A continuing decline in the use of domestic landline phones has cast the spotlight on a decades-old practice of pollsters using only home numbers to gather responses. Residential landlines are looking less and less attractive to locals, thanks to the rise of mobile devices, but collecting data solely through mobile numbers has its own statistical problems, researchers warn. That is due to the prevalence of mobile phones among non-locals and the likelihood that some users own more than one number, meaning their chances of being called for a telephone survey are higher, according to Dr Chung Kim-wah, director of Polytechnic University's Centre for Social Policy Studies. "It is difficult to confine the population frame if we call mobiles," he said. "It is not yet time to launch mobile-based polling." The number of mobile phones in service hit 12.3 million at the end of last year, the Communications Authority says. The data also showed the number of residential landlines dropped over the past decade, from 2.14 million in 2006 to 1.48 million in March this year. But Chung noted almost 80 per cent of the population still had access to a landline. In the United States, major pollsters retained their reliance on home numbers, he added, although the penetration rate was even lower - at only 60 per cent of the population. Chung said his centre had tried four times since 2013 to conduct polls via mobile numbers, but the process had been tough. "Sometimes, respondents would ask whether we could pay their mobile charges as their fees were calculated on a call-by-call basis," he said. University of Hong Kong lead pollster Dr Robert Chung Ting-yiu agreed, saying the current methodology was "so far so good". "It would not be wise to introduce new changes when running political reform surveys," Chung, director of the HKU public opinion programme, said. But pollsters could move beyond interviews via landlines and mobile phones, he said, by adopting voice-cum-text surveys incorporating landlines, traditional mobile phones and smartphones. That could be the next phase in public polling, he suggested. Ahead of a legislative vote this week on the government's proposal about a popular ballot for the 2017 chief executive election, PolyU, HKU and Chinese University have been jointly conducting a rolling survey to determine the level of support for the blueprint.