Saanich-Gulf Islands Poll - Oversampling recalculation
According to Elections Canada voting data for the 2008 election, 82.5% of Saanich-Gulf Islands voters resided in Saanich and 17.5 per cent on the Gulf Islands.
Due to an oversampling in the poll from the Gulf Islands (37.9 per cent of all respondents), GPS reweighted the results in our reporting for questions two (generic party result) and five (candidate with party).
.
This was done to ensure that the results from these two questions more accurately reflected the demographic breakdown of SGI and the voting intent of respondents at the time of this poll in our reporting of it, as there are significant differences in party and candidate support between the two sub-regions of the riding. After reweighting, the two results are below:
|
Q3 Crosstabs with 17.5% Gulf Island weighting adjustment
If a federal Election were held today, which Party would you most likely vote for?
|
| |
Liberal |
NDP |
Conservative |
Green |
|
| Full riding |
37.0% |
18.6% |
25.3% |
19.1% |
100.0% |
|
|
Q5 with 17.5% Gulf Island weighting adjustment
If a federal Election was held today, which of the following Party’s candidates would you vote for in your
Riding of Saanich Gulf Islands?
|
| |
Hetherington
Liberal |
Loring-Kuhanga
NDP |
Lunn
Conservative |
May
Green |
|
| Full riding |
26.4% |
14.9% |
30.4% |
28.3% |
100.0% |
|
The results within the report which follows have not been changed and as such do not include GPS’s adjustment for oversampling. We encourage readers to review the cross tabulations to have a more accurate sense of intentions.
GreenPartyStrategy.com
Table of Contents
| Methodology & Logistics |
3 |
| Executive Summary |
4 |
| Results by Question |
12 |
| Crosstabulations |
23 |
Methodology & Logistics
The following represents the results of a public opinion survey of voting age residents of the federal Riding of Saanich-Gulf Islands conducted for GreenPartyStrategy.ca.
Study Sample/
A total of 400 voting age residents of the Saanich-Gulf Islands were interviewed in the survey. Interviews were conducted between the days of August 5th and August 10th 2010.
The margin of error for a (400 person) survey of this nature is +/- 4.9%, 19/20 times.
Survey Method
The survey was conducted using computer-assisted techniques of telephone interviewing (CATI) and random number selection.
A total of 30% of all interviews were monitored and the management of Oraclepoll Research supervised 100%.
Initial calls were made between the hours of 6:00 p.m. and 9:00 p.m. Subsequent callbacks of no-answers and busy numbers were made on a (staggered) daily rotating basis up to 5 times (from 10:00 a.m. to 9:00 p.m.) until contact was made. In addition, telephone interview appointments were attempted with those respondents unable to complete the survey at the time of contact. If no contact was made at a number after the fifth attempt, the number was discarded.
Executive Summary
Top of Mind
All respondents were first asked in an open ended question what they felt was the most important issue facing Canada at this time.
The most important (first named) top of mind issue 34% of residents related to the economy, followed by 15% that named the environment, 7% poverty / the homeless, 6% taxation, 6% leadership and 5% health care. A total of 5% did not know or had no opinion.
When those with an opinion were asked about another issue, 9% named the environment, 8% the economy, 7% leadership, 7% poverty, 6% immigration and 6% health care. 25% were unsure or did not know of a second issue.
Vote Intent
Respondents were first asked about their voting preference considering only the federal Party. A total of 35% of decided voters said they would vote Liberal, 26% Conservative, 21% Green and 18% NDP. A total of 18% were undecided.
When decided voters were asked bout their second choice party, 28% said they would back no other Party 24% would vote Liberal, 20% NDP, 12% Green, 9% Conservative, while 8% were unsure.
The following table illustrates how decided voters would cast a ballot for second choice.
Next, voters were asked about their voting preference considering both candidate and Party.
“If a federal Election was held today, which of the following Party’s candidates would you vote for in your Riding of Saanich-Gulf Islands?”
IF DO NOT KNOW ASK:
“Which Party’s candidate would you be leaning towards?”
When considering both candidate and Party, there is almost a three way split among decided voters, with those backing the Green-May ticket slightly leading the Conservatives and Lunn, followed by the Liberals and Hetherington.
Decided voters were also asked about their second choice of which, 32% would back no other Party and candidate, 23% would support the NDP ticket, 17% Liberal, 8% Green and 6% Conservative, while 14% did not know.
The following table illustrates how decided voters would cast a ballot for second choice.
Gary Lunn
All respondents were asked the following.
“Using a scale from one not at all familiar to five very familiar, how would you rate your familiarity with the federal Conservative
Party candidate in your Riding, Gary Lunn?
A total of 45% of residents were familiar with Gary Lunn, 20% had a neutral opinion (neither familiar nor unfamiliar) and 29% were not familiar with him.
Those familiar or neutral with Gary Lunn were then asked to rate their overall opinion of him of which 36% had a favorable view, 34% were neutral, 24% had a negative opinion and 7% did not know.
Elizabeth May
All respondents were asked the following.
"Using a scale from one not at all familiar to five very familiar, how would you rate your familiarity with
the federal Green Party candidate in your Riding, Elizabeth May?"
A total of 29% were familiar with Elizabeth May, 20% neutral (neither familiar nor unfamiliar) and 48% were unfamiliar, while 3% did not know.
Those familiar or neutral with May were then asked to rate their overall opinion of her. A total of 61% said they had a favorable opinion, 16% a poor one, 20% neutral and 2% were unsure.
Direction of Canada
All respondents were then asked about where they thought that Canada under the Conservative government of Stephen Harper was headed.
“…would you say that Canada is headed in the right direction, or is Canada off in the wrong direction?”
More voters were of the opinion that things are headed in the wrong direction compared to those that said in the right direction.
When asked about the Party that they most trusted to run Canada, 30% said the Conservatives, while the Liberals trailed with 19%, closely followed by the NDP at 18% and the Green Party with 18%. A total of 12% said no Party and 3% did not know.
Party – Best Job
All respondents were asked about which political Party they felt would
do the best job at each of the following.
The Green Party fared best with respect to representing the needs of the Riding and poorest in relation to making realistic promises.
Statements
Respondents were also asked to rate their level of agreement with two statements.
- A total of 46% of all respondents agreed that
it would be good to have a Green Party MP in Ottawa
- A 57% majority agreed
that it would be good to have Elizabeth May as an MP in Ottawa
Results by Question
Q3a. If a federal Election were held today, which Party would you most likely vote for?
IF DO NOT KNOW IN Q3A ASK:
Q3b. Which Party would you be leaning towards?
IF DON’T KNOW SKIP TO Q5a
IF DO NOT KNOW IN Q3 SKIP TO Q5
Q5a. If a federal Election was held today, which of the following Party’s candidates would you vote for in your Riding of Saanich-Gulf Islands?
IF DO NOT KNOW IN Q5A ASK:
Q5b. Which Party’s candidate would you be leaning towards?
IF DO NOT KNOW IN Q3 SKIP TO Q7
Q6. What Party’s candidate would be your second choice?
Crosstabulations
 |
Comments
Graphs in the wrong place
FYI, it appears that the graphs showing familiarity and direction of Canada are listed below the wrong questions.
Jim Johnston
Thanks Jim
I believe we hae it fixed now.
Dermod
Not big enough sample
I could only wish that there were 1,000 respondents. Yes, I know, it costs a lot of money, but you are cloaking the paucity of respondants with a lot of admittedly intrigueing, but seriously misleading statistics. Even splitting out the male/female numbers leaves you with too small a sample to draw any conclusions.
The SGI campaign should replace their phone canvas with a survey like this, and get a few thousand responses. Who knows, maybe they will, or already did. All those staffers must have being doing something with $300,000+ ...
M Day
a sample of n=400 is a better use of resources than n=1,000
Hi Anonymous, I disagree that the statistics in this poll are misleading. I am a professional research analyst (at a different agency) and the sample size here is perfectly fine. Even with the breaks n=400 is statistically defensible.In the US, we will often use n=500 to get read of the entire country of 300 million people. Unless you have a lot of money to throw around, n=400 is a sound decision for Saanich Gulf Islands.
You are very right that sound polling intelligence is critical to winning an election. But perhaps the limited resources of the Green Party are better spent on two polls of n=400 over time to track change rather than a single n=1,000 sample. Even better, I would like to see four polls of n=400 over time and some focus groups. But hey, all in time...
If you like getting bored with numbers, below is a link to a statistical calculator for polls.
http://www.raosoft.com/samplesize.html
You'll see that the margin of error of a n=1000 poll at a 95% confidence level is 3.1 percent. But the cost would be thousands of dollar higher than a n=400 sample.
And while the margin of error of an n=400 poll is plus/minus 4.9 percent at a 95% confidence level and 50/50 break -- it is actually plus/minus 3.8 percent at a 90 percent confidence level with response distribution of 30/70 like the numbers in the above company poll. This is pretty good for strategic use.
Sample size, n=400 is fine for the big question but
Oct 6 Anon, I agree that a sample size of 400 is fine for the big question, but when you start looking at the smaller subsets, you are drifting into very small numbers. Assuming appr. 50/50 male female split, drawing conslusions about female populations based on 200 samples is getting shaky. Then further dividing by income, or various other factors, you simply cannot trust the conclusions you draw. Better off to focus group, and at least have some in depth 'anecdotal' evidence to draw on. As far as multiple polls of 400, absolutely, if you want to track voting intentions, and a real honest to goodness campaign, that's the way to go. Just save yourself some money, and do the polling in-house, provided you can draught the questions, and actually manage the samp[ling and phone work and gather the 20 or 30 volunteers needed.
M. DAY
Post new comment