Abstract
In December 2016, NHK introduced a dual frame telephone survey that used mobile phones in addition to conventional land-line phones. This survey method does not weight samples to re-balance the data set based on sampling probabilities and demographics such as genders and ages but calculates the survey results by simply aggregating the figures from the land-line and the mobile surveys. However, we conducted an experimental survey in March 2019 to examine whether it would be appropriate to keep this approach even if the number of land-line owners further decreases and that of mobile phone owners further increases in the future. This was the second examination since December 2016 when the dual frame survey was introduced. The key results are as follows. - We were able to collect more responses from young male respondents in the mobile survey than the land-line survey, due to which the sample composition ratio by age and gender in the dual frame became closer to that in the census.
- There was almost no difference in response tendencies between the land-line survey and the mobile survey. - As to the ownerships of land-line phones and mobile phones, the result of the dual frame calculation without weighting was closer to the result of a public opinion survey (drop-off and pick-up method). - We compared the results of the dual frame calculation and three types of weight sample calculations and found little difference in the results for most of the questions. The above findings indicate that there is no need to change the conventional approach to the dual frame survey using land-line and mobile phones—simply adding up results from the land-line survey and the mobile survey without weighting—for the time being.
