Frequently Asked Questions
Where can I find the complete survey question text?
Do the survey questions stay the same every year?
What is the survey universe for the SHED?
What is the unit of analysis for SHED responses?
Why does SHED data have weights? Should I use them?
What is the educ_orig variable?
If a respondent does not answer a question, how are missing response values dealt with?
What does “other” mean in the race/ethnicity variables?
Why do some questions have a different number of unweighted respondents in appendix B?
What levels of geography are available in SHED data?
Can SHED data be used to create state-level estimates?
Why are there some variables in the public dataset that are not in the survey questionnaire?
Where can I find the complete survey question text?
Posted: July 14, 2026
Appendix A of each Economic Well-Being of U.S. Households report contains a reproduction of the survey instrument, including the full text, skip logic, and response options for each question.
Do the survey questions stay the same every year?
Posted: July 14, 2026
No, the survey instrument changes over time. While many questions remain on the survey from year to year, the survey changes annually to include questions on new topics. Additionally, there are some question modules that are only included every other year or every few years. The wording and response options of some questions have changed over time. Data users should refer to appendix A for each survey year to confirm what questions are included.
What is the survey universe for the SHED?
Posted: July 14, 2026
The sample for the survey consists of noninstitutionalized adults age 18 or older who reside in the United States. It excludes those living in prisons, jails, and institutionalized settings. Because survey recruitment is done through address-based sampling, it also excludes adults who did not reside at a physical address when recruitment occurred.
What is the unit of analysis for SHED responses?
Posted: July 14, 2026
All SHED results are representative of the U.S. adult population and all percentages in SHED results are reflective of the share of adults, unless specified otherwise. For some questions, respondents are asked for information about others, such as their spouses or partners (if any), although the unit of analysis remains U.S. adults.
Why does SHED data have weights? Should I use them?
Posted: July 14, 2026
SHED data include weights so that the sample population matches the U.S. adult population (excluding those in prisons, jails, and institutionalized settings). All results in the Economic Well-Being of U.S. Households reports are generated using survey weights. It is recommended that data users use survey weights when conducting analyses with the SHED. The main survey weight for every year of SHED data is titled “weight” in the public datasets. For most years of survey data, panel weights are included for those respondents who also took the survey in the previous year. The variable name of the panel weights is listed in the codebook in each year the dataset contains panel weights.
Do respondents take the survey in multiple years? How do I link responses for the same respondent over multiple surveys?
Posted: July 14, 2026
Some respondents take the survey in multiple years, and the shedid variable can be used to link responses for the same respondent over multiple survey years. Respondents are assigned one unique shedid value that is used for every survey they take. The first four digits of the nine-digit shedid identify the first year the respondent completed a SHED survey.
What is the educ_orig variable?
Posted: July 14, 2026
Educ_orig is the original answer provided by respondents to question ED0 about the respondent’s highest level of education. Since 2017, the variable ED0 is recoded using information from that question along with other questions in the survey. For example, respondents who said that their highest level of education in ED0 was a high school degree but who subsequently say that they are enrolled in college are recoded as some college. Results in the Economic Well-Being of U.S. Households reports are generated using recoded educational values, and the variable ED0 in the public dataset also uses these recoded values. For most users, we recommend using ED0 to determine a respondent’s education, but if data users wish to use survey respondents’ original responses, they can use the educ_orig variable.
If a respondent does not answer a question, how are missing response values dealt with?
Posted: July 14, 2026
Respondents are only included in the data if they reach the final question in the survey. However, respondents can skip questions throughout the survey that they do not want to answer. For most survey questions, the response rate is over 99 percent. Since 2021, when respondents do not answer non-open-ended questions, the missing responses are handled by imputation. Each imputed variable in the SHED public dataset has a corresponding imputation flag variable, ‘variable’_iflag, which has a value of 1 if the observation was imputed and a value of 0 otherwise. Users who want to only use non-imputed respondents can use the iflag variables to do so. For more information about the imputation approach in any year of the survey, see the “Description of the Survey” section of that year’s report.
What does “other” mean in the race/ethnicity variables?
Posted: July 14, 2026
The “other” race/ethnicity category consists of those respondents identifying as American Indian, Alaska Native, Native Hawaiian, or Pacific Islander as well as all respondents who report more than one racial identification. This group routinely has a small sample size, and its racial composition can change from year to year.
Why do some questions have a different number of unweighted respondents in appendix B?
Posted: July 14, 2026
Each question is coded to follow a specific question skip logic to avoid asking questions to respondents for whom the question does not apply. Some questions are asked to every respondent. Other questions are conditioned on prior survey questions. For example, respondents only receive some employment questions if they reported that they are employed. For question-specific skip logic, see appendix A of each report.
What levels of geography are available in SHED data?
Posted: July 14, 2026
The public dataset contains variables for state (ppstaten), census region (ppreg4), and census division (ppreg9). Census regions and divisions are geographies delineated by the U.S. Census Bureau. The 50 states are grouped into four census regions and nine census divisions.
Can SHED data be used to create state-level estimates?
Posted: July 14, 2026
While the SHED datasets include state identifiers, the survey is not specifically designed to give representative estimates at the state level. To look at how estimates differ across the United States, we recommend using the census region and census division variables. Data users who require state-level information should ensure that observable characteristics for the state are consistent with those seen in other state-level data. The state identifiers can be useful for other purposes, such as merging in outside measures that vary by state for analysis alongside SHED variables.
Why are there some variables in the public dataset that are not in the survey questionnaire?
Posted: July 14, 2026
The Board’s survey partner fields questions to respondents on other surveys throughout the year. Some of those background profile variables are included in the SHED dataset alongside questions that were fielded specifically as part of the SHED. In most cases, variables that begin with “pp” are ones that come from our survey partner and are not fielded in the annual SHED. In some cases, these background profile variables cover similar concepts to the information collected on the SHED. In these cases, we recommend using the questions fielded on the SHED when possible.
How should I cite SHED data?
Posted: July 14, 2026
Please cite the use of SHED data as the following: Board of Governors of the Federal Reserve System, Survey of Household Economics and Decisionmaking [dataset] (Washington: Board of Governors, 2025), https://doi.org/10.17016/datasets.002.