The Proof-of-Concept test was designed to test the Gemini redesign plan of the Consumer Expenditure Survey (CE) that was released in 2013. Details on the redesign can be found in the above report and on the Gemini webpage.
Prior to moving forward with further research regarding the components of the design plan, a Proof-of-Concept test was needed to ensure that the basic underlying structure and components of the new design were feasible. The Proof-of-Concept test (POC) was designed to mirror the proposed design to the fullest extent possible. It included the administration of one wave of the proposed design, consisting of a Visit 1 recall interview, electronic individual diaries for eligible household members with a paper back-up, and a Visit 2 records-focused interview. Additionally, the design used incentives as designed in the proposal report in order to motivate respondents with an expected reduction in the number of required contact attempts.
This page includes a description of the test design, and links to materials used during the test.
An advance letter was sent via priority mail, two weeks prior to each month of interviews. This letter included a $2 bill token cash incentive.
A personal visit was conducted via CAPI Interview that includes identifying household members and demographics, selected questions on recall-based expenditures, individual diary placement, and records training. Click on the links below to find the relevant interview questions or survey material for each section.
Upon the completion of all the sections in the first visit, the respondent received a $20 debit card from the Field Representative.
The day following the interview, all eligible household respondents were asked to keep an individual diary for one week. Eligible household members and their choice of mode (electronic or paper diary) were identified during the first visit. Each individual that completed a paper or electronic diary received a $20 debit card in the mail.
Materials for the Diary week for paper mode:
A personal visit conducted via CAPI Interview that included the pick-up and recall of diary expenses and sections focused on expenditures that can be found in records collected during the previous week.
The Proof of Concept Test (POC) was designed to ensure that the basic underlying structure and components of the original redesign of the CE Survey is feasible. The test was fielded July through October 2015 in four Census Regional Offices. The final response rate was 49.8% (defining completes as a complete recall interview, at least one expenditure from a diary, and a complete records interview). While lower than the comparable production response rates, the test performed better than previous field tests and met the target number of 520 completed cases.
Results indicate improvements in data quality as shown by fewer missing expenses, fewer expenses that needed editing and less rounding compared to Consumer Expenditure Survey production data. The POC sample had higher overall expenditures reported in the interview survey, though this was not significant after controlling for other factors. Although the test had diary expenditure totals that were not significantly higher, the number of personal diary entries were significantly higher than production even when controlling for other factors.
Report 1: Field results (Not publicly available). Results on training, field procedures, and interviewer debriefing: The report included recommendations for training, protocols, and design as well as preliminary response rate estimates for the Proof of Concept test.
Report 2: Results on Respondent Cooperation, Select Interview and Diary Characteristics, and Respondent Experience: The report includes a detailed analysis of BLS defined response rates, contact attempts, diary use, and analysis of interviewer and respondent debriefing questions.
Report 3: Results on Interview and Diary Expenditure Reports and Data Quality: This report analyzes the expenditure data from the interview and diary for the POC test and compares it to production data to measure any effects on data quality.
Last Modified Date: April 13, 2018