The future of telephone surveys in Hong Kong

Abstract

Purpose This paper aims to compare residential fixed-line telephone surveys with cell phone surveys for assessing the extent of the potential undercoverage issue evaluating the necessity and feasibility of conducting cell phone surveys or dual-frame telephone surveys in Hong Kong. Design/methodology/approach The research team simultaneously carried out a conventional fixed-line telephone survey and a cell phone survey in 2015 with similar features on survey design, sampling and data collection procedures. Two samples with sample size of 801 and 1,203 were achieved separately. Data collected were analysed to see to what extent survey findings will be biased if the sampling frame of telephone surveys is solely based on residential fixed-line numbers in Hong Kong, and if such a bias does exist, whether a survey conducted through cell phones or by adding a cell phone-only (CO) group would be an ideal solution for it. Findings The findings show that the cooperation rates for the cell phone survey were much lower than those of the fixed-line telephone survey. The respondents from two surveys were fairly different. However, estimates of most commonly used socio-demographic characteristics from the latter group had the least bias compared with population statistics. Supplanting the traditional fixed-line survey with a cell phone survey or supplementing it with a CO group will not make the resulting sample a better representative of the population but it will amplify the sample bias on the major social socio-demographic characteristics. Originality/value This paper empirically compares the two types of telephone surveys in a well-designed scientific study.

The association of problematic smartphone use with family well-being mediated by family communication in Chinese adults: A population-based study

“…Perceived sufficiency and quality of family communication partially mediated the association of PSU with family well-being. We observed that the mobile sample included more young, highly educated, unmarried, employed, and high-income respondents, which was consistent with the findings of another study comparing two sampling frames in Hong Kong (Chiu & Jiang, 2017). The traditional random-digit-dialing landline survey may raise the undercoverage issue as the method cannot reach the increasing mobile-phone-only population (Chiu & Jiang, 2017;Keeter, Kennedy, Clark, Tompson, & Mokrzycki, 2007), which can be partially supported by our findings of more smartphone owners in the mobile sample than the landline sample (93.5% vs. 76.5%; p < .001).…”

Health of Youth in Transition in Hong Kong

“…This low response rate might be due to the length of survey (about 1 hr) and participants' inclusive criteria (17-23 years old). Therefore, our response rate was lower than that (about 50%) of other Hong Kong telephone surveys targeting the general population (i.e., 18 years or older) [27]. Still, our study's response rate was closer to that of another study [33] using telephone survey in Hong Kong (17%) and within the range of response rates for telephone polling in the US [34].…”

“…The survey utilized a computer-assisted telephone interviewing (CATI) system for selecting local household telephone numbers. Chiu and Jiang [27] did a study comparing residential fixed-line telephone surveys with cell phone surveys to assess the extent of the potential undercoverage issue in Hong Kong. Results found that with increasing the number of mobile phone-only households, the household telephone survey was still a better representative of the Hong Kong population as most estimates from the telephone survey sample were very similar to the population statistics.…”

Physical, mental, and behavioral health after adverse and benevolent childhood experiences: A comparison of two adversity conceptualizations.

“…Third, a telephone survey was conducted to eliminate household contact during COVID-19. Although the dual-frame approach (landline and mobile) may result in sampling bias on demographic characteristics (Chiu & Jiang, 2017), the age and gender distributions of the present sample closely approximate those of the population census (Hong Kong Census and Statistics Department, 2022). Last, this cross-sectional survey precluded drawing conclusions about causality.…”