Symposium: Advances in Survey Methods for Cohort Studies 1
Consent to Data Linkage in Longitudinal Studies: Patterns, Predictors and Change Over Time (Konstantinos Tsigridis, UCL Centre for Longitudinal Studies)
Linking survey data with administrative records enables richer research while reducing respondent burden, but often requires explicit consent which can introduce the risk of consent bias if consenters differ from non-consenters.
Evidence on how best to request and maintain consent in longitudinal surveys remains limited. Predictors of consent are inconsistent across studies and over time, and willingness to consent is not fixed. Participants who refuse at one wave may consent later, and consent rates may vary by mode, with lower rates often observed in web surveys.
This study uses data from four UK cohort studies – the National Child Development Study (NCDS), 1970 British Cohort Study (BCS70), Next Steps and the Millennium Cohort Study – to examine consent to data linkage over time.
We assess overall consent rates, transitions from non-consent to consent, predictors of consent, and differences by mode of data collection.
Evaluating the impact of short online follow-ups of non-respondents in UK cohort studies (Alessandra Gaia, UCL Centre for Longitudinal Studies)
Attrition in longitudinal studies threatens representativeness, reduces analytical power, and undermines the value of earlier data collection. As response rates decline, identifying effective strategies to improve participation is critical. One approach is to reduce response burden by offering non-respondents a shorter follow-up questionnaire towards the end of fieldwork.
This paper evaluates the effectiveness of short non-response surveys across several major UK cohort studies. We assess the impact of inviting participants who have not completed the main survey to complete a shorter questionnaire by examining changes in response rates and representativeness.
Using data from multiple cohorts, we explore variation in effectiveness across study contexts, longer-term effects on subsequent participation and the role of incentives. The findings provide new evidence on the value of non-response follow-up surveys and their implications for survey design and managing response burden.
Use of AI in Questionnaire Development in Generation New Era (Julia Pye, Ipsos)
Generation New Era (GNE), the first UK-wide birth cohort in 25 years, aims to be inclusive, overrepresenting families in Scotland, Wales, Northern Ireland, low-income areas and ethnic minority groups, and recruiting both mothers and fathers.
Questionnaire specifications were complex and required rapid, thorough review before programming. AI supported this process by checking readability, consistency and routing logic at speed. AI personas were also developed to simulate how different parents might interpret and answer questions, assessing inclusivity for key groups, including parents of disabled children, multigenerational families, and respondents across all four UK nations. These personas tested questions, highlighting potential misunderstandings and problematic wording or assumptions.
These checks complemented internal accessibility reviews and consultation with parents with lived experience. We discuss the rationale for using AI in questionnaire development, and evaluate its strengths and limitations, including comparison with findings from real parents.
Using an Innovative Smartphone App (BabySteps) in the Children of the 2020s Birth Cohort Study (Marialivia Bernardi, Clinical, Educational & Health Psychology, UCL)
Smartphone apps can complement longitudinal cohort studies by providing high-frequency data collection between main survey waves. This paper presents methodological contributions from BabySteps, a smartphone app implemented within the Children of the 2020s (COT20s) birth cohort, which follows 8,628 families with children born in England in autumn 2021 across five survey waves. The app was introduced at baseline, with 74% of primary caregivers registering and approximately 50-30% completing app-based research activities each month. First, we examine participation patterns and develop weighting approaches to address non-response and attrition. Second, we estimate propensity scores to assess whether app participation was associated with response in subsequent main survey waves. Third, we construct developmental trajectories from repeated monthly measures. These methods provide a framework for integrating longitudinal app data into large-scale cohort studies.
