I admit to being puzzled initially by the many options available in the Data Download System. (You can access this system once you have become a registered user). Through trial and error I eventually figured out what I needed and how all the files work together. Below is a high-level summary of the main points about these downloads.

HRS Biennial Data Products

On the left side of the page are the “Biennial Data Products” which are the files HRS produces from the surveys conducted every two years. This is where you will find data from the core interviews and exit interviews, as well as the off-year studies (e.g., CAMS). There are some download options. First, you could download the Distribution Set (e.g., h10core.zip), which contains all the files and sort out later which files you really need. Second, you can download just the Data Files (e.g., h10da.zip) and Data Descriptions needed for your software to read in the raw data (e.g., h10sta.zip). For some time I only downloaded the data and data descriptions, but I now recommend a third option, which is to also download the documentation. This is an extra step, but downloading the Codebook Files (e.g., h10cb.zip) and the User Information (e.g., h10dd.pdf) will save you time later. Although all the documentation is available online, there are times when we don’t have access to the internet and, speaking from experience, it’s frustrating to be in the midst of data management or analysis and not be able to check the codebook for some detail.

Which is the RAND FILE?!?

Although there are files in the first column of downloadable data with “RAND” in the description, these are the “Fat Files” and not the RAND HRS, which is found in the second column under RAND Contributed Files. The Fat Files are created by RAND, but they are wave-specific and contain most of the variables from each section of the survey for that year (RAND HRS contains only a subset of HRS variables). Another useful feature of the RAND enhanced Fat File is that the data are collapsed into a respondent-level dataset that is easily merged with other waves and with the RAND HRS. Because the Fat Files contain nearly all the variables in a given wave, the files are quite large and may take a few minutes to download.

If you intend to download the RAND data you likely want the “RAND HRS Data File”. But, as I’ve mentioned previously, the best way to use the HRS is to combine the RAND HRS with the rest of the HRS. This is easily done by merging the RAND HRS Data File with the Fat Files. The downside here is that the more waves you combine the larger your dataset gets. If you’re a Stata user you already know that larger files take longer to download and work with, so this can be a concern. If you only need a few variables across waves you can get those from downloading the HRS Core data files and merging across the specific sections of the questionnaire that contain your variable(s) of interest. When I want sleep variables, for instance, I only merge the files containing these variables (section C) with the RAND file. But if you want many different variables from different sections of the survey, and only for a few waves, the Fat Files are the way to go.

Additional Files

There are other files you may be interested in that can be found in the column on the right side of the page. This is where you can find the Tracker File, for instance. In addition, HRS provides some researcher contributed files. I highly recommend thoroughly reading the accompanying documentation before using these files – because HRS did not produce these files they do not support these files. If you have applied and been approved for use of sensitive data products, Biomarker Data for instance, these will also show up in the right-side column under “HRS Special Access Files.”