Menu Close

Comparing the Regional Chair Survey to the Election Results

Anthony Piscitelli

From October 16 to 17, 2018, ThreeHundredThirtyEight.com conducted a survey to assess support for the candidates in the Waterloo Region Chair race. The results showed Karen Redman in the lead but a large number of voters undecided. Ultimately, Karen Redman was successful during the election held from October 22 to 23, 2018. This post looks back to assess how accurate the survey was at predicting the election results.

In the original reporting of the race, we reported a margin of error of +/-4.25%. For simplicity sake, when reporting a margin of error a single value is typically shared. However, the calculation for margin of error actually varies based on the observed proportion. Results close to 50% have higher margins of error than results close to 10%. Forum Research breaks down the rough margins in a handy table by sample size and observed proportions. For example, according to Forum Research’s table, with a sample size of 300, the margin of error can vary between 3.4% (at 10% or 90% proportion) and 5.7% (at a 50% proportion).

In this post, one margin of error per sample is reported calculated at the 95% level (i.e. the results are considered accurate 19 times out of 20). However, in the commentary assessing the accuracy of the results the margin of error for the individual proportions were calculated using an online calculator.

The obvious place to start is to assess the accuracy of the top line results as reported on October 18, 2018. Here we see that the results overall did quite well. The results of each candidate are within the margin of error except for Jan d’Ailly who slightly outperformed. His margin of error was 2.8 percentage points, yet he received 9.7% of the vote, a result 3.0 percentage points above his 6.7% predicted support. The tracking error on this model also performed quite well at 8.4 percentage points. The tracking error was calculated by taking the election results and subtracting them from the survey results, then adding the absolute value of each of these numbers.

The reported results included leaning and decided voters. It is also possible to compare only using decided voters.  Once again in this approach, all results except for those involving Jan d’Ailly are within the margin of error. However, in this model, the tracking error increases to 11.2 percentage points.

A model was also created to predict likely voters. In this model, the results do not work as well. Both Karen Redman and Jan d’Ailly are outside of the margin of error in this model and the tracking error increases to 15.3 percentage points. Interestingly, using only unlikely voters all candidates results are within the margin of error. The small sample size for this group increases the margin of error. The tracking error amongst unlikely voters is 13.1 percentage points.

When the results of leaning and likely voters are broken down by city/township they all fall within the margin of error. However, it should be noted some of these sample sizes are very small creating very large margins of error. It is worth noting with respect to Karen Redman, the Cambridge results were at the edge of the Margin of Error at the 95% level.

The tracking error was lowest in the townships at 2.7 percentage points, followed by Kitchener at 5.9 percentage points, then Waterloo at 9.3 percentage points, and then Cambridge at 18.5 percentage points.

One final comparison was made. The results reported publically were weighted by age, gender, and city/township of residence. However, it is also possible to compare the unweighted survey results to the actual election results. This approach finds all of the results well within the margin of error and a tracking error of 4.8.

Overall the results of the survey were a fairly good predictor of the actual election results. Indeed, even the breakdown by city/township showed results that were a reasonable predictor of the actual election results. However, the likely voter model was a poor predictor of the election results. It is fortunate this model was not used. It is also interesting to note that weighting the variables did not improve the predictive power of the survey.



Three Reasons Why the Regional Chair Poll May Not Predict the Election Results

Anthony Piscitelli

Election polling results are interesting because we like to use them to predict an election. However, polls represent a snapshot in time so extrapolating results to a future event may lead to faulty predictions. With respect to the Regional Chair race poll, we released this week the need for caution is even greater, as this was a single poll in a low turnout election with a response rate of 0.2%. Three issues warrant particular consideration.

1) The Survey Could be a 1 in 20 Outcome

Public opinion surveys are an attempt to ascertain what the population believes about a topic by asking a small group of people. When reporting a single poll a margin of error is typically given at the 95% confidence level, indicating a range of plus or minus a few percentage points within which the population’s belief actually falls. So, for example, in the case of the poll for Regional Chair, 36.5% of respondents were undecided with a range of plus or minus 4.3%. Meaning statistically the poll predicts somewhere between 32.2% and 40.8% of voters were undecided 19 times out of 20. This caution of 19 times out of 20 is an important one to note as statistically speaking, even if the poll is a perfectly random sample 1 in 20 times the actual result is expected to fall outside of the margin of error (i.e. 1 time in 20 the poll will simply be wrong).

The best defence against this problem is multiple polls on the same topic by multiple polling firms. A single poll showing a result may be an outlier, multiple polls showing the same result is unlikely. Multiple polling firms tackling the same topic also decreases the likelihood that a bias may be built into the sampling procedure (i.e. how people are selected to participate in the poll). Yet, even when multiple poll results show the same result, the polls may not be predictive of future events. The most recent presidential election in the United States provides a case study in the need for caution when extrapolating from polls to a future election.



2) The Undecided Voters

The survey revealed that 36.5% of voters have yet to make up their mind. These individuals were first asked who they would support if the election were held today, then asked if they were leaning towards a particular candidate. Asking vote preference twice is considered best practice and tends to capture even soft supporters for a candidate. The 36.5% in this survey who were undecided, are therefore individuals who are likely quite open to being persuaded by at least two of the candidates. An additional 8.7% said they would prefer not to share their preferred candidate, meaning we have no idea how they will vote. Finally, of those who indicated a preference, 8.0% were only leaning towards a candidate (i.e. when first asked who they would support they said they did not know). These results indicate that over 50% of voters preferences are either unknown or open to changing. Undecided voters could therefore dramatically change the election results. How dramatically? Extrapolating from these numbers, and assuming the poll results are accurate and not an outlier, Karen Redman’s support on Election Day could fall anywhere between 27.6% and 89.3%.

3) The Sample May Not Be Accurate

It is also possible that the results of the poll may be off because the sampling method introduced an unknown error into the process. The sample of landlines was created by purchasing an electronic list of listed landlines from www.telephonelists.biz. Landline numbers were then randomly sampled from this list. In addition, a list of likely cellphones and unlisted landlines was added into the sample. This list was created using data published at www.cnac.ca on the area code and exchange (NPA & NXX). The data from the Canadian Number Administrator can be used to ascertain the first six digits of phone numbers that when originally activated belonged to someone who activated a phone within Waterloo Region. However, with number portability, it is possible to keep your phone number when moving into or out of Waterloo Region. The survey revealed 6% of respondents no longer live in Waterloo Region. These individuals were excluded from the final results. However, there was no correction made to introduce people with cell phones originally activated outside of Waterloo Region. It is impossible to know what percentage of phone numbers were excluded from the survey because they were not included in our sample and it is possible that these individuals support different candidates than those included in the survey.

The second problem with respect to the sample is those who choose to participate in the survey. The response rate was less than one-quarter of a percent. It is possible that the 99.8% of the population that did not complete the survey are different than those who did participate. A total of 86.5% of people called did not answer the phone. It is not possible to know if these people are somehow different. Perhaps the reason they were all busy on a Tuesday and Wednesday evening makes them predisposed to a particular candidate. There is no way to know. Of the 13.5% of people who did answer the phone when called; only 1.5% completed the entire survey. Again, it is not possible to know if these people are somehow different. Perhaps these people exhibited a Shy Tory effect and were unwilling to participate because they did not want to admit their preferences to a (liberal) Conestoga College professor.

Karen Redman Leads in Waterloo Region Chair Race

Anthony Piscitelli

According to a survey of 530 Waterloo Region respondents conducted on October 16 and 17, 2018, by Anthony Piscitelli of ThreeHundredThirtyEight.com  Karen Redman has a large lead in the race for Waterloo Region Chair but 37% of voters remain undecided.

Undecided Voters 37%
Decided and leaning Voters 55%
Prefer Not to Answer 9%

The race for Waterloo Region Chair is between four candidates: Jay Aissa, Karen Redman, Jan D’Ailly, and Rob Deutschmann. Before this survey, there was no public polling data available to indicate the popularity of any of these candidates.

Amongst respondents who indicated a preference Karen Redman (67%) is in first, Rob Deutschmann (18%) is in second, Jay Aissa (9%) is in third, and Jan d’Ailly (7%) is in fourth (note results do not add up to 100 due to rounding).

Candidate Decided Voters
Jay Aissa 9%
Jan d”Ailly 7%
Rob Deutschmann 18%
Karen Redman 67%

The survey also asked respondents what is the most important issue to them in this election. Keeping taxes low (23%), improving housing affordability (22%) and increasing social services (16%) were the top three answers.

Issue Support
Keeping taxes low 23%
Improving housing affordability 22%
Increasing social services 17%
Lowering crime 7%
Improving public transit 6%
Building more roads 3%
Improving cycling infrastructure 3%
Something else 10%

Candidate Comparisons

When comparing support for the candidates by gender, statistically significant differences in support were discovered. Karen Redman was the most popular candidate among males and females but her support was stronger among females (79%) than males (58%). Jay Aissa showed the largest discrepancy in support with males (14%) much more supportive than females (1%).

Sample (n) Jay Aissa Jan D’Ailly Rob Deutschmann Karen Redman
Female 144 1% 4% 15% 79%
Male 151 14% 8% 21% 58%

Comparing support for candidates to support for the top three issues also revealed statistically significant differences. Once again, Karen Redman remains the most popular candidate when looking at all issues. However, her support is higher for those who also support increasing social services (84%) and improving housing affordability (71%) than for those who support keeping taxes low (46%).  Those who support keeping taxes low gave more support proportionally to Rob Deutchman (27%) and Jay Aissa (20%). Jan d’Ailly’s support was consistent across all three categories (7%).

Sample (n) Jay Aissa Jan d’Ailly Rob Deutschmann Karen Redman
Keeping Taxes Low 71 20% 7% 27% 46%
Improving Housing Affordability 76 5% 7% 17% 71%
Increasing Social Services 55 0% 7% 9% 84%
Other 85 7% 5% 19% 70%

The results also show statistically significant difference between likely and unlikely voters. A likely voters model was created by combining those who said yes or not eligible to the question asking “Did you vote in the 2014 municipal election” with those who also indicated they were certain or likely to vote on the question asking how “On election day are you certain to vote, likely, unlikely, or certain not to vote”. Not surprisingly unlikely voters were much more likely to be uncertain of who they would support. A separate comparison removing those who were uncertain did not reveal statistically significant differences in levels of support between unlikely and likely voters.

Sample (n) Jay Aissa Jan d’Ailly Rob Deutschmann Karen Redman Uncertain
Unlikely Voter 214 2% 3% 7% 18% 69%
Likely Voter 312 6% 3% 10% 46% 35%

The results showed no statistically significant differences for candidate support by age or by city/township.

Sample (n) Jay Aissa Jan D’Ailly Rob Deutschmann Karen Redman
18 to 34 57 7% 7% 23% 63%
35 to 49 73 11% 11% 19% 59%
50 to 64 99 9% 4% 22% 65%
65 plus 77 6% 8% 9% 77%
Sample (n) Jay Aissa Jan D’Ailly Rob Deutschmann Karen Redman
Cambridge 57 7% 7% 23% 63%
Kitchener 150 9% 5% 16% 69%
Waterloo 67 9% 7% 16% 67%
Township 38 8% 11% 21% 61%

Issue Comparisons

The top three issues for respondents was also compared on age, gender, and cities/townships. There were no statistically significant differences amongst age groups or cities/townships with respect to the priority of different issues. However, gender showed statistically significant differences (at the 0.05 level). Keeping taxes low was the top issue for males while improving housing affordability and increasing social services were higher priorities for females.

Sample (n) Keeping Taxes Low Improving Housing Affordability Increasing Social Services Other
Female 235 18% 28% 21% 32%
Male 217 33% 21% 13% 33%

Survey Details

The Interactive Voice Response (IVR) survey was conducted by Professor Anthony Piscitelli with assistance from students in the Conestoga College Public Service Program. On Wednesday, October 10, 2018, students in the program participated in a workshop to learn about IVR surveys. The students then completed tasks such as creating the phone list, recording the survey, and gathering demographic data to build the survey weights. The survey was crowd funded by Craig Radcliffe in an attempt to understand the dynamics of the chair race and to bring more attention to the importance of this election.

Sampling Approach

The sample was created by randomly selecting Waterloo Region landlines as listed in a digital phone book. A sample of likely cellphone numbers was added by randomly selecting phone numbers that according to the Canadian Numbering Administrator were originally assigned to Waterloo Region. Sampling error exists as a result of this approach due to the mismatch created by the random dialling of phone numbers as opposed to randomly sampling actual Waterloo Region residents.

Response rate

The survey had a contact rate of 15%, indicating that of the 230,228 live lines that were called 35,681were answered by live potential respondents. The response rate was 0.23%, which is based on 530 respondents who completed the entire survey. The third question, which asked about candidate preferences, was answered by 585 respondents. All respondents who answered the first three questions were included in the results. It is worth noting that 13% of respondents were not eligible to participate due to being under 18 or not living in Waterloo Region.

Weights

Results of this survey have been weighted by age, gender, and city/township according to the 2016 census. The full weights are posted along with the raw data on OpenIcpsr.org and can be found by visiting: http://doi.org/10.3886/E106864V1

Margin of Error

Results are considered accurate +/-4.25%, 19 times out of 20. The margin of error on subsamples is higher.

Raw Data

Raw survey data is available on OpenIcpsr.org. The data can be found at: http://doi.org/10.3886/E106864V1

Disclaimer

The survey results will exhibit sampling error as a result of the mismatch created by the random dialling of phone numbers as opposed to randomly sampling actual Waterloo Region residents. These results also represent a snapshot in time and may not be indicative of the final election results. This survey was approved by the Conestoga College Research Ethics Board.

*Data updated on October 21, 2018 to correct undecided voters table totals