The CE Annual Data Quality Profile — 2024

Authors: Grayson Armstrong, Gray Jones, and Laura Petit

This paper was published as part of the Consumer Expenditure Surveys Program Report Series.

Table of Contents

Overview

Highlights

A. CE Quarterly Interview Survey

1. Final disposition rates of eligible sample units

2. Records Use

3. Information Booklet use

4. Expenditure edit rates

5. Income imputation rates

6. Respondent burden

7. Survey response time

8. Survey mode

B. CE Diary Survey

1. Final disposition rates of eligible sample units

2. Information Booklet use

3. Expenditure edit rates

4. Income imputation rates

5. Survey response time

6. Survey mode

7. Online Diary Paradata

Summary

References

In this section, the BLS highlights noteworthy metric trends from the past year of available data. This time frame covers the first quarter of 2024 to the fourth quarter of 2024 for the CE Diary Survey, and the second quarter of 2024 to the first quarter of 2025 for the CE Interview Survey. Subsequent sections will describe all the individual metrics with detailed data tables and graphical representations.

Recent Trends of Note

• Interview and Diary Survey response rates trended downward in the most recent year of available data.

• The Interview Survey other nonresponse rate was 19.0 percent in the first quarter of 2025. The persistently high rate of other nonresponse in recent quarters was expected as cost-reduction measures implemented by the BLS beginning in the fourth quarter of 2023 remained in effect through this time period. (See Chart A1.1)^[7]

• The Diary Survey other nonresponse rate was 19.8 percent in the fourth quarter of 2024. This change in other nonresponse rate was also the result of the cost reduction measures that were implemented by the BLS beginning in the fourth quarter of 2023.(See Chart B1.1)

• Three new metric series were produced to analyze online diary expenditure reporting patterns, using paradata in Section B7.

• The first new metric analyzes online diary logins and shows that since the paradata were made available in the first quarter of 2021, the average number of times a respondent successfully logged in to the online diary instrument over the two-week diary-keeping period fluctuated between 6.7 and 8.0 times. (See Chart B7.1).

• The second new metric analyzes device use and shows that that 21.6 percent of respondents used only a mobile device to enter expenses, while 59.7 percent used only a desktop, laptop, or tablet device, and the remaining 18.7 percent used both types of devices (See Chart B7.2).

• The third new metric examines time spent in the online diary. Following the second quarter of 2021, the series has fluctuated between 35.0 minutes and 40.9 minutes with no clear trend (See Chart B7.3).

A. CE Quarterly Interview Survey

1. Final disposition rates of eligible sample units

Final disposition rates of eligible sample units in the Interview Survey represent the final participation outcomes of field staff's survey recruitment efforts. The BLS classifies the "final outcome" of eligible sample units into the following four main categories:

1. Completed interview
2. Nonresponse due to refusal
3. Nonresponse due to noncontact
4. Nonresponse due to other reasons

More details on how these categories are defined and how BLS staff calculate response rates can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

The primary interest in response rates is their relationship with nonresponse bias.^[8] That is, low response rates can indicate the potential for nonresponse bias in expenditure estimates if the cause of nonresponse is correlated with spending. While recent research on nonresponse bias in the CE Interview Survey has not shown bias in the CE survey estimates (Ash & Rothbaum, 2024), BLS continually monitors nonresponse bias risk in the CE.

Response rates in this report are presented as unweighted values. This is because weights are used to estimate results from the population, and the goal of this analysis is to measure the effectiveness of our actual data collection efforts rather than make inferences about the wider population.

Final disposition rate summary

• The percentage of eligible sample units resulting in complete interviews experienced an overall change of 0.9 percentage points in the most recent year of available data, moving from 40.9 percent in the first quarter of 2024 to 40.0 percent in the first quarter of 2025 (See Chart A1.1). The percentage of complete interviews also experienced some minor fluctuations between the second and fourth quarters of 2024.^[9]

• Refusal rates experienced an overall decline of 3.3 percentage points, moving from 30.4 percent in the first quarter of 2024 to 27.1 percent in the first quarter of 2025. The noted drop off in refusal rates coincided with the implementation of cost containment measures in Fiscal Year 2024 (FY24), which targeted Wave 4 cases that were likely to result in a refusal to participate, effectively shifting these cases to be coded as other nonresponse. ^[10] A similar dynamic between refusals and other nonresponse can be seen in the third quarter of 2022, during temporary fielding changes in Fiscal Year 2023 (FY23).^[11]

• After reaching a series high of 23.0 percent in the fourth quarter of 2022, noncontact rates have generally fallen over the past two years of available data. In the first quarter of 2025, noncontact rates were measured at 13.9 percent.

• Other nonresponse rates experienced an overall change of 3.9 percentage points over the past year, moving from 15.1 percent in the first quarter of 2024 to 19.0 percent in the first quarter of 2025.

• The high percentage of other nonresponse since the fourth quarter of 2023, relative to the full series, resulted from the implementation of cost containment measures in FY24 aimed at lowering data collection costs. The increase in other nonresponse was largely driven by cases targeted by the cost containment measures, specifically 4th wave interview cases that had resulted in non-interviews in previous interview waves.

• The previous instance of high other nonresponse was related to the implementation of a similar set of cost containment measures by the BLS in the third quarter of 2022. This resulted in a 17.0 percentage point change from 2.0 percent in the second quarter of 2022 to 19.0 percent, before returning to 2.0 percent in the fourth quarter of 2022.

• The only other recorded instance of fluctuations of this magnitude in the component parts of Final Disposition (not related to data collection cost containment) was during the onset of the COVID-19 pandemic in 2020, when all in-person data collection was temporarily halted.^[12]

2. Records use

The Records Use metric measures the proportion of respondents who refer to records while answering the Interview Survey questions, as reported by Census Field Representatives (FRs). Examples of records include, but are not limited to: receipts; bills; checkbooks; and bank statements. The interviewer documents records use retrospectively, at the end of the interview. Past research has shown that respondents who use expenditure records report a more complete expenditure picture with lower rates of missing data (Abdirizak, Erhard, Lee, and McBride, 2017). Thus, a higher records use rate is desirable. Metrics in this section are presented by survey wave.^[13] More information on the Records Use metric can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Record use summary

• The records use rate for interview respondents decreased nominally across all waves in the period between the first quarter of 2024 and the first quarter of 2025 (See Chart A2.1).

• Wave 1 records use had an overall change of 0.2 percentage points, from 62.6 percent in the first quarter of 2024 to 62.4 percent in the first quarter of 2025 but did experience larger fluctuation in the interim.

• The records use rate by respondents in Waves 2 & 3 had an overall change of 2.1 percentage points from 60.5 percent in the first quarter of 2024 to 58.4 percent in the first quarter of 2025. Like in Wave 1, the Waves 2 & 3 series experienced slightly greater fluctuation in between the second and fourth quarters of 2024.

• Wave 4 records use experienced an overall change of 4.4 percentage points from 63.1 percent in the first quarter of 2024 to 58.7 percent in the first quarter of 2025, with some fluctuation in the interim.

3. Information Booklet use

The Information Booklet is a recall aid that the Census FR provides to respondents. The Information Booklet provides response options for demographic questions and bracket response options for income questions. Additionally, Survey respondents can use the Information Booklet to view clarifying examples for specific expenditures that each section/item code is intended to collect.

This metric identifies the prevalence of Information Booklet use among respondents during their interviews, according to Census FRs. For those who are interviewed in-person, the FR shares directly the physical Information Booklet when needed. However, respondents who are interviewed by telephone typically do not have ready access to the Information Booklet (although a PDF version is available on the BLS website).

Thus, this metric should be interpreted in conjunction with the rise in telephone interviews during the COVID-19 pandemic. Higher Information Booklet usage rates are encouraged, as usage can improve reporting quality by clarifying concepts and providing contextual examples. More information on the Information Booklet use metric can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Information booklet use summary

• The rate of Information Booklet use among Wave 1 respondents saw an overall shift of 4.6 percentage points in the most recent year of available data, moving from 36.9 percent in the first quarter of 2024 to 32.3 percent in the first quarter of 2025. This series experienced fluctuations between the second and fourth quarters of 2024, most notably going below 30 percent for the first time since the first quarter of 2022 (29.8 percent) (See Chart A3.1).

• The rate of Information Booklet use among Wave 2 & 3 respondents experienced a change of 1.8 percentage points in the most recent year of available data, moving from 28.2 percent in the first quarter of 2024 to 26.4 percent in the first quarter of 2025. Similar to Wave 1, Information Booklet use among Wave 2 & 3 respondents had fluctuations between the second and fourth quarters of 2024 and saw a two-year low in this time frame (24.8 percent).

• Information Booklet use among Wave 4 respondents experienced an overall change of 3.3 percentage points in the most recent year of data moving from 26.4 percent in the first quarter of 2024 to 23.1 percent in the first quarter of 2025. Like in the previous interview wave groups, Information Booklet use among Wave 4 respondent hit a two-year low during this time period (23.1 percent).

• Information Booklet use by respondents across all waves continues to be below pre-COVID-19 pandemic levels.^[14]

4. Expenditure edit rates

The Expenditure edit rates metric measures the proportion of reported expenditure data that are edited. Edits are defined as changes made to reported expenditure data during CE data processing, excluding changes due to: time period conversion calculations; top-coding; or suppression of reported values. Top-coding and suppression (redacting reported values) are done to protect respondent confidentiality in the public use microdata (PUMD).^[15] Additional information on top-coding and suppression is available on the CE Website.

The Interview Survey expenditure edit rates are broken down into three categories: Imputation; Allocation; and Manual Edits:

• Imputation replaces missing or invalid responses with a valid value.

• Allocation edits are applied when respondents provide insufficient detail to meet tabulation requirements. For example, if a respondent provides a non-itemized total expenditure report for the category of fuels and utilities, that total amount will be allocated to the target items mentioned by the respondent (such as natural gas and electricity).

• Manual edits occur whenever responses are directly edited by BLS economists based on their analysis and expert judgment.

Imputation of CE data results from item nonresponse. Allocation is a consequence of responses lacking the required details for items asked about in the survey. Lower edit rates are preferred, as it lowers the risk of processing error. However, edits based on sound methodology can improve the completeness of the data, and thereby reduce the risk of measurement error and nonresponse bias in survey estimates. Additional information on expenditure edits is available in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Expenditure edit rate summary

• The rate of allocated expenditures saw an overall change of 0.4 percentage points in the most recent year of data, moving from 8.3 percent in the first quarter of 2024 to 8.7 percent in the first quarter of 2025, with some minor fluctuation in between (See Chart A4.1).

• The rate of imputed expenditures, imputed and allocated expenditures, and manually edited expenditures varied little in the most recent year of available data.

• The total rate of unedited expenditures hovered around 87.0 percent for most of 2024 and early 2025, after reaching a series high of 87.7 percent in the first quarter of 2024.

5. Income imputation rates

The Income imputation rates metric describes edits performed on a consumer unit's (CU's) nonresponse to at least one source of income. This edit is based on three imputation methods:

1. Bracket response imputation: when the respondent reports receipt of an income source, but only reports that income as falling within a specified range.
2. Model-based imputation: when the respondent reports receipt of an income source but fails to report any information about the amount.
3. All valid blank (AVB) conversion: when the respondent reports no receipt of income from any source, but the BLS imputes receipt from at least one source.

After imputation, income is summed from each source for which receipt is reported or imputed to compute total income before taxes for the CU as a whole. In the following text, income before taxes is defined as “unimputed income” if no component of total income required imputation for one of the three reasons identified above.

The need for imputation arises from either item nonresponse or from response with insufficient detail was provided (e.g., the respondent provided a range of income like “between $40,000 and $50,000,”). Higher response rates for actual values are associated with less potential nonresponse bias in the data for imputation to address. However, imputation based on sound methodology produces a complete dataset, and reduces the risk of nonresponse bias since incomplete cases are no longer dropped from the dataset. Further details on the income imputation methodology can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025) and the User's Guide to Income Imputation in the CE (Paulin, Reyes-Morales, and Fisher, 2018).

Income imputation rate summary

• The rate of unimputed total income before taxes experienced an overall change of 1.6 percentage points in the most recent year of data, from 58.3 percent in the first quarter of 2024 to 59.9 percent in the first quarter of 2025, with some additional fluctuations in the quarters in between (See Chart A5.1).

• The rate of model-based imputation had an overall change of 0.5 percentage points in the most recent year of available data, from 17.4 percent in the first quarter of 2024 to 16.9 percent in the first quarter of 2025, with some minor fluctuations in the interim.

6. Respondent burden

Respondent burden measures the perceived level of effort exerted by respondents in answering survey questions. This metric is only recorded in the Interview Survey. Survey designers are concerned about respondent burden as it may negatively affect response rates and overall response quality. As of April 2017, at the end of the Wave 4 interview, the BLS introduced a question on respondent burden in the Interview Survey with response options describing five different levels of burden. The respondent burden metric is derived from this question and maps five burden categories to three metric values: respondents perceiving the survey to be very burdensome, perceiving the survey to pose some burden, and perceiving the survey as not burdensome. Please see the DQP Reference Guide (Armstrong, Jones, and Petit, 2025) for more details on the question wording and the burden categories.

A caveat about interpreting respondent burden is that since the burden question is only asked at the end of Wave 4, the metric may underestimate survey burden due to self-selection bias. That is, respondents who have agreed to participate in the final wave of the survey presumably find the survey less burdensome than sample units who had dropped out at any point prior to completing the final survey wave. However, it is also possible that the respondent answering this question did not participate in prior interview waves. For example, the respondent who participated in the first three surveys might have moved and been replaced by someone participating and answering the burden question for their first and only interview.

Respondent burden summary

• The percentage of respondents perceiving the survey as not burdensome saw an overall change of 2.7 percentage points in the most recent year of data, from 29.0 percent in the first quarter of 2024 to 26.3 percent in the first quarter of 2025. The series experienced some additional fluctuations between the second and fourth quarters of 2024, most notably in the third quarter of 2024 when the metric exceeded 30 percent for the first time since mid-2020 (See Chart A6.1).

• The percentage of respondents perceiving the survey to pose some burden had an overall change of 2.8 percentage points in the most recent year of data, from 53.2 percent in the first quarter of 2024 to 56.0 percent in the first quarter of 2025. The percentage of respondents reporting some burden saw fluctuation between the second and fourth quarters of 2024, but these values did not stray beyond those in the first quarter of 2024 or 2025 described above.

• The percentage of respondents perceiving the survey to be very burdensome had an overall change of 0.3 percentage points in the most recent year of data, from 16.3 percent in the first quarter of 2024 to 16.0 percent in the first quarter of 2025. Despite the relatively small overall change, the series did experience notable variation between the second and fourth quarters of 2024, reaching a high of 16.9 percent in the fourth quarter of 2024.

7. Survey response time

For the Interview Survey, the survey response time metric is the median number of minutes required to complete the interview. In the Interview Survey, Wave 1 and 4 interviews are typically longer because Census FRs collect additional information, mainly household demographics (Wave 1) and assets and liabilities (Wave 4). Longer survey response time could (in some circumstances) represent an indicator the respondent experienced more burden. Past internal CE research has found that higher respondent burden negatively affects both response rates and data quality. However, survey response time could also reflect the respondent’s degree of engagement. Engaged and conscientious respondents might take longer to complete the survey because they report more thoroughly or use records more extensively. Regardless, tracking the median survey response time can be useful for assessing the effect of changes in the survey design. More information on the Survey Response time metric can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Survey response time summary

• Median interview time in Wave 1 saw a 2.0 minute change in the most recent year of data from 81.8 minutes in the first quarter of 2024 to 79.8 minutes in the first quarter of 2025, with some additional variation present between the second and fourth quarters of 2024 (See Chart A7.1).

• In Waves 2 & 3, the median interview time experienced an overall change of 1.4 minutes from 59.7 minutes in the first quarter of 2024 to 58.3 minutes in the first quarter of 2025, with some additional variation evident between the second and fourth quarters of 2024.

• Wave 4 median interview time experienced the smallest overall change in the most recent year of data, from 64.6 minutes in the first quarter of 2024 to 65.4 minutes in the first quarter of 2025. While this change in median time was less than a minute, the series did experience some fluctuation between the second and fourth quarters of 2024.

8. Survey mode

The Interview Survey was designed to be conducted in-person. However, the interviewer can also collect data over the phone, or by a combination of the two modes. In general, survey practitioners prefer higher rates of in-person data collection, because the interviewer can actively prompt the respondent, as well as encourage the use of recall aids, thereby reducing the risk of measurement error. Conducting Wave 1 interviews in person is especially important as this is typically the respondent’s first exposure to the survey. This serves as an opportunity for the Census FR to build rapport with the respondent As a data quality measure, the BLS has inter-agency agreements with the Census Bureau stipulating that the average CE telephone interview rate, defined as interviews in which 50 percent or more of the survey sections are completed by telephone, should be limited to 25 percent for Wave 1 interviews, and 50 percent for all subsequent interview waves, or as feasible, to be determined by factors that may restrict or limit in-person interviewing in selected geographies. More information on how the mode metrics are calculated is available in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Survey mode summary

• The rate of in-person interviews, across all waves, experienced an overall change of 0.7 percentage points in the most recent year of data from 38.0 percent in the first quarter of 2024 to 37.3 percent in the first quarter of 2025 (See Chart A8.1). While the overall change was less than one percentage point, the series did experience a fluctuation of nearly 4 percentage points in that time frame.

• The rate of Wave 1 in-person interviews also had an overall change of 0.7 in the most recent year of data from 49.5 percent in the first quarter of 2024 to 48.8 percent in the first quarter of 2025 (See Chart A8.2). While the overall change was less than one percentage point, the series experienced a fluctuation of roughly 5 percentage points in that time frame.

• The rate of in-person interviews for Waves 2 & 3 saw slightly more overall change than in Wave 1 interviews. The in-person rate for Waves 2 & 3 was 34.2 percent in the first quarter of 2024 and 35.2 in the first quarter of 2025 with a fluctuation of nearly 4 percentage points in the quarters in between (See Chart A8.2).

• Wave 4 in-person interviews experienced the greatest overall change in the most recent year of available data from 33.8 percent in the first quarter of 2024 to 29.8 in the first quarter of 2025 (See Chart A8.2).

• In-person interview rates across all waves continue to be below pre-COVID-19 pandemic levels. The similarity of these findings to the trend discussed in the Information booklet section are linked as interviewers typically provide the information booklet to respondents during in-person interviews.^[16]

B. Diary Survey Metrics

1. Final disposition rates of eligible sample units

Final disposition rates of eligible sample units in the Diary Survey represent the final participation outcomes of field staff’s survey recruitment efforts and are calculated using the same methods as the CE Interview Survey.

While recent research on nonresponse bias for the CE Diary Survey has not shown significant bias in the CE survey estimates during the COVID-19 pandemic (Ash, Nix, and Steinberg, 2022), BLS continues to monitor this risk.

Like in the Interview Survey results (section A1) response rates are presented as unweighted values.

Additional information on how BLS staff calculate response rates can be found in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).

Final disposition rate summary

• The rate of complete Diary Survey cases experienced an overall change of 0.8 percentage points in the most recent year of data, moving from 38.2 percent in the fourth quarter of 2023, and 39.0 percent in the fourth quarter of 2024 (See Chart B1.1). During this time frame the percentage of complete diaries varied between 41.0 percent and 38.2 percent.

• Refusal rates experienced an overall change of 6.4 percentage points in the most recent year of data, moving from 38.4 percent in the fourth quarter of 2023 to 32.0 percent in the fourth quarter of 2024, after reaching a series high of 40.5 percent in the third quarter of 2023.

• The recent change in refusal rates was likely related to the introduction of cost-containment measures by the BLS in FY24 and coincided with the change in the rate of other nonresponse described below.

• Noncontact rates ranged between 11.2 percent in the fourth quarter of 2023 and 8.8 percent in the fourth quarter of 2024.

• Other nonresponse rates have continued to experience fluctuations since cost containment measures were implemented in the fourth quarter of 2023. In the most recent quarter, the rate of other nonresponse moved 6.3 percentage points from 13.5 percent in the third quarter of 2024 to 19.8 percent in the fourth quarter of 2024.^[17] This represented the highest level of other nonresponse since CE fielding operations were impacted by the COVID-19 pandemic in 2020.^[18]

2. Information Booklet use

As presented in Section A3, the Information Booklet is a recall aid that the Census FR provides to respondents. For the CE Diary Survey, the Information Booklet provides response options for demographic questions and bracket response options for income questions.

This metric measures Information Booklet use among respondents during diary placement interviews, according to Census FRs. Typically, for diary placement interviews conducted over the phone, the Information Booklet is not readily available to the respondent (although a PDF version is available on the BLS website). Like in the Interview Survey, this metric should be interpreted in conjunction with the rise in telephone use as the most common modality for diary placement interviews since the COVID-19 pandemic. Higher rates of Information Booklet usage are encouraged, as use can improve reporting quality by clarifying concepts and providing examples.

Information Booklet summary

• Information Booklet use among Diary Survey respondents experienced an overall change of 5.1 percentage points in the most recent year of data from 27.4 percent in the fourth quarter of 2023 and 22.3 percent in the fourth quarter of 2024 (See Chart B2.1), with some fluctuations in the interim.

• Information booklet use by respondents continues to be below pre-COVID-19 pandemic levels.^[19]

3. Expenditure edit rates

The Diary Expenditure edit rates metric measures the proportion of reported expenditure data that are edited, similar to the Interview Survey

The Diary Survey expenditure edit rates are broken down into two categories: Allocations and Other Edits. The vast majority of edits in the Diary Survey are allocations. Table B3.1 below shows the “Other Edits” category, which covers all other expenditure edits including imputation and manual edits. According to the data in Table B3.1, these data edits are relatively rare.

Beginning in 2022, the BLS changed the way expenditure edit rates are calculated for Diary Survey data. Changes to the alcohol cost flag are now considered an expenditure edit under the “Other Edits” category. This change was retroactively applied to the full metric series and has led to comparatively higher estimates for “Other Edits” and lower estimates for “Unedited” compared to previous reports.

Additional information on expenditure edits is available in the DQP Reference Guide (Armstrong, Jones, and Petit, 2025).