1。是取样分布不均。选择范围太区域,打电话的工作人员喜欢顺着打,随机性不够。
2。是取样不够大。选择范围太小。这么个大面积的国家,取样1200人,太少。若打1万2千,结果可能完全不一样。
3。误差太大。你看BC省的技术误差是8%,这个民调还有意义吗?
4。大家的阅读有问题,一般只看倾向性支持率,即第一节说的那些数字,而不看实绩已决定投票率。例如,保守党的支持率30.8%,但已确定投他的只占39%, 也就说已确定投他的占全部被访人员的30.8% x 39% (即15.1%)。倾向性支持者中间的不确定性还有61%,这就是说,3分之2的的倾向支持者还不一定投他。这就使得名调的结果将与实际的投票结果会有巨大的差距。
5。实绩当选差额很小。例如安省John Carmichael的上次当选, 只比对手多了1.14%,仅多了661票,这比民调承认的安省误差率达5.6%相差十万八千。
Liberal | Rob Oliphant | 22,351 | 41.79% |
Conservative | John Carmichael | 22,962 | 42.93% |
-----------《 加拿大大选第二次辩论后的民调出炉》
周四晚间第二次辩论结束, CTV 和The Globe and Mail委派Nanos机构作最新民调,截至6个小时前,对1070名合资格选民实地或电话问答,误差3%,有效问答比率19:20, 结论是:
--- 保守党 the Conservatives, 30.8%
--- 自由党 the Liberals, 30.3%
--- 新民主党 NDP, 29.1%
--- 绿党 The Greens, 5.1%
--- 魁省党 the Bloc, 3.7% (Quebec only).
问的是:“If a federal election were held today, could you please rank your top two current local voting preferences?"(如果今天投票,你会分序投哪两个党?”
按地区分布:
- 在大西洋沿岸省份,The Liberals领先,53.3%, NDP 29.0%坐二
- 在Quebec, the NDP领先,占45.7%
- 在中部的草原省份,the Conservatives领先,占50.5%
- 在大票藏Ontario省, the Liberals 37.3%,the Conservatives 36.0%,NDP 21.1%
- 在西部的BC/British Columbia省,The Conservatives 领先,35.5%, NDP 殿二 30.8%.
- 地区误差:BC省: 8.0%; 草原省份: 6.7%; Ontario省: 5.5%,Qebec省: 6.1%, 大西洋省份: 9.6%
当问到“在你支持的党中,你已经确定了一定要选这个党的?”(占该党支持率的),定了的:投自由党的49.0%,领先;会投NDP的45.5%;会投保守党的39%;会投绿党的24%;在魁百克省的支持者中,定了要投Bloc Quebecois的占26.3%。 --- 这些都是各党的铁军。而其他支持者只是算倾向票,随时可能会变。
Second Choice
When asked to rank their second choice:
- Of those who ranked the Liberals first, 53 per cent picked the NDP second.
- Of those who chose the NDP as their top pick, 49 per cent said the Liberals would be their second choice.
- Among those who favoured the Conservatives, 50 per cent said they have no second choice, 30 per cent said they would pick Liberals as their second choice, and 13 per cent said they would pick the NDP as their second choice.
谁当总理最合适
-- 保守党的Stephen Harper(哈帕)30%
-- 自由党的Justin Trudeau(小特鲁)26.6%
-- NDP的Tom Mulcair (汤玛凯)25.5%
-- 绿党的Elizabeth May(梅)5.2%
-- Qbec党的 Gilles Duceppe (杜塞)1.9% (在Quebec only)
-- 10.8%的被问者说:还没确定
Poll Methodology:
Nanos Research在第二次辩论之后的三天内,在全国实地或电话访问了1200人,1070是合资格选民。
Nanos Research分成了三个组,每组访问400人。