 61, article 100618. DOI: 10.1016/j.alcr.2024.100618. https://doi.org/10.1016/j.alcr.2024.100618