In the mid-1980's, the Bureau of Labor Statistics (BLS) pioneered the application of touch-tone data entry (TDE) to business establishment surveys. The success of TDE to improve response rates and timely reporting for the Current Employment Statistics (CES) has been well-documented (Clayton et al, 1989). More recent work has documented increased respondent satisfaction and reduced survey costs for TDE reporters in the CES (Rosen et al, 1993).
The success of TDE for establishment surveys, not only at BLS but the Bureau of the Census (Bond et al, 1993), led us to explore the possible application of this mode of data reporting to household surveys. In late 1991, work was begun on a pilot TDE instrument for household survey reporters. The BLS survey utilized for this research was a short form of the redesigned Current Population Survey (CPS). The wording of the questions in the short form of CPS is identical to that in regular CPS.
The short form of the CPS questionnaire asks the minimum number of questions necessary to assign household members to one of three labor force categories: Employed; Unemployed; Not In Labor Force. The shortest possible series, for those who did any paid work last week, is 2 questions long. Approximately 58% of respondents will be classifiable as to labor force status after answering the 2-question series. The CPS Touch-tone Data Entry program allows the respondent to answer open-ended questions as well as those be answered by "yes" and "no." Cognitive research on the pilot version indicated that respondents of widely-differing educational status followed the same rules as CPS interviewers in coding their open-ended responses. After the respondent has answered the minimum number of questions necessary to be classified as to labor force status, the program assigns labor force status to that respondent.
Thus, if a respondent indicates that he/she could not have returned to work last week, the program will request that the respondent state the reason for this inability*. After speaking the reason into the telephone, the respondent will be asked to self-code his/her open-ended answer as to why he/she could not have returned to work. If the respondent codes the open-ended answer as meaning that he/she was ill, and presses "1," the program classifies the respondent as "Unemployed." If the respondent codes the answer as "2," a reason other than illness, the program classifies the respondent as "Not In Labor Force."
The first phase of cognitive laboratory research on the CPS-TDE instrument was conducted in 1993. Nineteen respondents were interviewed using the short form of the CPS questionnaire, following which the respondents were asked to complete the same survey by keying their responses to a telephone version of the interview. Two respondents unwittingly pressed incorrect keys during their data entry, found themselves on the wrong paths, and were not able to complete their reports. In addition, respondents had a great deal of difficulty in recognizing the "line number" when using the "TDE Reporter's Guide" to report their household roster. The TDE instrument and "TDE Reporter's Guide" were revised to correct these problems.