Why was the cohort set up?
Population ageing and longevity - as observed in Germany - is not simply a phenomenon
of an increasing number of old people: it also involves a range of qualitative and
structural changes affecting older people. For this reason the German Federal Government,
aiming to improve the quality of its monitoring efforts on older people in Germany,
launched the German Ageing Survey (DEAS) in the mid 1990s, under the auspices of the
Federal Ministry for Family Affairs and Senior Citizens (BMFuS, now German Federal
Ministry for Family Affairs, Senior Citizens, Women, and Youth, BMFSFJ). In 1996,
the first wave of the survey was conducted by two collaborating research groups: the
Research Group on Ageing and the Life Course at the Freie Universität in Berlin; and
the Research Group on Psycho-Gerontology at the University of Nijmegen in The Netherlands.
From the year 2002 onwards, the German Centre of Gerontology in Berlin (DZA) has been
responsible for the conduct and ongoing development of the study. The DEAS is funded
by the BMFSFJ. Fieldwork for all waves (1996–2014) is carried out by the Bonn-based
Institute for Applied Social Sciences (infas).
The primary goal of the project is to provide a representative national database containing
information describing the living conditions of the country’s middle-aged and older
population and to study diversity within the older section of the population, the
process of ageing as it affects individuals and processes of social change as they
relate to old age and ageing. For this purpose, a cohort-sequential design has been
set up combining large cross-sectional samples with longitudinal samples. The design
of the DEAS permits three different perspectives for analyses: (i) analysis of social
change; (ii) analysis of intra-individual change; and (iii) analysis of historical
changes affecting individual ageing trajectories.
Who is in the cohort?
Cross-sectional samples (baseline samples) are drawn up every 6 years (in 1996, 2002,
2008 and 2014 so far). Each baseline sample is followed over time, so that longitudinal
data were collected in the years 2002, 2008, 2011 and 2014 (see Figure 1). The next
panel data collection will take place in 2017 and it is planned to draw up the next
baseline sample in 2020.
Figure 1.
DEAS sample design.
The DEAS baseline samples are nationally representative for adults aged from 40 to
85 years. The samples are based on a two-stage sampling methodology: Firstly, a random
sample of 290 municipalities was drawn in 1996 from the total of 12 000 municipalities
that existed in Germany at that time. In each baseline year, the local population
registries of these 290 municipalities provided the basis used to sample the population
of people living in the community in private households, aged between 40 and 85 years.
All baseline samples have been disproportionally stratified into three age groups
(40–54, 55–69 and 70–85 years), gender and region (East and West Germany). The oldest
age group, along with the group of men and of East Germans, were oversampled to ensure
that there would be a reasonable number of participants in pertinent demographic subgroups,
such as men in old age living in East Germany, and for follow-up. Cross-sectional
weights are computed for every baseline year. They are used to adjust for the disproportional
stratification against the population distribution in each relevant year, which is
obtained from the national micro census data (obtained from the Federal Statistical
Office, based in Wiesbaden).
The target population in 1996 was defined as community-dwelling German citizens (n = 4838).
In 2002, a sample of German citizens (n = 3084) was drawn up, along with a separate
sample of non-German citizens residing in Germany (n = 586). Since 2008, however,
a single sample containing both German and non-German citizens has been drawn up from
the community-dwelling population in Germany (2008: n = 6205; 2014: n = 6002).
The response rates for the baseline samples are defined as the proportion of respondents
with whom valid interviews were conducted against the gross sample of eligible individuals.
DEAS response rates are low as compared with other European surveys on ageing,
1
but are similar to other surveys conducted in Germany. Over recent years one can see
a trend of decreasing participation rates in surveys all over the Western world, but
this phenomenon is most pronounced in Germany.
2
The DEAS aimed to mitigate this decrease in participation rates by, for example, increasing
the incentives for respondents: €10 have been paid to participants since 2008. Despite
this, however, response rates for baseline samples decreased from 50.3% to 27.1% between
1996 and 2014 (see Table 1).
Table 1.
DEAS baseline samples 1996, 2002, 2008 and 2014
Survey years
1996
2002
2008
2014
Germans
Germans
Non-Germans
Germans and non-Germans
Germans and non-Germans
Sample sizes (n
)
Gross sample of eligible people
a
9613
8164
2343
17 366
22 139
Respondents: valid face-to-face interviews
4838
3084
586
6205
6002
Respondents: additional questionnaires
4034
2787
484
4442
4295
Rates (%)
Response rate
b
50.3
37.8
25.0
35.7
27.1
Additional questionnaires
c
83.4
90.4
82.6
71.6
71.6
Other
Birth cohorts
1911–56
1917–62
1917–62
1923–68
1929–74
Age at interview
40–85
40–85
40–85
40–85
40–85
Average duration of face-to-face interview (in minutes)
67
82
82
83
100
aGross sample of municipal registries excluding non-eligible persons (those living
in an institutional setting such as a nursing home and persons who do not speak German.
bValid interviews as a proportion of the gross sample of eligible people.
cNumber of questionnaires filled in as a proportion of valid face-to-face interviews.
Sample selectivity can be observed, as participation rates tend to be lower in large
cities, among women and in both the middle-aged (40 to 54 years) and oldest age groups
(70 to 85 years). Such selectivity effects are, however, small and the distribution
of such socio-demographic characteristics as family composition, marital status, household
size, employment status and education in the weighted baseline samples is very close
to the distribution within the population of Germany.
3
Up to 2014, a total of 20 715 individuals aged between 40 and 85 years at first interview
have participated in the DEAS.
How often have they been followed up?
Baseline participants who gave written consent were re-contacted for further waves
of data collection in 2002, 2008, 2011 and 2014. To enhance the quality of the longitudinal
sample,
4
people willing to participate in the panel are generally allowed to miss out on one
or more panel waves without being excluded from the address pool.
Panel attrition is high in the first re-interview but attenuates in subsequent follow-ups
(see Table 2), a phenomenon that is familiar to other panel studies.
5
To reduce attrition, the interval between panel waves was reduced from 6 to 3 years
from 2008 on, and efforts were intensified to update the address pool and to increase
participants’ compliance (e.g. by sending greeting cards and booklets along with study
results). As a consequence, the retention rate - defined as valid re-interviews as
a proportion of the number of valid interviews in the baseline year - has increased
(see Table 2). The most obvious sign of improvement is the fact that the retention
rate based on the 2008 baseline sample was considerably higher (41.4%) in 2014 than
the 2002 figure based on the 1996 baseline sample (31.5%). Up to 2014, a total of
6622 individuals had participated at least twice. Vital statistics were updated in
every follow-up for all respondents who gave their written consent to further contact
in relation to the study. For those participants who were discovered to have died
between waves, the date of death was obtained either from the relevant registration
office or from relatives.
Table 2.
DEAS longitudinal samples 2002, 2008, 2011 and 2014
Survey year
2002
2008
2008
2011
2011
2011
2014
2014
2014
Baseline year
1996
1996
2002
1996
2002
2008
1996
2002
2008
Time span in years
6
12
6
15
9
3
18
12
6
Sample sizes (n)
Valid re-interviews
1524
991
1000
1039
957
2858
887
866
2569
Additional questionnaires
1437
818
829
876
791
2338
749
729
2179
Rates (%)
Retention rate
a
31.5
20.5
32.4
21.5
31.0
46.1
18.3
28.1
41.4
Valid questionnaires
b
94.3
82.5
82.9
84.3
82.7
81.8
84.4
84.2
84.8
Other
Birth cohorts
c
1911–56
1911–56
1917–62
1911–56
1917–62
1923–68
1911–56
1917–62
1923–68
Age at interview (years)
46–91
52–96
46–89
55–98
49–92
43–88
58–96
52–95
46–91
Note: aValid interviews in the panel year as a proportion of valid interviews in baseline
wave.
bPanel questionnaires completed as a proportion of valid face-to-face panel interviews.
cPossible range of birth cohorts.
As is the practice with other panel studies, attrition rates are graded in relation
to a variety of demographic and socioeconomic characteristics. In the DEAS, we observe
that panel participants tend to be younger, healthier and better educated, and to
have larger incomes and larger informal networks than respondents who drop out. However,
it could be seen that these selectivity effects diminished in size after the interval
between waves was reduced from 2008 on.
3
To allow adjustments to be made for differential non-response in the successive panel
waves, longitudinal weights are available.
What has been measured?
Face-to-face interviews are conducted in each wave. These interviews usually take
place in the respondents’ homes and are conducted by professional interviewers using
a standardized questionnaire. In 1996 and 2002 the data were collected using paper-and-pencil
interviews (PAPI). From 2008 on this approach was replaced by computer-assisted personal
interviews (CAPI). In all waves, respondents are asked to fill out an additional written
questionnaire (either immediately or a few days after the oral interview). The language
of both the interview and questionnaire is exclusively German. No proxy interviews
are permitted. However, where a respondent requests, he or she may fill out the questionnaire
with the help of the interviewer. In addition, those who do not have a sufficient
knowledge of the German language may be supported by family members or other people
present.
A very broad range of topics is covered in the DEAS. For reasons of comparability,
many measures remain unchanged from wave to wave. In some cases, however, measures
have had to be modified, deleted, or added - as required by current political or scientific
discussion or in response to new findings in survey methodology and research on ageing.
The interview mainly seeks to record data on household composition, family relationships,
occupational status, personal network and health. The written questionnaire mainly
deals with psychological measures and questions on such sensitive issues as religious
or political affiliation, attitudes, income and sexual orientation. Two different
objective tests are conducted to provide an indication of the respondent’s cognitive
and physical capacity: a digit-symbol test to assess perceptual speed
6
and a lung functioning test using a spirometer.
7
In addition to the data obtained directly during the survey, interviewer ratings and
structural context data are available in order to describe respondent’s home and neighbourhood.
Table 3 shows the content of the DEAS data and gives examples of measures and indicators.
Most - but not all - measures and indicators listed are available for every wave of
the study.
Table 3.
DEAS data 1996–2014
Topics
Examples
(Socio-) Demographics
Age, gender, household composition, parents, siblings, education, marital status,
citizenship
Employment
Employment status, job details (ISCO, working hours, job quality), retirement
Activities
Leisure activities, voluntary work, religion
Family and social network
Numbers and demographics for children and grandchildren, quality of intergenerational
relationships, intimate partner, kin relations, social network
Support
Provision and reception of informal (emotional, cognitive, financial and practical)
help and care
Health
List of illnesses, visits to the doctor, subjective health, pain, sleep, functional
health, health-related behaviour (smoking, physical activity, health care, medication)
Subjective well-being
Life satisfaction, emotional well-being, depressive symptoms, loneliness
Psychological resources
Self-efficacy, coping strategies
Housing
Characteristics of private household (owner/tenant, size, costs), characteristics
related to retirement home, residential environment (infrastructure, shopping facilities,
services for seniors)
Finances
Income (sources, amount, personal and household income), assets, debts
Attitudes, norms, values, stereotypes
Positive and negative self-perceptions of ageing, religiosity, political orientation,
attitudes toward social security
Objective measurements (tests)
Digit-symbol test (since 2002)
Lung function test (since 2008)
Context data
Structural data at district level (NUTS-3) (e.g., unemployment rate, average household
income, population density). Structural data for place of residence (e.g., availability
of doctors, public transport). Interviewers’ rating to describe respondent’s home
and neighbourhood
NUTS, nomenclature of territorial units for statistics.
What has it found? Key findings and publications
DEAS data are used on a regular basis for social reports to the BMFSFJ on ageing and
old age in Germany. In addition, researchers all over the world use them in their
scientific work. A full list of known publications based on DEAS data is provided
on the DZA Research Data Centre’s website [www.dza.de/en/fdz/german-ageing-survey/publications.html].
To follow, we provide a selection of findings from the core topics of the DEAS: health
and well-being, work and income, and family and social relationships.
Health and well-being
How healthy older people are, how their state of health evolves over time and what
can be done to promote good health are major topics dealt with in research on ageing
and old age. DEAS data provide the opportunity to investigate a number of indicators
of health and well-being, as well as to analyse longitudinal trajectories, age and
cohort differences and predictors of health and mortality. Findings show that later-born
cohorts are healthier than earlier ones. Since 2008, this trend has only been detected
for persons aged 65 and older, whereas it has actually reversed for those aged 40
to 64 in relation to functional health and depressive symptoms.
8
In relation to self-rated health, the study has shown that what people include when
they rate their state of health will depend both on their age and on differences between
birth cohorts.
9
Health and well-being are dependent on a wide range of personal and contextual resources
and are also prone to social inequalities. DEAS data provide evidence of socioeconomic
differences in health
10
and in risk factors for bad health, as well as in mortality predictions. In an effort
to search for possible mechanisms by which socioeconomic status may be connected with
health in later life, the role of negative emotion was analysed using DEAS data. It
was found that negative emotion predicted health changes only in individuals with
lower educational levels.
11
Several studies have pointed to the role of regional resources (gross domestic product
per capita, supply of primary care) in health and well-being outcomes in later life.
12
In more prosperous districts, for example, more people aged 40 and older pursue regular
exercise
13
than in less prosperous areas. It has been shown that positive affect predicts mortality
in older adults even after controlling for self-rated health and physical activity.
14
Social and psychological resources for health and well-being have also been studied
using DEAS data. There is evidence of a differential in the instrumental support provided
by kin (with kinship relationships exposing a negative effect) as against non-kin
(with non-kin relationships showing a positive effect) for the well-being of older
adults.
15
Moreover, a number of studies have provided evidence in favour of the impact of self-perceptions
of ageing on health and healthy behaviour - even in the face of a serious health event.
16–19
Work, volunteering and income
To the extensive literature on older workers, the DEAS has contributed two new lines
of results. First, although the rate of employment of older workers has increased
since the mid 1990s, a majority continues to retire before the standard retirement
age and individual plans and expectations in relation to retirement tend to lag behind
changes in the regulations set out for retirement. Despite this, a considerable minority
of people in Germany plan to retire before age 65 or even before age 60.
20
,
21
Second, remaining in employment after completing the transition into retirement has
become increasingly common over the past few years. Using data from the DEAS and the
English Longitudinal Study of Ageing (ELSA), it was shown that fewer people are forced
to pursue post-retirement employment in Germany than in England, mainly for institutional
and/or structural reasons.
22
,
23
Besides the occupational, financial and health factors that affect post-retirement
employment,
24
psychological experiences of ageing also influence the decision to continue working
after retirement.
25
Comparing the 1996, 2002 and 2008 waves of the DEAS study, the financial situation
of the majority of people aged 40 years and older in Germany appears to be quite satisfactory.
However, income growth has been greater for individuals gainfully employed as compared
with people in retirement who are dependent entirely on their pension income.
26
In addition, an increase was detected in people reporting low living standards and
lack of financial resources between 1996 and 2008. To be more specific, widows who
spent many years as housewives or in unemployment during their working life are at
a higher risk of living in poverty in old age.
27
Volunteering is more widespread among the highly educated than among those DEAS respondents
with low educational status.
28
In addition, there are pronounced regional differences in rates of volunteering and
social participation. In economically stronger districts the rate of volunteering
is higher than in economically weaker ones, even after one has controlled for social
inequality at an individual level.
29
Between 1996 and 2014, not only did the rates of volunteering in organizations increase
in general, but also the rates for people volunteering in organizations with a particular
focus on older people.
30
From a longitudinal perspective, it has been shown that volunteering affects subjective
well-being differentially in the second half of life. Whereas volunteering affects
well-being directly for people aged 45–84 years, it is only in the age groups around
retirement (55–74 years) that volunteering turns out to be beneficial for subjective
well-being not just by its direct effects, but also indirectly via its effects on
self-efficacy.
31
Family, social relationships and support
Family relationships are strongly linked with physical and mental well-being. In old
age in particular, relationships with partners and children are an important asset
in helping to cope with everyday life and preserving autonomy. In the Western world,
the range of private living arrangements have become more diverse and to some extent
more fragile over the past decades. DEAS data demonstrate that ever fewer older parents
are living close to their adult children. Nevertheless, the quality of relationships
remains very high in terms of emotional closeness, frequency of contact and mutual
assistance between the generations.
32
,
33
Inheritances from parents to children are common but such bequests have been found
to be positively related to income position, thus implying a tendency to increase
social inequality in the subsequent generation.
34
Migrant families in Germany, apart from exhibiting slightly tighter relationships,
show patterns of parent-child relationships quite similar to those of non-migrant
families. There are, however, differences from migrant group to migrant group and
such differences depend heavily, in their turn, on differences in terms of educational
and financial resources.
35
Due to the rich information about grandparenthood, DEAS-based studies have revealed
that it is both being delayed and becoming less likely.
36
At the same time, the data show that the grandparent role is highly valued
37
and that the quality of relationships with grandchildren is positively related to
the grandparents’ level of life satisfaction and emotional well-being.
38
In general, relationships between grandparents and their older grandchildren are emotionally
close, contact is frequent and financial transfers from grandparents to grandchildren
have become more common.
39
However, the middle generation - the adult children - play a crucial role in these
sorts of relationship: they are decisive in enabling or preventing contact between
grandparent and grandchildren.
40
Due to contemporary high rates of childlessness, a growing attention is being paid
to elderly people who remained childless. Recent DEAS findings suggest that childless
people are in a position to substitute missing offspring with non-relatives and are
able in old age to rely on such ties, which appear to be very efficient in providing
support.
41
,
42
In general, friends seem to be gaining in importance as providers of social integration
and support to older people.
43
Health and life satisfaction in the second half of life depend, among other things,
on the social engagement and emotional support received from within one’s own social
network.
44
However, such effects tend to vary depending on the type of relationship. In old age,
for example, social activities with friends tend to promote positive feelings whereas
activities with family members are more likely to produce ambivalence.
45
What are the main strengths and weaknesses?
The main strengths of the DEAS are: (i) its cohort-sequential design, which equips
its users to analyse both societal trends and individual trajectories (embedded within
societal trends) and to disentangle age effects from cohort effects; (ii) its broad
range of topics, which cover major aspects of the living situation of individuals
in the second half of life, including their psychological resources; and (iii) the
large and representative samples of the community-dwelling population in Germany.
There are three main weaknesses of the DEAS: (i) most of the information gathered
is based on self-reported data; (ii) the language of the interview is exclusively
German; and (iii) there is a high rate of attrition between the first and second interviews.
Can I get hold of the data? Where can I find out more?
Data from completed DEAS waves are put through a process of editing and anonymization.
They are available to the scientific community free of charge through the DZA’s Research
Data Centre (FDZ). The questionnaires and a variety of data-documentations are published
online, but for reasons of data protection it will be necessary to sign a data distribution
contract before obtaining download access to the data. Access to the data is permitted
exclusively for use in a scientific, non-profit context. An application form can be
obtained from FDZ-DZA website [www.dza.de/en/fdz/german-ageing-survey.html] along
with further details on the data set. Context data may only be used on site at the
DZA, since the survey data may become sensitive when used in combination with regional
context data. We recommend the use of the GeroStat information system [www.gerostat.de/en/index.html]
to obtain a general overview of the content and to get an initial glimpse of the distributions
of major indicators differentiated by year, age group, gender and region.
The DEAS in a nutshell
The German Ageing Survey (DEAS) is a nationwide, longitudinal cohort study of the
community-dwelling population in Germany aged 40 years and older. It provides a unique
database for interdisciplinary research on changes and diversity in the living conditions
of the middle-aged and older population as well as on the multifaceted processes of
individual ageing.
The DEAS uses a cohort-sequential design. The initial wave of data collection for
people aged 40 to 85 years took place in 1996 (n = 4838). New baseline samples were
drawn up in 2002 (n = 3670), 2008 (n = 6205) and 2014 (n = 6002). Panel assessments
took place in 2002, 2008, 2011 and 2014. Retention rates in the first re-interview
range from 31.5% (baseline sample 1996) to 46.1% (baseline sample 2008). Currently,
more than 33 000 interviews have been conducted with 20 715 individuals aged 40 and
older.
The DEAS covers a wide range of topics. The data obtained provide information on socioeconomic
and demographic characteristics, household composition, housing, family structure,
social networks, psychological resources, attitudes and physical and mental health.
Funding
The German Ageing Survey (DEAS) is funded by the German Federal Ministry for Family
Affairs, Senior Citizens, Women and Youth (BMFSFJ; grant number: 301-6083-05/003*2).
Conflict of interest: The authors declare no competing interests.