Longitudinal Data Resources
What is longitudinal data?
Longitudinal data is collected from the same sample at different points in time. The sample can consist of individuals, households, establishments, and other units of observation and/or analysis. Using longitudinal data is a great way to measure change.
NACDA has longitudinal data organized by series and study, and even dataset within study.
For example, data organized by series means that we have several studies (usually 2-3 or more) that can be used together (and/or were intended to be used together) because they have the same questions across years, or because the studies have the same sample of respondents. Therefore, users can see all of the studies that are intended to be analyzed together by the principal investigator and have the components to do so (such as a consistent ID variable to sort and merge by). This also means that users will often need to download files from each study page in order to merge them, as there may not be a merged file already created/provided.
The SWAN series is an example of multiple waves by study within series, in addition to MIDUS and MIDJA, and NSHAP.
Data that are organized by dataset within study means that a single study was created and all of the waves and/or components of the whole study are downloadable from that same study page. The datasets are clearly meant to be used together, and there should be consistent variables to sort and merge by. Users may still need to download all of the study files or multiple files, however, they will only need to do so from a single study page.
SATSA, SEBAS, and the NLTCS are examples of multiple waves by datasets within a single study.
So what does a merged longitudinal file look like? We do receive some studies in this manner, one example is the American Changing Lives study (ACL). Another example would be the WHO studies on Global AGEing and Adult Health, as there are multiple years included in the datasets for each country.
Study | Start Year | Country | Sample | Age Group | Topic: Cognition | Topic: Biomeatures | Topic: Caregiving | Topic: Physical Health | Topic: Dementia | Topic: Depression |
|---|---|---|---|---|---|---|---|---|---|---|
1986 | U.S. National | Multi-stage | 25+ | X | X | X | X | na | X | |
HRS* | 1992 | U.S. National | Multi-stage | 51-61 | x | x | na | x | x | x |
1994 | U.S. National | Site-specific, women | 40-55 | na | x | x | x | na | x | |
1995 | U.S. National | General population, plus oversamples | 25-74 | x | x | x | x | x | x | |
2005 | U.S. National | Complex | Adults born 1920-1947 | x | x | x | x | na | x | |
NHATS** | 2011 | U.S. National | Medicare beneficiaries ages 65+ | 65+ | x | x | x | x | x | x |
1984 | Sweden | All pairs of twins from the Swedish Twin Reg. separated before age 10 | 25+ | x | x | na | x | x | x | |
1998 | China | Randomly selected | Centenarians and older | na | x | x | x | x | x | |
2002 | 6 Countries | Representative | 18+ | x | x | x | x | na | x | |
2004 | Costa Rica | Census drawn | Adults born 1945 or earlier | x | x | na | x | na | x | |
2004 | 27 European countries + Israel | Probability | 50+ | x | x | na | x | x | x | |
MIDJA (part of MIDUS series) | 2008 | Japan | Probability | 30-79 | x | x | na | x | x | x |
2009 | Ireland | Complex | 50+ | x | x | na | x | x | x | |
2014 | South Africa | INDEPTH, Census and WHO SAGE based | 40+ | x | x | x | x | na | x |
*HRS and NHATS are discoverable from NACDA; users will need to access the data through the HRS and NHATS repositories.
**Caregiving can be found in the NSOC
There are many helpful resources available from various centers and universities, as well as from statistical software agencies. Here are a few examples:
ICPSR YouTube: videos on how to deposit data, how to read ASCII data, and more!
Inside NIA: A Blog for Researchers: news about NIA research priorities and funding policies
