In the past many years, POP has been conducting popularity ratings for a variety of political figures and groups. The research design of the top 10 political groups is explained as follows:
The surveys are at present conducted once every three months, in two different phases. In the first stage, respondents were requested to nominate, unprompted, all political groups that they could think of. They could name up to 5 groups before June 1999, and up to 10 groups after that date.
POP would then select from this list of unprompted nominees the most frequently cited names for the next stage survey. Before June 1999, 7 groups would be admitted. The quota was increased to 12 after June 1999.
During the second stage survey, respondents would be asked to rate each of the selected groups using a 0-100 scale. 0 indicates absolutely no support, 100 indicates absolute support, and 50 means half-half. After calculation, the least well-known groups would be dropped. Only the ten most well-known political groups remained. (Before June 1999, only the top five political groups remained.) Finally, any group which failed to reach the 50% benchmark recognition rate would also be dropped (or retained only for reference). The list would then be sorted according to support ratings. In other words, those political groups outside the top 10 may have very high or very low support ratings. They are not in the list because they are not the most well-known groups among the public.
Using this method, the POP Team is able to screen out the most popular legislators and political groups both in terms of popular recognition and support. Those who failed to reach the top must have failed in either or both aspects.
To facilitate better understanding of our "Top 10" series, starting from September 2005, the POP Team would also include the following two pieces of supplementary information in each related release: (1) The ratings of political figures or groups being dropped in the final stage of the rating exercise due to their relatively low recognition rates, provided that they still attain 50% recognition rate; (2) Overall ratings ranked according to results obtained over the past 12 months. The groups' overall rankings were first determined by their number of times on top 10, and then their average ratings. In other words, the more frequent the groups on top 10, and the higher their average ratings are, the higher their overall rankings are.
It should be noted that because political groups are not yet legal entities in Hong Kong, such definitions are rather vague, and so-called political groups are constantly evolving. As a result, strange names may appear in the list of groups mentioned by respondents in Stage One surveys. In order to avoid personal bias, our research team will eliminate groups which fall outside the popular definition only after the first stage of the survey.
All stage one naming surveys and stage two rating surveys mentioned are telephone surveys conducted by interviewers under close supervision.
Target population of these surveys are Cantonese speakers in Hong Kong of age 18 or above.
Starting from mid-January 2014, telephone numbers are randomly generated using known prefixes assigned to telecommunication services providers under the Numbering Plan provided by the Office of the Communications Authority (OFCA). Invalid numbers are then eliminated according to computer and manual dialing records to produce the final sample.
Before mid-January 2014, telephone numbers are first selected randomly from telephone directories as seed numbers. Another set of numbers is then generated by using the plus/minus one/two method. Duplicate numbers are then screened out, and the remaining numbers mixed in a random order to become the final sample.
Upon successful contact being made with a target household, one member of the household is selected among those present using the "next birthday" rule.