Data Quality

When a sample rather than the entire population is surveyed, estimates differ from the true population values they represent. This difference, or sampling error, occurs by chance, and its variability is measured by the standard error of the estimate. In addition, nonsampling errors can occur at any phase of the survey process. Examples of potential sources of nonsampling error include failure to identify and obtain responses from all persons within samples households, inability or unwillingness to provide accurate information, errors made in recording or entering data, and the uncertainty involved in estimating values for missing data.

Estimates of standard errors for major employment and unemployment estimates can be found in Employment and Earnings, along with procedures for computing standard errors for other statistics. These estimated standard errors are, primarily, measures of sampling error but also partially reflect some nonsampling errors.

Estimates of nonsampling error cannot be directly computed. Therefore, our focus is on preventing or minimizing nonsampling errors, and monitoring statistics indicative of nonsampling errors, e.g., item nonresponse rates and refusal rates. Each month, various summary reports are generated to monitor production processes. Field supervisors, regional office staff, centralized telephone center managers, and coding supervisors monitor response rates, coding referral rates, and other process measures. While individual performance is monitored to determine if additional training is needed, the review is also intended to detect problems with current procedures and areas to be targeted for quality improvement.

For example, we once observed a rise in supplement noninterview rates. To prevent this from becoming a serious problem, we formed a team of Washington and regional office staff, who examined data and enacted several changes to operating procedures. All interviewers received a memorandum on the importance of maintaining high response rates for both the labor force and supplement portions of the interview; some interviewers received personal counseling. A new set of monthly reports was developed, and interviewers began receiving monthly feedback.

Source and Accuracy Statement

Basic Monthly Survey Methodology and Documentation Page

CPS Main Page