In recent years, low response rate and coverage has been important issues in traditional telephone surveys (CATI) using household landlines. One major reason is the increasing cell-only population who cannot be reached using household landlines. In Taiwan, change of life style and the fraud calls also hinder people from responding to surveys. Among CATI respondents, it is common to find a disproportional low share of people aged between 20 to 39 years old. A low share of younger respondents in telephone surveys may be attributed to their cell phone dependency for some, and night-shift or overworking jobs for others. Weighting or analytical methods are, therefore, needed to adjust the data for further inference.
While the inclusion of cell-phones in the sampling frame of CATI may increase the likelihood of reaching these young population, the additional cost often prevents its use. This study seeks an alternative to improve the contact and response rates of young respondents in CATI. Using a large national telephone survey as an example, we attempt to increase the probability of selecting eligible respondents aged 20-39 years old during the within-household sampling procedure. A parallel survey was conducted on an online panel, among whom a cell-only sample were identified using application software. Results of the CATI and APP surveys are reported. A special focus is on the comparability of survey results from the young respondents in the two samples; i.e. CATI and APP. The use parallel surveys in different modes as compromise to low shares of younger respondents in telephone surveys is discussed.