注意事項Points to Note返回
注意事項
 | 特首評分調查所用的提問方式一般如此：「請你用O至1OO分評價你對特首董建華既支持程度，O分代表絕對唔支持，1OO分代表絕對支持，5O分代表一半半，你會俾幾多分特首董建華呢？」此種提問方法有別於西方慣常採用之「假設明天投票你是否支持某人？」的方法。後者一般適用於普選產生的政治人物。民意研究計劃在九一年開展有關調查時便考慮了有關因素。 | |
| 本網頁數表上所載的標準誤差是根據評分數值的分佈統計而得，與一般抽樣誤差不同。在九成半的置信水平下， 實際的評分數值應在所載平均數值正負兩個標準誤差之間。「九成半置信水平」的涵義是在二十次推算中理應出 現一次失誤。綜觀本研究計劃的各次特首評分結果，在九成半的置信水平下，評分誤差都能控制在一分半至兩分之間。 | |
| 網頁上按季計算的數表及圖表是根據有關季度內各月份的平均數字再次平均計算，即平均再平均。 | |
| 由於資源所限，民意研究計劃的成員只能向各界人士解答有關調查所採用的方法，而不會就有關數據另作評述。 |
Points to Note
|
The question used in these surveys, in general, is: "Please rate your support of the Chief Executive Tung Chee-hwa using a 0-100 scale. 0 stands for absolutely no support, 100 stands for absolute support, 50 stands for half-half. How would you rate the Chief Executive Tung Chee-hwa?" | |
|
The standard figures given in the tables herewith have been computed according to the distribution of the rating figures. They are different from the sampling errors of percentages normally reported in other surveys. At 95% confidence level, the real figures should be within plus/minus two standard errors from the average rating figures reported. "At 95% confidence level" means that one out of twenty projects should be wrong. According to our figures, the possible error of CE's rating across our surveys should be within 1.5 to 2.0 points at 95% confidence level. | |
|
The quarterly figures presented in the relevant tables and charts are using average figures calculated from monthly averages. They are averages of averages. | |
|
Due to limited resources, our research team members will only be able to entertain questions on our research design, but not request for comment on the figures. |
