Research Method返回
This "Audience Survey on Hong Kong Harbour Fest" was conducted between 6 and 11 November 2003. On 6, 7 and 9 November 2003, a total of 9 workers were deployed by the research team to hand leaflets to the audience of the relevant Harbour Fest concerts.
| |
The leaflets, each printed with unique code for identification purpose, were distributed at three locations, namely, the main entrance of the Tamar Site, the Citic Tower, as well as the joint of Tim Mei Avenue and Lunch Wui Road. A total of 12,065 leaflets were distributed, which invited the Harbour Fest audience to visit the prescribed website http://hkupop.pori.hk/harbourfest to complete an online survey.
| |
As the deadline for submissions was set to be 3 days after the last concert of the Harbour Fest, which was held on November 9, 2003, late submissions after the midnight of November 11 were not processed, except for 5 respondents who had made a personal request to the research team via email or phone call, with personal contact and specific reasons provided. A total of 902 submissions were kept as preliminary valid cases.
| |
Verification procedures were then performed by the research team on the 902 preliminary valid cases. First, 17 cases without personal particulars were dropped, and the remaining 885 cases entered the next stage.
| |
For the 8 cases with repeated leaflet codes, only the last submissions of each case was kept. Another 156 cases with null leaflet codes were also dropped, leaving behind 721 valid cases.
| |
The 20 skeptical cases with repetitive IP addresses were then examined in terms of their timechop, nature of their outgoing website, and content of their submissions. Finally, 3 cases were confirmed to be repetitive and were eliminated from the dataset. The remaining 718 submissions were regarded as the final sample size (Figure 1).
| |
Figure 1. Verification procedures for online submissions 
| |
As shown from Table 1 below, the overall response rate of this online survey was calculated to be 6.0%, and the standard sampling error for percentages based on this sample was less than 1.9 percentage points. In other words, the sampling error for all percentages was less than plus/minus 3.8 percentage points at 95% confidence level. However, due to the very low response rate, significant systematic bias was bound to have occurred, and the findings reflected herewith would not be representative of all Harbour Fest concert-goers.
| |
Table 1 Calculation of overall response rate
|
