Documentation » Question Concordance » User Notes
This page provides additional information on how the Concordance works, including an overview of how Concordance records are selected and processed and how they relate to overall study content. If you are having difficulty in retrieving a desired record set, you should carefully review the Filtering section. It provides useful tools for broadening and/or narrowing your search.
If you discover errors in the Concordance, please forward a description of the problems you encounter to the HRS Help Desk.
Dataset/Interest Area Selection
The Health and Retirement Study Question Concordance provides links to Core and Exit data collected in all biennial interview years. Since the Concordance is under continuous development, you may find instances where information produced by this application does not match information in released datasets. For this reason, when using the Concordance, please refer to the actual codebook in order to confirm the presence and/or location of variables in the data. You can reach the online codebooks at the HRS Public Website Documentation page or by clicking on the appropriate codebook link from the Concordance page.
The Concordance contains one record for each variable in the HRS datasets distributed to the public. Multiple response and looped questions are an exception to this rule in that only one instance of each is included. Every concordance record is coded with three search terms: Category, Sub-category and Detail. Category search terms reflect the content of major Section areas. Sub-Category search terms match minor areas within the content defined by Category. Detail focuses on question/variable content. As noted below, if you enter a text string for retrieval purposes it will be compared to each of these categories (as well as question text, if available).
The HRS datasets upon which the Concordance is based have common research goals which produce questions with similar content. Unfortunately, changes in methodology and variations in questionnaire design have tended to blur the boundaries between Category and Sub-category. To correct this problem, we have introduced the concept of Interest Area. This additional search term cuts across Category and Sub-Category, producing concordance record groupings designed to mirror the research interests of potential users.
HRS variable cross-reference information has been added to selected records in the Concordance database. This feature allows users to compare similar questions across waves for 1995 through 2000 (Core) and 2002 onward (Core and Exit). Look for the Xref link in the retrieval set returned by your query.
Certain records in the Concordance do not have question text assigned. These records are flagged as "TBD" (to be determined).
Choose only the data sets and interest areas that you absolutely need. There are over 50,000 records in the Concordance database. If you select all data sets and all interest areas, the result set, in HTML format, may cause difficulties for your web browser. Because of this, the default retrieval set is limited to 500 records, after all filters have been applied. If you wish to override this value, just type a new value in the Retrieval Limit box.
Whenever possible, each record in a retrieval set contains a link to the appropriate codebook table. To see details for a variable (jump instructions, question text, user notes, and codeframe) click on the variable name to access the codebook.
The retrieval set is displayed in a table. You can sort individual columns in the table by clicking on the column header
The records in this version of the Concordance are based on the contents of these HRS Public Release files:
Dataset Date Version HRS 1992 (Wave 1) May 1995 Final v1.01 1 AHEAD 1993 (Wave 1) Mar 1998 Final v2.10 HRS 1994 (Wave 2) Core + Exit Jun 1998 Final v1.00 2 AHEAD 1995 (Wave 2) Core May 2002 Final v2.00 AHEAD 1995 (Wave 2) Exit Feb 2003 Final v2.00 HRS 1996 (Wave 3) Core Sep 2003 Final v4.00 HRS 1996 (Wave 3) Exit Sep 2003 Final v1.00 HRS 1998 Core May 2003 Final v2.00 HRS 1998 Exit Jan 2005 Final v1.00 HRS 2000 Core Sep 2002 Final v1.00 HRS 2000 Exit Mar 2005 Final v1.00 HRS 2002 Core Jun 2003 Final v1.00 HRS 2002 Exit Sep 2003 Early HRS 2004 Core Aug 2006 Final v1.00 HRS 2004 Exit Jan 2006 Early HRS 2006 Core Sep 2008 Final v1.0 HRS 2006 Exit Oct 2007 Early v1.0 HRS 2008 Core Aug 2009 Early v1.0 HRS 2008 Exit Dec 2009 Early v1.0 HRS 2010 Core Jul 2011 Early v1.0 HRS 2010 Exit Oct 2011 Early v1.0 HRS 2012 Core Dec 2014 Final v1.0 HRS 2012 Exit Apr 2014 Early v1.0 Note 1: A newer version of HRS 1992 is being distributed. Read the data description for a description of the differences between v1.01 and the latest version. Note 2: A newer version of HRS 1994 is being distributed. Read the data description for a description of the differences between v1.00 and the latest version.
Interest area boundaries do not necessarily match sections. If you want to retrieve a particular section within a data set, do not select interest areas (take the default of all) and instead filter on Section. Note: Selecting one or more Interest Areas together with one or more Filter statements will limit your search (implied Boolean AND).
Cross-Reference: Interest Area by Biennial Product
Interest Area sets are retrieved by matching on either of two fields in the Concordance record. These fields contain code value pairs that are designed to allow retrieval of record sets that cut across section boundaries.
Interest Area Code List (PDF)
Filtering with Search Fields
Once you have made your dataset and major category selection, you can refine your retrieval set by searching the question text, sub-category and detail fields for a string or a pattern. You can also select individual sections as well as short label (a.k.a. question number) ranges. An additional selection can be made on release type which flags variables as public or masked (restricted). If you use two or more filtering options, the result is a Boolean AND. As noted below, words must be 4 characters or more to be found via freetext search. Thus if you enter a 3-character word no filtering will occur.
- all of these words:
Enter one or more words (case insensitive) all of which may be present for for a record to be returned. For example:
- this exact phrase:
Enter a phrase (case insensitive) that must be present for a record to be returned. For example:
- any of these words:
Enter one or more words (case insensitive). If any word in the list is found, the record will be returned. For example:
- none of these words:
Enter one or more words (case insensitive). If any word in the list is found, the record will not be returned. For example, to return health questions that do not include certain conditions (e.g., cancer or diabetes), check the "Health Status" box and enter:
Users should be aware of two peculiarities of the freetext search process:
- Any word that is too short is ignored. The default minimum length of words that will be found by full-text searches is four characters.
- Words in the stopword list are ignored. A stopword is a word such as "the" or "some" that is so common that it is considered to have zero semantic value.
- An overview of section contents for each biennial survey is contained on the Questionnaires page. Since sections do not necessarily match across datasets. If you select multiple datasets, be careful to choose the correct section(s) from the Questionnaires page.
- Enter section letters separated by commas or spaces: e.g., A,B,N,M1 M2
- Interest area boundaries do not necessarily match sections. If your retrieval request results in an empty set, make sure that the section that you are seeking exists within the areas that you have selected.
- To retrieve all modules for a given year, enter "Module" in the string search field (see below).
Enter all or part of the variable name that you are seeking. All questions containing the character string you enter will be returned. Example:
returns records where variable name contains the submitted string.
D240, D2407, D2408, D2409
Beginning in 2002, variables replicated across waves are given a common base name with a prefix that corresponds to the Wave indicator. If you enter a base name, all questions containing that character string will be returned. Example:
returns "PURCHASE PRICE OF HOME" variables for all Core interviews beginning with 2002.
For biennial interview years 2002 and beyond, Variable Name and Short Label values are identical. For years 2000 and earlier, Short Labels have the standard form of
Primary alpha code (upper case) Primary label number Secondary alpha code (lower case) Secondary label number
The Low/High fields under Short Label allow the user to retrieve a series of questions. Example:
Low: A23 High: A25
A23, A24, A24a, A24a1, A24a2, A25.
Note: In some HRS documents, "Short Label" is referred to as "Question Name".
Release Type Selection:
Variables flagged as "masked" have been deleted from the public dataset. In certain cases these variables may be available as Restricted Data.
You can determine the format of your retrieval set in the following ways:
Sort Order: Selecting Dataset, Sequence Number will produce a listing that mirrors the order of variables within and actual dataset. Category, Subcategory, Detail, Dataset results in questions being grouped by dataset within the three search catagory levels.
Output Format: Table produces an HTML display, while Listing removes tabular formatting to produce a very simple display. This option is the fastest way to display your retrieval set. Use Delimited if you want to produce a retrieval set in comma-quote delimited format (e.g., for importation into a database or spreadsheet).
Retrieval Limit: Although the default value is 500, you can limit your retrieval set to any number of records in the range 1 to 59,574. This option is particularly useful if you are selecting many datasets and/or categories and are not sure how many records will be returned by the program. Note that this limit is applied after dataset selection, category selection, and filtering.