Conceived in the 1930s to meet the increased needs for unemployment statistics, the Current Population Survey (CPS) has been conducted in its present form since 1948, making it one of the oldest and most recognized surveys in the nation. Sponsored jointly by the U.S. Census Bureau and the U.S. Bureau of Labor Statistics (BLS), it is a primary source of labor force statistics for the U.S population, producing a key Principal Federal economic indicator. Additionally, the data are used to inform policymakers on the social and economic well-being of the nation and its people by adding a set of supplemental questions to the monthly basic CPS questions. A primary example of these supplements is the Annual Social and Economic Supplement which provides income, poverty, and health insurance statistics.
The CPS program leadership from the Census Bureau and BLS have agreed that modernization is needed to ensure the sustainability of the CPS. The Census Bureau and the Bureau of Labor Statistics have started a multi-year effort to improve and modernize the operations of CPS. One of the largest efforts of this process is the introduction of an Internet self-response mode by 2027.
Over the years CPS has made changes to measure the ever-changing population and labor force characteristics of the US and implement the latest in survey methodology practices. The last major change to the CPS was in the early 1990s when it went from paper data collection to computer assisted interviewing.
The response rates for all household surveys, including CPS, have been steadily declining for many years. Ten years ago, CPS regularly achieved response rates in the high 80 percentile. However, that is no longer true in the current environment. Issues like privacy concerns, challenges contacting respondents in cellphone only households, and respondents’ availability when contact is made have all contributed to the decline1. By adding a new Internet self-response mode, it is anticipated that response rates can be increased or at the very least stabilized.
Declines in response rates are not uniquely a United States issue. International labor surveys have all seen a gradual decrease in participation rates. The quarterly United Kingdom Labour Force Survey (LFS) has seen an even more drastic decrease in response rates over the same ten years when compared to CPS.
Even the mandatory Labour Force Survey (LFS) administered by the Canadian govenment was not spared from these data collection challenges.
Across the board, surveys have face higher non-response rates. Please see BLS's main webpage for a more in-depth review on response rates. Household and establishment survey response rates : U.S. Bureau of Labor Statistics (bls.gov).
In addition, it has been found that a self -response mode can lead different types of households to respond to government-sponsored surveys. For people not interested in an interviewer-based data collection method, they may be more receptive to a self-response method. The American Community Survey (ACS) has done extensive research on this topic.
As demonstrated in Table 15 and 16 of ACS's first field test review conducted in 2011, internet respondents tended to be “younger, female, Asian, other race, with higher education and more likely to speak a language other than English at home”2.
Additionally, households with children under five tended to see an increase in self-response rate when offered an internet option. “Younger adults and households with a child under five -- we found that both of these groups were significantly more likely to respond by Internet, but only households with a child under five also saw a significant increase in self-response.”3
Internet modes have also been shown to perform comparably to mail modes as it pertains to numeric outlier analysis, correlations, and differences in response error.4, 5
Lastly, research has shown that offering respondents multiple modes does not dampen response rates or have a significant impact on mode paralysis.2,6
The first step in modernization is to add an Internet Self-Response (ISR) Mode. This section will review the testing and planning needed for this implementation. This is a multi-year project with the goal of extensively reviewing and testing ISR as a viable mode for CPS. The goal of testing is to ensure that changes, including the option to complete the survey via ISR, do not affect the reliability of the estimates produced by CPS. This schedule seeks to bring the ISR mode to CPS as quickly as possible while doing the due diligence needed for such an important indicator. Our schedule and activities may change to accommodate unforeseen obstacles, including but not limited to funding restrictions, discovery of issues during testing and analysis, and project reprioritization.
The general timeline is as follows:
Starting in 2022, Census and BLS conducted research and began developing the new Internet Self-Response (ISR) option. This included a multi-faceted approach. The following is an outline of methods we conducted.
Survey methodologists from the Census Bureau and Bureau of Labor Statistics performed an expert review of the existing CPS questionnaire to identify areas where changes may be necessary for an internet self-response mode. For example, several CPS items require an interviewer to ask follow-up questions as necessary to elicit sufficient detail from respondents, such as descriptions of job titles and job duties. There are also questions that are interviewer-coded based on respondents providing open-ended answers (e.g., a question on which techniques respondents looking for work used to try to find a job). Since there is no interviewer present in an internet self-response mode, the format of these questions needs to be adapted. As part of the review, survey methodologists made recommendations for changes and identified areas for future research. Subject matter experts reviewed these recommendations and determined which would be implemented.
There are several key features of the CPS that impact multiple questions. For example, many questions refer to a specific reference period (e.g., last week). In the current interviewer-administered CPS, interviewers state the reference period out loud, often at the beginning of a question. However, it may or may not be most effective to follow the same convention in an internet instrument. As another example, in the current instrument, help text is available for interviewers to provide if a respondent needs it. On the internet, respondents may not know they need the help text, or they may choose not to read it. Determining how to best display and incorporate help text is important. To explore these kinds of issues, survey methodologists from the Census Bureau and Bureau of Labor Statistics created an online version of the CPS with follow-up questions embedded in the instrument. Two versions were tested with multiple design features. This was fielded to a non-probability sample of respondents for research purposes.
Survey methodologists at the Bureau of Labor Statistics conducted one-on-one cognitive interviews with members of the public to learn more about the potential challenges in an internet self-response mode of the CPS. The goal of the cognitive interviews is to learn more about the thought process respondents undergo while answering questions and suggest ways in which the survey questions can be improved. These early interviews have been particularly helpful for providing preliminary insight into more complex areas for future research. Cognitive testing continues with further refined versions of an internet response mode CPS.
The introduction of an internet self-response mode is well documented in the Census Bureau. In 2011 the ACS conducted two internet tests to evaluate the feasibility of introducing an internet mode to the survey. Both tests in April7 and in November8 concluded that the introduction of an internet mode combined with a paper follow-up increases the self-administered response rate compared to paper only data collection method. As a result, ACS fully integrated internet self-response to its data collection strategy in 20139. One of the key highlights from the 2013 report was how Internet responses composed 55.7% of all self-responses. Respondents are taking advantage of the internet mode for participating in our surveys.
CPS incorporated the latest research, best practices, and lessons learned from the ACS and other online surveys into its initial testing.
The CPS instrument was programmed into the new CPS enterprise data collection solution. Since this new solution includes a new type of data collection instrument all housed within Census, all modes are being reprogrammed. Integration into the new systems was reviewed even if there were no changes to the survey content. After each round of testing, changes were reviewed and corrected for accuracy.
Usability testing on the CPS ISR instrument was completed. The goal of usability testing is to make sure respondents can use the ISR mode efficiently and intuitively by understanding the user interface of the screen and the flow of the instrument. The testing also assessed whether respondents could understand the questions without help from an interviewer.
For results on research conducted over this time, see the results section below.
The first field test focused on reviewing ISR as a viable mode for data collection. It conducted various experiments on how to make an internet mode as successful as possible. The field test evaluated different wording options, contact strategies, and other procedural changes in an environment similar to production. The goal was to examine what kind of responses were received and determine which contact strategies were most effective at encouraging respondents to use ISR, while supporting efficient and accurate reporting.
The field test evaluated both ISR and computer-assisted personal interviewing (CAPI) to determine the optimal mix of data collection modes for response rate, data quality, and cost effectiveness.
The test ran from May 2025 to September 2025. It included a sample of about 16,000 households. Households were eligible to be interviewed up to four times. As under existing CPS procedures, the first month continued to use personal interviews conducted by an interviewer to establish legitimacy, build rapport, and provide support during initial collection for the most high-quality data. In the following months, cases could be eligible to respond by ISR with a CAPI follow-up, continuing to use dependent interviewing for ease and consistency of response.
The test served as proof of concept, demonstrating that respondents would choose to respond via the web and do so within the required timeframe. The results supported the viability of ISR as a data collection mode. See the Results section below for full accounting of the test.
In 2026, modernization activities included the continued evaluation of the 2025 field test results and conducting a second field test involving about 10,000 households. This sample included 5,000 newly selected households, 4,000 households that participated in the first field test, and another 1,000 households that had recently completed their eight months in the CPS. (Data collected in this test were not used in official CPS estimates.) This phase tested updated questions, added help text, piloted texting as a contact strategy, and evaluated longitudinal cases and revised earnings questions in ISR. By the end of 2026, ISR and other system development will be substantially complete, while work continues refining ISR logistics and assessing mode effects.
Findings will be posted in the results section at a later date as analysis is completed.
In 2027, many different paths of modernization are planned.
Census will conduct the first field test for the Annual Social and Economic Supplement (ASEC) in early 2027. This will be the first field test for ASEC which has also been developing, researching and internally testing an ISR mode as part of overall modernization.
In summer 2027, Census will begin data collection in our first production environment. Once the final test analysis is complete and the new production protocols are approved, CPS will collect a limited parallel sample using the new data collection system and internet mode. Because the test samples were insufficient to measure the mode impact of these changes on official estimates, the most effective approach for implementation is to collect an additional sample that can be processed alongside the standard production sample. This will allow for comparisons without affecting current estimates. (The results will not be used in any official estimates.)
The parallel sample will run concurrently with CPS production for as long as necessary to identify and minimize potential impacts on estimates. After a sufficient sample has been reviewed and analyzed, and any necessary corrections have been implemented, the new system will be adopted. Results from the parallel sample, along with supporting documentation, will be made available to the public so that the changes can be fully understood before the transition begins. A plan for full transition is still being developed.
In addition to the new internet mode, other CPS modernization efforts are underway.
This site will be updated as next steps happen and new reviews or research occur. Please check back here for updates on the project’s progress, webinars, and conferences we will be attending.
The following is a list of findings and results from research and testing. These are working papers or presentations for users to consider but are not official Census documents. As more are completed, they will be shared here.
Your feedback is important. If you have any questions or feedback, please email us at [email protected].
If any, what differences will there be between Basic CPS collected in CAPI/CATI versus internet self-response?
Mode effects are generally unavoidable but through thorough testing we hope to identify any that occur and minimize them. We expect that some questions will need to be modified, provide more instructions, or be displayed differently.
Will Field Interviewers be involved?
Field Interviewers are an integral part of review and testing. They will give guidance on how the current instrument is performing as well as be part of all tests. Part of the update of internet mode will be to review how the interviewers will be able to take advantage of a more manageable workload.
How will CPS supplements be affected?
We do plan on continuing supplements. There may be an impact on some collection during this transition in order to allocate resources properly. Supplements will need to add the self-response mode as well.
What steps will be taken to ensure output from the old data collection solution is consistent with the new solution?
Extensive output testing is planned with review of production keying to the new solution. The new instrument will not be used until the agencies feel all changes have been made correctly. As an extra layer of care, the new solution will be phased in and kept separate from the production data used to make the monthly estimates. This allows a longer review in a production environment at no risk.
Will there be documentation of programming changes between the new and old solution?
Along the entire process, Census and BLS pledge to provide updates on developments, testing results, and programmatic decisions for full transparency. Upon final determinations, a full technical paper with will be released.
What risks are there to the modernization effort?
These ambitious changes will be challenging to execute within current funding allocations. We will need to be flexible with plans based on funding received. We are working to try and minimize this as an issue, but we need to be aware that this will be a challenge over the next few years to manage and may lead to changes in scope or timing of the development.