All time references are in CEST
Surveying the Elderly: Methodological Issues |
|
| Session Organisers | Ms Nicole Hameister (German Centre of Gerontology) Stefan Stuth (German Centre of Gerontology) |
| Time | Tuesday 18 July, 09:00 - 10:30 |
| Room | |
European societies have been ageing for many decades now and will continue to do so. This dynamic holds challenges and chances, but in order to understand these changes, surveys are essential to assess the life situations of the older population. National and international surveys that target and cover the ageing populations can deliver this important knowledge but, at the same time, face a range of very specific methodological issues. These include, but are not limited to:
• fieldwork management, contact strategies and participant engagement for older target group
• age-related coverage errors of different sampling frames
• nonresponse bias and measurement error due to declining cognitive and physical (listening, reading) abilities
• restricted use of modes (online, mobile) and mixed-mode surveys for older people
• age-related interview challenges and the need for specific interviewer skills and training
• the use of proxy interviews or 'triad interviews' and consequences for data quality
• conversion of gate keepers who refuse / hesitate to let an older person be interviewed
• tracking /tracing of panel members, e.g., for people moving to institutions or the deceased
• data linkage: developments and new sources, for example to public health / death records
• age differences in the quality of data in face-to-face or self-administered interview modes
• the question of “vulnerability” of older respondents and ethical issues in general
• new survey tools/new data resources to improve data collection and quality among older people.
We invite researchers to address these or related issues of surveying the older population. We also encourage presentations on the potential adjustments necessary or beneficial to data collection.
Keywords: ageing, old, elderly, challenges, data quality,
Dr Margherita Silan (Department of Statistical Sciences, University of Padova) - Presenting Author
Miss Sophie Grace Parolin (Department of Statistical Sciences, University of Padova)
Miss Giorgia Bernardinis (Department of Statistical Sciences, University of Padova)
Dr Manuela Scioni (Department of Statistical Sciences, University of Padova)
The aim of this work is to explore the use of the Internet and social networks among people over 60 years of age, with a particular focus on Facebook, using data collected through a survey conducted through advertisements on the Meta platform. A fundamental methodological aspect of this work is the weighting of the data collected through the questionnaire, necessary to address the selection bias that often emerges in surveys conducted on platforms like Facebook. Three weighting methods were applied: post-stratification, raking, and quasi-randomization, incorporating both sociodemographic variables and online skill indicators. The choice of the most effective combination of method and variables to reduce distortion has been guided by the observation of three benchmark variables in the survey sample and in a nationally representative survey. These variables were collected using questions formulated in the same way in both the Aspects of Daily Life survey and the conducted survey and they regard the use of online tools to perform operations in the field of public administration. Our analysis highlighted the importance of including online skill variables in weighting to achieve a more accurate representation and reduce sampling bias. Furthermore, at the end of the questionnaire, respondents were asked to select an image that represented their relationship with social media platforms, using the same figures included in the ad that promoted the survey. Images were created with artificial intelligence to represent different aspects of relationships with social media platforms. The distribution of respondents according to the clicked images and the preferred ones is different, showing a possible distortion included by the algorithm to share advertisements. Visual representations of social media relationships showed that most respondents view these platforms to connect with younger family members.
Dr Michael Weinhardt (German Center of Gerontology) - Presenting Author
Dr Mareike Bünning (German Center of Gerontology)
Should individuals be contacted again for new waves of a panel study if they have not participated over several waves? This question is discussed on the basis of empirical data from the German Ageing Survey (DEAS). The DEAS has been conducted initially every six years since 1996 and every three years since 2008; the most recent wave was conducted in 2023. The survey is based on probability samples of people aged 40 to 85 drawn from municipal registration offices (1996, 2002, 2008, 2014). It also exhibits a panel component that aims to re-interview all respondents in future waves. So far, all individuals remain in the DEAS sample until they explicitly withdraw panel consent or permanently drop out for other reasons (e.g., permanent illness, moving abroad, death). From a cost-benefit perspective, the question arises to what extent such a strategy should be maintained. On the one hand, experience shows that a small amount of respondents can still be won back for an interview even after many years of panel absence. On the other hand, persons with multiple spells of non-participation are much less likely to participate in future waves and may pose difficulties for the computation of longitudinal weights. Moreover, the question arises what added value such cases with long gaps in the data series offer for longitudinal analyses. In this presentation, we investigate whether contacting repeated non-participants disproportionally increases the workload for interviewers and the survey institute (e.g., based on the number of contact attempts). We especially discuss this question in the context of studies of older people, as they are, e.g., more prone to temporary absences due to illness.
Professor Lea Ellwardt (University of Cologne) - Presenting Author
Social support from personal networks is a crucial predictor of healthy aging. Modules for assessing personal networks are therefore an integral part of many large-scale surveys on aging, such as NSHAP, SHARE, DEAS, and LASA. These modules typically include measures of the direct relationships between the respondent and their network members (ego-alter ties) but often omit relationships between the network members themselves (alter-alter ties). However, a sociometric analysis of key network concepts and dynamics—such as homophily and transitivity—requires data on both ego-alter ties and alter-alter ties.
This study aims to present a research design considered feasible for collecting comprehensive personal network data in a large, population-based random sample. To demonstrate the utility of this type of data, we provide basic descriptive statistics alongside a simple illustration of how two critical social mechanisms, transitivity and gender homophily, can be tested.
We collected primary data from 938 community-dwelling older adults (egos) and their 4,610 network members (alters) within a representative Dutch sample from the Longitudinal Aging Study Amsterdam (LASA) during 2018–2019. Data collection comprised two elements: a personal interview and a self-administered questionnaire sent approximately one month later, which included additional questions about respondents' personal networks. The resulting personal networks capture perceptual data on the relationships between egos and alters, as well as among alters themselves.
We used the R packages egor and ergm.ego to make inferences for exponential-family random graph models based on egocentrically sampled data. The results supported the presence of both gender homophily and transitivity. Notably, excluding transitivity from the model led to an overestimation of other network parameters.
Professor Jolanta Perek-Białas (Jagiellonian University, Kraków, Poland) - Presenting Author
Ms Paulina Skórska (Jagiellonian University, Kraków, Poland)
Ms Milena Maj (Jagiellonian University, Kraków, Poland)
Professor Jan K. Kazak (Wrocław University of Environmental and Life Science)
Dr Jeroen Dikken (The Hague University of Applied Science)
Professor Joost van Hoof (The Hague University of Applied Science)
Doing surveys among the older population (65+) is more and more challenging. Not only COVID-19 but also the protection of personal data and access to the respondents (GDPR) with cases of older persons being cheated by phone are reasons which need to be taken in account. The aim of the presentation is to give practical recommendations on the use of CATI vs CAPI, and how the choice for a certain method could possibly influence results. The survey used probability samples ensuring representativeness with regard to the districts of residence. The sample distribution reflected the demographic spread of the population aged 65 years and over. The study took place in 2023 and was either performed using the CATI or the face-to-face CAPI research method. The surveys were conducted mainly over the phone (57.3% of the questionnaires). Also, persons aged 80 years and over with whom CATI surveys were completed, accounted for 16.7% of the total sample (n = 134). All data collection was done in an anonymised manner. We completed n = 392 interviews in Kraków and n = 409 in Wrocław (total 801), which translates into response rates of 32.7% and 34.1% respectively. After completion of the study, the data were cross-checked by the research group members by listening to recordings of the study for at least randomly selected 10% of the sample. The aim of the study was to validate the AFCCQ scale (Dikken et al, 2020) for both cities (Perek-Białas, et al, 2024). In this presentation there will be shown how the results could differ taking into account the way of contacting older respondents (CATI vs CAPI) and if it has a significant impact on the results of AFCCQ. At the end, we are able to give some practical