First a disclaimer. I choose the title of this post to reflect how I think others view these data products. These are not, in fact, competing data products, nor are they alternatives to one another. I view these data products as complementary, each with individual strengths and weaknesses, but ultimately a powerful data analysis resource when combined. Below I briefly describe the difference between the two data products and outline the pros and cons of using each (though my ultimate recommendation is to combine the two data products for a “pros-only” data set).
The most frequent point of clarification I have to make about the HRS is to explain the differences between the data produced by HRS and the HRS longitudinal data file produced by RAND (a.k.a. RAND HRS). Sometimes users characterize RAND as having produced HRS data, but this is incorrect. That impressive feat can only be attributed to the HRS staff at the University of Michigan and the dozen or so Investigators at UM and elsewhere. All HRS data are collected, processed, and distributed by HRS.
RAND then takes that data and does something also very impressive — they create a user-friendly longitudinal data file (this is not easy to do!). There are few studies as complex as the HRS that have such an easy-to-use version of the data as the RAND HRS. Having a cleaned, user-friendly version of the HRS is a boon to both new and experienced HRS users.
The RAND HRS is a terrific way to get started using HRS data. But for all its advantages, the RAND file is quite limited relative to the full scope of all HRS data and many users will soon find that their data needs extend beyond the limits of the RAND HRS. On the other hand, users will also find out that using at least some parts of the RAND HRS will save them days, or even weeks, that would have been spent merging files and constructing variables. The more waves of data you’re using, the larger the advantage in using the RAND HRS.
So what are the advantages and disadvantages of the different data products? I’ve included a brief list of the “pros” and “cons” of each data product in the table below. But as I said, these products are complementary – the limitations of one data product are overcome by the other – so you need only combine the two for a “pros-only” data set. Fortunately, integration between the files provided by HRS and the RAND HRS file is fairly seamless. I usually start off my research projects by merging the RAND HRS with other files distributed by HRS, such as the Tracker File, Core Files, Biomarker, etc.